gcc/ChangeLog

   1 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
   2
   3         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
   4         fields to the reg_addr array that describes the valid addressing
   5         mode for any register, general purpose registers, floating point
   6         registers, and Altivec registers.
   7         (FIRST_RELOAD_REG_CLASS): Likewise.
   8         (LAST_RELOAD_REG_CLASS): Likewise.
   9         (struct reload_reg_map_type): Likewise.
  10         (reload_reg_map_type): Likewise.
  11         (RELOAD_REG_VALID): Likewise.
  12         (RELOAD_REG_MULTIPLE): Likewise.
  13         (RELOAD_REG_INDEXED): Likewise.
  14         (RELOAD_REG_OFFSET): Likewise.
  15         (RELOAD_REG_PRE_INCDEC): Likewise.
  16         (RELOAD_REG_PRE_MODIFY): Likewise.
  17         (reg_addr): Likewise.
  18         (mode_supports_pre_incdec_p): New helper functions to say whether
  19         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
  20         (mode_supports_pre_modify_p): Likewise.
  21         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
  22         print the valid address mode bits for each mode.
  23         (rs6000_debug_print_mode): Likewise.
  24         (rs6000_debug_reg_global): Likewise.
  25         (rs6000_setup_reg_addr_masks): New function to set up the address
  26         mask bits for each type.
  27         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
  28         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
  29         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
  30         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
  31         PRE_MODIFY are supported.
  32         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
  33         registers, instead of {src,dest}_av_p.
  34         (rs6000_print_options_internal): Tweak the debug output slightly.
  35
  36 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
  37
  38         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
  39         isa attribute.
  40
  41 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
  42
  43         * tree-flow.h (struct omp_region): Move to omp-low.c.
  44         Remove omp_ prototypes and variables.
  45         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
  46         (copy_var_decl): Relocate prototype from tree-flow.h.
  47         * gimple.c (copy_var_decl): Relocate from omp-low.c.
  48         * tree.h: Move prototype to omp-low.h.
  49         * omp-low.h: New File.  Relocate prototypes here.
  50         * omp-low.c (struct omp_region): Make local here.
  51         (root_omp_region): Make static.
  52         (copy_var_decl) Move to gimple.c.
  53         (new_omp_region): Make static.
  54         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
  55         * tree-cfg.c: Include omp-low.h.
  56         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
  57         * gimplify.c: Include omp-low.h.
  58         * tree-parloops.c: Likewise.
  59
  60 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
  61
  62         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
  63         isa attribute.
  64
  65 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
  66
  67         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
  68         operands are in memory, prefer to force non-matched operand 1 to
  69         the register.
  70
  71 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
  72
  73         PR target/58673
  74         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
  75         restrict TImode addresses to single indirect registers if both
  76         -mquad-memory and -mvsx-timode are used.
  77         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
  78         we should emit load/store quad.  Remove using %y for quad memory
  79         addresses.
  80
  81         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
  82         constraints to allow load/store quad on machines where TImode is
  83         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
  84         for TImode to load integer constants.
  85
  86 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  87
  88         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
  89
  90 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
  91
  92         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
  93         handling of STACK_REG.
  94
  95 2013-10-17  Richard Biener  <rguenther@suse.de>
  96
  97         PR tree-optimization/58143
  98         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
  99         New function.
 100         (rewrite_to_defined_overflow): Likewise.
 101         (move_computations_dom_walker::before_dom): Rewrite stmts
 102         with undefined signed overflow that are not always executed
 103         into unsigned arithmetic.
 104
 105 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
 106
 107         PR target/57756
 108         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
 109         explicit isa flag to be an options variable, instead of using
 110         global_options_set.  Remove define from rs6000.h.
 111         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
 112
 113         * config/rs6000/rs6000.c (rs6000_option_override_internal):
 114         Initialize rs6000_isa_flags_explicit.
 115         (rs6000_function_specific_save): Add gcc_options* parameter, so
 116         that the powerpc builds after the 2013-10-15 changes.
 117         (rs6000_function_specific_restore): Likewise.
 118
 119 2013-10-16  DJ Delorie  <dj@redhat.com>
 120
 121         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
 122         op is a REG before checking REGNO.
 123         (rl78_alloc_physical_registers): Verify pattern is a SET before
 124         checking SET_SRC.
 125
 126 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 127
 128         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
 129         endianness.
 130         (vec_unpacks_lo_v4sf): Likewise.
 131         (vec_unpacks_float_hi_v4si): Likewise.
 132         (vec_unpacks_float_lo_v4si): Likewise.
 133         (vec_unpacku_float_hi_v4si): Likewise.
 134         (vec_unpacku_float_lo_v4si): Likewise.
 135
 136 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 137
 138         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
 139         (vsx_concat_v2sf): Likewise.
 140
 141 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
 142
 143         * config/aarch64/aarch64.md
 144         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
 145
 146 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
 147
 148         PR tree-optimization/58697
 149         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
 150         estimated_loop_iterations_int.
 151         (max_stmt_executions_int): Call get_max_loop_iterations_int.
 152         (get_max_loop_iterations_int): New.  HWINT version of
 153         get_max_loop_iterations.
 154         * cfgloop.h: Add prototypes.
 155         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
 156         * loop-unroll.c (decide_peel_once_rolling): Call
 157         get_estimated_loop_iterations_int.
 158         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
 159         * tree-ssa-loop-niter.h: Tweak prototypes.
 160
 161 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
 162
 163         * gengtype-parse.c (struct_field_seq): Ignore access-control
 164         keywords ("public:" etc).
 165
 166 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
 167
 168         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
 169         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
 170
 171 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
 172
 173         * config/arm/arm.opt (mlra): New option.
 174         * config/arm/arm.c (arm_lra_p): New function.
 175         (TARGET_LRA_P): Define.
 176
 177 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
 178
 179         * tree-core.h (tree_code_name): Remove.
 180         * tree.h (get_tree_code_name): New prototype.
 181         * tree.c (tree_code_name): Make static.
 182         (get_tree_code_name): New function.
 183         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
 184         tree_class_check_failed, tree_range_check_failed,
 185         tree_not_class_check_failed, omp_clause_check_failed,
 186         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
 187         wrapper get_tree_code_name instead of calling tree_code_name directly.
 188         * tree-vrp.c (dump_asserts_for): Likewise.
 189         * tree-dump.c (dequeue_and_dump): Likewise.
 190         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
 191         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
 192         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
 193         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
 194         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
 195         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
 196         dump_gimple_omp_for): Likewise.
 197         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
 198         * tree-ssa-pre.c (print_pre_expr): Likewise.
 199         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
 200         * print-tree.c (print_node_brief, print_node): Likewise.
 201         * gimple.c (gimple_check_failed): Likewise.
 202         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
 203         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
 204         * config/mep/mep.c (mep_validate_vliw): Likewise.
 205         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
 206         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
 207
 208 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
 209
 210         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
 211         for AMD bdver3.
 212
 213 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
 214
 215         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
 216         (MULTILIB_MATCHES): Add multilib for -march=v8.
 217
 218 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
 219
 220         PR target/57756
 221         * optc-save-gen.awk: Add extra parameter to the save and restore
 222         target calls.
 223         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
 224         * tree.c (build_optimization_node): New parameter.  Add extra parameter
 225         to call to cl_optimization_save.
 226         (build_target_option_node): New parameter. Add extra parameter
 227         to call to cl_target_option_save.
 228         * tree.h (build_optimization_node): New parameter.
 229         (build_target_option_node): New parameter.
 230         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
 231         build_optimization_node and build_target_option_node.
 232         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
 233         (handle_pragma_push_options): Ditto.
 234         * toplev.c (process_options): Ditto.
 235         * opts.c (init_options_struct): Check for opts_set non-null.
 236         * target.def (target_option.save): New parameter.
 237         (target_option.restore): New parameter.
 238         * tm.texi: Generate.
 239         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
 240         (ix86_pragma_target_parse): Ditto.
 241         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
 242         parameters.
 243         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
 244         to  build_optimization_node and build_target_option_node.
 245         (rs6000_valid_attribute_p): Ditto.
 246         (rs6000_pragma_target_parse): Ditto.
 247         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
 248         data.
 249         * config/i386/i386.h:
 250         TARGET_64BIT_P: New Macro
 251         TARGET_MMX_P: New Macro.
 252         TARGET_3DNOW_P: New Macro.
 253         TARGET_3DNOW_A_P: New Macro.
 254         TARGET_SSE_P: New Macro.
 255         TARGET_SSE2_P: New Macro.
 256         TARGET_SSE3_P: New Macro.
 257         TARGET_SSSE3_P: New Macro.
 258         TARGET_SSE4_1_P: New Macro.
 259         TARGET_SSE4_2_P: New Macro.
 260         TARGET_AVX_P: New Macro.
 261         TARGET_AVX2_P: New Macro.
 262         TARGET_AVX512F_P: New Macro.
 263         TARGET_AVX512PF_P: New Macro.
 264         TARGET_AVX512ER_P: New Macro.
 265         TARGET_AVX512CD_P: New Macro.
 266         TARGET_FMA_P: New Macro.
 267         TARGET_SSE4A_P: New Macro.
 268         TARGET_FMA4_P: New Macro.
 269         TARGET_XOP_P: New Macro.
 270         TARGET_LWP_P: New Macro.
 271         TARGET_ABM_P: New Macro.
 272         TARGET_BMI_P: New Macro.
 273         TARGET_BMI2_P: New Macro.
 274         TARGET_LZCNT_P: New Macro.
 275         TARGET_TBM_P: New Macro.
 276         TARGET_POPCNT_P: New Macro.
 277         TARGET_SAHF_P: New Macro.
 278         TARGET_MOVBE_P: New Macro.
 279         TARGET_CRC32_P: New Macro.
 280         TARGET_AES_P: New Macro.
 281         TARGET_PCLMUL_P: New Macro.
 282         TARGET_CMPXCHG16B_P: New Macro.
 283         TARGET_FSGSBASE_P: New Macro.
 284         TARGET_RDRND_P: New Macro.
 285         TARGET_F16C_P: New Macro.
 286         TARGET_RTM_P: New Macro.
 287         TARGET_HLE_P: New Macro.
 288         TARGET_RDSEED_P: New Macro.
 289         TARGET_PRFCHW_P: New Macro.
 290         TARGET_ADX_P: New Macro.
 291         TARGET_FXSR_P: New Macro.
 292         TARGET_XSAVE_P: New Macro.
 293         TARGET_XSAVEOPT_P: New Macro.
 294         TARGET_LP64_P: New Macro.
 295         TARGET_X32_P: New Macro.
 296         TARGET_FPMATH_DEFAULT_P: New Macro.
 297         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
 298         * config/i386/i386.c (ix86_option_override_internal): New parameters.
 299         opts and opts_set.
 300         Change ix86_tune_string to access opts->x_ix86_tune_string.
 301         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
 302         Change ix86_arch_string to access opts->x_ix86_arch_string.
 303         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
 304         Change ix86_pmode to access opts->x_ix86_pmode.
 305         Change ix86_abi to access opts->x_ix86_abi.
 306         Change ix86_cmodel to access opts->x_ix86_cmodel.
 307         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
 308         Change ix86_isa_flags_explicit to access
 309         opts->x_ix86_isa_flags_explicit.
 310         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
 311         Change ix86_regparm to access opts->x_ix86_regparm.
 312         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
 313         Change ix86_preferred_stack_boundary_arg to access
 314         opts->x_ix86_preferred_stack_boundary_arg.
 315         Change ix86_force_align_arg_pointer to access
 316         opts->x_ix86_force_align_arg_pointer.
 317         Change ix86_incoming_stack_boundar_arg to access
 318         opts->x_ix86_incoming_stack_boundar_arg.
 319         Change ix86_fpmath to access opts->x_ix86_fpmath.
 320         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
 321         Change ix86_recip_name to access opts->x_ix86_recip_name.
 322         Change ix86_stack_protector_guard to access
 323         opts->x_ix86_stack_protector_guard.
 324         Change ix86_tune_memcpy_strategy to access
 325         opts->x_ix86_tune_memcpy_strategy.
 326         Change ix86_tune_memset_strategy to access
 327         opts->x_ix86_tune_memset_strategy.
 328         Change global_options to access opts.
 329         Change global_options_set to access opts_set.
 330         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
 331         Change TARGET_MMX to TARGET_MMX_P (opts->...).
 332         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
 333         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
 334         Change TARGET_SSE to TARGET_SSE_P (opts->...).
 335         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
 336         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
 337         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
 338         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
 339         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
 340         Change TARGET_AVX to TARGET_AVX_P (opts->...).
 341         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
 342         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
 343         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
 344         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
 345         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
 346         Change TARGET_FMA to TARGET_FMA_P (opts->...).
 347         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
 348         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
 349         Change TARGET_XOP to TARGET_XOP_P (opts->...).
 350         Change TARGET_LWP to TARGET_LWP_P (opts->...).
 351         Change TARGET_ABM to TARGET_ABM_P (opts->...).
 352         Change TARGET_BMI to TARGET_BMI_P (opts->...).
 353         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
 354         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
 355         Change TARGET_TBM to TARGET_TBM_P (opts->...).
 356         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
 357         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
 358         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
 359         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
 360         Change TARGET_AES to TARGET_AES_P (opts->...).
 361         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
 362         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
 363         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
 364         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
 365         Change TARGET_F16C to TARGET_F16C_P (opts->...).
 366         Change TARGET_RTM to TARGET_RTM_P (opts->...).
 367         Change TARGET_HLE to TARGET_HLE_P (opts->...).
 368         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
 369         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
 370         Change TARGET_ADX to TARGET_ADX_P (opts->...).
 371         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
 372         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
 373         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
 374         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
 375         Change TARGET_X32 to TARGET_X32_P (opts->...).
 376         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
 377         Change TARGET_FLOAT_RETURNS_IN_80387 to
 378         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
 379         (ix86_function_specific_save): New parameter. Use opts-> fields
 380         to replace global fields.
 381         (ix86_function_specific_restore): Ditto.
 382         (ix86_valid_target_attribute_inner_p): New parameters.
 383         Fix recursive call.
 384         Fix call to ix86_handle_option and set_option.
 385         (ix86_valid_target_attribute_tree): New parameters.
 386         Change global_options to access opts.
 387         Change global_options_set to access opts_set.
 388         Fix call to ix86_valid_target_attribute_inner_p.
 389         Change ix86_tune_string to access opts->x_ix86_tune_string.
 390         Change ix86_arch_string to access opts->x_ix86_arch_string.
 391         Change ix86_fpmath to access opts->x_ix86_fpmath
 392         Fix call to ix86_option_override_internal.
 393         Fix call to ix86_add_new_builtins.
 394         Fix calls to build_optimization_node and build_target_option_node.
 395         (ix86_valid_target_attribute_p): Remove access to global_options.
 396         Use new gcc_options structure func_options.
 397         Fix call to ix86_valid_target_attribute_tree.
 398         Fix call to  build_optimization_node.
 399         (get_builtin_code_for_version): Fix call to
 400         ix86_valid_target_attribute_tree.
 401
 402 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
 403
 404         * Makefile.in (PICFLAG): New.
 405         (enable_host_shared): New.
 406         (INTERNAL_CFLAGS): Use PICFLAG.
 407         (LIBIBERTY): Use pic build of libiberty.a if configured with
 408         --enable-host-shared.
 409         * configure.ac: Add --enable-host-shared, setting up new
 410         PICFLAG variable.
 411         * configure: Regenerate.
 412         * doc/install.texi (--enable-shared): Add note contrasting it with ...
 413         (--enable-host-shared): New option.
 414
 415 2013-10-15  Richard Biener  <rguenther@suse.de>
 416
 417         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
 418         for built-in functions.
 419
 420 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
 421
 422         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
 423         (optimize_range_tests_1): New function,
 424         extracted from optimize_range_tests.
 425         (optimize_range_tests_xor): Similarly.
 426         (optimize_range_tests_diff): New function.
 427         (optimize_range_tests): Use optimize_range_tests_1.
 428
 429 2013-10-15  Cong Hou  <congh@google.com>
 430
 431         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
 432         requirement of the reduction pattern so that one operand of the
 433         reduction operation can come from outside of the loop.
 434
 435 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 436
 437         * config/arm/neon-schedgen.ml: Remove.
 438         * config/arm/cortex-a9-neon.md: Remove comment regarding
 439         neon-schedgen.ml.
 440
 441 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 442
 443         * config/arm/types: Remove old neon types.
 444
 445 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 446
 447         * config/arm/cortex-a7.md
 448         (cortex_a7_neon_type): New.
 449         (cortex_a7_neon_mul): Update for new types.
 450         (cortex_a7_neon_mla): Likewise.
 451         (cortex_a7_neon): Likewise.
 452
 453 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 454
 455         * config/arm/cortex-a15-neon.md
 456         (cortex_a15_neon_type): New,
 457
 458         (cortex_a15_neon_int_1): Remove.
 459         (cortex_a15_neon_int_2): Likewise.
 460         (cortex_a15_neon_int_3): Likewise.
 461         (cortex_a15_neon_int_4): Likewise.
 462         (cortex_a15_neon_int_5): Likewise.
 463         (cortex_a15_neon_vqneg_vqabs): Likewise.
 464         (cortex_a15_neon_vmov): Likewise.
 465         (cortex_a15_neon_vaba): Likewise.
 466         (cortex_a15_neon_vaba_qqq): Likewise.
 467         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 468         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 469         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 470         Likewise.
 471         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 472         (cortex_a15_neon_mla_qqq_8_16): Likewise.
 473         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
 474         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 475         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 476         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
 477         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 478         (cortex_a15_neon_shift_1): Likewise.
 479         (cortex_a15_neon_shift_2): Likewise.
 480         (cortex_a15_neon_shift_3): Likewise.
 481         (cortex_a15_neon_vshl_ddd): Likewise.
 482         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 483         (cortex_a15_neon_vsra_vrsra): Likewise.
 484         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
 485         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
 486         (cortex_a15_neon_bp_3cycle): Likewise.
 487         (cortex_a15_neon_ldm_2): Likewise.
 488         (cortex_a15_neon_stm_2): Likewise.
 489         (cortex_a15_neon_mcr): Likewise.
 490         (cortex_a15_neon_mrc): Likewise.
 491         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
 492         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
 493         (cortex_a15_neon_fp_vmul_ddd): Likewise.
 494         (cortex_a15_neon_fp_vmul_qqd): Likewise.
 495         (cortex_a15_neon_fp_vmla_ddd): Likewise.
 496         (cortex_a15_neon_fp_vmla_qqq): Likewise.
 497         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
 498         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
 499         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 500         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 501         (cortex_a15_neon_bp_simple): Likewise.
 502         (cortex_a15_neon_bp_2cycle): Likewise.
 503         (cortex_a15_neon_bp_3cycle): Likewise.
 504         (cortex_a15_neon_vld1_1_2_regs): Likewise.
 505         (cortex_a15_neon_vld1_3_4_regs): Likewise.
 506         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 507         (cortex_a15_neon_vld2_4_regs): Likewise.
 508         (cortex_a15_neon_vld3_vld4): Likewise.
 509         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 510         (cortex_a15_neon_vst1_3_4_regs): Likewise.
 511         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
 512         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
 513         (cortex_a15_neon_vst3_vst4): Rename to...
 514         (cortex_a15_neon_vst4): This, update for new attributes.
 515         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
 516         (cortex_a15_neon_vld3_vld4_lane): Likewise.
 517         (cortex_a15_neon_vst1_vst2_lane): Likewise.
 518         (cortex_a15_neon_vst3_vst4_lane): Likewise.
 519         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
 520         (cortex_a15_neon_ldm_2): Likewise.
 521         (cortex_a15_neon_stm_2): Likewise.
 522         (cortex_a15_neon_mcr): Likewise.
 523         (cortex_a15_neon_mcr_2_mcrr): Likewise.
 524         (cortex_a15_neon_mrc): Likewise.
 525         (cortex_a15_neon_mrrc): Likewise.
 526
 527         (cortex_a15_neon_abd): New.
 528         (cortex_a15_neon_abd_q): Likewise.
 529         (cortex_a15_neon_aba): Likewise.
 530         (cortex_a15_neon_aba_q): Likewise.
 531         (cortex_a15_neon_acc): Likewise.
 532         (cortex_a15_neon_acc_q): Likewise.
 533         (cortex_a15_neon_arith_basic): Likewise.
 534         (cortex_a15_neon_arith_complex): Likewise.
 535         (cortex_a15_neon_multiply): Likewise.
 536         (cortex_a15_neon_multiply_q): Likewise.
 537         (cortex_a15_neon_mla): Likewise.
 538         (cortex_a15_neon_mla_q): Likewise.
 539         (cortex_a15_neon_sat_mla_long): Likewise.
 540         (cortex_a15_neon_shift_acc): Likewise.
 541         (cortex_a15_neon_shift_imm_basic): Likewise.
 542         (cortex_a15_neon_shift_imm_complex): Likewise.
 543         (cortex_a15_neon_shift_reg_basic): Likewise.
 544         (cortex_a15_neon_shift_reg_basic_q): Likewise.
 545         (cortex_a15_neon_shift_reg_complex): Likewise.
 546         (cortex_a15_neon_shift_reg_complex_q): Likewise.
 547         (cortex_a15_neon_fp_negabs): Likewise
 548         (cortex_a15_neon_fp_arith): Likewise
 549         (cortex_a15_neon_fp_arith_q): Likewise
 550         (cortex_a15_neon_fp_cvt_int): Likewise
 551         (cortex_a15_neon_fp_cvt_int_q): Likewise
 552         (cortex_a15_neon_fp_cvt_16): Likewise
 553         (cortex_a15_neon_fp_mul): Likewise
 554         (cortex_a15_neon_fp_mul_q): Likewise
 555         (cortex_a15_neon_fp_mla): Likewise
 556         (cortex_a15_neon_fp_mla_q): Likewise
 557         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
 558         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
 559         (cortex_a15_neon_bitops): Likewise.
 560         (cortex_a15_neon_bitops_q): Likewise.
 561         (cortex_a15_neon_from_gp): Likewise.
 562         (cortex_a15_neon_from_gp_q): Likewise.
 563         (cortex_a15_neon_tbl3_tbl4): Likewise.
 564         (cortex_a15_neon_zip_q): Likewise.
 565         (cortex_a15_neon_to_gp): Likewise.
 566         (cortex_a15_neon_load_a): Likewise.
 567         (cortex_a15_neon_load_b): Likewise.
 568         (cortex_a15_neon_load_c): Likewise.
 569         (cortex_a15_neon_load_d): Likewise.
 570         (cortex_a15_neon_load_e): Likewise.
 571         (cortex_a15_neon_load_f): Likewise.
 572         (cortex_a15_neon_store_a): Likewise.
 573         (cortex_a15_neon_store_b): Likewise.
 574         (cortex_a15_neon_store_c): Likewise.
 575         (cortex_a15_neon_store_d): Likewise.
 576         (cortex_a15_neon_store_e): Likewise.
 577         (cortex_a15_neon_store_f): Likewise.
 578         (cortex_a15_neon_store_g): Likewise.
 579         (cortex_a15_neon_store_h): Likewise.
 580         (cortex_a15_vfp_to_from_gp): Likewise.
 581
 582 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 583
 584         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
 585
 586         (cortex_a9_neon_vshl_ddd): Remove.
 587         (cortex_a9_neon_vst3_vst4): Likewise.
 588         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
 589
 590         (cortex_a9_neon_bit_ops_q): New.
 591
 592         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
 593         (cortex_a9_neon_int_2): Likewise.
 594         (cortex_a9_neon_int_3): Likewise.
 595         (cortex_a9_neon_int_4): Likewise.
 596         (cortex_a9_neon_int_5): Likewise.
 597         (cortex_a9_neon_vqneg_vqabs): Likewise.
 598         (cortex_a9_neon_vmov): Likewise.
 599         (cortex_a9_neon_vaba): Likewise.
 600         (cortex_a9_neon_vaba_qqq): Likewise.
 601         (cortex_a9_neon_shift_1): Likewise.
 602         (cortex_a9_neon_shift_2): Likewise.
 603         (cortex_a9_neon_shift_3): Likewise.
 604         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 605         (cortex_a9_neon_vsra_vrsra): Likewise.
 606         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 607         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 608         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 609         Likewise.
 610         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 611         (cortex_a9_neon_mla_qqq_8_16): Likewise.
 612         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
 613         Likewise.
 614         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 615         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 616         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
 617         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 618         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
 619         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
 620         (cortex_a9_neon_fp_vsum): Likewise.
 621         (cortex_a9_neon_fp_vmul_ddd): Likewise.
 622         (cortex_a9_neon_fp_vmul_qqd): Likewise.
 623         (cortex_a9_neon_fp_vmla_ddd): Likewise.
 624         (cortex_a9_neon_fp_vmla_qqq): Likewise.
 625         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
 626         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
 627         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 628         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 629         (cortex_a9_neon_bp_simple): Likewise.
 630         (cortex_a9_neon_bp_2cycle): Likewise.
 631         (cortex_a9_neon_bp_3cycle): Likewise.
 632         (cortex_a9_neon_ldr): Likewise.
 633         (cortex_a9_neon_str): Likewise.
 634         (cortex_a9_neon_vld1_1_2_regs): Likewise.
 635         (cortex_a9_neon_vld1_3_4_regs): Likewise.
 636         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 637         (cortex_a9_neon_vld2_4_regs): Likewise.
 638         (cortex_a9_neon_vld3_vld4): Likewise.
 639         (cortex_a9_neon_vld1_vld2_lane): Likewise.
 640         (cortex_a9_neon_vld3_vld4_lane): Likewise.
 641         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
 642         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 643         (cortex_a9_neon_vst1_3_4_regs): Likewise.
 644         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
 645         (cortex_a9_neon_vst1_vst2_lane): Likewise.
 646         (cortex_a9_neon_vst3_vst4_lane): Likewise.
 647         (cortex_a9_neon_mcr): Likewise.
 648         (cortex_a9_neon_mcr_2_mcrr): Likewise.
 649         (cortex_a9_neon_mrc): Likewise.
 650         (cortex_a9_neon_mrrc): Likewise.
 651
 652 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 653
 654         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
 655
 656         (cortex_a8_neon_vshl_ddd): Remove.
 657         (cortex_a8_neon_vst3_vst4): Likewise.
 658         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
 659
 660         (cortex_a8_neon_bit_ops_q): New.
 661
 662         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
 663         (cortex_a8_neon_int_2): Likewise..
 664         (cortex_a8_neon_int_3): Likewise.
 665         (cortex_a8_neon_int_5): Likewise.
 666         (cortex_a8_neon_vqneg_vqabs): Likewise.
 667         (cortex_a8_neon_int_4): Likewise.
 668         (cortex_a8_neon_vaba): Likewise.
 669         (cortex_a8_neon_vaba_qqq): Likewise.
 670         (cortex_a8_neon_shift_1): Likewise.
 671         (cortex_a8_neon_shift_2): Likewise.
 672         (cortex_a8_neon_shift_3): Likewise.
 673         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 674         (cortex_a8_neon_vsra_vrsra): Likewise.
 675         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 676         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 677         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 678         Likewise.
 679         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 680         (cortex_a8_neon_mla_qqq_8_16): Likewise.
 681         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
 682         Likewise.
 683         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 684         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 685         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
 686         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 687         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
 688         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
 689         (cortex_a8_neon_fp_vsum): Likewise.
 690         (cortex_a8_neon_fp_vmul_ddd): Likewise.
 691         (cortex_a8_neon_fp_vmul_qqd): Likewise.
 692         (cortex_a8_neon_fp_vmla_ddd): Likewise.
 693         (cortex_a8_neon_fp_vmla_qqq): Likewise.
 694         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
 695         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
 696         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 697         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 698         (cortex_a8_neon_bp_simple): Likewise.
 699         (cortex_a8_neon_bp_2cycle): Likewise.
 700         (cortex_a8_neon_bp_3cycle): Likewise.
 701         (cortex_a8_neon_ldr): Likewise.
 702         (cortex_a8_neon_str): Likewise.
 703         (cortex_a8_neon_vld1_1_2_regs): Likewise.
 704         (cortex_a8_neon_vld1_3_4_regs): Likewise.
 705         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 706         (cortex_a8_neon_vld2_4_regs): Likewise.
 707         (cortex_a8_neon_vld3_vld4): Likewise.
 708         (cortex_a8_neon_vld1_vld2_lane): Likewise.
 709         (cortex_a8_neon_vld3_vld4_lane): Likewise.
 710         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 711         (cortex_a8_neon_vst1_3_4_regs): Likewise.
 712         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
 713         (cortex_a8_neon_vst1_vst2_lane): Likewise.
 714         (cortex_a8_neon_vst3_vst4_lane): Likewise.
 715         (cortex_a8_neon_mcr): Likewise.
 716         (cortex_a8_neon_mcr_2_mcrr): Likewise.
 717         (cortex_a8_neon_mrc): Likewise.
 718         (cortex_a8_neon_mrrc): Likewise.
 719
 720 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 721
 722         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
 723         (fp): New.
 724         * config/aarch64/aarch64-simd.md (neon_type): Remove.
 725         (aarch64_simd_dup<mode>): Add "type" attribute.
 726         (aarch64_dup_lane<mode>): Likewise.
 727         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
 728         (*aarch64_simd_mov<mode>): Likewise.
 729         (aarch64_simd_mov_from_<mode>low): Likewise.
 730         (aarch64_simd_mov_from_<mode>high): Likewise.
 731         (orn<mode>3): Likewise.
 732         (bic<mode>3): Likewise.
 733         (add<mode>3): Likewise.
 734         (sub<mode>3): Likewise.
 735         (mul<mode>3): Likewise.
 736         (*aarch64_mul3_elt<mode>): Likewise.
 737         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
 738         (*aarch64_mul3_elt_to_128df): Likewise.
 739         (*aarch64_mul3_elt_to_64v2df): Likewise.
 740         (neg<mode>2): Likewise.
 741         (abs<mode>2): Likewise.
 742         (abd<mode>_3): Likewise.
 743         (aba<mode>_3): Likewise.
 744         (fabd<mode>_3): Likewise.
 745         (*fabd_scalar<mode>3): Likewise.
 746         (and<mode>3): Likewise.
 747         (ior<mode>3): Likewise.
 748         (xor<mode>3): Likewise.
 749         (one_cmpl<mode>2): Likewise.
 750         (aarch64_simd_vec_set<mode>): Likewise.
 751         (aarch64_simd_lshr<mode>): Likewise.
 752         (aarch64_simd_ashr<mode>): Likewise.
 753         (aarch64_simd_imm_shl<mode>): Likewise.
 754         (aarch64_simd_reg_sshl<mode): Likewise.
 755         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
 756         (aarch64_simd_reg_shl<mode>_signed): Likewise.
 757         (aarch64_simd_vec_setv2di): Likewise.
 758         (aarch64_simd_vec_set<mode>): Likewise.
 759         (aarch64_mla<mode>): Likewise.
 760         (*aarch64_mla_elt<mode>): Likewise.
 761         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
 762         (aarch64_mls<mode>): Likewise.
 763         (*aarch64_mls_elt<mode>): Likewise.
 764         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
 765         (<su><maxmin><mode>3): Likewise.
 766         (move_lo_quad_<mode>): Likewise.
 767         (aarch64_simd_move_hi_quad_<mode>): Likewise.
 768         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
 769         (vec_pack_trunc_<mode>): Likewise.
 770         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
 771         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
 772         (*aarch64_<su>mlal_lo<mode>): Likewise.
 773         (*aarch64_<su>mlal_hi<mode>): Likewise.
 774         (*aarch64_<su>mlsl_lo<mode>): Likewise.
 775         (*aarch64_<su>mlsl_hi<mode>): Likewise.
 776         (*aarch64_<su>mlal<mode>): Likewise.
 777         (*aarch64_<su>mlsl<mode>): Likewise.
 778         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
 779         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
 780         (add<mode>3): Likewise.
 781         (sub<mode>3): Likewise.
 782         (mul<mode>3): Likewise.
 783         (div<mode>3): Likewise.
 784         (neg<mode>2): Likewise.
 785         (abs<mode>2): Likewise.
 786         (fma<mode>4): Likewise.
 787         (*aarch64_fma4_elt<mode>): Likewise.
 788         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
 789         (*aarch64_fma4_elt_to_128df): Likewise.
 790         (*aarch64_fma4_elt_to_64v2df): Likewise.
 791         (fnma<mode>4): Likewise.
 792         (*aarch64_fnma4_elt<mode>): Likewise.
 793         (*aarch64_fnma4_elt_<vswap_width_name><mode>
 794         (*aarch64_fnma4_elt_to_128df): Likewise.
 795         (*aarch64_fnma4_elt_to_64v2df): Likewise.
 796         (<frint_pattern><mode>2): Likewise.
 797         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
 798         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
 799         (vec_unpacks_lo_v4sf): Likewise.
 800         (aarch64_float_extend_lo_v2df): Likewise.
 801         (vec_unpacks_hi_v4sf): Likewise.
 802         (aarch64_float_truncate_lo_v2sf): Likewise.
 803         (aarch64_float_truncate_hi_v4sf): Likewise.
 804         (aarch64_vmls<mode>): Likewise.
 805         (<su><maxmin><mode>3): Likewise.
 806         (<maxmin_uns><mode>3): Likewise.
 807         (reduc_<sur>plus_<mode>): Likewise.
 808         (reduc_<sur>plus_v2di): Likewise.
 809         (reduc_<sur>plus_v2si): Likewise.
 810         (reduc_<sur>plus_<mode>): Likewise.
 811         (aarch64_addpv4sf): Likewise.
 812         (clz<mode>2): Likewise.
 813         (reduc_<maxmin_uns>_<mode>): Likewise.
 814         (reduc_<maxmin_uns>_v2di): Likewise.
 815         (reduc_<maxmin_uns>_v2si): Likewise.
 816         (reduc_<maxmin_uns>_<mode>): Likewise.
 817         (reduc_<maxmin_uns>_v4sf): Likewise.
 818         (aarch64_simd_bsl<mode>_internal): Likewise.
 819         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
 820         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
 821         (aarch64_get_lane<mode>): Likewise.
 822         (*aarch64_combinez<mode>): Likewise.
 823         (aarch64_combine<mode>): Likewise.
 824         (aarch64_simd_combine<mode>): Likewise.
 825         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
 826         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
 827         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
 828         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
 829         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
 830         (aarch64_<sur>h<addsub><mode>): Likewise.
 831         (aarch64_<sur><addsub>hn<mode>): Likewise.
 832         (aarch64_<sur><addsub>hn2<mode>): Likewise.
 833         (aarch64_pmul<mode>): Likewise.
 834         (aarch64_<su_optab><optab><mode>): Likewise.
 835         (aarch64_<sur>qadd<mode>): Likewise.
 836         (aarch64_sqmovun<mode>): Likewise.
 837         (aarch64_<sur>qmovn<mode>): Likewise.
 838         (aarch64_s<optab><mode>): Likewise.
 839         (aarch64_sq<r>dmulh<mode>): Likewise.
 840         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
 841         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
 842         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
 843         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
 844         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 845         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 846         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
 847         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
 848         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
 849         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
 850         (aarch64_sqdmull<mode>): Likewise.
 851         (aarch64_sqdmull_lane<mode>_internal): Likewise.
 852         (aarch64_sqdmull_n<mode>): Likewise.
 853         (aarch64_sqdmull2<mode>_internal): Likewise.
 854         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
 855         (aarch64_sqdmull2_n<mode>_internal): Likewise.
 856         (aarch64_<sur>shl<mode>): Likewise.
 857         (aarch64_<sur>q<r>shl<mode>
 858         (aarch64_<sur>shll_n<mode>): Likewise.
 859         (aarch64_<sur>shll2_n<mode>): Likewise.
 860         (aarch64_<sur>shr_n<mode>): Likewise.
 861         (aarch64_<sur>sra_n<mode>): Likewise.
 862         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
 863         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
 864         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
 865         (aarch64_cm<optab><mode>): Likewise.
 866         (aarch64_cm<optab>di): Likewise.
 867         (aarch64_cm<optab><mode>): Likewise.
 868         (aarch64_cm<optab>di): Likewise.
 869         (aarch64_cmtst<mode>): Likewise.
 870         (aarch64_cmtstdi): Likewise.
 871         (aarch64_cm<optab><mode>): Likewise.
 872         (*aarch64_fac<optab><mode>): Likewise.
 873         (aarch64_addp<mode>): Likewise.
 874         (aarch64_addpdi): Likewise.
 875         (sqrt<mode>2): Likewise.
 876         (vec_load_lanesoi<mode>): Likewise.
 877         (vec_store_lanesoi<mode>): Likewise.
 878         (vec_load_lanesci<mode>): Likewise.
 879         (vec_store_lanesci<mode>): Likewise.
 880         (vec_load_lanesxi<mode>): Likewise.
 881         (vec_store_lanesxi<mode>): Likewise.
 882         (*aarch64_mov<mode>): Likewise.
 883         (aarch64_ld2<mode>_dreg): Likewise.
 884         (aarch64_ld2<mode>_dreg): Likewise.
 885         (aarch64_ld3<mode>_dreg): Likewise.
 886         (aarch64_ld3<mode>_dreg): Likewise.
 887         (aarch64_ld4<mode>_dreg): Likewise.
 888         (aarch64_ld4<mode>_dreg): Likewise.
 889         (aarch64_tbl1<mode>): Likewise.
 890         (aarch64_tbl2v16qi): Likewise.
 891         (aarch64_combinev16qi): Likewise.
 892         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
 893         (aarch64_st2<mode>_dreg): Likewise.
 894         (aarch64_st2<mode>_dreg): Likewise.
 895         (aarch64_st3<mode>_dreg): Likewise.
 896         (aarch64_st3<mode>_dreg): Likewise.
 897         (aarch64_st4<mode>_dreg): Likewise.
 898         (aarch64_st4<mode>_dreg): Likewise.
 899         (*aarch64_simd_ld1r<mode>): Likewise.
 900         (aarch64_frecpe<mode>): Likewise.
 901         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
 902         (aarch64_frecps<mode>): Likewise.
 903
 904 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 905
 906         * config/arm/iterators.md (V_elem_ch): New.
 907         (q): Likewise.
 908         (VQH_type): Likewise.
 909         * config/arm/arm.md (is_neon_type): New.
 910         (conds): Use is_neon_type.
 911         (anddi3_insn): Update type attribute.
 912         (xordi3_insn): Likewise.
 913         (one_cmpldi2): Likewise.
 914         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
 915         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
 916         (*movmisalign<mode>_neon_store): Likewise.
 917         (*movmisalign<mode>_neon_load): Likewise.
 918         (vec_set<mode>_internal): Likewise.
 919         (vec_set<mode>_internal): Likewise.
 920         (vec_setv2di_internal): Likewise.
 921         (vec_extract<mode>): Likewise.
 922         (vec_extract<mode>): Likewise.
 923         (vec_extractv2di): Likewise.
 924         (*add<mode>3_neon): Likewise.
 925         (adddi3_neon): Likewise.
 926         (*sub<mode>3_neon): Likewise.
 927         (subdi3_neon): Likewise.
 928         (fma<VCVTF:mode>4): Likewise.
 929         (fma<VCVTF:mode>4_intrinsic): Likewise.
 930         (*fmsub<VCVTF:mode>4): Likewise.
 931         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
 932         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
 933         (ior<mode>3): Likewise.
 934         (and<mode>3): Likewise.
 935         (orn<mode>3_neon): Likewise.
 936         (orndi3_neon): Likewise.
 937         (bic<mode>3_neon): Likewise.
 938         (bicdi3_neon): Likewise.
 939         (xor<mode>3): Likewise.
 940         (one_cmpl<mode>2): Likewise.
 941         (abs<mode>2): Likewise.
 942         (neg<mode>2): Likewise.
 943         (negdi2_neon): Likewise.
 944         (*umin<mode>3_neon): Likewise.
 945         (*umax<mode>3_neon): Likewise.
 946         (*smin<mode>3_neon): Likewise.
 947         (*smax<mode>3_neon): Likewise.
 948         (vashl<mode>3): Likewise.
 949         (vashr<mode>3_imm): Likewise.
 950         (vlshr<mode>3_imm): Likewise.
 951         (ashl<mode>3_signed): Likewise.
 952         (ashl<mode>3_unsigned): Likewise.
 953         (neon_load_count): Likewise.
 954         (ashldi3_neon_noclobber): Likewise.
 955         (ashldi3_neon): Likewise.
 956         (signed_shift_di3_neon): Likewise.
 957         (unsigned_shift_di3_neon): Likewise.
 958         (ashrdi3_neon_imm_noclobber): Likewise.
 959         (lshrdi3_neon_imm_noclobber): Likewise.
 960         (<shift>di3_neon): Likewise.
 961         (widen_ssum<mode>3): Likewise.
 962         (widen_usum<mode>3): Likewise.
 963         (quad_halves_<code>v4si): Likewise.
 964         (quad_halves_<code>v4sf): Likewise.
 965         (quad_halves_<code>v8hi): Likewise.
 966         (quad_halves_<code>v16qi): Likewise.
 967         (reduc_splus_v2di): Likewise.
 968         (neon_vpadd_internal<mode>): Likewise.
 969         (neon_vpsmin<mode>): Likewise.
 970         (neon_vpsmax<mode>): Likewise.
 971         (neon_vpumin<mode>): Likewise.
 972         (neon_vpumax<mode>): Likewise.
 973         (*ss_add<mode>_neon): Likewise.
 974         (*us_add<mode>_neon): Likewise.
 975         (*ss_sub<mode>_neon): Likewise.
 976         (*us_sub<mode>_neon): Likewise.
 977         (neon_vadd<mode>_unspec): Likewise.
 978         (neon_vaddl<mode>): Likewise.
 979         (neon_vaddw<mode>): Likewise.
 980         (neon_vhadd<mode>): Likewise.
 981         (neon_vqadd<mode>): Likewise.
 982         (neon_vaddhn<mode>): Likewise.
 983         (neon_vmul<mode>): Likewise.
 984         (neon_vfms<VCVTF:mode>): Likewise.
 985         (neon_vmlal<mode>): Likewise.
 986         (neon_vmls<mode>): Likewise.
 987         (neon_vmlsl<mode>): Likewise.
 988         (neon_vqdmulh<mode>): Likewise.
 989         (neon_vqdmlal<mode>): Likewise.
 990         (neon_vqdmlsl<mode>): Likewise.
 991         (neon_vmull<mode>): Likewise.
 992         (neon_vqdmull<mode>): Likewise.
 993         (neon_vsub<mode>_unspec): Likewise.
 994         (neon_vsubl<mode>): Likewise.
 995         (neon_vsubw<mode>): Likewise.
 996         (neon_vqsub<mode>): Likewise.
 997         (neon_vhsub<mode>): Likewise.
 998         (neon_vsubhn<mode>): Likewise.
 999         (neon_vceq<mode>): Likewise.
1000         (neon_vcge<mode>): Likewise.
1001         (neon_vcgeu<mode>): Likewise.
1002         (neon_vcgt<mode>): Likewise.
1003         (neon_vcgtu<mode>): Likewise.
1004         (neon_vcle<mode>): Likewise.
1005         (neon_vclt<mode>): Likewise.
1006         (neon_vcage<mode>): Likewise.
1007         (neon_vcagt<mode>): Likewise.
1008         (neon_vtst<mode>): Likewise.
1009         (neon_vabd<mode>): Likewise.
1010         (neon_vabdl<mode>): Likewise.
1011         (neon_vaba<mode>): Likewise.
1012         (neon_vabal<mode>): Likewise.
1013         (neon_vmax<mode>): Likewise.
1014         (neon_vmin<mode>): Likewise.
1015         (neon_vpaddl<mode>): Likewise.
1016         (neon_vpadal<mode>): Likewise.
1017         (neon_vpmax<mode>): Likewise.
1018         (neon_vpmin<mode>): Likewise.
1019         (neon_vrecps<mode>): Likewise.
1020         (neon_vrsqrts<mode>): Likewise.
1021         (neon_vqabs<mode>): Likewise.
1022         (neon_vqneg<mode>): Likewise.
1023         (neon_vcls<mode>): Likewise.
1024         (clz<mode>2): Likewise.
1025         (popcount<mode>2): Likewise.
1026         (neon_vrecpe<mode>): Likewise.
1027         (neon_vrsqrte<mode>): Likewise.
1028         (neon_vget_lane<mode>_sext_internal): Likewise.
1029         (neon_vget_lane<mode>_zext_internal): Likewise.
1030         (neon_vdup_n<mode>): Likewise.
1031         (neon_vdup_n<mode>): Likewise.
1032         (neon_vdup_nv2di): Likewise.
1033         (neon_vdup_lane<mode>_interal): Likewise.
1034         (*neon_vswp<mode>): Likewise.
1035         (neon_vcombine<mode>): Likewise.
1036         (float<mode><V_cvtto>2): Likewise.
1037         (floatuns<mode><V_cvtto>2): Likewise.
1038         (fix_trunc<mode><V_cvtto>2): Likewise.
1039         (fixuns_trunc<mode><V_cvtto>2
1040         (neon_vcvt<mode>): Likewise.
1041         (neon_vcvt<mode>): Likewise.
1042         (neon_vcvtv4sfv4hf): Likewise.
1043         (neon_vcvtv4hfv4sf): Likewise.
1044         (neon_vcvt_n<mode>): Likewise.
1045         (neon_vcvt_n<mode>): Likewise.
1046         (neon_vmovn<mode>): Likewise.
1047         (neon_vqmovn<mode>): Likewise.
1048         (neon_vqmovun<mode>): Likewise.
1049         (neon_vmovl<mode>): Likewise.
1050         (neon_vmul_lane<mode>): Likewise.
1051         (neon_vmul_lane<mode>): Likewise.
1052         (neon_vmull_lane<mode>): Likewise.
1053         (neon_vqdmull_lane<mode>): Likewise.
1054         (neon_vqdmulh_lane<mode>): Likewise.
1055         (neon_vqdmulh_lane<mode>): Likewise.
1056         (neon_vmla_lane<mode>): Likewise.
1057         (neon_vmla_lane<mode>): Likewise.
1058         (neon_vmlal_lane<mode>): Likewise.
1059         (neon_vqdmlal_lane<mode>): Likewise.
1060         (neon_vmls_lane<mode>): Likewise.
1061         (neon_vmls_lane<mode>): Likewise.
1062         (neon_vmlsl_lane<mode>): Likewise.
1063         (neon_vqdmlsl_lane<mode>): Likewise.
1064         (neon_vext<mode>): Likewise.
1065         (neon_vrev64<mode>): Likewise.
1066         (neon_vrev32<mode>): Likewise.
1067         (neon_vrev16<mode>): Likewise.
1068         (neon_vbsl<mode>_internal): Likewise.
1069         (neon_vshl<mode>): Likewise.
1070         (neon_vqshl<mode>): Likewise.
1071         (neon_vshr_n<mode>): Likewise.
1072         (neon_vshrn_n<mode>): Likewise.
1073         (neon_vqshrn_n<mode>): Likewise.
1074         (neon_vqshrun_n<mode>): Likewise.
1075         (neon_vshl_n<mode>): Likewise.
1076         (neon_vqshl_n<mode>): Likewise.
1077         (neon_vqshlu_n<mode>): Likewise.
1078         (neon_vshll_n<mode>): Likewise.
1079         (neon_vsra_n<mode>): Likewise.
1080         (neon_vsri_n<mode>): Likewise.
1081         (neon_vsli_n<mode>): Likewise.
1082         (neon_vtbl1v8qi): Likewise.
1083         (neon_vtbl2v8qi): Likewise.
1084         (neon_vtbl3v8qi): Likewise.
1085         (neon_vtbl4v8qi): Likewise.
1086         (neon_vtbl1v16qi): Likewise.
1087         (neon_vtbl2v16qi): Likewise.
1088         (neon_vcombinev16qi): Likewise.
1089         (neon_vtbx1v8qi): Likewise.
1090         (neon_vtbx2v8qi): Likewise.
1091         (neon_vtbx3v8qi): Likewise.
1092         (neon_vtbx4v8qi): Likewise.
1093         (*neon_vtrn<mode>_insn): Likewise.
1094         (*neon_vzip<mode>_insn): Likewise.
1095         (*neon_vuzp<mode>_insn): Likewise.
1096         (neon_vld1<mode>): Likewise.
1097         (neon_vld1_lane<mode>): Likewise.
1098         (neon_vld1_lane<mode>): Likewise.
1099         (neon_vld1_dup<mode>): Likewise.
1100         (neon_vld1_dup<mode>): Likewise.
1101         (neon_vld1_dupv2di): Likewise.
1102         (neon_vst1<mode>): Likewise.
1103         (neon_vst1_lane<mode>): Likewise.
1104         (neon_vst1_lane<mode>): Likewise.
1105         (neon_vld2<mode>): Likewise.
1106         (neon_vld2<mode>): Likewise.
1107         (neon_vld2_lane<mode>): Likewise.
1108         (neon_vld2_lane<mode>): Likewise.
1109         (neon_vld2_dup<mode>): Likewise.
1110         (neon_vst2<mode>): Likewise.
1111         (neon_vst2<mode>): Likewise.
1112         (neon_vst2_lane<mode>): Likewise.
1113         (neon_vst2_lane<mode>): Likewise.
1114         (neon_vld3<mode>): Likewise.
1115         (neon_vld3qa<mode>): Likewise.
1116         (neon_vld3qb<mode>): Likewise.
1117         (neon_vld3_lane<mode>): Likewise.
1118         (neon_vld3_lane<mode>): Likewise.
1119         (neon_vld3_dup<mode>): Likewise.
1120         (neon_vst3<mode>): Likewise.
1121         (neon_vst3qa<mode>): Likewise.
1122         (neon_vst3qb<mode>): Likewise.
1123         (neon_vst3_lane<mode>): Likewise.
1124         (neon_vst3_lane<mode>): Likewise.
1125         (neon_vld4<mode>): Likewise.
1126         (neon_vld4qa<mode>): Likewise.
1127         (neon_vld4qb<mode>): Likewise.
1128         (neon_vld4_lane<mode>): Likewise.
1129         (neon_vld4_lane<mode>): Likewise.
1130         (neon_vld4_dup<mode>): Likewise.
1131         (neon_vst4<mode>): Likewise.
1132         (neon_vst4qa<mode>): Likewise.
1133         (neon_vst4qb<mode>): Likewise.
1134         (neon_vst4_lane<mode>): Likewise.
1135         (neon_vst4_lane<mode>): Likewise.
1136         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1137         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1138         (neon_vec_<US>mult_lo_<mode>): Likewise.
1139         (neon_vec_<US>mult_hi_<mode>): Likewise.
1140         (neon_vec_<US>shiftl_<mode>): Likewise.
1141         (neon_unpack<US>_<mode>): Likewise.
1142         (neon_vec_<US>mult_<mode>): Likewise.
1143         (vec_pack_trunc_<mode>): Likewise.
1144         (neon_vec_pack_trunc_<mode>): Likewise.
1145         (neon_vabd<mode>_2): Likewise.
1146         (neon_vabd<mode>_3): Likewise.
1147
1148 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1149
1150         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1151         (load_pair): Update type attribute.
1152         (store_pair): Update type attribute.
1153         * config/aarch64/iterators.md (q): New.
1154
1155 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1156
1157         * config/arm/types.md: Add new types for Neon insns.
1158
1159 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1160             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1161             Sergey Lega  <sergey.s.lega@intel.com>
1162             Anna Tikhonova  <anna.tikhonova@intel.com>
1163             Ilya Tocar  <ilya.tocar@intel.com>
1164             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1165             Ilya Verbin  <ilya.verbin@intel.com>
1166             Kirill Yukhin  <kirill.yukhin@intel.com>
1167             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1168
1169         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1170         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1171         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1172         (rcp14<mode>): New.
1173         (srcp14<mode>): Ditto.
1174         (rsqrt14<mode>): Ditto.
1175         (rsqrt14<mode>): Ditto.
1176         (avx512f_vmscalef<mode>): Ditto.
1177         (avx512f_scalef<mode>): Ditto.
1178         (avx512f_getexp<mode>): Ditto.
1179         (avx512f_sgetexp<mode>): Ditto.
1180         (avx512f_fixupimm<mode>): Ditto.
1181         (avx512f_sfixupimm<mode>): Ditto.
1182         (avx512f_rndscale<mode>): Ditto.
1183         (*avx512er_exp2<mode>): Ditto.
1184         (*avx512er_rcp28<mode>): Ditto.
1185         (avx512er_rsqrt28<mode>): Ditto.
1186         (avx512f_getmant<mode>): Ditto.
1187         (avx512f_getmant<mode>): Ditto.
1188         (avx512f_rndscale<mode>): Fix formatting.
1189
1190 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1191
1192         * ipa-utils.h (ipa_edge_within_scc): Declare.
1193         * ipa-cp.c (edge_within_scc): Moved...
1194         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1195
1196 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1197             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1198             Sergey Lega  <sergey.s.lega@intel.com>
1199             Anna Tikhonova  <anna.tikhonova@intel.com>
1200             Ilya Tocar  <ilya.tocar@intel.com>
1201             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1202             Ilya Verbin  <ilya.verbin@intel.com>
1203             Kirill Yukhin  <kirill.yukhin@intel.com>
1204             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1205
1206         * config/i386/predicates.md (const_8_to_15_operand): New.
1207         (const_16_to_31_operand): Ditto.
1208         * config/i386/sse.md (V8FI): New.
1209         (V16FI): Ditto.
1210         (reduc_splus_v8df): Ditto.
1211         (reduc_splus_v16sf): Ditto.
1212         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1213         (vec_extract_hi_<mode>): Ditto.
1214         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1215         (vec_set_lo_<mode>): Ditto.
1216         (vec_set_hi_<mode>): Ditto.
1217         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1218         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1219         (avx512f_pshufd_1): Ditto.
1220         (avx512f_broadcast<mode>): Ditto.
1221         (avx512f_broadcast<mode>): Ditto.
1222         (define_split): Split vec_extract_lo into move.
1223         (ssequartermode): Ditto.
1224         (ssedoublemode): Extened with wider modes.
1225         (vec_extract_lo_<mode>): Ditto.
1226
1227 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1228             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1229             Sergey Lega  <sergey.s.lega@intel.com>
1230             Anna Tikhonova  <anna.tikhonova@intel.com>
1231             Ilya Tocar  <ilya.tocar@intel.com>
1232             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1233             Ilya Verbin  <ilya.verbin@intel.com>
1234             Kirill Yukhin  <kirill.yukhin@intel.com>
1235             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1236
1237         * config/i386/predicates.md (register_or_constm1_operand): New.
1238         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1239         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1240         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1241         UNSPEC_SCATTER_PREFETCH
1242         (VI48_512): New.
1243         (avx512f_ucmp<mode>3): Ditto.
1244         (avx512f_vternlog<mode>): Ditto.
1245         (avx512f_align<mode>): Ditto.
1246         (<shift_insn><mode>3): Ditto.
1247         (avx512f_<rotate>v<mode>): Ditto.
1248         (avx512f_<rotate><mode>): Ditto.
1249         (avx512f_eq<mode>3): Ditto.
1250         (avx512f_eq<mode>3_1): Ditto.
1251         (avx512f_gt<mode>3): Ditto.
1252         (avx512f_testm<mode>3): Ditto.
1253         (avx512f_testnm<mode>3): Ditto.
1254         (avx512pf_gatherpf<mode>): Ditto.
1255         (*avx512pf_gatherpf<mode>_mask): Ditto.
1256         (*avx512pf_gatherpf<mode>): Ditto.
1257         (avx512pf_scatterpf<mode>): Ditto.
1258         (*avx512pf_scatterpf<mode>_mask): Ditto.
1259         (*avx512pf_scatterpf<mode>): Ditto.
1260         (avx512f_vec_dup_gpr<mode>): Ditto.
1261         (clz<mode>2): Ditto.
1262         (conflict<mode>): Ditto.
1263         (REDUC_SMINMAX_MODE): Extened with wider modes.
1264         (reduc_<code>_<mode>): Ditto.
1265         (vlshr<mode>3): Ditto.
1266         (vashl<mode>3): Ditto.
1267
1268 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1269             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1270             Sergey Lega  <sergey.s.lega@intel.com>
1271             Anna Tikhonova  <anna.tikhonova@intel.com>
1272             Ilya Tocar  <ilya.tocar@intel.com>
1273             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1274             Ilya Verbin  <ilya.verbin@intel.com>
1275             Kirill Yukhin  <kirill.yukhin@intel.com>
1276             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1277
1278         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1279         UNSPEC_SCATTER.
1280         (VI48F_512): New.
1281         (avx512fmaskmode): Ditto.
1282         (bcstscalarsuff): Ditto.
1283         (avx512f_blendm<mode>): Ditto.
1284         (cmp_imm_predicate): Ditto.
1285         (avx512f_cmp<mode>3): Ditto.
1286         (avx512f_vec_dup<mode>): Ditto.
1287         (avx512f_vec_dup_mem<mode>): Ditto.
1288         (avx512f_vpermi2var<mode>3): Ditto.
1289         (avx512f_vpermt2var<mode>3): Ditto.
1290         (vec_init<mode>): Ditto.
1291         (avx512f_gathersi<mode>): Ditto.
1292         (*avx512f_gathersi<mode>): Ditto.
1293         (*avx512f_gathersi<mode>_2): Ditto.
1294         (avx512f_gatherdi<mode>): Ditto.
1295         (*avx512f_gatherdi<mode>): Ditto.
1296         (*avx512f_gatherdi<mode>_2): Ditto.
1297         (avx512f_scattersi<mode>): Ditto.
1298         (*avx512f_scattersi<mode>): Ditto.
1299         (avx512f_scatterdi<mode>): Ditto.
1300         (*avx512f_scatterdi<mode>): Ditto.
1301         (sseintprefix): Extened with wider modes.
1302         (VEC_GATHER_IDXSI): Ditto.
1303         (VEC_GATHER_IDXDI): Ditto.
1304         (VEC_GATHER_SRCDI): Ditto.
1305
1306 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1307             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1308
1309         * config/arm/t-aprofile: New file.
1310         * config.gcc: Handle --with-multilib-list option.
1311
1312 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1313
1314         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1315         the input for r2.
1316
1317 2013-10-15  Richard Biener  <rguenther@suse.de>
1318
1319         * tree-loop-distribution.c (build_empty_rdg): Inline into
1320         single user.
1321         (rdg_flag_vertex): Inline into single user.
1322         (rdg_flag_vertex_and_dependent): Likewise.
1323         (build_rdg_partition_for_vertex): Remove processed bitmap.
1324         (rdg_build_partitions): Simplify.
1325
1326 2013-10-15  Richard Biener  <rguenther@suse.de>
1327
1328         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1329         Restructure forwarding through conversions and copies to
1330         avoid performing copy-propagation the wrong way.  Adjust
1331         recursion invocations.
1332         (forward_propagate_addr_expr): Add argument stating if we
1333         are recursing from a single-use.
1334         (ssa_forward_propagate_and_combine): Adjust.
1335
1336 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1337
1338         * dumpfile.h (gcc::dump_manager): New class, to hold state
1339         relating to dumpfile management.
1340         (get_dump_file_name): Remove in favor of method of dump_manager.
1341         (dump_initialized_p): Likewise.
1342         (dump_start): Likewise.
1343         (dump_finish): Likewise.
1344         (dump_switch_p): Likewise.
1345         (dump_register): Likewise.
1346         (get_dump_file_info): Likewise.
1347         * context.c (gcc::context::context): Construct the dump_manager
1348         instance.
1349         * context.h (gcc::context::get_dumps): New.
1350         (gcc::context::m_dumps): New.
1351         * coverage.c (coverage_init): Port to dump_manager API.
1352         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1353         (extra_dump_files_in_use): Likewise.
1354         (extra_dump_files_alloced): Likewise.
1355         (gcc::dump_manager::dump_manager): New.
1356         (dump_register): Convert to...
1357         (gcc::dump_manager::dump_register): ...method, replacing
1358         function-static next_dump with m_next_dump field.
1359         (get_dump_file_info): Convert to...
1360         (gcc::dump_manager::get_dump_file_info): ...method.
1361         (get_dump_file_name): Convert to...
1362         (gcc::dump_manager::get_dump_file_name): ...method.
1363         (dump_start): Convert to...
1364         (gcc::dump_manager::dump_start): ...method.
1365         (dump_finish): Convert to...
1366         (gcc::dump_manager::dump_finish): ...method.
1367         (dump_begin): Replace body with...
1368         (gcc::dump_manager::dump_begin): ...new method.
1369         (dump_phase_enabled_p): Convert to...
1370         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1371         (dump_phase_enabled_p): Convert to...
1372         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1373         (dump_initialized_p):  Convert to...
1374         (gcc::dump_manager::dump_initialized_p): ...method.
1375         (dump_flag_name): Replace body with...
1376         (gcc::dump_manager::dump_flag_name): ...new method.
1377         (dump_enable_all): Convert to...
1378         (gcc::dump_manager::dump_enable_all): ...new method.
1379         (opt_info_enable_passes): Convert to...
1380         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1381         (dump_switch_p_1): Convert to...
1382         (gcc::dump_manager::dump_switch_p_1): ...new method.
1383         (dump_switch_p):  Convert to...
1384         (gcc::dump_manager::dump_switch_p): ...new method.
1385         (opt_info_switch_p): Port to dump_manager API.
1386         (enable_rtl_dump_file): Likewise.
1387         * opts-global.c (handle_common_deferred_options): Port to new
1388         dump_manager API.
1389         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1390         (pass_manager::register_one_dump_file): Likewise.
1391         (pass_manager::register_pass): Likewise.
1392         (pass_init_dump_file): Likewise.
1393         (pass_fini_dump_file): Likewise.
1394         * statistics.c (statistics_early_init): Likewise.
1395
1396 2013-10-14  Richard Biener  <rguenther@suse.de>
1397
1398         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1399         iterative_hash_canonical_type, gimple_canonical_type_hash,
1400         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1401         gimple_register_canonical_type, print_gimple_types_stats,
1402         free_gimple_type_tables): Move to lto/lto.c
1403         (gt-gimple.h): Do not include.
1404         * gimple.h (gimple_register_canonical_type,
1405         print_gimple_types_stats, free_gimple_type_tables): Remove.
1406         * Makefile.in (GTFILES): Remove gimple.c.
1407
1408 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1409
1410         PR target/58716
1411         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1412         scanning for msp430x targets.
1413
1414 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1415
1416         PR bootstrap/58509
1417         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1418         (registers_ok_for_ldd_peep): Move around.
1419         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1420         * config/sparc/sparc.md (widening peepholes): Use it.
1421
1422 2013-10-14  Richard Biener  <rguenther@suse.de>
1423
1424         PR middle-end/58712
1425         PR middle-end/55358
1426         * gimple.c (iterative_hash_canonical_type): Make sure to
1427         record the hash into the correct hashtable slot.
1428
1429 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1430
1431         PR rtl-optimization/58662
1432         * combine.c (try_combine): Take into account death nodes on I2 when
1433         splitting a PARALLEL of two independent SETs.  Fix dump message.
1434
1435 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1436
1437         PR target/51244
1438         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1439         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1440         * config/sh/t-sh (sh_treg_combine.o): New entry.
1441         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1442         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1443         (register_sh_passes): New function.  Register sh_treg_combine pass.
1444         (sh_option_override): Invoke it.
1445         (sh_canonicalize_comparison): Handle op0_preserve_value.
1446         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1447         opportunities.  Canonicalize branch condition.
1448         (nott): Allow only if pseudos can be created for non-SH2A.
1449
1450 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1451
1452         PR target/58690
1453         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1454         (ix86_expand_movmem): Replace copy_addr_to_reg with
1455         ix86_copy_addr_to_reg.
1456         (ix86_expand_setmem): Likewise.
1457
1458 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1459
1460         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1461         provided by ix86_fp_compare_mode instead of CCFPUmode.
1462
1463 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1464
1465         * config/aarch64/arm_neon.h
1466         (vtbx<1,3>_<psu>8): Fix register constriants.
1467
1468 2013-10-11  Jeff Law  <law@redhat.com>
1469
1470         PR tree-optimization/58640
1471         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1472         threading paths that cross over two loop entry points.
1473
1474 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1475
1476         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1477         handle vector float as well.
1478         (*vsx_le_perm_load_v4si): Likewise.
1479         (*vsx_le_perm_store_v2di): Likewise.
1480         (*vsx_le_perm_store_v4si): Likewise.
1481
1482 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1483
1484         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1485         directly to circumvent subtract from splat{31} workaround.
1486         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1487         prototype.
1488         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1489         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1490         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1491         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1492         separate big and little endian logic.
1493         (*altivec_vperm_<mode>_internal): New define_insn.
1494         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1495         separate big and little endian logic.
1496         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1497         (vec_permv16qi): Add little endian logic.
1498
1499 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1500
1501         * doc/extend.texi (returns_nonnull): Remove arguments.
1502
1503 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1504             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1505             Sergey Lega  <sergey.s.lega@intel.com>
1506             Anna Tikhonova  <anna.tikhonova@intel.com>
1507             Ilya Tocar  <ilya.tocar@intel.com>
1508             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1509             Ilya Verbin  <ilya.verbin@intel.com>
1510             Kirill Yukhin  <kirill.yukhin@intel.com>
1511             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1512
1513         * config/i386/sse.md (VI48F_256_512): New.
1514         (avx2_permvar<mode>): Change to ...
1515         (<avx2_avx512f>_permvar<mode>): This.
1516
1517 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1518             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1519             Sergey Lega  <sergey.s.lega@intel.com>
1520             Anna Tikhonova  <anna.tikhonova@intel.com>
1521             Ilya Tocar  <ilya.tocar@intel.com>
1522             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1523             Ilya Verbin  <ilya.verbin@intel.com>
1524             Kirill Yukhin  <kirill.yukhin@intel.com>
1525             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1526
1527         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1528         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1529         * config/i386/sse.md (VI4_AVX): New.
1530         (sf2simodelower): Ditto.
1531         (sse2_cvtps2dq): Change to ...
1532         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1533
1534 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1535             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1536             Sergey Lega  <sergey.s.lega@intel.com>
1537             Anna Tikhonova  <anna.tikhonova@intel.com>
1538             Ilya Tocar  <ilya.tocar@intel.com>
1539             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1540             Ilya Verbin  <ilya.verbin@intel.com>
1541             Kirill Yukhin  <kirill.yukhin@intel.com>
1542             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1543
1544         * config/i386/sse.md (V_512): New.
1545         (VI_512): Ditto.
1546         (vcond<V_512:mode><VF_512:mode>): Ditto.
1547         (vcond<V_512:mode><VI_512:mode>): Ditto.
1548         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1549
1550 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1551             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1552             Sergey Lega  <sergey.s.lega@intel.com>
1553             Anna Tikhonova  <anna.tikhonova@intel.com>
1554             Ilya Tocar  <ilya.tocar@intel.com>
1555             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1556             Ilya Verbin  <ilya.verbin@intel.com>
1557             Kirill Yukhin  <kirill.yukhin@intel.com>
1558             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1559
1560         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1561         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1562         (FMAMODE): Ditto.
1563         (fma<mode>4): Removed condition.
1564         (fms<mode>4): Ditto.
1565         (fnma<mode>4): Ditto.
1566         (fnms<mode>4): Ditto.
1567         (fma4i_fmadd_<mode>): Ditto.
1568         (*fma_fmadd_<mode>): Ditto.
1569         (*fma_fmsub_<mode>): Ditto.
1570         (*fma_fnmadd_<mode>): Ditto.
1571         (*fma_fnmsub_<mode>): Ditto.
1572         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1573         (*fma_fmaddsub_<mode>): Ditto.
1574         (*fma_fmsubadd_<mode>): Ditto.
1575         (*fmai_fmadd_<mode>): Ditto.
1576         (*fmai_fmsub_<mode>): Ditto.
1577         (*fmai_fnmadd_<mode>): Ditto.
1578         (*fmai_fnmsub_<mode>): Ditto.
1579
1580 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1581             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1582             Sergey Lega  <sergey.s.lega@intel.com>
1583             Anna Tikhonova  <anna.tikhonova@intel.com>
1584             Ilya Tocar  <ilya.tocar@intel.com>
1585             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1586             Ilya Verbin  <ilya.verbin@intel.com>
1587             Kirill Yukhin  <kirill.yukhin@intel.com>
1588             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1589
1590         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1591         (VI124_256): Changed to ...
1592         (VI124_256_48_512): This.
1593         (ssepackmode): Extended with wider modes.
1594         (<code><mode>3): Changed iterator.
1595         (*avx2_<code><mode>3): Ditto.
1596         (vec_pack_trunc_<mode>): Ditto.
1597
1598 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1599             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1600             Sergey Lega  <sergey.s.lega@intel.com>
1601             Anna Tikhonova  <anna.tikhonova@intel.com>
1602             Ilya Tocar  <ilya.tocar@intel.com>
1603             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1604             Ilya Verbin  <ilya.verbin@intel.com>
1605             Kirill Yukhin  <kirill.yukhin@intel.com>
1606             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1607
1608         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1609         (VI8F_256_512): Ditto.
1610         (abs<mode>2): Changed iterator.
1611         (avx2_perm<mode>): Changed to ...
1612         (<avx2_avx512f>_perm<mode>): This.
1613         (avx2_perm<mode>_1): Changed to ...
1614         (<avx2_avx512f>_perm<mode>_1): This.
1615
1616 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1617             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1618             Sergey Lega  <sergey.s.lega@intel.com>
1619             Anna Tikhonova  <anna.tikhonova@intel.com>
1620             Ilya Tocar  <ilya.tocar@intel.com>
1621             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1622             Ilya Verbin  <ilya.verbin@intel.com>
1623             Kirill Yukhin  <kirill.yukhin@intel.com>
1624             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1625
1626         * config/i386/sse.md (VI48_AVX512F): New.
1627         (VI48_AVX2): Changed to ...
1628         (VI48_AVX2_48_AVX512F): This.
1629         (avx2_ashrv<mode>): Changed to ...
1630         (<avx2_avx512f>_ashrv<mode>): This.
1631         (avx2_<shift_insn>v<mode>): Changed to ...
1632         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1633
1634 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1635             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1636             Sergey Lega  <sergey.s.lega@intel.com>
1637             Anna Tikhonova  <anna.tikhonova@intel.com>
1638             Ilya Tocar  <ilya.tocar@intel.com>
1639             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1640             Ilya Verbin  <ilya.verbin@intel.com>
1641             Kirill Yukhin  <kirill.yukhin@intel.com>
1642             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1643
1644         * config/i386/sse.md (VI4_AVX512F): New.
1645         (VI8_AVX2_AVX512F): Ditto.
1646         (mul<mode>3): Extended with wider modes.
1647         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1648         (mul<mode>3): Ditto.
1649         (vec_widen_<s>mult_odd_<mode>): Ditto.
1650
1651 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1652             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1653             Sergey Lega  <sergey.s.lega@intel.com>
1654             Anna Tikhonova  <anna.tikhonova@intel.com>
1655             Ilya Tocar  <ilya.tocar@intel.com>
1656             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1657             Ilya Verbin  <ilya.verbin@intel.com>
1658             Kirill Yukhin  <kirill.yukhin@intel.com>
1659             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1660
1661         * config/i386/sse.md (VI2_AVX512F): New.
1662         (VI124_AVX512F): Ditto.
1663         (sseunpackmode): Extended with wider modes.
1664         (sseunpackfltmode): Ditto.
1665         (vec_unpacks_float_hi_<mode>): Ditto.
1666         (vec_unpacks_float_lo_<mode>): Ditto.
1667         (vec_unpacku_float_hi_<mode>): Ditto.
1668         (vec_unpacku_float_lo_<mode>): Ditto.
1669         (vec_unpacks_lo_<mode>): Ditto.
1670         (vec_unpacks_hi_<mode>): Ditto.
1671         (vec_unpacku_lo_<mode>): Ditto.
1672         (vec_unpacku_hi_<mode>): Ditto.
1673
1674 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1675             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1676             Sergey Lega  <sergey.s.lega@intel.com>
1677             Anna Tikhonova  <anna.tikhonova@intel.com>
1678             Ilya Tocar  <ilya.tocar@intel.com>
1679             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1680             Ilya Verbin  <ilya.verbin@intel.com>
1681             Kirill Yukhin  <kirill.yukhin@intel.com>
1682             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1683
1684         * config/i386/i386.md (multdiv): New.
1685         (multdiv_mnemonic): Ditto.
1686         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1687         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1688         (<sse>_vmdiv<mode>3): Removed.
1689
1690 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1691             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1692             Sergey Lega  <sergey.s.lega@intel.com>
1693             Anna Tikhonova  <anna.tikhonova@intel.com>
1694             Ilya Tocar  <ilya.tocar@intel.com>
1695             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1696             Ilya Verbin  <ilya.verbin@intel.com>
1697             Kirill Yukhin  <kirill.yukhin@intel.com>
1698             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1699
1700         * config/i386/sse.md (V): Extended with wider modes.
1701         (VF2): Ditto.
1702         (ssehalfvecmode): Ditto.
1703         (i128): Ditto.
1704         (ssepackfltmode): Ditto.
1705         (avx_vec_concat<mode>): Ditto.
1706         (V_256_512): New iterator.
1707         (VF2_512_256): Ditto.
1708         (si2dfmode): New attribute.
1709         (si2dfmodelower): Ditto.
1710         (sf2dfmode): Ditto.
1711         (concat_tg_mode): Ditto.
1712         (floatv4siv4df2): Changed to ...
1713         (float<si2dfmodelower><mode>2): This.
1714         (avx_cvtps2pd256): Changed to ...
1715         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1716         (vec_pack_trunc_v4df): Changed to ...
1717         (vec_pack_trunc_<mode>): This.
1718         (avx_vpermil<mode>): Changed to ...
1719         (<sse2_avx_avx512f>_vpermil<mode>): This.
1720         (<fixsuffix>fix_truncv8dfv8si2): New.
1721         (vec_pack_sfix_trunc_v8df): Ditto.
1722         (avx512f_rndscale<mode>): Ditto.
1723         (avx512f_roundpd512): Ditto.
1724         (vec_pack_ufix_trunc_<mode>): Updated iterator.
1725
1726 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1727             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1728             Sergey Lega  <sergey.s.lega@intel.com>
1729             Anna Tikhonova  <anna.tikhonova@intel.com>
1730             Ilya Tocar  <ilya.tocar@intel.com>
1731             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1732             Ilya Verbin  <ilya.verbin@intel.com>
1733             Kirill Yukhin  <kirill.yukhin@intel.com>
1734             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1735
1736         * config/i386/i386.md (any_fix): New iterator.
1737         (fixsuffix): New attribute.
1738         * config/i386/sse.md (VF1): Extened with wider modes.
1739         (VI): Ditto.
1740         (VI_AVX2): Ditto.
1741         (VI8): Ditto.
1742         (sseintvecmodelower): Ditto.
1743         (ssescalarmode): Ditto.
1744         (ssescalarnum): Ditto.
1745         (VF1_128_256): New.
1746         (ssexmmmode): Ditto.
1747         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
1748         (<sse>_rcp<mode>2): Change iterator.
1749         (rsqrt<mode>2): Ditto.
1750         (<sse>_rsqrt<mode>2): Ditto.
1751         (avx2_vec_dup<mode>): Ditto.
1752         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1753         (round<mode>2_sfix): Ditto.
1754         (avx2_pbroadcast<mode>): Ditto.
1755         (*andnot<mode>3): Handle XI mode.
1756         (*<code><mode>3): Ditto.
1757         (AVXTOSSEMODE): Removed.
1758         (avx_vpermil<mode>): Changed to ...
1759         (<sse2_avx_avx512f>_vpermil<mode>): This.
1760
1761 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1762             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1763             Sergey Lega  <sergey.s.lega@intel.com>
1764             Anna Tikhonova  <anna.tikhonova@intel.com>
1765             Ilya Tocar  <ilya.tocar@intel.com>
1766             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1767             Ilya Verbin  <ilya.verbin@intel.com>
1768             Kirill Yukhin  <kirill.yukhin@intel.com>
1769             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1770
1771         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
1772         (<sse>_comi): Ditto.
1773         (<sse>_ucomi): Ditto.
1774         (sse_cvtss2siq_2): Ditto.
1775         (sse2_cvtsd2si): Ditto.
1776         (sse2_cvtsd2siq): Ditto.
1777         (sse2_cvttsd2si): Ditto.
1778         (sse2_cvttsd2siq): Ditto.
1779         (<shift_insn><mode>3): Ditto.
1780         (sse2_cvtsi2sdq): Update constraint and prefix.
1781         (sse_cvtsi2ss): Update prefix.
1782         (sse_cvtsi2ssq): Ditto.
1783
1784 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
1785
1786         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
1787         ignore internal calls.
1788
1789 2013-10-11  Richard Biener  <rguenther@suse.de>
1790
1791         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
1792         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
1793         TREE_OVERFLOW is set.
1794
1795 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
1796
1797         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
1798
1799         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
1800         * gimple.h: Likewise.
1801
1802         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
1803
1804         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
1805         macros.
1806
1807         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
1808
1809         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
1810         * configure: Regenerate.
1811
1812 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
1813
1814         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
1815         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
1816         formatting fixes, use pp_colon instead of pp_character (..., ':'),
1817         similarly pp_right_paren.
1818         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
1819         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
1820         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
1821         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
1822         clauses.
1823         (omp_declare_simd_clauses_equal,
1824         omp_remove_redundant_declare_simd_attrs): New functions.
1825         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
1826         (walk_tree_1): Handle new OpenMP 4.0 clauses.
1827         * tree.h (OMP_LOOP_CHECK): Define.
1828         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
1829         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
1830         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
1831         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
1832         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
1833         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
1834         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
1835         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
1836         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
1837         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
1838         OMP_CLAUSE_SIMDLEN_EXPR): Define.
1839         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
1840         (omp_remove_redundant_declare_simd_attrs): New prototype.
1841         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
1842         GIMPLE_OMP_TEAMS): New codes.
1843         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
1844         * omp-low.c (struct omp_context): Add cancel_label and cancellable
1845         fields.
1846         (target_nesting_level): New variable.
1847         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
1848         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
1849         for collapse > 1 static schedule unless ordered.
1850         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
1851         (determine_parallel_type): Adjust get_ws_args_for caller.
1852         (install_var_field): Handle mask & 4 for double indirection.
1853         (scan_sharing_clauses): Ignore shared clause on teams construct.
1854         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
1855         (create_omp_child_function): If inside target or declare target
1856         constructs, set "omp declare target" attribute on the child function.
1857         (find_combined_for): New function.
1858         (scan_omp_parallel): Handle combined loops.
1859         (scan_omp_target, scan_omp_teams): New functions.
1860         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
1861         restrictions and set ctx->cancellable for cancellable constructs.
1862         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
1863         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
1864         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
1865         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
1866         (omp_clause_aligned_alignment): New function.
1867         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
1868         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
1869         on teams constructs.  Handle user defined reductions and new
1870         OpenMP 4.0 clauses.
1871         (lower_reduction_clauses): Don't set placeholder to address of ref
1872         if it has already the right type.
1873         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
1874         (expand_parallel_call): Use the new non-_start suffixed builtins,
1875         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
1876         and GOMP_parallel_end after the call.
1877         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
1878         (expand_omp_for_init_counts): Handle combined loops.
1879         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
1880         loops.
1881         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
1882         end of cancellable loops.
1883         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
1884         Likewise.  Handle collapse > 1 loops.
1885         (expand_omp_simd): Handle combined loops.
1886         (expand_omp_for): Add inner_stmt argument, adjust callers of
1887         expand_omp_for* functions, use expand_omp_for_static*chunk even
1888         for collapse > 1 unless ordered.
1889         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
1890         of cancellable sections.
1891         (expand_omp_single): Remove need_barrier variable, just rely on
1892         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
1893         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
1894         (expand_omp_atomic_load, expand_omp_atomic_store,
1895         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
1896         (expand_omp_target): New function.
1897         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
1898         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
1899         (build_omp_regions_1): Immediately close region for
1900         GF_OMP_TARGET_KIND_UPDATE.
1901         (maybe_add_implicit_barrier_cancel): New function.
1902         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
1903         cancellation.
1904         (lower_omp_single): Likewise.  Add clobber after the barrier.
1905         (lower_omp_taskgroup): New function.
1906         (lower_omp_for): Handle combined loops.  Adjust
1907         lower_rec_input_clauses caller.  Handle cancellation.
1908         (lower_depend_clauses): New function.
1909         (lower_omp_taskreg): Lower depend clauses.  Adjust
1910         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
1911         cancellation.
1912         (lower_omp_target, lower_omp_teams): New functions.
1913         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
1914         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
1915         and GOMP_cancellation_point calls.
1916         (lower_omp): Fold stmts inside of target region.
1917         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
1918         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
1919         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
1920         (BT_FN_VOID_OMPFN_PTR_UINT,
1921         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
1922         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
1923         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
1924         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
1925         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
1926         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
1927         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
1928         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
1929         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
1930         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
1931         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
1932         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
1933         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
1934         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
1935         BUILT_IN_GOMP_PARALLEL_END.
1936         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
1937         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
1938         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
1939         GF_OMP_FOR_KIND_DISTRIBUTE.
1940         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
1941         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
1942         (dump_gimple_omp_return): Print lhs if it has any.
1943         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
1944         gimple_omp_atomic_seq_cst_p.
1945         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
1946         and GIMPLE_OMP_TEAMS.
1947         * langhooks.c (lhd_omp_mappable_type): New function.
1948         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
1949         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
1950         hook.
1951         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
1952         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
1953         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
1954         (struct gimplify_omp_ctx): Add combined_loop field.
1955         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
1956         on stmts inside of target region.
1957         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
1958         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
1959         ORT_TARGET and ORT_TARGET_DATA.
1960         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
1961         Handle GOVD_MAP.
1962         (omp_notice_threadprivate_variable): Complain about threadprivate
1963         variables in target region.
1964         (omp_notice_variable): Complain about vars with non-mappable type
1965         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
1966         (omp_check_private): Ignore ORT_TARGET* regions.
1967         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
1968         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
1969         (find_combined_omp_for): New function.
1970         (gimplify_omp_for): Handle gimplification of combined loops.
1971         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
1972         OMP_TEAMS.
1973         (gimplify_omp_target_update): New function.
1974         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
1975         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
1976         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
1977         (gimplify_body): If fndecl has "omp declare target" attribute, add
1978         implicit ORT_TARGET context around it.
1979         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
1980         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
1981         * tree-nested.c (convert_nonlocal_reference_stmt,
1982         convert_local_reference_stmt, convert_gimple_call): Handle
1983         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
1984         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
1985         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
1986         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
1987         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
1988         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
1989         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
1990         BUILT_IN_GOMP_LOOP_END_CANCEL,
1991         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
1992         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
1993         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
1994         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
1995         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
1996         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
1997         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
1998         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
1999         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2000         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2001         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2002         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2003         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2004         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2005         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2006         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2007         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2008         gimple_build_omp_teams): New functions.
2009         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2010         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2011         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2012         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2013         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2014         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2015         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2016         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2017         GF_OMP_ATOMIC_SEQ_CST): New.
2018         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2019         gimple_build_omp_teams): New prototypes.
2020         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2021         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2022         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2023         GIMPLE_OMP_SINGLE as end of range.
2024         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2025         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2026         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2027         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2028         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2029         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2030         gimple_omp_target_kind, gimple_omp_target_set_kind,
2031         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2032         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2033         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2034         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2035         gimple_omp_teams_set_clauses): New inlines.
2036         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2037         and GIMPLE_OMP_TASKGROUP.
2038         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2039         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2040         enum omp_clause_proc_bind_kind): New.
2041         (union omp_clause_subcode): Add depend_kind, map_kind and
2042         proc_bind_kind fields.
2043         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2044         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2045         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2046         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2047         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2048
2049 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2050
2051         * predict.c (tree_estimate_probability): Add new parameter
2052         for estimate_bb_frequencies.
2053         (estimate_bb_frequencies): Add new parameter to force estimation.
2054         (rebuild_frequencies): When max frequency in function is small,
2055         recompute counts from frequencies.
2056         * predict.h (estimate_bb_frequencies): New parameter.
2057
2058 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2059
2060         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2061         optimizations are disabled.
2062
2063 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2064
2065         * coverage.c (coverage_finish): Fix leak of da_file_name.
2066
2067 2013-10-10  Jan Hubicka  <jh@suse.cz>
2068
2069         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2070         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2071         Bobcat and generic.
2072
2073 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2074
2075         PR middle-end/58670
2076         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2077         if any labels are in FALLTHRU_BB, use a special label emitted
2078         immediately after the asm goto insn rather than label_rtx
2079         of the LABEL_DECL.
2080         (expand_asm_stmt): Adjust caller.
2081         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2082         edge if the last insn in predecessor is a jump with single successor,
2083         but it isn't simplejump_p.
2084
2085 2013-10-10  Richard Biener  <rguenther@suse.de>
2086
2087         PR tree-optimization/58656
2088         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2089
2090 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2091
2092         * gimplify.c: Include expr.h and tm_p.h for targets with special
2093         va-arg padding requirements.
2094
2095 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2096
2097         * tree-flow.h: Move some prototypes to gimple.h.
2098         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2099         * gimple.h: Relocate some prototypes from tree-flow.h
2100         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2101         Move to gimplify.c.
2102         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2103         (build_va_arg_indirect_ref): Relocate and make static.
2104         (std_gimplify_va_arg_expr): Relocate here.
2105         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2106         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2107
2108 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2109
2110         * doc/md.texi: Document the mnemonic attribute.
2111
2112 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2113
2114         PR target/57377
2115         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2116         (set_attr_alternative x ...) when searching for user defined
2117         mnemonic attribute.
2118
2119 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2120
2121         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2122
2123 2013-10-09  Easwaran Raman <eraman@google.com>
2124
2125         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2126         * cfgexpand.c (defer_stack_allocation): ...use here
2127         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2128
2129 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2130
2131         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2132         (operand_equal_for_value_replacement): New function, extracted from
2133         value_replacement and enhanced to catch more cases.
2134         (value_replacement): Use operand_equal_for_value_replacement.
2135
2136 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2137
2138         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2139         get_max_loop_iterations.
2140
2141 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2142
2143         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2144         Place comment above function.
2145
2146 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2147
2148         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2149         are not related to tree-cfg.c.
2150         * tree-ssa-address.h: New file.  Relocate prototypes.
2151         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2152         (addr_for_mem_ref): New.  Combine call to get_address_description and
2153         return addr_for_mem_ref.
2154         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2155         * tree-ssa-live.h: Adjust prototypes.
2156         * passes.c: Include tree-ssa-live.h.
2157         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2158         * graphite.c (graphite_transform_loops): Make static.
2159         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2160         make_pass_graphite, pass_data_graphite_transforms,
2161         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2162         * ipa-pure-const.c (warn_function_noreturn): Make static.
2163         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2164         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2165         Relocate from tree-cfg.c
2166         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2167         static.
2168         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2169         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2170         Move to ipa-pure-const.c.
2171         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2172         Relocate from tree-optimize.c.
2173         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2174         make_pass_fixup_cfg): Move to tree-cfg.c.
2175         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2176         Relocate some prototypes.
2177         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2178         * tree-dump.c (dump_function_to_file): Remove prototype.
2179         Add tree-flow.h to the include file.
2180         * tree-dump.h: Remove prototype.
2181         * tree-parloops.h: New File.  Add prototypes.
2182         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2183         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2184         from tree-ssa-loop.c.
2185         * tree-predcom.c (run_tree_predictive_commoning,
2186         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2187         Relocate here from tree-ssa-loop.c.
2188         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2189         ssa_name_values.release ().
2190         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2191         (ssa_name_values): Relocate from tree-flow.h.
2192         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2193         * tree-ssa-loop.c (run_tree_predictive_commoning,
2194         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2195         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2196         make_pass_graphite, pass_data_graphite_transforms,
2197         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2198         tree_parallelize_loops, pass_data_parallelize_loops,
2199         make_pass_parallelize_loops): Move to other files.
2200         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2201         moved here.
2202         * tree.h: Remove prototypes from tree-address.c.
2203
2204 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2205
2206         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2207         (struct int_tree_map): Move to tree-hasher.h
2208         (SCALE, LABEL, PERCENT): Move to gimple.h
2209         * tree-flow-inline.h: Delete.  Move functions to other files.
2210         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2211         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2212         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2213         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2214         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2215         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2216         * tree-hasher.h: Don't include tree-flow.h.
2217         (struct int_tree_map): Relocate from tree-flow.h.
2218         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2219         tree-flow-inline.h and make static.
2220         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2221         tree-flow-inline.h.
2222         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2223         tree-flow-inline.h and make static.
2224         * tree.h (is_global_var, may_be_aliased): Relocate from
2225         tree-flow-inline.h.
2226         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2227         * value-prof.c: No longer include tree-flow-inline.h.
2228         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2229
2230 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2231
2232         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2233         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2234         (enum move_pos): Move to tree-ssa-loop-im.h
2235         * cfgloop.h: Move some prototypes.
2236         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2237         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2238         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2239         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2240         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2241         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2242         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2243         (enum move_pos): Relocate here.
2244         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2245         tree-ssa-loop.c.
2246         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2247         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2248         make_pass_lim): Relocate here from tree-ssa-loop.c.
2249         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2250         tree-ssa-loop.c.
2251         (loop_edge_to_cancel, unloop_loops): Make static.
2252         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2253         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2254         (tree_complete_unroll, gate_tree_complete_unroll,
2255         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2256         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2257         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2258         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2259         (stmt_invariant_in_loop_p): Remove unused function.
2260         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2261         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2262         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2263         (gcov_type_to_double_int): Move to cfgloop.h.
2264         (double_int_cmp, bound_index,
2265         estimate_numbers_of_iterations_loop): Make static.
2266         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2267         (max_loop_iterations): Factor out get_max_loop_iterations.
2268         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2269         cfgloop.c.
2270         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2271         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2272         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2273         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2274         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2275         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2276         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2277         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2278         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2279         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2280         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2281         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2282         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2283         pass_data_complete_unroll, make_pass_complete_unroll,
2284         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2285         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2286         tree-ssa-loop-ivcanon.c.
2287         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2288         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2289         tree-ssa-loop-prefetch.c.
2290         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2291         tree-ssa-loop-im.c.
2292         (get_lsm_tmp_name): Relocate and add suffix parameter.
2293         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2294         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2295         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2296         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2297         (get_estimated_loop_iterations): Factor out accessor from
2298         estimated_loop_iterations in tree-ssa-loop-niter.c.
2299         (get_max_loop_iterations): Factor out accessor from
2300         _max_loop_iterations in tree-ssa-niter.c.
2301         * loop-unroll.c (decide_unroll_constant_iterations,
2302         decide_unroll_runtime_iterations, decide_peel_simple,
2303         decide_unroll_stupid): Use new get_* accessors.
2304
2305 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2306
2307         PR tree-optimization/20318
2308         * doc/extend.texi (returns_nonnull): New function attribute.
2309         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2310         attribute.
2311         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2312         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2313
2314 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2315
2316         PR middle-end/58570
2317         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2318         false if both components are bitfields.
2319
2320 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2321
2322         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2323         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2324         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2325         * config/aarch64/aarch64.h
2326         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2327         * config/aarch64/aarch64-simd-builtins.def
2328         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2329
2330 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2331
2332         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2333
2334 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2335
2336         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2337
2338 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2339
2340         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2341         (vneg_f64): New intrinsic.
2342         (vneg_s8): Asm replaced with C.
2343         (vneg_s16): Likewise.
2344         (vneg_s32): Likewise.
2345         (vneg_s64): New intrinsic.
2346         (vnegq_f32): Asm replaced with C.
2347         (vnegq_f64): Likewise.
2348         (vnegq_s8): Likewise.
2349         (vnegq_s16): Likewise.
2350         (vnegq_s32): Likewise.
2351         (vnegq_s64): Likewise.
2352
2353 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2354
2355         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2356
2357 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2358
2359         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2360         packed stack special handling.
2361         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2362         back to fixed stack slots for FPRs saved due to stdarg.
2363
2364 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2365
2366         * config/s390/s390.c (s390_frame_info): Restructure function.
2367
2368 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2369
2370         * config/s390/s390.c (struct s390_frame_layout): New field
2371         gpr_save_slots.
2372         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2373         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2374         regs_ever_clobbered to char*.
2375         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2376         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2377         (s390_register_info): Enable FPR save slots.  Move/Copy some
2378         functionality into ...
2379         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2380         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2381         function.
2382         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2383         not imply a stack frame.
2384         (s390_init_frame_layout): Remove variable clobbered_regs.
2385         (s390_update_register_info): Remove function.
2386         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2387         cannot be used for register renaming.
2388         (s390_hard_regno_scratch_ok): New function.
2389         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2390         (s390_initial_elimination_offset): Change offset calculation of
2391         the return address pointer.
2392         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2393         (restore_gprs): Set frame related flag.
2394         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2395         (s390_emit_prologue): Call s390_register_info instead of
2396         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2397         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2398         (s390_optimize_prologue): Call s390_optimize_register_info.
2399         Try to remove also FPR slot save/restore INSNs.  Remove frame
2400         related flags from restore INSNs.
2401
2402 2013-10-08  DJ Delorie  <dj@redhat.com>
2403
2404         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2405         (movhi): Likewise.
2406
2407         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2408         avoid conflict with the MI use of %c.
2409         * config/rl78/rl78-real.md: change %c to %C throughout.
2410         * config/rl78/rl78-virt.md: Likewise.
2411
2412 2013-10-08  Jan Hubicka  <jh@suse.cz>
2413
2414         * config/i386/i386.c (ix86_option_override_internal): Switch
2415         to SSE math for -ffast-math when target ISA supports SSE2.
2416
2417 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2418
2419         * tree-flow.h: Remove some prototypes.
2420         * tree.h: Remove some protypes, add a couple.
2421         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2422         using_eh_for_cleanups_p): Add interface routines for front ends.
2423         * tree-eh.h: New file.  Add protoptyes.
2424         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2425         (add_stmt_to_eh_lp_fn): Make static.
2426         (lower_try_finally): Use new using_eh_for_cleanups_p.
2427         * emit-rtl.c: Include tree-eh.h.
2428         * gimple.h: Include tree-eh.h.
2429
2430 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2431
2432         PR tree-optimization/58480
2433         * tree-vrp.c (infer_nonnull_range): New function.
2434         (infer_value_range): Call infer_nonnull_range.
2435
2436 2013-10-08  Dehao Chen  <dehao@google.com>
2437
2438         PR tree-optimization/58619
2439         * tree-inline.c (copy_phis_for_bb): Combine location data
2440         only if non-null.
2441
2442 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2443
2444         PR target/58423
2445         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2446         RTX_FRAME_RELATED_P on INSN.
2447
2448 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2449
2450         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2451         (altivec_expand_vec_perm_const): Call it.
2452
2453 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2454
2455         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2456         sequences for LE VSX loads and stores at expand time.
2457         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2458         prototype.
2459         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2460         (rs6000_gen_le_vsx_permute): New.
2461         (rs6000_gen_le_vsx_load): New.
2462         (rs6000_gen_le_vsx_store): New.
2463         (rs6000_gen_le_vsx_move): New.
2464         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2465         (*vsx_le_perm_load_v4si): New.
2466         (*vsx_le_perm_load_v8hi): New.
2467         (*vsx_le_perm_load_v16qi): New.
2468         (*vsx_le_perm_store_v2di): New.
2469         (*vsx_le_perm_store_v4si): New.
2470         (*vsx_le_perm_store_v8hi): New.
2471         (*vsx_le_perm_store_v16qi): New.
2472         (*vsx_xxpermdi2_le_<mode>): New.
2473         (*vsx_xxpermdi4_le_<mode>): New.
2474         (*vsx_xxpermdi8_le_V8HI): New.
2475         (*vsx_xxpermdi16_le_V16QI): New.
2476         (*vsx_lxvd2x2_le_<mode>): New.
2477         (*vsx_lxvd2x4_le_<mode>): New.
2478         (*vsx_lxvd2x8_le_V8HI): New.
2479         (*vsx_lxvd2x16_le_V16QI): New.
2480         (*vsx_stxvd2x2_le_<mode>): New.
2481         (*vsx_stxvd2x4_le_<mode>): New.
2482         (*vsx_stxvd2x8_le_V8HI): New.
2483         (*vsx_stxvd2x16_le_V16QI): New.
2484
2485 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2486
2487         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2488
2489 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2490
2491         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2492         loop to work also for 31bit ABI.
2493         Save the stack pointer for frame_size > 0.
2494
2495 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2496
2497         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2498         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2499         constraint letters from expanders.
2500         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2501         retry count to general_operand.
2502         ("tabort"): Give operand 0 a mode.
2503         ("tabort_1"): Add mode and constraint letter for operand 0.
2504         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2505
2506 2013-10-04  Jeff Law  <law@redhat.com>
2507
2508         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2509
2510         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2511         out of ...
2512         (thread_across_edge): Here.  Call it.
2513
2514 2013-10-04  Cary Coutant  <ccoutant@google.com>
2515
2516         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2517         discarding a location list expression (or a piece of one).
2518
2519 2013-10-03  Jan Hubicka  <jh@suse.cz>
2520
2521         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2522         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2523         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2524
2525 2013-10-03  Jan Hubicka  <jh@suse.cz>
2526
2527         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2528         accumulate-outgoing-args when producing unwind info.
2529
2530 2013-10-03  Wei Mi  <wmi@google.com>
2531
2532         * lra-constraints.c (insert_move_for_subreg): New function
2533         extracted from simplify_operand_subreg.
2534         (simplify_operand_subreg): Add reload for paradoxical subreg.
2535
2536 2013-10-03  Rong Xu  <xur@google.com>
2537
2538         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2539         the candidate of builtin_expect such that we should fix the
2540         size/time estimation.
2541         (estimate_function_body_sizes): Do the acutally size/time fix-up
2542         for builtin_expect.
2543
2544 2013-10-03  Rong Xu  <xur@google.com>
2545
2546         * predict.c (tree_predict_by_opcode): Get the probability
2547         for builtin_expect from param builtin_expect_probability.
2548         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2549         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2550         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2551
2552 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2553
2554         PR c++/19476
2555         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2556         * calls.c (alloca_call_p): Use get_callee_fndecl.
2557         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2558         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2559         Likewise.
2560         (vrp_visit_stmt): Remove duplicated code.
2561
2562 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2563
2564         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2565         ceildf2, btruncdf2, instead of vsx_* name.
2566
2567         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2568         iterators to only do V2DF and V4SF here.  Move the DF code to
2569         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2570         just 'v' since only vector operations are handled with these insns
2571         after moving the DF support to rs6000.md.
2572         (vsx_sub<mode>3): Likewise.
2573         (vsx_mul<mode>3): Likewise.
2574         (vsx_div<mode>3): Likewise.
2575         (vsx_fre<mode>2): Likewise.
2576         (vsx_neg<mode>2): Likewise.
2577         (vsx_abs<mode>2): Likewise.
2578         (vsx_nabs<mode>2): Likewise.
2579         (vsx_smax<mode>3): Likewise.
2580         (vsx_smin<mode>3): Likewise.
2581         (vsx_sqrt<mode>2): Likewise.
2582         (vsx_rsqrte<mode>2): Likewise.
2583         (vsx_fms<mode>4): Likewise.
2584         (vsx_nfma<mode>4): Likewise.
2585         (vsx_copysign<mode>3): Likewise.
2586         (vsx_btrunc<mode>2): Likewise.
2587         (vsx_floor<mode>2): Likewise.
2588         (vsx_ceil<mode>2): Likewise.
2589         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2590         (vsx_sminsf3): Likewise.
2591         (vsx_fmadf4): Likewise.
2592         (vsx_fmsdf4): Likewise.
2593         (vsx_nfmadf4): Likewise.
2594         (vsx_nfmsdf4): Likewise.
2595         (vsx_cmpdf_internal1): Likewise.
2596
2597         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2598         simpler to select whether a target has SPE or traditional floating
2599         point support in iterators.
2600         (TARGET_DF_SPE): Likewise.
2601         (TARGET_SF_FPR): Likewise.
2602         (TARGET_DF_FPR): Likewise.
2603         (TARGET_SF_INSN): Macros to say whether floating point support
2604         exists for a given operation for expanders.
2605         (TARGET_DF_INSN): Likewise.
2606
2607         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2608         combining of SF/DF mode operations, using both traditional and VSX
2609         registers.
2610         (Fvsx): Likewise.
2611         (Ff): Likewise.
2612         (Fv): Likewise.
2613         (Fs): Likewise.
2614         (Ffre): Likewise.
2615         (FFRE): Likewise.
2616         (abs<mode>2): Combine SF/DF modes using traditional floating point
2617         instructions.  Add support for using the upper DF registers with
2618         VSX support, and SF registers with power8-vector support.  Update
2619         expanders for operations supported by both the SPE and traditional
2620         floating point units.
2621         (abs<mode>2_fpr): Likewise.
2622         (nabs<mode>2): Likewise.
2623         (nabs<mode>2_fpr): Likewise.
2624         (neg<mode>2): Likewise.
2625         (neg<mode>2_fpr): Likewise.
2626         (add<mode>3): Likewise.
2627         (add<mode>3_fpr): Likewise.
2628         (sub<mode>3): Likewise.
2629         (sub<mode>3_fpr): Likewise.
2630         (mul<mode>3): Likewise.
2631         (mul<mode>3_fpr): Likewise.
2632         (div<mode>3): Likewise.
2633         (div<mode>3_fpr): Likewise.
2634         (sqrt<mode>3): Likewise.
2635         (sqrt<mode>3_fpr): Likewise.
2636         (fre<Fs>): Likewise.
2637         (rsqrt<mode>2): Likewise.
2638         (cmp<mode>_fpr): Likewise.
2639         (smax<mode>3): Likewise.
2640         (smin<mode>3): Likewise.
2641         (smax<mode>3_vsx): Likewise.
2642         (smin<mode>3_vsx): Likewise.
2643         (negsf2): Delete SF operations that are merged with DF.
2644         (abssf2): Likewise.
2645         (addsf3): Likewise.
2646         (subsf3): Likewise.
2647         (mulsf3): Likewise.
2648         (divsf3): Likewise.
2649         (fres): Likewise.
2650         (fmasf4_fpr): Likewise.
2651         (fmssf4_fpr): Likewise.
2652         (nfmasf4_fpr): Likewise.
2653         (nfmssf4_fpr): Likewise.
2654         (sqrtsf2): Likewise.
2655         (rsqrtsf_internal1): Likewise.
2656         (smaxsf3): Likewise.
2657         (sminsf3): Likewise.
2658         (cmpsf_internal1): Likewise.
2659         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2660         (negdf2): Delete DF operations that are merged with SF.
2661         (absdf2): Likewise.
2662         (nabsdf2): Likewise.
2663         (adddf3): Likewise.
2664         (subdf3): Likewise.
2665         (muldf3): Likewise.
2666         (divdf3): Likewise.
2667         (fred): Likewise.
2668         (rsqrtdf_internal1): Likewise.
2669         (fmadf4_fpr): Likewise.
2670         (fmsdf4_fpr): Likewise.
2671         (nfmadf4_fpr): Likewise.
2672         (nfmsdf4_fpr): Likewise.
2673         (sqrtdf2): Likewise.
2674         (smaxdf3): Likewise.
2675         (smindf3): Likewise.
2676         (cmpdf_internal1): Likewise.
2677         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2678         (btrunc<mode>2): Delete separate expander, and combine with the
2679         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2680         (btrunc<mode>2_fpr): Likewise.
2681         (ceil<mode>2): Likewise.
2682         (ceil<mode>2_fpr): Likewise.
2683         (floor<mode>2): Likewise.
2684         (floor<mode>2_fpr): Likewise.
2685         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2686         Add support for using the upper registers with VSX and
2687         power8-vector.  Move insns to be closer to the define_expands. On
2688         VSX systems, prefer the traditional form of FMA over the VSX
2689         version, since the traditional form allows the target not to
2690         overlap with the inputs.
2691         (fms<mode>4_fpr): Likewise.
2692         (nfma<mode>4_fpr): Likewise.
2693         (nfms<mode>4_fpr): Likewise.
2694
2695 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2696             Richard Earnshaw  <richard.earnshaw@arm.com>
2697
2698         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2699         (struct mult_cost_table): Likewise.
2700         (struct mem_cost_table): Likewise.
2701         (struct fp_cost_table): Likewise.
2702         (struct vector_cost_table): Likewise.
2703         (cpu_cost_table): Likewise.
2704         * config/arm/arm.opt (mold-rts-costs): New option.
2705         (mnew-generic-costs): Likewise.
2706         * config/arm/arm.c (generic_extra_costs): New table.
2707         (cortexa15_extra_costs): Likewise.
2708         (arm_slowmul_tune): Use NULL as new costs.
2709         (arm_fastmul_tune): Likewise.
2710         (arm_strongarm_tune): Likewise.
2711         (arm_xscale_tune): Likewise.
2712         (arm_9e_tune): Likewise.
2713         (arm_v6t2_tune): Likewise.
2714         (arm_cortex_a5_tune): Likewise.
2715         (arm_cortex_a9_tune): Likewise.
2716         (arm_v6m_tune): Likewise.
2717         (arm_fa726te_tune): Likewise.
2718         (arm_cortex_a15_tune): Use cortex15_extra_costs.
2719         (arm_cortex_tune): Use generict_extra_costs.
2720         (shifter_op_p): New function.
2721         (arm_unspec_cost): Likewise.
2722         (LIBCALL_COST): Define.
2723         (arm_new_rtx_costs): New function.
2724         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
2725         table is available. Use old costs otherwise unless mnew-generic-costs
2726         is specified.
2727         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
2728         (cpu_cost_table): Declare.
2729
2730 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2731
2732         PR target/58460
2733         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
2734         (*subs_mul_imm_<mode>)
2735         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
2736         (*sub_<shift>_<mode>)
2737         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
2738         Remove k constraint.
2739
2740 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
2741
2742         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
2743         code.
2744         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
2745
2746 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2747
2748         * predict.c (probably_never_executed): New function.
2749         (probably_never_executed_bb_p): Invoke probably_never_executed.
2750         (probably_never_executed_edge_p): Ditto.
2751         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
2752         Treat profile insanities conservatively.
2753
2754 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
2755
2756         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
2757
2758 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
2759
2760         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
2761         use it.  Use smaller increase for scratch.  Don't increase reject
2762         for early clobber scratch.
2763         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
2764         setting eliminated regs except setting fp from hfp.
2765         (lra_eliminate): Check lra_insn_recog_data on NULL.
2766
2767 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2768
2769         PR target/58587
2770         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
2771         setting -mvsx-timode by default until the underlying problem is fixed.
2772         (RS6000_CPU, power7 defaults): Likewise.
2773
2774 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2775
2776         * config/x-linux (host-linux.o): Remove header dependencies.
2777         Use $(COMPILE) and $(POSTCOMPILE).
2778         * config/t-linux-android (linux-android.o): Ditto.
2779
2780 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2781
2782         * Makefile.in (expmed.o-warn): Remove.
2783
2784 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2785
2786         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
2787         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
2788
2789 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2790
2791         * dojump.c (do_jump_1): Divide probability between
2792         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
2793
2794 2013-10-02  Tom Tromey  <tromey@redhat.com>
2795
2796         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
2797
2798 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2799
2800         * tree-flow.h: Remove some prototypes.
2801         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
2802         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
2803         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
2804         mark_virtual_phi_result_for_renaming): Relocate here.
2805         * tree-into-ssa.h: Add prototypes.
2806         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
2807         single_pred_before_succ_order.
2808         (blocks_in_phiopt_order): Rename and move to cfganal.c.
2809         (nonfreeing_call_p) Move to gimple.c.
2810         * cfganal.c (single_pred_before_succ_order): Move and renamed from
2811         tree-ssa-phiopt.c.
2812         * basic-block.h (single_pred_before_succ_order): Add prototype.
2813         * gimple.c (nonfreeing_call_p): Relocate here.
2814         * gimple.h: Add prototype.
2815         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
2816         * tree-ssa-dom.h: New file.  Relocate prototypes here.
2817         * tree-ssa.h: Include tree-ssa-dom.h.
2818
2819 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2820
2821         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
2822         Use $(COMPILE) and $(POSTCOMPILE).
2823
2824         * config/alpha/x-alpha (driver-alpha.o): Ditto.
2825
2826 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2827
2828         * tree-flow.h: Remove some prototypes.
2829         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
2830         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
2831         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
2832         propagate_tree_value*): Move from here to...
2833         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
2834         propagate_tree_value*): Relocate here.
2835         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
2836         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
2837         * gimple-fold.c: Remove gimple-fold.h from include list.
2838         * tree-vrp.c: Remove gimple-fold.h from include list.
2839         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
2840         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
2841         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
2842         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
2843         * sese.c: Add tree-ssa-propagate.h to include list.
2844
2845 2013-10-02  Richard Biener  <rguenther@suse.de>
2846
2847         * tree-loop-distribution.c: Include tree-vectorizer.h for
2848         find_loop_location.
2849         (enum partition_kind): Remove PKIND_REDUCTION.
2850         (struct partition_s): Remove has_writes member, add reduction_p member.
2851         (partition_alloc): Adjust.
2852         (partition_builtin_p): Likewise.
2853         (partition_has_writes): Remove.
2854         (partition_reduction_p): New function.
2855         (partition_merge_into): Likewise.
2856         (generate_code_for_partition): Commonize builtin partition
2857         handling tail.
2858         (rdg_cannot_recompute_vertex_p): Remove.
2859         (already_processed_vertex_p): Likewise.
2860         (rdg_flag_vertex): Do not set has_writes.
2861         (classify_partition): Adjust.
2862         (rdg_build_partitions): Do not set has_writes, treat all
2863         partitions as useful.
2864         (distribute_loop): Record number of library calls generated.  Adjust.
2865         (tree_loop_distribution): Report number of loops and library
2866         calls generated as opt-info.
2867
2868 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2869
2870         * tree-flow.h: Include new .h files.  Move prototypes.
2871         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
2872         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
2873         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
2874         * tree-pretty-print.h: Add prototypes from tree-flow.h.
2875         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
2876         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
2877         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
2878         * tree.h (get_ref_base_and_extent): Move prototype out.
2879         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
2880         tree-dfa.h.
2881         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
2882         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
2883         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
2884         * tree-scalar-evolution.c: Include tree.h.
2885         * sese.c: Include tree.h.
2886         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
2887         * dwarf2out.c: Include tree-dfa.h.
2888         * tree-chrec.c: Include tree.h.
2889         * tree-data-ref.c: Include tree.h.
2890
2891 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
2892
2893         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
2894         Fix whitespace.
2895
2896 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2897
2898         * config/t-sol2 (sol2-c.o): Remove header dependencies.
2899         Use $(COMPILE) and $(POSTCOMPILE).
2900         (sol2-cxx.o): Likewise.
2901         (sol2-stubs.o): Likewise.
2902         (sol2.o): Likewise.
2903         * config/x-solaris (host-solaris.o): Likewise.
2904
2905         * config/sparc/t-sparc (sparc.o): Remove.
2906         (sparc-c.o): Remove header dependencies.
2907         Use $(COMPILE) and $(POSTCOMPILE).
2908         * config/sparc/x-sparc: Likewise.
2909
2910 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
2911
2912         * config/arc/arc-opts.h: Add 2013 to Copyright years.
2913         * config/arc/arc700.md: Likewise.
2914         * config/arc/arc-modes.def: Likewise.
2915         * config/arc/arc-simd.h: Likewise.
2916         * config/arc/t-arc-uClibc: Likewise.
2917         * config/arc/t-arc-newlib: Likewise.
2918
2919 2013-10-02  Renlin Li  <renlin.li@arm.com>
2920
2921         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2922         plus_constant.
2923         (aarch64_expand_epilogue): Likewise.
2924
2925 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2926             Yufeng Zhang  <yufeng.zhang@arm.com>
2927
2928         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
2929         declaration.
2930         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
2931         'base_in' represent a conversion and legal_cast_p_1 holds; set
2932         'base_in' with the returned value from get_unwidened.
2933
2934 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2935
2936         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
2937         plus_constant is not used.
2938
2939 2013-10-01  Wei Mi  <wmi@google.com>
2940
2941         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
2942         * config/i386/i386.md: Add define_peephole2 to
2943         break partial reg stall for cvtss2sd/cvtsd2ss.
2944
2945 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2946
2947         * config/arc/arc.c (pass_arc_ifcvt::clone):
2948         Update for ctxt_ -> m_ctxt change.
2949
2950 2013-10-01  Jeff Law  <law@redhat.com>
2951
2952         * tree-ssa-threadupdate.c (struct redirection_data): Delete
2953         outgoing_edge and intermediate_edge fields.  Instead store the path.
2954         (redirection_data::hash): Hash on the last edge's destination index.
2955         (redirection_data::equal): Check the entire thread path.
2956         (lookup_redirectio_data): Corresponding changes.
2957         (create_edge_and_update_destination_phis): Likewise.
2958         (thread_single_edge): Likewise.
2959
2960 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2961             Diego Novillo <dnovillo@google.com>
2962
2963         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
2964         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
2965         (vld32wh_insn, vld32wl_insn): Delete commented-out old
2966         versions of these patterns.
2967
2968         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
2969         (__builtin_arc_aligned): Likewise.
2970
2971         * config/arc/arc.md: Expand adc_0 comment stating the intended
2972         purpose and why it isn't ready.
2973         Replace commented out call_value_via_label_mixed with a
2974         plain comment about bl_s.
2975
2976         * config/arc/arc.c (stdio.h): Don't include directly.
2977         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
2978         Remove if (1) condition.
2979         (arc_encode_section_info): Fix comment.
2980
2981 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2982
2983         * config/arc/arc.c (arc_conditional_register_usage):
2984         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
2985         Also set reg_alloc_order for DMA config regs.
2986
2987 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2988             Jeremy Bennett  <jeremy.bennett@embecosm.com>
2989
2990         * doc/install.texi (--with-cpu): Mention ARC.
2991         (arc-*-elf32): New paragraph.
2992         (arc-linux-uclibc): Likewise.
2993         * doc/md.texi (Machine Constraints): Add ARC part.
2994         * doc/invoke.texi: (menu): Add ARC Options.
2995         (Machine Dependent Options) <ARC Options>: Add synopsis.
2996         (node ARC Options): Add.
2997         * doc/extend.texi (long_call / short_call attribute): Add ARC.
2998         (ARC Built-in Functions): New section defining
2999         generic ARC built-in functions.
3000         (ARC SIMD Built-in Functions): New section defining SIMD specific
3001         built-in functions.
3002         (Declaring Attributes of Functions): Extended
3003         description of short_call and long_call attributes for ARC and
3004         added index entries.
3005
3006 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3007             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3008             Joern Rennecke  <joern.rennecke@embecosm.com>
3009             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3010             Brendan Kehoe  <brendan@zen.org>
3011             Michael Eager  <eager@eagercon.com>
3012             Simon Cook  <simon.cook@embecosm.com>
3013             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3014
3015         * config/arc, common/config/arc: New directories.
3016
3017 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3018             Brendan Kehoe  <brendan@zen.org>
3019             Simon Cook  <simon.cook@embecosm.com>
3020
3021         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3022
3023 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3024
3025         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3026         * tree-ssa-coalesce.h: New. Move prototype to here.
3027         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3028         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3029         (gimple_can_coalesce_p): Move to...
3030         * gimple.c (gimple_can_coalesce_p): Here.
3031
3032 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3033
3034         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3035         (dump_decl_set): Move to gimple.c.
3036         * gimple.h: Don't include tree-ssa-operands.h.
3037         (dump_decl_set): Add prototype.
3038         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3039         Move to gimple-ssa.h.
3040         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3041         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3042         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3043         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3044         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3045         than PHI_ARG_DEF.
3046         (dump_decl_set): Relocate here.
3047         * gimple-ssa.h: New file.
3048         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3049         Relocate from gimple.h.
3050         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3051         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3052         swap_tree_operands and remove non-ssa path.
3053         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3054         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3055         swap_ssa_operands.
3056         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3057         vect_is_simple_reduction_1): Use swap_ssa_operands.
3058         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3059         (enum need_phi_state): Move to tree-into-ssa.c.
3060         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3061         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3062         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3063         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3064         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3065         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3066         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3067         num_imm_uses): Move to ssa-iterators.h.
3068         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3069         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3070         tree-phinodes.h.
3071         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3072         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3073         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3074         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3075         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3076         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3077         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3078         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3079         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3080         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3081         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3082         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3083         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3084         (set_phi_nodes): Move to tree-phinodes.h.
3085         * tree-ssa-operands.h (enum ssa_op_iter_type,
3086         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3087         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3088         (dump_decl_set): Remove prototype.
3089         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3090         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3091         (set_phi_nodes): Relocate from tree-flow-inline.h.
3092         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3093         tree-flow-inline.h
3094         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3095         include list.  Temporarily add gimple.h to include list.
3096         * ssa-iterators.h: New file.
3097         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3098         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3099         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3100         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3101         Relocate from tree-ssa-operands.h.
3102         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3103         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3104         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3105         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3106         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3107         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3108         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3109         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3110         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3111         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3112         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3113         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3114         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3115         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3116         Relocate from tree-flow-inline.h.
3117         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3118
3119 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3120
3121         * aarch64-simd.md
3122         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3123         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3124         Insert '\t' to output template.
3125         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3126         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3127         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3128         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3129
3130 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3131
3132         * doc/install.texi (Host/target specific installation notes for GCC):
3133         Put @anchor before @heading.
3134         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3135         Use @email for email addresses.
3136
3137 2013-10-01  Jeff Law  <law@redhat.com>
3138
3139         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3140         a vec.  Only delete the path if we create one without successfully
3141         registering a jump thread.
3142         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3143         as a pointer.
3144         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3145         (paths): New vector of jump threading paths.
3146         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3147         (THREAD_PATH): New accessor macro for the entire thread path.
3148         (lookup_redirection_data): Get intermediate and final outgoing edge
3149         from the thread path.
3150         (create_edge_and_update_destination_phis): Copy the threading path.
3151         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3152         jump threading path.
3153         (ssa_redirect_edges): Get edges and block types from the jump threading
3154         path.  Free the path vector.
3155         (thread_block): Get edges from the jump threading path.  Look at the
3156         entire path to see if we thread to a loop exit.  If we cancel a jump
3157         thread request, then free the path vector.
3158         (thread_single_edge): Get edges and block types from the jump threading
3159         path.  Free the path vector.
3160         (thread_through_loop_header): Get edges and block types from the jump
3161         threading path.  Free the path vector.
3162         (mark_threaded_blocks): Iterate over the vector of paths and store
3163         the path on the appropriate edge.  Get edges and block types from the
3164         jump threading path.
3165         (mark_threaded_blocks): Get edges and block types from the jump
3166         threading path.  Free the path vector.
3167         (thread_through_all_blocks): Use the vector of paths rather than
3168         a vector of 3-edge sets.
3169         (register_jump_thread): Accept pointer to a path vector rather
3170         than the path vector itself.  Store the path vector for later use.
3171         Simplify.
3172
3173 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3174             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3175
3176         PR target/58574
3177         * config/s390/s390.c (s390_split_branches): Modify check for table
3178         jump insns.
3179         (s390_chunkify_start): Rearrange table jump insn check in order to
3180         deal with compare and branch insns correctly.
3181
3182 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3183
3184         PR target/58578
3185         Revert
3186         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3187         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3188         define_insn_and_split.
3189         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3190         (shiftsi3_compare): New pattern.
3191         (rrx): New pattern.
3192         * config/arm/unspecs.md (UNSPEC_RRX): New.
3193
3194 2013-10-01  Alan Modra  <amodra@gmail.com>
3195
3196         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3197         casing inout operands.
3198
3199 2013-10-01  Richard Biener  <rguenther@suse.de>
3200
3201         PR tree-optimization/58553
3202         * tree-loop-distribution.c (struct partition_s): Add niter member.
3203         (classify_partition): Populate niter member for the partition
3204         and properly identify whether the relevant store happens before
3205         or after the loop exit.
3206         (generate_memset_builtin): Use niter member from the partition.
3207         (generate_memcpy_builtin): Likewise.
3208
3209 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3210
3211         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3212         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3213
3214 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3215
3216         * basic-block.h (edge_list): Prefix member names with "m_".
3217         * context.h (context): Likewise.
3218         * domwalk.h (dom_walker): Likewise.
3219         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3220         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3221         * hash-table.h (hash_table): Likewise.
3222         * machmode.h (bit_field_mode_iterator): Likewise.
3223         * pass_manager.h (pass_list): Likewise.
3224         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3225         * tree-pass.h (pass_data): Likewise.
3226         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3227         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3228         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3229         * asan.c (pass_data_asan): Update accordingly.
3230         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3231         (control_dependences::control_dependences): Likewise.
3232         (control_dependences::~control_dependences): Likewise.
3233         (control_dependences::~control_dependences): Likewise.
3234         (control_dependences::get_edges_dependent_on): Likewise.
3235         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3236         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3237         * context.c (gcc::context::context): Likewise.
3238         * cprop.c (pass_rtl_cprop::clone): Likewise.
3239         * domwalk.c (dom_walker::walk): Likewise.
3240         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3241         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3242         * mode-switching.c (pass_mode_switching::clone): Likewise.
3243         * passes.c (opt_pass::opt_pass): Likewise.
3244         (pass_manager::pass_manager): Likewise.
3245         * predict.c (pass_strip_predict_hints::clone): Likewise.
3246         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3247         (pass_split_all_insns::clone): Likewise.
3248         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3249         Likewise.
3250         (bit_field_mode_iterator::next_mode): Likewise.
3251         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3252         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3253         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3254         * tree-complex.c (pass_lower_complex::clone): Likewise.
3255         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3256         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3257         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3258         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3259         (pass_fold_builtins::clone): Likewise.
3260         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3261         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3262         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3263         (pass_cd_dce::clone): Likewise.
3264         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3265         (pass_phi_only_cprop::clone): Likewise.
3266         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3267         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3268         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3269         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3270         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3271         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3272         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3273         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3274         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3275         * tree-vrp.c (pass_vrp::clone): Likewise.
3276         * tsan.c (pass_tsan::clone): Likewise.
3277
3278 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3279
3280         PR middle-end/58564
3281         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3282         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3283
3284         PR middle-end/58564
3285         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3286         optimization, punt if sign_bit_p looked through any zero extension.
3287
3288 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3289
3290         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3291         Update redirected out edge count in joiner case.
3292         (ssa_redirect_edges): Common the joiner and non-joiner cases
3293         so that joiner case gets profile updates.
3294
3295 2013-09-30  Richard Biener  <rguenther@suse.de>
3296
3297         PR tree-optimization/58554
3298         * tree-loop-distribution.c (classify_partition): Require
3299         unconditionally executed stores for memcpy and memset recognition.
3300         (tree_loop_distribution): Calculate dominance info.
3301
3302 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3303
3304         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3305         (NO_PROFILE_COUNTERS): Likewise.
3306         (PROFILE_HOOK): Likewise.
3307         (FUNCTION_PROFILER): Likewise.
3308         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3309
3310 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3311
3312         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3313         calls and defines in TARGET_MACHO conditional.
3314         (load_macho_picbase_di): Likewise.
3315         (reload_macho_picbase): Likewise.
3316         (reload_macho_picbase_si): Likewise.
3317         (reload_macho_picbase_di): Likewise.
3318         (nonlocal_goto_receiver): Likewise.
3319
3320 2013-09-30  Nick Clifton  <nickc@redhat.com>
3321
3322         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3323         that use the MSP430X ISA.
3324         (msp430_option_override): Scan -mmcu command line option for any
3325         MCU name that supports the MSP430X ISA.
3326         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3327         -mmcu options which enable the MSP430X ISA.
3328
3329 2013-09-30  Richard Biener  <rguenther@suse.de>
3330
3331         PR middle-end/58532
3332         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3333         before looking for setjmp-like calls.
3334
3335 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3336
3337         PR target/10901
3338         * config/darwin-protos.h (machopic_get_function_picbase): New.
3339         * config/darwin.c (machopic_get_function_picbase): New.
3340         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3341         label for a new func.  (load_macho_picbase_di): Likewise.
3342         (reload_macho_picbase): New expand.
3343         (reload_macho_picbase_si): New insn.
3344         (reload_macho_picbase_di): New insn.
3345         (nonlocal_goto_receiver): New define and split.
3346         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3347         (unspecv enum): Add UNSPECV_NLGR.
3348
3349 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3350
3351         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3352         that altivec registers are correctly sized on Darwin.
3353
3354 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3355
3356         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3357         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3358         * config/x-darwin (host-darwin.o): Likewise.
3359         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3360         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3361         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3362
3363 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3364
3365         * doc/invoke.texi: Fix usage of @tie{} command.
3366
3367 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3368
3369         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3370
3371 2013-09-28  Jan Hubicka  <jh@suse.cz>
3372
3373         * config/i386/x86-tune.def: Add documentation for each of the options;
3374         add whitespace.
3375
3376 2013-09-28  Jan Hubicka  <jh@suse.cz>
3377
3378         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3379         generic.
3380         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3381         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3382         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3383         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3384         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3385
3386 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3387
3388         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3389         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3390         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3391         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3392         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3393         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3394         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3395         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3396         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3397         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3398         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3399         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3400         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3401         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3402         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3403         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3404         gimple.h, godump.c, graphite-clast-to-gimple.c,
3405         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3406         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3407         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3408         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3409         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3410         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3411         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3412         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3413         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3414         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3415         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3416         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3417         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3418         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3419         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3420         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3421         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3422         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3423         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3424         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3425         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3426         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3427         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3428         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3429         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3430         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3431         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3432         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3433         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3434         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3435         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3436         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3437         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3438         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3439         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3440         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3441         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3442         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3443         whitespace before "(".
3444
3445 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3446
3447         * expr.h (extract_bit_field): Remove packedp parameter.
3448         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3449         from forward declaration.
3450         (store_split_bit_field): Remove packedp arg from calls to
3451         extract_fixed_bit_field.
3452         (extract_bit_field_1): Remove packedp parameter and packedp
3453         argument from recursive calls and calls to extract_fixed_bit_field.
3454         (extract_bit_field): Remove packedp parameter and corresponding
3455         arg to extract_bit_field_1.
3456         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3457         to issue warnings.
3458         (extract_split_bit_field): Remove packedp arg from call to
3459         extract_fixed_bit_field.
3460         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3461         (copy_blkmode_from_reg): Likewise.
3462         (copy_blkmode_to_reg): Likewise.
3463         (read_complex_part): Likewise.
3464         (store_field): Likewise.
3465         (expand_expr_real_1): Likewise.
3466         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3467         to extract_bit_field.
3468         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3469         call to extract_bit_field.
3470         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3471         call to extract_bit_field.
3472         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3473         and special packedp behavior from -fstrict-volatile-bitfields
3474         documentation.
3475
3476 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3477
3478         * lra-eliminations.c (init_elim_table): Guard value_p.
3479
3480 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3481
3482         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3483         DFmode, DImode, and SFmode in the upper VSX registers based on the
3484         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3485         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3486         of the VSX register handling.
3487
3488         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3489         (f32_sv): Likewise.
3490         (zero_extendsidi2_lfiwzx): Add support for loading into the
3491         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3492         only do VSX memory options on Altivec registers.
3493         (extendsidi2_lfiwax): Likewise.
3494         (extendsfdf2_fpr): Likewise.
3495         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3496         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3497         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3498         (movdi_internal64): Likewise.
3499
3500 2013-09-27  Xinliang David Li  <davidxl@google.com>
3501
3502         * opts.c (finish_options): Adjust parameters
3503         according to vect cost model.
3504         (common_handle_option): Set dynamic vect cost
3505         model for FDO.
3506         targhooks.c (default_add_stmt_cost): Compute stmt cost
3507         unconditionally.
3508         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3509         Use helper function.
3510         * tree-vectorizer.h (unlimited_cost_model): New function.
3511         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3512         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3513         function.
3514         (vect_enhance_data_refs_alignment): Ditto.
3515         * flag-types.h: New enum.
3516         * common/config/i386/i386-common.c (ix86_option_init_struct):
3517         No need to initialize vect_cost_model flag.
3518         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3519         unconditionally.
3520
3521 2013-09-27  Diego Novillo  <dnovillo@google.com>
3522
3523         * gimple.h (enum ssa_mode): Remove.
3524
3525 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3526
3527         * cfgloop.h (number_of_loops): Fix typo in check for null.
3528
3529 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3530
3531         PR middle-end/58551
3532         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3533         are children of outermost saved_cfun's loop, and set it up to
3534         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3535         if loop != loop0 and SESE region contains bbs that belong to loop0.
3536
3537 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3538
3539         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3540         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3541         (set_address_segment, set_address_base, set_address_index)
3542         (set_address_disp): Accept the argument unconditionally.
3543         (baseness): Remove must_be_base_p and must_be_index_p checks.
3544         (decompose_normal_address): Classify as much as possible in the
3545         main loop.
3546
3547 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3548
3549         * cse.c (count_reg_usage): Handle INT_LIST.
3550         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3551         * reginfo.c (reg_scan_mark_refs): Likewise.
3552         * reload1.c (eliminate_regs_1): Likewise.
3553
3554 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3555
3556         PR middle-end/58547
3557         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3558         signed.
3559
3560 2013-09-27  Richard Biener  <rguenther@suse.de>
3561
3562         PR tree-optimization/58459
3563         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3564         restriction not propagating into loops.
3565
3566 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3567
3568         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3569         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3570         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3571
3572 2013-09-26  Richard Biener  <rguenther@suse.de>
3573
3574         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3575
3576 2013-09-26  Richard Biener  <rguenther@suse.de>
3577
3578         * alias.h (component_uses_parent_alias_set): Rename to ...
3579         (component_uses_parent_alias_set_from): ... this.
3580         * alias.c (component_uses_parent_alias_set): Rename to ...
3581         (component_uses_parent_alias_set_from): ... this and return
3582         the desired parent.
3583         (reference_alias_ptr_type_1): Use the result from
3584         component_uses_parent_alias_set_from instead of stripping
3585         components one at a time.
3586         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3587
3588 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3589
3590         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3591         Move prototypes to...
3592         * tree-ssa-ter.h: New File.  Move prototypes here.
3593         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3594         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3595         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3596         from is_replaceable_p.
3597         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3598         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3599         (process_replaceable): Use ter_is_replaceable_p.
3600         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3601         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3602         newly refactored ssa_is_replaceable_p.
3603         * cfgexpand.c: Include tree-outof-ssa.h.
3604         * ssaexpand.h: Delete.
3605
3606 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3607
3608         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3609         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3610         tree-ssa.c
3611         (create_gimple_tmp): Delete.
3612         (get_expr_type, build_assign, build_type_cast): Move to...
3613         * gimple-builder.c: New File.
3614         (get_expr_type): Relocate from gimple.c.
3615         (build_assign, build_type_cast): Change to only create ssanames.
3616         * gimple.h: Move prototypes to...
3617         * gimple-builder.h: New File. Here.
3618         * tree-ssa.h: And here.
3619         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3620         count_uses_and_derefs): Relocate from gimple.c.
3621         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3622         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3623         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3624         gimple_replace_ssa_lhs.
3625         * asan.c: Include gimple-builder.h.
3626         * Makefile.in: Add gimple-builder.o.
3627
3628 2013-09-26  Richard Biener  <rguenther@suse.de>
3629
3630         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3631         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3632         (loe_visit_block): Use gcc_checking_assert.
3633         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3634         gimple_assign_ssa_name_copy_p.
3635         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3636         change.
3637
3638 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3639
3640         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3641         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3642
3643 2013-09-26  Richard Biener  <rguenther@suse.de>
3644
3645         PR tree-optimization/58539
3646         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3647         the fact that debug statements are not taking part in loop-closed
3648         SSA construction.
3649
3650 2013-09-26  Nick Clifton  <nickc@redhat.com>
3651
3652         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3653         time warning message.
3654         (msp430_print_operand_raw): Delete unused letter parameter.
3655         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3656         (msp430_print_operand_address): New function.
3657         (msp430_print_operand): Move address printing code from here to
3658         new function.
3659         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3660         assembler.
3661         (zero_extendpsisi2): Likewise.
3662         (extendpsisi2): New pattern.
3663         (andneghi3): New pattern.
3664
3665 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3666
3667         * config/aarch64/aarch64.opt (mlra): New option.
3668         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3669         (TARGET_LRA_P): Define.
3670
3671 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3672
3673         * expr.c (expand_assignment): Remove obsolete comment.
3674
3675 2013-09-25  Jeff Law  <law@redhat.com>
3676
3677         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3678         tree-ssa-threadupdate.h.
3679         (register_jump_thread): Similarly.
3680         * tree-ssa-threadupdate.h: New header file.
3681         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3682         * tree-vrp.c: Likewise.
3683         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3684         (thread_around_empty_blocks): Change type of path vector argument to
3685         an edge,type pair from just an edge.  Initialize both elements when
3686         appending to a jump threading path.  Tweak references to elements
3687         appropriately.
3688         (thread_across_edge): Similarly.  Release memory for the elements
3689         as needed.
3690         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3691         (dump_jump_thread_path): New function broken out from
3692         register_jump_thread.
3693         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3694         path vector entries.  Search the path for NULL edges and dump
3695         the path if one is found.  Tweak the conversion of path to 3-edge
3696         form to use the block copy type information embedded in the path.
3697
3698 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3699
3700         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3701
3702 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3703             Vladimir Makarov  <vmakarov@redhat.com>
3704
3705         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3706         from the least significant bit.
3707         (strip_address_mutations): Add bitfield operations handling.
3708         (must_be_index_p): Add shifting and rotate operations handling.
3709         (set_address_base): Use must_be_base_p predicate.
3710         (set_address_index): Use must_be_index_p predicate.
3711
3712 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3713             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3714             Sergey Lega  <sergey.s.lega@intel.com>
3715             Anna Tikhonova  <anna.tikhonova@intel.com>
3716             Ilya Tocar  <ilya.tocar@intel.com>
3717             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3718             Ilya Verbin  <ilya.verbin@intel.com>
3719             Kirill Yukhin  <kirill.yukhin@intel.com>
3720             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3721
3722         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
3723         Use new names.
3724         (ix86_expand_vector_move_misalign): Support new unaligned load and
3725         stores and use new names.
3726         (CODE_FOR_sse2_storedqu): Rename to ...
3727         (CODE_FOR_sse2_storedquv16qi): ... this.
3728         (CODE_FOR_sse2_loaddqu): Rename to ...
3729         (CODE_FOR_sse2_loaddquv16qi): ... this.
3730         (CODE_FOR_avx_loaddqu256): Rename to ...
3731         (CODE_FOR_avx_loaddquv32qi): ... this.
3732         (CODE_FOR_avx_storedqu256): Rename to ...
3733         (CODE_FOR_avx_storedquv32qi): ... this.
3734         * config/i386/i386.md (fpint_logic): New.
3735         * config/i386/sse.md (VMOVE): Extend for AVX512.
3736         (VF): Ditto.
3737         (VF_128_256): New.
3738         (VF_512): Ditto.
3739         (VI_UNALIGNED_LOADSTORE): Ditto.
3740         (sse2_avx_avx512f): Ditto.
3741         (sse2_avx2): Extend for AVX512.
3742         (sse4_1_avx2): Ditto.
3743         (avx2_avx512f): New.
3744         (sse): Extend for AVX512.
3745         (sse2): Ditto.
3746         (sse4_1): Ditto.
3747         (avxsizesuffix): Ditto.
3748         (sseintvecmode): Ditto.
3749         (ssePSmode): Ditto.
3750         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
3751         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
3752         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3753         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
3754         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3755         (<sse2_avx_avx512f>_storedqu<mode): ... this.
3756         (<sse>_movnt<mode>): Replace constraint "x" with "v".
3757         (STORENT_MODE): Extend for AVX512.
3758         (*absneg<mode>2): Replace constraint "x" with "v".
3759         (*mul<mode>3): Ditto.
3760         (*ieee_smin<mode>3): Ditto.
3761         (*ieee_smax<mode>3): Ditto.
3762         (avx_cmp<mode>3): Replace VF with VF_128_256.
3763         (*<sse>_maskcmp<mode>3_comm): Ditto.
3764         (<sse>_maskcmp<mode>3): Ditto.
3765         (<sse>_andnot<mode>3): Extend for AVX512.
3766         (<code><mode>3, anylogic): Replace VF with VF_128_256.
3767         (<code><mode>3, fpint_logic): New.
3768         (*<code><mode>3): Extend for AVX512.
3769         (avx512flogicsuff): New.
3770         (avx512f_<logic><mode>): Ditto.
3771         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
3772         VF_128_256.
3773         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
3774         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
3775         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
3776         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
3777         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
3778         (xop_vpermil2<mode>3): Ditto.
3779         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
3780         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
3781         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
3782         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
3783
3784 2013-09-25  Tom Tromey  <tromey@redhat.com>
3785
3786         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
3787         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
3788         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
3789         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
3790         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
3791         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
3792         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
3793         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
3794         (GRAPHITE_HTAB_H): Remove.
3795
3796 2013-09-25  Tom Tromey  <tromey@redhat.com>
3797
3798         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
3799
3800 2013-09-25  Tom Tromey  <tromey@redhat.com>
3801
3802         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
3803
3804 2013-09-25  Tom Tromey  <tromey@redhat.com>
3805
3806         * config/i386/t-i386 (i386.o): Remove.
3807         (i386-c.o): Use COMPILE and POSTCOMPILE.
3808
3809 2013-09-25  Tom Tromey  <tromey@redhat.com>
3810
3811         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
3812
3813 2013-09-25  Tom Tromey  <tromey@redhat.com>
3814
3815         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
3816         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
3817         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
3818         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
3819         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
3820         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
3821         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
3822         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
3823         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
3824         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
3825         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
3826         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
3827         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
3828         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
3829         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
3830         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
3831         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
3832         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
3833         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
3834         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
3835         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
3836         (tree-ssa-pre.o, tree-ssa-sccvn.o)
3837         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
3838         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
3839         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
3840         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
3841         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
3842         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
3843         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
3844         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
3845         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
3846         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
3847         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
3848         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
3849         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
3850         (graphite-clast-to-gimple.o, graphite-dependences.o)
3851         (graphite-interchange.o, graphite-poly.o)
3852         (graphite-scop-detection.o, graphite-sese-to-poly.o)
3853         (graphite-optimize-isl.o, tree-vect-loop.o)
3854         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
3855         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
3856         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
3857         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
3858         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
3859         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
3860         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
3861         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
3862         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
3863         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
3864         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
3865         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
3866         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
3867         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
3868         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
3869         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
3870         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
3871         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
3872         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
3873         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
3874         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
3875         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
3876         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
3877         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
3878         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
3879         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
3880         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
3881         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
3882         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
3883         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
3884         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
3885         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
3886         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
3887         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
3888         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
3889         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
3890         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
3891         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
3892         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
3893         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
3894         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
3895         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
3896         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
3897         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
3898         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
3899         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
3900         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
3901         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
3902         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
3903         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
3904         (gcov-dump.o): Remove.
3905         (default-c.o): Use COMPILE and POSTCOMPILE.
3906         (CFLAGS-gcc.o): New variable.
3907         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
3908
3909 2013-09-25  Tom Tromey  <tromey@redhat.com>
3910
3911         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
3912         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
3913         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
3914         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
3915         (c-family/c-pragma.o, c-family/c-pretty-print.o)
3916         (c-family/c-semantics.o, c-family/c-ada-spec.o)
3917         (c-family/array-notation-common.o, c-family/stub-objc.o)
3918         (c-family/c-ubsan.o): Remove.
3919
3920 2013-09-25  Tom Tromey  <tromey@redhat.com>
3921
3922         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
3923
3924 2013-09-25  Tom Tromey  <tromey@redhat.com>
3925
3926         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
3927         to add -DENABLE_SHARED_LIBGCC.
3928         (gcc.o): Don't use subshell.
3929
3930 2013-09-25  Tom Tromey  <tromey@redhat.com>
3931
3932         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
3933         * configure.ac: Don't invoke AM_PROG_CC_C_O.
3934         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
3935         * configure, config.in: Rebuild.
3936
3937 2013-09-25  Tom Tromey  <tromey@redhat.com>
3938
3939         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
3940         (COMPILE, POSTCOMPILE): New variables.
3941         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
3942         (DEPFILES): New variable.
3943         Include ".Po" files.
3944         * configure.ac: Add checks for dependency checking.
3945         * configure, aclocal.m4: Regenerate.
3946
3947 2013-09-25  Tom Tromey  <tromey@redhat.com>
3948
3949         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
3950         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
3951
3952 2013-09-25  Tom Tromey  <tromey@redhat.com>
3953
3954         * Makefile.in (generated_files): Add options.h,
3955         target-hooks-def.h, insn-opinit.h,
3956         common/common-target-hooks-def.h, pass-instances.def,
3957         c-family/c-target-hooks-def.h.
3958
3959 2013-09-25  Jeff Law  <law@redhat.com>
3960
3961         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
3962         than foo[foo.length () - 1] to access last member in a vec.
3963         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
3964
3965 2013-09-25  Richard Biener  <rguenther@suse.de>
3966
3967         PR middle-end/58521
3968         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
3969
3970 2013-09-25  Jan Hubicka  <jh@suse.cz>
3971
3972         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
3973         test.
3974
3975 2013-09-25  Marek Polacek  <polacek@redhat.com>
3976
3977         PR sanitizer/58420
3978         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
3979         when determining the type name.
3980
3981 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
3982
3983         * config/sh/sh.md: Fix formatting.
3984
3985 2013-09-24  Xinliang David Li  <davidxl@google.com>
3986
3987         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
3988         max peel iterations parameter.
3989         * param.def: New parameter.
3990         * doc/invoke.texi: Document New parameter.
3991
3992 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
3993
3994         * gimple-pretty-print.c: Various whitespace tweaks.
3995         * tree-core.h: Likewise.
3996         * tree-pretty-print.c: Likewise.
3997         * tree-ssa-alias.c: Likewise.
3998         * tree-ssa-copy.c: Likewise.
3999         * tree-ssanames.c: Likewise.
4000         * tree-ssanames.h: Likewise.
4001         * tree-vrp.c: Likewise.
4002
4003 2013-09-24  Alan Modra  <amodra@gmail.com>
4004
4005         PR middle-end/57134
4006         PR middle-end/57586
4007         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4008         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4009         inout operands.
4010
4011 2013-09-24  Richard Biener  <rguenther@suse.de>
4012
4013         PR middle-end/58513
4014         * tree.c (reference_alias_ptr_type): Move ...
4015         * alias.c (reference_alias_ptr_type): ... here and implement
4016         in terms of the new reference_alias_ptr_type_1.
4017         (ref_all_alias_ptr_type_p): New helper.
4018         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4019         use ref_all_alias_ptr_type_p.
4020         (get_deref_alias_set): Add flag_strict_aliasing check here.
4021         (reference_alias_ptr_type_1): New function, split out from ...
4022         (get_alias_set): ... here.
4023         (alias_ptr_types_compatible_p): New function.
4024         * alias.h (reference_alias_ptr_type): Declare.
4025         (alias_ptr_types_compatible_p): Likewise.
4026         * tree.h (reference_alias_ptr_type): Remove.
4027         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4028         to compare MEM_REF alias types.
4029
4030 2013-09-24  Richard Biener  <rguenther@suse.de>
4031
4032         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4033
4034 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4035
4036         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4037         reload helper function arrays into a single array reg_addr.
4038         (reload_fpr_gpr): Likewise.
4039         (reload_gpr_vsx): Likewise.
4040         (reload_vsx_gpr): Likewise.
4041         (struct rs6000_reg_addr): Likewise.
4042         (reg_addr): Likewise.
4043         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4044         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4045         (rs6000_init_hard_regno_mode_ok): Likewise.
4046         (rs6000_secondary_reload_direct_move): Likewise.
4047         (rs6000_secondary_reload): Likewise.
4048
4049         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4050         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4051         power8 changes.  Put wg constraint in alphabetical order.
4052
4053         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4054         for future work to add ISA 2.07 VSX single precision support.
4055         (-mvsx-scalar-double): Change default from -1 to 1, update
4056         documentation comment.
4057         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4058         (-mupper-regs-df): New debug switch to control whether DF values
4059         can go in the traditional Altivec registers.
4060         (-mupper-regs-sf): New debug switch to control whether SF values
4061         can go in the traditional Altivec registers.
4062
4063         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4064         and wy constraints.
4065         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4066         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4067         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4068         (rs6000_debug_legitimate_address_p): Print if we are running
4069         before, during, or after reload.
4070         (rs6000_secondary_reload): Add a comment.
4071         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4072
4073         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4074         constraints.  Update documentation string.
4075         (wd constraint): Likewise.
4076         (wf constraint): Likewise.
4077         (wg constraint): Likewise.
4078         (wn constraint): Likewise.
4079         (ws constraint): Likewise.
4080         (wt constraint): Likewise.
4081         (wx constraint): Likewise.
4082         (wz constraint): Likewise.
4083         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4084         instructions.
4085         (ww constraint): Likewise.
4086         (wy constraint): Likewise.
4087         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4088         in the previous submissions.
4089         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4090
4091 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4092
4093         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4094         (REG_BR_PROB): Say that the probability is stored in an int_list.
4095         * reg-notes.def: Update commentary to mention INT_LIST.
4096         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4097         (INT_LIST): New rtx.
4098         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4099         * rtlanal.c (int_reg_note_p): New function.
4100         (alloc_reg_note): Assert that the note does not have an int argument.
4101         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4102         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4103         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4104         rather than an INSN_LIST.  Handle INT_LIST.
4105         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4106         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4107         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4108         Manipulate them as ints rather than rtxes.
4109         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4110         * regmove.c (copy_src_to_dest): Likewise.
4111         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4112
4113         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4114         into the cases that need it.
4115         * config/arm/arm.c (arm_unwind_emit): Likewise.
4116
4117         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4118         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4119         * loop-doloop.c (add_test, doloop_modify): Likewise.
4120         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4121         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4122         * predict.c (combine_predictions_for_insn): Likewise.
4123         * print-rtl.c (print_rtx): Handle INT_LIST.
4124         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4125         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4126         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4127         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4128         (ix86_split_fp_branch, predict_jump): Likewise.
4129         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4130         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4131         * config/spu/spu.c (ea_load_store_inline): Likewise.
4132
4133         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4134         value of a REG_BR_PROB note.
4135         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4136         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4137         * emit-rtl.c (try_split): Likewise.
4138         * predict.c (br_prob_note_reliable_p): Likewise.
4139         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4140         * reorg.c (mostly_true_jump): Likewise.
4141         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4142         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4143         * config/i386/i386.c (ix86_print_operand): Likewise.
4144         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4145         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4146         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4147         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4148         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4149         * config/sparc/sparc.c (output_cbranch): Likewise.
4150         * config/spu/spu.c (get_branch_target): Likewise.
4151         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4152         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4153
4154 2013-09-23  Jan Hubicka  <jh@suse.cz>
4155
4156         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4157         for ipa-devirt.
4158         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4159         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4160         of external calls
4161         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4162         (gimple_fold_call): Dump inconsistent devirtualizations; add
4163         sanity check for type based devirtualizations.
4164         * ipa-prop.c: Include ipa-utils.h
4165         (ipa_intraprocedural_devirtualization): Add sanity check.
4166         (try_make_edge_direct_virtual_call): Likewise.
4167
4168 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4169
4170         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4171         assignment statements.
4172
4173 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4174
4175         * gimple-pretty-print.c (dump_ssaname_info): New function.
4176         (dump_gimple_phi): Call it.
4177         (pp_gimple_stmt_1): Likewise.
4178         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4179         (range_info_def): Declare.
4180         * tree-pretty-print.c (pp_double_int): New function.
4181         (dump_generic_node): Call it.
4182         * tree-pretty-print.h (pp_double_int): Declare.
4183         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4184         * tree-ssanames.h (range_info_def): New structure.
4185         (value_range_type): Move definition here.
4186         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4187         Declare.
4188         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4189         initialization.
4190         (set_range_info): New function.
4191         (get_range_info): Likewise.
4192         (duplicate_ssa_name_range_info): Likewise.
4193         (duplicate_ssa_name_fn): Check pointer type and call
4194         duplicate_ssa_name_range_info.
4195         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4196         * tree-vrp.c (value_range_type): Remove definition, now in
4197         tree-ssanames.h.
4198         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4199         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4200         (SSA_NAME_RANGE_INFO): New macro.
4201
4202 2013-09-23  Richard Biener  <rguenther@suse.de>
4203
4204         PR tree-optimization/58464
4205         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4206         (phi_trans_add): Change to add conditionally on being not present.
4207         (phi_translate_1): Remove recursion detection here.
4208         (phi_translate): Pre-seed the cache with NULL to catch
4209         recursion here in a more generic way.
4210         (bitmap_find_leader): Adjust comment.
4211         (get_representative_for): Dump value-numbers.
4212         (create_expression_by_pieces): Likewise.
4213         (insert_into_preds_of_block): Likewise.
4214
4215 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4216
4217         PR target/58475
4218         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4219         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4220
4221 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4222
4223         Revert r202780:
4224         2013-09-20  Renlin Li  <renlin.li@arm.com>
4225
4226         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4227         plus_constant.
4228         (aarch64_expand_epilogue): Likewise.
4229         (aarch64_legitimize_reload_address): Likewise.
4230
4231 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4232
4233         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4234         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4235
4236 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4237
4238         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4239         comparison operators when -fno-trapping-math is in effect.
4240         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4241         comparison operators in TFmode and assert that unsupported operators
4242         cannot reach here.
4243         (ia64_print_operand): Likewise.
4244
4245 2013-09-21  Jan Hubicka  <jh@suse.cz>
4246
4247         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4248         (sse_typeless_stores): Enable for core
4249         (sse_load0_by_pxor): Likewise.
4250         (four_jump_limit): Disable for core.
4251         (pad_returns): Likewise.
4252         (avoid_vector_decode): Likewise.
4253         (fuse_cmp_and_branch): Enable for cores.
4254         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4255
4256 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4257
4258         PR middle-end/56791
4259         * config/pa/pa.c (pa_option_override): Disable auto increment and
4260         decrement instructions until reload is completed.
4261
4262         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4263         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4264         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4265
4266 2013-09-20  DJ Delorie  <dj@redhat.com>
4267             Nick Clifton  <nickc@redhat.com>
4268
4269         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4270         (need_to_save): Save bank 0 on interrupts.
4271         (characterize_address): Strip far address wrappers.
4272         (rl78_as_legitimate_address): Likewise.
4273         (transcode_memory_rtx): Likewise.
4274         (rl78_peep_movhi_p): Disable this peephole after devirt.
4275         (rl78_propogate_register_origins): Forget all origins when a
4276         CLOBBER is seen.
4277         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4278         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4279         peephole2's.
4280         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4281         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4282         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4283         relocating.
4284         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4285
4286 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4287
4288         * config/pa/pa.md: In "scc" insn patterns, change output template to
4289         handle const0_rtx in reg_or_0_operand operands.
4290
4291 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4292
4293         PR target/56875
4294         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4295         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4296
4297 2013-09-20  Richard Biener  <rguenther@suse.de>
4298
4299         PR middle-end/58484
4300         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4301         remembering SSA name version and block index.
4302         (new_scev_info_str): Adjust.
4303         (hash_scev_info): Likewise.  Also hash the block index.
4304         (eq_scev_info): Adjust.
4305         (find_var_scev_info): Likewise.
4306         (struct instantiate_cache_entry): Remove.
4307         (struct instantiate_cache_type): Use a htab to map name, block
4308         to chrec.
4309         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4310         (get_instantiated_value_entry): Likewise.
4311         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4312         (instantiate_scev_name): Adjust.
4313
4314 2013-09-20  Jeff Law  <law@redhat.com>
4315
4316         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4317
4318 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4319
4320         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4321         Call aarch64_simd_expand_args to update op[argc].
4322
4323 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4324
4325         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4326         plugin argument.
4327
4328 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4329
4330         * gengtype.c (file_rules): Added rule for *.cc files.
4331         (get_output_file_with_visibility): Give fatal message when no
4332         rules found.
4333
4334 2013-09-20  Renlin Li  <renlin.li@arm.com>
4335
4336         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4337         (aarch64_expand_epilogue): Likewise.
4338         (aarch64_legitimize_reload_address): Likewise.
4339
4340 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4341
4342         PR middle-end/57748
4343         * expr.c (expand_assignment): Remove misalignp code path.
4344
4345 2013-09-20  Marek Polacek  <polacek@redhat.com>
4346
4347         PR sanitizer/58413
4348         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4349         TYPE_PRECISION.  Add asserts.
4350
4351 2013-09-20  Richard Biener  <rguenther@suse.de>
4352
4353         PR tree-optimization/58453
4354         * tree-loop-distribution.c (distribute_loop): Apply the cost
4355         model for -ftree-loop-distribute-patterns, too.
4356
4357 2013-09-20  Richard Biener  <rguenther@suse.de>
4358
4359         PR middle-end/58473
4360         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4361         make type comparison less strict.
4362
4363 2013-09-20  Alan Modra  <amodra@gmail.com>
4364
4365         * configure: Regenerate.
4366         * aclocal.m4: Regenerate.
4367
4368 2013-09-20  Marek Polacek  <polacek@redhat.com>
4369
4370         PR other/58467
4371         * doc/extend.texi: Document that attribute used is meant to be used
4372         on variables with static storage duration.
4373
4374 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4375
4376         PR tree-optimization/58472
4377         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4378         simd_lane_access set inv_p = false.
4379         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4380         the simduid magic VAR_DECL.
4381
4382 2013-09-19  Jan Hubicka  <jh@suse.cz>
4383
4384         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4385
4386 2013-09-17  Jeff Law  <law@redhat.com>
4387
4388         * tree-ssa-dom.c (record_temporary_equivalences): New function
4389         split out of dom_opt_dom_walker::after_dom_children.
4390         (dom_opt_dom_walker::thread_across_edge): Move common code
4391         in here from dom_opt_dom_walker::after_dom_children.
4392         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4393
4394 2013-09-19  Jan Hubicka  <jh@suse.cz>
4395
4396         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4397         (TARGET_GENERIC): Use PROCESOR_GENERIC
4398         (enum processor_type): Unify generic32 and 64.
4399         * i386.md (cpu): Likewise.
4400         * x86-tune.def (use_leave): Enable for generic32.
4401         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4402         * athlon.md: Change generic64 to generic in all occurences.
4403         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4404         (ix86_target_macros_internal): Likewise.
4405         * driver-i386.c (host_detect_local_cpu): Likewise.
4406         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4407         to ..
4408         (generic_memcpy, generic_memset, generic_cost): This one.
4409         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4410         (m_GENERIC32, m_GENERIC64): Remove.
4411         (m_GENERIC): Turn into one flag.
4412         (processor_target): Unify generic tunnings.
4413         (ix86_option_override_internal): Replace generic32/64 by generic.
4414         (ix86_issue_rate): Likewise.
4415         (ix86_adjust_cost): Likewise.
4416
4417 2013-09-19  Jan Hubicka  <jh@suse.cz>
4418
4419         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4420         of speculative flag.
4421
4422 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4423
4424         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4425         GOMP_sections_start, even if !exit_reachable.
4426
4427 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4428
4429         * lra-constraints.c (need_for_all_save_p): Use macro
4430         HARD_REGNO_CALL_PART_CLOBBERED.
4431         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4432         set up pseudo conflict hard regs.
4433
4434 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4435
4436         PR target/58452
4437         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4438         operands.
4439
4440 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4441
4442         PR rtl-optimization/58438
4443         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4444         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4445         from previous subpasses.
4446
4447 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4448
4449         * arm.c (arm_get_frame_offsets): Validate architecture supports
4450         LDRD/STRD before accepting the tuning preference.
4451         (arm_expand_prologue): Likewise.
4452         (arm_expand_epilogue): Likewise.
4453
4454 2013-09-18  Richard Biener  <rguenther@suse.de>
4455
4456         PR tree-optimization/58417
4457         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4458         have chrecs with symbols defined in the loop as operands.
4459         (chrec_fold_multiply): Likewise.
4460         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4461         parameters before folding binary operations.
4462         (struct instantiate_cache_entry_hasher): Remove.
4463         (struct instantiate_cache_type): Use a pointer-map.
4464         (instantiate_cache_type::instantiate_cache_type): New function.
4465         (instantiate_cache_type::get): Likewise.
4466         (instantiate_cache_type::set): Likewise.
4467         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4468         (get_instantiated_value_entry): Likewise.
4469         (global_cache): New global.
4470         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4471         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4472         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4473         (instantiate_scev_name): Adjust.
4474         (instantiate_scev): Construct global instead of local cache.
4475         (resolve_mixers): Likewise.
4476
4477 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4478             Paolo Carlini  <paolo.carlini@oracle.com>
4479
4480         PR c++/58458
4481         * doc/implement-cxx.texi: Fix references to the C++ standards.
4482
4483 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4484
4485         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4486         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4487         simd_lane_access drs, update also DR_ALIGNED_TO.
4488
4489 2013-09-18  Marek Polacek  <polacek@redhat.com>
4490
4491         PR sanitizer/58411
4492         * doc/extend.texi: Document no_sanitize_undefined attribute.
4493         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4494         no_sanitize_undefined attribute.
4495
4496 2013-09-18  Nick Clifton  <nickc@redhat.com>
4497
4498         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4499         (ASM_DECLARE_FUNCTION_NAME): Define.
4500
4501 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4502
4503         * compare-elim.c (find_comparison_dom_walker): New class
4504         (find_comparisons_in_bb): Rename to
4505         find_comparison_dom_walker::before_dom_children
4506         (find_comparisons): Adjust
4507         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4508         adjust.
4509         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4510         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4511         (init_walk_dominator_tree): Remove declaration.
4512         (fini_walk_dominator_tree): Remove declaration.
4513         * fwprop.c (single_def_use_dom_walker): New class
4514         (single_def_use_enter_block): Convert to
4515         single_def_use_dom_walker::before_dom_children.
4516         (single_def_use_leave_block): Convert to
4517         single_def_use_dom_walker::after_dom_children.
4518         (build_single_def_use_links): Adjust.
4519         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4520         class.
4521         (find_candidates_in_block): Convert to
4522         find_candidates_dom_walker::before_dom_children.
4523         (execute_strength_reduction): Adjust.
4524         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4525         (sese_dom_walker): New class.
4526         (sese_dom_walker::sese_dom_walker): New constructor.
4527         (sese_dom_walker::~sese_dom_walker): New destructor.
4528         (build_sese_conditions_before): Convert to
4529         sese_dom_walker::before_dom_children.
4530         (build_sese_conditions_after): Convert to
4531         sese_dom_walker::after_dom_children.
4532         (build_poly_scop): Adjust
4533         * tree-into-ssa.c (rewrite_dom_walker): New class
4534         (rewrite_enter_block): Convert to
4535         rewrite_dom_walker::before_dom_children.
4536         (rewrite_leave_block): Convert to
4537         rewrite_dom_walker::after_dom_children.
4538         (rewrite_update_dom_walker): New class.
4539         (rewrite_update_enter_block): Convert to
4540         rewrite_update_dom_walker::before_dom_children.
4541         (rewrite_update_leave_block): Convert to
4542         rewrite_update_dom_walker::after_dom_children.
4543         (rewrite_blocks, rewrite_into_ssa): Adjust.
4544         (mark_def_dom_walker): New class.
4545         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4546         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4547         (mark_def_sites_blocks): Convert to
4548         mark_def_dom_walker::before_dom_children.
4549         (mark_def_site_blocks): Remove.
4550         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4551         (tree_ssa_dominator_optimize): Adjust.
4552         (dom_thread_across_edge): Convert to method
4553         dom_opt_dom_walker::thread_across_edge.
4554         (dom_opt_enter_block): Convert to member function
4555         dom_opt_dom_walker::before_dom_children.
4556         (dom_opt_leave_block): Convert to member function
4557         dom_opt_dom_walker::after_dom_children.
4558         * tree-ssa-dse.c (dse_dom_walker): New class.
4559         (dse_enter_block): Convert to member function
4560         dse_dom_walker::before_dom_children.
4561         (tree_ssa_dse): Adjust.
4562         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4563         (determine_invariantness_stmt): Convert to method
4564         invariantness_dom_walker::before_dom_children.
4565         (determine_invariantness): Remove
4566         (move_computations_dom_walker): New class.
4567         (move_computations_stmt): Convert to method
4568         move_computations_dom_walker::before_dom_children.
4569         (move_computations, tree_ssa_lim): Adjust.
4570         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4571         (nt_init_block): Make method
4572         notrappping_dom_walker::before_dom_children.
4573         (nt_fini_block): Make
4574         method nontrapping_dom_walker::after_dom_children.
4575         (get_non_trapping): Adjust.
4576         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4577         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4578         (eliminate_leave_block): Make method.
4579         eliminate_dom_walker::after_dom_children.
4580         (eliminate): Adjust
4581         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4582         (strlen_enter_block): Make method
4583         strlen_dom_walker::before_dom_children.
4584         (strlen_leave_block): Make
4585         method strlen_dom_walker::after_dom_children.
4586         (tree_ssa_strlen): Adjust.
4587         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4588         (tree_ssa_uncprop): Adjust.
4589         (uncprop_leave_block): Make method
4590         uncprop_dom_walker::after_dom_children.
4591         (uncprop_leave_block): Make method
4592         uncprop_dom_walker::before_dom_children.
4593
4594 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4595
4596         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4597         branch in same basic block.  Check both src and dest of the move insn.
4598
4599 2013-09-17  Nick Clifton  <nickc@redhat.com>
4600
4601         * config/rl78/rl78-real.md (bf): New pattern.
4602         (bt): New pattern.
4603         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4604         (rl78_print_operand): Do not put a # before a %B.
4605         * config/rl78/rl78.opt: Tweak doc strings.
4606
4607 2013-09-17  DJ Delorie  <dj@redhat.com>
4608
4609         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4610         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4611         (rl78_option_override): Likewise, if -mallregs.
4612         (is_virtual_register): Likewise.
4613         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4614         (REGNO_OK_FOR_BASE_P): Likewise.
4615         * config/rl78/rl78.opt (-mallregs): New.
4616
4617 2013-09-17  Nick Clifton  <nickc@redhat.com>
4618
4619         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4620         For interrupt functions: save all call clobbered registers if the
4621         interrupt handler is not a leaf function.
4622         (rl78_expand_prologue): Always recompute the frame information.
4623         For interrupt functions: only select bank 0 if one of the bank 0
4624         registers is going to be psuhed.
4625
4626 2013-09-17  DJ Delorie  <dj@redhat.com>
4627
4628         * config/rl78/constraints.md: For each W* constraint, rename to C*
4629         and create a W* constraint that checks for an optional ES: prefix
4630         pattern also.
4631         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4632         (es_addr): New.  Used to wrap far addresses.
4633         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4634         (rl78_es_base): New.
4635         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4636         wrapped far addresses.
4637         (rl78_print_operand_1): Unwrap far addresses before processing.
4638         (rl78_lo16): Wrap far addresses in unspecs.
4639         (rl78_es_addr): New.
4640         (rl78_es_base): New.
4641         (insn_ok_now): Check for not-yet-wrapped far addresses.
4642         (transcode_memory_rtx): Properly re-wrap far addresses.
4643
4644 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4645
4646         * config/sparc/t-rtems: Add leon3 multilibs.
4647
4648 2013-09-17  Cong Hou  <congh@google.com>
4649
4650         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4651         when checking the dot production pattern. The type of rhs operand
4652         of multiply is now checked correctly.
4653
4654 2013-09-17  Jeff Law  <law@redhat.com>
4655
4656         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4657         edge implied equivalences into successor phis.
4658         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4659         here from tree-ssa-threadedge.c.
4660         (mark_threaded_blocks): When threading through a joiner, if both
4661         successors of the joiner's clone reach the same block, verify the
4662         PHI arguments are equal.  If not, cancel the jump threading request.
4663         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4664         tree-ssa-threadupdate.c
4665         (thread_across_edge): Don't check PHI argument equality when
4666         threading through joiner block here.
4667
4668 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4669
4670         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4671         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4672         (warn_uninit, warn_uninitialized_vars,
4673         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4674         Move to tree-ssa-uninit.c.
4675         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4676         (has_undefined_value_p): New.  Pass dependant parts of
4677         ssa_undefined_value_p.
4678         (uninit_undefined_value_p): Use has_undefined_value_p.
4679         (warn_uninit, warn_uninitialized_vars,
4680         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4681         Move from tree-ssa.c.
4682         * tree-ssa.h: Adjust prototypes.
4683
4684 2013-09-17  Jan Hubicka  <jh@suse.cz>
4685
4686         PR middle-end/58332
4687         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4688         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4689         FUNCTION_NOT_OPTIMIZED.
4690         * ipa-inline-analysis.c (compute_inline_parameters): Function
4691         not optimized is not inlinable unless it is alwaysinline.
4692         (inline_analyze_function): Force calls in not optimized
4693         function not inlinable.
4694
4695 2013-09-17  Jan Hubicka  <jh@suse.cz>
4696
4697         PR middle-end/58329
4698         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4699         to return NULL.
4700         * ipa.c (function_and_variable_visibility): Likewise.
4701         * ipa-profile.c (ipa_profile): Likewise.
4702
4703 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4704
4705         PR ipa/58398
4706         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4707         attribute, and don't inline the resolver in this case.
4708
4709 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4710
4711         * coverage.c (get_coverage_counts): Add missing newline.
4712
4713 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4714
4715         PR tree-optimization/58088
4716         * fold-const.c (mask_with_trailing_zeros): New function.
4717         (fold_binary_loc): Make sure we don't recurse infinitely
4718         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
4719         Use mask_with_trailing_zeros where appropriate.
4720
4721 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
4722
4723         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
4724         of 'prev' var to get better distance estimation.
4725
4726 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
4727
4728         * tree-inline.h (struct copy_body_data): Add transform_parameter.
4729         * tree-inline.c (is_parameter_of): New predicate.
4730         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
4731         a parameter has been remapped.
4732         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
4733         (optimize_inline_calls): Initialize transform_parameter.
4734         (copy_gimple_seq_and_replace_locals): Likewise.
4735         (tree_function_versioning): Likewise.
4736         (maybe_inline_call_in_expr): Likewise.
4737
4738 2013-09-17  Nick Clifton  <nickc@redhat.com>
4739
4740         * config/msp430/msp430-protos.h: Add prototypes for new functions.
4741         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
4742         interrupt handlers.
4743         (is_attr_func): New function.
4744         (msp430_is_interrupt_func): New function.
4745         (is_naked_func): New function.
4746         (is_reentrant_func): New function.
4747         (is_critical_func): New function.
4748         (msp430_start_function): Add annotations for function attributes.
4749         (msp430_attr): New function.
4750         (msp430_attribute_table): New.
4751         (msp430_function_section): New function.
4752         (TARGET_ASM_FUNCTION_SECTION): Define.
4753         (msp430_builtin): New enum.
4754         (msp430_init_builtins): New function.
4755         (msp430_builtin_devl): New function.
4756         (msp430_expand_builtin): New function.
4757         (TARGET_INIT_BUILTINS): Define.
4758         (TARGET_EXPAND_BUILTINS): Define.
4759         (TARGET_BUILTIN_DECL): Define.
4760         (msp430_expand_prologue): Add support for naked, interrupt,
4761         critical and reentrant functions.
4762         (msp430_expand_epilogue): Likewise.
4763         (msp430_print_operand): Handle 'O' character.
4764         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
4765         NO_TRAMPOLINES.
4766         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
4767         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
4768         (pushm): Use a 'n' rather than an 'i' constraint.
4769         (msp_return): Add generation of the interrupt return instruction.
4770         (disable_interrupts): New pattern.
4771         (enable_interrupts): New pattern.
4772         (push_intr_state): New pattern.
4773         (pop_intr_state): New pattern.
4774         (bic_SR): New pattern.
4775         (bis_SR): New pattern.
4776         * doc/extend.texi: Document MSP430 function attributes and builtin
4777         functions.
4778
4779 2013-09-17  Richard Biener  <rguenther@suse.de>
4780
4781         PR tree-optimization/58432
4782         * tree-loop-distribution.c (tree_loop_distribution): Also
4783         scan PHIs for outside loop uses and seed a partition from them.
4784
4785 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
4786
4787         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
4788         (restructure_reference): Call backtrace_base_for_ref.
4789
4790 2013-09-17  Alan Modra  <amodra@gmail.com>
4791
4792         PR target/57589
4793         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
4794         patch.
4795
4796 2013-09-16  DJ Delorie  <dj@redhat.com>
4797
4798         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
4799         vregs location for RL78/G10.
4800         (rl78_expand_prologue): Avoid SEL on G10.
4801         (rl78_expand_epilogue): Likewise.
4802         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
4803         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
4804         __RL78_G10__ when appropriate.
4805         (ASM_SPEC): Pass -mg10 along to the assembler.
4806         * config/rl78/rl78.md (sel_rb): Disable for G10.
4807         * config/rl78/rl78.opt: Add -mg10 option.
4808         * config/rl78/t-rl78: Add -mg10 multilib.
4809
4810 2013-09-16  Xinliang David Li  <davidxl@google.com>
4811
4812         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
4813         * omp-low.c (omp_max_vf): Ditto.
4814         (expand_omp_simd): Ditto.
4815         * tree-vectorizer.c (vectorize_loops): Ditto.
4816         (gate_vect_slp): Ditto.
4817         (gate_increase_alignment): Ditto.
4818         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
4819         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
4820         (gate_tree_vectorize): Name change.
4821         (tree_vectorize): Ditto.
4822         (pass_vectorize::gate): Call new function.
4823         (pass_vectorize::execute): Ditto.
4824         * opts.c: O3 default setting change.
4825         (finish_options): Check new flag.
4826         * doc/invoke.texi: Document new flags.
4827         * common.opt: New flags.
4828
4829 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
4830
4831         * doc/tm.texi.in (Cond Exec Macros): Remove node.
4832         (Condition Code): Don't reference it.
4833         * doc/tm.texi: Regenerate.
4834
4835 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
4836
4837         PR middle-end/58418
4838         * lra-constraints.c (undo_optional_reloads): Consider all optional
4839         reload even if it did not get a hard reg.
4840
4841 2013-09-16  Teresa Johnson  <tejohnson@google.com>
4842
4843         * dumpfile.c (dump_loc): Remove newline emission.
4844         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
4845         emission to dump_printf_loc calls where missing.
4846         (vect_mark_for_runtime_alias_test): Ditto.
4847         (vect_analyze_data_ref_dependence): Ditto.
4848         (vect_analyze_data_ref_dependences): Ditto.
4849         (vect_slp_analyze_data_ref_dependence): Ditto.
4850         (vect_slp_analyze_data_ref_dependences): Ditto.
4851         (vect_compute_data_ref_alignment): Ditto.
4852         (vect_update_misalignment_for_peel): Ditto.
4853         (vect_verify_datarefs_alignment): Ditto.
4854         (vector_alignment_reachable_p): Ditto.
4855         (vect_get_data_access_cost): Ditto.
4856         (vect_enhance_data_refs_alignment): Ditto.
4857         (vect_find_same_alignment_drs): Ditto.
4858         (vect_analyze_data_refs_alignment): Ditto.
4859         (vect_analyze_group_access): Ditto.
4860         (vect_analyze_data_ref_access): Ditto.
4861         (vect_analyze_data_ref_accesses): Ditto.
4862         (vect_prune_runtime_alias_test_list): Ditto.
4863         (vect_analyze_data_refs): Ditto.
4864         (vect_create_addr_base_for_vector_ref): Ditto.
4865         (vect_create_data_ref_ptr): Ditto.
4866         (vect_grouped_store_supported): Ditto.
4867         (vect_grouped_load_supported): Ditto.
4868         * value-prof.c (check_counter): Ditto.
4869         (check_ic_target): Ditto.
4870         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
4871         (vect_recog_widen_mult_pattern): Ditto.
4872         (vect_recog_widen_sum_pattern): Ditto.
4873         (vect_recog_over_widening_pattern): Ditto.
4874         (vect_recog_widen_shift_pattern): Ditto.
4875         (vect_recog_rotate_pattern): Ditto.
4876         (vect_recog_vector_vector_shift_pattern): Ditto.
4877         (vect_recog_divmod_pattern): Ditto.
4878         (vect_recog_mixed_size_cond_pattern): Ditto.
4879         (vect_recog_bool_pattern): Ditto.
4880         (vect_pattern_recog_1): Ditto.
4881         (vect_pattern_recog): Ditto.
4882         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
4883         (vect_is_simple_iv_evolution): Ditto.
4884         (vect_analyze_scalar_cycles_1): Ditto.
4885         (vect_get_loop_niters): Ditto.
4886         (vect_analyze_loop_1): Ditto.
4887         (vect_analyze_loop_form): Ditto.
4888         (vect_analyze_loop_operations): Ditto.
4889         (vect_analyze_loop_2): Ditto.
4890         (vect_analyze_loop): Ditto.
4891         (report_vect_op): Ditto.
4892         (vect_is_slp_reduction): Ditto.
4893         (vect_is_simple_reduction_1): Ditto.
4894         (vect_get_known_peeling_cost): Ditto.
4895         (vect_estimate_min_profitable_iters): Ditto.
4896         (vect_model_reduction_cost): Ditto.
4897         (vect_model_induction_cost): Ditto.
4898         (get_initial_def_for_induction): Ditto.
4899         (vect_create_epilog_for_reduction): Ditto.
4900         (vectorizable_reduction): Ditto.
4901         (vectorizable_induction): Ditto.
4902         (vectorizable_live_operation): Ditto.
4903         (vect_loop_kill_debug_uses): Ditto.
4904         (vect_transform_loop): Ditto.
4905         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
4906         (vect_stmt_relevant_p): Ditto.
4907         (process_use): Ditto.
4908         (vect_mark_stmts_to_be_vectorized): Ditto.
4909         (vect_model_simple_cost): Ditto.
4910         (vect_model_promotion_demotion_cost): Ditto.
4911         (vect_model_store_cost): Ditto.
4912         (vect_get_store_cost): Ditto.
4913         (vect_model_load_cost): Ditto.
4914         (vect_get_load_cost): Ditto.
4915         (vect_init_vector_1): Ditto.
4916         (vect_get_vec_def_for_operand): Ditto.
4917         (vect_finish_stmt_generation): Ditto.
4918         (vectorizable_call): Ditto.
4919         (vectorizable_conversion): Ditto.
4920         (vectorizable_assignment): Ditto.
4921         (vectorizable_shift): Ditto.
4922         (vectorizable_operation): Ditto.
4923         (vectorizable_store): Ditto.
4924         (vectorizable_load): Ditto.
4925         (vectorizable_condition): Ditto.
4926         (vect_analyze_stmt): Ditto.
4927         (vect_transform_stmt): Ditto.
4928         (vect_is_simple_use): Ditto.
4929         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
4930         (vect_can_advance_ivs_p): Ditto.
4931         (vect_update_ivs_after_vectorizer): Ditto.
4932         (vect_do_peeling_for_loop_bound): Ditto.
4933         (vect_gen_niters_for_prolog_loop): Ditto.
4934         (vect_update_inits_of_drs): Ditto.
4935         (vect_create_cond_for_alias_checks): Ditto.
4936         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
4937         (vect_build_slp_tree_1): Ditto.
4938         (vect_supported_load_permutation_p): Ditto.
4939         (vect_analyze_slp_instance): Ditto.
4940         (vect_analyze_slp): Ditto.
4941         (vect_make_slp_decision): Ditto.
4942         (vect_detect_hybrid_slp): Ditto.
4943         (vect_bb_vectorization_profitable_p): Ditto.
4944         (vect_slp_analyze_bb_1): Ditto.
4945         (vect_update_slp_costs_according_to_vf): Ditto.
4946         (vect_get_mask_element): Ditto.
4947         (vect_transform_slp_perm_load): Ditto.
4948         (vect_schedule_slp_instance): Ditto.
4949         (vect_schedule_slp): Ditto.
4950         (vect_slp_transform_bb): Ditto.
4951         * profile.c (read_profile_edge_counts): Ditto.
4952         (compute_branch_probabilities): Ditto.
4953         * coverage.c (get_coverage_counts): Ditto.
4954
4955 2013-09-16  Diego Novillo  <dnovillo@google.com>
4956
4957         * tree-core.h: Add missing comment lines from refactoring of tree.h.
4958
4959 2013-09-16  Jan Hubicka  <jh@suse.cz>
4960
4961         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
4962         abstract functions; for static functions check the presence of body.
4963
4964 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
4965
4966         * config/aarch64/aarch64-simd-builtins.def (fma): New.
4967         * config/aarch64/aarch64-simd.md
4968         (aarch64_mla_elt<mode>): New.
4969         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4970         (aarch64_mls_elt<mode>): Likewise.
4971         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4972         (aarch64_fma4_elt<mode>): Likewise.
4973         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4974         (aarch64_fma4_elt_to_128v2df): Likewise.
4975         (aarch64_fma4_elt_to_64df): Likewise.
4976         (fnma<mode>4): Likewise.
4977         (aarch64_fnma4_elt<mode>): Likewise.
4978         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4979         (aarch64_fnma4_elt_to_128v2df): Likewise.
4980         (aarch64_fnma4_elt_to_64df): Likewise.
4981         * config/aarch64/iterators.md (VDQSF): New.
4982         * config/aarch64/arm_neon.h
4983         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
4984         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
4985
4986 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
4987
4988         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
4989         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4990         (aarch64_mul3_elt_to_128df): Likewise.
4991         (aarch64_mul3_elt_to_64v2df): Likewise.
4992         * config/aarch64/iterators.md (VEL): Also handle DFmode.
4993         (VMUL): New.
4994         (VMUL_CHANGE_NLANES) Likewise.
4995         (h_con): Likewise.
4996         (f): Likewise.
4997         * config/aarch64/arm_neon.h
4998         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
4999
5000 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5001
5002         * config/aarch64/arm_neon.h
5003         (vcvtx_high_f32_f64): Fix parameters.
5004
5005 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5006             Uros Bizjak  <ubizjak@gmail.com>
5007
5008         * config/alpha.c: Include tree-ssanames.h.
5009
5010 2013-09-16  Richard Biener  <rguenther@suse.de>
5011
5012         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5013         (dot_rdg_1): Handle control_dd.
5014         (create_edge_for_control_dependence): New function.
5015         (create_rdg_edges): Add control dependences if asked for.
5016         (build_rdg): Likewise.
5017         (generate_loops_for_partition): If there are not necessary
5018         control stmts remove all their dependencies.
5019         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5020         (distribute_loop): Pass on control dependences.
5021         (tree_loop_distribution): Compute control dependences and remove
5022         restriction on number of loop nodes.
5023
5024 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5025
5026         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5027         for internal calls.
5028
5029 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5030
5031         * cse.c (try_const_anchors): Punt on CC modes.
5032
5033 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5034
5035         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5036
5037 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5038
5039         PR target/58382
5040         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5041         calls to word_mode.
5042
5043 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5044
5045         PR target/48094
5046         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5047         seen.
5048         (darwin_objc1_section): Likewise.
5049         (darwin_file_end): Emit Image Info section when required.
5050
5051 2013-09-14  Jan Hubicka  <jh@suse.cz>
5052
5053         * tree-into-ssa.c (gate_into_ssa): New.
5054         (pass_data_build_ssa): Use it.
5055         * cgraph.h (expand_thunk): Update prototype.
5056         * cgraphunit.c (analyze_function): Expand thunks early.
5057         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5058         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5059         set lowered flag; do not add new function.
5060         (assemble_thunks_and_aliases): Update.
5061         * tree-ssa.c (gate_init_datastructures): New gate.
5062         (pass_data_init_datastructures): Use it.
5063
5064 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5065
5066         PR target/58269
5067         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5068         xmm register set described in the psABI.
5069
5070 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5071
5072         * dwarf2out.c (should_emit_struct_debug): Add check
5073         for type_decl variable is not NULL.
5074
5075 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5076
5077         * config.gcc: Use new winnt-c.c target hooks
5078         * config/t-winnt: New file
5079         * config/winnt-c.c: New file
5080         * doc/tm.texi.in: Document new hook
5081         * doc/tm.texi: Regenerated
5082
5083 2013-09-13  Jan Hubicka  <jh@suse.cz>
5084
5085         PR middle-end/58094
5086         * ipa-inline.c (check_callers): New function.
5087         (check_caller_edge): Remove.
5088         (want_inline_function_to_all_callers_p): Also permit alises that are
5089         called dirrectly.
5090         (inline_to_all_callers): Terminate the walk when devirtualization
5091         introduce new calls.
5092
5093 2013-09-13  Jan Hubicka  <jh@suse.cz>
5094
5095         * ipa-inline-analysis.c (struct growth_data): Add node.
5096         (do_estimate_growth_1): Fix detection of recursion.
5097
5098 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5099
5100         PR tree-optimization/58392
5101         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5102         to avoid shadowing of outer loop variable.  If
5103         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5104         replace_by_duplicate_decl simduid of loops that have it set and
5105         set dest_cfun->has_simduid_loops and/or
5106         dest_cfun->has_force_vect_loops.
5107         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5108         instead of maybe_lookup_decl.
5109         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5110         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5111         call.  Copy over force_vect and copy and remap simduid.  Set
5112         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5113         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5114         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5115         caller.  Don't set cfun->has_simduid_loops and/or
5116         cfun->has_force_vect_loops here.
5117         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5118         caller.
5119         (expand_call_inline, tree_function_versioning): Adjust copy_body
5120         callers.
5121
5122 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5123
5124         PR bootstrap/58388
5125         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5126         the assert if the edge was a speculative one.
5127
5128 2013-09-13  Richard Biener  <rguenther@suse.de>
5129
5130         * tree-data-ref.h (known_dependences_p): Move here ...
5131         * tree-loop-distribution.c (known_dependences_p): ... from here.
5132         (dump_rdg_component, debug_rdg_component): Remove.
5133         (dump_rdg): Adjust.
5134         (generate_loops_for_partition): Use gimple_uid instead of
5135         relying on matching stmt visit order.
5136         (rdg_build_partitions): Take starting stmt vector.
5137         (ldist_gen): Merge into ...
5138         (distribute_loop): ... this function.  Do not compute starting
5139         vertices vector.
5140         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5141
5142 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5143
5144         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5145         Set type attribute correctly. Set predicable_short_it attribute.
5146         (cmpsi_shiftsi): Remove %? from output template.
5147
5148 2013-09-13  Richard Biener  <rguenther@suse.de>
5149
5150         * tree-loop-distribution.c (struct rdg_component,
5151         rdg_defs_used_in_other_loops_p, free_rdg_components,
5152         rdg_build_components): Remove.
5153         (stmts_from_loop): Do not record virtual PHIs.
5154         (generate_loops_for_partition): Skip virtual PHIs.
5155         (build_rdg_partition_for_component): Rename to ...
5156         (build_rdg_partition_for_vertex): ... this and adjust.
5157         (rdg_build_partitions): Take a vector of starting vertices
5158         instead of components.  Remove unnecessary leftover handling.
5159         (ldist_gen): Do not build components or record other stores.
5160         (distribute_loop): Do not distribute loops containing stmts
5161         with side-effects.
5162
5163 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5164
5165         PR target/58314
5166         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5167
5168 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5169
5170         * config.gcc: Separate cases for mingw and cygwin targets,
5171         and add 64-bit cygwin target case.
5172
5173         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5174         dll-export inline-functions.
5175         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5176
5177
5178 2013-09-13  Jeff Law  <law@redhat.com>
5179
5180         PR middle-end/58387
5181         Revert:
5182         2013-09-06  Jeff Law  <law@redhat.com>
5183
5184         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5185         edge implied equivalences into successor phis.
5186
5187 2013-09-12  DJ Delorie  <dj@redhat.com>
5188
5189         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5190         separators.
5191
5192 2013-09-12  Brooks Moses  <bmoses@google.com>
5193
5194         PR driver/42955
5195         * Makefile.in: Do not install driver binaries in $(target)/bin.
5196
5197 2013-09-12  DJ Delorie  <dj@redhat.com>
5198
5199         * config/rl78/rl78.opt (mrelax): New.
5200         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5201         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5202
5203         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5204         between SP and FP.
5205         (rl78_expand_epilogue): Likewise.
5206
5207 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5208
5209         PR middle-end/58335
5210         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5211         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5212         pointer elimination with using remove_reg_equal_offset_note.
5213
5214 2013-09-12  DJ Delorie  <dj@redhat.com>
5215
5216         * config/msp430/: New port.
5217         * config.gcc (msp430): Added.
5218         * doc/invoke.texi: Document MSP430 options.
5219         * doc/install.texi: Document msp430-elf
5220         * doc/md.texi: Document msp430-elf
5221         * doc/contrib.texi: Document msp430-elf
5222
5223         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5224         PSImode.
5225
5226 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5227
5228         PR ipa/58389
5229         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5230         reference descriptor is NULL.
5231         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5232         edge to NULL.
5233
5234 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5235
5236         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5237         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5238         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5239         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5240         make_temp_ssa_name): move to tree-ssanames.h
5241         * tree-ssa-alias.h: Move prototype.
5242         * tree-ssa.h: Include tree-ssanames.h.
5243         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5244         * tree-ssanames.h: New.  Move items from tree-flow*.h
5245         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5246
5247 2013-09-12  Richard Biener  <rguenther@suse.de>
5248
5249         PR tree-optimization/58404
5250         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5251         propagate non-invariant addresses into dereferences wrapped
5252         in component references.
5253
5254 2013-09-12  Richard Biener  <rguenther@suse.de>
5255
5256         PR tree-optimization/58402
5257         * passes.def: Move pass_late_warn_uninitialized later.
5258
5259 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5260
5261         * tree-ssa.h: New.  Move content from tree-flow.h and
5262         tree-flow-inline.h.
5263         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5264         Move prototypes belonging to tree-ssa.c.
5265         * tree-flow-inline.h (redirect_edge_var_map_def,
5266         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5267         tree-ssa.h.
5268         * gimple.h: Adjust prototypes.
5269         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5270         to...
5271         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5272         * tree.h: Move prototype to tree-ssa.h.
5273         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5274         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5275         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5276         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5277         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5278         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5279         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5280         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5281         graphite-blocking.c, graphite-clast-to-gimple.c,
5282         graphite-dependences.c, graphite-interchange.c,
5283         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5284         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5285         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5286         ipa-reference.c, ipa-split.c, ipa-utils.c,
5287         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5288         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5289         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5290         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5291         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5292         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5293         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5294         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5295         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5296         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5297         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5298         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5299         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5300         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5301         value-prof.c, var-tracking.c,
5302         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5303
5304 2013-09-12  Richard Biener  <rguenther@suse.de>
5305
5306         PR tree-optimization/58396
5307         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5308         (build_rdg): Take a loop-nest parameter, fix memleaks.
5309         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5310
5311 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5312
5313         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5314         for SLM.
5315
5316 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5317
5318         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5319         The FMA instruction names should have a 'v' prefix.
5320
5321 2013-09-12  Richard Biener  <rguenther@suse.de>
5322
5323         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5324         (dot_rdg): Use popen instead of system in optional code.
5325         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5326         (already_processed_vertex_p): Adjust.
5327         (has_anti_or_output_dependence, predecessor_has_mem_write,
5328         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5329         rdg_flag_uses): Remove.
5330         (rdg_flag_vertex): Simplify.
5331         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5332         remove recursion.
5333         (build_rdg_partition_for_component): Process the first vertex
5334         of a component only.
5335         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5336
5337 2013-09-12  Alan Modra  <amodra@gmail.com>
5338
5339         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5340
5341 2013-09-11  DJ Delorie  <dj@redhat.com>
5342             Nick Clifton  <nickc@redhat.com>
5343
5344         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5345         (rl78_stack_based_mem): New.
5346         * config/rl78/constraints.md (Iv08): New.
5347         (Iv16): New.
5348         (Iv24): New.
5349         (Is09): New.
5350         (Is17): New.
5351         (Is25): New.
5352         (ISsi): New.
5353         (IShi): New.
5354         (ISqi): New.
5355         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5356         (movhi): Likewise.
5357         (movsi): Change from expand to insn-and-split.
5358         (ashrsi3): Clobber AX.
5359         (lshrsi3): New.
5360         (ashlsi3): New.
5361         (cbranchsi4): New.
5362         * config/rl78/rl78.md (CC_REG): Fix.
5363         (addsi3): Allow memory and immediate operands.
5364         (addsi3_internal): Split into...
5365         (addsi3_internal_virt): ...new, and ...
5366         (addsi3_internal_real): ...new.
5367         (subsi): New.
5368         (subsi3_internal_virt): New.
5369         (subsi3_internal_real): New.
5370         (mulsi3): Add memory operand.
5371         (mulsi3_rl78): Likewise.
5372         (mulsi3_g13): Likewise.
5373         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5374         (cbranchqi4_real): Add more constraint options.
5375         (cbranchhi4_real): Expand pattern.
5376         (cbranchhi4_real_signed): New.
5377         (cbranchhi4_real_inverted): New.
5378         (cbranchsi4_real_lt): New.
5379         (cbranchsi4_real_ge): New.
5380         (cbranchsi4_real_signed): New.
5381         (cbranchsi4_real): New.
5382         (peephole2): New.
5383         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5384         constant shifts.
5385         (lshrsi3_virt): Likewise.
5386         (ashlsi3_virt): Likewise.
5387         (cbranchqi4_virt_signed): New.
5388         (cbranchhi4_virt_signed): New.
5389         (cbranchsi4_virt): New.
5390         * config/rl78/rl78.c: Whitespace fixes throughout.
5391         (move_elim_pass): New.
5392         (pass_data_rl78_move_elim): New.
5393         (pass_rl78_move_elim): New.
5394         (make_pass_rl78_move_elim): New.
5395         (rl78_devirt_info): Run devirt earlier.
5396         (rl78_move_elim_info): New.
5397         (rl78_asm_file_start): Register it.
5398         (rl78_split_movsi): New.
5399         (rl78_as_legitimate_address): Allow virtual base registers when
5400         appropriate.
5401         (rl78_addr_space_convert): Remove spurious debug stuff.
5402         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5403         (rl78_print_operand): More cases for not printing '#'.
5404         (rl78_expand_compare): Remove most of the logic.
5405         (content_memory): New.
5406         (clear_content_memory): New.
5407         (get_content_index): New.
5408         (get_content_name): New.
5409         (display_content_memory): New.
5410         (update_content): New.
5411         (record_content): New.
5412         (already_contains): New.
5413         (insn_ok_now): Re-recog insns with virtual registers.
5414         (add_postponed_content_update): New.
5415         (process_postponed_content_update): New.
5416         (gen_and_emit_move): New.
5417         (transcode_memory_rtx): Record new location content.
5418         Use gen_and_emit_move.
5419         (force_into_acc): New.
5420         (move_to_acc): Use gen_and_emit_move.
5421         (move_from_acc): Likewise.
5422         (move_acc_to_reg): Likewise.
5423         (move_to_x): Likewise.
5424         (move_to_hl): Likewise.
5425         (move_to_de): Likewise.
5426         (rl78_alloc_physical_registers_op1): Record location content.
5427         (has_constraint): New.
5428         (rl78_alloc_physical_registers_op2): Record location content.
5429         Optimize use of HL.
5430         (rl78_alloc_physical_registers_ro1): Likewise.
5431         (rl78_alloc_physical_registers_cmp): Likewise.
5432         (rl78_alloc_physical_registers_umul): Likewise.
5433         (rl78_alloc_address_registers_macax): New.
5434         (rl78_alloc_physical_registers): Initialize and set location
5435         content memory as needed.
5436         (rl78_reorg): Make sure split2 is called.
5437         (rl78_rtx_costs): New.
5438
5439 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5440
5441         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5442         for (not (neg ...)) and (neg (not ...)) cases.
5443
5444 2013-09-11  Richard Biener  <rguenther@suse.de>
5445
5446         PR middle-end/58377
5447         * passes.def: Split critical edges before late uninit warning passes.
5448         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5449
5450 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5451
5452         PR tree-optimization/58385
5453         * fold-const.c (build_range_check): If both low and high are NULL,
5454         use omit_one_operand_loc to preserve exp side-effects.
5455
5456 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5457
5458         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5459
5460 2013-09-11  Richard Biener  <rguenther@suse.de>
5461
5462         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5463         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5464         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5465         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5466         stmts_from_loop, known_dependences_p, build_empty_rdg,
5467         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5468         * tree-loop-distribution.c: ... here.
5469         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5470         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5471         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5472         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5473         * tree-loop-distribution.c: ... here.
5474         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5475         (struct partition_s): Add loops member.
5476         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5477         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5478         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5479
5480 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5481             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5482             Sergey Lega  <sergey.s.lega@intel.com>
5483             Anna Tikhonova  <anna.tikhonova@intel.com>
5484             Ilya Tocar  <ilya.tocar@intel.com>
5485             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5486             Ilya Verbin  <ilya.verbin@intel.com>
5487             Kirill Yukhin  <kirill.yukhin@intel.com>
5488             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5489
5490         * config/i386/constraints.md (k): New.
5491         (Yk): Ditto.
5492         * config/i386/i386.c (const regclass_map): Add new mask registers.
5493         (dbx_register_map): Ditto.
5494         (dbx64_register_map): Ditto.
5495         (svr4_dbx_register_map): Ditto.
5496         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5497         disabled.
5498         (ix86_preferred_reload_class): Disable constants for mask registers.
5499         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5500         (ix86_hard_regno_mode_ok): Support new mask registers.
5501         (x86_order_regs_for_local_alloc): Ditto.
5502         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5503         (FIXED_REGISTERS): Add new mask registers.
5504         (CALL_USED_REGISTERS): Ditto.
5505         (REG_ALLOC_ORDER): Ditto.
5506         (VALID_MASK_REG_MODE): New.
5507         (FIRST_MASK_REG): Ditto.
5508         (LAST_MASK_REG): Ditto.
5509         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5510         (MAYBE_MASK_CLASS_P): New.
5511         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5512         (REG_CLASS_CONTENTS): Ditto.
5513         (MASK_REGNO_P): New.
5514         (ANY_MASK_REG_P): Ditto.
5515         (HI_REGISTER_NAMES): Add new mask registers.
5516         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5517         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5518         mask registers.
5519         (attribute "type"): Add mskmov, msklog.
5520         (attribute "length_immediate"): Support them.
5521         (attribute "memory"): Ditto.
5522         (attribute "prefix_0f"): Ditto.
5523         (*movhi_internal): Support new mask registers.
5524         (*movqi_internal): Ditto.
5525         (define_split): Split out clobber pattern is a logic
5526         insn on mask registers.
5527         (*k<logic><mode>): New.
5528         (*andhi_1): Extend to support mask regs.
5529         (*andqi_1): Extend to support mask regs.
5530         (kandn<mode>): New.
5531         (define_split): Split and-not to and and not if operands
5532         are not mask regs.
5533         (*<code><mode>_1): Separate HI mode to new pattern...
5534         (*<code>hi_1): This.
5535         (*<code>qi_1): Extend to support mask regs.
5536         (kxnor<mode>): New.
5537         (kortestzhi): Ditto.
5538         (kortestchi): Ditto.
5539         (kunpckhi): Ditto.
5540         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5541         (*one_cmplhi2_1): ...Here, now with mask registers support.
5542         (*one_cmplqi2_1): Support new mask registers.
5543         (HI/QImode arithmetics splitter): Don't split if mask registers
5544         are used.
5545         (HI/QImode not splitter): Ditto.
5546         * config/i386/predicated.md (mask_reg_operand): New.
5547         (general_reg_operand): Ditto.
5548
5549 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5550
5551         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5552         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5553
5554 2013-09-10  Jeff Law  <law@redhat.com>
5555
5556         PR tree-optimization/58380
5557         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5558         of threading through a buried loop header.
5559
5560         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5561         return value for single successor case.
5562
5563 2013-09-10  Jan Hubicka  <jh@suse.cz>
5564
5565         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5566
5567 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5568
5569         PR target/58361
5570         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5571         support conditional execution.
5572         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5573
5574 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5575
5576         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5577         constraint pass.
5578         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5579         Check destination too to check move insn.
5580         (undo_optional_reloads): Add check that the original peudo did not
5581         changed its allocation and the optional reload was inherited on last
5582         inheritance pass.  Break loop after deciding to keep optional reload.
5583         (lra_undo_inheritance): Add check that inherited pseudo still in
5584         memory.
5585
5586 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5587
5588         * config/aarch64/aarch64.md (generic_sched): New.
5589         * config/aarch64/aarch64-generic.md (load): Make conditional
5590         on generic_sched attribute.
5591         (nonload): Likewise.
5592
5593 2013-09-10  Jan Hubicka  <jh@suse.cz>
5594
5595         * lto-cgraph.c: Include ipa-utils.h.
5596         (compute_ltrans_boundary): Also add possible targets into the boundary.
5597
5598 2013-09-10  Jan Hubicka  <jh@suse.cz>
5599
5600         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5601         VAR_DECL of vtable rather than full expression.
5602
5603 2013-09-10  Jan Hubicka  <jh@suse.cz>
5604             Paolo Carlini  <paolo.carlini@oracle.com>
5605
5606         * cgraphunit.c (analyze_functions): Save input_location, set it
5607         to UNKNOWN_LOCATION and restore it at the end.
5608
5609 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5610
5611         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5612         represented by a thunk.
5613
5614 2013-09-10  Jeff Law  <law@redhat.com>
5615
5616         PR tree-optimization/58343
5617         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5618         jump threading requests through loop headers buried in the middle
5619         of a jump threading path.
5620
5621         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5622         in return value/type.
5623
5624 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5625
5626         PR rtl-optimization/58365
5627         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5628         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5629         it differs.
5630
5631 2013-09-10  Richard Biener  <rguenther@suse.de>
5632
5633         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5634         * tree-data-ref.c (create_rdg_vertices): Collect all data
5635         references, signal failure to the caller, use data-ref API.
5636         (build_rdg): Compute data references only once.  Maintain lifetime
5637         of data references and data dependences from within RDG.
5638         (free_rdg): Free dependence relations.
5639         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5640         inventing extra dependences.
5641         (distribute_loop): Update for RDG API changes.
5642
5643 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5644
5645         * doc/invoke.texi (fms-extensions): Document changed
5646         behavior for ms-abi targets.
5647         * config/i386/i386.c (ix86_option_override_internal):
5648         Set default value of option -fms-extension for ms-abi targets.
5649
5650 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5651
5652         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5653
5654 2013-09-10  Alan Modra  <amodra@gmail.com>
5655
5656         PR target/58330
5657         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5658
5659 2013-09-10  Alan Modra  <amodra@gmail.com>
5660
5661         * config/rs6000/predicates.md (add_cint_operand): New.
5662         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5663         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5664         using add_cint_operand.
5665         (largetoc_high_plus_aix): Likewise.
5666
5667 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5668
5669         PR tree-optimization/58364
5670         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5671         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5672         the current range can't be an unconditional true or false.
5673
5674 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5675
5676         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5677
5678 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5679
5680         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5681
5682 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5683
5684         PR c++/43452
5685         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5686
5687 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5688
5689         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5690         NO_REGS for immediate that can't be moved directly into FP_REGS.
5691
5692 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5693
5694         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5695         comparison with negated operand.
5696         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5697         RTL form.
5698
5699 2013-09-09  Richard Biener  <rguenther@suse.de>
5700
5701         PR middle-end/58326
5702         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5703         of a subloop record all its block as affecting loop-closed SSA form.
5704
5705 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5706
5707         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5708         of an rtx/bitpos pair.
5709         (store_fixed_bit_field): Update accordingly.
5710
5711 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5712
5713         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5714         GEN_INT.
5715         * builtins.c (expand_errno_check): Likewise.
5716         * dwarf2cfi.c (init_return_column_size): Likewise.
5717         * except.c (sjlj_mark_call_sites): Likewise.
5718         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
5719         * lra-constraints.c (emit_inc): Likewise.
5720         * ree.c (combine_set_extension): Likewise.
5721         * regmove.c (fixup_match_2): Likewise.
5722         * reload1.c (inc_for_reload): Likewise.
5723
5724 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5725
5726         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
5727         (force_to_mode, simplify_shift_const_1, simplify_comparison):
5728         Use gen_int_mode with the mode of the associated simplify_* call.
5729         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
5730         * expmed.c (expand_shift_1): Likewise.
5731         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5732         * loop-iv.c (iv_number_of_iterations): Likewise.
5733         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5734         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
5735
5736 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5737
5738         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
5739         of the associated expand_* call.
5740         (asan_emit_stack_protection): Likewise.
5741         * builtins.c (round_trampoline_addr): Likewise.
5742         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
5743         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
5744         (emit_store_flag): Likewise.
5745         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
5746         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
5747         Likewise.
5748         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5749         * ifcvt.c (noce_try_store_flag_constants): Likewise.
5750         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5751         * modulo-sched.c (generate_prolog_epilog): Likewise.
5752         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
5753         (expand_ctz, expand_ffs, expand_unop): Likewise.
5754
5755 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5756
5757         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
5758         of the associated gen_rtx_* call.
5759         * caller-save.c (init_caller_save): Likewise.
5760         * combine.c (find_split_point, make_extraction): Likewise.
5761         (make_compound_operation): Likewise.
5762         * dwarf2out.c (mem_loc_descriptor): Likewise.
5763         * explow.c (plus_constant, probe_stack_range): Likewise.
5764         * expmed.c (expand_mult_const): Likewise.
5765         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
5766         * reload1.c (init_reload): Likewise.
5767         * valtrack.c (cleanup_auto_inc_dec): Likewise.
5768         * var-tracking.c (adjust_mems): Likewise.
5769         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
5770         rather than gen_rtx_fmt_ee.
5771
5772 2013-09-09  Jan Hubicka  <jh@suse.cz>
5773
5774         PR middle-end/58294
5775         * value-prof.c (gimple_ic): Copy also abnormal edges.
5776
5777 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5778
5779         * asan.c (asan_shadow_cst): Use gen_int_mode.
5780
5781 2013-09-08  Jan Hubicka  <jh@suse.cz>
5782
5783         * ipa-profile.c: Add toplevel comment.
5784         (ipa_propagate_frequency_1): Be more conservative when profile is read.
5785         (contains_hot_call_p): New function.
5786         (ipa_propagate_frequency): Set frequencies based on counts when
5787         profile is read.
5788         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
5789         profile; do not tamper with profile after inlining if it is read.
5790
5791 2013-09-08  Jan Hubicka  <jh@suse.cz>
5792
5793         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
5794         speculative edges.
5795
5796 2013-09-08  Jan Hubicka  <jh@suse.cz>
5797
5798         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
5799         summary generation.
5800
5801 2013-09-08  Jeff Law  <law@redhat.com>
5802
5803         PR bootstrap/58340
5804         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
5805         of 'found'.
5806
5807 2013-09-08  Andi Kleen  <ak@linux.intel.com>
5808
5809         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
5810
5811 2013-09-08  Jan Hubicka  <jh@suse.cz>
5812
5813         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
5814
5815 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
5816
5817         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
5818         for non-debug insns.
5819         * lra.c (new_insn_reg): Take the containing insn as a parameter.
5820         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
5821         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
5822         accordingly.
5823
5824 2013-09-08  Jan Hubicka  <jh@suse.cz>
5825
5826         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
5827         targets and devirtualize to BUILT_IN_UNREACHABLE.
5828         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
5829         * ipa.c (walk_polymorphic_call_targets): New function.
5830         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
5831         functions; use the new timevar.
5832         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
5833         was removed from the program.
5834         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
5835         not consider it in the walk when its vtable is dead.
5836         (possible_polymorphic_call_targets_1): Pass anonymous flag to
5837         record_binfo.
5838         (devirt_variable_node_removal_hook): New function.
5839         (possible_polymorphic_call_targets): Also register
5840         devirt_variable_node_removal_hook.
5841         (ipa_devirt): Do not do non-speculative devirtualization.
5842         (gate_ipa_devirt): One execute if devirtualizing speculatively.
5843
5844 2013-09-08  Jan Hubicka  <jh@suse.cz>
5845
5846         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
5847         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
5848         varpool_remove_variable_insertion_hook): Declare.
5849         * varpool.c (varpool_node_hook_list): New structure.
5850         (first_varpool_node_removal_hook,
5851         first_varpool_variable_insertion_hook): New variables.
5852         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
5853         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
5854         varpool_remove_variable_insertion_hook,
5855         varpool_call_variable_insertion_hooks): New functions.
5856         (varpool_remove_node): Use it.
5857
5858 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
5859
5860         PR c++/54941
5861         * diagnostic.c (diagnostic_build_prefix): When s.file is
5862         "<built-in>" don't output line and column numbers.
5863
5864 2013-09-06  Jan Hubicka  <jh@suse.cz>
5865
5866         * cgraphunit.c (expand_thunk): Get body before touching arguments.
5867         * lto-streamer-out.c: Stream thunks, too.
5868         * lto-streamer-in.c (input_function): Pop cfun here
5869         (lto_read_body): Instead of here.
5870
5871 2013-09-06  Caroline Tice  <cmtice@google.com>
5872
5873         * doc/install.texi: Add documentation for the --enable-vtable-verify
5874         and the --disable-libvtv configure options.
5875
5876 2013-09-06  Jeff Law  <law@redhat.com>
5877
5878         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5879         edge implied equivalences into successor phis.
5880
5881 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
5882
5883         * resource.c (mark_referenced_resources): Handle COND_EXEC.
5884
5885 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
5886
5887         * resource.c (mark_target_live_regs): Compute resources taking
5888         into account if a call is predicated or not.
5889
5890 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
5891
5892         * toplev.c (output_stack_usage): Be prepared for suffixes created by
5893         the compiler in the function names.
5894
5895 2013-09-06  Jan Hubicka  <jh@suse.cz>
5896
5897         PR middle-end/58094
5898         * ipa-inline.c (has_caller_p): New function.
5899         (want_inline_function_to_all_callers_p): Use it.
5900         (sum_callers, inline_to_all_callers): Break out from ...
5901         (ipa_inline): ... here.
5902
5903 2013-09-06  Jan Hubicka  <jh@suse.cz>
5904
5905         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
5906         only when AVX is enabled.
5907
5908 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5909
5910         * config/aarch64/aarch64.md
5911         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
5912         is fpsimd_<load/store>2.
5913         (load_pair<mode>): Likewise.
5914         (store_pair<mode>): Likewise.
5915
5916 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5917
5918         * config/arm/types.md (type): Add "mrs" type.
5919         * config/aarch64/aarch64.md
5920         (aarch64_load_tp_hard): Make type "mrs".
5921         * config/arm/arm.md
5922         (load_tp_hard): Make type "mrs".
5923         * config/arm/cortex-a15.md: Update with new attributes.
5924         * config/arm/cortex-a5.md: Update with new attributes.
5925         * config/arm/cortex-a53.md: Update with new attributes.
5926         * config/arm/cortex-a7.md: Update with new attributes.
5927         * config/arm/cortex-a8.md: Update with new attributes.
5928         * config/arm/cortex-a9.md: Update with new attributes.
5929         * config/arm/cortex-m4.md: Update with new attributes.
5930         * config/arm/cortex-r4.md: Update with new attributes.
5931         * config/arm/fa526.md: Update with new attributes.
5932         * config/arm/fa606te.md: Update with new attributes.
5933         * config/arm/fa626te.md: Update with new attributes.
5934         * config/arm/fa726te.md: Update with new attributes.
5935
5936 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5937
5938         * config/aarch64/aarch64.md
5939         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
5940         (*movtf_aarch64): Likewise.
5941         * config/arm/arm.md
5942         (thumb1_movdi_insn): Use "multiple" for type where more than one
5943         instruction is used for a move.
5944         (*arm32_movhf): Likewise.
5945         (*thumb_movdf_insn): Likewise.
5946
5947 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5948
5949         * config/arm/types.md (type): Rename fcpys to fmov.
5950         * config/arm/vfp.md
5951         (*arm_movsi_vfp): Rename type fcpys as fmov.
5952         (*thumb2_movsi_vfp): Likewise
5953         (*movhf_vfp_neon): Likewise
5954         (*movhf_vfp): Likewise
5955         (*movsf_vfp): Likewise
5956         (*thumb2_movsf_vfp): Likewise
5957         (*movsfcc_vfp): Likewise
5958         (*thumb2_movsfcc_vfp): Likewise
5959         * config/aarch64/aarch64-simd.md
5960         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
5961         * config/aarch64/aarch64.md
5962         (*movsi_aarch64): Replace type mov_reg with fmovs.
5963         (*movdi_aarch64): Likewise
5964         (*movsf_aarch64): Likewise
5965         (*movdf_aarch64): Likewise
5966         * config/arm/arm.c
5967         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
5968         * config/arm/iwmmxt.md
5969         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
5970         * config/arm/arm1020e.md: Update with new attributes.
5971         * config/arm/cortex-a15-neon.md: Update with new attributes.
5972         * config/arm/cortex-a5.md: Update with new attributes.
5973         * config/arm/cortex-a53.md: Update with new attributes.
5974         * config/arm/cortex-a7.md: Update with new attributes.
5975         * config/arm/cortex-a8-neon.md: Update with new attributes.
5976         * config/arm/cortex-a9.md: Update with new attributes.
5977         * config/arm/cortex-m4-fpu.md: Update with new attributes.
5978         * config/arm/cortex-r4f.md: Update with new attributes.
5979         * config/arm/marvell-pj4.md: Update with new attributes.
5980         * config/arm/vfp11.md: Update with new attributes.
5981
5982 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5983
5984         * config/aarch64/aarch64.md
5985         (*madd<mode>): Fix type attribute.
5986         (*maddsi_uxtw): Likewise.
5987         (*msub<mode>): Likewise.
5988         (*msubsi_uxtw): Likewise.
5989         (<su_optab>maddsidi4): Likewise.
5990         (<su_optab>msubsidi4): Likewise.
5991
5992 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5993
5994         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
5995         * config/arm/arm.md (core_cycles): Remove fdiv.
5996         * config/arm/vfp.md:
5997         (*sqrtsf2_vfp): Update for attribute changes.
5998         (*sqrtdf2_vfp): Likewise.
5999         * config/aarch64/aarch64.md:
6000         (sqrt<mode>2): Update for attribute changes.
6001         * config/arm/arm1020e.md: Update with new attributes.
6002         * config/arm/cortex-a15-neon.md: Update with new attributes.
6003         * config/arm/cortex-a5.md: Update with new attributes.
6004         * config/arm/cortex-a53.md: Update with new attributes.
6005         * config/arm/cortex-a7.md: Update with new attributes.
6006         * config/arm/cortex-a8-neon.md: Update with new attributes.
6007         * config/arm/cortex-a9.md: Update with new attributes.
6008         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6009         * config/arm/cortex-r4f.md: Update with new attributes.
6010         * config/arm/marvell-pj4.md: Update with new attributes.
6011         * config/arm/vfp11.md: Update with new attributes.
6012
6013 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6014
6015         * config/arm/types.md
6016         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6017         * config/aarch64/aarch64.md
6018         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6019         new attributes.
6020         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6021         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6022         (float<GPI:mode><GPF:mode>2): Likewise.
6023         * config/arm/vfp.md
6024         (*truncsisf2_vfp): Update with new attributes.
6025         (*truncsidf2_vfp): Likewise.
6026         (fixuns_truncsfsi2): Likewise.
6027         (fixuns_truncdfsi2): Likewise.
6028         (*floatsisf2_vfp): Likewise.
6029         (*floatsidf2_vfp): Likewise.
6030         (floatunssisf2): Likewise.
6031         (floatunssidf2): Likewise.
6032         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6033         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6034         * config/arm/arm1020e.md: Update with new attributes.
6035         * config/arm/cortex-a15-neon.md: Update with new attributes.
6036         * config/arm/cortex-a5.md: Update with new attributes.
6037         * config/arm/cortex-a53.md: Update with new attributes.
6038         * config/arm/cortex-a7.md: Update with new attributes.
6039         * config/arm/cortex-a8-neon.md: Update with new attributes.
6040         * config/arm/cortex-a9.md: Update with new attributes.
6041         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6042         * config/arm/cortex-r4f.md: Update with new attributes.
6043         * config/arm/marvell-pj4.md: Update with new attributes.
6044         * config/arm/vfp11.md: Update with new attributes.
6045
6046 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6047
6048         * config/aarch64/arm_neon.h
6049         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6050         (vqtbx<1,2,3,4><q>_s8): Likewise.
6051
6052 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6053
6054         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6055         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6056         (add<mode>3): Add type attribute.
6057         (add<mode>3): Likewise.
6058         (usadd<mode>3): Likewise.
6059         (ssadd<mode>3): Likewise.
6060         (sub<mode>3): Likewise.
6061         (sub<mode>3): Likewise.
6062         (ussub<mode>3): Likewise.
6063         (sssub<mode>3): Likewise.
6064         (ssmulsa3): Likewise.
6065         (usmulusa3): Likewise.
6066         (arm_usatsihi): Likewise.
6067         * config/arm/vfp.md
6068         (*movdi_vfp): Add types for all instructions.
6069         (*movdi_vfp_cortexa8): Likewise.
6070         (*movhf_vfp_neon): Likewise.
6071         (*movhf_vfp): Likewise.
6072         (*movdf_vfp): Likewise.
6073         (*thumb2_movdf_vfp): Likewise.
6074         (*thumb2_movdfcc_vfp): Likewise.
6075         * config/arm/arm.md: Add type attribute to all insn patterns.
6076         (*thumb1_adddi3): Add type attribute.
6077         (*arm_adddi3): Likewise.
6078         (*adddi_sesidi_di): Likewise.
6079         (*adddi_zesidi_di): Likewise.
6080         (*thumb1_addsi3): Likewise.
6081         (addsi3_compare0): Likewise.
6082         (*addsi3_compare0_scratch): Likewise.
6083         (*compare_negsi_si): Likewise.
6084         (cmpsi2_addneg): Likewise.
6085         (*addsi3_carryin_<optab>): Likewise.
6086         (*addsi3_carryin_alt2_<optab>): Likewise.
6087         (*addsi3_carryin_clobercc_<optab>): Likewise.
6088         (*subsi3_carryin): Likewise.
6089         (*subsi3_carryin_const): Likewise.
6090         (*subsi3_carryin_compare): Likewise.
6091         (*subsi3_carryin_compare_const): Likewise.
6092         (*arm_subdi3): Likewise.
6093         (*thumb_subdi3): Likewise.
6094         (*subdi_di_zesidi): Likewise.
6095         (*subdi_di_sesidi): Likewise.
6096         (*subdi_zesidi_di): Likewise.
6097         (*subdi_sesidi_di): Likewise.
6098         (*subdi_zesidi_ze): Likewise.
6099         (thumb1_subsi3_insn): Likewise.
6100         (*arm_subsi3_insn): Likewise.
6101         (*anddi3_insn): Likewise.
6102         (*anddi_zesidi_di): Likewise.
6103         (*anddi_sesdi_di): Likewise.
6104         (*ne_zeroextracts): Likewise.
6105         (*ne_zeroextracts): Likewise.
6106         (*ite_ne_zeroextr): Likewise.
6107         (*ite_ne_zeroextr): Likewise.
6108         (*anddi_notdi_di): Likewise.
6109         (*anddi_notzesidi): Likewise.
6110         (*anddi_notsesidi): Likewise.
6111         (andsi_notsi_si): Likewise.
6112         (thumb1_bicsi3): Likewise.
6113         (*iordi3_insn): Likewise.
6114         (*iordi_zesidi_di): Likewise.
6115         (*iordi_sesidi_di): Likewise.
6116         (*thumb1_iorsi3_insn): Likewise.
6117         (*xordi3_insn): Likewise.
6118         (*xordi_zesidi_di): Likewise.
6119         (*xordi_sesidi_di): Likewise.
6120         (*arm_xorsi3): Likewise.
6121         (*andsi_iorsi3_no): Likewise.
6122         (*smax_0): Likewise.
6123         (*smax_m1): Likewise.
6124         (*arm_smax_insn): Likewise.
6125         (*smin_0): Likewise.
6126         (*arm_smin_insn): Likewise.
6127         (*arm_umaxsi3): Likewise.
6128         (*arm_uminsi3): Likewise.
6129         (*minmax_arithsi): Likewise.
6130         (*minmax_arithsi_): Likewise.
6131         (*satsi_<SAT:code>): Likewise.
6132         (arm_ashldi3_1bit): Likewise.
6133         (arm_ashrdi3_1bit): Likewise.
6134         (arm_lshrdi3_1bit): Likewise.
6135         (*arm_negdi2): Likewise.
6136         (*thumb1_negdi2): Likewise.
6137         (*arm_negsi2): Likewise.
6138         (*thumb1_negsi2): Likewise.
6139         (*negdi_extendsid): Likewise.
6140         (*negdi_zero_extend): Likewise.
6141         (*arm_abssi2): Likewise.
6142         (*thumb1_abssi2): Likewise.
6143         (*arm_neg_abssi2): Likewise.
6144         (*thumb1_neg_abss): Likewise.
6145         (one_cmpldi2): Likewise.
6146         (extend<mode>di2): Likewise.
6147         (*compareqi_eq0): Likewise.
6148         (*arm_extendhisi2addsi): Likewise.
6149         (*arm_movdi): Likewise.
6150         (*thumb1_movdi_insn): Likewise.
6151         (*arm_movt): Likewise.
6152         (*thumb1_movsi_insn): Likewise.
6153         (pic_add_dot_plus_four): Likewise.
6154         (pic_add_dot_plus_eight): Likewise.
6155         (tls_load_dot_plus_eight): Likewise.
6156         (*thumb1_movhi_insn): Likewise.
6157         (*thumb1_movsf_insn): Likewise.
6158         (*movdf_soft_insn): Likewise.
6159         (*thumb_movdf_insn): Likewise.
6160         (cbranchsi4_insn): Likewise.
6161         (cbranchsi4_scratch): Likewise.
6162         (*negated_cbranchsi4): Likewise.
6163         (*tbit_cbranch): Likewise.
6164         (*tlobits_cbranch): Likewise.
6165         (*tstsi3_cbranch): Likewise.
6166         (*cbranchne_decr1): Likewise.
6167         (*addsi3_cbranch): Likewise.
6168         (*addsi3_cbranch_scratch): Likewise.
6169         (*arm_cmpdi_insn): Likewise.
6170         (*arm_cmpdi_unsig): Likewise.
6171         (*arm_cmpdi_zero): Likewise.
6172         (*thumb_cmpdi_zero): Likewise.
6173         (*deleted_compare): Likewise.
6174         (*mov_scc): Likewise.
6175         (*mov_negscc): Likewise.
6176         (*mov_notscc): Likewise.
6177         (*cstoresi_eq0_thumb1_insn): Likewise.
6178         (cstoresi_nltu_thumb1): Likewise.
6179         (cstoresi_ltu_thu): Likewise.
6180         (thumb1_addsi3_addgeu): Likewise.
6181         (*arm_jump): Likewise.
6182         (*thumb_jump): Likewise.
6183         (*check_arch2): Likewise.
6184         (arm_casesi_internal): Likewise.
6185         (thumb1_casesi_dispatch): Likewise.
6186         (*arm_indirect_jump): Likewise.
6187         (*thumb1_indirect_jump): Likewise.
6188         (nop): Likewise.
6189         (*and_scc): Likewise.
6190         (*ior_scc): Likewise.
6191         (*compare_scc): Likewise.
6192         (*cond_move): Likewise.
6193         (*cond_arith): Likewise.
6194         (*cond_sub): Likewise.
6195         (*cmp_ite0): Likewise.
6196         (*cmp_ite1): Likewise.
6197         (*cmp_and): Likewise.
6198         (*cmp_ior): Likewise.
6199         (*ior_scc_scc): Likewise.
6200         (*ior_scc_scc_cmp): Likewise.
6201         (*and_scc_scc): Likewise.
6202         (*and_scc_scc_cmp): Likewise.
6203         (*and_scc_scc_nod): Likewise.
6204         (*negscc): Likewise.
6205         (movcond_addsi): Likewise.
6206         (movcond): Likewise.
6207         (*ifcompare_plus_move): Likewise.
6208         (*if_plus_move): Likewise.
6209         (*ifcompare_move_plus): Likewise.
6210         (*if_move_plus): Likewise.
6211         (*ifcompare_arith_arith): Likewise.
6212         (*if_arith_arith): Likewise.
6213         (*ifcompare_arith_move): Likewise.
6214         (*if_arith_move): Likewise.
6215         (*ifcompare_move_arith): Likewise.
6216         (*if_move_arith): Likewise.
6217         (*ifcompare_move_not): Likewise.
6218         (*if_move_not): Likewise.
6219         (*ifcompare_not_move): Likewise.
6220         (*if_not_move): Likewise.
6221         (*ifcompare_shift_move): Likewise.
6222         (*if_shift_move): Likewise.
6223         (*ifcompare_move_shift): Likewise.
6224         (*if_move_shift): Likewise.
6225         (*ifcompare_shift_shift): Likewise.
6226         (*ifcompare_not_arith): Likewise.
6227         (*ifcompare_arith_not): Likewise.
6228         (*if_arith_not): Likewise.
6229         (*ifcompare_neg_move): Likewise.
6230         (*if_neg_move): Likewise.
6231         (*ifcompare_move_neg): Likewise.
6232         (*if_move_neg): Likewise.
6233         (prologue_thumb1_interwork): Likewise.
6234         (*cond_move_not): Likewise.
6235         (*sign_extract_onebit): Likewise.
6236         (*not_signextract_onebit): Likewise.
6237         (stack_tie): Likewise.
6238         (align_4): Likewise.
6239         (align_8): Likewise.
6240         (consttable_end): Likewise.
6241         (consttable_1): Likewise.
6242         (consttable_2): Likewise.
6243         (consttable_4): Likewise.
6244         (consttable_8): Likewise.
6245         (consttable_16): Likewise.
6246         (*thumb1_tablejump): Likewise.
6247         (prefetch): Likewise.
6248         (force_register_use): Likewise.
6249         (thumb_eh_return): Likewise.
6250         (load_tp_hard): Likewise.
6251         (load_tp_soft): Likewise.
6252         (tlscall): Likewise.
6253         (*arm_movtas_ze): Likewise.
6254         (*arm_rev): Likewise.
6255         (*arm_revsh): Likewise.
6256         (*arm_rev16): Likewise.
6257         * config/arm/thumb2.md
6258         (*thumb2_smaxsi3): Likewise.
6259         (*thumb2_sminsi3): Likewise.
6260         (*thumb32_umaxsi3): Likewise.
6261         (*thumb2_uminsi3): Likewise.
6262         (*thumb2_negdi2): Likewise.
6263         (*thumb2_abssi2): Likewise.
6264         (*thumb2_neg_abss): Likewise.
6265         (*thumb2_movsi_insn): Likewise.
6266         (tls_load_dot_plus_four): Likewise.
6267         (*thumb2_movhi_insn): Likewise.
6268         (*thumb2_mov_scc): Likewise.
6269         (*thumb2_mov_negs): Likewise.
6270         (*thumb2_mov_negs): Likewise.
6271         (*thumb2_mov_nots): Likewise.
6272         (*thumb2_mov_nots): Likewise.
6273         (*thumb2_movsicc_): Likewise.
6274         (*thumb2_movsfcc_soft_insn): Likewise.
6275         (*thumb2_indirect_jump): Likewise.
6276         (*thumb2_and_scc): Likewise.
6277         (*thumb2_ior_scc): Likewise.
6278         (*thumb2_ior_scc_strict_it): Likewise.
6279         (*thumb2_cond_move): Likewise.
6280         (*thumb2_cond_arith): Likewise.
6281         (*thumb2_cond_ari): Likewise.
6282         (*thumb2_cond_sub): Likewise.
6283         (*thumb2_negscc): Likewise.
6284         (*thumb2_movcond): Likewise.
6285         (thumb2_casesi_internal): Likewise.
6286         (thumb2_casesi_internal_pic): Likewise.
6287         (*thumb2_alusi3_short): Likewise.
6288         (*thumb2_mov<mode>_shortim): Likewise.
6289         (*thumb2_addsi_short): Likewise.
6290         (*thumb2_subsi_short): Likewise.
6291         (thumb2_addsi3_compare0): Likewise.
6292         (*thumb2_cbz): Likewise.
6293         (*thumb2_cbnz): Likewise.
6294         (*thumb2_one_cmplsi2_short): Likewise.
6295         (*thumb2_negsi2_short): Likewise.
6296         (*orsi_notsi_si): Likewise.
6297         * config/arm/arm1020e.md: Update with new attributes.
6298         * config/arm/arm1026ejs.md: Update with new attributes.
6299         * config/arm/arm1136jfs.md: Update with new attributes.
6300         * config/arm/arm926ejs.md: Update with new attributes.
6301         * config/arm/cortex-a15.md: Update with new attributes.
6302         * config/arm/cortex-a5.md: Update with new attributes.
6303         * config/arm/cortex-a53.md: Update with new attributes.
6304         * config/arm/cortex-a7.md: Update with new attributes.
6305         * config/arm/cortex-a8.md: Update with new attributes.
6306         * config/arm/cortex-a9.md: Update with new attributes.
6307         * config/arm/cortex-m4.md: Update with new attributes.
6308         * config/arm/cortex-r4.md: Update with new attributes.
6309         * config/arm/fa526.md: Update with new attributes.
6310         * config/arm/fa606te.md: Update with new attributes.
6311         * config/arm/fa626te.md: Update with new attributes.
6312         * config/arm/fa726te.md: Update with new attributes.
6313
6314 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6315
6316         * config/aarch64/aarch64-simd.md
6317         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6318         <vwx> iterator to ensure correct register choice.
6319         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6320         (aarch64_sqdmull_n<mode>): Likewise.
6321         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6322         * config/aarch64/arm_neon.h
6323         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6324         (vml<as><q>_n_<su>16): Likewise.
6325         (vml<as>l_high_lane<q>_<su>16): Likewise.
6326         (vml<as>l_high_n_<su>16): Likewise.
6327         (vml<as>l_lane<q>_<su>16): Likewise.
6328         (vml<as>l_n_<su>16): Likewise.
6329         (vmul<q>_lane<q>_<su>16): Likewise.
6330         (vmul<q>_n_<su>16): Likewise.
6331         (vmull_lane<q>_<su>16): Likewise.
6332         (vmull_n_<su>16): Likewise.
6333         (vmull_high_lane<q>_<su>16): Likewise.
6334         (vmull_high_n_<su>16): Likewise.
6335         (vqrdmulh<q>_n_s16): Likewise.
6336
6337 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6338
6339         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6340         have the correct lane parameter.
6341
6342 2013-09-06  Richard Biener <rguenther@suse.de>
6343
6344         * cfganal.c (control_dependences::~control_dependences):
6345         Properly free all of the vector.
6346
6347 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6348
6349         PR target/58269
6350         * config/i386/i386.c (ix86_conditional_register_usage):
6351         Proper initialize extended SSE registers.
6352
6353 2013-09-06  Jan Hubicka  <jh@suse.cz>
6354
6355         PR tree-optimization/58311
6356         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6357
6358 2013-09-06  Jan Hubicka  <jh@suse.cz>
6359
6360         * Makefile.in (tree-sra.o): Update dependencies.
6361         * tree-sra.c: Include ipa-utils.h
6362         (scan_function): Use recursive_call_p.
6363         (has_caller_p): New function.
6364         (cgraph_for_node_and_aliases): Count also callers of aliases.
6365
6366 2013-09-06  Jan Hubicka  <jh@suse.cz>
6367
6368         PR middle-end/58094
6369         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6370         * tree-tailcall.c: Include ipa-utils.h.
6371         (find_tail_calls): Use it.
6372         * ipa-pure-const.c (check_call): Likewise.
6373         * ipa-utils.c (recursive_call_p): New function.
6374         * ipa-utils.h (recursive_call_p): Dclare.
6375         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6376         (symtab_semantically_equivalent_p): New function.
6377         * Makefile.in (tree-tailcall.o): Update dependencies.
6378
6379 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6380
6381         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6382         non-inlinable part.
6383
6384 2013-09-06  Richard Biener  <rguenther@suse.de>
6385
6386         * lto-streamer.h (lto_global_var_decls): Remove.
6387         * Makefile.in (OBJS): Remove lto-symtab.o.
6388         (lto-symtab.o): Remove.
6389         (GTFILES): Remove lto-symtab.c
6390         * lto-symtab.c: Move to lto/
6391
6392 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6393
6394         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6395         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6396         (UNSPEC_FPINT_RINT): New constant definitions.
6397         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6398         definition with 2 attributes.
6399         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6400         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6401         definitions.
6402
6403 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6404
6405         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6406         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6407         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6408         attribute to "z196_alone".
6409         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6410         "zEC12_simple".
6411
6412 2013-09-06  Richard Biener  <rguenther@suse.de>
6413
6414         * basic-block.h (class control_dependences): New.
6415         * tree-ssa-dce.c (control_dependence_map): Remove.
6416         (cd): New global.
6417         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6418         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6419         find_pdom, find_control_dependence, find_all_control_dependences):
6420         Move to cfganal.c.
6421         (mark_control_dependent_edges_necessary,
6422         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6423         tree_dce_done, perform_tree_ssa_dce): Adjust.
6424         * cfganal.c (set_control_dependence_map_bit,
6425         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6426         find_all_control_dependences): Move from tree-ssa-dce.c and
6427         implement as methods of control_dependences class.
6428         (control_dependences::control_dependences): New.
6429         (control_dependences::~control_dependences): Likewise.
6430         (control_dependences::get_edges_dependent_on): Likewise.
6431         (control_dependences::get_edge): Likewise.
6432
6433 2013-09-04  Jan Hubicka  <jh@suse.cz>
6434
6435         * tree.c (types_same_for_odr): Drop overactive check.
6436         * ipa-devirt.c (hash_type_name): Likewise.
6437
6438 2013-09-04  Jan Hubicka  <jh@suse.cz>
6439
6440         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6441         (analyze_functions): ... here.
6442
6443 2013-09-04  Jan Hubicka  <jh@suse.cz>
6444
6445         PR middle-end/58201
6446         * cgraphunit.c (analyze_functions): Clear AUX fields
6447         after processing; initialize assembler name has.
6448
6449 2013-09-05  Jeff Law  <law@redhat.com>
6450
6451         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6452         from thread_around_empty_block.  Record threading path into PATH.
6453         Recurse if threading through the initial block is successful.
6454         (thread_across_edge): Corresponding changes to slightly simplify.
6455
6456 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6457
6458         * config/aarch64/aarch64.md
6459         (type): Remove frecpe, frecps, frecpx.
6460         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6461         fix to be a TARGET_SIMD instruction.
6462         (aarch64_frecps): Remove.
6463         * config/aarch64/aarch64-simd.md
6464         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6465         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6466
6467 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6468             Sofiane Naci  <sofiane.naci@arm.com>
6469
6470         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6471         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6472         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6473         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6474         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6475         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6476         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6477         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6478         into "clz, "rbit".  Rename "shift" to "shift_imm".
6479         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6480         changes.  Update for attribute changes all occurrences of arlo_* and
6481         shift* types.
6482         * config/arm/arm-fixed.md: Update for attribute changes
6483         all occurrences of arlo_* types.
6484         * config/arm/thumb2.md: Update for attribute changes all occurrences
6485         of arlo_* types.
6486         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6487         (cortexa7_older_only): Likewise.
6488         (cortexa7_younger):  Likewise.
6489         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6490         (1020alu_shift_op): Likewise.
6491         (1020alu_shift_reg_op): Likewise.
6492         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6493         (alu_shift_op): Likewise.
6494         (alu_shift_reg_op): Likewise.
6495         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6496         (11_alu_shift_op): Likewise.
6497         (11_alu_shift_reg_op): Likewise.
6498         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6499         (9_alu_shift_reg_op): Likewise.
6500         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6501         attribute changes.
6502         (cortex_a15_alu_shift): Likewise.
6503         (cortex_a15_alu_shift_reg): Likewise.
6504         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6505         attribute changes.
6506         (cortex_a5_alu_shift): Likewise.
6507         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6508         attribute changes.
6509         (cortex_a53_alu_shift): Likewise.
6510         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6511         attribute changes.
6512         (cortex_a7_alu_reg): Likewise.
6513         (cortex_a7_alu_shift): Likewise.
6514         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6515         attribute changes.
6516         (cortex_a8_alu_shift): Likewise.
6517         (cortex_a8_alu_shift_reg): Likewise.
6518         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6519         (cortex_a9_dp_shift): Likewise.
6520         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6521         attribute changes.
6522         * config/arm/cortex-r4.md
6523         (cortex_r4_alu): Update for attribute changes.
6524         (cortex_r4_mov): Likewise.
6525         (cortex_r4_alu_shift_reg): Likewise.
6526         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6527         (526_alu_shift_op): Likewise.
6528         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6529         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6530         (626te_alu_shift_op): Likewise.
6531         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6532         (726te_alu_shift_op): Likewise.
6533         (726te_alu_shift_reg_op): Likewise.
6534         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6535         (mp626_alu_shift_op): Likewise.
6536         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6537         (pj4_alu_conds): Likewise.
6538         (pj4_shift): Likewise.
6539         (pj4_shift_conds): Likewise.
6540         (pj4_alu_shift): Likewise.
6541         (pj4_alu_shift_conds): Likewise.
6542         * config/aarch64/aarch64.md: Update for attribute change
6543         all occurrences of arlo_* and shift* types.
6544
6545 2013-09-05  Mike Stump  <mikestump@comcast.net>
6546
6547         * tree.h: Move documentation for tree_function_decl to tree-core.h
6548         with the declaration.
6549
6550 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6551
6552         PR target/58139
6553         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6554         looking for widest mode.
6555
6556 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6557
6558         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6559
6560 2013-09-05  Richard Biener  <rguenther@suse.de>
6561
6562         PR tree-optimization/58137
6563         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6564         Do not create vectors of pointers.
6565         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6566         types for the components of the vector initializer.
6567         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6568         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6569
6570 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6571
6572         * ipa-prop.c (remove_described_reference): Accept missing references,
6573         return false if that hppens, otherwise return true.
6574         (cgraph_node_for_jfunc): New function.
6575         (try_decrement_rdesc_refcount): Likewise.
6576         (try_make_edge_direct_simple_call): Use them.
6577         (ipa_edge_removal_hook): Remove references from rdescs.
6578         (ipa_edge_duplication_hook): Clone rdescs and their references
6579         when the new edge has the same caller as the old one.
6580         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6581         reference before removing any edges.
6582
6583 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6584
6585         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6586         initial store.
6587         * thumb2.md (thumb2_storewb_parisi): New pattern.
6588
6589 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6590
6591         * config/aarch64/aarch64-option-extensions.def: Add
6592         AARCH64_OPT_EXTENSION of 'crc'.
6593         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6594         (AARCH64_ISA_CRC): Ditto.
6595         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6596         description of the CRC extension.
6597
6598 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6599
6600         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6601         * config/rs6000/linux.h: Ditto.
6602         * alpha/linux.h: Ditto.
6603         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6604         no_c99_libc_has_function.
6605         * config/c6x/uclinux-elf.h: Ditto.
6606         * config/lm32/uclinux-elf.h: Ditto.
6607         * config/m68k/uclinux.h: Ditto.
6608         * config/moxie/uclinux.h: Ditto.
6609         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6610         (crisv32-*-linux*, cris-*-linux*): Ditto.
6611         * config/bfin/bfin.c: Include "tm_p.h".
6612
6613 2013-09-05  Richard Biener  <rguenther@suse.de>
6614
6615         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6616         check for a definition without a basic-block.
6617
6618 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6619             Sofiane Naci <sofiane.naci@arm.com>
6620
6621         * config/aarch64/aarch64.md
6622         (*movti_aarch64): Rename r_2_f and f_2_r.
6623         (*movsf_aarch64): Likewise.
6624         (*movdf_aarch64): Likewise.
6625         (*movtf_aarch64): Likewise.
6626         (aarch64_movdi_<mode>low): Likewise.
6627         (aarch64_movdi_<mode>high): Likewise.
6628         (aarch64_mov<mode>high_di): Likewise.
6629         (aarch64_mov<mode>low_di): Likewise.
6630         (aarch64_movtilow_tilow): Likewise.
6631         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6632         values to config/arm/types.md
6633         (attribute "conds"): Update for attribute change.
6634         (anddi3_insn): Likewise.
6635         (iordi3_insn): Likewise.
6636         (xordi3_insn): Likewise.
6637         (one_cmpldi2): Likewise.
6638         * config/arm/types.md (type): Add Neon types.
6639         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6640         use "type" attribute.
6641         (movmisalign<mode>_neon_store): Likewise.
6642         (movmisalign<mode>_neon_load): Likewise.
6643         (vec_set<mode>_internal): Likewise.
6644         (vec_setv2di_internal): Likewise.
6645         (vec_extract<mode>): Likewise.
6646         (vec_extractv2di): Likewise.
6647         (add<mode>3_neon): Likewise.
6648         (adddi3_neon): Likewise.
6649         (sub<mode>3_neon): Likewise.
6650         (subdi3_neon): Likewise.
6651         (mul<mode>3_neon): Likewise.
6652         (mul<mode>3add<mode>_neon): Likewise.
6653         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6654         (fma<VCVTF:mode>4)): Likewise.
6655         (fma<VCVTF:mode>4_intrinsic): Likewise.
6656         (fmsub<VCVTF:mode>4)): Likewise.
6657         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6658         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6659         (ior<mode>3): Likewise.
6660         (and<mode>3): Likewise.
6661         (anddi3_neon): Likewise.
6662         (orn<mode>3_neon): Likewise.
6663         (orndi3_neon): Likewise.
6664         (bic<mode>3_neon): Likewise.
6665         (bicdi3_neon): Likewise.
6666         (xor<mode>3): Likewise.
6667         (one_cmpl<mode>2): Likewise.
6668         (abs<mode>2): Likewise.
6669         (neg<mode>2): Likewise.
6670         (umin<mode>3_neon): Likewise.
6671         (umax<mode>3_neon): Likewise.
6672         (smin<mode>3_neon): Likewise.
6673         (smax<mode>3_neon): Likewise.
6674         (vashl<mode>3): Likewise.
6675         (vashr<mode>3_imm): Likewise.
6676         (vlshr<mode>3_imm): Likewise.
6677         (ashl<mode>3_signed): Likewise.
6678         (ashl<mode>3_unsigned): Likewise.
6679         (neon_load_count): Likewise.
6680         (ashldi3_neon_noclobber): Likewise.
6681         (signed_shift_di3_neon): Likewise.
6682         (unsigned_shift_di3_neon): Likewise.
6683         (ashrdi3_neon_imm_noclobber): Likewise.
6684         (lshrdi3_neon_imm_noclobber): Likewise.
6685         (widen_ssum<mode>3): Likewise.
6686         (widen_usum<mode>3): Likewise.
6687         (quad_halves_<code>v4si): Likewise.
6688         (quad_halves_<code>v4sf): Likewise.
6689         (quad_halves_<code>v8hi): Likewise.
6690         (quad_halves_<code>v16qi): Likewise.
6691         (reduc_splus_v2di): Likewise.
6692         (neon_vpadd_internal<mode>): Likewise.
6693         (neon_vpsmin<mode>): Likewise.
6694         (neon_vpsmax<mode>): Likewise.
6695         (neon_vpumin<mode>): Likewise.
6696         (neon_vpumax<mode>): Likewise.
6697         (ss_add<mode>_neon): Likewise.
6698         (us_add<mode>_neon): Likewise.
6699         (ss_sub<mode>_neon): Likewise.
6700         (us_sub<mode>_neon): Likewise.
6701         (neon_vadd<mode>_unspec): Likewise.
6702         (neon_vaddl<mode>): Likewise.
6703         (neon_vaddw<mode>): Likewise.
6704         (neon_vhadd<mode>): Likewise.
6705         (neon_vqadd<mode>): Likewise.
6706         (neon_vaddhn<mode>): Likewise.
6707         (neon_vmul<mode>): Likewise.
6708         (neon_vmla<mode>): Likewise.
6709         (neon_vmlal<mode>): Likewise.
6710         (neon_vmls<mode>): Likewise.
6711         (neon_vmlsl<mode>): Likewise.
6712         (neon_vqdmulh<mode>): Likewise.
6713         (neon_vqdmlal<mode>): Likewise.
6714         (neon_vqdmlsl<mode>): Likewise.
6715         (neon_vmull<mode>): Likewise.
6716         (neon_vqdmull<mode>): Likewise.
6717         (neon_vsub<mode>_unspec): Likewise.
6718         (neon_vsubl<mode>): Likewise.
6719         (neon_vsubw<mode>): Likewise.
6720         (neon_vqsub<mode>): Likewise.
6721         (neon_vhsub<mode>): Likewise.
6722         (neon_vsubhn<mode>): Likewise.
6723         (neon_vceq<mode>): Likewise.
6724         (neon_vcge<mode>): Likewise.
6725         (neon_vcgeu<mode>): Likewise.
6726         (neon_vcgt<mode>): Likewise.
6727         (neon_vcgtu<mode>): Likewise.
6728         (neon_vcle<mode>): Likewise.
6729         (neon_vclt<mode>): Likewise.
6730         (neon_vcage<mode>): Likewise.
6731         (neon_vcagt<mode>): Likewise.
6732         (neon_vtst<mode>): Likewise.
6733         (neon_vabd<mode>): Likewise.
6734         (neon_vabdl<mode>): Likewise.
6735         (neon_vaba<mode>): Likewise.
6736         (neon_vabal<mode>): Likewise.
6737         (neon_vmax<mode>): Likewise.
6738         (neon_vmin<mode>): Likewise.
6739         (neon_vpaddl<mode>): Likewise.
6740         (neon_vpadal<mode>): Likewise.
6741         (neon_vpmax<mode>): Likewise.
6742         (neon_vpmin<mode>): Likewise.
6743         (neon_vrecps<mode>): Likewise.
6744         (neon_vrsqrts<mode>): Likewise.
6745         (neon_vqabs<mode>): Likewise.
6746         (neon_vqneg<mode>): Likewise.
6747         (neon_vcls<mode>): Likewise.
6748         (clz<mode>2): Likewise.
6749         (popcount<mode>2): Likewise.
6750         (neon_vrecpe): Likewise.
6751         (neon_vrsqrte): Likewise.
6752         (neon_vget_lane<mode>_sext_internal): Likewise.
6753         (neon_vget_lane<mode>_zext_internal): Likewise.
6754         (neon_vdup_n<mode>): Likewise.
6755         (neon_vdup_nv2di): Likewise.
6756         (neon_vdpu_lane<mode>_internal): Likewise.
6757         (neon_vswp<mode>): Likewise.
6758         (float<mode><V_cvtto>2): Likewise.
6759         (floatuns<mode><V_cvtto>2): Likewise.
6760         (fix_trunc<mode><V_cvtto>)2): Likewise
6761         (fixuns_trunc<mode><V_cvtto)2): Likewise.
6762         (neon_vcvt<mode>): Likewise.
6763         (neon_vcvtv4sfv4hf): Likewise.
6764         (neon_vcvtv4hfv4sf): Likewise.
6765         (neon_vcvt_n<mode>): Likewise.
6766         (neon_vmovn<mode>): Likewise.
6767         (neon_vqmovn<mode>): Likewise.
6768         (neon_vqmovun<mode>): Likewise.
6769         (neon_vmovl<mode>): Likewise.
6770         (neon_vmul_lane<mode>): Likewise.
6771         (neon_vmull_lane<mode>): Likewise.
6772         (neon_vqdmull_lane<mode>): Likewise.
6773         (neon_vqdmulh_lane<mode>): Likewise.
6774         (neon_vmla_lane<mode>): Likewise.
6775         (neon_vmlal_lane<mode>): Likewise.
6776         (neon_vqdmlal_lane<mode>): Likewise.
6777         (neon_vmls_lane<mode>): Likewise.
6778         (neon_vmlsl_lane<mode>): Likewise.
6779         (neon_vqdmlsl_lane<mode>): Likewise.
6780         (neon_vext<mode>): Likewise.
6781         (neon_vrev64<mode>): Likewise.
6782         (neon_vrev32<mode>): Likewise.
6783         (neon_vrev16<mode>): Likewise.
6784         (neon_vbsl<mode>_internal): Likewise.
6785         (neon_vshl<mode>): Likewise.
6786         (neon_vqshl<mode>): Likewise.
6787         (neon_vshr_n<mode>): Likewise.
6788         (neon_vshrn_n<mode>): Likewise.
6789         (neon_vqshrn_n<mode>): Likewise.
6790         (neon_vqshrun_n<mode>): Likewise.
6791         (neon_vshl_n<mode>): Likewise.
6792         (neon_vqshl_n<mode>): Likewise.
6793         (neon_vqshlu_n<mode>): Likewise.
6794         (neon_vshll_n<mode>): Likewise.
6795         (neon_vsra_n<mode>): Likewise.
6796         (neon_vsri_n<mode>): Likewise.
6797         (neon_vsli_n<mode>): Likewise.
6798         (neon_vtbl1v8qi): Likewise.
6799         (neon_vtbl2v8qi): Likewise.
6800         (neon_vtbl3v8qi): Likewise.
6801         (neon_vtbl4v8qi): Likewise.
6802         (neon_vtbx1v8qi): Likewise.
6803         (neon_vtbx2v8qi): Likewise.
6804         (neon_vtbx3v8qi): Likewise.
6805         (neon_vtbx4v8qi): Likewise.
6806         (neon_vtrn<mode>_internal): Likewise.
6807         (neon_vzip<mode>_internal): Likewise.
6808         (neon_vuzp<mode>_internal): Likewise.
6809         (neon_vld1<mode>): Likewise.
6810         (neon_vld1_lane<mode>): Likewise.
6811         (neon_vld1_dup<mode>): Likewise.
6812         (neon_vld1_dupv2di): Likewise.
6813         (neon_vst1<mode>): Likewise.
6814         (neon_vst1_lane<mode>): Likewise.
6815         (neon_vld2<mode>): Likewise.
6816         (neon_vld2_lane<mode>): Likewise.
6817         (neon_vld2_dup<mode>): Likewise.
6818         (neon_vst2<mode>): Likewise.
6819         (neon_vst2_lane<mode>): Likewise.
6820         (neon_vld3<mode>): Likewise.
6821         (neon_vld3qa<mode>): Likewise.
6822         (neon_vld3qb<mode>): Likewise.
6823         (neon_vld3_lane<mode>): Likewise.
6824         (neon_vld3_dup<mode>): Likewise.
6825         (neon_vst3<mode>): Likewise.
6826         (neon_vst3qa<mode>): Likewise.
6827         (neon_vst3qb<mode>): Likewise.
6828         (neon_vst3_lane<mode>): Likewise.
6829         (neon_vld4<mode>): Likewise.
6830         (neon_vld4qa<mode>): Likewise.
6831         (neon_vld4qb<mode>): Likewise.
6832         (neon_vld4_lane<mode>): Likewise.
6833         (neon_vld4_dup<mode>): Likewise.
6834         (neon_vst4<mode>): Likewise.
6835         (neon_vst4qa<mode>): Likewise.
6836         (neon_vst4qb<mode>): Likewise.
6837         (neon_vst4_lane<mode>): Likewise.
6838         (neon_vec_unpack<US>_lo_<mode>): Likewise.
6839         (neon_vec_unpack<US>_hi_<mode>): Likewise.
6840         (neon_vec_<US>mult_lo_<mode>): Likewise.
6841         (neon_vec_<US>mult_hi_<mode>): Likewise.
6842         (neon_vec_<US>shiftl_<mode>): Likewise.
6843         (neon_unpack<US>_<mode>): Likewise.
6844         (neon_vec_<US>mult_<mode>): Likewise.
6845         (vec_pack_trunc_<mode>): Likewise.
6846         (neon_vec_pack_trunk_<mode>): Likewise.
6847         (neon_vabd<mode>_2): Likewise.
6848         (neon_vabd<mode>_3): Likewise.
6849         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
6850         (thumb2_movsi_vfp): Likewise.
6851         (movdi_vfp): Likewise.
6852         (movdi_vfp_cortexa8): Likewise.
6853         (movhf_vfp_neon): Likewise.
6854         (movhf_vfp): Likewiwse.
6855         (movsf_vfp): Likewiwse.
6856         (thumb2_movsf_vfp): Likewiwse.
6857         (movdf_vfp): Likewise.
6858         (thumb2_movdf_vfp): Likewise.
6859         (movsfcc_vfp): Likewise.
6860         (thumb2_movsfcc_vfp): Likewise.
6861         (movdfcc_vfp): Likewise.
6862         (thumb2_movdfcc_vfp): Likewise.
6863         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
6864         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
6865         (v10_v2c): Likewise.
6866         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
6867         attribute change.
6868         (cortex_a15_neon_int_2): Likewise.
6869         (cortex_a15_neon_int_3): Likewise.
6870         (cortex_a15_neon_int_4): Likewise.
6871         (cortex_a15_neon_int_5): Likewise.
6872         (cortex_a15_neon_vqneg_vqabs): Likewise.
6873         (cortex_a15_neon_vmov): Likewise.
6874         (cortex_a15_neon_vaba): Likewise.
6875         (cortex_a15_neon_vaba_qqq): Likewise.
6876         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6877         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
6878         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
6879         scalar_64_32_long_scalar): Likewise.
6880         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6881         (cortex_a15_neon_mla_qqq_8_16): Likewise.
6882         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
6883         lotype_qdd_64_32_long): Likewise.
6884         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
6885         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
6886         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
6887         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
6888         (cortex_a15_neon_shift_1): Likewise.
6889         (cortex_a15_neon_shift_2): Likewise.
6890         (cortex_a15_neon_shift_3): Likewise.
6891         (cortex_a15_neon_vshl_ddd): Likewise.
6892         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
6893         (cortex_a15_neon_vsra_vrsra): Likewise.
6894         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
6895         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
6896         (cortex_a15_neon_fp_vmul_ddd): Likewise.
6897         (cortex_a15_neon_fp_vmul_qqd): Likewise.
6898         (cortex_a15_neon_fp_vmla_ddd): Likewise.
6899         (cortex_a15_neon_fp_vmla_qqq): Likewise.
6900         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
6901         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
6902         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
6903         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
6904         (cortex_a15_neon_bp_simple): Likewise.
6905         (cortex_a15_neon_bp_2cycle): Likewise.
6906         (cortex_a15_neon_bp_3cycle): Likewise.
6907         (cortex_a15_neon_vld1_1_2_regs): Likewise.
6908         (cortex_a15_neon_vld1_3_4_regs): Likewise.
6909         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
6910         (cortex_a15_neon_vld2_4_regs): Likewise.
6911         (cortex_a15_neon_vld3_vld4): Likewise.
6912         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
6913         (cortex_a15_neon_vst1_3_4_regs): Likewise.
6914         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
6915         (cortex_a15_neon_vst3_vst4): Likewise.
6916         (cortex_a15_neon_vld1_vld2_lane): Likewise.
6917         (cortex_a15_neon_vld3_vld4_lane" 10
6918         (cortex_a15_neon_vst1_vst2_lane): Likewise.
6919         (cortex_a15_neon_vst3_vst4_lane): Likewise.
6920         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
6921         (cortex_a15_neon_ldm_2): Likewise.0
6922         (cortex_a15_neon_stm_2): Likewise.
6923         (cortex_a15_neon_mcr): Likewise.
6924         (cortex_a15_neon_mcr_2_mcrr): Likewise.
6925         (cortex_a15_neon_mrc): Likewise.
6926         (cortex_a15_neon_mrrc): Likewise.
6927         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6928         change.
6929         (cortex_a15_alu_shift): Likewise.
6930         (cortex_a15_alu_shift_reg): Likewise.
6931         (cortex_a15_mult32): Likewise.
6932         (cortex_a15_mult64): Likewise.
6933         (cortex_a15_block): Likewise.
6934         (cortex_a15_branch): Likewise.
6935         (cortex_a15_load1): Likewise.
6936         (cortex_a15_load3): Likewise.
6937         (cortex_a15_store1): Likewise.
6938         (cortex_a15_store3): Likewise.
6939         (cortex_a15_call): Likewise.
6940         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
6941         (cortex_a5_f2r): Likewise.
6942         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
6943         change.
6944         (cortex_a53_f2r): Likewise.
6945         * config/arm/cortex-a7.md
6946         (cortex_a7_branch): Update for attribute change.
6947         (cortex_a7_call): Likewise.
6948         (cortex_a7_alu_imm): Likewise.
6949         (cortex_a7_alu_reg): Likewise.
6950         (cortex_a7_alu_shift): Likewise.
6951         (cortex_a7_mul): Likewise.
6952         (cortex_a7_load1): Likewise.
6953         (cortex_a7_store1): Likewise.
6954         (cortex_a7_load2): Likewise.
6955         (cortex_a7_store2): Likewise.
6956         (cortex_a7_load3): Likewise.
6957         (cortex_a7_store3): Likewise.
6958         (cortex_a7_load4): Likewise.
6959         (cortex_a7_store4): Likewise.
6960         (cortex_a7_fpalu): Likewise.
6961         (cortex_a7_fconst): Likewise.
6962         (cortex_a7_fpmuls): Likewise.
6963         (cortex_a7_neon_mul): Likewise.
6964         (cortex_a7_fpmacs): Likewise.
6965         (cortex_a7_neon_mla: Likewise.
6966         (cortex_a7_fpmuld: Likewise.
6967         (cortex_a7_fpmacd: Likewise.
6968         (cortex_a7_fpfmad: Likewise.
6969         (cortex_a7_fdivs: Likewise.
6970         (cortex_a7_fdivd: Likewise.
6971         (cortex_a7_r2f: Likewise.
6972         (cortex_a7_f2r: Likewise.
6973         (cortex_a7_f_flags: Likewise.
6974         (cortex_a7_f_loads: Likewise.
6975         (cortex_a7_f_loadd: Likewise.
6976         (cortex_a7_f_stores: Likewise.
6977         (cortex_a7_f_stored: Likewise.
6978         (cortex_a7_neon): Likewise.
6979         * config/arm/cortex-a8-neon.md
6980         (cortex_a8_neon_mrc): Update for attribute change.
6981         (cortex_a8_neon_mrrc): Likewise.
6982         (cortex_a8_neon_int_1): Likewise.
6983         (cortex_a8_neon_int_2): Likewise.
6984         (cortex_a8_neon_int_3): Likewise.
6985         (cortex_a8_neon_int_4): Likewise.
6986         (cortex_a8_neon_int_5): Likewise.
6987         (cortex_a8_neon_vqneg_vqabs): Likewise.
6988         (cortex_a8_neon_vmov): Likewise.
6989         (cortex_a8_neon_vaba): Likewise.
6990         (cortex_a8_neon_vaba_qqq): Likewise.
6991         (cortex_a8_neon_vsma): Likewise.
6992         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6993         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
6994         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
6995         long_scalar): Likewise.
6996         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6997         (cortex_a8_neon_mla_qqq_8_16): Likewise.
6998         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
6999         long_scalar_qdd_64_32_long): Likewise.
7000         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7001         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7002         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7003         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7004         (cortex_a8_neon_shift_1): Likewise.
7005         (cortex_a8_neon_shift_2): Likewise.
7006         (cortex_a8_neon_shift_3): Likewise.
7007         (cortex_a8_neon_vshl_ddd): Likewise.
7008         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7009         (cortex_a8_neon_vsra_vrsra): Likewise.
7010         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7011         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7012         (cortex_a8_neon_fp_vsum): Likewise.
7013         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7014         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7015         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7016         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7017         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7018         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7019         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7020         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7021         (cortex_a8_neon_bp_simple): Likewise.
7022         (cortex_a8_neon_bp_2cycle): Likewise.
7023         (cortex_a8_neon_bp_3cycle): Likewise.
7024         (cortex_a8_neon_ldr): Likewise.
7025         (cortex_a8_neon_str): Likewise.
7026         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7027         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7028         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7029         (cortex_a8_neon_vld2_4_regs): Likewise.
7030         (cortex_a8_neon_vld3_vld4): Likewise.
7031         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7032         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7033         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7034         (cortex_a8_neon_vst3_vst4): Likewise.
7035         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7036         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7037         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7038         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7039         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7040         (cortex_a8_neon_mcr): Likewise.
7041         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7042         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7043         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7044         change.
7045         (ca9_neon_mrrc): Likewise.
7046         (cortex_a9_neon_int_1): Likewise.
7047         (cortex_a9_neon_int_2): Likewise.
7048         (cortex_a9_neon_int_3): Likewise.
7049         (cortex_a9_neon_int_4): Likewise.
7050         (cortex_a9_neon_int_5): Likewise.
7051         (cortex_a9_neon_vqneg_vqabs): Likewise.
7052         (cortex_a9_neon_vmov): Likewise.
7053         (cortex_a9_neon_vaba): Likewise.
7054         (cortex_a9_neon_vaba_qqq): Likewise.
7055         (cortex_a9_neon_vsma): Likewise.
7056         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7057         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7058         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7059         long_scalar): Likewise.
7060         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7061         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7062         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7063         long_scalar_qdd_64_32_long): Likewise.
7064         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7065         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7066         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7067         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7068         (cortex_a9_neon_shift_1): Likewise.
7069         (cortex_a9_neon_shift_2): Likewise.
7070         (cortex_a9_neon_shift_3): Likewise.
7071         (cortex_a9_neon_vshl_ddd): Likewise.
7072         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7073         (cortex_a9_neon_vsra_vrsra): Likewise.
7074         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7075         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7076         (cortex_a9_neon_fp_vsum): Likewise.
7077         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7078         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7079         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7080         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7081         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7082         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7083         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7084         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7085         (cortex_a9_neon_bp_simple): Likewise.
7086         (cortex_a9_neon_bp_2cycle): Likewise.
7087         (cortex_a9_neon_bp_3cycle): Likewise.
7088         (cortex_a9_neon_ldr): Likewise.
7089         (cortex_a9_neon_str): Likewise.
7090         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7091         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7092         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7093         (cortex_a9_neon_vld2_4_regs): Likewise.
7094         (cortex_a9_neon_vld3_vld4): Likewise.
7095         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7096         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7097         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7098         (cortex_a9_neon_vst3_vst4): Likewise.
7099         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7100         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7101         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7102         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7103         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7104         (cortex_a9_neon_mcr): Likewise.
7105         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7106         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7107         (cortex_a9_fps): Likewise.
7108         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7109         change.
7110         (cortex_m4_fmuls): Likewise.
7111         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7112         change.
7113         (cortex_r4_mrc): Likewise.
7114         * config/arm/iterators.md: Update comment referring to neon_type.
7115         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7116         (iwmmxt_movsi_insn): Likewise.
7117         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7118         attribute change.
7119         (pj4_core_to_vfp): Likewise.
7120         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7121         attribute change.
7122         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7123         (vfp_fstore): Likewise.
7124         * doc/md.texi: Change references to neon_type to refer to type.
7125
7126 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7127
7128         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7129
7130 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7131
7132         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7133         lglobl if not weak.
7134
7135 2013-09-04  Easwaran Raman  <eraman@google.com>
7136
7137         PR middle-end/57370
7138         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7139         (build_and_add_sum): Use it.
7140         (appears_later_in_bb): Simplify code.
7141
7142 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7143
7144         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7145
7146 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7147
7148         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7149
7150 2013-09-04  Jan Hubicka  <jh@suse.cz>
7151
7152         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7153         * ipa-devirt.c: Include diganostic.h
7154         (odr_type_d): Add types and types_set.
7155         (hash_type_name): Work for types with vtables during LTO.
7156         (odr_hasher::remove): Fix comment; destroy types_set.
7157         (add_type_duplicate): New function,
7158         (get_odr_type): Use it.
7159         (dump_type_inheritance_graph): Dump type duplicates.
7160         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7161         graph.
7162         * tree.c (types_same_for_odr): Give exact answers on types with
7163         virtual tables.
7164
7165 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7166
7167         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7168         explaining their differences.
7169
7170 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7171
7172         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7173
7174 2013-09-03  Jeff Law  <law@redhat.com>
7175
7176         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7177         when not threading through a joiner block.  Pass joiner/no joiner
7178         state to register_jump_thread.
7179         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7180         state from argument rather than implying on path length.
7181         Dump the entire jump thread path into debugging dump.
7182         * tree-flow.h (register_jump_thread): Update prototype.
7183
7184 2013-08-29  Xinliang David Li  <davidxl@google.com>
7185
7186         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7187         Remove a trivial gcc_assert.
7188
7189 2013-08-29  Xinliang David Li  <davidxl@google.com>
7190
7191         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7192         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7193         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7194         Delay base decl alignment adjustment.
7195         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7196         * tree-vectorizer.h: New data structure.
7197         (set_dr_misalignment): New function.
7198         (dr_misalignment): Ditto.
7199         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7200         (vectorizable_load): Ditto.
7201         (ensure_base_align): New function.
7202         (vectorize_loops): Add dbg_cnt support.
7203         (execute_vect_slp): Ditto.
7204         * dbgcnt.def: New debug counter.
7205         * Makefile: New dependency.
7206
7207 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7208
7209         Revert:
7210
7211         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7212
7213         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7214
7215 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7216
7217         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7218         function descriptor.
7219
7220 2013-09-03  Richard Biener  <rguenther@suse.de>
7221
7222         * tree-affine.c (add_elt_to_tree): Fix association issue,
7223         avoid useless converts and make sure to always return a
7224         properly typed result.
7225
7226 2013-09-03  Richard Biener  <rguenther@suse.de>
7227
7228         PR middle-end/57656
7229         * fold-const.c (negate_expr_p): Fix division case.
7230         (negate_expr): Likewise.
7231
7232 2013-09-03  Richard Biener  <rguenther@suse.de>
7233
7234         PR lto/58285
7235         * tree-streamer-out.c: Include tm.h.
7236         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7237
7238 2013-09-03  Jan Hubicka  <jh@suse.cz>
7239
7240         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7241
7242 2013-09-03  Alan Modra  <amodra@gmail.com>
7243
7244         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7245         multilibs to big-endian target and vice versa.
7246         * config/rs6000/t-linux64: Use := assignment on all vars.
7247         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7248         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7249         * config/rs6000/t-linux64le: New file.
7250         * config/rs6000/t-linux64bele: New file.
7251         * config/rs6000/t-linux64lebe: New file.
7252
7253 2013-09-02  Jan Hubicka  <jh@suse.cz>
7254
7255         * ipa-inline-transform.c (inline_transform): Do not
7256         optimize_inline_calls when not optimizing.
7257
7258 2013-09-02  Jan Hubicka  <jh@suse.cz>
7259
7260         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7261         duplicated nodes for assembler names.
7262         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7263         hard registers from assembler name hash.
7264
7265 2013-09-02  Jan Hubicka  <jh@suse.cz>
7266
7267         * ipa-split.c (execute_split_functions): Split externally visible
7268         functions called once.
7269
7270 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7271
7272         PR ipa/58106
7273         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7274         linked list.  When finding the correct duplicate, also consider also
7275         the caller in additon to its inlined_to node.
7276
7277 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7278
7279         * config/aarch64/aarch64-simd-builtins.def
7280         (dup_lane_scalar): Remove.
7281         * config/aarch64/aarch64-simd.md
7282         (aarch64_simd_dup): Add 'w->w' alternative.
7283         (aarch64_dup_lane<mode>): Allow for VALL.
7284         (aarch64_dup_lane_scalar<mode>): Remove.
7285         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7286         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7287         (aarch64_get_lane_unsigned<mode>): Likewise.
7288         (aarch64_get_lane<mode>): Likewise.
7289         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7290         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7291         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7292         (VCON): Change container of V2SF.
7293         (vswap_width_name): Likewise.
7294         * config/aarch64/arm_neon.h
7295         (__aarch64_vdup_lane_any): New.
7296         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7297         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7298         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7299
7300 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7301
7302         PR middle-end/56382
7303         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7304         the source or the destination is a single hard register.
7305
7306 2013-09-02  Richard Biener  <rguenther@suse.de>
7307
7308         PR middle-end/57511
7309         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7310         non-linear SCEVs.
7311
7312 2013-09-02  Richard Biener  <rguenther@suse.de>
7313
7314         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7315         arithmetic to sizetype.
7316
7317 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7318
7319         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7320         Find auto-increment use both before and after candidate.
7321
7322 2013-09-02  Marek Polacek  <polacek@redhat.com>
7323
7324         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7325
7326 2013-09-01  Jan Hubicka  <jh@suse.cz>
7327
7328         * Makefile.in: Add ipa-profile.o
7329         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7330         * cgraph.c (struct cgraph_propagate_frequency_data,
7331         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7332         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7333         * cgraph.h (cgraph_propagate_frequency): Remove.
7334         * ipa-inline-analysis.c: Include ipa-utils.h;
7335         drop duplicated cfgloop.h.
7336         (inline_update_callee_summaries): Update.
7337         * ipa-profile.c: New file.
7338         * ipa-utils.h (ipa_propagate_frequency): Declare.
7339         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7340         data-streamer.h, value-prof.h.
7341         (symtab_remove_unreachable_nodes): Update profile.
7342         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7343         account_time_size, cmp_counts, dump_histogram,
7344         ipa_profile_generate_summary, ipa_profile_write_summary,
7345         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7346         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7347         Move to ipa-profile.c.
7348
7349 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7350
7351         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7352
7353 2013-09-01  Jan Hubicka  <jh@suse.cz>
7354
7355         * common.opt (fdevirtualize-speculatively): New function.
7356         * invoke.texi (fdevirtualize-speculatively): Document.
7357         * ipa-devirt.c: Include ipa-inline.h
7358         (likely_target_p): New function.
7359         (ipa_devirt): New function.
7360         (gate_ipa_devirt): New function.
7361         (pass_data_ipa_devirt): New static var.
7362         (pass_ipa_devirt): Likewise.
7363         (make_pass_ipa_devirt): New function.
7364         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7365         (common_handle_option): Disable devirtualization when
7366         value range profiling is available.
7367         * passes.def (pass_ipa_devirt): Add.
7368         * timever.def (TV_IPA_DEVIRT): New timevar.
7369         * tree-pass.h (make_pass_ipa_devirt):
7370
7371 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7372
7373         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7374         include sanitize(undefined).
7375
7376 2013-08-31  Diego Novillo  <dnovillo@google.com>
7377
7378         * Makefile.in (TREE_CORE_H): Define.
7379         (TREE_H): Use.
7380         (GTFILES): Add tree-core.h.
7381         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7382         size the array.
7383         * tree-core.h: New file.
7384         Move all data structures, enum, typedefs, global
7385         declarations and constants from ...
7386         * tree.h: ... here.
7387
7388 2013-08-31  Jan Hubicka  <jh@suse.cz>
7389
7390         * bulitins.c (expand_builtin): Do not early exit for gcov
7391         instrumented functions.
7392
7393 2013-08-31  Marek Polacek  <polacek@redhat.com>
7394
7395         * ubsan.c: Include tm_p.h.
7396
7397 2013-08-31  Jan Hubicka  <jh@suse.cz>
7398
7399         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7400         warning.
7401
7402         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7403         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7404         * tree-cfg.c (verify_gimple_label): ... here.
7405         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7406         (ipa_merge_profiles): New function.
7407         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7408         (lto_input_function_body): Likewise.
7409         * ipa-utils.h (ipa_merge_profiles): Declare.
7410         * lto-streamer.h (lto_input_function_body): Update prototype.
7411         (emit_label_in_global_context_p): Remove.
7412         * lto-symtab.c: Include ipa-utils.h
7413         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7414
7415 2013-08-31  Jan Hubicka  <jh@suse.cz>
7416
7417         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7418
7419 2013-08-31  Jan Hubicka  <jh@suse.cz>
7420
7421         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7422
7423 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7424
7425         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7426         "cmp" RTX before signed_comparison_operator check to account
7427         for "code" changes.
7428
7429 2013-08-30  Jan Hubicka  <jh@suse.cz>
7430
7431         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7432         (detect_type_change_1): Rename to ...
7433         (detect_type_change): ... this one; early return on non-polymorphic
7434         types.
7435         (detect_type_change_ssa): Add comp_type parameter; update
7436         use of detect_type_change.
7437         (compute_complex_assign_jump_func): Add param_type parameter;
7438         update use of detect_type_change_ssa.
7439         (compute_complex_ancestor_jump_func): Likewise.
7440         (ipa_get_callee_param_type): New function.
7441         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7442         update calls to the jump function computation functions.
7443
7444 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7445             Steven Bosscher  <steven@gcc.gnu.org>
7446
7447         * cfgrtl.c (fixup_new_cold_bb): New routine.
7448         (commit_edge_insertions): Invoke fixup_partitions.
7449         (find_partition_fixes): New routine.
7450         (fixup_partitions): Ditto.
7451         (verify_hot_cold_block_grouping): Update comments.
7452         (rtl_verify_edges): Invoke find_partition_fixes.
7453         (rtl_verify_bb_pointers): Update comments.
7454         (rtl_verify_bb_layout): Ditto.
7455         * basic-block.h (probably_never_executed_edge_p): Declare.
7456         (fixup_partitions): Ditto.
7457         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7458         * bb-reorder.c (sanitize_hot_paths): New function.
7459         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7460         sanitize_hot_paths.
7461         * predict.c (probably_never_executed_edge_p): New routine.
7462         * cfg.c (check_bb_profile): Add partition insanity warnings.
7463
7464 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7465
7466         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7467
7468 2013-08-30  Marek Polacek  <polacek@redhat.com>
7469
7470         * Makefile.in (ubsan.o): Add.
7471         (c-family/c-ubsan.o): Add.
7472         (builtins.o): Add ubsan.h dependency.
7473         * ubsan.h: New file.
7474         * ubsan.c: New file.
7475         * common.opt: Add -fsanitize=undefined option.
7476         (flag_sanitize): Add variable.
7477         (fsanitize=): Add option.  Add Driver.
7478         (fsanitize=thread): Remove option.
7479         (fsanitize=address): Likewise.
7480         (static-libubsan): New option.
7481         * doc/invoke.texi: Document the new flag and -static-libubsan.
7482         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7483         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7484         * builtin-attrs.def (ATTR_COLD): Define.
7485         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7486         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7487         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7488         * flag-types.h (sanitize_code): New enum.
7489         * opts.c (common_handle_option): Parse command line arguments
7490         of -fsanitize=.  Add -fsanitize=unreachable option.
7491         * varasm.c (get_variable_section): Adjust.
7492         (assemble_noswitch_variable): Likewise.
7493         (assemble_variable): Likewise.
7494         (output_constant_def_contents): Likewise.
7495         (categorize_decl_for_section): Likewise.
7496         (place_block_symbol): Likewise.
7497         (output_object_block): Likewise.
7498         * builtins.def: Likewise.
7499         * toplev.c (compile_file): Likewise.
7500         (process_options): Likewise.
7501         * cppbuiltin.c: Likewise.
7502         * tsan.c (tsan_pass): Likewise.
7503         (tsan_gate): Likewise.
7504         (tsan_gate_O0): Likewise.
7505         * cfgexpand.c (partition_stack_vars): Likewise.
7506         (expand_stack_vars): Likewise.
7507         (defer_stack_allocation): Likewise.
7508         (expand_used_vars): Likewise.
7509         * cfgcleanup.c (old_insns_match_p): Likewise.
7510         * asan.c (asan_finish_file): Likewise.
7511         (asan_instrument): Likewise.
7512         (gate_asan): Likewise.
7513         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7514         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7515         (asan_global_struct): Use pointer_sized_int_node instead
7516         calling build_nonstandard_integer_type.
7517         (initialize_sanitizer_builtins): Likewise.
7518         (asan_finish_file): Likewise.
7519         * gcc.c: Document %{%:function(args):X}.
7520         (static_spec_functions): Add sanitize.
7521         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7522         store funcval != NULL there.
7523         (do_spec_1): Adjust handle_spec_function caller.
7524         (handle_braces): Allow %:function(args) as condition.
7525         (sanitize_spec_function): New function.
7526         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7527         (LIBUBSAN_SPEC): Likewise.
7528         (LIBUBSAN_EARLY_SPEC): Likewise.
7529         (SANITIZER_SPEC): Handle libubsan.
7530         (SANITIZER_EARLY_SPEC): Likewise.
7531         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7532         instead of fsanitize=address.
7533         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7534         instead of fsanitize=address*.
7535         * builtins.c: Include ubsan.h.
7536         (fold_builtin_0): Instrument __builtin_unreachable.
7537         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7538         instead of flag_asan.
7539         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7540         (pointer_sized_int_node): Define.
7541         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7542
7543 2013-08-30  Mike Stump  <mikestump@comcast.net>
7544
7545         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7546         with RE patterns.
7547
7548 2013-08-29  Jan Hubicka  <jh@suse.cz>
7549
7550         * cgraph.c (cgraph_function_body_availability): Handle weakref
7551         correctly.
7552         * passes.def: Remove pass_fixup_cfg.
7553         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7554         track when we need to remove functions.
7555         (gate_ipa_inline): Execute inlining always; add comment why.
7556         (pass_data_ipa_inline): Remove TODO_remove_functions.
7557         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7558         do not produce summaries.
7559         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7560         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7561         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7562         constant pool and vtable.
7563
7564 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7565
7566         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7567         New arm_neon.h's internal macros to specify 64-bit constants.
7568         Avoid using stdint.h's macros.
7569
7570 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7571
7572         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7573         before applying REG_P.
7574
7575 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7576
7577         PR tree-optimization/58277
7578         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7579         after seeing too many stmts with vdef in between dombb and current
7580         bb, invalidate everything.
7581
7582 2013-08-30  Richard Biener  <rguenther@suse.de>
7583
7584         * fold-const.c (fold_single_bit_test): Fix overflow test.
7585
7586 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7587
7588         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7589         and which can live in a register, always retrieve the value on entry.
7590         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7591         passed by invisible reference specially.
7592         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7593         (vt_add_function_parameter): Correctly deal with a parameter passed by
7594         invisible reference.
7595
7596 2013-08-30  Jan Hubicka  <jh@suse.cz>
7597
7598         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7599
7600 2013-08-30  Richard Biener  <rguenther@suse.de>
7601
7602         PR tree-optimization/58228
7603         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7604         allow invariant loads in nested loop vectorization.
7605
7606 2013-08-30  Richard Biener  <rguenther@suse.de>
7607
7608         PR tree-optimization/58223
7609         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7610         (has_anti_or_output_dependence): ... this and adjust to also
7611         look for output dependences.
7612         (mark_nodes_having_upstream_mem_writes): Adjust.
7613         (rdg_flag_uses): Likewise.
7614
7615 2013-08-30  Richard Biener  <rguenther@suse.de>
7616
7617         PR tree-optimization/58010
7618         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7619         assert that we have a loop-closed PHI.
7620
7621 2013-08-29  Jan Hubicka  <jh@suse.cz>
7622
7623         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7624         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7625         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7626         (lto_free_function_in_decl_state_for_node): New function.
7627
7628 2013-08-29  Xinliang David Li  <davidxl@google.com>
7629
7630         * loop-unroll.c (report_unroll_peel): Minor message change.
7631         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7632         Emit alignment peeling message with default -fopt-info.
7633         (vect_loop_versioning): Emit loop version info message.
7634         * tree-vectorizer.c (vectorize_loops): Minor message change.
7635         (execute_vect_slp): Ditto.
7636
7637 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7638
7639         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7640         of the clone from the DECL_NAME of the original function.
7641
7642 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7643
7644         * passes.c (register_pass): Add overload.
7645         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7646
7647 2013-08-29  Jan Hubicka  <jh@suse.cz>
7648
7649         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7650         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7651         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7652         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7653         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7654         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7655         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7656         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7657
7658 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7659
7660         * dumpfile.c (dump_loc): Output column number.
7661         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7662         * doc/invoke.texi: Document optall -fopt-info flag.
7663         * profile.c (read_profile_edge_counts): Use new dump framework.
7664         (compute_branch_probabilities): Ditto.
7665         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7666         when pass not in any opt group.
7667         * pass_manager.h (pass_manager::get_pass_profile): New method.
7668         * value-prof.c (check_counter): Use new dump framework.
7669         (check_ic_target): Ditto.
7670         * coverage.c (get_coverage_counts): Ditto.
7671         (coverage_init): Setup new dump framework.
7672
7673 2013-08-29  Richard Biener  <rguenther@suse.de>
7674
7675         PR tree-optimization/58246
7676         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7677         handle the dominance check inside a basic-block.
7678
7679 2013-08-29  Richard Biener  <rguenther@suse.de>
7680
7681         PR middle-end/57287
7682         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7683         of default defs that appear in abnormal PHI nodes.
7684
7685 2013-08-29  Richard Biener  <rguenther@suse.de>
7686
7687         PR tree-optimization/57685
7688         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7689         single-use operands to avoid exponential complexity.
7690
7691 2013-08-28  Dehao Chen  <dehao@google.com>
7692
7693         * ipa-inline.c (edge_badness): Fix integer underflow.
7694
7695 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7696
7697         * gtm-builtins.def (_ITM_free): Declare leaf.
7698
7699 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7700
7701         PR target/58067
7702         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7703         (*tls_local_dynamic_base_64_largepic): Likewise.
7704         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7705         Remove predicate from call operand.
7706         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7707         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7708
7709 2013-08-28  Jeff Law  <law@redhat.com>
7710
7711         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7712         checks for the number of predecessors and successors allowed.
7713         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7714         which require copying a joiner block if there is a request which
7715         is a subpath that requires no joiner block copying.
7716
7717 2013-08-28  Jan Hubicka  <jh@suse.cz>
7718
7719         * lto-streamer-out.c (DFS_write_tree_body): Drop
7720         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
7721         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
7722         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
7723         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
7724         Do not read DECL_ERROR_ISSUED.
7725         (unpack_ts_decl_with_vis_value_fields): Do not read
7726         DECL_DEFER_OUTPUT.
7727         (lto_input_ts_binfo_tree_pointers): Do not read
7728         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
7729         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
7730         write DECL_ERROR_ISSUED..
7731         (pack_ts_decl_with_vis_value_fields): Do not write
7732         DECL_DEFER_OUTPUT.
7733         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
7734         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
7735         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
7736         * tree.h (tree_decl_common): Update comment.
7737         (DECL_ERROR_ISSUED): Remove.
7738
7739 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7740
7741         PR middle-end/58257
7742         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
7743
7744 2013-08-28  Jan Hubicka  <jh@suse.cz>
7745
7746         * builtins.def (free): Declare leaf.
7747
7748 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
7749
7750         * gdbhooks.py: New.
7751         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
7752         * configure: Regenerate.
7753
7754 2013-08-27  Martin Jambor  <mjambor@suse.cz>
7755
7756         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
7757         (ipa_ancestor_jf_data): Likewise.
7758         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
7759         (ipa_get_jf_pass_through_type_preserved): New function.
7760         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
7761         (ipa_get_jf_ancestor_type_preserved): New function.
7762         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
7763         (ipa_get_jf_ancestor_result): Likewise.
7764         (propagate_vals_accross_pass_through): Use
7765         ipa_get_jf_pass_through_result to do all the value mappings.
7766         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
7767         type_preserved flag.
7768         (ipa_set_jf_cst_copy): New function.
7769         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
7770         (ipa_set_jf_arith_pass_through): Likewise.
7771         (ipa_set_ancestor_jf): Likewise.
7772         (compute_complex_assign_jump_func): Set type_preserved instead of
7773         punting.
7774         (ipa_compute_jump_functions_for_edge): Likewise.
7775         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
7776         (update_jump_functions_after_inlining): Update type_preserved.
7777         Explicitely create jump functions when combining one with pass_through.
7778         (ipa_write_jump_function): Stream the type_preserved flags.
7779         (ipa_read_jump_function): Likewise.
7780
7781 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
7782             Aldy Hernandez  <aldyh@redhat.com>
7783
7784         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
7785         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
7786         * function.h (struct function): Add has_force_vect_loops and
7787         has_simduid_loops.
7788         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
7789         * gimple.c (gimple_build_omp_critical): Add KIND argument and
7790         handle it.
7791         * gimple.def: Update CLAUSES comments.
7792         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
7793         (gimple_build_omp_for): Add argument to prototype.
7794         (gimple_omp_for_kind): New.
7795         (gimple_omp_for_set_kind): New.
7796         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
7797         GOVD_DATA_SHARE_CLASS.
7798         (enum omp_region_type): Add ORT_SIMD.
7799         (gimple_add_tmp_var): Handle ORT_SIMD.
7800         (gimplify_var_or_parm_decl): Same.
7801         (is_gimple_stmt): Same.
7802         (omp_firstprivatize_variable): Same.
7803         (omp_add_variable): Only use splay_tree_insert if lookup failed.
7804         (omp_notice_variable): Handle ORT_SIMD.
7805         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
7806         (omp_check_private): Handle ORT_SIMD.
7807         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
7808         OMP_CLAUSE_SAFELEN.
7809         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
7810         Handle OMP_CLAUSE_LASTPRIVATE.
7811         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
7812         OMP_CLAUSE_SAFELEN.
7813         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
7814         (gimplify_expr): Handle OMP_SIMD.
7815         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
7816         (expand_GOMP_SIMD_VF): New.
7817         (expand_GOMP_SIMD_LAST_LANE): New.
7818         * internal-fn.def (GOMP_SIMD_LANE): New.
7819         (GOMP_SIMD_VF): New.
7820         (GOMP_SIMD_LAST_LANE): New.
7821         * omp-low.c: Include target.h.
7822         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
7823         OMP_CLAUSE_SAFELEN.
7824         (check_omp_nesting_restrictions): Same.
7825         (omp_max_vf): New.
7826         (lower_rec_simd_input_clauses): New.
7827         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
7828         OMP_CLAUSE_LINEAR.
7829         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
7830         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
7831         (expand_omp_build_assign): New.
7832         (expand_omp_for_init_counts): New.
7833         (expand_omp_for_init_vars): New.
7834         (extract_omp_for_update_vars): New.
7835         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
7836         and rewrite accordingly.
7837         (expand_omp_simd): New.
7838         (expand_omp_for): Use expand_omp_simd.
7839         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
7840         (lower_omp_for): Do not lower the body.
7841         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
7842         in their own loops.
7843         * tree-flow.h (find_omp_clause): Remove prototype.
7844         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
7845         forcing vectorization of the loop, or if flag_tree_vectorize.
7846         (gate_tree_if_conversion): Similarly.
7847         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
7848         gimple_build_omp_for.
7849         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
7850         * tree-parloops (create_parallel_loop): Pass kind argument to
7851         gimple_build_omp_for.
7852         * tree-pretty-print.c (dump_omp_clause): Add cases for
7853         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
7854         OMP_CLAUSE__SIMDUID_.
7855         (dump_generic_node): Handle OMP_SIMD.
7856         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
7857         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
7858         unroll OMP_SIMD loops here.
7859         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
7860         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
7861         loop->safelen.
7862         (vect_analyze_data_refs): Handle simd loops.
7863         * tree-vect-loop.c (vectorizable_live_operation): Handle
7864         IFN_GOMP_SIMD*.
7865         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
7866         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
7867         (vectorizable_load): Same.
7868         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
7869         (struct simduid_to_vf): New.
7870         (simduid_to_vf::hash): New.
7871         (simduid_to-vf::equal): New.
7872         (struct simd_array_to_simduid): New.
7873         (simd_array_to_simduid::hash): New.
7874         (simd_array_to_simduid::equal): New.
7875         (adjust_simduid_builtins): New.
7876         (struct note_simd_array_uses_struct): New.
7877         (note_simd_array_uses_cb): New.
7878         (note_simd_array_uses): New.
7879         (vectorize_loops): Handle simd hints and adjust simd builtins
7880         accordingly.
7881         * tree-vectorizer.h (struct _stmt_vec_info): Add
7882         simd_lane_access_p field.
7883         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
7884         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
7885         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
7886         (omp_clause_code_name): Same.
7887         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
7888         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
7889         * tree.def (OMP_SIMD): New entry.
7890         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
7891         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
7892         (OMP_CLAUSE_DECL): Adjust range for new clauses.
7893         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
7894         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
7895         (OMP_CLAUSE_LINEAR_STEP): New.
7896         (OMP_CLAUSE_SAFELEN_EXPR): New.
7897         (OMP_CLAUSE__SIMDUID__DECL): New.
7898         (find_omp_clause): New prototype.
7899
7900 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
7901
7902         * config/i386/driver-i386.c (host_detect_local_cpu): Update
7903         Haswell processor detection.
7904
7905 2013-08-27  Christian Widmer  <shadow@umbrox.de>
7906
7907         PR target/57927
7908         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
7909         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
7910         AVX2 capable processors.
7911
7912 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
7913
7914         * config/aarch64/arm_neon.h: Replace all inline asm implementations
7915         of vget_low_* with implementations in terms of other intrinsics.
7916
7917 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
7918
7919         PR middle-end/57219
7920         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
7921         values to -1, 0 and 1.
7922
7923 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
7924
7925         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
7926         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
7927         (<optab><mode>3_insn): Remove.
7928         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
7929         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
7930         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
7931         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
7932         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
7933         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
7934         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
7935         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
7936         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
7937         (ror<mode>3_insn): Likewise.
7938         * config/aarch64/predicates.md (aarch64_simd_register): New.
7939
7940 2013-08-27  Richard Biener  <rguenther@suse.de>
7941
7942         PR tree-optimization/57521
7943         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
7944         one edge is non-critical.
7945         (find_phi_replacement_condition): Make sure to use a non-critical
7946         edge.  Cleanup and remove old bug workarounds.
7947         (bb_postdominates_preds): Remove.
7948         (if_convertible_loop_p_1): Do not compute post-dominators.
7949         (combine_blocks): Do not free post-dominators.
7950         (main_tree_if_conversion): Likewise.
7951         (pass_data_if_conversion): Add TODO_verify_ssa.
7952
7953 2013-08-27  DJ Delorie  <dj@redhat.com>
7954
7955         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
7956
7957 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
7958
7959         * function.c (assign_parm_find_data_types): Set passed_mode and
7960         nominal_mode to the TYPE_MODE of nominal_type for the built
7961         pointer type in case of the struct-pass-by-reference.
7962
7963 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
7964
7965         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
7966         (UINT16_TYPE): Change default to "unsigned int".
7967
7968         * config/avr/avr.opt (mfract-convert-truncate): New option.
7969         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
7970         is set, round negative fractional integers according to n1169
7971         when converting to integer types.
7972
7973 2013-08-26  Jan Hubicka  <jh@suse.cz>
7974
7975         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
7976         methods can not be called indirectly when their address is not taken.
7977
7978 2013-08-26  Jan Hubicka  <jh@suse.cz>
7979
7980         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
7981         ctor_for_folding.
7982
7983 2013-08-26  Jan Hubicka  <jh@suse.cz>
7984
7985         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
7986         can be unshared.
7987
7988 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
7989
7990         * reload.c (find_valid_class): Allow classes that do not include
7991         FIRST_PSEUDO_REGISTER - 1.
7992
7993 2013-08-26  Jan Hubicka  <jh@suse.cz>
7994
7995         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
7996         fix edge count/frequency when speculation failed; fix type check
7997         for the direct call.
7998
7999 2013-08-26  Jan Hubicka  <jh@suse.cz>
8000
8001         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8002
8003 2013-08-26  Jan Hubicka  <jh@suse.cz>
8004
8005         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8006         to be changed by edge redirection.
8007
8008 2013-08-26  Jan Hubicka  <jh@suse.cz>
8009
8010         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8011         formating; add sanity check.
8012         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8013         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8014         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8015         (resolve_noninline_speculation): Update callee keys, too.
8016
8017 2013-08-26  Jan Hubicka  <jh@suse.cz>
8018
8019         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8020         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8021
8022 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8023
8024         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8025         into proper place.
8026
8027 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8028
8029         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8030         (x86_64_elf_select_section): Ditto.
8031         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8032         arguments.
8033         (ix86_pass_by_reference): Ditto.
8034         (output_set_got): Ditto.
8035         (ix86_unary_operator_ok): Ditto.
8036         (ix86_expand_builtin): Ditto.
8037
8038 2013-08-23  Jan Hubicka  <jh@suse.cz>
8039
8040         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8041
8042 2013-08-23  Jan Hubicka  <jh@suse.cz>
8043
8044         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8045         (tree_decl_with_vis): Add FINAL field.
8046
8047 2013-08-23  Jeff Law  <law@redhat.com>
8048
8049         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8050         the debugging dump when the expression is fully redundant.
8051
8052 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8053
8054         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8055         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8056         * pretty-print.c (pp_formatted_text_data): Likewise.
8057         (pp_write_text_to_stream): Likewise.
8058         (pp_write_text_as_dot_label_to_stream): Likewise.
8059         (pp_append_r): Likewise.
8060         (pp_format): Likewise.
8061         (pp_flush): Likewise.
8062         (pp_clear_output_area): Likewise.
8063         (pp_append_text): Likewise.
8064         (pp_formatted_text): Likewise.
8065         (pp_remaining_character_count_for_line): Likewise.
8066         (pp_newline): Likewise.
8067         (pp_character): Likewise.
8068         (output_buffer::~output_buffer): Define.
8069         (pretty_printer::~pretty_printer): Destruct output buffer.
8070         * pretty-print.h (output_buffer::~output_buffer): Declare.
8071         (pretty_printer::~pretty_printer): Declare virtual.
8072
8073 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8074
8075         PR other/57324
8076         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8077         HOST_WIDE_INT_M1U): New macros.
8078         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8079         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8080         unsigned -1 for lshift.
8081         * cse.c (cse_insn): Likewise.
8082         * double-int.c (rshift_double, lshift_double): Likewise.
8083         * builtins.c (fold_builtin_bitop): Likewise.
8084         * combine.c (force_to_mode): Likewise.
8085         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8086         * simplify-rtx.c (simplify_const_unary_operation,
8087         simplify_const_binary_operation): Likewise.
8088         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8089         check_va_list_escapes): Likewise.
8090         * rtlanal.c (nonzero_bits1): Likewise.
8091         * expmed.c (expand_smod_pow2): Likewise.
8092         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8093
8094 2013-08-23  Jan Hubicka  <jh@suse.cz>
8095
8096         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8097         as having address taken.
8098
8099 2013-08-23  Jan Hubicka  <jh@suse.cz>
8100
8101         * ipa-utils.h (method_class_type): Declare.
8102         * ipa-devirt.c (method_class_type): Export.
8103
8104         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8105         do not walk base classes of anonymous types.
8106
8107 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8108
8109         PR rtl-optimization/58220
8110         PR regression/58221
8111         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8112         handle SEQUENCE insns properly.
8113
8114 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8115
8116         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8117         definition.
8118         (pp_newline_and_indent): Likewise.
8119         (pp_separate_with): Likewise.
8120         * pretty-print.c (pp_newline_and_flush): Define.
8121         (pp_newline_and_indent): Likewise.
8122         (pp_separate_with): Likewise.
8123
8124 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8125
8126         PR target/58218
8127         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8128         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8129
8130 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8131
8132         * config/i386/predicates.md (ext_sse_reg_operand): New.
8133         * config/i386/i386.md (*movti_internal): Use
8134         predicate to determine if EVEX is needed.
8135         (*movsi_internal): Ditto.
8136         (*movdf_internal): Ditto.
8137         (*movsf_internal): Ditto.
8138         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8139
8140 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8141
8142         PR tree-optimization/58209
8143         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8144         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8145         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8146         emit POINTER_PLUS_EXPR.
8147         (create_tailcall_accumulator): For pointer result type accumulate in
8148         sizetype type.
8149
8150 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8151
8152         * configure.ac: Add backslashes missing from the last change.
8153         * configure: Regenerate.
8154
8155 2013-08-22  Jan Hubicka  <jh@susue.cz>
8156
8157         * ipa.c (function_and_variable_visibility): First remember function
8158         was global and then make it local.
8159
8160 2013-08-22  Julian Brown  <julian@codesourcery.com>
8161
8162         * configure.ac: Add aarch64 to list of arches which use "nop" in
8163         debug_line test.
8164         * configure: Regenerate.
8165
8166 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8167
8168         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8169         gnu_libc_has_function.
8170         * config/s390/tpf.h: Likewise.
8171
8172 2013-08-22  Jan Hubicka  <jh@susue.cz>
8173
8174         * timevar.c (validate_phases): Add cast.
8175
8176 2013-08-22  Jan Hubicka  <jh@susue.cz>
8177
8178         * timevar.c (validate_phases): Use size_t for memory.
8179         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8180
8181 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8182
8183         * pretty-print.h (output_buffer::output_buffer): Declare.
8184         (pretty_printer::pretty_printer): Likewise.
8185         (pp_construct): Remove.
8186         * pretty-print.c (output_buffer::output_buffer): Define.
8187         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8188         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8189         (print_gimple_expr): Likewise.
8190         (print_gimple_seq): Likewise.
8191         (gimple_dump_bb): Likewise.
8192         * sched-vis.c (dump_value_slim): Likewise.
8193         (dump_insn_slim): Likewise.
8194         (dump_rtl_slim): Likewise.
8195         (str_pattern_slim): Likewise.
8196         * tree-mudflap.c (mf_varname_tree): Likewise.
8197         * graph.c (print_graph_cfg): Likewise.
8198         (start_graph_dump): Likewise.
8199         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8200         placement-new.
8201         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8202         pretty printer initialization.
8203         * coretypes.h (diagnostic_context): Remove superflous type alias
8204         declaration.
8205         (pretty_printer): Likewise.  Declare directly as a class.
8206         (pretty_print_info): Remove declaration as class.
8207         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8208         and pp_clear_output_area.
8209         (asan_add_global): Likewise.
8210
8211 2013-08-22  Jan Hubicka  <jh@suse.cz>
8212
8213         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8214         * ipa-utils.h (update_type_inheritance_graph): Declare.
8215         (possible_polymorphic_call_target_p): Declare.
8216         (possible_polymorphic_call_target_p): New.
8217         * ipa-devirt.c: Update toplevel comments.
8218         (cached_polymorphic_call_targets): Move up.
8219         (odr_type_d): Move ID down.
8220         (polymorphic_type_binfo_p): Update comment.
8221         (odr_hasher::remove): Likewise;
8222         (get_odr_type): Set anonymous_namespace.
8223         (dump_odr_type): Dump it.
8224         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8225         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8226         (record_binfo): Add comment.
8227         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8228         built.
8229         (devirt_node_removal_hook): Do not iCE when cache is freed.
8230         (possible_polymorphic_call_target_p): New predicate.
8231         (update_type_inheritance_graph): New function.
8232
8233 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8234             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8235             Sergey Lega  <sergey.s.lega@intel.com>
8236             Anna Tikhonova  <anna.tikhonova@intel.com>
8237             Ilya Tocar  <ilya.tocar@intel.com>
8238             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8239             Ilya Verbin  <ilya.verbin@intel.com>
8240             Kirill Yukhin  <kirill.yukhin@intel.com>
8241             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8242
8243         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8244         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8245         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8246         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8247         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8248         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8249         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8250         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8251         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8252         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8253         OPT_mavx512pf, OPT_mavx512er cases.
8254         * config/i386/constraints.md (v): New constraint.
8255         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8256         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8257         (bit_AVX512CD): New.
8258         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8259         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8260         * config/i386/i386-c.c (ix86_target_macros_internal):
8261         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8262         __AVX512PF__.
8263         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8264         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8265         * config/i386/i386.c (regclass_map, dbx_register_map)
8266         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8267         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8268         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8269         -mavx512pf options.
8270         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8271         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8272         -mavx512pf options.  Fix formatting.
8273         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8274         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8275         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8276         -mavx512cd, -mavx512pf options.
8277         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8278         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8279         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8280         ALL_SSE_REGS.
8281         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8282         (ix86_set_reg_reg_cost): Ditto.
8283         (x86_order_regs_for_local_alloc): Ditto.
8284         (MAX_VECT_LEN): Extend to 64-byte.
8285         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8286         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8287         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8288         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8289         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8290         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8291         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8292         (SSE_REG_MODE_P): Support new modes.
8293         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8294         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8295         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8296         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8297         (REG_CLASS_CONTENTS): Add new registers.
8298         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8299         (EXT_REX_SSE_REGNO_P): New.
8300         (HI_REGISTER_NAMES): Add new registers.
8301         * config/i386/i386.md: Define constants for new registers.
8302         (mode): Add new 512-bit modes.
8303         (prefix): Support evex prefix.
8304         (isa): Support avx512f, noavx512f, fma_avx512f.
8305         (ssemodesuffix): Add new 512-bit modes.
8306         (movxi): New.
8307         (*movxi_internal_avx512f): Ditto.
8308         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8309         Support MODE_XI.
8310         (*movsi_internal): Likewise.
8311         (*movdf_internal): Likewise.
8312         (*movsf_internal): Likewise.
8313         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8314         (<code><mode>3): Likewise.
8315         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8316         New.
8317         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8318         with the new constraint "v".
8319         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8320         modes.
8321         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8322         with the new constraint "v".
8323         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8324         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8325         (*<plusminus_insn><mode>3): Likewise.
8326         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8327         (*mul<mode>3): Likewise.
8328         (<sse>_vmmul<mode>3): Likewise.
8329         (<sse>_div<mode>3): Likewise.
8330         (<sse>_vmdiv<mode>3): Likewise.
8331         (<sse>_sqrt<mode>2): Likewise.
8332         (<sse>_vmsqrt<mode>2): Likewise.
8333         (*<code><mode>3_finite): Likewise.
8334         (*<code><mode>3) <smaxmin>: Likewise.
8335         (<sse>_vm<code><mode>3): Likewise.
8336         (*<code><mode>3) <any_logic>: Likewise.
8337         (*fma_fmadd_<mode>): Likewise.
8338         (*fma_fmsub_<mode>): Likewise.
8339         (*fma_fnmadd_<mode>): Likewise.
8340         (*fma_fnmsub_<mode>): Likewise.
8341         (*fma_fmaddsub_<mode>): Likewise.
8342         (*fma_fmsubadd_<mode>): Likewise.
8343         (*fmai_fmadd_<mode>): Likewise.
8344         (*fmai_fmsub_<mode>): Likewise.
8345         (*fmai_fnmadd_<mode>): Likewise.
8346         (*fmai_fnmsub_<mode>): Likewise.
8347         (sse_cvtsi2ss): Likewise.
8348         (sse_cvtsi2ssq): Likewise.
8349         (sse_cvtss2si): Likewise.
8350         (sse_cvtss2si_2): Likewise.
8351         (sse_cvtss2siq): Likewise.
8352         (sse_cvtss2siq_2): Likewise.
8353         (sse_cvttss2si): Likewise.
8354         (sse_cvtss2siq_2): Likewise.
8355         (float<sseintvecmodelower><mode>2): Likewise.
8356         (sse2_cvtsd2si_2): Likewise.
8357         (sse2_cvtsd2siq_2): Likewise.
8358         (*<plusminus_insn><mode>3): Likewise.
8359         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8360         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8361         (ashr<mode>3): Likewise.
8362         (<shift_insn><mode>3): Likewise.
8363         (avx2_<code><mode>3): Likewise.
8364         (*avx2_<code><mode>3): Likewise.
8365         (*andnot<mode>3): Likewise.
8366         (*<code><mode>3) <any_logic>: Likewise.
8367         (abs<mode>2): Likewise.
8368         (avx2_permvar<mode>): Likewise.
8369         (avx2_perm<mode>_1): Likewise.
8370         (*avx_vpermilp<mode>): Likewise.
8371         (avx_vpermilvar<mode>3): Likewise.
8372         (avx2_ashrv<mode>): Likewise.
8373         (avx2_<shift_insn>v<mode>): Likewise.
8374         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8375         -mavx512cd.
8376         * doc/rtl.texi: Document XImode.
8377
8378 2013-08-21  Jeff Law  <law@redhat.com>
8379
8380         * tree-flow.h (register_jump_thread): Pass vector of edges
8381         instead of each important edge.
8382         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8383         thread path into a vector and pass that to register_jump_thread.
8384         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8385         passed in edge vector to the current 3-edge form.
8386
8387         Revert:
8388         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8389
8390         * dce.c (fini_dce): Call df_analyze again just in case
8391         delete_unmarked_insns removed anything.
8392
8393 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8394
8395         * reload.h (struct reg_equivs): Rename to ..
8396         (struct reg_equivs_s): .. this.
8397
8398 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8399
8400         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8401
8402 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8403
8404         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8405
8406 2013-08-21  Jeff Law  <law@redhat.com>
8407
8408         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8409         simplify assignments too.  If the RHS collapses to a singleton
8410         range, then return the value for the range.
8411
8412 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8413
8414         * config/i386/sse.md (V16): Rename to...
8415         (VMOVE): this.
8416         (mov<mode>): Update iterator name.
8417         (*mov<mode>_internal): Ditto.
8418         (push<mode>1): Ditto.
8419         (movmisalign<mode>): Ditto.
8420
8421 2013-08-20  Jan Hubicka  <jh@suse.cz>
8422
8423         PR bootstrap/58186
8424         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8425         entry for direct edges.
8426         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8427
8428 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8429
8430         Revert my last two changes, r201865 and r201864:
8431
8432         Revert r201865:
8433         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8434
8435         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8436         instances can own GC refs.
8437
8438         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8439         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8440         (gcc::context::gt_pch_nx): Likewise.
8441         (gcc::context::gt_pch_nx):  Likewise.
8442         * ggc.h (gt_ggc_mx <T>): New.
8443         (gt_pch_nx_with_op <T>): New.
8444         (gt_pch_nx <T>): New.
8445         * passes.c (opt_pass::gt_ggc_mx): New.
8446         (opt_pass::gt_pch_nx): New.
8447         (opt_pass::gt_pch_nx_with_op): New.
8448         (pass_manager::gt_ggc_mx): New.
8449         (pass_manager::gt_pch_nx): New.
8450         (pass_manager::gt_pch_nx_with_op): New.
8451         (pass_manager::operator new): Use
8452         ggc_internal_cleared_alloc_stat rather than xcalloc.
8453         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8454         (pass_manager::gt_ggc_mx): New.
8455         (pass_manager::gt_pch_nx): New.
8456         (pass_manager::gt_pch_nx_with_op): New.
8457         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8458         (opt_pass::operator new): New.
8459         (opt_pass::gt_ggc_mx): New.
8460         (opt_pass::gt_pch_nx): New.
8461         (opt_pass::gt_pch_nx_with_op): New.
8462
8463         Revert r201864:
8464         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8465
8466         * Makefile.in (GTFILES): Add context.h.
8467         * context.c (gcc::context::operator new): New.
8468         (gcc::context::gt_ggc_mx): New.
8469         (gcc::context::gt_pch_nx): New.
8470         (gcc::context::gt_pch_nx): New.
8471         * context.h (gcc::context): Add GTY((user)) marking.
8472         (gcc::context::operator new): New.
8473         (gcc::context::gt_ggc_mx): New.
8474         (gcc::context::gt_pch_nx): New.
8475         (gcc::context::gt_pch_nx): New.
8476         (g): Add GTY marking.
8477         (gt_ggc_mx (gcc::context *)): New.
8478         (gt_pch_nx (gcc::context *)): New.
8479         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8480         void *cookie)): New.
8481         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8482
8483 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8484
8485         * dce.c (fini_dce): Call df_analyze again just in case
8486         delete_unmarked_insns removed anything.
8487
8488 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8489
8490         PR rtl-optimizations/57451
8491         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8492         from crossing split section boundaries.
8493
8494 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8495
8496         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8497         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8498         with MULTLIB_DEFAULTS.
8499
8500 2013-08-20  Nick Clifton  <nickc@redhat.com>
8501
8502         * target.def (narrow_volatile_bitfield): Note that the default
8503         value is false, not !TARGET_STRICT_ALIGN.
8504         * doc/tm.texi: Regenerate.
8505
8506 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8507
8508         Fix LIB_SPEC for systems without libpthread.
8509
8510         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8511         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8512         for Android.
8513         * config/i386/linux-common.h: Likewise.
8514         * config/mips/linux-common.h: Likewise.
8515
8516 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8517
8518         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8519         checks.
8520
8521 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8522
8523         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8524         instances can own GC refs.
8525
8526         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8527         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8528         (gcc::context::gt_pch_nx): Likewise.
8529         (gcc::context::gt_pch_nx):  Likewise.
8530         * ggc.h (gt_ggc_mx <T>): New.
8531         (gt_pch_nx_with_op <T>): New.
8532         (gt_pch_nx <T>): New.
8533         * passes.c (opt_pass::gt_ggc_mx): New.
8534         (opt_pass::gt_pch_nx): New.
8535         (opt_pass::gt_pch_nx_with_op): New.
8536         (pass_manager::gt_ggc_mx): New.
8537         (pass_manager::gt_pch_nx): New.
8538         (pass_manager::gt_pch_nx_with_op): New.
8539         (pass_manager::operator new): Use
8540         ggc_internal_cleared_alloc_stat rather than xcalloc.
8541         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8542         (pass_manager::gt_ggc_mx): New.
8543         (pass_manager::gt_pch_nx): New.
8544         (pass_manager::gt_pch_nx_with_op): New.
8545         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8546         (opt_pass::operator new): New.
8547         (opt_pass::gt_ggc_mx): New.
8548         (opt_pass::gt_pch_nx): New.
8549         (opt_pass::gt_pch_nx_with_op): New.
8550
8551 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8552
8553         * Makefile.in (GTFILES): Add context.h.
8554         * context.c (gcc::context::operator new): New.
8555         (gcc::context::gt_ggc_mx): New.
8556         (gcc::context::gt_pch_nx): New.
8557         (gcc::context::gt_pch_nx): New.
8558         * context.h (gcc::context): Add GTY((user)) marking.
8559         (gcc::context::operator new): New.
8560         (gcc::context::gt_ggc_mx): New.
8561         (gcc::context::gt_pch_nx): New.
8562         (gcc::context::gt_pch_nx): New.
8563         (g): Add GTY marking.
8564         (gt_ggc_mx (gcc::context *)): New.
8565         (gt_pch_nx (gcc::context *)): New.
8566         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8567         void *cookie)): New.
8568         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8569
8570 2013-08-20  Alan Modra  <amodra@gmail.com>
8571
8572         PR target/57865
8573         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8574         (rs6000_emit_epilogue): Likewise.
8575
8576 2013-08-19  Dehao Chen  <dehao@google.com>
8577
8578         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8579
8580 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8581             Jakub Jelinek  <jakub@redhat.com>
8582
8583         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8584         (BUILT_IN_FABSD64): Likewise.
8585         (BUILT_IN_FABSD128): Likewise.
8586         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8587         (fold_builtin_1): Likewise.
8588         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8589         destination and source operands.
8590         (*abstd2_fpr): Likewise.
8591         (*nabstd2_fpr): Likewise.
8592
8593 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8594
8595         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8596         JUMP_P and INSN_P.
8597
8598 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8599
8600         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8601         incomplete.
8602
8603 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8604
8605         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8606         * builtins.c (default_libc_has_function): New.
8607         (gnu_libc_has_function): Ditto.
8608         (no_c99_libc_has_function): Ditto.
8609         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8610         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8611         (fold_builtin_sincos): Likewise.
8612         (fold_builtin_cexp): Likewise.
8613         * builtins.def (DEF_C94_BUILTIN): Likewise.
8614         (DEF_C99_BUILTIN): Likewise.
8615         (DEF_C99_C90RES_BUILTIN): Likewise.
8616         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8617         definitions to using this define.
8618         * config/darwin-protos.h (darwin_libc_has_function): New.
8619         * config/darwin.c (darwin_libc_has_function): Ditto.
8620         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8621         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8622         * config/darwin.h: Ditto.
8623         * config/elfos.h: Ditto.
8624         * config/freebsd.h: Ditto.
8625         * config/i386/cygming.h: Ditto.
8626         * config/i386/djgpp.h: Ditto.
8627         * config/i386/i386-interix.h: Ditto.
8628         * config/microblaze/microblaze.h: Ditto.
8629         * config/mmix/mmix.h: Ditto.
8630         * config/gnu-user.h: Ditto.
8631         * config/ia64/hpux.h: Ditto.
8632         * config/pa/pa-hpux.h: Ditto.
8633         * config/pdp11/pdp11.h: Ditto.
8634         * config/picochip/picochip.h: Ditto.
8635         * config/linux.h: Ditto.
8636         * config/netbsd.h: Ditto.
8637         * config/openbsd.h: Ditto.
8638         * config/rs6000/aix43.h: Ditto.
8639         * config/rs6000/aix51.h: Ditto.
8640         * config/rs6000/aix52.h: Ditto.
8641         * config/rs6000/aix53.h: Ditto.
8642         * config/rs6000/aix61.h: Ditto.
8643         * config/rs6000/darwin.h: Ditto.
8644         * config/rs6000/linux.h: Ditto.
8645         * config/rs6000/linux64.h: Ditto.
8646         * config/s390/tpf.h: Ditto.
8647         * config/sol2-10.h: Ditto.
8648         * config/sol2.h: Ditto.
8649         * config/vms/vms.h: Ditto.
8650         * config/vxworks.h: Ditto.
8651         * config/linux-android.c (linux_android_libc_has_function):
8652         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8653         * config/linux-protos.h (linux_android_libc_has_function):
8654         New declaration.
8655         * config/i386/i386.c (ix86_libc_has_function): New.
8656         * config/i386/i386-protos.h
8657         (ix86_libc_has_function): New declaration.
8658         * config/i386/i386.md
8659         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8660         ("isinf<mode>2): Likewise.
8661         * convert.c (convert_to_integer): Using new target hook
8662         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8663         TARGET_C99_FUNCTIONS.
8664         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8665         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8666         * coretypes.h (function_class): New enum for different
8667         classes of functions.
8668         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8669         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8670         (TARGET_HAS_SINCOS): Likewise.
8671         (TARGET_LIBC_HAS_FUNCTION): New.
8672         * doc/tm.texi: Regenerated.
8673         * targhooks.h (default_libc_has_function): New declaration.
8674         (no_c99_libc_has_function): Ditto.
8675         (gnu_libc_has_function): Ditto.
8676         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8677         and TARGET_HAS_SINCOS.
8678
8679 2013-08-18  Jan Hubicka  <jh@suse.cz>
8680
8681         * Makeifle-in (ipa-devirt.o): New.
8682         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8683         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8684         (analyze_functions): Look into possible targets of polymorphic call.
8685         * dumpfile.c (dump_files): Add type-inheritance dump.
8686         * dumpfile.h (TDI_inheritance): New.
8687         * ipa-devirt.c: New file.
8688         * ipa-utils.h (odr_type_d): Forward declare.
8689         (odr_type): New type.
8690         (build_type_inheritance_graph): Declare.
8691         (possible_polymorphic_call_targets): Declare and introduce inline
8692         variant when only edge is pased.
8693         (dump_possible_polymorphic_call_targets): Likewise.
8694         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8695         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8696         (types_same_for_odr): ... here.
8697         * tree.h (type_in_anonymous_namespace_p): Declare.
8698
8699 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8700
8701         PR tree-optimization/58006
8702         * tree-parloops.c (take_address_of): Don't ICE if get_name
8703         returns NULL.
8704         (eliminate_local_variables_stmt): Remove clobber stmts.
8705
8706 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8707
8708         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8709         error message is issued for an alias to undefined symbol.
8710
8711 2013-08-18  Jan Hubicka  <jh@suse.cz>
8712
8713         * cgraph.c (cgraph_create_indirect_edge): Discover
8714         polymorphic calls and record basic info into indirect_info.
8715         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8716         devirtualization, ignore objc function calls.
8717         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
8718         call with no parm index info.
8719         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
8720         * tree.c (virtual_method_call_p): New function.
8721         * tree.h (virtual_method_call_p): Declare.
8722
8723 2013-08-16  Jan Hubicka  <jh@suse.cz>
8724
8725         PR middle-end/58179
8726         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
8727
8728 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
8729
8730         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
8731         attribute.
8732
8733 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8734
8735         * gengtype.c (type_for_name): Add special-case support for
8736         locating types within the "gcc::" namespace.
8737         (open_base_files): Emit a "using namespace gcc" directive.
8738
8739 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8740
8741         PR target/58160
8742         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
8743         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
8744
8745         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
8746         array instead of each individual operand as a separate argument.
8747         (emit_fusion_gpr_load): Likewise.
8748         (expand_fusion_gpr_load): Add new function declaration.
8749
8750         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
8751         signature to have the operands passed as an array, instead of as
8752         separate arguments.  Allow ZERO_EXTEND to be in the memory
8753         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
8754         depend on the register live/dead flags when peepholes are run.
8755         (expand_fusion_gpr_load): New function to be called from the
8756         peephole2 pass, to change the register that addis sets to be the
8757         target register.
8758         (emit_fusion_gpr_load): Change the calling signature to have the
8759         operands passed as an array, instead of as separate arguments.
8760         Allow ZERO_EXTEND to be in the memory address, and also
8761         SIGN_EXTEND if -mpower8-fusion-sign.
8762
8763         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
8764         unspec enumeration.
8765         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
8766         adjust the register addis loads up in the peephole2 pass.  Do not
8767         depend on the register live/dead state when the peephole pass is done.
8768
8769 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8770
8771         * gengtype.c (create_user_defined_type): Ensure that the kind
8772         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
8773         declaration is seen before the GTY((user)) marking.
8774
8775 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8776
8777         PR target/58105
8778         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
8779
8780 2013-08-16  Jan Hubicka  <jh@suse.cz>
8781
8782         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
8783         arugment expected_type.
8784         (gimple_fold_call): Use it.
8785         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
8786         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
8787         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
8788         (try_make_edge_direct_virtual_call): Likewise.
8789         * tree.c (obj_type_ref_class): New.
8790         * tree.h (obj_type_ref_class): Use it.
8791
8792 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8793
8794         * sched-vis.c (rtl_slim_pp_initialized): Remove.
8795         (rtl_slim_pp): Likewise.
8796         (init_rtl_slim_pretty_print): Likewise.
8797         (dump_value_slim):  Don't call it.  Use local pretty printer.
8798         (dump_insn_slim): Likewise.
8799         (dump_rtl_slim): Likewise.
8800         (str_pattern_slim): Likewise.
8801         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
8802         Simplify.
8803
8804 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
8805
8806         PR tree-optimization/58164
8807         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
8808         walk gimple_goto_dest of GIMPLE_GOTO.
8809
8810         PR tree-optimization/58165
8811         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
8812         bi_call must be the last stmt in a bb, don't split_block, instead
8813         use fallthru edge from it and give up if there is none.
8814         Release conds vector when returning early.
8815
8816 2013-08-14  Xinliang David Li  <davidxl@google.com>
8817
8818         * config/i386/i386.c (ix86_option_override_internal):
8819         Remove unused variable and field.
8820
8821 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8822
8823         PR target/57949
8824         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
8825         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
8826         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
8827         and Linux, correct BLKmode alignment when 128-bit alignment is
8828         required and compatibility flag is not set.
8829         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
8830         for zero-size arguments when compatibility flag is not set.
8831
8832 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
8833
8834         PR tree-optimization/58145
8835         * tree-sra.c (build_ref_for_offset): If prev_base has
8836         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
8837
8838 2013-08-14  Xinliang David Li  <davidxl@google.com>
8839
8840         * config/i386/i386.c (ix86_option_override_internal):
8841         Fix uninitialized variable error.
8842
8843 2013-08-14  Xinliang David Li  <davidxl@google.com>
8844
8845         * config/i386/i386.opt: Define two new options.
8846         * config/i386/x86-tune.def: Add arch selector field in macros.
8847         * config/i386/i386.h: Adjust macro definition.
8848         * config/i386/i386.c (ix86_option_override_internal):
8849         Refactor the code.
8850         (parse_mtune_ctrl_str): New function.
8851         (set_ix86_tune_features): New function.
8852         (ix86_function_specific_restore): Call the new helper function.
8853
8854 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
8855
8856         PR rtl-optimization/57662
8857         * sel-sched.c (code_motion_process_successors): When the current insn
8858         is removed after the recursive traversal, break from the loop.
8859         Add comments and debug printouts.
8860
8861 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
8862             Alexandre Oliva  <aoliva@redhat.com>
8863
8864         PR target/58067
8865         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
8866         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
8867         there also UNSPEC_PLTOFF.
8868
8869 2013-08-14  Marek Polacek  <polacek@redhat.com>
8870
8871         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
8872         NUM_CONDITIONS bit positions.
8873
8874 2013-08-13  Cary Coutant  <ccoutant@google.com>
8875
8876         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
8877         (attr_checksum): Hash vector contents instead of pointer.
8878         (attr_checksum_ordered): Likewise.
8879
8880 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
8881
8882         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
8883         when Pmode != word_mode.  Add length_address attribute.
8884         (sse3_monitor_<mode>): Merge from sse3_monitor and
8885         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
8886         Pmode != word_mode.  Update insn length attribute.
8887         * config/i386/i386.c (ix86_option_override_internal): Update
8888         ix86_gen_monitor selection for merged sse3_monitor insn.
8889
8890 2013-08-13  Julian Brown  <julian@codesourcery.com>
8891
8892         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
8893         perform invalid legitimization on greater-than-word-size modes for
8894         TARGET_E500_DOUBLE.
8895
8896 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
8897
8898         * ira.c (setup_class_translate_array): Use aclass instead of cl
8899         for classes not fully covered by allocno classes.
8900
8901 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
8902
8903         PR tree-optimization/57661
8904         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
8905         * tree-inline.c (tree_function_versioning): Initialize it.
8906         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
8907         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
8908         that is not being copied.
8909
8910         PR sanitizer/56417
8911         * asan.c (instrument_strlen_call): Fix typo in comment.
8912         Use char * type even for the lhs of POINTER_PLUS_EXPR.
8913
8914 2013-08-13  Steve Ellcey  <sellcey@mips.com>
8915
8916         * config/mips/mips.md (prefetch): Use lw instead of ld on
8917         loongson in 32bit mode.
8918
8919 2013-08-13  Nick Clifton  <nickc@redhat.com>
8920
8921         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
8922
8923 2013-08-13  Jan Hubicka  <jh@suse.cz>
8924
8925         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
8926         introduced edge; fix typo in sanity check.
8927         (cgraph_resolve_speculation): Export; improve diagnostic.
8928         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
8929         speculation at type mismatch.
8930         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
8931         (cgraph_resolve_speculation): Declare.
8932         (symtab_can_be_discarded): New function.
8933         * value-prof.c (gimple_ic_transform): Remove actual transform code.
8934         * ipa-inline-transform.c (speculation_removed): New global var.
8935         (clone_inlined_nodes): See if speculation can be removed.
8936         (inline_call): If speculations was removed, we growths may not match.
8937         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
8938         (speculation_useful_p): New function.
8939         (resolve_noninline_speculation): New function.
8940         (inline_small_functions): Resolve useless speculations.
8941         * ipa-inline.h (speculation_useful_p): Declare
8942         * ipa.c (can_replace_by_local_alias): Simplify.
8943         (ipa_profile): Produce speculative calls in non-lto, too;
8944         add simple cost model; produce local aliases.
8945
8946 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
8947
8948         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
8949         PASS_MANAGER_H.
8950
8951 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
8952
8953         * config/i386/i386.c (ix86_function_versions): Use error + inform.
8954
8955 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
8956
8957         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
8958         iterator instead of X87MODEF.
8959
8960 2013-08-12  Perez Read  <netfirewall@gmail.com>
8961
8962         PR target/58132
8963         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
8964         operand 0 for intel asm alternative.
8965         (*movabs<mode>_2): Ditto for operand 1.
8966
8967 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
8968
8969         * config/aarch64/arm_none.h
8970         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
8971
8972 2013-08-12  Nick Clifton  <nickc@redhat.com>
8973
8974         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
8975
8976 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8977
8978         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
8979         expand for QI/HImode operand to produce more effictive code for
8980         unsigned char(short) --> float(double) conversion.
8981
8982 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
8983
8984         * doc/invoke.texi: Mention that -ftls-model does not force the final
8985         model.
8986
8987 2013-08-12  Marek Polacek  <polacek@redhat.com>
8988             Marc Glisse  <marc.glisse@inria.fr>
8989
8990         PR tree-optimization/57980
8991         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
8992         when creating -1 constant.
8993
8994 2013-08-10  Jan Hubicka  <jh@suse.cz>
8995
8996         Workaround binutils PR14342.
8997         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
8998         (gimple_init_edge_profiler): Likewise.
8999         (gimple_gen_ic_func_profiler): Likewise.
9000
9001 2013-08-09  Jan Hubicka  <jh@suse.cz>
9002
9003         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9004
9005 2013-08-09  Xinliang David Li  <davidxl@google.com>
9006
9007         * config/i386/stringop.def: New file.
9008         * config/i386/stringop.opt: New file.
9009         * config/i386/i386-opts.h: Include stringopt.def.
9010         * config/i386/i386.opt: Include stringopt.opt.
9011         * config/i386/i386.c (ix86_option_override_internal):
9012         Override default size based stringop inline strategies with options.
9013         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9014         New function.
9015
9016 2013-08-09  Jan Hubicka  <jh@suse.cz>
9017
9018         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9019
9020 2013-08-09  Jan Hubicka  <jh@suse.cz>
9021
9022         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9023         CGRAPH_FREQ_MAX.
9024         (dump_cgraph_node): Dump profile-id.
9025         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9026         and common_target_probability.
9027         * lto-cgraph.c (lto_output_edge): Stream common targets.
9028         (lto_output_node): Stream profile ids.
9029         (input_node): Stream profile ids.
9030         (input_edge): Stream common targets.
9031         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9032         * ipa.c: Include value-prof.h
9033         (ipa_profile_generate_summary): Turn indirect call statement histograms
9034         into common targets.
9035         (ipa_profile): Turn common targets into speculative edges.
9036
9037 2013-08-09  Jan Hubicka  <jh@suse.cz>
9038
9039         * cgraph.h (cgraph_node): Add profile_id.
9040         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9041         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9042         (del_node_map): Update.
9043         (find_func_by_funcdef_no): Replace by ...
9044         (find_func_by_profile_id): ... this one.
9045         (gimple_ic_transform): Do not remove useful histograms when
9046         speculation is not done; dump info when indirect call removal
9047         can happen at LTO.
9048         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9049         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9050         (__gcov_indirect_call_profiler_v2): .. this one.
9051         * profile.h (init_node_map): Update.
9052         * coverage.c (coverage_compute_profile_id): New function.
9053         * coverage.h (coverage_compute_profile_id): Declare.
9054         * tree-profile.c (init_ic_make_global_vars): Make
9055         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9056         (gimple_init_edge_profiler): Update prototype of
9057         __gcov_indirect_call_profiler.
9058         (gimple_gen_ic_func_profiler): Simplify.
9059         (tree_profiling): Use init_node_map
9060
9061 2013-08-09  Jan Hubicka  <jh@suse.cz>
9062
9063         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9064         non-speculative refs.
9065         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9066         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9067         (cgraph_set_call_stmt): Likewise.
9068         (cgraph_create_edge_1): Fix release checking compilatoin;
9069         clear lto_stmt_uid.
9070         (cgraph_free_edge): Free indirect info.
9071         (cgraph_turn_edge_to_speculative): New function.
9072         (cgraph_speculative_call_info): New function.
9073         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9074         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9075         (dump_cgraph_node): Dump speculation.
9076         (verify_edge_count_and_frequency): Accept speculative edges.
9077         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9078         (verify_cgraph_node): Handle speculation.
9079         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9080         (cgraph_set_call_stmt): Update prototype.
9081         (cgraph_make_edge_direct): Update prototype.
9082         (cgraph_speculative_call_info): Declare.
9083         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9084         to change; update call of ipa_find_references.
9085         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9086         lto_stmt_uid and speculative flags.
9087         (ipa_dump_references): Dump speculation.
9088         (ipa_clone_references): Clone speculative flag.
9089         (ipa_clone_referring): Likewise.
9090         (ipa_clone_ref): New function.
9091         (ipa_find_reference): Look into lto_stmt_uids
9092         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9093         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9094         (ipa_find_reference): Update declaration.
9095         (ipa_clone_ref): Declare.
9096         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9097         stream speculative flag.
9098         (lto_output_ref): Stream statements uids and speculation.
9099         (input_ref): Likewise.
9100         (input_edge): Stream speuclation.
9101         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9102         (cgraph_set_call_stmt_including_clones): Handle speculation.
9103         * ipa-inline.c (heap_edge_removal_hook): New function.
9104         (inline_small_functions): Register it.
9105         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9106         also initialize refs.
9107         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9108         edge to change.
9109         (try_make_edge_direct_simple_call): Likewise.
9110         (try_make_edge_direct_simple_call): Likewise.
9111         (update_indirect_edges_after_inlining): Likewise.
9112         (remove_described_reference): Look proper lto_stmt_uid.
9113         (propagate_controlled_uses): Likewise.
9114         (propagate_controlled_uses): Liekwise.
9115         * tree-inline.c (copy_bb): Copy speculative edges.
9116         (redirect_all_calls): New function.
9117         (copy_cfg_body): Do redirection after loop info is updated.
9118         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9119
9120 2013-08-09  Jan Hubicka  <jh@suse.cz>
9121
9122         * lto-streamer-out.c (output_function): Renumber PHIs.
9123         * lto-streamer-in.c (input_function): Likewise.
9124
9125 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9126
9127         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9128         (get_lane_unsigned): Likewise.
9129         (dup_lane_scalar): Likewise.
9130         (get_lane): enable for VALL.
9131         * config/aarch64/aarch64-simd.md
9132         (aarch64_dup_lane_scalar<mode>): Remove.
9133         (aarch64_get_lane_signed<mode>): Likewise.
9134         (aarch64_get_lane_unsigned<mode>): Likewise.
9135         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9136         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9137         (aarch64_get_lane<mode>): Enable for all vector modes.
9138         (aarch64_get_lanedi): Remove misleading constraints.
9139         * config/aarch64/arm_neon.h
9140         (__aarch64_vget_lane_any): Define.
9141         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9142         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9143         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9144         * config/aarch64/iterators.md (VDQQH): New.
9145         (VDQQHS): Likewise.
9146         (vwcore): Likewise.
9147
9148 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9149
9150         * configure.ac: Add GAS check for LEON instructions on SPARC.
9151         * configure: Regenerate.
9152         * config.in: Likewise.
9153         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9154         sparc*-*-* block.
9155         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9156         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9157         for LEON or LEON3.
9158         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9159         (AS_LEON_FLAG): New macro.
9160         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9161         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9162         Deal with LEON and LEON3 for the memory model.
9163         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9164         (atomic_compare_and_swap<mode>_1): Likewise.
9165         (*atomic_compare_and_swap<mode>_1): Likewise.
9166
9167 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9168
9169         * config/arm/neon.md (vcond): Fix floating-point vector
9170         comparisons against 0.
9171
9172 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9173
9174         * lra-constraints.c (emit_spill_move): Remove assert.
9175         (process_alt_operands): Add more debugging
9176         output.  Increase reject for spilling into memory.  Decrease
9177         reject for reloading scratch.
9178         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9179
9180 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9181
9182         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9183         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9184         micromips incompatible.  Add nan2008.
9185         (MULTILIB_DIRNAMES): Add nan2008.
9186         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9187         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9188         and micromips incompatible.  Add nan2008.
9189         (MULTILIB_DIRNAMES): Add nan2008.
9190         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9191
9192 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9193
9194         PR rtl-optimization/58079
9195         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9196         simplify_comparison has widened a comparison with an integer.
9197
9198 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9199
9200         * config/arm/neon.md (movmisalign<mode>): Disable when we
9201         don't allow unaligned accesses.
9202         (*movmisalign<mode>_neon_store): Likewise.
9203         (*movmisalign<mode>_neon_load): Likewise.
9204         (*movmisalign<mode>_neon_store): Likewise.
9205         (*movmisalign<mode>_neon_load): Likewise.
9206
9207 2013-08-08  Jan Hubicka  <jh@suse.cz>
9208
9209         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9210         (make_pass_rebuild_cgraph_edges): Also clear references.
9211         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9212         * ipa-inline-transform.c (inline_transform): Remove all references
9213         after inlining.
9214         * cgraphunit.c (expand_function): Remove all references after
9215         expansion.
9216         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9217         (ipa_find_reference): Rewrite to iterator.
9218         (remove_stmt_references): Likewise.
9219         (ipa_clear_stmts_in_references): New function.
9220         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9221         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9222         clear references.
9223         * ipa-split.c (split_function): Remove references in split function.
9224
9225 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9226
9227         PR target/57431
9228         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9229         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9230
9231 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9232
9233         PR target/56979
9234         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9235         suggested mode for the assignment isn't compatible with the
9236         registers required.
9237
9238 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9239
9240         PR target/58065
9241         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9242
9243 2013-08-07  Xinliang David Li  <davidxl@google.com>
9244
9245         * config/i386/i386.opt: New option -mtune-ctrl=.
9246         * config/i386/x86-tune.def: New file.
9247         * config/i386/i386.h: include x86-tune.def.
9248         * config/i386/i386.c (ix86_option_override_internal):
9249         Parsing -mtune-ctrl= option and set tune features.
9250
9251 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9252
9253         PR other/12081
9254         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9255         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9256         to gen_2arg_fn_t.
9257
9258 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9259
9260         * rtl.h (update_alignments): Declare.
9261         * final.c (grow_label_align): New function extracted from...
9262         (shorten_branches): ...here.  Call it.
9263         (update_alignments): New function.
9264         * reorg.c (sibling_labels): New variable.
9265         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9266         the new label along with it onto the sibling_labels vector.
9267         (fill_simple_delay_slots): Adjust call to get_label_before.
9268         (fill_slots_from_thread): Likewise.
9269         (relax_delay_slots): Likewise.
9270         (make_return_insns): Likewise.
9271         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9272
9273 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9274
9275         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9276         document its semantics.
9277         (diagnostic_report_diagnostic): Adjust accordingly.
9278
9279 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9280
9281         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9282         (sparc_option_override): ...and port to new C++ pass API.
9283         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9284
9285 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9286
9287         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9288
9289 2013-08-06  Caroline Tice  <cmtice@google.com>
9290
9291         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9292         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9293         * tree-pass.h: Add pass_vtable_verify.
9294         * varasm.c (assemble_variable): Add code to properly set the comdat
9295         section and name for the .vtable_map_vars section.
9296         (assemble_vtyv_preinit_initializer): New function.
9297         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9298         LINK_ONCE flag.
9299         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9300         * vtable-verify.c: New file.
9301         * vtable-verify.h: New file.
9302         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9303         initialiation levels.
9304         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9305         * passes.def: Insert pass_vtable_verify.
9306         * aclocal.m4: Reorder includes.
9307         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9308         -fvtv-counts options.
9309         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9310         as appropriate, if -fvtable-verify=... is used.
9311         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9312         -fvtable-verify=... is used.
9313         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9314         (vtable-verify.o): Add new build rule.
9315         (GTFILES): Add vtable-verify.c to list.
9316         * common.opt (fvtable-verify=): New flag.
9317         (vtv_priority): Values for fvtable-verify= flag.
9318         (fvtv-counts): New flag.
9319         (fvtv-debug): New flag.
9320         * tree.h (save_vtable_map_decl): New extern function decl.
9321
9322 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9323
9324         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9325         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9326         (pass_data_rl78_devirt): ...new pass_data instance and...
9327         (make_pass_rl78_devirt): ...new function.
9328         (rl78_asm_file_start): Port pass registration to new C++ API.
9329
9330 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9331
9332         * coretypes.h (rtl_opt_pass): Add.
9333         (gcc::context): Add.
9334         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9335         (epiphany_init): Port to new C++ pass API.
9336         (epiphany_optimize_mode_switching): Likewise.
9337         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9338         (pass_manager::get_pass_mode_switching): New.
9339         (pass_manager::get_pass_peephole2): New.
9340         * mode-switching.c (pass_mode_switching): Add clone method.
9341         * recog.c (pass_peephole2): Add clone method.
9342         (pass_split_all_insns): Add clone method.
9343
9344 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9345
9346         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9347         (mips_option_override): ...here, porting to new C++ API for passes.
9348
9349 2013-08-06  Jan Hubicka  <jh@suse.cz>
9350
9351         * cgraph.c (cgraph_get_body): New function based on lto.c
9352         implementation.
9353         * cgraph.h (cgraph_get_body): Declare.
9354         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9355         LTO paths.
9356         * cgraphunit.c (expand_function): Get body prior expanding.
9357         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9358         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9359         really need.
9360         * passes.c (do_per_function_toporder): Get body.
9361         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9362         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9363
9364 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9365
9366         PR fortran/57987
9367         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9368         is not re-finalized.  Rename second parameter to no_collect.
9369
9370 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9371
9372         PR middle-end/58041
9373         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9374         MEM_REF has proper alignment information.
9375
9376 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9377
9378         PR other/12081
9379         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9380         class insn_gen_fn.
9381         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9382         rtx (*) (rtx, ...) with insn_gen_fn.
9383         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9384         insn_gen_fn::stored_funcptr.  Add initializer braces.
9385
9386 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9387
9388         Rewrite how instances of passes are cloned to remove assumptions
9389         about their sizes (thus allowing pass subclasses to have
9390         additional data fields, albeit non-GC-managed ones at this point).
9391
9392         * passes.c (make_pass_instance): Now that passes have clone
9393         methods, rewrite this function to eliminate XNEW and memcpy
9394         calls that used hardcoded sizes.  Since this function no longer
9395         creates pass instances, rename it to...
9396         (add_pass_instance): ...this.  Document the old way that passes were
9397         numbered and flagged, and rework this function to continue using it.
9398         (next_pass_1): Add an initial_pass argument for use by
9399         add_pass_instance.
9400         (position_pass): When adding multiple instances of a pass, use
9401         the pass's clone method, rather than relying on the XNEW/memcpy
9402         within the former make_pass_instance (now add_pass_instance).
9403         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9404         the initial instance of the current pass within the pass manager.
9405
9406 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9407
9408         This is the automated part of the conversion of passes from C
9409         structs to C++ classes.
9410
9411         Patch autogenerated by refactor_passes.py from
9412         https://github.com/davidmalcolm/gcc-refactoring-scripts
9413         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9414
9415         * asan.c (pass_asan): Convert from a global struct to a subclass of
9416         gimple_opt_pass along with...
9417         (pass_data_asan): ...new pass_data instance and...
9418         (make_pass_asan): ...new function.
9419         (pass_asan_O0): Convert from a global struct to a subclass of
9420         gimple_opt_pass along with...
9421         (pass_data_asan_O0): ...new pass_data instance and...
9422         (make_pass_asan_O0): ...new function.
9423         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9424         subclass of rtl_opt_pass along with...
9425         (pass_data_inc_dec): ...new pass_data instance and...
9426         (make_pass_inc_dec): ...new function.
9427         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9428         a subclass of rtl_opt_pass along with...
9429         (pass_data_reorder_blocks): ...new pass_data instance and...
9430         (make_pass_reorder_blocks): ...new function.
9431         (pass_duplicate_computed_gotos): Convert from a global struct to a
9432         subclass of rtl_opt_pass along with...
9433         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9434         (make_pass_duplicate_computed_gotos): ...new function.
9435         (pass_partition_blocks): Convert from a global struct to a subclass of
9436         rtl_opt_pass along with...
9437         (pass_data_partition_blocks): ...new pass_data instance and...
9438         (make_pass_partition_blocks): ...new function.
9439         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9440         struct to a subclass of rtl_opt_pass along with...
9441         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9442         and...
9443         (make_pass_branch_target_load_optimize1): ...new function.
9444         (pass_branch_target_load_optimize2): Convert from a global struct to a
9445         subclass of rtl_opt_pass along with...
9446         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9447         and...
9448         (make_pass_branch_target_load_optimize2): ...new function.
9449         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9450         of rtl_opt_pass along with...
9451         (pass_data_jump): ...new pass_data instance and...
9452         (make_pass_jump): ...new function.
9453         (pass_jump2): Convert from a global struct to a subclass of
9454         rtl_opt_pass along with...
9455         (pass_data_jump2): ...new pass_data instance and...
9456         (make_pass_jump2): ...new function.
9457         * cfgexpand.c (pass_expand): Convert from a global struct to a
9458         subclass of rtl_opt_pass along with...
9459         (pass_data_expand): ...new pass_data instance and...
9460         (make_pass_expand): ...new function.
9461         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9462         of rtl_opt_pass along with...
9463         (pass_data_free_cfg): ...new pass_data instance and...
9464         (make_pass_free_cfg): ...new function.
9465         (pass_into_cfg_layout_mode): Convert from a global struct to a
9466         subclass of rtl_opt_pass along with...
9467         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9468         (make_pass_into_cfg_layout_mode): ...new function.
9469         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9470         subclass of rtl_opt_pass along with...
9471         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9472         (make_pass_outof_cfg_layout_mode): ...new function.
9473         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9474         struct to a subclass of gimple_opt_pass along with...
9475         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9476         (make_pass_build_cgraph_edges): ...new function.
9477         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9478         subclass of gimple_opt_pass along with...
9479         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9480         (make_pass_rebuild_cgraph_edges): ...new function.
9481         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9482         subclass of gimple_opt_pass along with...
9483         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9484         and...
9485         (make_pass_remove_cgraph_callee_edges): ...new function.
9486         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9487         struct to a subclass of rtl_opt_pass along with...
9488         (pass_data_stack_adjustments): ...new pass_data instance and...
9489         (make_pass_stack_adjustments): ...new function.
9490         * combine.c (pass_combine): Convert from a global struct to a subclass
9491         of rtl_opt_pass along with...
9492         (pass_data_combine): ...new pass_data instance and...
9493         (make_pass_combine): ...new function.
9494         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9495         global struct to a subclass of rtl_opt_pass along with...
9496         (pass_data_compare_elim_after_reload): ...new pass_data instance
9497         and...
9498         (make_pass_compare_elim_after_reload): ...new function.
9499         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9500         of rtl_opt_pass along with...
9501         (pass_data_rtl_cprop): ...new pass_data instance and...
9502         (make_pass_rtl_cprop): ...new function.
9503         * cse.c (pass_cse): Convert from a global struct to a subclass of
9504         rtl_opt_pass along with...
9505         (pass_data_cse): ...new pass_data instance and...
9506         (make_pass_cse): ...new function.
9507         (pass_cse2): Convert from a global struct to a subclass of
9508         rtl_opt_pass along with...
9509         (pass_data_cse2): ...new pass_data instance and...
9510         (make_pass_cse2): ...new function.
9511         (pass_cse_after_global_opts): Convert from a global struct to a
9512         subclass of rtl_opt_pass along with...
9513         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9514         (make_pass_cse_after_global_opts): ...new function.
9515         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9516         of rtl_opt_pass along with...
9517         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9518         (make_pass_ud_rtl_dce): ...new function.
9519         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9520         rtl_opt_pass along with...
9521         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9522         (make_pass_fast_rtl_dce): ...new function.
9523         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9524         a subclass of rtl_opt_pass along with...
9525         (pass_data_df_initialize_opt): ...new pass_data instance and...
9526         (make_pass_df_initialize_opt): ...new function.
9527         (pass_df_initialize_no_opt): Convert from a global struct to a
9528         subclass of rtl_opt_pass along with...
9529         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9530         (make_pass_df_initialize_no_opt): ...new function.
9531         (pass_df_finish): Convert from a global struct to a subclass of
9532         rtl_opt_pass along with...
9533         (pass_data_df_finish): ...new pass_data instance and...
9534         (make_pass_df_finish): ...new function.
9535         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9536         rtl_opt_pass along with...
9537         (pass_data_rtl_dse1): ...new pass_data instance and...
9538         (make_pass_rtl_dse1): ...new function.
9539         (pass_rtl_dse2): Convert from a global struct to a subclass of
9540         rtl_opt_pass along with...
9541         (pass_data_rtl_dse2): ...new pass_data instance and...
9542         (make_pass_rtl_dse2): ...new function.
9543         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9544         subclass of rtl_opt_pass along with...
9545         (pass_data_dwarf2_frame): ...new pass_data instance and...
9546         (make_pass_dwarf2_frame): ...new function.
9547         * except.c (pass_set_nothrow_function_flags): Convert from a global
9548         struct to a subclass of rtl_opt_pass along with...
9549         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9550         and...
9551         (make_pass_set_nothrow_function_flags): ...new function.
9552         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9553         subclass of rtl_opt_pass along with...
9554         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9555         and...
9556         (make_pass_convert_to_eh_region_ranges): ...new function.
9557         * final.c (pass_compute_alignments): Convert from a global struct to a
9558         subclass of rtl_opt_pass along with...
9559         (pass_data_compute_alignments): ...new pass_data instance and...
9560         (make_pass_compute_alignments): ...new function.
9561         (pass_final): Convert from a global struct to a subclass of
9562         rtl_opt_pass along with...
9563         (pass_data_final): ...new pass_data instance and...
9564         (make_pass_final): ...new function.
9565         (pass_shorten_branches): Convert from a global struct to a subclass of
9566         rtl_opt_pass along with...
9567         (pass_data_shorten_branches): ...new pass_data instance and...
9568         (make_pass_shorten_branches): ...new function.
9569         (pass_clean_state): Convert from a global struct to a subclass of
9570         rtl_opt_pass along with...
9571         (pass_data_clean_state): ...new pass_data instance and...
9572         (make_pass_clean_state): ...new function.
9573         * function.c (pass_instantiate_virtual_regs): Convert from a global
9574         struct to a subclass of rtl_opt_pass along with...
9575         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9576         (make_pass_instantiate_virtual_regs): ...new function.
9577         (pass_leaf_regs): Convert from a global struct to a subclass of
9578         rtl_opt_pass along with...
9579         (pass_data_leaf_regs): ...new pass_data instance and...
9580         (make_pass_leaf_regs): ...new function.
9581         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9582         subclass of rtl_opt_pass along with...
9583         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9584         and...
9585         (make_pass_thread_prologue_and_epilogue): ...new function.
9586         (pass_match_asm_constraints): Convert from a global struct to a
9587         subclass of rtl_opt_pass along with...
9588         (pass_data_match_asm_constraints): ...new pass_data instance and...
9589         (make_pass_match_asm_constraints): ...new function.
9590         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9591         subclass of rtl_opt_pass along with...
9592         (pass_data_rtl_fwprop): ...new pass_data instance and...
9593         (make_pass_rtl_fwprop): ...new function.
9594         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9595         rtl_opt_pass along with...
9596         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9597         (make_pass_rtl_fwprop_addr): ...new function.
9598         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9599         rtl_opt_pass along with...
9600         (pass_data_rtl_pre): ...new pass_data instance and...
9601         (make_pass_rtl_pre): ...new function.
9602         (pass_rtl_hoist): Convert from a global struct to a subclass of
9603         rtl_opt_pass along with...
9604         (pass_data_rtl_hoist): ...new pass_data instance and...
9605         (make_pass_rtl_hoist): ...new function.
9606         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9607         subclass of gimple_opt_pass along with...
9608         (pass_data_lower_cf): ...new pass_data instance and...
9609         (make_pass_lower_cf): ...new function.
9610         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9611         from a global struct to a subclass of gimple_opt_pass along with...
9612         (pass_data_strength_reduction): ...new pass_data instance and...
9613         (make_pass_strength_reduction): ...new function.
9614         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9615         of rtl_opt_pass along with...
9616         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9617         (make_pass_rtl_ifcvt): ...new function.
9618         (pass_if_after_combine): Convert from a global struct to a subclass of
9619         rtl_opt_pass along with...
9620         (pass_data_if_after_combine): ...new pass_data instance and...
9621         (make_pass_if_after_combine): ...new function.
9622         (pass_if_after_reload): Convert from a global struct to a subclass of
9623         rtl_opt_pass along with...
9624         (pass_data_if_after_reload): ...new pass_data instance and...
9625         (make_pass_if_after_reload): ...new function.
9626         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9627         a subclass of rtl_opt_pass along with...
9628         (pass_data_initialize_regs): ...new pass_data instance and...
9629         (make_pass_initialize_regs): ...new function.
9630         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9631         of ipa_opt_pass_d along with...
9632         (pass_data_ipa_cp): ...new pass_data instance and...
9633         (make_pass_ipa_cp): ...new function.
9634         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9635         global struct to a subclass of gimple_opt_pass along with...
9636         (pass_data_inline_parameters): ...new pass_data instance and...
9637         (make_pass_inline_parameters): ...new function.
9638         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9639         subclass of gimple_opt_pass along with...
9640         (pass_data_early_inline): ...new pass_data instance and...
9641         (make_pass_early_inline): ...new function.
9642         (pass_ipa_inline): Convert from a global struct to a subclass of
9643         ipa_opt_pass_d along with...
9644         (pass_data_ipa_inline): ...new pass_data instance and...
9645         (make_pass_ipa_inline): ...new function.
9646         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9647         struct to a subclass of gimple_opt_pass along with...
9648         (pass_data_local_pure_const): ...new pass_data instance and...
9649         (make_pass_local_pure_const): ...new function.
9650         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9651         ipa_opt_pass_d along with...
9652         (pass_data_ipa_pure_const): ...new pass_data instance and...
9653         (make_pass_ipa_pure_const): ...new function.
9654         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9655         to a subclass of ipa_opt_pass_d along with...
9656         (pass_data_ipa_reference): ...new pass_data instance and...
9657         (make_pass_ipa_reference): ...new function.
9658         * ipa-split.c (pass_split_functions): Convert from a global struct to
9659         a subclass of gimple_opt_pass along with...
9660         (pass_data_split_functions): ...new pass_data instance and...
9661         (make_pass_split_functions): ...new function.
9662         (pass_feedback_split_functions): Convert from a global struct to a
9663         subclass of gimple_opt_pass along with...
9664         (pass_data_feedback_split_functions): ...new pass_data instance and...
9665         (make_pass_feedback_split_functions): ...new function.
9666         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9667         global struct to a subclass of simple_ipa_opt_pass along with...
9668         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9669         instance and...
9670         (make_pass_ipa_function_and_variable_visibility): ...new function.
9671         (pass_ipa_free_inline_summary): Convert from a global struct to a
9672         subclass of simple_ipa_opt_pass along with...
9673         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9674         (make_pass_ipa_free_inline_summary): ...new function.
9675         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9676         subclass of ipa_opt_pass_d along with...
9677         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9678         and...
9679         (make_pass_ipa_whole_program_visibility): ...new function.
9680         (pass_ipa_profile): Convert from a global struct to a subclass of
9681         ipa_opt_pass_d along with...
9682         (pass_data_ipa_profile): ...new pass_data instance and...
9683         (make_pass_ipa_profile): ...new function.
9684         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9685         ipa_opt_pass_d along with...
9686         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9687         (make_pass_ipa_cdtor_merge): ...new function.
9688         * ira.c (pass_ira): Convert from a global struct to a subclass of
9689         rtl_opt_pass along with...
9690         (pass_data_ira): ...new pass_data instance and...
9691         (make_pass_ira): ...new function.
9692         (pass_reload): Convert from a global struct to a subclass of
9693         rtl_opt_pass along with...
9694         (pass_data_reload): ...new pass_data instance and...
9695         (make_pass_reload): ...new function.
9696         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9697         subclass of rtl_opt_pass along with...
9698         (pass_data_cleanup_barriers): ...new pass_data instance and...
9699         (make_pass_cleanup_barriers): ...new function.
9700         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9701         of rtl_opt_pass along with...
9702         (pass_data_loop2): ...new pass_data instance and...
9703         (make_pass_loop2): ...new function.
9704         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9705         rtl_opt_pass along with...
9706         (pass_data_rtl_loop_init): ...new pass_data instance and...
9707         (make_pass_rtl_loop_init): ...new function.
9708         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9709         rtl_opt_pass along with...
9710         (pass_data_rtl_loop_done): ...new pass_data instance and...
9711         (make_pass_rtl_loop_done): ...new function.
9712         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9713         subclass of rtl_opt_pass along with...
9714         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9715         (make_pass_rtl_move_loop_invariants): ...new function.
9716         (pass_rtl_unswitch): Convert from a global struct to a subclass of
9717         rtl_opt_pass along with...
9718         (pass_data_rtl_unswitch): ...new pass_data instance and...
9719         (make_pass_rtl_unswitch): ...new function.
9720         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
9721         subclass of rtl_opt_pass along with...
9722         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
9723         and...
9724         (make_pass_rtl_unroll_and_peel_loops): ...new function.
9725         (pass_rtl_doloop): Convert from a global struct to a subclass of
9726         rtl_opt_pass along with...
9727         (pass_data_rtl_doloop): ...new pass_data instance and...
9728         (make_pass_rtl_doloop): ...new function.
9729         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
9730         a subclass of rtl_opt_pass along with...
9731         (pass_data_lower_subreg): ...new pass_data instance and...
9732         (make_pass_lower_subreg): ...new function.
9733         (pass_lower_subreg2): Convert from a global struct to a subclass of
9734         rtl_opt_pass along with...
9735         (pass_data_lower_subreg2): ...new pass_data instance and...
9736         (make_pass_lower_subreg2): ...new function.
9737         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
9738         struct to a subclass of ipa_opt_pass_d along with...
9739         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
9740         (make_pass_ipa_lto_gimple_out): ...new function.
9741         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
9742         of ipa_opt_pass_d along with...
9743         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
9744         (make_pass_ipa_lto_finish_out): ...new function.
9745         * mode-switching.c (pass_mode_switching): Convert from a global struct
9746         to a subclass of rtl_opt_pass along with...
9747         (pass_data_mode_switching): ...new pass_data instance and...
9748         (make_pass_mode_switching): ...new function.
9749         * modulo-sched.c (pass_sms): Convert from a global struct to a
9750         subclass of rtl_opt_pass along with...
9751         (pass_data_sms): ...new pass_data instance and...
9752         (make_pass_sms): ...new function.
9753         * omp-low.c (pass_expand_omp): Convert from a global struct to a
9754         subclass of gimple_opt_pass along with...
9755         (pass_data_expand_omp): ...new pass_data instance and...
9756         (make_pass_expand_omp): ...new function.
9757         (pass_lower_omp): Convert from a global struct to a subclass of
9758         gimple_opt_pass along with...
9759         (pass_data_lower_omp): ...new pass_data instance and...
9760         (make_pass_lower_omp): ...new function.
9761         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
9762         of gimple_opt_pass along with...
9763         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
9764         (make_pass_diagnose_omp_blocks): ...new function.
9765         * passes.c (pass_early_local_passes): Convert from a global struct to
9766         a subclass of simple_ipa_opt_pass along with...
9767         (pass_data_early_local_passes): ...new pass_data instance and...
9768         (make_pass_early_local_passes): ...new function.
9769         (pass_all_early_optimizations): Convert from a global struct to a
9770         subclass of gimple_opt_pass along with...
9771         (pass_data_all_early_optimizations): ...new pass_data instance and...
9772         (make_pass_all_early_optimizations): ...new function.
9773         (pass_all_optimizations): Convert from a global struct to a subclass
9774         of gimple_opt_pass along with...
9775         (pass_data_all_optimizations): ...new pass_data instance and...
9776         (make_pass_all_optimizations): ...new function.
9777         (pass_all_optimizations_g): Convert from a global struct to a subclass
9778         of gimple_opt_pass along with...
9779         (pass_data_all_optimizations_g): ...new pass_data instance and...
9780         (make_pass_all_optimizations_g): ...new function.
9781         (pass_rest_of_compilation): Convert from a global struct to a subclass
9782         of rtl_opt_pass along with...
9783         (pass_data_rest_of_compilation): ...new pass_data instance and...
9784         (make_pass_rest_of_compilation): ...new function.
9785         (pass_postreload): Convert from a global struct to a subclass of
9786         rtl_opt_pass along with...
9787         (pass_data_postreload): ...new pass_data instance and...
9788         (make_pass_postreload): ...new function.
9789         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
9790         subclass of rtl_opt_pass along with...
9791         (pass_data_gcse2): ...new pass_data instance and...
9792         (make_pass_gcse2): ...new function.
9793         * postreload.c (pass_postreload_cse): Convert from a global struct to
9794         a subclass of rtl_opt_pass along with...
9795         (pass_data_postreload_cse): ...new pass_data instance and...
9796         (make_pass_postreload_cse): ...new function.
9797         * predict.c (pass_profile): Convert from a global struct to a subclass
9798         of gimple_opt_pass along with...
9799         (pass_data_profile): ...new pass_data instance and...
9800         (make_pass_profile): ...new function.
9801         (pass_strip_predict_hints): Convert from a global struct to a subclass
9802         of gimple_opt_pass along with...
9803         (pass_data_strip_predict_hints): ...new pass_data instance and...
9804         (make_pass_strip_predict_hints): ...new function.
9805         * recog.c (pass_peephole2): Convert from a global struct to a subclass
9806         of rtl_opt_pass along with...
9807         (pass_data_peephole2): ...new pass_data instance and...
9808         (make_pass_peephole2): ...new function.
9809         (pass_split_all_insns): Convert from a global struct to a subclass of
9810         rtl_opt_pass along with...
9811         (pass_data_split_all_insns): ...new pass_data instance and...
9812         (make_pass_split_all_insns): ...new function.
9813         (pass_split_after_reload): Convert from a global struct to a subclass
9814         of rtl_opt_pass along with...
9815         (pass_data_split_after_reload): ...new pass_data instance and...
9816         (make_pass_split_after_reload): ...new function.
9817         (pass_split_before_regstack): Convert from a global struct to a
9818         subclass of rtl_opt_pass along with...
9819         (pass_data_split_before_regstack): ...new pass_data instance and...
9820         (make_pass_split_before_regstack): ...new function.
9821         (pass_split_before_sched2): Convert from a global struct to a subclass
9822         of rtl_opt_pass along with...
9823         (pass_data_split_before_sched2): ...new pass_data instance and...
9824         (make_pass_split_before_sched2): ...new function.
9825         (pass_split_for_shorten_branches): Convert from a global struct to a
9826         subclass of rtl_opt_pass along with...
9827         (pass_data_split_for_shorten_branches): ...new pass_data instance
9828         and...
9829         (make_pass_split_for_shorten_branches): ...new function.
9830         * ree.c (pass_ree): Convert from a global struct to a subclass of
9831         rtl_opt_pass along with...
9832         (pass_data_ree): ...new pass_data instance and...
9833         (make_pass_ree): ...new function.
9834         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
9835         subclass of rtl_opt_pass along with...
9836         (pass_data_stack_regs): ...new pass_data instance and...
9837         (make_pass_stack_regs): ...new function.
9838         (pass_stack_regs_run): Convert from a global struct to a subclass of
9839         rtl_opt_pass along with...
9840         (pass_data_stack_regs_run): ...new pass_data instance and...
9841         (make_pass_stack_regs_run): ...new function.
9842         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
9843         subclass of rtl_opt_pass along with...
9844         (pass_data_cprop_hardreg): ...new pass_data instance and...
9845         (make_pass_cprop_hardreg): ...new function.
9846         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
9847         subclass of rtl_opt_pass along with...
9848         (pass_data_reginfo_init): ...new pass_data instance and...
9849         (make_pass_reginfo_init): ...new function.
9850         * regmove.c (pass_regmove): Convert from a global struct to a subclass
9851         of rtl_opt_pass along with...
9852         (pass_data_regmove): ...new pass_data instance and...
9853         (make_pass_regmove): ...new function.
9854         * regrename.c (pass_regrename): Convert from a global struct to a
9855         subclass of rtl_opt_pass along with...
9856         (pass_data_regrename): ...new pass_data instance and...
9857         (make_pass_regrename): ...new function.
9858         * reorg.c (pass_delay_slots): Convert from a global struct to a
9859         subclass of rtl_opt_pass along with...
9860         (pass_data_delay_slots): ...new pass_data instance and...
9861         (make_pass_delay_slots): ...new function.
9862         (pass_machine_reorg): Convert from a global struct to a subclass of
9863         rtl_opt_pass along with...
9864         (pass_data_machine_reorg): ...new pass_data instance and...
9865         (make_pass_machine_reorg): ...new function.
9866         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
9867         of rtl_opt_pass along with...
9868         (pass_data_sched): ...new pass_data instance and...
9869         (make_pass_sched): ...new function.
9870         (pass_sched2): Convert from a global struct to a subclass of
9871         rtl_opt_pass along with...
9872         (pass_data_sched2): ...new pass_data instance and...
9873         (make_pass_sched2): ...new function.
9874         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
9875         to a subclass of rtl_opt_pass along with...
9876         (pass_data_stack_ptr_mod): ...new pass_data instance and...
9877         (make_pass_stack_ptr_mod): ...new function.
9878         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
9879         to a subclass of rtl_opt_pass along with...
9880         (pass_data_rtl_store_motion): ...new pass_data instance and...
9881         (make_pass_rtl_store_motion): ...new function.
9882         * tracer.c (pass_tracer): Convert from a global struct to a subclass
9883         of gimple_opt_pass along with...
9884         (pass_data_tracer): ...new pass_data instance and...
9885         (make_pass_tracer): ...new function.
9886         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
9887         to a subclass of gimple_opt_pass along with...
9888         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
9889         (make_pass_diagnose_tm_blocks): ...new function.
9890         (pass_lower_tm): Convert from a global struct to a subclass of
9891         gimple_opt_pass along with...
9892         (pass_data_lower_tm): ...new pass_data instance and...
9893         (make_pass_lower_tm): ...new function.
9894         (pass_tm_init): Convert from a global struct to a subclass of
9895         gimple_opt_pass along with...
9896         (pass_data_tm_init): ...new pass_data instance and...
9897         (make_pass_tm_init): ...new function.
9898         (pass_tm_mark): Convert from a global struct to a subclass of
9899         gimple_opt_pass along with...
9900         (pass_data_tm_mark): ...new pass_data instance and...
9901         (make_pass_tm_mark): ...new function.
9902         (pass_tm_edges): Convert from a global struct to a subclass of
9903         gimple_opt_pass along with...
9904         (pass_data_tm_edges): ...new pass_data instance and...
9905         (make_pass_tm_edges): ...new function.
9906         (pass_tm_memopt): Convert from a global struct to a subclass of
9907         gimple_opt_pass along with...
9908         (pass_data_tm_memopt): ...new pass_data instance and...
9909         (make_pass_tm_memopt): ...new function.
9910         (pass_ipa_tm): Convert from a global struct to a subclass of
9911         simple_ipa_opt_pass along with...
9912         (pass_data_ipa_tm): ...new pass_data instance and...
9913         (make_pass_ipa_tm): ...new function.
9914         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
9915         subclass of gimple_opt_pass along with...
9916         (pass_data_call_cdce): ...new pass_data instance and...
9917         (make_pass_call_cdce): ...new function.
9918         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
9919         subclass of gimple_opt_pass along with...
9920         (pass_data_build_cfg): ...new pass_data instance and...
9921         (make_pass_build_cfg): ...new function.
9922         (pass_split_crit_edges): Convert from a global struct to a subclass of
9923         gimple_opt_pass along with...
9924         (pass_data_split_crit_edges): ...new pass_data instance and...
9925         (make_pass_split_crit_edges): ...new function.
9926         (pass_warn_function_return): Convert from a global struct to a
9927         subclass of gimple_opt_pass along with...
9928         (pass_data_warn_function_return): ...new pass_data instance and...
9929         (make_pass_warn_function_return): ...new function.
9930         (pass_warn_function_noreturn): Convert from a global struct to a
9931         subclass of gimple_opt_pass along with...
9932         (pass_data_warn_function_noreturn): ...new pass_data instance and...
9933         (make_pass_warn_function_noreturn): ...new function.
9934         (pass_warn_unused_result): Convert from a global struct to a subclass
9935         of gimple_opt_pass along with...
9936         (pass_data_warn_unused_result): ...new pass_data instance and...
9937         (make_pass_warn_unused_result): ...new function.
9938         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
9939         a subclass of gimple_opt_pass along with...
9940         (pass_data_merge_phi): ...new pass_data instance and...
9941         (make_pass_merge_phi): ...new function.
9942         * tree-complex.c (pass_lower_complex): Convert from a global struct to
9943         a subclass of gimple_opt_pass along with...
9944         (pass_data_lower_complex): ...new pass_data instance and...
9945         (make_pass_lower_complex): ...new function.
9946         (pass_lower_complex_O0): Convert from a global struct to a subclass of
9947         gimple_opt_pass along with...
9948         (pass_data_lower_complex_O0): ...new pass_data instance and...
9949         (make_pass_lower_complex_O0): ...new function.
9950         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
9951         subclass of gimple_opt_pass along with...
9952         (pass_data_lower_eh): ...new pass_data instance and...
9953         (make_pass_lower_eh): ...new function.
9954         (pass_refactor_eh): Convert from a global struct to a subclass of
9955         gimple_opt_pass along with...
9956         (pass_data_refactor_eh): ...new pass_data instance and...
9957         (make_pass_refactor_eh): ...new function.
9958         (pass_lower_resx): Convert from a global struct to a subclass of
9959         gimple_opt_pass along with...
9960         (pass_data_lower_resx): ...new pass_data instance and...
9961         (make_pass_lower_resx): ...new function.
9962         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
9963         of gimple_opt_pass along with...
9964         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
9965         (make_pass_lower_eh_dispatch): ...new function.
9966         (pass_cleanup_eh): Convert from a global struct to a subclass of
9967         gimple_opt_pass along with...
9968         (pass_data_cleanup_eh): ...new pass_data instance and...
9969         (make_pass_cleanup_eh): ...new function.
9970         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
9971         to a subclass of simple_ipa_opt_pass along with...
9972         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
9973         (make_pass_ipa_lower_emutls): ...new function.
9974         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
9975         a subclass of gimple_opt_pass along with...
9976         (pass_data_if_conversion): ...new pass_data instance and...
9977         (make_pass_if_conversion): ...new function.
9978         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
9979         subclass of gimple_opt_pass along with...
9980         (pass_data_build_ssa): ...new pass_data instance and...
9981         (make_pass_build_ssa): ...new function.
9982         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
9983         global struct to a subclass of gimple_opt_pass along with...
9984         (pass_data_loop_distribution): ...new pass_data instance and...
9985         (make_pass_loop_distribution): ...new function.
9986         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
9987         subclass of gimple_opt_pass along with...
9988         (pass_data_mudflap_1): ...new pass_data instance and...
9989         (make_pass_mudflap_1): ...new function.
9990         (pass_mudflap_2): Convert from a global struct to a subclass of
9991         gimple_opt_pass along with...
9992         (pass_data_mudflap_2): ...new pass_data instance and...
9993         (make_pass_mudflap_2): ...new function.
9994         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
9995         subclass of gimple_opt_pass along with...
9996         (pass_data_mudflap_1): ...new pass_data instance and...
9997         (make_pass_mudflap_1): ...new function.
9998         (pass_mudflap_2): Convert from a global struct to a subclass of
9999         gimple_opt_pass along with...
10000         (pass_data_mudflap_2): ...new pass_data instance and...
10001         (make_pass_mudflap_2): ...new function.
10002         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10003         gimple_opt_pass along with...
10004         (pass_data_nrv): ...new pass_data instance and...
10005         (make_pass_nrv): ...new function.
10006         (pass_return_slot): Convert from a global struct to a subclass of
10007         gimple_opt_pass along with...
10008         (pass_data_return_slot): ...new pass_data instance and...
10009         (make_pass_return_slot): ...new function.
10010         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10011         to a subclass of gimple_opt_pass along with...
10012         (pass_data_object_sizes): ...new pass_data instance and...
10013         (make_pass_object_sizes): ...new function.
10014         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10015         global struct to a subclass of gimple_opt_pass along with...
10016         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10017         and...
10018         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10019         (pass_fixup_cfg): Convert from a global struct to a subclass of
10020         gimple_opt_pass along with...
10021         (pass_data_fixup_cfg): ...new pass_data instance and...
10022         (make_pass_fixup_cfg): ...new function.
10023         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10024         (make_pass_mudflap_1): ...new function.
10025         (pass_mudflap_2): Replace declaration with that of...
10026         (make_pass_mudflap_2): ...new function.
10027         (pass_asan): Replace declaration with that of...
10028         (make_pass_asan): ...new function.
10029         (pass_asan_O0): Replace declaration with that of...
10030         (make_pass_asan_O0): ...new function.
10031         (pass_tsan): Replace declaration with that of...
10032         (make_pass_tsan): ...new function.
10033         (pass_tsan_O0): Replace declaration with that of...
10034         (make_pass_tsan_O0): ...new function.
10035         (pass_lower_cf): Replace declaration with that of...
10036         (make_pass_lower_cf): ...new function.
10037         (pass_refactor_eh): Replace declaration with that of...
10038         (make_pass_refactor_eh): ...new function.
10039         (pass_lower_eh): Replace declaration with that of...
10040         (make_pass_lower_eh): ...new function.
10041         (pass_lower_eh_dispatch): Replace declaration with that of...
10042         (make_pass_lower_eh_dispatch): ...new function.
10043         (pass_lower_resx): Replace declaration with that of...
10044         (make_pass_lower_resx): ...new function.
10045         (pass_build_cfg): Replace declaration with that of...
10046         (make_pass_build_cfg): ...new function.
10047         (pass_early_tree_profile): Replace declaration with that of...
10048         (make_pass_early_tree_profile): ...new function.
10049         (pass_cleanup_eh): Replace declaration with that of...
10050         (make_pass_cleanup_eh): ...new function.
10051         (pass_sra): Replace declaration with that of...
10052         (make_pass_sra): ...new function.
10053         (pass_sra_early): Replace declaration with that of...
10054         (make_pass_sra_early): ...new function.
10055         (pass_early_ipa_sra): Replace declaration with that of...
10056         (make_pass_early_ipa_sra): ...new function.
10057         (pass_tail_recursion): Replace declaration with that of...
10058         (make_pass_tail_recursion): ...new function.
10059         (pass_tail_calls): Replace declaration with that of...
10060         (make_pass_tail_calls): ...new function.
10061         (pass_tree_loop): Replace declaration with that of...
10062         (make_pass_tree_loop): ...new function.
10063         (pass_tree_loop_init): Replace declaration with that of...
10064         (make_pass_tree_loop_init): ...new function.
10065         (pass_lim): Replace declaration with that of...
10066         (make_pass_lim): ...new function.
10067         (pass_tree_unswitch): Replace declaration with that of...
10068         (make_pass_tree_unswitch): ...new function.
10069         (pass_predcom): Replace declaration with that of...
10070         (make_pass_predcom): ...new function.
10071         (pass_iv_canon): Replace declaration with that of...
10072         (make_pass_iv_canon): ...new function.
10073         (pass_scev_cprop): Replace declaration with that of...
10074         (make_pass_scev_cprop): ...new function.
10075         (pass_empty_loop): Replace declaration with that of...
10076         (make_pass_empty_loop): ...new function.
10077         (pass_record_bounds): Replace declaration with that of...
10078         (make_pass_record_bounds): ...new function.
10079         (pass_graphite): Replace declaration with that of...
10080         (make_pass_graphite): ...new function.
10081         (pass_graphite_transforms): Replace declaration with that of...
10082         (make_pass_graphite_transforms): ...new function.
10083         (pass_if_conversion): Replace declaration with that of...
10084         (make_pass_if_conversion): ...new function.
10085         (pass_loop_distribution): Replace declaration with that of...
10086         (make_pass_loop_distribution): ...new function.
10087         (pass_vectorize): Replace declaration with that of...
10088         (make_pass_vectorize): ...new function.
10089         (pass_slp_vectorize): Replace declaration with that of...
10090         (make_pass_slp_vectorize): ...new function.
10091         (pass_complete_unroll): Replace declaration with that of...
10092         (make_pass_complete_unroll): ...new function.
10093         (pass_complete_unrolli): Replace declaration with that of...
10094         (make_pass_complete_unrolli): ...new function.
10095         (pass_parallelize_loops): Replace declaration with that of...
10096         (make_pass_parallelize_loops): ...new function.
10097         (pass_loop_prefetch): Replace declaration with that of...
10098         (make_pass_loop_prefetch): ...new function.
10099         (pass_iv_optimize): Replace declaration with that of...
10100         (make_pass_iv_optimize): ...new function.
10101         (pass_tree_loop_done): Replace declaration with that of...
10102         (make_pass_tree_loop_done): ...new function.
10103         (pass_ch): Replace declaration with that of...
10104         (make_pass_ch): ...new function.
10105         (pass_ccp): Replace declaration with that of...
10106         (make_pass_ccp): ...new function.
10107         (pass_phi_only_cprop): Replace declaration with that of...
10108         (make_pass_phi_only_cprop): ...new function.
10109         (pass_build_ssa): Replace declaration with that of...
10110         (make_pass_build_ssa): ...new function.
10111         (pass_build_alias): Replace declaration with that of...
10112         (make_pass_build_alias): ...new function.
10113         (pass_build_ealias): Replace declaration with that of...
10114         (make_pass_build_ealias): ...new function.
10115         (pass_dominator): Replace declaration with that of...
10116         (make_pass_dominator): ...new function.
10117         (pass_dce): Replace declaration with that of...
10118         (make_pass_dce): ...new function.
10119         (pass_dce_loop): Replace declaration with that of...
10120         (make_pass_dce_loop): ...new function.
10121         (pass_cd_dce): Replace declaration with that of...
10122         (make_pass_cd_dce): ...new function.
10123         (pass_call_cdce): Replace declaration with that of...
10124         (make_pass_call_cdce): ...new function.
10125         (pass_merge_phi): Replace declaration with that of...
10126         (make_pass_merge_phi): ...new function.
10127         (pass_split_crit_edges): Replace declaration with that of...
10128         (make_pass_split_crit_edges): ...new function.
10129         (pass_pre): Replace declaration with that of...
10130         (make_pass_pre): ...new function.
10131         (pass_profile): Replace declaration with that of...
10132         (make_pass_profile): ...new function.
10133         (pass_strip_predict_hints): Replace declaration with that of...
10134         (make_pass_strip_predict_hints): ...new function.
10135         (pass_lower_complex_O0): Replace declaration with that of...
10136         (make_pass_lower_complex_O0): ...new function.
10137         (pass_lower_complex): Replace declaration with that of...
10138         (make_pass_lower_complex): ...new function.
10139         (pass_lower_vector): Replace declaration with that of...
10140         (make_pass_lower_vector): ...new function.
10141         (pass_lower_vector_ssa): Replace declaration with that of...
10142         (make_pass_lower_vector_ssa): ...new function.
10143         (pass_lower_omp): Replace declaration with that of...
10144         (make_pass_lower_omp): ...new function.
10145         (pass_diagnose_omp_blocks): Replace declaration with that of...
10146         (make_pass_diagnose_omp_blocks): ...new function.
10147         (pass_expand_omp): Replace declaration with that of...
10148         (make_pass_expand_omp): ...new function.
10149         (pass_expand_omp_ssa): Replace declaration with that of...
10150         (make_pass_expand_omp_ssa): ...new function.
10151         (pass_object_sizes): Replace declaration with that of...
10152         (make_pass_object_sizes): ...new function.
10153         (pass_strlen): Replace declaration with that of...
10154         (make_pass_strlen): ...new function.
10155         (pass_fold_builtins): Replace declaration with that of...
10156         (make_pass_fold_builtins): ...new function.
10157         (pass_stdarg): Replace declaration with that of...
10158         (make_pass_stdarg): ...new function.
10159         (pass_early_warn_uninitialized): Replace declaration with that of...
10160         (make_pass_early_warn_uninitialized): ...new function.
10161         (pass_late_warn_uninitialized): Replace declaration with that of...
10162         (make_pass_late_warn_uninitialized): ...new function.
10163         (pass_cse_reciprocals): Replace declaration with that of...
10164         (make_pass_cse_reciprocals): ...new function.
10165         (pass_cse_sincos): Replace declaration with that of...
10166         (make_pass_cse_sincos): ...new function.
10167         (pass_optimize_bswap): Replace declaration with that of...
10168         (make_pass_optimize_bswap): ...new function.
10169         (pass_optimize_widening_mul): Replace declaration with that of...
10170         (make_pass_optimize_widening_mul): ...new function.
10171         (pass_warn_function_return): Replace declaration with that of...
10172         (make_pass_warn_function_return): ...new function.
10173         (pass_warn_function_noreturn): Replace declaration with that of...
10174         (make_pass_warn_function_noreturn): ...new function.
10175         (pass_cselim): Replace declaration with that of...
10176         (make_pass_cselim): ...new function.
10177         (pass_phiopt): Replace declaration with that of...
10178         (make_pass_phiopt): ...new function.
10179         (pass_forwprop): Replace declaration with that of...
10180         (make_pass_forwprop): ...new function.
10181         (pass_phiprop): Replace declaration with that of...
10182         (make_pass_phiprop): ...new function.
10183         (pass_tree_ifcombine): Replace declaration with that of...
10184         (make_pass_tree_ifcombine): ...new function.
10185         (pass_dse): Replace declaration with that of...
10186         (make_pass_dse): ...new function.
10187         (pass_nrv): Replace declaration with that of...
10188         (make_pass_nrv): ...new function.
10189         (pass_rename_ssa_copies): Replace declaration with that of...
10190         (make_pass_rename_ssa_copies): ...new function.
10191         (pass_sink_code): Replace declaration with that of...
10192         (make_pass_sink_code): ...new function.
10193         (pass_fre): Replace declaration with that of...
10194         (make_pass_fre): ...new function.
10195         (pass_check_data_deps): Replace declaration with that of...
10196         (make_pass_check_data_deps): ...new function.
10197         (pass_copy_prop): Replace declaration with that of...
10198         (make_pass_copy_prop): ...new function.
10199         (pass_vrp): Replace declaration with that of...
10200         (make_pass_vrp): ...new function.
10201         (pass_uncprop): Replace declaration with that of...
10202         (make_pass_uncprop): ...new function.
10203         (pass_return_slot): Replace declaration with that of...
10204         (make_pass_return_slot): ...new function.
10205         (pass_reassoc): Replace declaration with that of...
10206         (make_pass_reassoc): ...new function.
10207         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10208         (make_pass_rebuild_cgraph_edges): ...new function.
10209         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10210         (make_pass_remove_cgraph_callee_edges): ...new function.
10211         (pass_build_cgraph_edges): Replace declaration with that of...
10212         (make_pass_build_cgraph_edges): ...new function.
10213         (pass_local_pure_const): Replace declaration with that of...
10214         (make_pass_local_pure_const): ...new function.
10215         (pass_tracer): Replace declaration with that of...
10216         (make_pass_tracer): ...new function.
10217         (pass_warn_unused_result): Replace declaration with that of...
10218         (make_pass_warn_unused_result): ...new function.
10219         (pass_diagnose_tm_blocks): Replace declaration with that of...
10220         (make_pass_diagnose_tm_blocks): ...new function.
10221         (pass_lower_tm): Replace declaration with that of...
10222         (make_pass_lower_tm): ...new function.
10223         (pass_tm_init): Replace declaration with that of...
10224         (make_pass_tm_init): ...new function.
10225         (pass_tm_mark): Replace declaration with that of...
10226         (make_pass_tm_mark): ...new function.
10227         (pass_tm_memopt): Replace declaration with that of...
10228         (make_pass_tm_memopt): ...new function.
10229         (pass_tm_edges): Replace declaration with that of...
10230         (make_pass_tm_edges): ...new function.
10231         (pass_split_functions): Replace declaration with that of...
10232         (make_pass_split_functions): ...new function.
10233         (pass_feedback_split_functions): Replace declaration with that of...
10234         (make_pass_feedback_split_functions): ...new function.
10235         (pass_strength_reduction): Replace declaration with that of...
10236         (make_pass_strength_reduction): ...new function.
10237         (pass_ipa_lower_emutls): Replace declaration with that of...
10238         (make_pass_ipa_lower_emutls): ...new function.
10239         (pass_ipa_function_and_variable_visibility): Replace declaration with
10240         that of...
10241         (make_pass_ipa_function_and_variable_visibility): ...new function.
10242         (pass_ipa_tree_profile): Replace declaration with that of...
10243         (make_pass_ipa_tree_profile): ...new function.
10244         (pass_early_local_passes): Replace declaration with that of...
10245         (make_pass_early_local_passes): ...new function.
10246         (pass_ipa_whole_program_visibility): Replace declaration with that
10247         of...
10248         (make_pass_ipa_whole_program_visibility): ...new function.
10249         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10250         (make_pass_ipa_lto_gimple_out): ...new function.
10251         (pass_ipa_increase_alignment): Replace declaration with that of...
10252         (make_pass_ipa_increase_alignment): ...new function.
10253         (pass_ipa_inline): Replace declaration with that of...
10254         (make_pass_ipa_inline): ...new function.
10255         (pass_ipa_free_lang_data): Replace declaration with that of...
10256         (make_pass_ipa_free_lang_data): ...new function.
10257         (pass_ipa_free_inline_summary): Replace declaration with that of...
10258         (make_pass_ipa_free_inline_summary): ...new function.
10259         (pass_ipa_cp): Replace declaration with that of...
10260         (make_pass_ipa_cp): ...new function.
10261         (pass_ipa_reference): Replace declaration with that of...
10262         (make_pass_ipa_reference): ...new function.
10263         (pass_ipa_pure_const): Replace declaration with that of...
10264         (make_pass_ipa_pure_const): ...new function.
10265         (pass_ipa_pta): Replace declaration with that of...
10266         (make_pass_ipa_pta): ...new function.
10267         (pass_ipa_lto_finish_out): Replace declaration with that of...
10268         (make_pass_ipa_lto_finish_out): ...new function.
10269         (pass_ipa_tm): Replace declaration with that of...
10270         (make_pass_ipa_tm): ...new function.
10271         (pass_ipa_profile): Replace declaration with that of...
10272         (make_pass_ipa_profile): ...new function.
10273         (pass_ipa_cdtor_merge): Replace declaration with that of...
10274         (make_pass_ipa_cdtor_merge): ...new function.
10275         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10276         of...
10277         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10278         (pass_init_datastructures): Replace declaration with that of...
10279         (make_pass_init_datastructures): ...new function.
10280         (pass_fixup_cfg): Replace declaration with that of...
10281         (make_pass_fixup_cfg): ...new function.
10282         (pass_expand): Replace declaration with that of...
10283         (make_pass_expand): ...new function.
10284         (pass_instantiate_virtual_regs): Replace declaration with that of...
10285         (make_pass_instantiate_virtual_regs): ...new function.
10286         (pass_rtl_fwprop): Replace declaration with that of...
10287         (make_pass_rtl_fwprop): ...new function.
10288         (pass_rtl_fwprop_addr): Replace declaration with that of...
10289         (make_pass_rtl_fwprop_addr): ...new function.
10290         (pass_jump): Replace declaration with that of...
10291         (make_pass_jump): ...new function.
10292         (pass_jump2): Replace declaration with that of...
10293         (make_pass_jump2): ...new function.
10294         (pass_lower_subreg): Replace declaration with that of...
10295         (make_pass_lower_subreg): ...new function.
10296         (pass_cse): Replace declaration with that of...
10297         (make_pass_cse): ...new function.
10298         (pass_fast_rtl_dce): Replace declaration with that of...
10299         (make_pass_fast_rtl_dce): ...new function.
10300         (pass_ud_rtl_dce): Replace declaration with that of...
10301         (make_pass_ud_rtl_dce): ...new function.
10302         (pass_rtl_dce): Replace declaration with that of...
10303         (make_pass_rtl_dce): ...new function.
10304         (pass_rtl_dse1): Replace declaration with that of...
10305         (make_pass_rtl_dse1): ...new function.
10306         (pass_rtl_dse2): Replace declaration with that of...
10307         (make_pass_rtl_dse2): ...new function.
10308         (pass_rtl_dse3): Replace declaration with that of...
10309         (make_pass_rtl_dse3): ...new function.
10310         (pass_rtl_cprop): Replace declaration with that of...
10311         (make_pass_rtl_cprop): ...new function.
10312         (pass_rtl_pre): Replace declaration with that of...
10313         (make_pass_rtl_pre): ...new function.
10314         (pass_rtl_hoist): Replace declaration with that of...
10315         (make_pass_rtl_hoist): ...new function.
10316         (pass_rtl_store_motion): Replace declaration with that of...
10317         (make_pass_rtl_store_motion): ...new function.
10318         (pass_cse_after_global_opts): Replace declaration with that of...
10319         (make_pass_cse_after_global_opts): ...new function.
10320         (pass_rtl_ifcvt): Replace declaration with that of...
10321         (make_pass_rtl_ifcvt): ...new function.
10322         (pass_into_cfg_layout_mode): Replace declaration with that of...
10323         (make_pass_into_cfg_layout_mode): ...new function.
10324         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10325         (make_pass_outof_cfg_layout_mode): ...new function.
10326         (pass_loop2): Replace declaration with that of...
10327         (make_pass_loop2): ...new function.
10328         (pass_rtl_loop_init): Replace declaration with that of...
10329         (make_pass_rtl_loop_init): ...new function.
10330         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10331         (make_pass_rtl_move_loop_invariants): ...new function.
10332         (pass_rtl_unswitch): Replace declaration with that of...
10333         (make_pass_rtl_unswitch): ...new function.
10334         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10335         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10336         (pass_rtl_doloop): Replace declaration with that of...
10337         (make_pass_rtl_doloop): ...new function.
10338         (pass_rtl_loop_done): Replace declaration with that of...
10339         (make_pass_rtl_loop_done): ...new function.
10340         (pass_web): Replace declaration with that of...
10341         (make_pass_web): ...new function.
10342         (pass_cse2): Replace declaration with that of...
10343         (make_pass_cse2): ...new function.
10344         (pass_df_initialize_opt): Replace declaration with that of...
10345         (make_pass_df_initialize_opt): ...new function.
10346         (pass_df_initialize_no_opt): Replace declaration with that of...
10347         (make_pass_df_initialize_no_opt): ...new function.
10348         (pass_reginfo_init): Replace declaration with that of...
10349         (make_pass_reginfo_init): ...new function.
10350         (pass_inc_dec): Replace declaration with that of...
10351         (make_pass_inc_dec): ...new function.
10352         (pass_stack_ptr_mod): Replace declaration with that of...
10353         (make_pass_stack_ptr_mod): ...new function.
10354         (pass_initialize_regs): Replace declaration with that of...
10355         (make_pass_initialize_regs): ...new function.
10356         (pass_combine): Replace declaration with that of...
10357         (make_pass_combine): ...new function.
10358         (pass_if_after_combine): Replace declaration with that of...
10359         (make_pass_if_after_combine): ...new function.
10360         (pass_ree): Replace declaration with that of...
10361         (make_pass_ree): ...new function.
10362         (pass_partition_blocks): Replace declaration with that of...
10363         (make_pass_partition_blocks): ...new function.
10364         (pass_match_asm_constraints): Replace declaration with that of...
10365         (make_pass_match_asm_constraints): ...new function.
10366         (pass_regmove): Replace declaration with that of...
10367         (make_pass_regmove): ...new function.
10368         (pass_split_all_insns): Replace declaration with that of...
10369         (make_pass_split_all_insns): ...new function.
10370         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10371         (make_pass_fast_rtl_byte_dce): ...new function.
10372         (pass_lower_subreg2): Replace declaration with that of...
10373         (make_pass_lower_subreg2): ...new function.
10374         (pass_mode_switching): Replace declaration with that of...
10375         (make_pass_mode_switching): ...new function.
10376         (pass_sms): Replace declaration with that of...
10377         (make_pass_sms): ...new function.
10378         (pass_sched): Replace declaration with that of...
10379         (make_pass_sched): ...new function.
10380         (pass_ira): Replace declaration with that of...
10381         (make_pass_ira): ...new function.
10382         (pass_reload): Replace declaration with that of...
10383         (make_pass_reload): ...new function.
10384         (pass_clean_state): Replace declaration with that of...
10385         (make_pass_clean_state): ...new function.
10386         (pass_branch_prob): Replace declaration with that of...
10387         (make_pass_branch_prob): ...new function.
10388         (pass_value_profile_transformations): Replace declaration with that
10389         of...
10390         (make_pass_value_profile_transformations): ...new function.
10391         (pass_postreload_cse): Replace declaration with that of...
10392         (make_pass_postreload_cse): ...new function.
10393         (pass_gcse2): Replace declaration with that of...
10394         (make_pass_gcse2): ...new function.
10395         (pass_split_after_reload): Replace declaration with that of...
10396         (make_pass_split_after_reload): ...new function.
10397         (pass_branch_target_load_optimize1): Replace declaration with that
10398         of...
10399         (make_pass_branch_target_load_optimize1): ...new function.
10400         (pass_thread_prologue_and_epilogue): Replace declaration with that
10401         of...
10402         (make_pass_thread_prologue_and_epilogue): ...new function.
10403         (pass_stack_adjustments): Replace declaration with that of...
10404         (make_pass_stack_adjustments): ...new function.
10405         (pass_peephole2): Replace declaration with that of...
10406         (make_pass_peephole2): ...new function.
10407         (pass_if_after_reload): Replace declaration with that of...
10408         (make_pass_if_after_reload): ...new function.
10409         (pass_regrename): Replace declaration with that of...
10410         (make_pass_regrename): ...new function.
10411         (pass_cprop_hardreg): Replace declaration with that of...
10412         (make_pass_cprop_hardreg): ...new function.
10413         (pass_reorder_blocks): Replace declaration with that of...
10414         (make_pass_reorder_blocks): ...new function.
10415         (pass_branch_target_load_optimize2): Replace declaration with that
10416         of...
10417         (make_pass_branch_target_load_optimize2): ...new function.
10418         (pass_leaf_regs): Replace declaration with that of...
10419         (make_pass_leaf_regs): ...new function.
10420         (pass_split_before_sched2): Replace declaration with that of...
10421         (make_pass_split_before_sched2): ...new function.
10422         (pass_compare_elim_after_reload): Replace declaration with that of...
10423         (make_pass_compare_elim_after_reload): ...new function.
10424         (pass_sched2): Replace declaration with that of...
10425         (make_pass_sched2): ...new function.
10426         (pass_stack_regs): Replace declaration with that of...
10427         (make_pass_stack_regs): ...new function.
10428         (pass_stack_regs_run): Replace declaration with that of...
10429         (make_pass_stack_regs_run): ...new function.
10430         (pass_df_finish): Replace declaration with that of...
10431         (make_pass_df_finish): ...new function.
10432         (pass_compute_alignments): Replace declaration with that of...
10433         (make_pass_compute_alignments): ...new function.
10434         (pass_duplicate_computed_gotos): Replace declaration with that of...
10435         (make_pass_duplicate_computed_gotos): ...new function.
10436         (pass_variable_tracking): Replace declaration with that of...
10437         (make_pass_variable_tracking): ...new function.
10438         (pass_free_cfg): Replace declaration with that of...
10439         (make_pass_free_cfg): ...new function.
10440         (pass_machine_reorg): Replace declaration with that of...
10441         (make_pass_machine_reorg): ...new function.
10442         (pass_cleanup_barriers): Replace declaration with that of...
10443         (make_pass_cleanup_barriers): ...new function.
10444         (pass_delay_slots): Replace declaration with that of...
10445         (make_pass_delay_slots): ...new function.
10446         (pass_split_for_shorten_branches): Replace declaration with that of...
10447         (make_pass_split_for_shorten_branches): ...new function.
10448         (pass_split_before_regstack): Replace declaration with that of...
10449         (make_pass_split_before_regstack): ...new function.
10450         (pass_convert_to_eh_region_ranges): Replace declaration with that
10451         of...
10452         (make_pass_convert_to_eh_region_ranges): ...new function.
10453         (pass_shorten_branches): Replace declaration with that of...
10454         (make_pass_shorten_branches): ...new function.
10455         (pass_set_nothrow_function_flags): Replace declaration with that of...
10456         (make_pass_set_nothrow_function_flags): ...new function.
10457         (pass_dwarf2_frame): Replace declaration with that of...
10458         (make_pass_dwarf2_frame): ...new function.
10459         (pass_final): Replace declaration with that of...
10460         (make_pass_final): ...new function.
10461         (pass_rtl_seqabstr): Replace declaration with that of...
10462         (make_pass_rtl_seqabstr): ...new function.
10463         (pass_release_ssa_names): Replace declaration with that of...
10464         (make_pass_release_ssa_names): ...new function.
10465         (pass_early_inline): Replace declaration with that of...
10466         (make_pass_early_inline): ...new function.
10467         (pass_inline_parameters): Replace declaration with that of...
10468         (make_pass_inline_parameters): ...new function.
10469         (pass_update_address_taken): Replace declaration with that of...
10470         (make_pass_update_address_taken): ...new function.
10471         (pass_convert_switch): Replace declaration with that of...
10472         (make_pass_convert_switch): ...new function.
10473         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10474         to a subclass of simple_ipa_opt_pass along with...
10475         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10476         (make_pass_ipa_tree_profile): ...new function.
10477         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10478         subclass of gimple_opt_pass along with...
10479         (pass_data_sra_early): ...new pass_data instance and...
10480         (make_pass_sra_early): ...new function.
10481         (pass_sra): Convert from a global struct to a subclass of
10482         gimple_opt_pass along with...
10483         (pass_data_sra): ...new pass_data instance and...
10484         (make_pass_sra): ...new function.
10485         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10486         gimple_opt_pass along with...
10487         (pass_data_early_ipa_sra): ...new pass_data instance and...
10488         (make_pass_early_ipa_sra): ...new function.
10489         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10490         subclass of gimple_opt_pass along with...
10491         (pass_data_ccp): ...new pass_data instance and...
10492         (make_pass_ccp): ...new function.
10493         (pass_fold_builtins): Convert from a global struct to a subclass of
10494         gimple_opt_pass along with...
10495         (pass_data_fold_builtins): ...new pass_data instance and...
10496         (make_pass_fold_builtins): ...new function.
10497         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10498         subclass of gimple_opt_pass along with...
10499         (pass_data_copy_prop): ...new pass_data instance and...
10500         (make_pass_copy_prop): ...new function.
10501         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10502         global struct to a subclass of gimple_opt_pass along with...
10503         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10504         (make_pass_rename_ssa_copies): ...new function.
10505         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10506         subclass of gimple_opt_pass along with...
10507         (pass_data_dce): ...new pass_data instance and...
10508         (make_pass_dce): ...new function.
10509         (pass_dce_loop): Convert from a global struct to a subclass of
10510         gimple_opt_pass along with...
10511         (pass_data_dce_loop): ...new pass_data instance and...
10512         (make_pass_dce_loop): ...new function.
10513         (pass_cd_dce): Convert from a global struct to a subclass of
10514         gimple_opt_pass along with...
10515         (pass_data_cd_dce): ...new pass_data instance and...
10516         (make_pass_cd_dce): ...new function.
10517         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10518         subclass of gimple_opt_pass along with...
10519         (pass_data_dominator): ...new pass_data instance and...
10520         (make_pass_dominator): ...new function.
10521         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10522         gimple_opt_pass along with...
10523         (pass_data_phi_only_cprop): ...new pass_data instance and...
10524         (make_pass_phi_only_cprop): ...new function.
10525         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10526         subclass of gimple_opt_pass along with...
10527         (pass_data_dse): ...new pass_data instance and...
10528         (make_pass_dse): ...new function.
10529         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10530         a subclass of gimple_opt_pass along with...
10531         (pass_data_forwprop): ...new pass_data instance and...
10532         (make_pass_forwprop): ...new function.
10533         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10534         struct to a subclass of gimple_opt_pass along with...
10535         (pass_data_tree_ifcombine): ...new pass_data instance and...
10536         (make_pass_tree_ifcombine): ...new function.
10537         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10538         subclass of gimple_opt_pass along with...
10539         (pass_data_ch): ...new pass_data instance and...
10540         (make_pass_ch): ...new function.
10541         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10542         subclass of gimple_opt_pass along with...
10543         (pass_data_tree_loop): ...new pass_data instance and...
10544         (make_pass_tree_loop): ...new function.
10545         (pass_tree_loop_init): Convert from a global struct to a subclass of
10546         gimple_opt_pass along with...
10547         (pass_data_tree_loop_init): ...new pass_data instance and...
10548         (make_pass_tree_loop_init): ...new function.
10549         (pass_lim): Convert from a global struct to a subclass of
10550         gimple_opt_pass along with...
10551         (pass_data_lim): ...new pass_data instance and...
10552         (make_pass_lim): ...new function.
10553         (pass_tree_unswitch): Convert from a global struct to a subclass of
10554         gimple_opt_pass along with...
10555         (pass_data_tree_unswitch): ...new pass_data instance and...
10556         (make_pass_tree_unswitch): ...new function.
10557         (pass_predcom): Convert from a global struct to a subclass of
10558         gimple_opt_pass along with...
10559         (pass_data_predcom): ...new pass_data instance and...
10560         (make_pass_predcom): ...new function.
10561         (pass_vectorize): Convert from a global struct to a subclass of
10562         gimple_opt_pass along with...
10563         (pass_data_vectorize): ...new pass_data instance and...
10564         (make_pass_vectorize): ...new function.
10565         (pass_graphite): Convert from a global struct to a subclass of
10566         gimple_opt_pass along with...
10567         (pass_data_graphite): ...new pass_data instance and...
10568         (make_pass_graphite): ...new function.
10569         (pass_graphite_transforms): Convert from a global struct to a subclass
10570         of gimple_opt_pass along with...
10571         (pass_data_graphite_transforms): ...new pass_data instance and...
10572         (make_pass_graphite_transforms): ...new function.
10573         (pass_check_data_deps): Convert from a global struct to a subclass of
10574         gimple_opt_pass along with...
10575         (pass_data_check_data_deps): ...new pass_data instance and...
10576         (make_pass_check_data_deps): ...new function.
10577         (pass_iv_canon): Convert from a global struct to a subclass of
10578         gimple_opt_pass along with...
10579         (pass_data_iv_canon): ...new pass_data instance and...
10580         (make_pass_iv_canon): ...new function.
10581         (pass_scev_cprop): Convert from a global struct to a subclass of
10582         gimple_opt_pass along with...
10583         (pass_data_scev_cprop): ...new pass_data instance and...
10584         (make_pass_scev_cprop): ...new function.
10585         (pass_record_bounds): Convert from a global struct to a subclass of
10586         gimple_opt_pass along with...
10587         (pass_data_record_bounds): ...new pass_data instance and...
10588         (make_pass_record_bounds): ...new function.
10589         (pass_complete_unroll): Convert from a global struct to a subclass of
10590         gimple_opt_pass along with...
10591         (pass_data_complete_unroll): ...new pass_data instance and...
10592         (make_pass_complete_unroll): ...new function.
10593         (pass_complete_unrolli): Convert from a global struct to a subclass of
10594         gimple_opt_pass along with...
10595         (pass_data_complete_unrolli): ...new pass_data instance and...
10596         (make_pass_complete_unrolli): ...new function.
10597         (pass_parallelize_loops): Convert from a global struct to a subclass
10598         of gimple_opt_pass along with...
10599         (pass_data_parallelize_loops): ...new pass_data instance and...
10600         (make_pass_parallelize_loops): ...new function.
10601         (pass_loop_prefetch): Convert from a global struct to a subclass of
10602         gimple_opt_pass along with...
10603         (pass_data_loop_prefetch): ...new pass_data instance and...
10604         (make_pass_loop_prefetch): ...new function.
10605         (pass_iv_optimize): Convert from a global struct to a subclass of
10606         gimple_opt_pass along with...
10607         (pass_data_iv_optimize): ...new pass_data instance and...
10608         (make_pass_iv_optimize): ...new function.
10609         (pass_tree_loop_done): Convert from a global struct to a subclass of
10610         gimple_opt_pass along with...
10611         (pass_data_tree_loop_done): ...new pass_data instance and...
10612         (make_pass_tree_loop_done): ...new function.
10613         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10614         struct to a subclass of gimple_opt_pass along with...
10615         (pass_data_cse_reciprocals): ...new pass_data instance and...
10616         (make_pass_cse_reciprocals): ...new function.
10617         (pass_cse_sincos): Convert from a global struct to a subclass of
10618         gimple_opt_pass along with...
10619         (pass_data_cse_sincos): ...new pass_data instance and...
10620         (make_pass_cse_sincos): ...new function.
10621         (pass_optimize_bswap): Convert from a global struct to a subclass of
10622         gimple_opt_pass along with...
10623         (pass_data_optimize_bswap): ...new pass_data instance and...
10624         (make_pass_optimize_bswap): ...new function.
10625         (pass_optimize_widening_mul): Convert from a global struct to a
10626         subclass of gimple_opt_pass along with...
10627         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10628         (make_pass_optimize_widening_mul): ...new function.
10629         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10630         subclass of gimple_opt_pass along with...
10631         (pass_data_phiopt): ...new pass_data instance and...
10632         (make_pass_phiopt): ...new function.
10633         (pass_cselim): Convert from a global struct to a subclass of
10634         gimple_opt_pass along with...
10635         (pass_data_cselim): ...new pass_data instance and...
10636         (make_pass_cselim): ...new function.
10637         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10638         subclass of gimple_opt_pass along with...
10639         (pass_data_phiprop): ...new pass_data instance and...
10640         (make_pass_phiprop): ...new function.
10641         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10642         subclass of gimple_opt_pass along with...
10643         (pass_data_pre): ...new pass_data instance and...
10644         (make_pass_pre): ...new function.
10645         (pass_fre): Convert from a global struct to a subclass of
10646         gimple_opt_pass along with...
10647         (pass_data_fre): ...new pass_data instance and...
10648         (make_pass_fre): ...new function.
10649         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10650         subclass of gimple_opt_pass along with...
10651         (pass_data_reassoc): ...new pass_data instance and...
10652         (make_pass_reassoc): ...new function.
10653         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10654         subclass of gimple_opt_pass along with...
10655         (pass_data_sink_code): ...new pass_data instance and...
10656         (make_pass_sink_code): ...new function.
10657         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10658         subclass of gimple_opt_pass along with...
10659         (pass_data_strlen): ...new pass_data instance and...
10660         (make_pass_strlen): ...new function.
10661         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10662         struct to a subclass of gimple_opt_pass along with...
10663         (pass_data_build_alias): ...new pass_data instance and...
10664         (make_pass_build_alias): ...new function.
10665         (pass_build_ealias): Convert from a global struct to a subclass of
10666         gimple_opt_pass along with...
10667         (pass_data_build_ealias): ...new pass_data instance and...
10668         (make_pass_build_ealias): ...new function.
10669         (pass_ipa_pta): Convert from a global struct to a subclass of
10670         simple_ipa_opt_pass along with...
10671         (pass_data_ipa_pta): ...new pass_data instance and...
10672         (make_pass_ipa_pta): ...new function.
10673         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10674         subclass of gimple_opt_pass along with...
10675         (pass_data_uncprop): ...new pass_data instance and...
10676         (make_pass_uncprop): ...new function.
10677         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10678         global struct to a subclass of gimple_opt_pass along with...
10679         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10680         (make_pass_late_warn_uninitialized): ...new function.
10681         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10682         to a subclass of gimple_opt_pass along with...
10683         (pass_data_init_datastructures): ...new pass_data instance and...
10684         (make_pass_init_datastructures): ...new function.
10685         (pass_early_warn_uninitialized): Convert from a global struct to a
10686         subclass of gimple_opt_pass along with...
10687         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10688         (make_pass_early_warn_uninitialized): ...new function.
10689         (pass_update_address_taken): Convert from a global struct to a
10690         subclass of gimple_opt_pass along with...
10691         (pass_data_update_address_taken): ...new pass_data instance and...
10692         (make_pass_update_address_taken): ...new function.
10693         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10694         struct to a subclass of gimple_opt_pass along with...
10695         (pass_data_release_ssa_names): ...new pass_data instance and...
10696         (make_pass_release_ssa_names): ...new function.
10697         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10698         subclass of gimple_opt_pass along with...
10699         (pass_data_stdarg): ...new pass_data instance and...
10700         (make_pass_stdarg): ...new function.
10701         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10702         global struct to a subclass of gimple_opt_pass along with...
10703         (pass_data_convert_switch): ...new pass_data instance and...
10704         (make_pass_convert_switch): ...new function.
10705         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10706         to a subclass of gimple_opt_pass along with...
10707         (pass_data_tail_recursion): ...new pass_data instance and...
10708         (make_pass_tail_recursion): ...new function.
10709         (pass_tail_calls): Convert from a global struct to a subclass of
10710         gimple_opt_pass along with...
10711         (pass_data_tail_calls): ...new pass_data instance and...
10712         (make_pass_tail_calls): ...new function.
10713         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10714         struct to a subclass of gimple_opt_pass along with...
10715         (pass_data_lower_vector): ...new pass_data instance and...
10716         (make_pass_lower_vector): ...new function.
10717         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
10718         gimple_opt_pass along with...
10719         (pass_data_lower_vector_ssa): ...new pass_data instance and...
10720         (make_pass_lower_vector_ssa): ...new function.
10721         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
10722         to a subclass of gimple_opt_pass along with...
10723         (pass_data_slp_vectorize): ...new pass_data instance and...
10724         (make_pass_slp_vectorize): ...new function.
10725         (pass_ipa_increase_alignment): Convert from a global struct to a
10726         subclass of simple_ipa_opt_pass along with...
10727         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
10728         (make_pass_ipa_increase_alignment): ...new function.
10729         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
10730         gimple_opt_pass along with...
10731         (pass_data_vrp): ...new pass_data instance and...
10732         (make_pass_vrp): ...new function.
10733         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
10734         subclass of simple_ipa_opt_pass along with...
10735         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
10736         (make_pass_ipa_free_lang_data): ...new function.
10737         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
10738         gimple_opt_pass along with...
10739         (pass_data_tsan): ...new pass_data instance and...
10740         (make_pass_tsan): ...new function.
10741         (pass_tsan_O0): Convert from a global struct to a subclass of
10742         gimple_opt_pass along with...
10743         (pass_data_tsan_O0): ...new pass_data instance and...
10744         (make_pass_tsan_O0): ...new function.
10745         * var-tracking.c (pass_variable_tracking): Convert from a global
10746         struct to a subclass of rtl_opt_pass along with...
10747         (pass_data_variable_tracking): ...new pass_data instance and...
10748         (make_pass_variable_tracking): ...new function.
10749         * web.c (pass_web): Convert from a global struct to a subclass of
10750         rtl_opt_pass along with...
10751         (pass_data_web): ...new pass_data instance and...
10752         (make_pass_web): ...new function.
10753         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
10754         declaration with that of...
10755         (make_pass_mode_switch_use): ...new function.
10756         (pass_resolve_sw_modes): Replace declaration with that of...
10757         (make_pass_resolve_sw_modes): ...new function.
10758         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
10759         from a global struct to a subclass of rtl_opt_pass along with...
10760         (pass_data_mode_switch_use): ...new pass_data instance and...
10761         (make_pass_mode_switch_use): ...new function.
10762         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
10763         from a global struct to a subclass of rtl_opt_pass along with...
10764         (pass_data_resolve_sw_modes): ...new pass_data instance and...
10765         (make_pass_resolve_sw_modes): ...new function.
10766         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
10767         struct to a subclass of rtl_opt_pass along with...
10768         (pass_data_insert_vzeroupper): ...new pass_data instance and...
10769         (make_pass_insert_vzeroupper): ...new function.
10770         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
10771         global struct to a subclass of rtl_opt_pass along with...
10772         (pass_data_work_around_errata): ...new pass_data instance and...
10773         (make_pass_work_around_errata): ...new function.
10774         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
10775         struct to a subclass of rtl_opt_pass along with...
10776         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
10777         (make_pass_mips_machine_reorg2): ...new function.
10778
10779 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10780
10781         * passes.c (pass_manager::operator new): New.
10782
10783 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10784
10785         Handwritten part of conversion of passes to C++ classes.
10786
10787         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
10788         (toplev.o): Add dep on PASS_MANAGER_H.
10789         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
10790         of early local pases to reflect this moving from a global to a
10791         member of gcc::pass_manager.
10792         (cgraph_add_new_function): Likewise.
10793         * lto-cgraph.c (lto_output_node): Update for conversion of
10794         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
10795         * passes.c (opt_pass::clone): New.
10796         (opt_pass::gate): New.
10797         (opt_pass::execute): New.
10798         (opt_pass::opt_pass): New.
10799         (pass_manager::execute_early_local_passes): New.
10800         (pass_manager::execute_pass_mode_switching): new.
10801         (finish_optimization_passes): Convert to...
10802         (pass_manager::finish_optimization_passes): ...this.
10803         (finish_optimization_passes): Update for conversion of passes to
10804         C++ classes.
10805         (register_dump_files_1): Use has_gate since we cannot portably
10806         check a vtable entry against NULL.
10807         (dump_one_pass): Likewise.
10808         (ipa_write_summaries_2): Likewise.
10809         (ipa_write_optimization_summaries_1): Likewise.
10810         (ipa_read_summaries_1): Likewise.
10811         (ipa_read_optimization_summaries_1): Likewise.
10812         (execute_ipa_stmt_fixups): Likewise.
10813         (pass_manager::pass_manager): Rewrite pass-creation, invoking
10814         pass-creation functions rather than wiring up globals, and
10815         storing the results in fields of pass_manager generated using
10816         pass-instances.def.
10817         (pass_manager::dump_profile_report): Update for conversion of
10818         passes to C++ classes.
10819         (pass_manager::execute_ipa_summary_passes): Likewise.
10820         (execute_one_ipa_transform_pass): Likewise.
10821         (execute_one_pass): Use has_gate and has_execute since we cannot
10822         portably check a vtable entry against NULL.
10823         * pass_manager.h (pass_manager::finish_optimization_passes): New.
10824         (pass_manager): Use pass-instances.def to add fields for the
10825         various pass instances.
10826         * toplev.c (finalize): Update for move of
10827         finish_optimization_passes to a method of gcc::pass_manager.
10828         * toplev.h (finish_optimization_passes): Move to method of class
10829         pass_manager.
10830         * tree-pass.h (struct pass_data): New.
10831         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
10832         (opt_pass::gate): Convert to virtual function.
10833         (opt_pass::~opt_pass): New.
10834         (opt_pass::clone): New.
10835         (opt_pass::execute): Convert to virtual function.
10836         (opt_pass::opt_pass): New.
10837         (opt_pass::ctxt_): new.
10838         (gimple_opt_pass): Convert to subclass of opt_pass.
10839         (gimple_opt_pass::gimple_opt_pass): New.
10840         (rtl_opt_pass): Convert to subclass of opt_pass.
10841         (rtl_opt_pass::rtl_opt_pass): New.
10842         (ipa_opt_pass_d): Convert to subclass of opt_pass.
10843         (ipa_opt_pass_d::ipa_opt_pass_d): New.
10844         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
10845         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
10846         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
10847         invocation of pass_mode_switching to reflect this moving from a
10848         global to a member of gcc::pass_manager.
10849         (ix86_option_override): Rework how pass_insert_vzeroupper is
10850         added to the pass_manager to reflect autogenerated changes.
10851         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
10852
10853 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
10854
10855         PR rtl-optimization/57708
10856         * recog.c (peep2_find_free_register): Validate all regs in a
10857         multi-reg mode.
10858
10859 2013-08-05  Jan Hubicka  <jh@suse.cz>
10860
10861         PR lto/57602
10862         * cgraph.c (verify_cgraph_node): Accept local flags from other
10863         partitions.
10864         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
10865         (function_and_variable_visibility): Likewise.
10866         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
10867
10868 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10869
10870         * graph.c (init_graph_slim_pretty_print): Remove.
10871         (print_graph_cfg): Do not call it.  Use local pretty printer.
10872         (start_graph_dump): Likewise.
10873
10874 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10875
10876         * gimple-pretty-print.c (buffer): Remove.
10877         (initialized): Likewise.
10878         (maybe_init_pretty_print): Likewise.
10879         (print_gimple_stmt): Do not call it.  Use non-static local
10880         pretty_printer variable.
10881         (print_gimple_expr): Likewise.
10882         (print_gimple_seq): Likewise.
10883         (gimple_dump_bb): Likewise.
10884
10885 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10886
10887         * asan.c (asan_pp): Remove.
10888         (asan_pp_initialized): Likewise.
10889         (asan_pp_initialize): Likewise.
10890         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
10891         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
10892         (asan_add_global): Likewise.
10893
10894 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10895
10896         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
10897         * diagnostic.h (diagnostic_flush_buffer): Adjust.
10898         * pretty-print.c (pp_formatted_text_data): Likewise.
10899         (pp_indent): Rename from pp_base_indent.
10900         (pp_format): Rename from pp_base_format.
10901         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
10902         (pp_format_verbatim): Rename from pp_base_format_verbatim.
10903         (pp_flush): Rename from pp_base_flush.
10904         (pp_set_line_maximum_length): Rename from
10905         pp_base_set_line_maximum_length.
10906         (pp_clear_output_area): Rename from pp_base_clear_output_area.
10907         (pp_set_prefix): Rename from pp_base_set_prefix.
10908         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
10909         (pp_emit_prefix): Rename from pp_base_emit_prefix.
10910         (pp_append_text): Rename from pp_base_append_text.
10911         (pp_formatted_text): Rename from pp_base_formatted_text.
10912         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
10913         (pp_remaining_character_count_for_line): Rename from
10914         pp_base_remaining_character_count_for_line.
10915         (pp_newline): Rename from pp_base_newline.
10916         (pp_character): Rename from pp_base_character.
10917         (pp_string): Rename from pp_base_string.
10918         (pp_maybe_space): Rename from pp_base_maybe_space.
10919         * asan.c (asan_pp_string): Adjust.
10920         (asan_emit_stack_protection): Likewise.
10921         (asan_add_global): Likewise.
10922         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
10923         * tree-mudflap.c (mf_varname_tree): Likewise.
10924         * tree-pretty-print.c (pp_tree_identifier): Rename from
10925         pp_base_tree_identifier.
10926         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
10927         Declare as function.
10928
10929 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10930
10931         * pretty-print.h (pp_bar_bar): New.
10932         (pp_ampersand_ampersand): Likewise.
10933         (pp_less_equal): Likewise.
10934         (pp_greater_equal): Likewise.
10935         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
10936         printer functions instead of pp_string or operators and punctuators.
10937         (dump_gimple_call): Likewise.
10938         (dump_gimple_omp_for): Likewise.
10939         (dump_gimple_transaction): Likewise.
10940         (dump_gimple_phi): Likewise.
10941         (pp_gimple_stmt_1): Likewise.
10942         * sched-vis.c (print_insn): Likewise.
10943         * tree-mudflap.c (mf_varname_tree): Likewise.
10944         * tree-pretty-print.c (dump_block_node): Likewise.
10945         (dump_generic_node): Likewise.
10946
10947 2013-08-02  Jan Hubicka  <jh@suse.cz>
10948
10949         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
10950         boundaries.
10951         * lto-streamer-out.c (tree_is_indexable): Results decls and
10952         parm decls are not indexable.
10953         (DFS_write_tree_body): Do not follow args and results.
10954         (hash_tree): Likewise.
10955         (output_functions): Rearrange so struct function is needed
10956         only when real body is output; be able to also ouptut abstract
10957         functions; output DECL_ARGUMENTS and DECL_RESULT.
10958         (lto_output): When not in WPA, ale store abstract functions.
10959         (write_symbol): Do not care about RESULT_DECL.
10960         (output_symbol_p): Handle correctly sbtract decls.
10961         * lto-streamer-in.c (input_function): Rearrange so struct
10962         function can be NULL at entry; allow streaming of
10963         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
10964         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
10965         sanity check during LTO.
10966         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
10967         RESULT_DECl and DECL_ARGUMENTS.
10968         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10969         Likewise.
10970
10971 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10972
10973         * pretty-print.h (pp_underscore): New.
10974         (pp_comma): Tidy.
10975         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
10976         printer functions instead of pp_character.
10977         (dump_binary_rhs): Likewise.
10978         (dump_ternary_rhs): Likewise.
10979         (dump_gimple_call_args): Likewise.
10980         (pp_points_to_solution): Likewise.
10981         (dump_gimple_call): Likewise.
10982         (dump_gimple_switch): Likewise.
10983         (dump_gimple_cond): Likewise.
10984         (dump_gimple_bind): Likewise.
10985         (dump_gimple_try): Likewise.
10986         (dump_gimple_omp_for): Likewise.
10987         (dump_gimple_omp_continue): Likewise.
10988         (dump_gimple_omp_single): Likewise.
10989         (dump_gimple_omp_sections): Likewise.
10990         (dump_gimple_omp_block): Likewise.
10991         (dump_gimple_omp_critical): Likewise.
10992         (dump_gimple_transaction): Likewise.
10993         (dump_gimple_asm): Likewise.
10994         (dump_gimple_phi): Likewise.
10995         (dump_gimple_omp_parallel): Likewise.
10996         (dump_gimple_omp_task): Likewise.
10997         (dump_gimple_omp_atomic_load): Likewise.
10998         (dump_gimple_omp_atomic_store): Likewise.
10999         (dump_gimple_mem_ops): Likewise.
11000         (pp_gimple_stmt_1): Likewise.
11001         (pp_cfg_jump): Likewise.
11002         (dump_implicit_edges): Likewise.
11003         (gimple_dump_bb_for_graph): Likewise.
11004         * graph.c (draw_cfg_node): Likewise.
11005         * langhooks.c (lhd_print_error_function): Likewise.
11006         * sched-vis.c (print_exp): Likewise.
11007         (print_value): Likewise.
11008         (print_pattern): Likewise.
11009         (print_insn): Likewise.
11010         (rtl_dump_bb_for_graph): Likewise.
11011         * tree-pretty-print.c (dump_function_declaration): Likewise.
11012         (dump_array_domain): Likewise.
11013         (dump_omp_clause): Likewise.
11014         (dump_location): Likewise.
11015         (dump_generic_node): Likewise.
11016         (print_struct_decl): Likewise.
11017         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11018
11019 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11020
11021         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11022         candidate table when replacing a candidate statement.
11023         (replace_rhs_if_not_dup): Likewise.
11024         (replace_one_candidate): Likewise.
11025
11026 2013-08-02  Jan Hubicka  <jh@suse.cz>
11027             Martin Liska  <marxin.liska@gmail.com>
11028
11029         * cgraphunit.c (add_new_function): Fix logic when adding from
11030         late IPA pass.
11031         (assemble_thunk): Rename to ...
11032         (expand_thunk); .. this one; export; get it working with
11033         general functions; make produced gimple valid.
11034         * cgraph.h (expand_thunk): Declare.
11035
11036 2013-08-02  Jan Hubicka  <jh@suse.cz>
11037
11038         * ipa-cp.c (gather_context_independent_values): Use
11039         ipa_get_param_move_cost.
11040         (get_replacement_map): Remove PARAM; move parameter folding
11041         into tree-inline.c
11042         (create_specialized_node): Update.
11043         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11044         assert that we have gimple body; update move_cost.
11045         (count_formal_params): Assert that we have gimple body.
11046         (ipa_dump_param): New function.
11047         (ipa_alloc_node_params): Break out from ...
11048         (ipa_initialize_node_params): ... here.
11049         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11050         (ipa_write_node_info): Stream move costs.
11051         (ipa_read_node_info): Read move costs.
11052         (ipa_update_after_lto_read): Do not recompute node params.
11053         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11054         (ipa_get_param): Check we are not in WPA.
11055         (ipa_get_param_move_cost): New.
11056         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11057         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11058         parm numbers to be present.
11059
11060 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11061
11062         PR rtl-optimization/58048
11063         * lra-constraints.c (process_alt_operands): Don't check asm
11064         operand on register.
11065
11066 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11067
11068         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11069         the implied StoreLoad barrier for atomic operations if before.
11070
11071 2013-08-02  Jan Hubicka  <jh@suse.cz>
11072             Martin Liska  <marxin.liska@gmail.com>
11073
11074         * cgraph.c (cgraph_function_body_availability): Do not check
11075         cgraph flags.
11076         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11077         symtab_node_availability): Declare.
11078         * ipa.c (can_replace_by_local_alias): New.
11079         (function_and_variable_visibility): Use it.
11080         * symtab.c (symtab_for_node_and_aliases,
11081         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11082
11083 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11084
11085         PR rtl-optimization/57963
11086         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11087         (lra_constraints): Use them.
11088
11089 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11090
11091         * config/arm/types.md (define_attr "type"): Add "load_acq"
11092         and "store_rel".
11093         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11094         changes.
11095         (cortex_a53_store1): Likewise.
11096
11097 2013-08-01  Jan Hubicka  <jh@suse.cz>
11098
11099         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11100         partitions are not needed.
11101
11102 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11103
11104         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11105         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11106         MAYBE_NON_Q_CLASS_P where appropriate.
11107
11108 2013-08-01  Jan Hubicka  <jh@suse.cz>
11109
11110         * cgraph.h (release_function_body): Declare.
11111         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11112         of unused delcarations.
11113
11114 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11115
11116         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11117         RTL form when subtracting a constant.
11118
11119 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11120
11121         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11122         Generate canonical plus rtx with negated immediate instead of minus
11123         where appropriate.
11124         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11125
11126 2013-08-01  Jan Hubicka  <jh@suse.cz>
11127
11128         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11129         (cgraph_release_function_body): Likewise.
11130         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11131         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11132         to used_as_abstract_origin.
11133         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11134         symbols used as abstract origins.
11135         * cgraphunit.c (analyze_functions): Update.
11136         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11137         used_as_abstract_origin.
11138         * tree-inline.c (tree_function_versioning): Update
11139         used_as_abstract_origin; be ready for DECL_RESULT and
11140         DECL_ARGUMENTS to be NULL.
11141
11142         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11143         for abstract functions.
11144         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11145         real symbols.
11146
11147 2013-08-01  Jan Hubicka  <jh@suse.cz>
11148
11149         * profile.c (compute_value_histograms): Fix thinko.
11150
11151 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11152
11153         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11154         aarch-common-protos.h to extra_headers.
11155         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11156         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11157         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11158
11159 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11160
11161         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11162         Include "../arm/types.md".  Define "type" attribute for all patterns.
11163         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11164         attribute changes.
11165
11166 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11167
11168         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11169         to support power8 load fusion.
11170         (fusion_gpr_mem_load): Likewise.
11171
11172         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11173
11174         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11175         declarations for power8 load fusion.
11176         (emit_fusion_gpr_load): Likewise.
11177
11178         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11179         tuning for power8, turn on fusion mode by default.  Turn on sign
11180         extending fusion mode if normal fusion mode is on, and we are at
11181         -O2 or -O3.
11182         (fusion_gpr_load_p): New function, return true if we can fuse an
11183         addis instruction with a dependent load to a GPR.
11184         (emit_fusion_gpr_load): Emit the instructions for power8 load
11185         fusion to GPRs.
11186
11187         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11188         (VSX load fusion peepholes): New peepholes to fuse together an
11189         addi instruction with a VSX load instruction.
11190
11191         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11192         peepholes to fuse an addis instruction with a load to a GPR base
11193         register.  If we are supporting sign extending fusions, convert
11194         sign extending loads to zero extending loads and add an explicit
11195         sign extension.
11196
11197 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11198
11199         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11200         aarch-common-protos.h to extra_headers.
11201         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11202         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11203         (arm_early_store_addr_dep): Likewise.
11204         (arm_no_early_alu_shift_dep): Likewise.
11205         (arm_no_early_alu_shift_value_dep): Likewise.
11206         (arm_no_early_mul_dep): Likewise.
11207         (arm_no_early_store_addr_dep): Likewise.
11208         (arm_mac_accumulator_is_mul_result): Likewise.
11209         (arm_mac_accumulator_is_result): Likewise.
11210         * config/arm/aarch-common.c: ... here.  New file.
11211         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11212         here to ...
11213         (arm_early_store_addr_dep): Likewise.
11214         (arm_no_early_alu_shift_dep): Likewise.
11215         (arm_no_early_alu_shift_value_dep): Likewise.
11216         (arm_no_early_mul_dep): Likewise.
11217         (arm_no_early_store_addr_dep): Likewise.
11218         (arm_mac_accumulator_is_mul_result): Likewise.
11219         (arm_mac_accumulator_is_result): Likewise.
11220         * config/arm/aarch-common-protos.h: ... here.  New file.
11221         * config/arm/t-arm (aarch-common.o): Define.
11222
11223 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11224
11225         * config/arm/arm.md: Include new file "types.md".
11226         (define_attr "type"): Move from here to ...
11227         (define_attr "mul32"): Likewise.
11228         (define_attr "mul64"): Likewise.
11229         * config/arm/types.md: ... here.  New file.
11230
11231 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11232
11233         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11234         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11235
11236 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11237
11238         * gen-pass-instances.awk: Fix offset of substr().
11239
11240 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11241
11242         * Makefile.in (pass-instances.def): New.
11243         (passes.o): Replace dependency on passes.def with one on
11244         pass-instances.def
11245
11246         * gen-pass-instances.awk: New.
11247
11248         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11249         rather than passes.def, updating local definition of NEXT_PASS
11250         macro to add an extra NUM parameter (currently unused).
11251
11252 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11253
11254         * Makefile.in (PASS_MANAGER_H): New.
11255         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11256         (passes.o): Likewise.
11257         (statistics.o): Likewise.
11258         (cgraphunit.o): Likewise.
11259         (context.o): Depend on PASS_MANAGER_H.
11260
11261         * pass_manager.h: New.
11262
11263         * cgraphunit.c (cgraph_add_new_function): Update for moves
11264         of globals to fields of pass_manager.
11265         (analyze_function): Likewise.
11266         (expand_function): Likewise.
11267         (ipa_passes): Likewise.
11268         (compile): Likewise.
11269
11270         * context.c (context::context): New.
11271         * context.h  (context::context): New.
11272         (context::get_passes): New.
11273         (context::passes_): New.
11274
11275         * lto-cgraph.c (input_node): Update for moves of globals to
11276         fields of pass_manager.
11277
11278         * passes.c (all_passes): Remove, in favor of a field of the
11279         same name within the new class pass_manager.
11280         (all_small_ipa_passes): Likewise.
11281         (all_lowering_passes): Likewise.
11282         (all_regular_ipa_passes): Likewise.
11283         (all_late_ipa_passes): Likewise.
11284         (all_lto_gen_passes): Likewise.
11285         (passes_by_id): Likewise.
11286         (passes_by_id_size): Likewise.
11287         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11288         the new class pass_manager.
11289         (set_pass_for_id): Convert to...
11290         (pass_manager::set_pass_for_id): ...method.
11291         (get_pass_for_id): Convert to...
11292         (pass_manager::get_pass_for_id): ...method.
11293         (register_one_dump_file): Move body of implementation into...
11294         (pass_manager::register_one_dump_file): ...here.
11295         (register_dump_files_1): Convert to...
11296         (pass_manager::register_dump_files_1): ...method.
11297         (register_dump_files): Convert to...
11298         (pass_manager::register_dump_files): ...method.
11299         (create_pass_tab): Update for moves of globals to fields of
11300         pass_manager.
11301         (dump_passes): Move body of implementation into...
11302         (pass_manager::dump_passes): ...here.
11303         (register_pass): Move body of implementation into...
11304         (pass_manager::register_pass): ...here.
11305         (init_optimization_passes): Convert into...
11306         (pass_manager::pass_manager): ...constructor for new
11307         pass_manager class, and initialize the pass_lists array.
11308         (check_profile_consistency): Update for moves of globals to
11309         fields of pass_manager.
11310         (dump_profile_report): Move body of implementation into...
11311         (pass_manager::dump_profile_report): ...here.
11312         (ipa_write_summaries_1): Update for moves of pass lists from
11313         being globals to fields of pass_manager.
11314         (ipa_write_optimization_summaries): Likewise.
11315         (ipa_read_summaries):  Likewise.
11316         (ipa_read_optimization_summaries): Likewise.
11317         (execute_all_ipa_stmt_fixups): Likewise.
11318
11319         * statistics.c (statistics_fini): Update for moves of globals to
11320         fields of pass_manager.
11321
11322         * toplev.c (general_init): Replace call to
11323         init_optimization_passes with construction of the pass_manager
11324         instance.
11325
11326         * tree-pass.h (all_passes): Remove, in favor of a field of the
11327         same name within the new class pass_manager.
11328         (all_small_ipa_passes): Likewise.
11329         (all_lowering_passes): Likewise.
11330         (all_regular_ipa_passes): Likewise.
11331         (all_lto_gen_passes): Likewise.
11332         (all_late_ipa_passes): Likewise.
11333         (passes_by_id): Likewise.
11334         (passes_by_id_size): Likewise.
11335         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11336         the new class pass_manager.
11337         (get_pass_for_id): Remove.
11338
11339 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11340
11341         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11342         configs.
11343
11344 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11345
11346         * arm.md (mulhi3): New expand pattern.
11347
11348 2013-07-30  Jan Hubicka  <jh@suse.cz>
11349             Martin Liska  <marxin.liska@gmail.com>
11350
11351         * profile.c (compute_value_histograms): Do not ICE when
11352         there is mismatch only on some counters.
11353
11354 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11355
11356         PR rtl-optimization/57637
11357         * function.c (move_insn_for_shrink_wrap): Also check the
11358         GEN set of the LIVE problem for the liveness analysis
11359         if it exists, otherwise give up.
11360
11361 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11362
11363         PR tree-optimization/57993
11364         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11365         replaced statement in the candidate table.
11366         (phi_add_costs): Return infinite cost when the hidden basis does
11367         not dominate all phis on which the candidate is dependent.
11368         (replace_one_candidate): Record replaced statement in the
11369         candidate table.
11370
11371 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11372
11373         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11374         (ashlv2si3): New expander.
11375         (*ashlv2si3_i): New define_insn_and_split.
11376         * predicates.md (float_operation): Allow patterns with three
11377         basic sub-patterns.
11378
11379         PR rtl-optimization/58021
11380         * mode-switching.c (create_pre_exit): Always split off preceding
11381         insns if we are not at the basic block head.
11382
11383 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11384
11385         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11386         (UCLIBC_DYNAMIC_LINKER): New macro.
11387         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11388         `-mnan=2008'.
11389         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11390         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11391         `-mnan=2008'.
11392         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11393         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11394         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11395         for SF and DF modes.  Use ieee_quad_format for TF mode.
11396         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11397         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11398         (mips_option_override): Handle `-mnan=legacy'.
11399         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11400         `-mabs=2008' and `-mnan=2008'.
11401         (OPTION_DEFAULT_SPECS): Add "nan" default.
11402         (ASM_SPEC): Handle `-mnan='.
11403         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11404         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11405         comment accordingly.
11406         (neg<mode>2): Likewise.
11407         * config/mips/mips.opt (mabs, mnan): New options.
11408         * doc/install.texi (Configuration): Document `--with-nan=' option.
11409         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11410         `-mnan=' options.
11411         (MIPS Options): Document them.
11412         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11413         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11414         * configure: Regenerate.
11415         * config.in: Regenerate.
11416
11417 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11418
11419         * config/i386/i386.md (float post-reload splitters): Do not check
11420         for subregs of SSE registers.
11421
11422 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11423             H.J. Lu  <hongjiu.lu@intel.com>
11424
11425         PR target/57954
11426         PR target/57988
11427         * config/i386/i386.md (post-reload splitter
11428         to avoid partial SSE reg dependency stalls): New pattern.
11429
11430 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11431
11432         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11433
11434 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11435
11436         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11437         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11438
11439 2013-07-26  Cary Coutant  <ccoutant@google.com>
11440
11441         * dwarf2out.c (die_checksum_ordered): Don't include template
11442         instantiations in signature.
11443         (is_template_parameter): New function.
11444         (is_template_instantiation): New function.
11445         (generate_skeleton_bottom_up): Don't include template instantiations
11446         in type unit DIE.
11447         (generate_skeleton): Likewise.
11448         (break_out_comdat_types): Move recursive call to break out nested
11449         types earlier.
11450         (prune_unused_types_mark_generic_parms_dies): Call
11451         is_template_parameter.
11452
11453 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11454
11455         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11456         uses vector registers.
11457         * config/aarch64/iterators.md: Add attributes rtn and vas.
11458
11459 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11460             Richard Earnshaw  <richard.earnshaw@arm.com>
11461
11462         * combine.c (simplify_comparison): Re-canonicalize operands
11463         where appropriate.
11464         * config/arm/arm.md (movcond_addsi): New splitter.
11465
11466 2013-07-25  Sterling Augustine  <saugustine@google.com>
11467
11468         * dwarf2out.c (size_of_pubnames): Move code to...
11469         (include_pubname_in_output): ...here.  New.
11470         (want_pubnames): Rearrange.
11471         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11472
11473 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11474
11475         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11476
11477 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11478
11479         PR target/38836
11480         * doc/extend.texi: Remove obsolete builtins. Fix
11481         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11482
11483 2013-07-25  Jan Hubicka  <jh@suse.cz>
11484
11485         * cgraph.c (release_function_body): Break out from ...
11486         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11487         and DECL_ARGUMENTS.
11488         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11489         old_tree in the map.
11490         (create_specialized_node): Update.
11491         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11492         into index.
11493         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11494         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11495         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11496         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11497         DECL_RESULT.
11498
11499 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11500
11501         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11502         addsi3_carryin_alt2_<optab>): Correct output template.
11503
11504 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11505
11506         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11507         Adjust for arm_restrict_it.
11508         Remove trailing whitespace.
11509
11510 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
11511
11512         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11513         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11514
11515         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11516
11517 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11518
11519         PR rtl-optimization/57960
11520         * lra-constraints.c (process_alt_operands): Use the right mode
11521         when checking strict_low.
11522
11523 2013-07-25  Jan Hubicka  <jh@suse.cz>
11524
11525         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11526         * cgraph.c (cgraph_remove_node): Do not release function body
11527         when in cgraph streaming.
11528         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11529         in other partitions are not considered reachable; fix handling of
11530         clones.
11531
11532 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11533
11534         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11535
11536 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11537
11538         PR target/19599
11539         PR target/57731
11540         PR target/57837
11541         * config/arm/arm.md ("*sibcall_insn): Replace use of
11542         Ss with US. Adjust output for v5 and v4t.
11543         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11544
11545         * config/arm/constraints.md ("Ss"): Rename to US.
11546
11547 2013-07-25  Terry Guo  <terry.guo@arm.com>
11548
11549         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11550         shift_add/shift_sub0/shift_sub1 RTXs.
11551
11552 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11553             Anton Blanchard  <anton@au1.ibm.com>
11554
11555         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11556         (altivec_vpks<VI_char>ss): Likewise.
11557         (altivec_vpks<VI_char>us): Likewise.
11558         (altivec_vpku<VI_char>us): Likewise.
11559         (altivec_vpku<VI_char>um): Likewise.
11560
11561 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11562
11563         Introduce context class.
11564
11565         * Makefile.in (CONTEXT_H): New.
11566         (OBJS): Add context.o.
11567         (toplev.o): Add CONTEXT_H to dependencies.
11568         (context.o): New.
11569
11570         * toplev.c (general_init): Create the singleton gcc::context instance.
11571
11572         * context.c: New.
11573
11574         * context.h: New.
11575
11576 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11577
11578         PR rtl-optimization/57968
11579         * mode-switching.c (create_pre_exit): Allow instructions that
11580         don't set a return register to need a non-exit mode.
11581
11582 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11583             Anton Blanchard  <anton@au1.ibm.com>
11584
11585         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11586         operands to vperm for little endian.
11587         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11588         of lvsl to create the control mask for a vperm for little endian.
11589
11590 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11591             Anton Blanchard  <anton@au1.ibm.com>
11592
11593         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11594         two operands for little-endian.
11595
11596 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11597
11598         * config/mips/mips.c (mips_case_values_threshold): New.
11599         (TARGET_CASE_VALUES_THRESHOLD): Define.
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): Correct
11605         selection of field for vector splat in little endian mode.
11606
11607 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11608
11609         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11610         expanders to rs6000.md.
11611         (ior<mode>3): Likewise.
11612         (and<mode>3): Likewise.
11613         (one_cmpl<mode>2): Likewise.
11614         (nor<mode>3): Likewise.
11615         (andc<mode>3): Likewise.
11616         (eqv<mode>3): Likewise.
11617         (nand<mode>3): Likewise.
11618         (orc<mode>3): Likewise.
11619
11620         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11621         declaration.
11622
11623         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11624         to split multi-word logical operations.
11625         (rs6000_split_logical_di): Likewise.
11626         (rs6000_split_logical): Likewise.
11627
11628         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11629         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11630         and allow TImode operations in 32-bit.
11631         (vsx_and<mode>3_64bit): Likewise.
11632         (vsx_ior<mode>3_32bit): Likewise.
11633         (vsx_ior<mode>3_64bit): Likewise.
11634         (vsx_xor<mode>3_32bit): Likewise.
11635         (vsx_xor<mode>3_64bit): Likewise.
11636         (vsx_one_cmpl<mode>2_32bit): Likewise.
11637         (vsx_one_cmpl<mode>2_64bit): Likewise.
11638         (vsx_nor<mode>3_32bit): Likewise.
11639         (vsx_nor<mode>3_64bit): Likewise.
11640         (vsx_andc<mode>3_32bit): Likewise.
11641         (vsx_andc<mode>3_64bit): Likewise.
11642         (vsx_eqv<mode>3_32bit): Likewise.
11643         (vsx_eqv<mode>3_64bit): Likewise.
11644         (vsx_nand<mode>3_32bit): Likewise.
11645         (vsx_nand<mode>3_64bit): Likewise.
11646         (vsx_orc<mode>3_32bit): Likewise.
11647         (vsx_orc<mode>3_64bit): Likewise.
11648
11649         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11650         logical types in GPRs.
11651
11652         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11653         logical insns to rs6000.md, and allow TImode operations in
11654         32-bit.
11655         (altivec_ior<mode>3): Likewise.
11656         (altivec_xor<mode>3): Likewise.
11657         (altivec_one_cmpl<mode>2): Likewise.
11658         (altivec_nor<mode>3): Likewise.
11659         (altivec_andc<mode>3): Likewise.
11660
11661         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11662         attributes for moving the 128-bit logical operations into
11663         rs6000.md.
11664         (BOOL_REGS_OUTPUT): Likewise.
11665         (BOOL_REGS_OP1): Likewise.
11666         (BOOL_REGS_OP2): Likewise.
11667         (BOOL_REGS_UNARY): Likewise.
11668         (BOOL_REGS_AND_CR0): Likewise.
11669         (one_cmpl<mode>2): Add support for DI logical operations on
11670         32-bit, splitting the operations to 32-bit.
11671         (anddi3): Likewise.
11672         (iordi3): Likewise.
11673         (xordi3): Likewise.
11674         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11675         changes to combine the 32/64-bit code, allow logical operations on
11676         TI mode in 32-bit, and to use similar match_operator patterns like
11677         scalar mode uses.  Combine the Altivec and VSX code for logical
11678         operations, and move it here.
11679         (ior<mode>3, 128-bit types): Likewise.
11680         (xor<mode>3, 128-bit types): Likewise.
11681         (one_cmpl<mode>3, 128-bit types): Likewise.
11682         (nor<mode>3, 128-bit types): Likewise.
11683         (andc<mode>3, 128-bit types): Likewise.
11684         (eqv<mode>3, 128-bit types): Likewise.
11685         (nand<mode>3, 128-bit types): Likewise.
11686         (orc<mode>3, 128-bit types): Likewise.
11687         (and<mode>3_internal): Likewise.
11688         (bool<mode>3_internal): Likewise.
11689         (boolc<mode>3_internal1): Likewise.
11690         (boolc<mode>3_internal2): Likewise.
11691         (boolcc<mode>3_internal1): Likewise.
11692         (boolcc<mode>3_internal2): Likewise.
11693         (eqv<mode>3_internal1): Likewise.
11694         (eqv<mode>3_internal2): Likewise.
11695         (one_cmpl1<mode>3_internal): Likewise.
11696
11697 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11698
11699         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11700         Rename flag_stack_usage to flag_stack_usage_info.
11701
11702 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11703
11704         * config/microblaze/sync.md: New file.
11705         * config/microblaze/microblaze.md: Include sync.md
11706         * config/microblaze/microblaze.c: Add print_operand 'y'.
11707         * config/microblaze/constraints.md: Add memory_contraint
11708         'Q' which is a single register.
11709
11710 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11711
11712         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11713
11714 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11715
11716         * reload.c (find_reloads): Exit loop once we find this operand
11717         cannot be reloaded somehow for this alternative.
11718
11719         * reload.c (find_reloads): Exit loop once we find a hard register.
11720
11721         * rtlanal.c (computed_jump_p): Exit loop once we find label
11722         reference is used.
11723
11724         * i386.c (ix86_pad_returns): Exit loop after setting replace.
11725
11726         * cfgloopmanip.c (remove_path): Exit loop after setting
11727         irred_invalidated.
11728
11729         * gensupport.c (subst_dup): Avoid loop if code is not
11730         MATCH_DUP nor MATCH_OP_DUP.
11731
11732 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
11733
11734         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
11735
11736 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11737
11738         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
11739         true for SP_REGNUM if mode == ptr_mode.
11740         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
11741         with value R0_REGNUM + 31.
11742
11743 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11744
11745         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
11746         pad pointer-typed argument downward.
11747
11748 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11749
11750         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
11751         and __ILP32__ when the ILP32 model is in use.
11752
11753 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11754
11755         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
11756         (aarch64_load_symref_appropriately): In the case of
11757         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
11758         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
11759         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
11760         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
11761         if 'mode' doesn't equal to 'ptr_mode'.
11762         (aarch64_output_mi_thunk): Add an assertion on the alignment of
11763         'vcall_offset'; change to call aarch64_emit_move differently depending
11764         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
11765         to calculate the upper bound of 'vcall_offset'.
11766         (aarch64_cannot_force_const_mem): Change to also return true if
11767         mode != ptr_mode.
11768         (aarch64_legitimize_reload_address): In the case of large
11769         displacements, add new local variable 'xmode' and an assertion
11770         based on it; change to use 'xmode' to generate the new rtx and
11771         reload.
11772         (aarch64_asm_trampoline_template): Change to generate the template
11773         differently depending on TARGET_ILP32 or not; change to use
11774         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
11775         (aarch64_trampoline_size): Removed.
11776         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
11777         and replace immediate literals with it.  Change to use 'ptr_mode'
11778         instead of 'DImode' and call convert_memory_address if the mode
11779         of 'fnaddr' doesn't equal to 'ptr_mode'.
11780         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
11781         to output symbol.
11782         (aarch64_elf_asm_destructor): Likewise.
11783         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
11784         on TARGET_ILP32 instead of aarch64_trampoline_size.
11785         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
11786         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
11787         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
11788         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
11789         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
11790         (storewb_pair<GPI:mode>_<P:mode>): ... this.
11791         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
11792         depending on the value of 'mode'.
11793         (add_losym_<mode>): New.
11794         (ldr_got_small_<mode>): New, based on ldr_got_small.
11795         (ldr_got_small): Remove.
11796         (ldr_got_small_sidi): New.
11797         * config/aarch64/iterators.md (P): New.
11798         (PTR): Change to 'ptr_mode' in the condition.
11799
11800 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11801
11802         * config.gcc (aarch64*-*-*): Support --with-abi.
11803         (aarch64*-*-elf): Support --with-multilib-list.
11804         (aarch64*-*-linux*): Likewise.
11805         (supported_defaults): Add abi to aarch64*-*-*.
11806         * configure.ac: Mention AArch64 for --with-multilib-list.
11807         * configure: Re-generated.
11808         * config/aarch64/biarchilp32.h: New file.
11809         * config/aarch64/biarchlp64.h: New file.
11810         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
11811         (ABI_SPEC): Ditto.
11812         (MULTILIB_DEFAULTS): Ditto.
11813         (DRIVER_SELF_SPECS): Ditto.
11814         (ASM_SPEC): Update to also substitute -mabi.
11815         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
11816         file whose name depends on -mabi= and -mbig-endian.
11817         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
11818         TARGET_ILP32.
11819         (POINTER_SIZE): New define.
11820         (POINTERS_EXTEND_UNSIGNED): Ditto.
11821         (enum aarch64_abi_type): New enumeration tag.
11822         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
11823         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
11824         (TARGET_ILP32): New define.
11825         * config/aarch64/aarch64.opt (mabi): New.
11826         (aarch64_abi): New.
11827         (ilp32, lp64): New values for -mabi.
11828         * config/aarch64/t-aarch64 (comma): New define.
11829         (MULTILIB_OPTIONS): Ditto.
11830         (MULTILIB_DIRNAMES): Ditto.
11831         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
11832         * doc/invoke.texi: Document -mabi for AArch64.
11833
11834 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
11835
11836         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
11837
11838 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11839             Anton Blanchard  <anton@au1.ibm.com>
11840
11841         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
11842         endianness when selecting field to splat.
11843
11844 2013-07-22  Eric Christopher  <echristo@gmail.com>
11845
11846         * dwarf2out.c (die_odr_checksum): New function to use
11847         CHECKSUM_ macros and ULEB128 for DIE tag.
11848         (generate_type_signature): Use.
11849
11850 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
11851
11852         * config.gcc (sparc*-*-*): Accept leon3 processor.
11853         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
11854         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
11855         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
11856         * config/sparc/sparc.opt (enum processor_type): Add leon3.
11857         (mfix-ut699): Adjust comment.
11858         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
11859         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
11860         (CPP_CPU_SPEC): Likewise.
11861         (ASM_CPU_SPEC): Likewise.
11862         * config/sparc/sparc.c (leon3_cost): New constant.
11863         (sparc_option_override): Add leon3 support.
11864         (mem_ref): New function.
11865         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
11866         (sparc_do_work_around_errata): Look into the instruction in the delay
11867         slot and adjust accordingly.  Add fix for the data cache nullify issues
11868         of the UT699.  Change insertion position for the NOP.
11869         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
11870         (leon3_load): New reservation.
11871         (leon_store): Bump latency to 2.
11872         (grfpu): New automaton.
11873         (grfpu_alu): New unit.
11874         (grfpu_ds): Likewise.
11875         (leon_fp_alu): Adjust.
11876         (leon_fp_mult): Delete.
11877         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
11878         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
11879         * config/sparc/sparc.md (cpu): Add leon3.
11880         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
11881         (swapsi): Likewise.
11882         (atomic_test_and_set): Likewise.
11883         (ldstub): Likewise.
11884
11885 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
11886
11887         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
11888         default for R5900 targets.
11889         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
11890         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
11891         * config/mips/mips.c (mips_option_override): Report an error for
11892         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
11893         for -march=r5900 -mhard-float.
11894
11895 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11896
11897         * df-problems.c (can_move_insns_across): Exit loop once we
11898         find a non-fixed, non-global register.
11899
11900         * ipa-pure-const.c (propagate_nothrow): Exit loop after
11901         setting can_throw.
11902
11903         * omega.c (omega_eliminate_red): Break after setting red_found.
11904         (omega_problem_has_red_equations): Similarly after setting found.
11905         (omega_query_variable): Similarly after setting coupled.
11906
11907 2013-07-22  Marek Polacek  <polacek@redhat.com>
11908
11909         * gimplify.c: Don't include gimple.h twice.
11910
11911 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11912
11913         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
11914         instead of TARGET_THUMB1.
11915         (Pz): New constraint.
11916         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
11917         encodings.
11918         (compare_negsi_si): Likewise.
11919         (compare_addsi2_op0): Likewise.
11920         (compare_addsi2_op1): Likewise.
11921         (addsi3_carryin_<optab>): Likewise.
11922         (addsi3_carryin_alt2_<optab>): Likewise.
11923         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
11924         for arm_restrict_it.
11925         (subsi3_carryin): Likewise.
11926         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
11927         (minmax_arithsi): Disable for arm_restrict_it.
11928         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
11929         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
11930         (satsi_<SAT:code>_shift): Likewise.
11931         (arm_shiftsi3): Add alternative for 16-bit encoding.
11932         (arm32_movhf): Disable for arm_restrict_it.
11933         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
11934         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
11935
11936 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
11937
11938         * config/arm/arm.md (attribute "insn"): Delete.
11939         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
11940         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
11941         (not_shiftsi): Update for attribute change.
11942         (not_shiftsi_compare0): Likewise.
11943         (not_shiftsi_compare0_scratch): Likewise.
11944         (arm_one_cmplsi2): Likewise.
11945         (thumb1_one_cmplsi2): Likewise.
11946         (notsi_compare0): Likewise.
11947         (notsi_compare0_scratch): Likewise.
11948         (thumb1_movdi_insn): Likewise.
11949         (arm_movsi_insn): Likewise.
11950         (movhi_insn_arch4): Likewise.
11951         (movhi_bytes): Likewise.
11952         (arm_movqi_insn): Likewise.
11953         (thumb1_movqi_insn): Likewise.
11954         (arm32_movhf): Likewise.
11955         (thumb1_movhf): Likewise.
11956         (arm_movsf_soft_insn): Likewise.
11957         (thumb1_movsf_insn): Likewise.
11958         (thumb_movdf_insn): Likewise.
11959         (movsicc_insn): Likewise.
11960         (movsfcc_soft_insn): Likewise.
11961         (and_scc): Likewise.
11962         (cond_move): Likewise.
11963         (if_move_not): Likewise.
11964         (if_not_move): Likewise.
11965         (if_shift_move): Likewise.
11966         (if_move_shift): Likewise.
11967         (if_shift_shift): Likewise.
11968         (if_not_arith): Likewise.
11969         (if_arith_not): Likewise.
11970         (cond_move_not): Likewise.
11971         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
11972         (neon_mov<mode>): Likewise.
11973         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
11974         (thumb2_movsi_vfp): Likewise.
11975         (movsf_vfp): Likewise.
11976         (thumb2_movsf_vfp): Likewise.
11977         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
11978         change.
11979         (cortexa7_older_only): Likewise.
11980         (cortexa7_younger): Likewise.
11981         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
11982         (1020alu_shift_op): Likewise.
11983         (1020alu_shift_reg_op): Likewise.
11984         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
11985         (alu_shift_op): Likewise.
11986         (alu_shift_reg_op): Likewise.
11987         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
11988         (11_alu_shift_op): Likewise.
11989         (11_alu_shift_reg_op): Likewise.
11990         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
11991         (9_alu_shift_reg_op): Likewise.
11992         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
11993         change.
11994         (cortex_a15_alu_shift): Likewise.
11995         (cortex_a15_alu_shift_reg): Likewise.
11996         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
11997         (cortex_a5_alu_shift): Likewise.
11998         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
11999         change.
12000         (cortex_a53_alu_shift): Likewise.
12001         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12002         change.
12003         (cortex_a7_alu_reg): Likewise.
12004         (cortex_a7_alu_shift): Likewise.
12005         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12006         (cortex_a8_alu_shift): Likewise.
12007         (cortex_a8_alu_shift_reg): Likewise.
12008         (cortex_a8_mov): Likewise.
12009         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12010         (cortex_a9_dp_shift): Likewise.
12011         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12012         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12013         (cortex_r4_mov): Likewise.
12014         (cortex_r4_alu_shift): Likewise.
12015         (cortex_r4_alu_shift_reg): Likewise.
12016         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12017         (526_alu_shift_op): Likewise.
12018         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12019         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12020         (626te_alu_shift_op): Likewise.
12021         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12022         (726te_alu_op): Likewise.
12023         (726te_alu_shift_op): Likewise.
12024         (726te_alu_shift_reg_op): Likewise.
12025         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12026         (mp626_alu_shift_op): Likewise.
12027         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12028         (pj4_alu_e1_conds): Likewise.
12029         (pj4_alu): Likewise.
12030         (pj4_alu_conds): Likewise.
12031         (pj4_shift): Likewise.
12032         (pj4_shift_conds): Likewise.
12033         (pj4_alu_shift): Likewise.
12034         (pj4_alu_shift_conds): Likewise.
12035
12036 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12037
12038         * config/arm/predicates.md (shiftable_operator_strict_it):
12039         New predicate.
12040         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12041         Disable cond_exec version for arm_restrict_it.
12042         (thumb2_smaxsi3): Convert to generate cond_exec.
12043         (thumb2_sminsi3): Likewise.
12044         (thumb32_umaxsi3): Likewise.
12045         (thumb2_uminsi3): Likewise.
12046         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12047         (thumb2_neg_abssi2): Likewise.
12048         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12049         (thumb2_movsicc_insn): Adjust alternatives.
12050         (thumb2_mov_negscc): Disable for arm_restrict_it.
12051         (thumb2_mov_negscc_strict_it): New pattern.
12052         (thumb2_mov_notscc_strict_it): New pattern.
12053         (thumb2_mov_notscc): Disable for arm_restrict_it.
12054         (thumb2_ior_scc): Likewise.
12055         (thumb2_ior_scc_strict_it): New pattern.
12056         (thumb2_cond_move): Adjust for arm_restrict_it.
12057         (thumb2_cond_arith): Disable for arm_restrict_it.
12058         (thumb2_cond_arith_strict_it): New pattern.
12059         (thumb2_cond_sub): Adjust for arm_restrict_it.
12060         (thumb2_movcond): Likewise.
12061         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12062         (thumb2_zero_extendhisi2_v6): Likewise.
12063         (thumb2_zero_extendqisi2_v6): Likewise.
12064         (orsi_notsi_si): Likewise.
12065         (orsi_not_shiftsi_si): Likewise.
12066
12067 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12068
12069         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12070         instruction sequence is 1 byte shorter.
12071
12072 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12073
12074         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12075         it is not needed after split.
12076
12077 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12078
12079         PR target/51784
12080         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12081         second label for nonlocal goto receivers. Don't output pic base labels
12082         unless we're producing PIC; mark that action unreachable().
12083         (ix86_save_reg): If the function contains a nonlocal label, save the
12084         PIC base reg.
12085         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12086         * config/darwin.c (emitted_pic_label_num): New GTY.
12087         (update_pic_label_number_if_needed): New.
12088         (machopic_output_function_base_name): Adjust for nonlocal receiver
12089         case.
12090         (machopic_should_output_picbase_label): New.
12091         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12092         (nonlocal_goto_receiver): New insn and split.
12093
12094 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12095
12096         * config/aarch64/aarch64-builtins.c
12097         (aarch64_fold_builtin): Fold abs in all modes.
12098         * config/aarch64/aarch64-simd-builtins.def
12099         (abs): Enable for all modes.
12100         * config/aarch64/arm_neon.h
12101         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12102         (vabs_f64): Add missing intrinsic.
12103
12104 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12105
12106         * config/aarch64/arm_neon.h (vabs_s64): New function
12107
12108 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12109
12110         PR target/57516
12111         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12112         * config/avr/avr.md (adjust_len): Add `round'.
12113         * config/avr/avr-protos.h (avr_out_round): New prototype.
12114         (avr_out_plus): Add `out_label' argument.
12115         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12116         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12117         Handle the case where `insn' is just a pattern.
12118         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12119         (avr_out_round): New function.
12120         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12121
12122 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12123
12124         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12125         Add check for flag_stack_usage to handle -fstack-usage support
12126
12127 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12128
12129         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12130         interaction for new Power8 flags and VSX.
12131
12132 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12133
12134         PR middle-end/57698
12135         * tree-inline.c (expand_call_inline): Emit errors during
12136         early_inlining only if optimization is not turned on.
12137
12138 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12139
12140         * passes.def: New.
12141
12142         * passes.c (init_optimization_passes): Move the construction of
12143         the pass hierarchy into a new passes.def file.
12144
12145         * Makefile.in (passes.o): Add dependency on passes.def.
12146
12147 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12148
12149         * passes.c (init_optimization_passes): Introduce macros for
12150         constructing the tree of passes (INSERT_PASSES_AFTER,
12151         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12152         TERMINATE_PASS_LIST).
12153
12154 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12155             Wei Mi  <wmi@google.com>
12156
12157         PR rtl-optimization/57878
12158         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12159         top.
12160         (reload_pseudo_compare_func): Check nregs first for reload
12161         pseudos.
12162
12163 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12164
12165         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12166
12167 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12168
12169         * read-rtl.c (validate_const_int): Once an invalid character is
12170         seen, quit the loop.
12171
12172         * gengtype.c (write_roots): Similarly once we find the "deletable"
12173         or "if_marked" option.
12174
12175 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12176
12177         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12178         "xtab" and "sat".  Move value "clz" from here to ...
12179         (attriubte "type"): ... here.
12180         (satsi_<SAT:code>): Delete "insn" attribute.
12181         (satsi_<SAT:code>_shift): Likewise.
12182         (arm_zero_extendqisi2addsi): Likewise.
12183         (arm_extendqisi2addsi): Likewise.
12184         (clzsi2): Update for attribute changes.
12185         (rbitsi2): Likewise.
12186         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12187         attribute.
12188         (arm_usatsihi): Likewise.
12189         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12190
12191 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12192
12193         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12194         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12195         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12196         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12197         in alphabetical order.
12198         (attribute "core_cycles"): Update for attribute changes.
12199         (arm_addsi3): Likewise.
12200         (addsi3_compare0): Likewise.
12201         (addsi3_compare0_scratch): Likewise.
12202         (addsi3_compare_op1): Likewise.
12203         (addsi3_compare_op2): Likewise.
12204         (compare_addsi2_op0): Likewise.
12205         (compare_addsi2_op1): Likewise.
12206         (addsi3_carryin_shift_<optab>): Likewise.
12207         (subsi3_carryin_shift): Likewise.
12208         (rsbsi3_carryin_shift): Likewise.
12209         (arm_subsi3_insn): Likewise.
12210         (subsi3_compare0): Likewise.
12211         (subsi3_compare): Likewise.
12212         (arm_andsi3_insn): Likewise.
12213         (thumb1_andsi3_insn): Likewise.
12214         (andsi3_compare0): Likewise.
12215         (andsi3_compare0_scratch): Likewise.
12216         (zeroextractsi_compare0_scratch
12217         (andsi_not_shiftsi_si): Likewise.
12218         (iorsi3_insn): Likewise.
12219         (iorsi3_compare0): Likewise.
12220         (iorsi3_compare0_scratch): Likewise.
12221         (arm_xorsi3): Likewise.
12222         (thumb1_xorsi3_insn): Likewise.
12223         (xorsi3_compare0): Likewise.
12224         (xorsi3_compare0_scratch): Likewise.
12225         (satsi_<SAT:code>_shift): Likewise.
12226         (rrx): Likewise.
12227         (arm_shiftsi3): Likewise.
12228         (shiftsi3_compare0): Likewise.
12229         (not_shiftsi): Likewise.
12230         (not_shiftsi_compare0): Likewise.
12231         (not_shiftsi_compare0_scratch): Likewise.
12232         (arm_one_cmplsi2): Likewise.
12233         (thumb_one_complsi2): Likewise.
12234         (notsi_compare0): Likewise.
12235         (notsi_compare0_scratch): Likewise.
12236         (thumb1_zero_extendhisi2): Likewise.
12237         (arm_zero_extendhisi2): Likewise.
12238         (arm_zero_extendhisi2_v6): Likewise.
12239         (arm_zero_extendhisi2addsi): Likewise.
12240         (thumb1_zero_extendqisi2): Likewise.
12241         (thumb1_zero_extendqisi2_v6): Likewise.
12242         (arm_zero_extendqisi2): Likewise.
12243         (arm_zero_extendqisi2_v6): Likewise.
12244         (arm_zero_extendqisi2addsi): Likewise.
12245         (thumb1_extendhisi2): Likewise.
12246         (arm_extendhisi2): Likewise.
12247         (arm_extendhisi2_v6): Likewise.
12248         (arm_extendqisi): Likewise.
12249         (arm_extendqisi_v6): Likewise.
12250         (arm_extendqisi2addsi): Likewise.
12251         (thumb1_extendqisi2): Likewise.
12252         (thumb1_movdi_insn): Likewise.
12253         (arm_movsi_insn): Likewise.
12254         (movsi_compare0): Likewise.
12255         (movhi_insn_arch4): Likewise.
12256         (movhi_bytes): Likewise.
12257         (arm_movqi_insn): Likewise.
12258         (thumb1_movqi_insn): Likewise.
12259         (arm32_movhf): Likewise.
12260         (thumb1_movhf): Likewise.
12261         (arm_movsf_soft_insn): Likewise.
12262         (thumb1_movsf_insn): Likewise.
12263         (movdf_soft_insn): Likewise.
12264         (thumb_movdf_insn): Likewise.
12265         (arm_cmpsi_insn): Likewise.
12266         (cmpsi_shiftsi): Likewise.
12267         (cmpsi_shiftsi_swp): Likewise.
12268         (arm_cmpsi_negshiftsi_si): Likewise.
12269         (movsicc_insn): Likewise.
12270         (movsfcc_soft_insn): Likewise.
12271         (arith_shiftsi): Likewise.
12272         (arith_shiftsi_compare0
12273         (arith_shiftsi_compare0_scratch
12274         (sub_shiftsi): Likewise.
12275         (sub_shiftsi_compare0
12276         (sub_shiftsi_compare0_scratch
12277         (and_scc): Likewise.
12278         (cond_move): Likewise.
12279         (if_plus_move): Likewise.
12280         (if_move_plus): Likewise.
12281         (if_move_not): Likewise.
12282         (if_not_move): Likewise.
12283         (if_shift_move): Likewise.
12284         (if_move_shift): Likewise.
12285         (if_shift_shift): Likewise.
12286         (if_not_arith): Likewise.
12287         (if_arith_not): Likewise.
12288         (cond_move_not): Likewise.
12289         (thumb1_ashlsi3): Set type attribute.
12290         (thumb1_ashrsi3): Likewise.
12291         (thumb1_lshrsi3): Likewise.
12292         (thumb1_rotrsi3): Likewise.
12293         (shiftsi3_compare0_scratch): Likewise.
12294         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12295         (neon_mov<mode>): Likewise.
12296         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12297         attribute changes.
12298         (thumb2_movsi_insn): Likewise.
12299         (thumb2_cmpsi_neg_shiftsi): Likewise.
12300         (thumb2_extendqisi_v6): Likewise.
12301         (thumb2_zero_extendhisi2_v6): Likewise.
12302         (thumb2_zero_extendqisi2_v6): Likewise.
12303         (thumb2_shiftsi3_short): Likewise.
12304         (thumb2_addsi3_compare0_scratch): Likewise.
12305         (orsi_not_shiftsi_si): Likewise.
12306         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12307         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12308         changes.
12309         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12310         (1020alu_shift_op): Likewise.
12311         (1020alu_shift_reg_op): Likewise.
12312         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12313         (alu_shift_op): Likewise.
12314         (alu_shift_reg_op): Likewise.
12315         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12316         (11_alu_shift_op): Likewise.
12317         (11_alu_shift_reg_op): Likewise.
12318         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12319         (9_alu_shift_reg_op): Likewise.
12320         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12321         changes.
12322         (cortex_a15_alu_shift): Likewise.
12323         (cortex_a15_alu_shift_reg): Likewise.
12324         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12325         changes.
12326         (cortex_a5_alu_shift): Likewise.
12327         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12328         changes.
12329         (cortex_a53_alu_shift): Likewise.
12330         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12331         changes.
12332         (cortex_a7_alu_reg): Likewise.
12333         (cortex_a7_alu_shift): Likewise.
12334         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12335         changes.
12336         (cortex_a8_alu_shift): Likewise.
12337         (cortex_a8_alu_shift_reg): Likewise.
12338         (cortex_a8_mov): Likewise.
12339         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12340         (cortex_a9_dp_shift): Likewise.
12341         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12342         changes.
12343         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12344         changes.
12345         (cortex_r4_mov): Likewise.
12346         (cortex_r4_alu_shift): Likewise.
12347         (cortex_r4_alu_shift_reg): Likewise.
12348         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12349         (526_alu_shift_op): Likewise.
12350         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12351         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12352         (626te_alu_shift_op): Likewise.
12353         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12354         (726te_alu_op): Likewise.
12355         (726te_alu_shift_op): Likewise.
12356         (726te_alu_shift_reg_op): Likewise.
12357         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12358         (mp626_alu_shift_op): Likewise.
12359         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12360         (pj4_alu_e1_conds): Likewise.
12361         (pj4_alu): Likewise.
12362         (pj4_alu_conds): Likewise.
12363         (pj4_shift): Likewise.
12364         (pj4_shift_conds): Likewise.
12365         (pj4_alu_shift): Likewise.
12366         (pj4_alu_shift_conds): Likewise.
12367         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12368         changes.
12369         (cortexa7_older_only): Likewise.
12370         (cortexa7_younger): Likewise.
12371
12372 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12373
12374         * ipa-pure-const.c (generate_summary): Rename to...
12375         (pure_const_generate_summary): ... this.
12376
12377 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12378
12379         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12380
12381 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12382
12383         PR target/57909
12384         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12385         usage in HI mode.
12386
12387 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12388
12389         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12390         enabled without -march=zEC12.
12391         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12392         flags to be set.
12393
12394 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12395
12396         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12397         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12398         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12399         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12400         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12401         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12402         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12403         ISA_HAS_FP4.
12404         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12405         and ISA_HAS_NMADD3_NMSUB3.
12406         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12407         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12408         (nmsub4<mode>, nmsub3<mode>): Likewise.
12409         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12410
12411 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12412
12413         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12414         TARGET_MIPS5400 checking.
12415
12416 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12417             Peter Bergner  <bergner@vnet.ibm.com>
12418
12419         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12420         registers in the comment.
12421         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12422         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12423         rather than FIRST_PSEUDO_REGISTERS.
12424
12425 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12426
12427         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12428         enable extra ISA flags with TARGET_HTM.
12429
12430 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12431
12432         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12433         Fix comment typos.
12434
12435 2013-07-15  Cong Hou  <congh@google.com>
12436
12437         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12438         in compare function for sorting.
12439
12440 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12441
12442         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12443         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12444         * config/rs6000/rs6000.opt: Add -mhtm option.
12445         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12446         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12447         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12448         __HTM__ if the HTM instructions are available.
12449         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12450         htm_spr_reg_operand): New define_predicates.
12451         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12452         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12453         Include htm.md.
12454         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12455         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12456         HTM builtin functions.
12457         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12458         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12459         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12460         (rs6000_builtin_mask_calculate): Likewise.
12461         (rs6000_option_override_internal): Likewise.
12462         (bdesc_htm): Add new HTM builtin support.
12463         (htm_spr_num): New function.
12464         (htm_spr_regno): Likewise.
12465         (rs6000_htm_spr_icode): Likewise.
12466         (htm_expand_builtin): Likewise.
12467         (htm_init_builtins): Likewise.
12468         (rs6000_expand_builtin): Add support for HTM builtin functions.
12469         (rs6000_init_builtins): Likewise.
12470         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12471         option.
12472         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12473         (TARGET_HTM, MASK_HTM): Define macros.
12474         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12475         (FIXED_REGISTERS): Likewise.
12476         (CALL_USED_REGISTERS): Likewise.
12477         (CALL_REALLY_USED_REGISTERS): Likewise.
12478         (REG_ALLOC_ORDER): Likewise.
12479         (enum reg_class): Likewise.
12480         (REG_CLASS_NAMES): Likewise.
12481         (REG_CLASS_CONTENTS): Likewise.
12482         (REGISTER_NAMES): Likewise.
12483         (ADDITIONAL_REGISTER_NAMES): Likewise.
12484         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12485         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12486         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12487         * config/rs6000/htm.md: New file.
12488         * config/rs6000/htmintrin.h: New file.
12489         * config/rs6000/htmxlintrin.h: New file.
12490
12491 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12492
12493         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12494         Define SYMBOL_TINY_GOT, update comment.
12495         * config/aarch64/aarch64.c
12496         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12497         (aarch64_expand_mov_immediate): Likewise.
12498         (aarch64_print_operand): Likewise.
12499         (aarch64_classify_symbol): Likewise.
12500         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12501         (ldr_got_tiny): Define.
12502
12503 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12504
12505         PR tree-optimization/54094
12506         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12507           scheduling dimension for the parallelism check from the polyhedral
12508           information in the AST.
12509         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12510           in 2D + 1 form.
12511
12512 2013-07-13  Jason Merrill  <jason@redhat.com>
12513
12514         * print-tree.c (debug_vec_tree): Use debug_raw.
12515         (debug_raw (vec<tree, va_gc> &)): New.
12516         (debug_raw (vec<tree, va_gc> *)): New.
12517         * tree.h: Declare them.
12518
12519 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12520
12521         * ifcvt.c (ifcvt_after_combine): New static variable.
12522         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12523         for size.
12524         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12525         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12526         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12527
12528 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12529
12530         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12531
12532 2013-07-12  Michael Matz  <matz@suse.de>
12533
12534         PR middle-end/55771
12535         * convert.c (convert_to_real): Reject non-float inner types.
12536
12537 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12538
12539         * config/aarch64/aarch64-protos.h
12540         (aarch64_simd_immediate_valid_for_move): Remove.
12541         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12542         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12543         cases.
12544         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12545
12546 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12547
12548         * config/mips/mips.c (mips_conditional_register_usage): Do not
12549         use t[0-7] registers in MIPS16 mode when optimizing for size.
12550
12551 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12552
12553         * config/i386/i386.c (dispatch_function_versions): Fix array
12554         indexing of function_version_info to match actual_versions.
12555
12556 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12557
12558         * vec.h (struct va_gc): Move release out-of-line.
12559         (va_gc::release): Call ggc_free on released vec.
12560
12561 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12562
12563         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12564         Require GOT register as additional operand in UNSPEC.
12565         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12566         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12567         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12568         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12569         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12570         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12571         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12572
12573 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12574
12575         PR target/57631
12576         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12577         name seen by assembler/linker rather if available.
12578
12579 2013-07-11  Andreas Schwab  <schwab@suse.de>
12580
12581         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12582
12583 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12584
12585         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12586
12587 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12588
12589         * doc/tm.texi.in: Move hook documentation to ....
12590         * target.def: ... here.
12591
12592         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12593         text on @hook line.
12594         * doc/tm.texi: Regenerate.
12595
12596 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12597
12598         PR c++/57869
12599         * doc/invoke.texi: Document -Wconditionally-supported.
12600
12601 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12602
12603         PR target/57844
12604         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12605         of my_fp.
12606
12607 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12608
12609         PR target/57506
12610         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12611         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12612         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12613         Remove duplicate devices.
12614         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12615         * config/avr/t-multilib: Regenerate.
12616         * config/avr/avr-tables.opt: Regenerate.
12617         * doc/avr-mmcu.texi: Regenerate.
12618
12619 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12620
12621         PR target/56987
12622         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12623
12624 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12625
12626         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12627         the cost of MULT when optimizing for size.
12628
12629 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12630
12631         * config/cr16/cr16-protos.h: Don't include target.h.
12632
12633 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12634
12635         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12636         adjust register size for TDmode and TFmode for VSX registers.
12637
12638 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12639
12640         PR target/56892
12641         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12642         hook_bool_const_tree_true.
12643
12644 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12645
12646         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12647         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12648         * config/s390/s390.md: Define FPR*_REGNUM constants.
12649         Fix FPR2_REGNUM constant (18 -> 17).
12650         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12651         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12652         ("trunc<BFP:mode><DFP_ALL:mode>2")
12653         ("trunc<DFP_ALL:mode><BFP:mode>2")
12654         ("*extend<BFP:mode><DFP_ALL:mode>2")
12655         ("*extend<DFP_ALL:mode><BFP:mode>2")
12656         ("extend<BFP:mode><DFP_ALL:mode>2")
12657         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12658         FPR4_REGNUM.
12659
12660 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12661
12662         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12663
12664 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12665
12666         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12667         and cfun_fpr_bit_p to cfun_fpr_save_p.
12668         (s390_frame_area, s390_register_info, s390_frame_info)
12669         (s390_emit_prologue, s390_emit_epilogue)
12670         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12671         register numbers.
12672         * config/s390/s390.h: Define *_REGNUM macros for floating point
12673         register numbers.
12674
12675 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12676
12677         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12678
12679 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12680
12681         PR rtl-optimization/57786
12682         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12683         and break out of the loop when it is set to false.
12684
12685 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12686
12687         PR target/57819
12688         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12689         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12690         (const_int 63)) 0)).
12691         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12692         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12693         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12694
12695         PR rtl-optimization/57829
12696         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12697         mask bits outside of mode are just sign-extension from mode to HWI.
12698
12699 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12700
12701         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12702         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12703         adjust_address instead of change_address to keep info about alignment.
12704         (emit_strmov): Remove.
12705         (emit_memmov): New function.
12706         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12707         (expand_movmem_epilogue): Likewise and return updated rtx for
12708         destination.
12709         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12710         destination and source.
12711         (decide_alignment): Refactor, handle vector_loop.
12712         (ix86_expand_movmem): Likewise.
12713         (ix86_expand_setmem): Likewise.
12714         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12715
12716 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
12717
12718         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
12719         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
12720
12721 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12722
12723         * config/i386/sse.md (sse_movlhps): Change alternative 3
12724         of operand 2 to "m".
12725
12726 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12727
12728         PR target/57807
12729         * config/i386/sse.md (iptr): New mode attribute.
12730         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
12731         (<sse>_vm<plusminus_insn><mode>3): Ditto.
12732         (<sse>_vmmul<mode>3): Ditto.
12733         (<sse>_vmdiv<mode>3): Ditto.
12734         (sse_vmrcpv4sf2): Ditto.
12735         (<sse>_vmsqrt<mode>2): Ditto.
12736         (sse_vmrsqrtv4sf2): Ditto.
12737         (<sse>_vm<code><mode>3): Ditto.
12738         (avx_vmcmp<mode>3): Ditto.
12739         (<sse>_vmmaskcmp<mode>3): Ditto.
12740         (<sse>_comi): Ditto.
12741         (<sse>_ucomi): Ditto.
12742         (*xop_vmfrcz_<mode>): Ditto.
12743         (*fmai_fmadd_<mode>): Ditto.
12744         (*fmai_fmsub_<mode>): Ditto.
12745         (*fmai_fnmadd_<mode>): Ditto.
12746         (*fmai_fnmsub_<mode>): Ditto.
12747         (*fma4i_vmfmadd_<mode>): Ditto.
12748         (*fma4i_vmfmsub_<mode>): Ditto.
12749         (*fma4i_vmfnmadd_<mode>): Ditto.
12750         (*fma4i_vmfnmsub_<mode>): Ditto.
12751         (*xop_vmfrcz_<mode>): Ditto.
12752         (sse_cvtps2pi): Ditto.
12753         (sse_cvttps2pi): Ditto.
12754         (sse_cvtss2si): Ditto.
12755         (sse_cvtss2si_2): Ditto.
12756         (sse_cvtss2siq_2): Ditto.
12757         (sse_cvttss2si): Ditto.
12758         (sse_cvttss2siq): Ditto.
12759         (sse_cvtsd2si): Ditto.
12760         (sse_cvtsd2si_2): Ditto.
12761         (sse_cvtsd2siq_2): Ditto.
12762         (sse_cvttsd2si): Ditto.
12763         (sse_cvttsd2siq): Ditto.
12764         (sse_cvtsd2ss): Ditto.
12765         (sse_cvtss2sd): Ditto.
12766         (avx2_pbroadcast<mode>): Ditto.
12767         (avx2_pbroadcast<mode>_1): Ditto.
12768         (*avx_vperm_broadcast_v4sf): Ditto.
12769
12770         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
12771         (sse_movlhps): Ditto.
12772         (sse_storehps): Ditto.
12773         (sse_loadhps): Ditto.
12774         (sse_storelps): Ditto.
12775         (sse_loadlps): Ditto.
12776         (*vec_concatv4sf): Ditto.
12777         (*vec_interleave_highv2df): Ditto.
12778         (*vec_interleave_lowv2df): Ditto.
12779         (*vec_extractv2df_1_sse): Ditto.
12780         (*vec_extractv2df_0_sse): Ditto.
12781         (sse2_storelpd): Ditto.
12782         (sse2_loadlpd): Ditto.
12783         (sse2_movsd): Ditto.
12784         (*vec_concatv4si): Ditto.
12785         (vec_concatv2di): Ditto.
12786
12787         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
12788         for Intel asm dialect.
12789         (mmx_punpcklwd): Ditto.
12790         (mmx_punpckldq): Ditto.
12791
12792         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
12793         for intel assembler dialect.
12794
12795 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
12796
12797         PR target/29776
12798         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
12799         for BUILT_IN_C{LZ,LRSB}*.
12800         * tree.h (CASE_INT_FN): Add FN##IMAX case.
12801         * tree-vrp.c (extract_range_basic): Handle
12802         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
12803         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
12804         fall thru to code calling set_value*.
12805         * builtins.c (expand_builtin): Remove *IMAX cases.
12806         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
12807         if width is bigger than 2*HWI.
12808
12809 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
12810
12811         PR rtl-optimization/55342
12812         * lra-int.h (lra_subreg_reload_pseudos): New.
12813         * lra.c: Add undoing optional reloads to the block diagram.
12814         (lra_subreg_reload_pseudos): New.
12815         (lra_optional_reload_pseudos): Change comments.
12816         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
12817         lra_optional_reload_pseudos after undo transformations.
12818         * lra-assigns.c (pseudo_prefix_title): New.
12819         (lra_setup_reg_renumber): Use it.
12820         (spill_for): Ditto.  Check subreg reload pseudos too.
12821         (assign_by_spills): Consider subreg reload pseudos too.
12822         * lra-constraints.c (simplify_operand_subreg): Use
12823         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
12824         (curr_insn_transform): Recognize and do optional reloads.
12825         (undo_optional_reloads): New.
12826         (lra_undo_inheritance): Call undo_optional_reloads.
12827
12828 2013-07-05  Thomas Quinot  <quinot@adacore.com>
12829
12830         * tree-complex.c (expand_complex_operations_1): Fix typo.
12831
12832 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
12833
12834         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
12835         (tune_params): New member 'const vec_costs'.
12836         * config/aarch64/aarch64.c (generic_vector_cost): New.
12837         (generic_tunings): New member 'generic_vector_cost'.
12838         (aarch64_builtin_vectorization_cost): New.
12839         (aarch64_add_stmt_cost): New.
12840         (TARGET_VECTORIZE_ADD_STMT_COST): New.
12841         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
12842
12843 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
12844
12845         PR target/57777
12846         * config/i386/predicates.md (vsib_address_operand): Disallow
12847         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
12848
12849 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
12850
12851         PR middle-end/55030
12852         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
12853         expand_builtin_setjmp_receiver.
12854         (expand_label): Adjust, call expand_builtin_setjmp_receiver
12855         with NULL for the label parameter.
12856         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
12857         the frame-pointer.  Adjust comments.
12858         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
12859         only if LABEL is non-NULL.
12860
12861 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
12862
12863         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
12864         (ARM_ABI_AAPCS64): Ditto.
12865         (arm_abi): Ditto.
12866         (ARM_DEFAULT_ABI): Ditto.
12867
12868 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
12869
12870         * config/aarch64/aarch64-builtins.c
12871         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
12872         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
12873         (st1): Likewise.
12874         * config/aarch64/aarch64-simd.md
12875         (aarch64_ld1<VALL:mode>): New.
12876         (aarch64_st1<VALL:mode>): Likewise.
12877         * config/aarch64/arm_neon.h
12878         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
12879
12880 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
12881
12882         * config/i386/i386.c (gate_insert_vzeroupper): Check if
12883         target ISA is AVX.
12884         (ix86_option_override_internal):Turn on all -mavx target flags by
12885         default as they are dependent on AVX anyway.
12886
12887 2013-07-02  Cary Coutant  <ccoutant@google.com>
12888
12889         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
12890         deterministic hash.
12891         (loc_checksum_ordered): Likewise.
12892         (hash_loc_operands): Remove inline keyword.
12893
12894 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
12895
12896         PR tree-optimization/57741
12897         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
12898         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
12899         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
12900         Allow REAL_CST step_exprs if flag_associative_math.
12901         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
12902
12903 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
12904
12905         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
12906
12907 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
12908
12909         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
12910
12911 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12912
12913         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
12914         encoding.
12915         (iorsi3_insn): Likewise.
12916         (arm_xorsi3): Likewise.
12917
12918 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
12919
12920         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
12921         to ...
12922         (attribute "type"): ... here, and prefix with "wmmx_".
12923         (attribute "core_cycles"): Update for attribute changes.
12924         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
12925         (tbcstv4hi): Likewise.
12926         (tbcstv2si): Likewise.
12927         (iwmmxt_iordi3): Likewise.
12928         (iwmmxt_xordi3): Likewise.
12929         (iwmmxt_anddi3): Likewise.
12930         (iwmmxt_nanddi3): Likewise.
12931         (iwmmxt_arm_movdi): Likewise.
12932         (iwmmxt_movsi_insn): Likewise.
12933         (mov<mode>_internal): Likewise.
12934         (and<mode>3_iwmmxt): Likewise.
12935         (ior<mode>3_iwmmxt): Likewise.
12936         (xor<mode>3_iwmmxt): Likewise.
12937         (add<mode>3_iwmmxt): Likewise.
12938         (ssaddv8qi3): Likewise.
12939         (ssaddv4hi3): Likewise.
12940         (ssaddv2si3): Likewise.
12941         (usaddv8qi3): Likewise.
12942         (usaddv4hi3): Likewise.
12943         (usaddv2si3): Likewise.
12944         (sub<mode>3_iwmmxt): Likewise.
12945         (sssubv8qi3): Likewise.
12946         (sssubv4hi3): Likewise.
12947         (sssubv2si3): Likewise.
12948         (ussubv8qi3): Likewise.
12949         (ussubv4hi3): Likewise.
12950         (ussubv2si3): Likewise.
12951         (mulv4hi3_iwmmxt): Likewise.
12952         (smulv4hi3_highpart): Likewise.
12953         (umulv4hi3_highpart): Likewise.
12954         (iwmmxt_wmacs): Likewise.
12955         (iwmmxt_wmacsz): Likewise.
12956         (iwmmxt_wmacu): Likewise.
12957         (iwmmxt_wmacuz): Likewise.
12958         (iwmmxt_clrdi): Likewise.
12959         (iwmmxt_clrv8qi): Likewise.
12960         (iwmmxt_clr4hi): Likewise.
12961         (iwmmxt_clr2si): Likewise.
12962         (iwmmxt_uavgrndv8qi3): Likewise.
12963         (iwmmxt_uavgrndv4hi3): Likewise.
12964         (iwmmxt_uavgv8qi3): Likewise.
12965         (iwmmxt_uavgv4hi3): Likewise.
12966         (iwmmxt_tinsrb): Likewise.
12967         (iwmmxt_tinsrh): Likewise.
12968         (iwmmxt_tinsrw): Likewise.
12969         (iwmmxt_textrmub): Likewise.
12970         (iwmmxt_textrmsb): Likewise.
12971         (iwmmxt_textrmuh): Likewise.
12972         (iwmmxt_textrmsh): Likewise.
12973         (iwmmxt_textrmw): Likewise.
12974         (iwmxxt_wshufh): Likewise.
12975         (eqv8qi3): Likewise.
12976         (eqv4hi3): Likewise.
12977         (eqv2si3): Likewise.
12978         (gtuv8qi3): Likewise.
12979         (gtuv4hi3): Likewise.
12980         (gtuv2si3): Likewise.
12981         (gtv8qi3): Likewise.
12982         (gtv4hi3): Likewise.
12983         (gtv2si3): Likewise.
12984         (smax<mode>3_iwmmxt): Likewise.
12985         (umax<mode>3_iwmmxt): Likewise.
12986         (smin<mode>3_iwmmxt): Likewise.
12987         (umin<mode>3_iwmmxt): Likewise.
12988         (iwmmxt_wpackhss): Likewise.
12989         (iwmmxt_wpackwss): Likewise.
12990         (iwmmxt_wpackdss): Likewise.
12991         (iwmmxt_wpackhus): Likewise.
12992         (iwmmxt_wpackwus): Likewise.
12993         (iwmmxt_wpackdus): Likewise.
12994         (iwmmxt_wunpckihb): Likewise.
12995         (iwmmxt_wunpckihh): Likewise.
12996         (iwmmxt_wunpckihw): Likewise.
12997         (iwmmxt_wunpckilb): Likewise.
12998         (iwmmxt_wunpckilh): Likewise.
12999         (iwmmxt_wunpckilw): Likewise.
13000         (iwmmxt_wunpckehub): Likewise.
13001         (iwmmxt_wunpckehuh): Likewise.
13002         (iwmmxt_wunpckehuw): Likewise.
13003         (iwmmxt_wunpckehsb): Likewise.
13004         (iwmmxt_wunpckehsh): Likewise.
13005         (iwmmxt_wunpckehsw): Likewise.
13006         (iwmmxt_wunpckelub): Likewise.
13007         (iwmmxt_wunpckeluh): Likewise.
13008         (iwmmxt_wunpckeluw): Likewise.
13009         (iwmmxt_wunpckelsb): Likewise.
13010         (iwmmxt_wunpckelsh): Likewise.
13011         (iwmmxt_wunpckelsw): Likewise.
13012         (ror<mode>3): Likewise.
13013         (ashr<mode>3_iwmmxt): Likewise.
13014         (lshr<mode>3_iwmmxt): Likewise.
13015         (ashl<mode>3_iwmmxt): Likewise.
13016         (ror<mode>3_di): Likewise.
13017         (ashr<mode>3_di): Likewise.
13018         (lshr<mode>3_di): Likewise.
13019         (ashl<mode>3_di): Likewise.
13020         (iwmmxt_wmadds): Likewise.
13021         (iwmmxt_wmaddu): Likewise.
13022         (iwmmxt_tmia): Likewise.
13023         (iwmmxt_tmiaph): Likewise.
13024         (iwmmxt_tmiabb): Likewise.
13025         (iwmmxt_tmiatb): Likewise.
13026         (iwmmxt_tmiabt): Likewise.
13027         (iwmmxt_tmiatt): Likewise.
13028         (iwmmxt_tmovmskb): Likewise.
13029         (iwmmxt_tmovmskh): Likewise.
13030         (iwmmxt_tmovmskw): Likewise.
13031         (iwmmxt_waccb): Likewise.
13032         (iwmmxt_wacch): Likewise.
13033         (iwmmxt_waccw): Likewise.
13034         (iwmmxt_waligni): Likewise.
13035         (iwmmxt_walignr): Likewise.
13036         (iwmmxt_walignr0): Likewise.
13037         (iwmmxt_walignr1): Likewise.
13038         (iwmmxt_walignr2): Likewise.
13039         (iwmmxt_walignr3): Likewise.
13040         (iwmmxt_wsadb): Likewise.
13041         (iwmmxt_wsadh): Likewise.
13042         (iwmmxt_wsadbz): Likewise.
13043         (iwmmxt_wsadhz): Likewise.
13044         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13045         (iwmmxt_wabsdiffb): Likewise.
13046         (iwmmxt_wabsdiffh): Likewise.
13047         (iwmmxt_wabsdiffw): Likewise.
13048         (iwmmxt_waddsubhx): Likewise
13049         (iwmmxt_wsubaddhx): Likewise.
13050         (addc<mode>3): Likewise.
13051         (iwmmxt_avg4): Likewise.
13052         (iwmmxt_avg4r): Likewise.
13053         (iwmmxt_wmaddsx): Likewise.
13054         (iwmmxt_wmaddux): Likewise.
13055         (iwmmxt_wmaddsn): Likewise.
13056         (iwmmxt_wmaddun): Likewise.
13057         (iwmmxt_wmulwsm): Likewise.
13058         (iwmmxt_wmulwum): Likewise.
13059         (iwmmxt_wmulsmr): Likewise.
13060         (iwmmxt_wmulumr): Likewise.
13061         (iwmmxt_wmulwsmr): Likewise.
13062         (iwmmxt_wmulwumr): Likewise.
13063         (iwmmxt_wmulwl): Likewise.
13064         (iwmmxt_wqmulm): Likewise.
13065         (iwmmxt_wqmulwm): Likewise.
13066         (iwmmxt_wqmulmr): Likewise.
13067         (iwmmxt_wqmulwmr): Likewise.
13068         (iwmmxt_waddbhusm): Likewise.
13069         (iwmmxt_waddbhusl): Likewise.
13070         (iwmmxt_wqmiabb): Likewise.
13071         (iwmmxt_wqmiabt): Likewise.
13072         (iwmmxt_wqmiatb): Likewise.
13073         (iwmmxt_wqmiatt): Likewise.
13074         (iwmmxt_wqmiabbn): Likewise.
13075         (iwmmxt_wqmiabtn): Likewise.
13076         (iwmmxt_wqmiatbn): Likewise.
13077         (iwmmxt_wqmiattn): Likewise.
13078         (iwmmxt_wmiabb): Likewise.
13079         (iwmmxt_wmiabt): Likewise.
13080         (iwmmxt_wmiatb): Likewise.
13081         (iwmmxt_wmiatt): Likewise.
13082         (iwmmxt_wmiabbn): Likewise.
13083         (iwmmxt_wmiabtn): Likewise.
13084         (iwmmxt_wmiatbn): Likewise.
13085         (iwmmxt_wmiattn): Likewise.
13086         (iwmmxt_wmiawbb): Likewise.
13087         (iwmmxt_wmiawbt): Likewise.
13088         (iwmmxt_wmiawtb): Likewise.
13089         (iwmmxt_wmiawtt): Likewise.
13090         (iwmmxt_wmiawbbn): Likewise.
13091         (iwmmxt_wmiawbtn): Likewise.
13092         (iwmmxt_wmiawtbn): Likewise.
13093         (iwmmxt_wmiawttn): Likewise.
13094         (iwmmxt_wmerge): Likewise.
13095         (iwmmxt_tandc<mode>3): Likewise.
13096         (iwmmxt_torc<mode>3): Likewise.
13097         (iwmmxt_torvsc<mode>3): Likewise.
13098         (iwmmxt_textrc<mode>3): Likewise.
13099         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13100         (wmmxt_pack): Likewise.
13101         (wmmxt_mult_c1): Likewise.
13102         (wmmxt_mult_c2): Likewise.
13103         (wmmxt_alu_c1): Likewise.
13104         (wmmxt_alu_c2): Likewise.
13105         (wmmxt_alu_c3): Likewise.
13106         (wmmxt_transfer_c1): Likewise.
13107         (wmmxt_transfer_c2): Likewise.
13108         (wmmxt_transfer_c3): Likewise.
13109         (marvell_f_iwmmxt_wstr): Likewise.
13110         (marvell_f_iwmmxt_wldr): Likewise.
13111
13112 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13113
13114         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13115
13116 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13117
13118         Revert:
13119         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13120         * lra-constraints.c (need_for_split_p): Check call used hard regs
13121         living through calls.
13122
13123         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13124         call used regs for call insn.
13125
13126 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13127
13128         PR target/57736
13129         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13130         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13131
13132 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13133
13134         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13135
13136 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13137
13138         * lra-constraints.c (need_for_split_p): Check call used hard regs
13139         living through calls.
13140
13141 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13142
13143         PR target/57744
13144         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13145         to tie with any other modes.  Eliminate Altivec vector mode tests,
13146         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13147         code, to return 0 if testing MODE2 for a condition, if we've
13148         already tested MODE1 for the same condition.
13149
13150 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13151
13152         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13153         layout.
13154
13155 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13156
13157         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13158         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13159
13160 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13161
13162         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13163         Define.
13164         (aarch64_symbolic_constant_p): Remove.
13165         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13166         static.  Fix line length and white space.
13167         (aarch64_symbolic_constant_p): Remove.
13168         * config/aarch64/predicates.md (aarch64_valid_symref):
13169         Use aarch64_classify_symbol_expression.
13170
13171 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13172
13173         * config/arm/constraints.md (Ts): New constraint.
13174         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13175         16-bit encodings.
13176         (compare_scc): Use "Ts" constraint for operand 0.
13177         (ior_scc_scc): Likewise.
13178         (and_scc_scc): Likewise.
13179         (and_scc_scc_nodom): Likewise.
13180         (ior_scc_scc_cmp): Likewise for operand 7.
13181         (and_scc_scc_cmp): Likewise.
13182         * config/arm/thumb2.md (thumb2_movsi_insn):
13183         Add alternatives for 16-bit encodings.
13184         (thumb2_movhi_insn): Likewise.
13185         (thumb2_movsicc_insn): Likewise.
13186         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13187         (thumb2_negscc): Use "Ts" constraint.
13188         Move mvn instruction outside cond_exec block.
13189         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13190         for 16-bit encodings.
13191
13192 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13193
13194         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13195         encoding.
13196         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13197         (mulsi3subsi): Likewise.
13198         (mulsidi3adddi): Likewise.
13199         (mulsidi3_v6): Likewise.
13200         (umulsidi3_v6): Likewise.
13201         (umulsidi3adddi_v6): Likewise.
13202         (smulsi3_highpart_v6): Likewise.
13203         (umulsi3_highpart_v6): Likewise.
13204         (mulhisi3tb): Likewise.
13205         (mulhisi3bt): Likewise.
13206         (mulhisi3tt): Likewise.
13207         (maddhisi4): Likewise.
13208         (maddhisi4tb): Likewise.
13209         (maddhisi4tt): Likewise.
13210         (maddhidi4): Likewise.
13211         (maddhidi4tb): Likewise.
13212         (maddhidi4tt): Likewise.
13213         (zeroextractsi_compare0_scratch): Likewise.
13214         (insv_zero): Likewise.
13215         (insv_t2): Likewise.
13216         (anddi_notzesidi_di): Likewise.
13217         (anddi_notsesidi_di): Likewise.
13218         (andsi_notsi_si): Likewise.
13219         (iordi_zesidi_di): Likewise.
13220         (xordi_zesidi_di): Likewise.
13221         (andsi_iorsi3_notsi): Likewise.
13222         (smax_0): Likewise.
13223         (smax_m1): Likewise.
13224         (smin_0): Likewise.
13225         (not_shiftsi): Likewise.
13226         (unaligned_loadsi): Likewise.
13227         (unaligned_loadhis): Likewise.
13228         (unaligned_loadhiu): Likewise.
13229         (unaligned_storesi): Likewise.
13230         (unaligned_storehi): Likewise.
13231         (extv_reg): Likewise.
13232         (extzv_t2): Likewise.
13233         (divsi3): Likewise.
13234         (udivsi3): Likewise.
13235         (arm_zero_extendhisi2addsi): Likewise.
13236         (arm_zero_extendqisi2addsi): Likewise.
13237         (compareqi_eq0): Likewise.
13238         (arm_extendhisi2_v6): Likewise.
13239         (arm_extendqisi2addsi): Likewise.
13240         (arm_movt): Likewise.
13241         (thumb2_ldrd): Likewise.
13242         (thumb2_ldrd_base): Likewise.
13243         (thumb2_ldrd_base_neg): Likewise.
13244         (thumb2_strd): Likewise.
13245         (thumb2_strd_base): Likewise.
13246         (thumb2_strd_base_neg): Likewise.
13247         (arm_negsi2): Add alternative for 16-bit encoding.
13248         (arm_one_cmplsi2): Likewise.
13249
13250 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13251
13252         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13253         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13254         (movdfcc): Likewise.
13255         * config/arm/vfp.md (*thumb2_movsf_vfp):
13256         Disable predication for arm_restrict_it.
13257         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13258         (*thumb2_movdfcc_vfp): Likewise.
13259         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13260         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13261         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13262         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13263         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13264         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13265         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13266         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13267         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13268         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13269         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13270         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13271         Disable predication for arm_restrict_it.
13272
13273 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13274
13275         * config/i386/bmiintrin.h (_bextr_u32): New.
13276         (_bextr_u64): Ditto.
13277
13278 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13279
13280         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13281         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13282         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13283         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13284         * config/mips/n32-elf.h: ...this new file.
13285
13286 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13287
13288         PR target/57224
13289         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13290         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13291
13292 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13293
13294         * config/mips/mips-tables.opt: Regenerate.
13295         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13296         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13297         * doc/invoke.texi: Add -m14kc.
13298
13299 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13300
13301         PR target/57623
13302         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13303         constraints of operand 1 and 2.
13304
13305         PR target/57623
13306         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13307         to match RTL canonicalization.  Swap predicates and
13308         constraints of operand 1 and 2.
13309
13310 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13311
13312         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13313         Process OP_INOUT regs for splitting too.
13314
13315 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13316
13317         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13318         decl before the loop, initialize to NULL.
13319         (vectorizable_load): Initialize ptr_incr to NULL.
13320
13321 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13322
13323         PR lto/57208
13324         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13325         * ipa-ref.c (ipa_maybe_record_reference): New function.
13326         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13327         * ipa-cp.c (create_specialized_node): Record potential references from
13328         aggvals.
13329         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13330
13331 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13332
13333         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13334         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13335         'mode' to force_reg.
13336         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13337         (aarch64_expand_mov_immediate): Likewise.
13338
13339 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13340
13341         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13342         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13343
13344 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13345
13346         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13347         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13348         (struct machine_function): Add tbegin_p.
13349         (s390_canonicalize_comparison): Fold CC mode compares to
13350         conditional jump if possible.
13351         (s390_emit_jump): Return the emitted jump.
13352         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13353         Handle CCRAWmode compares.
13354         (s390_option_override): Default to -mhtm if available.
13355         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13356         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13357         FPRs instead of df_regs_ever_live_p.
13358         (s390_optimize_nonescaping_tx): New function.
13359         (s390_init_frame_layout): Extend clobbered_regs array to cover
13360         FPRs as well.
13361         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13362         (s390_expand_tbegin): New function.
13363         (enum s390_builtin): New enum definition.
13364         (code_for_builtin): New array definition.
13365         (s390_init_builtins): New function.
13366         (s390_expand_builtin): New function.
13367         (TARGET_INIT_BUILTINS): Define.
13368         (TARGET_EXPAND_BUILTIN): Define.
13369         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13370         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13371         (s390_alc_comparison): Likewise.
13372         * config/s390/s390-modes.def: Add CCRAWmode.
13373         * config/s390/s390.h (processor_flags): Add PF_TX.
13374         (TARGET_CPU_HTM): Define macro.
13375         (TARGET_HTM): Define macro.
13376         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13377         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13378         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13379         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13380         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13381         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13382         ("*cc_to_int"): Move up.
13383         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13384         constants other than 0.
13385         ("*ccraw_to_int"): New insn and splitter definition.
13386         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13387         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13388         ("tx_assist"): New expander.
13389         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13390         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13391         * config/s390/s390.opt: Add -mhtm option.
13392         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13393         * config/s390/htmxlintrin.h: New file.
13394         * config/s390/htmintrin.h: New file.
13395         * config/s390/s390intrin.h: New file.
13396         * doc/extend.texi: Document htm builtins.
13397         * config.gcc: Add the new header files to extra_headers.
13398
13399 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13400
13401         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13402         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13403
13404 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13405             Pat Haugen  <pthaugen@us.ibm.com>
13406             Peter Bergner  <bergner@vnet.ibm.com>
13407
13408         * config/rs6000/power8.md: New.
13409         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13410         setting for power8 entry.
13411         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13412         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13413         test for Power4/Power5 only.
13414         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13415         support.
13416         (force_new_group): Adjust comment.
13417         * config/rs6000/rs6000.md: Include power8.md.
13418
13419 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13420
13421         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13422         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13423         declaration.
13424         (tune_params): Update comment.
13425         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13426         (arm_max_conditional_execute): New function.
13427         (thumb2_final_prescan_insn): Use max_insn_skipped and
13428         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13429
13430 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13431
13432         PR tree-optimization/57705
13433         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13434         SSA_NAME step, provided that it is not defined inside the loop.
13435         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13436         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13437
13438 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13439
13440         PR middle-end/57670
13441         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13442         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13443         calls in the dump.
13444         (ipa_note_param_call): Initialize member_ptr flag.
13445         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13446         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13447         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13448         (ipa_read_indirect_edge_info): Likewise.
13449
13450 2013-06-25  Richard Biener  <rguenther@suse.de>
13451
13452         PR middle-end/56977
13453         * passes.c (init_optimization_passes): Move pass_fold_builtins
13454         and pass_dce earlier with -Og.
13455
13456 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13457
13458         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13459         <BIT_FIELD_REF>: Remove trailing TAB.
13460         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13461         remove blank line.
13462
13463 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13464
13465         PR tree-optimization/57358
13466         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13467         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13468         (ipa_analyze_params_uses): Generate pessimistic info when true.
13469
13470 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13471
13472         PR tree-optimization/57539
13473         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13474         global.inlined_to of the new node to it.  All callers changed.
13475         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13476         inlining_into, pass it to cgraph_clone_node.
13477         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13478         ipa_free_edge_args_substructures.
13479         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13480         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13481         Assert we have found an rdesc in the rdesc list.
13482
13483 2013-06-24  Richard Biener  <rguenther@suse.de>
13484
13485         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13486         (pointer_set_lookup): Declare.
13487         (class pointer_map): New template class implementing a
13488         generic pointer to T map.
13489         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13490         pointer_map<T>::contains, pointer_map<T>::insert,
13491         pointer_map<T>::traverse): New functions.
13492         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13493         (pointer_set_lookup): New function.
13494         (pointer_set_contains): Use pointer_set_lookup.
13495         (pointer_set_insert): Likewise.
13496         (insert_aux): Remove.
13497         (struct pointer_map_t): Embed a pointer_set_t.
13498         (pointer_map_create): Adjust.
13499         (pointer_map_destroy): Likewise.
13500         (pointer_map_contains): Likewise.
13501         (pointer_map_insert): Likewise.
13502         (pointer_map_traverse): Likewise.
13503         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13504         pointer_map<unsigned> instead of a pointer_map_t.
13505         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13506         (streamer_tree_cache_lookup): Likewise.
13507         (streamer_tree_cache_create): Likewise.
13508         (streamer_tree_cache_delete): Likewise.
13509         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13510         pointer_map<unsigned> instead of a pointer_map_t.
13511         (lto_init_tree_ref_encoder): Adjust.
13512         (lto_destroy_tree_ref_encoder): Likewise.
13513         * lto-section-out.c (lto_output_decl_index): Likewise.
13514         (lto_record_function_out_decl_state): Likewise.
13515         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13516
13517 2013-06-24  Richard Biener  <rguenther@suse.de>
13518
13519         PR tree-optimization/57488
13520         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13521
13522 2013-06-24  Alan Modra  <amodra@gmail.com>
13523
13524         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13525         (gen_easy_altivec_constant): Likewise.
13526         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13527         easy_vector_constant_msb): Likewise.
13528
13529 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13530
13531         PR target/57688
13532         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13533         add missing return true.
13534
13535 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13536
13537         PR target/52483
13538         * config/sh/predicates.md (general_extend_operand): Invoke
13539         general_movsrc_operand for memory operands.
13540         (general_movsrc_operand): Allow reg+reg addressing, do not use
13541         general_operand for memory operands.
13542
13543 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13544
13545         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13546         when current target options does not apply.
13547         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13548         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13549         * config/i386/bmiintrin.h: Pass appropriate target
13550         attributes to header.
13551         * config/i386/mmintrin.h: Ditto.
13552         * config/i386/nmmintrin.h: Ditto.
13553         * config/i386/avx2intrin.h: Ditto.
13554         * config/i386/fxsrintrin.h: Ditto.
13555         * config/i386/tbmintrin.h: Ditto.
13556         * config/i386/xsaveintrin.h: Ditto.
13557         * config/i386/f16cintrin.h: Ditto.
13558         * config/i386/xtestintrin.h: Ditto.
13559         * config/i386/xsaveoptintrin.h: Ditto.
13560         * config/i386/bmi2intrin.h: Ditto.
13561         * config/i386/lzcntintrin.h: Ditto.
13562         * config/i386/smmintrin.h: Ditto.
13563         * config/i386/wmmintrin.h: Ditto.
13564         * config/i386/x86intrin.h: Remove all header include guards.
13565         * config/i386/prfchwintrin.h: Ditto.
13566         * config/i386/pmmintrin.h: Ditto.
13567         * config/i386/tmmintrin.h: Ditto.
13568         * config/i386/xmmintrin.h: Ditto.
13569         * config/i386/popcntintrin.h: Ditto.
13570         * config/i386/rdseedintrin.h: Ditto.
13571         * config/i386/ammintrin.h: Ditto.
13572         * config/i386/emmintrin.h: Ditto.
13573         * config/i386/immintrin.h: Remove all header include guards.
13574         * config/i386/fma4intrin.h: Ditto.
13575         * config/i386/lwpintrin.h: Ditto.
13576         * config/i386/xopintrin.h: Ditto.
13577         * config/i386/ia32intrin.h: Ditto.
13578         * config/i386/avxintrin.h: Ditto.
13579         * config/i386/rtmintrin.h: Ditto.
13580         * config/i386/fmaintrin.h: Ditto.
13581         * config/i386/mm3dnow.h: Ditto.
13582
13583 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13584
13585         * common/config/i386/i386-common.c: Handle LZCNT.
13586
13587 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13588
13589         * doc/extend.texi: Use __atomic_store_n instead of
13590         __atomic_store in HLE example.
13591
13592 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13593
13594         * config/sh/sh.c: Remove <cstdlib> workaround.
13595
13596 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13597
13598         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13599
13600 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13601
13602         * doc/extend.texi: Document that __atomic_clear and
13603           __atomic_test_and_set should only be used with bool.
13604
13605 2013-06-20  Jan Hubicka  <jh@suse.cz>
13606
13607         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13608         types_same_for_odr.
13609         * tree.c (decls_same_for_odr): New function.
13610         (same_for_edr): New function.
13611         (types_same_for_odr): New function.
13612         (get_binfo_at_offset): Use it.
13613         * tree.h (types_same_for_odr): Declare.
13614
13615 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13616             Jason Merrill  <jason@redhat.com>
13617
13618         * system.h: Include <cstdlib> as well as <stdlib.h>.
13619
13620 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13621
13622         PR target/57655
13623         * config/i386/i386.c (construct_container): Report error if
13624         long double is used with disabled x87 float returns.
13625
13626 2013-06-20  Jan Hubicka  <jh@suse.cz>
13627
13628         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13629
13630 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13631
13632         PR rtl-optimization/57425
13633         PR rtl-optimization/57569
13634         * alias.c (write_dependence_p): Remove parameters mem_mode and
13635         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13636         Changed all callers.
13637         (canon_anti_dependence): Get comments and semantics in sync.
13638         Add parameter mem_canonicalized.  Changed all callers.
13639         * rtl.h (canon_anti_dependence): Update prototype.
13640
13641 2013-06-20  Richard Biener  <rguenther@suse.de>
13642
13643         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13644         byte case, inline streamer_read_uchar and defer section
13645         overrun check.
13646
13647 2013-06-20  Richard Biener  <rguenther@suse.de>
13648
13649         PR tree-optimization/57584
13650         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13651         SSA names into the expanded expression that take part in
13652         abnormal coalescing.
13653
13654 2013-06-19  Sharad Singhai  <singhai@google.com>
13655
13656         * gcov.c (print_usage): Handle new option.
13657         (process_args): Ditto.
13658         (get_gcov_intermediate_filename): New function.
13659         (output_intermediate_file): New function.
13660         (output_gcov_file): New function
13661         (generate_results): Handle new option.
13662         (release_function): Relase demangled name.
13663         (read_graph_file): Handle demangled name.
13664         (output_lines): Ditto.
13665         * doc/gcov.texi: Document gcov intermediate format.
13666
13667 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13668
13669         PR bootstrap/57604
13670         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13671         (lra_emit_add): Use the functions.  Add comment about Y as an
13672         address segment.
13673
13674 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13675
13676         PR driver/57652
13677         * collect2.c (collect_atexit): New.
13678         (collect_exit): Delete.
13679         (main): Register collect_atexit with atexit.
13680         (collect_wait): Change collect_exit to exit.
13681         (do_wait): Same.
13682         * collect2.h (collect_exit): Delete.
13683         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13684
13685 2013-06-19  Wei Mi  <wmi@google.com>
13686
13687         PR rtl-optimization/57518
13688         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13689         if regno is used in paradoxical subreg.
13690         (update_equiv_regs): Check pdx_subregs[regno] before
13691         set a reg to be equivalent with a mem.
13692
13693 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13694
13695         PR driver/57651
13696         * file-find.h (find_a_file): Add a mode parameter.
13697         * file-find.c (find_a_file): Likewise.
13698         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13699         with X_OK for the executables.
13700         * collect2.c (main): Call find_a_file with X_OK.
13701
13702 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13703
13704         PR target/56942
13705         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13706         Use NEXT_INSN instead of next_real_insn.
13707
13708 2013-06-19  Jan Hubicka  <jh@suse.cz>
13709
13710         * cgraph.h (const_value_known_p): Replace by ...
13711         (ctor_for_folding): .. this one.
13712         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13713         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13714         * expr.c (expand_expr_real_1): Likewise.
13715         (string_constant): Likewise.
13716         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
13717         * ipa.c (process_references): Likewise.
13718         (symtab_remove_unreachable_nodes): Likewise.
13719         * ipa-inline-analysis.c (param_change_prob): Likewise.
13720         * gimple-fold.c (canonicalize_constructor_val): Likewise.
13721         (get_base_constructor): Likwise.
13722         * varpool.c (varpool_remove_node): Likewise.
13723         (varpool_remove_initializer): LIkewise.
13724         (dump_varpool_node): LIkwise.
13725         (const_value_known_p): Rewrite to ...
13726         (ctor_for_folding): ... this one.
13727
13728 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
13729
13730         PR driver/57651
13731         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
13732         PERSONALITY in $PATH derived prefixes.
13733
13734 2013-06-19  Jeff Law  <law@redhat.com>
13735
13736         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
13737         in comment.
13738
13739         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
13740         (simplify_bitwise_binary): Use it to simpify certain binary ops on
13741         booleans.
13742
13743 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
13744
13745         * config/arm/vfp.md: Move VFP instruction classification documentation
13746         to ...
13747         * config/arm/arm.md: ... here.  Update instruction classification
13748         documentation.
13749
13750 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
13751
13752         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
13753         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
13754         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
13755
13756 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
13757
13758         PR target/57609
13759         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13760         with NEXT_INSN.  Use tablejump_p to check for jump table data
13761         insns.
13762
13763 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
13764
13765         PR c++/56544
13766         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
13767         that now in C++ the value is correct per the C++ standards.
13768
13769 2013-06-19  Richard Biener  <rguenther@suse.de>
13770
13771         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
13772         for global context.
13773
13774 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13775
13776         Revert:
13777         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13778
13779         PR target/57609
13780         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13781         with next_active_insn.
13782
13783 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
13784
13785         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
13786         functions are inlined during failures to flag an error.
13787         * tree-inline.c (expand_call_inline): Allow the error to be flagged
13788         in early inline pass.
13789
13790 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
13791
13792         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
13793         in comments.
13794
13795 2013-06-18  Julian Brown  <julian@codesourcery.com>
13796
13797         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
13798         Permit virtual register pre-reload if !strict.
13799         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
13800         change.
13801         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
13802         prototype.
13803         * config/arm/neon.md (movmisalign<mode>): Use
13804         neon_perm_struct_or_reg_operand instead of
13805         neon_struct_or_register_operand.
13806         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
13807         neon_permissive_struct_operand instead of neon_struct_operand.
13808         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
13809         neon_vector_mem_operand.
13810         * config/arm/predicates.md (neon_struct_operand): Adjust call to
13811         neon_vector_mem_operand.
13812         (neon_permissive_struct_operand): New.
13813         (neon_struct_or_register_operand): Rename to...
13814         (neon_perm_struct_or_reg_operand): This. Adjust call to
13815         neon_vector_mem_operand.
13816
13817 2013-06-18  Richard Biener  <rguenther@suse.de>
13818
13819         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
13820         * lto-streamer.h: Include pointer-set.h.
13821         (struct lto_decl_slot): Remove.
13822         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
13823         Remove next_index entry.
13824         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
13825         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
13826         (lto_init_tree_ref_encoder): Adjust.
13827         (lto_destroy_tree_ref_encoder): Likewise.
13828         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
13829         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
13830         (lto_output_decl_index): Adjust.
13831         (lto_new_out_decl_state): Likewise.
13832         (lto_record_function_out_decl_state): Likewise.
13833         * lto-streamer-out.c (copy_function): Likewise.
13834
13835 2013-06-18  Richard Biener  <rguenther@suse.de>
13836
13837         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
13838         * cgraphunit.c: Include cfgloop.h.
13839         (init_lowered_empty_function): Initialize the loop tree.
13840         (assemble_thunk): Insert new BBs into loops.
13841
13842 2013-06-18  Richard Biener  <rguenther@suse.de>
13843
13844         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
13845         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
13846         the map from cache entry to cache index optional.
13847         (streamer_tree_cache_replace_tree): Adjust accordingly.
13848         (streamer_tree_cache_append): Likewise.
13849         (streamer_tree_cache_delete): Likewise.
13850         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
13851         streamer cache map from cache entry to cache index.
13852         * lto-streamer-out.c (create_output_block): Adjust.
13853
13854 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
13855
13856         * config/arm/arm.md (attribute "insn"): Move multiplication and
13857         division attributes to...
13858         (attribute "type"): ... here.  Remove mult.
13859         (attribute "mul32"): New attribute.
13860         (attribute "mul64"): Add umaal.
13861         (*arm_mulsi3): Update attributes.
13862         (*arm_mulsi3_v6): Likewise.
13863         (*thumb_mulsi3): Likewise.
13864         (*thumb_mulsi3_v6): Likewise.
13865         (*mulsi3_compare0): Likewise.
13866         (*mulsi3_compare0_v6): Likewise.
13867         (*mulsi_compare0_scratch): Likewise.
13868         (*mulsi_compare0_scratch_v6): Likewise.
13869         (*mulsi3addsi): Likewise.
13870         (*mulsi3addsi_v6): Likewise.
13871         (*mulsi3addsi_compare0): Likewise.
13872         (*mulsi3addsi_compare0_v6): Likewise.
13873         (*mulsi3addsi_compare0_scratch): Likewise.
13874         (*mulsi3addsi_compare0_scratch_v6): Likewise.
13875         (*mulsi3subsi): Likewise.
13876         (*mulsidi3adddi): Likewise.
13877         (*mulsi3addsi_v6): Likewise.
13878         (*mulsidi3adddi_v6): Likewise.
13879         (*mulsidi3_nov6): Likewise.
13880         (*mulsidi3_v6): Likewise.
13881         (*umulsidi3_nov6): Likewise.
13882         (*umulsidi3_v6): Likewise.
13883         (*umulsidi3adddi): Likewise.
13884         (*umulsidi3adddi_v6): Likewise.
13885         (*smulsi3_highpart_nov6): Likewise.
13886         (*smulsi3_highpart_v6): Likewise.
13887         (*umulsi3_highpart_nov6): Likewise.
13888         (*umulsi3_highpart_v6): Likewise.
13889         (mulhisi3): Likewise.
13890         (*mulhisi3tb): Likewise.
13891         (*mulhisi3bt): Likewise.
13892         (*mulhisi3tt): Likewise.
13893         (maddhisi4): Likewise.
13894         (*maddhisi4tb): Likewise.
13895         (*maddhisi4tt): Likewise.
13896         (maddhidi4): Likewise.
13897         (*maddhidi4tb): Likewise.
13898         (*maddhidi4tt): Likewise.
13899         (divsi3): Likewise.
13900         (udivsi3): Likewise.
13901         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
13902         (thumb2_mulsi_short_compare0): Likewise.
13903         (thumb2_mulsi_short_compare0_scratch): Likewise.
13904         * config/arm/arm1020e.md (1020mult1): Update attribute change.
13905         (1020mult2): Likewise.
13906         (1020mult3): Likewise.
13907         (1020mult4): Likewise.
13908         (1020mult5): Likewise.
13909         (1020mult6): Likewise.
13910         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
13911         change.
13912         (cortex_a15_mult64): Likewise.
13913         (cortex_a15_sdiv): Likewise.
13914         (cortex_a15_udiv): Likewise.
13915         * config/arm/arm1026ejs.md (mult1): Update attribute change.
13916         (mult2): Likewise.
13917         (mult3): Likewise.
13918         (mult4): Likewise.
13919         (mult5): Likewise.
13920         (mult6): Likewise.
13921         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
13922         (pj4_ir_div): Likewise.
13923         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
13924         (11_mult2): Likewise.
13925         (11_mult3): Likewise.
13926         (11_mult4): Likewise.
13927         (11_mult5): Likewise.
13928         (11_mult6): Likewise.
13929         (11_mult7): Likewise.
13930         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
13931         (cortex_a8_mla): Likewise.
13932         (cortex_a8_mull): Likewise.
13933         (cortex_a8_smulwy): Likewise.
13934         (cortex_a8_smlald): Likewise.
13935         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
13936         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
13937         (cortex_r4_mul_3): Likewise.
13938         (cortex_r4_mla_4): Likewise.
13939         (cortex_r4_mla_3): Likewise.
13940         (cortex_r4_smlald): Likewise.
13941         (cortex_r4_mull): Likewise.
13942         (cortex_r4_sdiv): Likewise.
13943         (cortex_r4_udiv): Likewise.
13944         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
13945         (cortex_a7_idiv): Likewise.
13946         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
13947         (9_mult2): Likewise.
13948         (9_mult3): Likewise.
13949         (9_mult4): Likewise.
13950         (9_mult5): Likewise.
13951         (9_mult6): Likewise.
13952         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
13953         (cortex_a53_sdiv): Likewise.
13954         (cortex_a53_udiv): Likewise.
13955         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
13956         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
13957         (mp626_mult2): Likewise.
13958         (mp626_mult3): Likewise.
13959         (mp626_mult4): Likewise.
13960         * config/arm/fa526.md (526_mult1): Update attribute change.
13961         (526_mult2): Likewise.
13962         * config/arm/arm-generic.md (mult): Update attribute change.
13963         (mult_ldsched_strongarm): Likewise.
13964         (mult_ldsched): Likewise.
13965         (multi_cycle): Likewise.
13966         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
13967         * config/arm/fa606te.md (606te_mult1): Update attribute change.
13968         (606te_mult2): Likewise.
13969         (606te_mult3): Likewise.
13970         (606te_mult4): Likewise.
13971         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
13972         (cortex_a9_mac16): Likewise.
13973         (cortex_a9_multiply): Likewise.
13974         (cortex_a9_mac): Likewise.
13975         (cortex_a9_multiply_long): Likewise.
13976         * config/arm/fa626te.md (626te_mult1): Update attribute change.
13977         (626te_mult2): Likewise.
13978         (626te_mult3): Likewise.
13979         (626te_mult4): Likewise.
13980
13981 2013-06-18  Richard Biener  <rguenther@suse.de>
13982
13983         PR lto/57334
13984         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
13985
13986 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13987
13988         PR target/57609
13989         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13990         with next_active_insn.
13991
13992 2013-06-18  Alan Modra  <amodra@gmail.com>
13993
13994         * config/rs6000/rs6000.h (enum data_align): New.
13995         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
13996         (DATA_ABI_ALIGNMENT): Define.
13997         (CONSTANT_ALIGNMENT): Correct comment.
13998         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
13999         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14000
14001 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14002
14003         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14004         ATTRIBUTE_UNUSED marking.
14005
14006 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14007
14008         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14009         alternative and update.
14010         (aarch64_dup_lanedi): Delete.
14011         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14012         * config/aarch64/aarch64-simd-builtins.def: Update.
14013
14014 2013-06-17  Richard Biener  <rguenther@suse.de>
14015
14016         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14017         (lto_input_scc): Declare.
14018         (lto_input_tree_1): Likewise.
14019         (struct lto_stats_d): Add num_tree_bodies_output and
14020         num_pickle_refs_output.
14021         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14022         (lto_read_tree_1): Split out from ...
14023         (lto_read_tree): ... this.
14024         (lto_input_scc): New function.
14025         (lto_input_tree_1): Split out from ...
14026         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14027         (lto_data_in_create): Create the streamer cache without hashes.
14028         * lto-streamer-out.c (create_output_block): Create the streamer
14029         cache with hashes when not doing WPA.
14030         (lto_write_tree_1): Split out from ...
14031         (lto_write_tree): ... this.
14032         (get_symbol_initial_value): New function.
14033         (lto_output_tree_1): Split out from ...
14034         (lto_output_tree): ... this.  Write trees as series of SCCs
14035         using a DFS walk via DFS_write_tree.
14036         (struct sccs, struct scc_entry): New types.
14037         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14038         (DFS_write_tree_body): New function.
14039         (DFS_write_tree): Likewise.
14040         (hash_tree): Likewise.
14041         (scc_entry_compare): Likewise.
14042         (hash_scc): Likewise.
14043         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14044         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14045         TREE_CHAIN as regular reference.
14046         (streamer_read_integer_cst): Remove.
14047         (streamer_get_pickled_tree): Adjust.
14048         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14049         of DECL_EXTERNALs in BLOCK_VARS for now.
14050         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14051         reference.
14052         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14053         Add hash value argument and record that if hashes are recorded
14054         in the cache.
14055         (streamer_tree_cache_insert_1): Adjust.
14056         (streamer_tree_cache_insert): Likewise.
14057         (streamer_tree_cache_insert_at): Rename to ...
14058         (streamer_tree_cache_replace_tree): ... this and adjust.
14059         (streamer_tree_cache_append): Adjust.
14060         (record_common_node): Likewise.
14061         (streamer_tree_cache_create): Add argument whether to
14062         record hash values together with trees.
14063         (streamer_tree_cache_delete): Adjust.
14064         * tree-streamer.h (struct streamer_tree_cache_d): Add
14065         vector of hashes.
14066         (streamer_read_integer_cst): Remove.
14067         (streamer_tree_cache_insert): Adjust.
14068         (streamer_tree_cache_append): Likewise.
14069         (streamer_tree_cache_insert_at): Rename to ...
14070         (streamer_tree_cache_replace_tree): ... this and adjust.
14071         (streamer_tree_cache_create): Add argument whether to record hashes.
14072         (streamer_tree_cache_get): Rename to ...
14073         (streamer_tree_cache_get_tree): ... this.
14074         (streamer_tree_cache_get_hash): New function.
14075         * tree.c (cache_integer_cst): New function.
14076         * tree.h (cache_integer_cst): Declare.
14077         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14078         * lto-symtab.c (lto_varpool_replace_node): Only release
14079         DECL_INITIAL of non-prevailing decls.
14080         * varpool.c (varpool_remove_initializer): Do not release
14081         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14082
14083 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14084
14085         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14086         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14087         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14088         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14089         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14090         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14091         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14092         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14093         instead of TARGET_64BIT.
14094         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14095         Require ISA_HAS_<D>DIV.
14096
14097 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14098
14099         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14100         (mips*-*-linux*): Move default with_llsc setting to where other
14101         defaults are set.
14102         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14103         with_arch block.
14104         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14105         Likewise.  Remove default with_tune setting.  Move default float
14106         setting to its own block.  Handle with_llsc in the same block as above.
14107
14108 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14109
14110         PR rtl-optimization/57425
14111         PR rtl-optimization/57569
14112         * alias.c (write_dependence_p): Add new parameters mem_mode,
14113         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14114         Changed all callers.
14115         (canon_anti_dependence): New function.
14116         * cse.c (check_dependence): Use canon_anti_dependence.
14117         * cselib.c (cselib_invalidate_mem): Likewise.
14118         * rtl.h (canon_anti_dependence): Declare.
14119
14120 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14121
14122         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14123         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14124         ".set mips3" for 64-bit targets.
14125
14126 2013-06-15  Dehao Chen  <dehao@google.com>
14127
14128         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14129         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14130         (gimple_check_call_args): Likewise.
14131         * value-prof.c (check_ic_target): Likewise.
14132         * ipa-inline.c (early_inliner): Likewise.
14133         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14134         * cgraph.c (cgraph_create_edge_1): Likewise.
14135         (cgraph_make_edge_direct): Likewise.
14136
14137 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14138
14139         PR target/57615
14140         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14141         rs6000_output_move_128bit to handle emitting quad memory
14142         operations.  Set attribute length to 8 bytes.
14143
14144 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14145
14146         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14147         New pattern.
14148         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14149         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14150         (aarch64_<su>mlsl<mode>): Likewise.
14151
14152 2013-06-14  Mike Stump  <mikestump@comcast.net>
14153
14154         * Makefile.in (TARGET_H): Add insn-codes.h.
14155
14156 2013-06-14  Alan Modra  <amodra@gmail.com>
14157
14158         PR middle-end/57134
14159         PR middle-end/57586
14160         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14161         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14162         bitfield expansion when EXPAND_MEMORY.
14163         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14164
14165 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14166
14167         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14168         test for clearing quad memory on 32-bit later.
14169
14170 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14171
14172         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14173         (fold_negate_expr): Likewise.
14174         (fold_real_zero_addition_p): Handle vectors.
14175         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14176
14177 2013-06-14  Alan Modra  <amodra@gmail.com>
14178
14179         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14180
14181 2013-06-13  Jan Hubicka  <jh@suse.cz>
14182
14183         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14184         Local comdats are not externally visible.
14185         * symtab.c (dump_symtab_base): Dump externally visible.
14186         (verify_symtab_base): Verify back links in the symtab hash.
14187
14188 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14189
14190         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14191         CONVERT_EXPR as equal nodes.
14192
14193 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14194
14195         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14196
14197 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14198
14199         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14200         Generalize to complex and vector.
14201         * tree.c (build_all_ones_cst): New function.
14202         * tree.h (build_all_ones_cst): Declare it.
14203
14204 2013-06-13  Alan Modra  <amodra@gmail.com>
14205
14206         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14207         * config/rs6000/rs6000.md (signbittf2): New insn.
14208         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14209         (abstf2_internal, cmptf_internal2): Likewise.
14210         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14211
14212 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14213             Pat Haugen  <pthaugen@us.ibm.com>
14214             Peter Bergner  <bergner@vnet.ibm.com>
14215
14216         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14217         power8 byte, half-word, and quad-word atomic instructions.
14218         (emit_store_conditional): Likewise.
14219         (rs6000_expand_atomic_compare_and_swap): Likewise.
14220         (rs6000_expand_atomic_op): Likewise.
14221
14222         * config/rs6000/sync.md (larx): Add new modes for power8.
14223         (stcx): Likewise.
14224         (AINT): New mode iterator to include TImode as well as normal
14225         integer modes on power8.
14226         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14227         that VSX registers are not considered.  Use AINT mode iterator
14228         instead of INT1 to allow inclusion of quad word atomic operations
14229         on power8.
14230         (load_locked<mode>): Likewise.
14231         (store_conditional<mode>): Likewise.
14232         (atomic_compare_and_swap<mode>): Likewise.
14233         (atomic_exchange<mode>): Likewise.
14234         (atomic_nand<mode>): Likewise.
14235         (atomic_fetch_<fetchop_name><mode>): Likewise.
14236         (atomic_nand_fetch<mode>): Likewise.
14237         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14238         each type.
14239         (ATOMIC): On power8, add QImode, HImode modes.
14240         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14241         modes that promote to SImode.
14242         (load_lockedti): Convert TImode arguments to PTImode, so that we
14243         get a guaranteed even/odd register pair.
14244         (load_lockedpti): Likewise.
14245         (store_conditionalti): Likewise.
14246         (store_conditionalpti): Likewise.
14247
14248         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14249         atomic load/store instructions.
14250         (HSI): Likewise.
14251
14252 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14253
14254         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14255         loads.
14256         (insn_count): New attribute, with most cases extracted from...
14257         (length): ...here.  Redefine most cases in terms of insn_count.
14258         (single_insn): Delete.
14259         (can_delay): Use insn_count to check for single instructions.
14260         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14261         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14262         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14263         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14264         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14265         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14266         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14267         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14268         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14269         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14270         rather than "length".
14271         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14272         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14273         Use "insn_count" rather than "length".
14274         * config/mips/mips-dsp.md
14275         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14276         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14277         length attributes.
14278
14279 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14280
14281         PR tree-optimization/57361
14282         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14283
14284 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14285
14286         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14287         to split.
14288         (aarch64_simd_combine<mode>): New instruction expansion.
14289         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14290         function prototype.
14291         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14292         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14293
14294 2013-06-12  Jan Hubicka  <jh@suse.cz>
14295
14296         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14297         decl has when in streaming stage.
14298         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14299         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14300
14301 2013-06-12  Roland Stigge  <stigge@antcom.de>
14302
14303         PR target/57578
14304         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14305
14306 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14307
14308         PR tree-optimization/57537
14309         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14310         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14311
14312 2013-06-12  Richard Biener  <rguenther@suse.de>
14313
14314         * data-streamer.h (streamer_write_char_stream): CSE
14315         obs->current_pointer.
14316         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14317         streamer_write_char_stream manually and optimize the resulting loop.
14318         (streamer_write_hwi_stream): Likewise.
14319
14320 2013-06-12  Jan Hubicka  <jh@suse.cz>
14321
14322         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14323         * cgraph.h (varpool_create_empty_node): Declare.
14324         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14325         duplicated nodes.
14326         * symtab.c (symtab_unregister_node): Be lax about missin entries
14327         in node hash.
14328         (symtab_get_node): Update comment.
14329         * varpool.c (varpool_create_empty_node): Break out from ...
14330         (varpool_node_for_decl): ... here.
14331         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14332
14333 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14334
14335         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14336         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14337         part.  Use straight-line flow at the end.
14338         <COMPONENT_REF>: Remove superfluous else.
14339         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14340
14341 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14342
14343         PR target/56564
14344         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14345         target hook even for !TREE_PUBLIC decls.  If no resolution info
14346         is available, return false for common and external decls.
14347
14348 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14349
14350         * config/rl78/constraints.md (U): New constraint.
14351         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14352         valloc attribute.
14353
14354 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14355
14356         PR target/57589
14357         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14358         to allow returning address to AT_PLATFORM name.
14359
14360 2013-06-11  Jan Hubicka  <jh@suse.cz>
14361
14362         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14363         * cgraph.h (symtab_node_base): Add weakref flag.
14364         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14365         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14366         (output_weakrefs): Use weakref flag.
14367         * fold-const.c (simple_operand_p): Handle WEAK.
14368         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14369         * ipa.c (varpool_externally_visible_p): Drop weakref.
14370         (function_and_variable_visibility): Update comment; fix weakref
14371         sanity checks; do not clear DECL_WEAK on them.
14372         * lto-cgraph.c (lto_output_node): update.
14373         (lto_output_varpool_node): Update.
14374         (input_overwrite_node): Update.
14375         (input_node): Update.
14376         (input_varpool_node): Update.
14377         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14378         (lto_symtab_merge_symbols): Add sanity check.
14379         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14380         * passes.c (rest_of_decl_compilation): Set static flag, too.
14381         * symtab.c (dump_symtab_base): Dump weakref.
14382         (verify_symtab_base): Sanity check weakrefs.
14383         (symtab_make_decl_local): Remove duplicated code.
14384         (symtab_alias_ultimate_target): Simplify.
14385         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14386
14387 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14388
14389         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14390         sequence_vect == NULL.
14391
14392 2013-06-11  DJ Delorie  <dj@redhat.com>
14393
14394         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14395         (rl78_unwind_word_mode): New.
14396
14397 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14398
14399         * final.c (debug_prefix_maps): Make static.
14400
14401 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14402
14403         * function.c (initial_trampoline): Remove stray copy.
14404
14405 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14406
14407         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14408
14409 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14410
14411         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14412         within bounds at the beginning of the function.
14413
14414 2013-06-11  Alan Modra  <amodra@gmail.com>
14415
14416         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14417         reporting.
14418         (get_named_section): Don't NULL !DECL_P decl.
14419
14420 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14421
14422         * doc/invoke.texi (core-avx2): Document.
14423         (slm): Likewise.
14424         (atom): Updated with MOVBE.
14425
14426 2013-06-11  Richard Biener  <rguenther@suse.de>
14427
14428         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14429
14430 2013-06-11  Anton Blanchard  <anton@samba.org>
14431
14432         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14433         correct shift value in little-endian mode.
14434
14435 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14436
14437         PR target/56564
14438         * varasm.c (get_variable_align): Move #endif to the right place.
14439
14440 2013-06-10  Cary Coutant  <ccoutant@google.com>
14441
14442         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14443         for hash so that hash table traversal order is deterministic.
14444
14445 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14446             Pat Haugen  <pthaugen@us.ibm.com>
14447             Peter Bergner  <bergner@vnet.ibm.com>
14448
14449         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14450         of vectors in GPRS if they are direct moves or quad word load or
14451         store moves.
14452
14453         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14454         declaration.
14455         (direct_move_p): Likewise.
14456         (quad_load_store_p): Likewise.
14457
14458         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14459         classes into bins based on the physical register type.
14460         (reg_class_to_reg_type): Likewise.
14461         (IS_STD_REG_TYPE): Likewise.
14462         (IS_FP_VECT_REG_TYPE): Likewise.
14463         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14464         use direct move instructions.
14465         (reload_gpr_vsx): Likewise.
14466         (reload_vsx_gpr): Likewise.
14467         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14468         information that is a simplification of register classes.  Also
14469         precalculate direct move reload helpers.
14470         (direct_move_p): New function to return true if the operation can
14471         be done as a direct move instruciton.
14472         (quad_load_store_p): New function to return true if the operation
14473         is a quad memory operation.
14474         (rs6000_legitimize_address): If quad memory, only allow register
14475         indirect for TImode addresses.
14476         (rs6000_legitimate_address_p): Likewise.
14477         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14478         (rs6000_reload_register_type): Likewise.
14479         (register_to_reg_type): Return register type.
14480         (rs6000_secondary_reload_simple_move): New helper function for
14481         secondary reload and secondary memory needed to identify anything
14482         that is a simple move, and does not need reloading.
14483         (rs6000_secondary_reload_direct_move): New helper function for
14484         secondary reload to identify cases that can be done with several
14485         instructions via the direct move instructions.
14486         (rs6000_secondary_reload_move): New helper function for secondary
14487         reload to identify moves between register types that can be done.
14488         (rs6000_secondary_reload): Add support for quad memory operations
14489         and for direct move.
14490         (rs6000_secondary_memory_needed): Likewise.
14491         (rs6000_debug_secondary_memory_needed): Change argument names.
14492         (rs6000_output_move_128bit): New function to return the move to
14493         use for 128-bit moves, including knowing about the various
14494         limitations of quad memory operations.
14495
14496         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14497         memory operations.  call rs6000_output_move_128bit for the actual
14498         instruciton(s) to generate.
14499         (vsx_movti_64bit): Likewise.
14500
14501         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14502         (UNSPEC_P8V_MTVSRWZ): Likewise.
14503         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14504         (UNSPEC_P8V_MTVSRD): Likewise.
14505         (UNSPEC_P8V_XXPERMDI): Likewise.
14506         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14507         (UNSPEC_FUSION_GPR): Likewise.
14508         (FMOVE128_GPR): New iterator for direct move.
14509         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14510         (f32_sv): Likewise.
14511         (f32_dm): Likewise.
14512         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14513         loads and direct move instructions.
14514         (zero_extendsidi2_lfiwzx): Likewise.
14515         (extendsidi2_lfiwax): Likewise.
14516         (extendsidi2_nocell): Likewise.
14517         (floatsi<mode>2_lfiwax): Likewise.
14518         (lfiwax): Likewise.
14519         (floatunssi<mode>2_lfiwzx): Likewise.
14520         (lfiwzx): Likewise.
14521         (fix_trunc<mode>_stfiwx): Likewise.
14522         (fixuns_trunc<mode>_stfiwx): Likewise.
14523         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14524         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14525         (parity<mode>2_cmpb): Set length/type attr.
14526         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14527         for 'mr.' to fast_compare.
14528         (bpermd_<mode>): Change type attr to popcnt.
14529         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14530         (p8_mtvsrwz_1): Likewise.
14531         (p8_mtvsrwz_2): Likewise.
14532         (reload_fpr_from_gpr<mode>): Likewise.
14533         (p8_mtvsrd_1): Likewise.
14534         (p8_mtvsrd_2): Likewise.
14535         (p8_xxpermdi_<mode>): Likewise.
14536         (reload_vsx_from_gpr<mode>): Likewise.
14537         (reload_vsx_from_gprsf): Likewise.
14538         (p8_mfvsrd_3_<mode>): LIkewise.
14539         (reload_gpr_from_vsx<mode>): Likewise.
14540         (reload_gpr_from_vsxsf): Likewise.
14541         (p8_mfvsrd_4_disf): Likewise.
14542         (multi-word GPR splits): Do not split direct moves or quad memory
14543         operations.
14544
14545 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14546
14547         * tree-into-ssa.c (interesting_blocks): Make static.
14548
14549 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14550
14551         PR target/56564
14552         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14553         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14554         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14555         (get_variable_align): New function.
14556         (get_variable_section, emit_bss, emit_common,
14557         assemble_variable_contents, place_block_symbol): Use
14558         get_variable_align instead of DECL_ALIGN.
14559         (assemble_noswitch_variable): Add align argument, use it
14560         instead of DECL_ALIGN.
14561         (assemble_variable): Adjust caller.  Use get_variable_align
14562         instead of DECL_ALIGN.
14563         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14564         caller.
14565         (DATA_ABI_ALIGNMENT): Define.
14566         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14567         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14568         opt is false, only return the psABI mandated alignment increase.
14569         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14570         (DATA_ABI_ALIGNMENT): ... this.
14571         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14572         (DATA_ABI_ALIGNMENT): ... this.
14573         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14574         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14575         (DATA_ABI_ALIGNMENT): ... this.
14576         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14577         * doc/tm.texi: Regenerated.
14578
14579 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14580
14581         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14582         cmp_code to construct REG_EQUAL note.
14583
14584 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14585
14586         PR target/57568
14587         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14588         that operands[2] doesn't overlap with operands[0].
14589
14590 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14591             Jan Hubicka  <jh@suse.cz>
14592
14593         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14594         hack to mark symbols as used.
14595
14596 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14597
14598         PR rtl-optimization/57559
14599         * lra-constraints.c (process_alt_operands): Don't discourage
14600         memory with known offset for offsetable memory constraint.
14601         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14602
14603 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14604
14605         * varasm.c (struct oc_local_state): Reorder fields.
14606         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14607         and adjust accordingly.
14608         (output_constructor): Reorder initialization code and adjust call to
14609         output_constructor_bitfield.
14610
14611 2013-06-07  Jan Hubicka  <jh@suse.cz>
14612
14613         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14614
14615 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14616
14617         * tree-object-size.c (unknown): Make const.
14618
14619 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14620
14621         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14622         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14623         for last alternative in the cpu_facility attribute.
14624
14625 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14626
14627         PR target/56315
14628         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14629         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14630         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14631         * config/arm/constraints.md (Dg): New constraint.
14632         * config/arm/neon.md (xordi3_neon): Remove.
14633         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14634         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14635
14636 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14637
14638         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14639         Clean up alternatives.
14640
14641 2013-06-07  Alan Modra  <amodra@gmail.com>
14642
14643         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14644         va_list_gpr_size.
14645
14646 2013-06-07  Alan Modra  <amodra@gmail.com>
14647
14648         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14649
14650 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14651
14652         * config/arm/constraints.md (Df): New constraint.
14653         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14654         Correct length attribute for last two alternatives.
14655
14656 2013-06-07  Alan Modra  <amodra@gmail.com>
14657
14658         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14659         override user -mfp-in-toc.
14660         (offsettable_ok_by_alignment): Consider just the current access
14661         rather than the whole object, unless BLKmode.  Handle
14662         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14663         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14664         for -mcmodel=medium.
14665         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14666         override user -mfp-in-toc or -msum-in-toc.  Default to
14667         -mno-fp-in-toc for -mcmodel=medium.
14668
14669 2013-06-06  DJ Delorie  <dj@redhat.com>
14670
14671         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14672         TARGET_VALID_POINTER_MODE.
14673
14674 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14675             Pat Haugen  <pthaugen@us.ibm.com>
14676             Peter Bergner  <bergner@vnet.ibm.com>
14677
14678         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14679         Document new power8 builtins.
14680
14681         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14682         condition code register, to allow 128-bit logical operations to be
14683         done in the VSX or GPR registers.
14684         (nor<mode>3): Use the canonical form for nor.
14685         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14686         vclz*, and vpopcnt* vector instructions.
14687         (nand<mode>3): Likewise.
14688         (orc<mode>3): Likewise.
14689         (clz<mode>2): LIkewise.
14690         (popcount<mode>2): Likewise.
14691
14692         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14693         that only the GPRs are recognized.
14694
14695         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14696         support for new power8 builtins.
14697
14698         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14699         builtin functions.
14700         (xscvdpspn): Likewise.
14701         (vclz): Likewise.
14702         (vclzb): Likewise.
14703         (vclzh): Likewise.
14704         (vclzw): Likewise.
14705         (vclzd): Likewise.
14706         (vpopcnt): Likewise.
14707         (vpopcntb): Likewise.
14708         (vpopcnth): Likewise.
14709         (vpopcntw): Likewise.
14710         (vpopcntd): Likewise.
14711         (vgbbd): Likewise.
14712         (vmrgew): Likewise.
14713         (vmrgow): Likewise.
14714         (eqv): Likewise.
14715         (eqv_v16qi3): Likewise.
14716         (eqv_v8hi3): Likewise.
14717         (eqv_v4si3): Likewise.
14718         (eqv_v2di3): Likewise.
14719         (eqv_v4sf3): Likewise.
14720         (eqv_v2df3): Likewise.
14721         (nand): Likewise.
14722         (nand_v16qi3): Likewise.
14723         (nand_v8hi3): Likewise.
14724         (nand_v4si3): Likewise.
14725         (nand_v2di3): Likewise.
14726         (nand_v4sf3): Likewise.
14727         (nand_v2df3): Likewise.
14728         (orc): Likewise.
14729         (orc_v16qi3): Likewise.
14730         (orc_v8hi3): Likewise.
14731         (orc_v4si3): Likewise.
14732         (orc_v2di3): Likewise.
14733         (orc_v4sf3): Likewise.
14734         (orc_v2df3): Likewise.
14735
14736         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
14737         allow power8 quad mode in 64-bit.
14738         (rs6000_builtin_vectorized_function): Add support to vectorize
14739         ISA 2.07 count leading zeros, population count builtins.
14740         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
14741         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
14742         (builtin_function_type): Add vgbbd builtin function which takes an
14743         unsigned argument.
14744         (altivec_expand_vec_perm_const): Add support for new power8 merge
14745         instructions.
14746
14747         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
14748         that does not include TImdoe for use with 32-bit.
14749         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
14750         instructions.
14751         (UNSPEC_VSX_CVDPSPN): Likewise.
14752         (vsx_xscvdpspn): Likewise.
14753         (vsx_xscvspdpn): Likewise.
14754         (vsx_xscvdpspn_scalar): Likewise.
14755         (vsx_xscvspdpn_directmove): Likewise.
14756         (vsx_and<mode>3): Split logical operations into 32-bit and
14757         64-bit. Add support to do logical operations on TImode as well as
14758         VSX vector types.  Allow logical operations to be done in either
14759         VSX registers or in general purpose registers in 64-bit mode.  Add
14760         splitters if GPRs were used. For AND, add clobber of CCmode to
14761         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
14762         encoding.
14763         (vsx_and<mode>3_32bit): Likewise.
14764         (vsx_and<mode>3_64bit): Likewise.
14765         (vsx_ior<mode>3): Likewise.
14766         (vsx_ior<mode>3_32bit): Likewise.
14767         (vsx_ior<mode>3_64bit): Likewise.
14768         (vsx_xor<mode>3): Likewise.
14769         (vsx_xor<mode>3_32bit): Likewise.
14770         (vsx_xor<mode>3_64bit): Likewise.
14771         (vsx_one_cmpl<mode>2): Likewise.
14772         (vsx_one_cmpl<mode>2_32bit): Likewise.
14773         (vsx_one_cmpl<mode>2_64bit): Likewise.
14774         (vsx_nor<mode>3): Likewise.
14775         (vsx_nor<mode>3_32bit): Likewise.
14776         (vsx_nor<mode>3_64bit): Likewise.
14777         (vsx_andc<mode>3): Likewise.
14778         (vsx_andc<mode>3_32bit): Likewise.
14779         (vsx_andc<mode>3_64bit): Likewise.
14780         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
14781         and xxlorc instructions.
14782         (vsx_eqv<mode>3_64bit): Likewise.
14783         (vsx_nand<mode>3_32bit): Likewise.
14784         (vsx_nand<mode>3_64bit): Likewise.
14785         (vsx_orc<mode>3_32bit): Likewise.
14786         (vsx_orc<mode>3_64bit): Likewise.
14787
14788         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
14789
14790         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
14791         instruction.
14792         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
14793         (p8_vmrgow): Likewise.
14794         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
14795         GPRs to be split under VSX.
14796         (p8v_clz<mode>2): Add power8 count leading zero support.
14797         (p8v_popcount<mode>2): Add power8 population count support.
14798         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
14799         support.
14800
14801         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
14802         instruction.
14803
14804         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
14805         builtin functions.
14806         (vec_nand): Likewise.
14807         (vec_vclz): Likewise.
14808         (vec_vclzb): Likewise.
14809         (vec_vclzd): Likewise.
14810         (vec_vclzh): Likewise.
14811         (vec_vclzw): Likewise.
14812         (vec_vgbbd): Likewise.
14813         (vec_vmrgew): Likewise.
14814         (vec_vmrgow): Likewise.
14815         (vec_vpopcnt): Likewise.
14816         (vec_vpopcntb): Likewise.
14817         (vec_vpopcntd): Likewise.
14818         (vec_vpopcnth): Likewise.
14819         (vec_vpopcntw): Likewise.
14820
14821 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14822
14823         PR rtl-optimization/57468
14824         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
14825         spilled pseudos.
14826
14827 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14828
14829         PR rtl-optimization/57459
14830         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
14831         type when setting live regs.
14832
14833 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14834
14835         * config/s390/s390.opt (mlra): New option.
14836         * config/s390/s390.c (s390_decompose_address): Check displacement
14837         for all registers for LRA.
14838         (s390_secondary_reload): Don't used secondary reloads for LRA.
14839         (s390_lra_p): New function.
14840         (TARGET_LRA_P): Define.
14841         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
14842         of attribute cpu_facility to zarch for the last alternative.
14843         (*cmpmem_short): Ditto.
14844
14845 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
14846
14847         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
14848         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
14849         (arm_expand_prologue): Likewise.
14850
14851 2013-06-06  Teresa Johnson  <tejohnson@google.com>
14852
14853         PR c++/53743
14854         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
14855         as this is now done by redirect_edge_and_branch_force.
14856         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
14857         barriers, and fix interaction with splitting.
14858         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
14859         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
14860         reflect changes made in the routine.
14861         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
14862         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
14863         since this is called in cfglayout mode, and replace partition fixup
14864         with assert as that is now done by force_nonfallthru_and_redirect.
14865         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
14866         already be marked with region crossing note.
14867         (insert_section_boundary_note): Make non-static, gate on flag
14868         has_bb_partition, rewrite to also check for multiple partitions.
14869         (rest_of_handle_reorder_blocks): Remove call to
14870         insert_section_boundary_note, now done later during free_cfg.
14871         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
14872         * bb-reorder.h (insert_section_boundary_note): Declare.
14873         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
14874         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
14875         invoke insert_section_boundary_note.
14876         (try_redirect_by_replacing_jump): Remove unnecessary
14877         check for region crossing note.
14878         (fixup_partition_crossing): New function.
14879         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
14880         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
14881         in non-cfglayout mode.
14882         (force_nonfallthru_and_redirect): Fixup partition boundaries,
14883         remove old code that tried to do this. Emit barrier correctly
14884         when we are in cfglayout mode.
14885         (last_bb_in_partition): New function.
14886         (rtl_split_edge): Correctly fixup partition boundaries.
14887         (commit_one_edge_insertion): Remove old code that tried to
14888         fixup region crossing edge since this is now handled in
14889         split_block, and set up insertion point correctly since
14890         block may now end in a jump.
14891         (verify_hot_cold_block_grouping): Guard against checking when not in
14892         linearized RTL mode.
14893         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
14894         notes.
14895         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
14896         rtl_verify_flow_info, so not called in cfglayout mode.
14897         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
14898         (fixup_reorder_chain): Remove old code that attempted to fixup region
14899         crossing note as this is now handled in force_nonfallthru_and_redirect.
14900         (duplicate_insn_chain): Don't duplicate switch section notes.
14901         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
14902         note.
14903         * basic-block.h (emit_barrier_after_bb): Declare.
14904
14905 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14906
14907         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
14908         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
14909         arm_usatsihi): Adjust alternatives for arm_restrict_it.
14910
14911 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14912
14913         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
14914         where appropriate.
14915         * config/arm/ldmstm.md: Regenerate.
14916
14917 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14918
14919         * config/arm/sync.md (atomic_loaddi_1):
14920         Disable predication for arm_restrict_it.
14921         (arm_load_exclusive<mode>): Likewise.
14922         (arm_load_exclusivesi): Likewise.
14923         (arm_load_exclusivedi): Likewise.
14924         (arm_load_acquire_exclusive<mode>): Likewise.
14925         (arm_load_acquire_exclusivesi): Likewise.
14926         (arm_load_acquire_exclusivedi): Likewise.
14927         (arm_store_exclusive<mode>): Likewise.
14928         (arm_store_exclusive<mode>): Likewise.
14929         (arm_store_release_exclusivedi): Likewise.
14930         (arm_store_release_exclusive<mode>): Likewise.
14931
14932 2013-06-06  Richard Biener  <rguenther@suse.de>
14933
14934         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
14935         after LTO_null.
14936         (lto_tag_is_tree_code_p): Adjust.
14937         (lto_tag_is_gimple_code_p): Likewise.
14938         (lto_gimple_code_to_tag): Likewise.
14939         (lto_tag_to_gimple_code): Likewise.
14940         (lto_tree_code_to_tag): Likewise.
14941         (lto_tag_to_tree_code): Likewise.
14942         * data-streamer.h (streamer_write_hwi_in_range): Use
14943         uhwi streaming to stream the normalized range.
14944         (streamer_read_hwi_in_range): Likewise.
14945
14946 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14947
14948         * config/arm/arm.md (enabled_for_depr_it): New attribute.
14949         (predicable_short_it): Likewise.
14950         (predicated): Likewise.
14951         (enabled): Handle above.
14952         (define_cond_exec): Set predicated attribute to yes.
14953
14954 2013-06-05  Mike Stump  <mikestump@comcast.net>
14955
14956         * gdbinit.in (__FUNCTION__): Add.
14957
14958 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
14959
14960         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
14961         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
14962
14963 2013-06-05  Jan Hubicka  <jh@suse.cz>
14964
14965         * varasm.c (mark_decl_referenced): Revert the removal until targets
14966         are fixed.
14967
14968 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
14969
14970         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
14971         instead of mark_decl_referenced.
14972
14973 2013-06-05  Jan Hubicka  <jh@suse.cz>
14974
14975         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
14976         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
14977         and symtab_used_from_object_file_p.
14978         (cgraph_make_node_local_1): Clear forced_by_abi.
14979         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
14980         * cgraph.h (symtab_node_base): Add forced_by_abi.
14981         (decide_is_variable_needed): Remove.
14982         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
14983         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
14984         (decide_is_symbol_needed): ... this one; handle symbols in general;
14985         always analyze virtuals; honnor forced_by_abi.
14986         (cgraph_finalize_function): Update.
14987         (varpool_finalize_decl): Update.
14988         (symbol_defined_and_needed): Remove.
14989         (analyze_functions): Update.
14990         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
14991         output_refs, input_overwrite_node): Handle forced_by_abi.
14992         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
14993         (address_taken_from_non_vtable_p): ... this one.
14994         (comdat_can_be_unshared_p_1): New function.
14995         (cgraph_comdat_can_be_unshared_p): Rename to ...
14996         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
14997         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
14998         (function_and_variable_visibility): Clear forced_by_abi as needed.
14999         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15000         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15001         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15002         * varpool.c (decide_is_variable_needed): Remove.
15003
15004 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15005
15006         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15007         (arm_option_override): Override arm_restrict_it where appropriate.
15008         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15009         * config/arm/arm.opt (mrestrict-it): New command-line option.
15010         * doc/invoke.texi: Document -mrestrict-it.
15011
15012 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15013
15014         * tsan.c (tsan_atomic_table): Make const.
15015
15016 2013-06-05  Richard Biener  <rguenther@suse.de>
15017
15018         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15019         index associated with the tree we are supposed to replace.
15020         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15021         TREE_ASM_WRITTEN as zero for everything but SSA names.
15022
15023 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15024
15025         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15026
15027 2013-06-04  Jan Hubicka  <jh@suse.cz>
15028
15029         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15030         (input_node, input_varpool_node): Handle correctly external same
15031         body aliases.
15032         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15033         nodes at ltrans stage.
15034
15035 2013-06-04  Jan Hubicka  <jh@suse.cz>
15036
15037         * ipa-inline.c (update_caller_keys): Fix availability test.
15038         (update_callee_keys): Likewise.
15039         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15040         to follow ELF standard.
15041
15042 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15043
15044         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15045         (mips64r5900el-*-elf*): New configurations.
15046         * config/mips/mips-cpus.def (r5900): New processor.
15047         * config/mips/mips-tables.opt: Regenerate.
15048         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15049         (mips_issue_rate): Handle PROCESSOR_R5900.
15050         (mips_reorg_process_insns): Force reorder mode for the R5900.
15051         * config/mips/mips.h (TARGET_MIPS5900): Define.
15052         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15053         TARGET_MIPS5900.
15054         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15055         TARGET_MIPS5900.
15056         * config/mips/mips.md (processor): Add r5900.
15057         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15058
15059 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15060
15061         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15062         into function to generate MOVI instruction.
15063         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15064         (aarch64_preferred_simd_mode): Turn into wrapper.
15065         (aarch64_output_scalar_simd_mov_immediate): New function.
15066         * config/aarch64/aarch64-protos.h: Add prototype for above.
15067
15068 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15069
15070         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15071         element_char member.
15072         (sizetochar): Return signed char.
15073         (aarch64_simd_valid_immediate): Remove elchar and other
15074         unnecessary variables.
15075         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15076         Calculate element_char as required.
15077         * config/aarch64/aarch64-protos.h: Update and move prototype
15078         for aarch64_output_simd_mov_immediate.
15079         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15080         Update arguments.
15081
15082 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15083
15084         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15085         information completed by aarch64_simd_valid_immediate.
15086         (aarch64_legitimate_constant_p): Update arguments.
15087         (aarch64_simd_valid_immediate): Work with struct rather than many
15088         pointers.
15089         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15090         (aarch64_simd_make_constant): Update arguments.
15091         (aarch64_output_simd_mov_immediate): Work with struct rather than
15092         many pointers.  Output immediate directly rather than as operand.
15093         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15094         Update prototype.
15095         * config/aarch64/constraints.md (Dn): Update arguments.
15096
15097 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15098
15099         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15100         longer static.
15101         (aarch64_simd_immediate_valid_for_move): Remove.
15102         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15103         (aarch64_simd_make_constant): Update call.
15104         (aarch64_output_simd_mov_immediate): Update call.
15105         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15106         Add prototype.
15107         * config/aarch64/constraints.md (Dn): Update call.
15108
15109 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15110
15111         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15112         return type to bool for prototype.
15113         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15114         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15115         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15116
15117 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15118
15119         * config/mips/mips.opt (meva): New.
15120         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15121         (ASM_SPEC): Handle -meva.
15122         * doc/invoke.texi (meva):  Document.
15123
15124 2013-06-04  Alan Modra  <amodra@gmail.com>
15125
15126         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15127         constant output.
15128
15129 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15130
15131         * rtl.def: Add extra fourth optional field to define_cond_exec.
15132         * gensupport.c (process_one_cond_exec): Process attributes from
15133         define_cond_exec.
15134         * doc/md.texi: Document fourth field in define_cond_exec.
15135
15136 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15137
15138         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15139         out the processing order as in store_bit_field_1.
15140
15141 2013-06-04  Jan Hubicka  <jh@suse.cz>
15142
15143         PR middle-end/57500
15144         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15145         non-VAR_DECL node if it does not exist yet.
15146
15147 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15148
15149         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15150         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15151         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15152         target_cpu_default setting.
15153
15154 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15155
15156         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15157         default to -fopt-info=optimized instead of all.
15158         * doc/invoke.texi: Ditto.
15159         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15160         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15161         (execute_vect_slp): Emit BB vectorization success under
15162         MSG_OPTIMIZED_LOCATIONS.
15163         * tree-vect-slp.c (vect_slp_transform_bb): Change
15164         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15165         * tree-vect-loop.c (vect_transform_loop): Ditto.
15166
15167 2013-06-03  Jason Merrill  <jason@redhat.com>
15168
15169         PR c++/57415
15170         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15171         Use TARGET_EXPR for C++.
15172
15173 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15174
15175         PR rtl-optimization/57268
15176         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15177         if DEBUG_INSN_P (insn).
15178
15179         Reapply
15180         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15181
15182         PR rtl-optimization/57268
15183         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15184         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15185
15186 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15187
15188         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15189         (ix86_avoid_lea_for_addr): Likewise.
15190         (exact_dependency_1): Likewise.
15191         (ix86_adjust_cost): Likewise.
15192         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15193         removed.
15194         (do_reorder_for_imul): Fix typo, formatting and
15195         !reload_completed check removed.
15196         (ix86_sched_reorder): Fix typo and formatting.
15197         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15198         list.
15199
15200 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15201
15202         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15203
15204 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15205
15206         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15207         <STRING_CST>: Likewise.
15208         <VECTOR_CST>: Likewise.
15209
15210 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15211             Mikael Morin  <mikael@gcc.gnu.org>
15212
15213         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15214         * config.in: Regenerated.
15215         * configure: Regenerated.
15216
15217 2013-06-01  Jan Hubicka  <jh@suse.cz>
15218
15219         PR middle-end/57366
15220         * cgraphunit.c (compile): When weakref is not supported,
15221         set up transparent aliases before final output pass.
15222         * varasm.c (assemble_alias): Do not try to do it here.
15223
15224 2013-06-01  Jan Hubicka  <jh@suse.cz>
15225
15226         PR middle-end/57467
15227         * passes.c (for_per_function): Skip unanalyzed functions.
15228
15229 2013-06-01  Jan Hubicka  <jh@suse.cz>
15230
15231         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15232         (lto_symtab_merge_symbols_1): ... this one.
15233         (lto_symtab_merge_cgraph_nodes): Rename to ...
15234         (lto_symtab_merge_symbols): ... this one; simplify.
15235         * cgraph.c (same_body_aliases_done): Rename to ...
15236         (cpp_implicit_aliases_done): ... this one.
15237         (cgraph_create_function_alias): Update.
15238         (cgraph_same_body_alias): Update.
15239         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15240         (verify_edge_corresponds_to_fndecl): Simplify.
15241         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15242         (cgraph_node): Remove same_body_alias.
15243         (varpool_node): Remove alias_of and extra_name_alias.
15244         (same_body_aliases_done): Rename to ..
15245         (cpp_implicit_aliases_done): ... this one.
15246         (symtab_alias_ultimate_target): Add default parameter.
15247         (symtab_resolve_alias): New function.
15248         (fixup_same_cpp_alias_visibility): Declare.
15249         (cgraph_function_node): Add default parameter.
15250         (cgraph_node_asm_name): Likewise.
15251         (cgraph_function_or_thunk_node): Add default parameter; do
15252         not ICE when it is NULL.
15253         (varpool_variable_node): Likewise.
15254         * tree-emutls.c (create_emultls_var): Update.
15255         (ipa_lower_emutls): Update.
15256         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15257         (cgraph_reset_node): Reset alias info.
15258         (cgraph_finalize_function): Update.
15259         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15260         (analyze_function): Simplify.
15261         (cgraph_process_same_body_aliases): Simplify.
15262         (analyze_functions): Fixup same body aliases.
15263         (handle_alias_pairs): Simplify.
15264         (assemble_thunk): Update.
15265         (assemble_thunks_and_aliases): Update.
15266         (output_weakrefs): Rewrite.
15267         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15268         (lto_output_varpool_node): Likewise.
15269         (compute_ltrans_boundary): Remve assert.
15270         (get_alias_symbol): New functoin.
15271         (input_node): Rewrite alias handling.
15272         (input_varpool_node): Likewise.
15273         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15274         * ipa.c (process_references): Handle weakrefs correctly.
15275         (symtab_remove_unreachable_nodes): Likewise.
15276         * trans-mem.c (get_cg_data): Update.
15277         (ipa_tm_create_version_alias): Update.
15278         (ipa_tm_execute): Update.
15279         * symtab.c (dump_symtab_base): Dump aliases.
15280         (verify_symtab_base): Verify aliases.
15281         (symtab_node_availability): New function.
15282         (symtab_alias_ultimate_target): Simplify.
15283         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15284         handle all the fixup cases.
15285         (symtab_resolve_alias): New function.
15286         * passes.c (ipa_write_summaries): Handle weakrefs.
15287         * varpool.c (varpool_analyze_node): Simplify.
15288         (assemble_aliases): Update.
15289         (varpool_create_variable_alias): Simplify.
15290         (varpool_extra_name_alias): Simplify.
15291         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15292         (lto_symtab_merge_symbols): ... this one.
15293
15294 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15295
15296         Revert
15297         PR rtl-optimization/57268
15298         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15299         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15300
15301 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15302
15303         Partially reverted:
15304         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15305
15306         PR middle-end/57073
15307         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15308         further up.
15309
15310 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15311
15312         PR rtl-optimization/57268
15313         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15314         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15315
15316 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15317
15318         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15319         unordered comparison operators when -fno-trapping-math is in effect
15320         on the e500.
15321         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15322         and implement unordered comparison operators properly on the e500.
15323
15324 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15325
15326         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15327         for constant scalar integers.
15328         (simplify_relational_operation_1): Likewise.
15329
15330 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15331
15332         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15333         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15334         Fix comment.
15335
15336 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15337             Igor Zamyatin  <igor.zamyatin@intel.com>
15338
15339         Silvermont (SLM) architecture performance tuning.
15340         * config/i386/i386.h (enum ix86_tune_indices): Add
15341         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15342         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15343
15344         * config/i386/i386.c (initial_ix86_tune_features)
15345         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15346         (ix86_lea_outperforms): Handle Silvermont tuning.
15347         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15348         call.
15349         (ix86_use_lea_for_mov): Likewise.
15350         (ix86_avoid_lea_for_addr): Likewise.
15351         (ix86_lea_for_add_ok): Likewise.
15352         (exact_dependency_1): New function.
15353         (exact_store_load_dependency): Likewise.
15354         (ix86_adjust_cost): Handle Silvermont tuning.
15355         (do_reoder_for_imul): Likewise.
15356         (swap_top_of_ready_list): New function.
15357         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15358
15359         * config/i386/i386.md (peepholes that split memory operand in fp
15360         converts): New.
15361
15362 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15363
15364         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15365         Remove un-necessary braces.
15366
15367 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15368
15369         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15370         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15371
15372 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15373
15374         PR middle-end/57073
15375         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15376
15377 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15378
15379         PR target/56315
15380         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15381         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15382         * config/arm/neon.md (iordi3_neon): Remove.
15383         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15384         * config/arm/predicates.md (imm_for_neon_logic_operand):
15385         Move to earlier in the file.
15386         (neon_logic_op2): Likewise.
15387         (arm_iordi_operand_neon): New predicate.
15388
15389 2013-05-31  Richard Biener  <rguenther@suse.de>
15390
15391         PR tree-optimization/57478
15392         PR tree-optimization/57453
15393         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15394         are life as well.
15395
15396 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15397
15398         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15399         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15400
15401 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15402             Thomas Koenig  <tkoenig@gcc.gnu.org>
15403
15404         PR middle-end/57073
15405         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15406         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15407
15408 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15409
15410         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15411
15412 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15413
15414         * target.def (register_usage_leveling_p): New hook.
15415         * targhooks.c (default_register_usage_leveling_p): New.
15416         * targhooks.h (default_register_usage_leveling_p): New prototype.
15417         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15418         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15419         * doc/tm.texi: Update.
15420         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15421
15422 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15423
15424         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15425         (*insv_reg<mode>): New define_insn.
15426
15427 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15428
15429         PR rtl-optimization/57439
15430         * postreload.c (move2add_valid_value_p): Check that we have
15431         a zero subreg_regno_offset when accessing the register in
15432         the requested mode.
15433
15434 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15435             Igor Zamyatin  <igor.zamyatin@intel.com>
15436
15437         Silvermont (SLM) architecture pipeline model, tuning and
15438         insn selection.
15439         * config.gcc: Add slm config options and target.
15440
15441         * config/i386/slm.md: New.
15442
15443         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15444
15445         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15446         PROCESSOR_SLM.
15447         (ix86_target_macros_internal): Likewise.
15448
15449         * config/i386/i386.c (slm_cost): New cost.
15450         (m_SLM): New macro flag.
15451         (initial_ix86_tune_features): Set m_SLM.
15452         (x86_accumulate_outgoing_args): Likewise.
15453         (x86_arch_always_fancy_math_387): Likewise.
15454         (processor_target_table): Add slm cost.
15455         (cpu_names): Add slm cpu name.
15456         (x86_option_override_internal): Set SLM ISA.
15457         (ix86_issue_rate): New case PROCESSOR_SLM.
15458         (ia32_multipass_dfa_lookahead): Likewise.
15459         (fold_builtin_cpu): Add slm.
15460
15461         * config/i386/i386.h (TARGET_SLM): New target macro.
15462         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15463         (processor_type): Add PROCESSOR_SLM.
15464
15465         * config/i386/i386.md (cpu): Add new value "slm".
15466         (slm.md): Include slm.md.
15467
15468 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15469             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15470
15471         * config/arm/arm-protos.h: Add and update function protos.
15472         * config/arm/arm.c (use_simple_return_p): New added.
15473         (thumb2_expand_return): Check simple_return flag.
15474         * config/arm/arm.md: Add simple_return and conditional simple_return.
15475         * config/arm/iterators.md: Add iterator for return and simple_return.
15476
15477 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15478
15479         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15480         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15481         (arm_emit_vfp_multi_reg_pop): Likewise.
15482         (thumb2_emit_ldrd_pop): Likewise.
15483         (arm_expand_epilogue): Add misc REG_CFA notes.
15484         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15485
15486 2013-05-29  Lawrence Crowl  <crowl@google.com>
15487
15488         * config/arm/t-arm: Update for below.
15489
15490         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15491         Change type to hash_table.  Update dependent calls and types.
15492
15493         * config/i386/t-cygming: Update for below.
15494
15495         * config/i386/t-interix: Update for below.
15496
15497         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15498         Change type to hash_table.  Update dependent calls and types.
15499         (i386_find_on_wrapper_list::wrappers): Likewise.
15500
15501         * config/ia64/t-ia64: Update for below.
15502
15503         * config/ia64/ia64.c (bundle_state_table):
15504         Change type to hash_table.  Update dependent calls and types.
15505
15506         * config/mips/mips.c (mips_reorg_process_insns::htab):
15507         Change type to hash_table.  Update dependent calls and types.
15508
15509         * config/sol2.c (solaris_comdat_htab):
15510         Change type to hash_table.  Update dependent calls and types.
15511
15512         * config/t-sol2: Update for above.
15513
15514 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15515
15516         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15517         functions are not yet marked as defined.
15518
15519 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15520             Pat Haugen  <pthaugen@us.ibm.com>
15521             Peter Bergner  <bergner@vnet.ibm.com>
15522
15523         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15524         instructions.
15525         (VEC_A): Likewise.
15526         (VEC_C): Likewise.
15527         (vrotl<mode>3): Likewise.
15528         (vashl<mode>3): Likewise.
15529         (vlshr<mode>3): Likewise.
15530         (vashr<mode>3): Likewise.
15531
15532         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15533         support for power8 V2DI builtins.
15534
15535         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15536         power8 V2DI builtins.
15537         (vupkhsw): Likewise.
15538         (vupklsw): Likewise.
15539         (vaddudm): Likewise.
15540         (vminsd): Likewise.
15541         (vmaxsd): Likewise.
15542         (vminud): Likewise.
15543         (vmaxud): Likewise.
15544         (vpkudum): Likewise.
15545         (vpksdss): Likewise.
15546         (vpkudus): Likewise.
15547         (vpksdus): Likewise.
15548         (vrld): Likewise.
15549         (vsld): Likewise.
15550         (vsrd): Likewise.
15551         (vsrad): Likewise.
15552         (vsubudm): Likewise.
15553         (vcmpequd): Likewise.
15554         (vcmpgtsd): Likewise.
15555         (vcmpgtud): Likewise.
15556         (vcmpequd_p): Likewise.
15557         (vcmpgtsd_p): Likewise.
15558         (vcmpgtud_p): Likewise.
15559         (vupkhsw): Likewise.
15560         (vupklsw): Likewise.
15561         (vaddudm): Likewise.
15562         (vmaxsd): Likewise.
15563         (vmaxud): Likewise.
15564         (vminsd): Likewise.
15565         (vminud): Likewise.
15566         (vpksdss): Likewise.
15567         (vpksdus): Likewise.
15568         (vpkudum): Likewise.
15569         (vpkudus): Likewise.
15570         (vrld): Likewise.
15571         (vsld): Likewise.
15572         (vsrad): Likewise.
15573         (vsrd): Likewise.
15574         (vsubudm): Likewise.
15575
15576         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15577         support for power8 V2DI instructions.
15578
15579         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15580         power8 V2DI instructions.  Combine pack and unpack insns to use an
15581         iterator for each mode.  Check whether a particular mode supports
15582         Altivec instructions instead of just checking TARGET_ALTIVEC.
15583         (UNSPEC_VPKUWUM): Likewise.
15584         (UNSPEC_VPKSHSS): Likewise.
15585         (UNSPEC_VPKSWSS): Likewise.
15586         (UNSPEC_VPKUHUS): Likewise.
15587         (UNSPEC_VPKSHUS): Likewise.
15588         (UNSPEC_VPKUWUS): Likewise.
15589         (UNSPEC_VPKSWUS): Likewise.
15590         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15591         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15592         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15593         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15594         (UNSPEC_VUPKHSB): Likewise.
15595         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15596         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15597         (UNSPEC_VUPKHSH): Likewise.
15598         (UNSPEC_VUPKLSB): Likewise.
15599         (UNSPEC_VUPKLSH): Likewise.
15600         (VI2): Likewise.
15601         (VI_char): Likewise.
15602         (VI_scalar): Likewise.
15603         (VI_unit): Likewise.
15604         (VP): Likewise.
15605         (VP_small): Likewise.
15606         (VP_small_lc): Likewise.
15607         (VU_char): Likewise.
15608         (add<mode>3): Likewise.
15609         (altivec_vaddcuw): Likewise.
15610         (altivec_vaddu<VI_char>s): Likewise.
15611         (altivec_vadds<VI_char>s): Likewise.
15612         (sub<mode>3): Likewise.
15613         (altivec_vsubcuw): Likewise.
15614         (altivec_vsubu<VI_char>s): Likewise.
15615         (altivec_vsubs<VI_char>s): Likewise.
15616         (altivec_vavgs<VI_char>): Likewise.
15617         (altivec_vcmpbfp): Likewise.
15618         (altivec_eq<mode>): Likewise.
15619         (altivec_gt<mode>): Likewise.
15620         (altivec_gtu<mode>): Likewise.
15621         (umax<mode>3): Likewise.
15622         (smax<mode>3): Likewise.
15623         (umin<mode>3): Likewise.
15624         (smin<mode>3): Likewise.
15625         (altivec_vpkuhum): Likewise.
15626         (altivec_vpkuwum): Likewise.
15627         (altivec_vpkshss): Likewise.
15628         (altivec_vpkswss): Likewise.
15629         (altivec_vpkuhus): Likewise.
15630         (altivec_vpkshus): Likewise.
15631         (altivec_vpkuwus): Likewise.
15632         (altivec_vpkswus): Likewise.
15633         (altivec_vpks<VI_char>ss): Likewise.
15634         (altivec_vpks<VI_char>us): Likewise.
15635         (altivec_vpku<VI_char>us): Likewise.
15636         (altivec_vpku<VI_char>um): Likewise.
15637         (altivec_vrl<VI_char>): Likewise.
15638         (altivec_vsl<VI_char>): Likewise.
15639         (altivec_vsr<VI_char>): Likewise.
15640         (altivec_vsra<VI_char>): Likewise.
15641         (altivec_vsldoi_<mode>): Likewise.
15642         (altivec_vupkhsb): Likewise.
15643         (altivec_vupkhs<VU_char>): Likewise.
15644         (altivec_vupkls<VU_char>): Likewise.
15645         (altivec_vupkhsh): Likewise.
15646         (altivec_vupklsb): Likewise.
15647         (altivec_vupklsh): Likewise.
15648         (altivec_vcmpequ<VI_char>_p): Likewise.
15649         (altivec_vcmpgts<VI_char>_p): Likewise.
15650         (altivec_vcmpgtu<VI_char>_p): Likewise.
15651         (abs<mode>2): Likewise.
15652         (vec_unpacks_hi_v16qi): Likewise.
15653         (vec_unpacks_hi_v8hi): Likewise.
15654         (vec_unpacks_lo_v16qi): Likewise.
15655         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15656         (vec_unpacks_lo_v8hi): Likewise.
15657         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15658         (vec_pack_trunc_v8h): Likewise.
15659         (vec_pack_trunc_v4si): Likewise.
15660         (vec_pack_trunc_<mode>): Likewise.
15661
15662         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15663         V2DI builtins.
15664         (vec_vmaxsd): Likewise.
15665         (vec_vmaxud): Likewise.
15666         (vec_vminsd): Likewise.
15667         (vec_vminud): Likewise.
15668         (vec_vpksdss): Likewise.
15669         (vec_vpksdus): Likewise.
15670         (vec_vpkudum): Likewise.
15671         (vec_vpkudus): Likewise.
15672         (vec_vrld): Likewise.
15673         (vec_vsld): Likewise.
15674         (vec_vsrad): Likewise.
15675         (vec_vsrd): Likewise.
15676         (vec_vsubudm): Likewise.
15677         (vec_vupkhsw): Likewise.
15678         (vec_vupklsw): Likewise.
15679
15680 2013-05-29  Jan Hubicka  <jh@suse.cz>
15681
15682         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15683         flags; reorder rest of fields in more consistent way.
15684         (varpool_node): Remove analyzed, finalized and alias.
15685         (cgraph_ndoe): Likewise.
15686         (symtab_alias_ultimate_target): New function.
15687         (cgraph_function_node): Move offline.
15688         (cgraph_reset_node): Declare.
15689         (cgraph_comdat_can_be_unshared_p): Remove.
15690         (varpool_remove_initializer): Declare.
15691         (varpool_first_defined_variable, varpool_next_defined_variable
15692         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15693         (cgraph_function_with_gimple_body_p): Update.
15694         (varpool_all_refs_explicit_p): Update.
15695         (symtab_alias_target): New function.
15696         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15697         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15698         (cgraph_function_or_thunk_node): Simplify using
15699         symtab_alias_ultimate_target.
15700         (varpool_variable_node): Likewise.
15701         * cgraph.c (cgraph_create_function_alias): Update.
15702         (cgraph_add_thunk): Update.
15703         (cgraph_remove_node): Update.
15704         (dump_cgraph_node): Do not dump removed flags.
15705         (cgraph_function_body_availability): Update.
15706         (cgraph_propagate_frequency): Update.
15707         (verify_cgraph_node): Check sanity of local flag.
15708         (cgraph_function_node): Move here from cgraph.h; revamp for
15709         cgraph_function_or_thunk_node.
15710         * lto-symtab.c (lto_varpool_replace_node): Update.
15711         (lto_symtab_resolve_can_prevail_p): Update.
15712         (lto_symtab_merge_cgraph_nodes): Update.
15713         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15714         propagate_constants_accross_call, devirtualization_time_bonus,
15715         ipcp_propagate_stage): Update.
15716         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
15717         * ipa-inline-transform.c (clone_inlined_nodes,
15718         preserve_function_body_p): Update.
15719         * ipa-reference.c (propagate): Update.
15720         (write_node_summary_p): Update.
15721         * toplev.c (wrapup_global_declaration_2): Update.
15722         * cgraphunit.c (cgraph_analyze_function): Rename to ...
15723         (analyze_function) ... this one.
15724         (cgraph_process_new_functions): Update.
15725         (cgraph_reset_node): Export.
15726         (cgraph_finalize_function): Update.
15727         (cgraph_add_new_function): Update.
15728         (process_function_and_variable_attributes): Update.
15729         (varpool_finalize_decl): Update.
15730         (symbol_finalized): Remove.
15731         (symbol_finalized_and_needed): Rename to ...
15732         (symbol_defined_and_needed): ... update.
15733         (cgraph_analyze_functions): Update.
15734         (handle_alias_pairs): Update.
15735         (mark_functions_to_output): Update.
15736         (assemble_thunk): Update.
15737         (output_in_order): Update.
15738         (output_weakrefs): Update.
15739         (finalize_compilation_unit): Update.
15740         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
15741         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
15742         input_node, input_varpool_node): Update.
15743         * dbxout.c (dbxout_expand_expr): Update.
15744         * cgraphclones.c (cgraph_clone_node): Update.
15745         (cgraph_copy_node_for_versioning): Update.
15746         (cgraph_materialize_clone): Update.
15747         (cgraph_materialize_all_clones): Update.
15748         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
15749         propagate_pure_const, propagate_nothrow): Update.
15750         * lto-streamer-out.c (lto_output, write_symbol): Update.
15751         * ipa-utils.c (ipa_reverse_postorder): Update.
15752         * ipa-inline.c (can_inline_edge_p): Update.
15753         (update_caller_keys, ipa_inline): Update.
15754         * dwarf2out.c (reference_to_unused,
15755         premark_types_used_by_global_vars_helper): Update.
15756         * tree-eh.c (tree_could_trap_p): Update.
15757         * ipa-split.c (consider_split, execute_split_functions): Update.
15758         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
15759         has_addr_references_p): Update; move ahead in file for better
15760         readability.
15761         (process_references): Simplify.
15762         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
15763         bodies are removed.
15764         (cgraph_comdat_can_be_unshared_p): Make static.
15765         (cgraph_externally_visible_p): Update.
15766         (varpool_externally_visible_p): Update.
15767         (function_and_variable_visibility): Update.
15768         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
15769         ipa_tm_mark_force_output_node): Update.
15770         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
15771         estimate_edge_devirt_benefit, inline_generate_summary,
15772         inline_write_summary): Update.
15773         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
15774         * ipa-prop.c (ipa_compute_jump_functions): Update.
15775         (ipa_print_node_params, ipa_prop_read_section,
15776         ipa_update_after_lto_read, read_replacements_section): Update.
15777         * varasm.c (mark_decl_referenced): Update.
15778         (assemble_alias, dump_tm_clone_pairs): Update.
15779         * tree-inline.c (copy_bb): Update.
15780         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
15781         Update.
15782         * symtab.c (dump_symtab_base): Print new flags.
15783         (verify_symtab_base): Verify new flags.
15784         (symtab_alias_ultimate_target): New function.
15785         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
15786         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
15787         Update.
15788         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
15789         Update.
15790         * i386.c (ix86_get_function_versions_dispatcher,
15791         ix86_generate_version_dispatcher_body): Update.
15792         (fold_builtin_cpu): Use varpool_add_new_variable.
15793         * varpool.c (varpool_remove_initializer): Break out from ...
15794         (varpool_remove_node): ... this one.
15795         (dump_varpool_node, varpool_node_for_asm,
15796         cgraph_variable_initializer_availability, varpool_analyze_node,
15797         varpool_assemble_decl, varpool_remove_unreferenced_decls,
15798         varpool_finalize_named_section_flags, varpool_create_variable_alias):
15799         Update.
15800
15801 2013-05-29  Jan Hubicka  <jh@suse.cz>
15802
15803         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
15804
15805 2013-05-29  Easwaran Raman  <eraman@google.com>
15806
15807         PR tree-optimization/57442
15808         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
15809         when control exits the main loop.
15810
15811 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
15812
15813         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
15814         and RX600.
15815         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
15816         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
15817         * rx/t-rx: Add rx100 under multi library matches option for nofpu
15818         option.
15819
15820 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15821
15822         PR tree-optimization/57441
15823         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
15824         Don't limit size of incr_vec to number of candidates.
15825
15826 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
15827
15828         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
15829         and mips16 directories.
15830         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
15831         (MULTILIB_DIRNAMES): Ditto.
15832         (MULTILIB_EXCEPTIONS): Add new exceptions.
15833         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
15834         (MULTILIB_DIRNAMES): Ditto.
15835         (MULTILIB_EXCEPTIONS): Add new exceptions.
15836
15837 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
15838             Marcus Shawcroft  <marcus.shawcroft@arm.com>
15839
15840         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
15841         SYMBOL_TINY_ABSOLUTE.
15842         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
15843         SYMBOL_TINY_ABSOLUTE.
15844         (aarch64_expand_mov_immediate): Likewise.
15845         (aarch64_classify_symbol): Likewise.
15846         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
15847         Permit SYMBOL_TINY_ABSOLUTE.
15848         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
15849
15850 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
15851             Marcus Shawcroft  <marcus.shawcroft@arm.com>
15852
15853         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
15854         Refactor if/switch.  Replace gcc_assert with if.
15855
15856 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
15857
15858         * config/i386/i386.c (initial_ix86_tune_features): Enable
15859         FP Reassociation for AMD bdver1 and bdver2.
15860
15861 2013-05-29  Martin Jambor  <mjambor@suse.cz>
15862
15863         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
15864         and IMAGPART_EXPR do not occur within other handled_components.
15865
15866 2013-05-29  Richard Biener  <rguenther@suse.de>
15867
15868         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
15869         access on whether the use is in the BB we currently try to
15870         vectorize.
15871         (vect_bb_vectorization_profitable_p): Pass the BB we currently
15872         vectorize to vect_bb_slp_scalar_cost.
15873
15874 2013-05-29  Richard Biener  <rguenther@suse.de>
15875
15876         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
15877         computing scalar cost offsetted by stmts that are kept live
15878         by scalar uses.
15879         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
15880         for computation of scalar cost.
15881
15882 2013-05-28  Steve Ellcey  <sellcey@mips.com>
15883
15884         * config/mips/mips-cpus.def (mips32r2): Change processor type.
15885
15886 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
15887
15888         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
15889         array notation built-in reduction functions.
15890         * doc/passes.texi (Passes): Added documentation about changes done
15891         for Cilk Plus.
15892         * doc/invoke.texi (C Dialect Options): Added documentation about
15893         the -fcilkplus flag.
15894         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
15895         (BUILTINS_DEF): Depend on cilkplus.def.
15896         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
15897         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
15898         * cilkplus.def: New file.
15899
15900 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
15901
15902         PR rtl-optimization/57439
15903         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
15904
15905 2013-05-28  Easwaran Raman  <eraman@google.com>
15906
15907         PR tree-optimization/57337
15908         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
15909         (find_insert_point): Correctly identify the insertion point
15910         when two statements with the same UID is compared.
15911
15912 2013-05-28  Richard Biener  <rguenther@suse.de>
15913
15914         PR tree-optimization/56787
15915         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
15916         from the list of data references.
15917         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
15918         clobbers.
15919         (vect_analyze_loop_operations): Likewise.
15920         (vect_transform_loop): Remove clobbers.
15921
15922 2013-05-28  Martin Jambor  <mjambor@suse.cz>
15923
15924         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
15925         and REALPART_EXPRs have scalar type.
15926
15927 2013-05-28  Richard Biener  <rguenther@suse.de>
15928
15929         PR tree-optimization/57411
15930         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
15931         virtual operands.
15932         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
15933         virtual operand propagation.
15934
15935 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
15936
15937         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
15938         destination register for bmasksi_vis.
15939         (vector_init_bshuffle): Likewise.
15940         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
15941
15942 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
15943
15944         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
15945         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
15946         mode if the instruction isn't available in the original mode.
15947         * config/sparc/sparc.opt (mfix-ut699): New option.
15948         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
15949         (divdf3): Turn into expander.
15950         (divdf3_nofix): New insn.
15951         (divdf3_fix): Likewise.
15952         (divsf3): Disable if -mfix-ut699.
15953         (sqrtdf2): Turn into expander.
15954         (sqrtdf2_nofix): New insn.
15955         (sqrtdf2_fix): Likewise.
15956         (sqrtsf2): Disable if -mfix-ut699.
15957
15958 2013-05-27  Richard Biener  <rguenther@suse.de>
15959
15960         PR middle-end/57412
15961         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
15962         block for the new loop.
15963
15964 2013-05-27  Richard Biener  <rguenther@suse.de>
15965
15966         PR tree-optimization/57343
15967         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
15968         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
15969         (number_of_iterations_cond): Do not build the folded tree.
15970
15971 2013-05-27  Richard Biener  <rguenther@suse.de>
15972
15973         Revert
15974         PR middle-end/57381
15975         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
15976         OEP_CONSTANT_ADDRESS_OF retained.
15977
15978         PR tree-optimization/57417
15979         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
15980         for unchanged base.
15981         (set_ssa_val_to): Compare addresses using
15982         get_addr_base_and_unit_offset.
15983
15984 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
15985
15986         PR rtl-optimization/56833
15987         * postreload.c (move2add_record_mode): New function.
15988         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
15989         (move2add_use_add2_insn): Use move2add_record_sym_value.
15990         (move2add_use_add3_insn): Likewise.
15991         (reload_cse_move2add): Use move2add_valid_value_p and
15992         move2add_record_mode.  Invalidate call-clobbered and REG_INC
15993         affected regs by setting reg_mode to VOIDmode.
15994         (move2add_note_store): Don't pretend the inside of a SUBREG is
15995         the actual destination.  Invalidate single/leading registers by
15996         setting reg_mode to VOIDmode.
15997         Use move2add_record_sym_value, move2add_valid_value_p and
15998         move2add_record_mode.
15999
16000 2013-05-27  Richard Biener  <rguenther@suse.de>
16001
16002         PR tree-optimization/57396
16003         * tree-affine.c (double_int_constant_multiple_p): Properly
16004         return false for val == 0 and div != 0.
16005
16006 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16007
16008         * config/mips/mips.h: Use #elif in preprocessor conditions.
16009
16010 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16011
16012         PR target/53916
16013         * config/mips/constraints.md (kl): New constraint.
16014         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16015         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16016         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16017         from LO for MIPS16.
16018         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16019
16020 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16021
16022         PR target/55777
16023         * config/mips/mips.c (mips_can_inline_p): New function.
16024         (TARGET_CAN_INLINE_P): Define.
16025
16026 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16027
16028         * sched-int.h (ds_t, dw_t): Make unsigned int.
16029         Fix documentation that describes how all the ds_t bits are used.
16030         Reserve the last bit for delayed-branch scheduling.
16031         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16032         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16033         (gen_dep_weak_1): Remove prototype.
16034         * sched-deps.c (get_dep_weak_1): Make static.
16035         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16036         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16037         * doc/tm.texi: Regenerate.
16038         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16039
16040 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16041
16042         PR debug/56950
16043         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16044
16045 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16046             Sandra Loosemore  <sandra@codesourcery.com>
16047
16048         * config.gcc (powerpc-*): Allow native for with-cpu.
16049
16050 2013-05-24  Jeff Law  <law@redhat.com>
16051
16052         PR tree-optimization/57124
16053         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16054         conversion feeding a condition if the range has an overflow
16055         if -fstrict-overflow.  Add warnings for when we do make the
16056         transformation.
16057
16058 2013-05-24  Dehao Chen  <dehao@google.com>
16059
16060         * tree-cfg.c (locus_discrim_map): Fix the typo.
16061         (locus_discrim_hasher): Likewise.
16062         (locus_discrim_hasher::hash): Likewise.
16063         (locus_discrim_hasher::equal): Likewise.
16064
16065 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16066
16067         PR tree-optimization/57294
16068         * cgraph.h (ipa_record_stmt_references): Declare.
16069         * cgraphbuild.c (ipa_record_stmt_references): New function.
16070         (build_cgraph_edges): Use ipa_record_stmt_references.
16071         (rebuild_cgraph_edges): Likewise.
16072         (cgraph_rebuild_references): Likewise.
16073         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16074         associated with the old statement and build references from the
16075         newly built statements.
16076         * ipa-ref.c (ipa_remove_stmt_references): New function.
16077         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16078
16079 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16080
16081         * lra-constraints.c (emit_spill_move): Use smaller mode for
16082         mem-mem moves.
16083         (check_and_process_move): Consider mem-reg moves for secondary
16084         too.
16085         (curr_insn_transform): Don't lose insns emitted before for
16086         secondary memory moves.
16087         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16088         reg set up in the current insn.
16089
16090 2013-05-24  Dehao Chen  <dehao@google.com>
16091
16092         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16093         hash function.
16094         (locus_descrim_hasher::equal): Likewise.
16095         (build_gimple_cfg): New discriminator assignment algorithm.
16096         (make_edges): Likewise.
16097         (next_discriminator_for_locus): Likewise.
16098         (same_line_p): Likewise.
16099         (assign_discriminators): Likewise.
16100         (make_cond_expr_edges): Likewise.
16101         (make_gimple_switch_edges): Likewise.
16102         (make_goto_expr_edges): Likewise.
16103         (make_gimple_asm_edges): Likewise.
16104
16105 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16106
16107         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16108         X format specifier to only display bottom 16 bits.
16109         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16110         immediate to match for operand 2, since it will be masked.
16111
16112 2013-05-24  Richard Biener  <rguenther@suse.de>
16113
16114         PR tree-optimization/57287
16115         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16116         all SSA names that occur in abnormal PHIs.
16117
16118 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16119
16120         PR tree-ssa/57385
16121         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16122         that index is not negative.
16123
16124 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16125
16126         PR rtl-optimization/55177
16127         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16128         (simplify_byte_swapping_operation): New.
16129         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16130         (simplify_relational_operation_1): Deal with BSWAP.
16131
16132 2013-05-23  Richard Henderson  <rth@redhat.com>
16133
16134         PR target/56742
16135         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16136         (ix86_reorg): Call it.
16137
16138 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16139
16140         PR target/57379
16141         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16142         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16143         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16144
16145 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16146
16147         PR debug/57351
16148         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16149
16150 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16151             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16152
16153         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16154         * config/aarch64/constraints.md (Usa): Remove.
16155         * doc/md.texi (AArch64 Usa): Remove.
16156
16157 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16158             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16159
16160         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16161         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16162         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16163         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16164
16165 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16166
16167         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16168         instruction (AdvSIMD).
16169         * config/aarch64/aarch64-builtins.c
16170         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16171         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16172
16173 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16174
16175         PR middle-end/57347
16176         * tree.h (contains_bitfld_component_ref_p): Declare.
16177         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16178         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16179         caller.
16180         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16181         not access a bit-field.  Assert all final offsets are byte-aligned.
16182
16183 2013-05-23  Richard Biener  <rguenther@suse.de>
16184
16185         PR tree-optimization/57380
16186         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16187         least one invariant or re-used load.
16188         * passes.c (init_optimization_passes): Move pass_phiprop before
16189         pass_forwprop.
16190
16191 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16192
16193         * config/aarch64/aarch64-simd.md
16194         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16195
16196 2013-05-23  Richard Biener  <rguenther@suse.de>
16197
16198         PR middle-end/57381
16199         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16200         OEP_CONSTANT_ADDRESS_OF retained.
16201
16202 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16203
16204         PR middle-end/57344
16205         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16206         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16207
16208 2013-05-23  Richard Biener  <rguenther@suse.de>
16209
16210         PR rtl-optimization/57341
16211         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16212         instead of true_dependence.
16213
16214 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16215
16216         * bb-reorder.c (branch_threshold): Make const.
16217         (exec_threshold): Ditto.
16218
16219 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16220             Pat Haugen  <pthaugen@us.ibm.com>
16221             Peter Bergner  <bergner@vnet.ibm.com>
16222
16223         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16224         documentation for the power8 crypto builtins.
16225
16226         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16227
16228         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16229         macros for defining power8 builtin functions.
16230         (BU_P8V_AV_2): Likewise.
16231         (BU_P8V_AV_P): Likewise.
16232         (BU_P8V_VSX_1): Likewise.
16233         (BU_P8V_OVERLOAD_1): Likewise.
16234         (BU_P8V_OVERLOAD_2): Likewise.
16235         (BU_CRYPTO_1): Likewise.
16236         (BU_CRYPTO_2): Likewise.
16237         (BU_CRYPTO_3): Likewise.
16238         (BU_CRYPTO_OVERLOAD_1): Likewise.
16239         (BU_CRYPTO_OVERLOAD_2): Likewise.
16240         (XSCVSPDP): Fix typo, point to the correct instruction.
16241         (VCIPHER): Add power8 crypto builtins.
16242         (VCIPHERLAST): Likewise.
16243         (VNCIPHER): Likewise.
16244         (VNCIPHERLAST): Likewise.
16245         (VPMSUMB): Likewise.
16246         (VPMSUMH): Likewise.
16247         (VPMSUMW): Likewise.
16248         (VPERMXOR_V2DI): Likewise.
16249         (VPERMXOR_V4SI: Likewise.
16250         (VPERMXOR_V8HI: Likewise.
16251         (VPERMXOR_V16QI: Likewise.
16252         (VSHASIGMAW): Likewise.
16253         (VSHASIGMAD): Likewise.
16254         (VPMSUM): Likewise.
16255         (VPERMXOR): Likewise.
16256         (VSHASIGMA): Likewise.
16257
16258         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16259         __CRYPTO__ if the crypto instructions are available.
16260         (altivec_overloaded_builtins): Add support for overloaded power8
16261         builtins.
16262
16263         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16264         support for power8 crypto builtins.
16265         (builtin_function_type): Likewise.
16266         (altivec_init_builtins): Add support for builtins that take vector
16267         long long (V2DI) arguments.
16268
16269         * config/rs6000/crypto.md: New file, define power8 crypto
16270         instructions.
16271
16272 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16273             Pat Haugen  <pthaugen@us.ibm.com>
16274             Peter Bergner  <bergner@vnet.ibm.com>
16275
16276         * doc/invoke.texi (Option Summary): Add power8 options.
16277         (RS/6000 and PowerPC Options): Likewise.
16278
16279         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16280         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16281         wm, wn, wr documentation.
16282
16283         * config/rs6000/constraints.md (wm): New constraint for VSX
16284         registers if direct move instructions are enabled.
16285         (wn): New constraint for no registers.
16286         (wq): New constraint for quad word even GPR registers.
16287         (wr): New constraint if 64-bit instructions are enabled.
16288         (wv): New constraint if power8 vector instructions are enabled.
16289         (wQ): New constraint for quad word memory locations.
16290
16291         * config/rs6000/predicates.md (const_0_to_15_operand): New
16292         constraint for 0..15 for crypto instructions.
16293         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16294         as GPR and floating point registers.
16295         (int_reg_operand): New predicate to match only GPR registers.
16296         (base_reg_operand): New predicate to match base registers.
16297         (quad_int_reg_operand): New predicate to match even GPR registers
16298         for quad memory operations.
16299         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16300         operations in both GPR and VSX registers.
16301         (quad_memory_operand): New predicate for quad memory operations.
16302         (reg_or_indexed_operand): New predicate for direct move support.
16303
16304         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16305         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16306         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16307         (POWERPC_MASKS): Add power8 options.
16308         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16309         various options.
16310
16311         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16312         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16313
16314         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16315         (-mpower8-fusion): New power8 options.
16316         (-mpower8-fusion-sign): Likewise.
16317         (-mpower8-vector): Likewise.
16318         (-mcrypto): Likewise.
16319         (-mdirect-move): Likewise.
16320         (-mquad-memory): Likewise.
16321
16322         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16323         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16324         registers.
16325         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16326         (rs6000_debug_vector_unit): Add p8_vector.
16327         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16328         definitions.  Also print fusion state.
16329         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16330         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16331         (rs6000_option_override_internal): Add support for power8.
16332         (rs6000_common_init_builtins): Add debugging for skipped builtins
16333         if -mdebug=builtin.
16334         (rs6000_adjust_cost): Add power8 support.
16335         (rs6000_issue_rate): Likewise.
16336         (insn_must_be_first_in_group): Likewise.
16337         (insn_must_be_last_in_group): Likewise.
16338         (force_new_group): Likewise.
16339         (rs6000_register_move_cost): Likewise.
16340         (rs6000_opt_masks): Likewise.
16341
16342         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16343         power8 capable assembler, default to power7 options.
16344         (TARGET_DIRECT_MOVE): Likewise.
16345         (TARGET_CRYPTO): Likewise.
16346         (TARGET_P8_VECTOR): Likewise.
16347         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16348         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16349         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16350         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16351         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16352         (TARGET_XSCVDPSPN): Likewise.
16353         (TARGET_XSCVSPDPN): Likewsie.
16354         (TARGET_SYNC_HI_QI): Likewise.
16355         (TARGET_SYNC_TI): Likewise.
16356         (MASK_CRYPTO): Likewise.
16357         (MASK_DIRECT_MOVE): Likewise.
16358         (MASK_P8_FUSION): Likewise.
16359         (MASK_P8_VECTOR): Likewise.
16360         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16361         temporary used by some of the direct move instructions to get two FP
16362         temporary registers does not force creation of a stack frame.
16363         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16364         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16365         that any VSX registers are tieable, even if they are also an
16366         Altivec vector mode.
16367         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16368         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16369         (RS6000_BTM_CRYPTO): Likewise.
16370         (RS6000_BTM_COMMON): Likewise.
16371
16372         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16373         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16374         (enum rs6000_vector): Add power8 vector support.
16375
16376 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16377
16378         PR target/19599
16379         PR target/57340
16380         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16381         (any_sibcall_could_use_r3): this and handle indirect calls.
16382         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16383
16384 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16385
16386         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16387
16388 2013-05-22  Richard Biener  <rguenther@suse.de>
16389
16390         PR middle-end/57349
16391         * profile.c (branch_prob): Do not split blocks that are
16392         abnormally receiving from ECF_RETURNS_TWICE functions.
16393
16394 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16395
16396         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16397         address mode.  Move pointer mode initialization to the same place.
16398
16399 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16400
16401         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16402         while it has any effect.
16403
16404 2013-05-21  Easwaran Raman  <eraman@google.com>
16405
16406         PR tree-optimization/57322
16407         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16408         UID of the statement added to the BB to be 1.
16409
16410 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16411
16412         PR tree-optimization/57331
16413         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16414         of conversion from pointer type to integral type with integer.
16415
16416 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16417
16418         PR lto/57289
16419         * ipa-prop.c (ipa_read_node_info): Process param_used and
16420         controlled_uses in the same order as when writing.
16421
16422 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16423
16424         PR plugins/56754
16425         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16426
16427 2013-05-21  Richard Biener  <rguenther@suse.de>
16428
16429         PR tree-optimization/57318
16430         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16431         estimate stmts with side-effects as likely eliminated.
16432
16433 2013-05-21  Richard Biener  <rguenther@suse.de>
16434
16435         PR tree-optimization/57330
16436         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16437         preserve the call stmts fntype.
16438
16439 2013-05-21  Richard Biener  <rguenther@suse.de>
16440
16441         PR tree-optimization/57303
16442         * tree-ssa-sink.c (statement_sink_location): Improve killing
16443         stmt detection and properly handle self-assignments.
16444
16445 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16446
16447         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16448         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16449         registers. Set register size out of the PARALLEL loop.
16450
16451 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16452
16453         PR target/56547
16454         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16455         (*fmasf4, *fmasf4_media): New insns.
16456
16457 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16458
16459         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16460         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16461         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16462         (mips_idiv_insns): Update the comments to say that the returned
16463         instruction counts are in units of BASE_INSN_LENGTH.
16464         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16465         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16466         using 2 rather than 4 as the length of indirect MIPS16 and
16467         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16468         length of a NOP.  Don't divide MIPS16 lengths by 2.
16469         (mips16_split_long_branches): Assume a branch is long if the
16470         length is greater than 4 rather than 8.
16471         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16472         rather than multiplying them by 2.  Multiply instruction counts
16473         by BASE_INSN_LENGTH rather than 4.
16474         (*jump_mips16, tls_get_tp_mips16_<mode>)
16475         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16476
16477 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16478
16479         * config/mips/mips.md (extended_mips16): Remove branch case.
16480         (length): Remove duplicated extended_mips16 test.
16481
16482 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16483
16484         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16485
16486 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16487
16488         * recog.h (Recog_data): Rename to...
16489         (recog_data_d): ...this.
16490         (recog_data): Update accordingly.
16491         * recog.c (recog_data): Likewise.
16492         * reload.c (save_recog_data): Likewise.
16493         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16494         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16495
16496 2013-05-17  Julian Brown  <julian@codesourcery.com>
16497
16498         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16499         found in a REG_EQUAL note, invalidate it.
16500
16501 2013-05-17  Easwaran Raman  <eraman@google.com>
16502
16503         * tree-ssa-reassoc.c (find_insert_point): New function.
16504         (insert_stmt_after): Likewise.
16505         (get_def_stmt): Likewise.
16506         (ensure_ops_are_available): Likewise.
16507         (not_dominated_by): Likewise.
16508         (rewrite_expr_tree): Do not move statements beyond what is
16509         necessary. Remove call to swap_ops_for_binary_stmt...
16510         (reassociate_bb): ... and move it here.
16511         (build_and_add_sum): Assign UIDs for new statements.
16512         (linearize_expr): Likewise.
16513         (do_reassoc): Renumber gimple statement UIDs.
16514
16515 2013-05-17  Jan Hubicka  <jh@suse.cz>
16516
16517         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16518         weakrefs.
16519         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16520         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16521         weakrefs.
16522         (output_weakrefs): Update.
16523
16524 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16525             Martin Jambor  <mjambor@suse.cz>
16526
16527         PR middle-end/57276
16528         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16529         value that corresponds to the given aggval is found in values vector.
16530
16531 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16532
16533         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16534         sse, sse2, sse3, ssse3 and sse4a flags to options.
16535
16536 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16537
16538         * gengtype-state.c: (s_expr_writer): New class, to handle
16539         prettifying of output layout of s-expressions.
16540         (state_writer): New class, to write out gtype.state.
16541         (state_written_type_count): Move this variable into member data of
16542         state_writer.
16543         (s_expr_writer::s_expr_writer): New code: constructor for new class
16544         (state_writer::state_writer(): ditto
16545         (s_expr_writer::write_new_line): New function
16546         (s_expr_writer::write_any_indent): ditto
16547         (s_expr_writer::begin_s_expr): ditto
16548         (s_expr_writer::end_s_expr): ditto
16549         (write_state_fileloc): convert to method of state_writer...
16550         (state_writer:: write_state_fileloc): ...and use methods of
16551         s_expr_writer to write indentation into the gtype.state output file
16552         to visually represent the hierarchical structure of the list
16553         structures
16554         (write_state_fields): ditto, renaming to...
16555         (state_writer::write_state_fields)
16556         (write_state_a_string): ditto, renaming to...
16557         (state_writer::write_state_a_string)
16558         (write_state_string_option): ditto, renaming to...
16559         (state_writer::write_state_string_option)
16560         (write_state_type_option): ditto, renaming to...
16561         (state_writer::write_state_type_option)
16562         (write_state_nested_option): ditto, renaming to...
16563         (state_writer::write_state_nested_option)
16564         (write_state_option): ditto, renaming to...
16565         (state_writer::write_state_option)
16566         (write_state_options): ditto, renaming to...
16567         (state_writer::write_state_options)
16568         (write_state_lang_bitmap): ditto, renaming to...
16569         (state_writer::write_state_lang_bitmap)
16570         (write_state_version): ditto, renaming to...
16571         (state_writer::write_state_version)
16572         (write_state_scalar_type): ditto, renaming to...
16573         (state_writer::write_state_scalar_type)
16574         (write_state_string_type): ditto, renaming to...
16575         (state_writer::write_state_string_type)
16576         (write_state_undefined_type): ditto, renaming to...
16577         (state_writer::write_state_undefined_type)
16578         (write_state_struct_union_type): ditto, renaming to...
16579         (state_writer::write_state_struct_union_type)
16580         (write_state_struct_type): ditto, renaming to...
16581         (state_writer::write_state_struct_type)
16582         (write_state_user_struct_type): ditto, renaming to...
16583         (state_writer::write_state_user_struct_type)
16584         (write_state_lang_struct_type): ditto, renaming to...
16585         (state_writer::write_state_lang_struct_type)
16586         (write_state_param_struct_type): ditto, renaming to...
16587         (state_writer::write_state_param_struct_type)
16588         (write_state_pointer_type): ditto, renaming to...
16589         (state_writer::write_state_pointer_type)
16590         (write_state_array_type): ditto, renaming to...
16591         (state_writer::write_state_array_type)
16592         (write_state_gc_used): ditto, renaming to...
16593         (state_writer::write_state_gc_used)
16594         (write_state_common_type_content): ditto, renaming to...
16595         (state_writer::write_state_common_type_content)
16596         (write_state_type): ditto, renaming to...
16597         (state_writer::write_state_type)
16598         (write_state_pair_list): ditto, renaming to...
16599         (state_writer::write_state_pair_list)
16600         (write_state_pair): ditto, renaming to...
16601         (state_writer::write_state_pair)
16602         (write_state_typedefs): ditto, renaming to...
16603         (state_writer::write_state_typedefs)
16604         (write_state_structures): ditto, renaming to...
16605         (state_writer::write_state_structures)
16606         (write_state_param_structs): ditto, renaming to...
16607         (state_writer::write_state_param_structs)
16608         (write_state_variables): ditto, renaming to...
16609         (state_writer::write_state_variables)
16610         (write_state_srcdir): ditto, renaming to...
16611         (state_writer::write_state_srcdir)
16612         (write_state_files_list): ditto, renaming to...
16613         (state_writer::write_state_files_list)
16614         (write_state_languages): ditto, renaming to...
16615         (state_writer::write_state_languages)
16616         (write_state): create a state_writer instance and use it when
16617         writing out the state file
16618
16619 2013-05-17  Mike Stump  <mikestump@comcast.net>
16620
16621         PR rtl-optimization/57304
16622         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16623         accessing DF_REF_REAL_LOC.
16624
16625 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16626
16627         PR rtl-optimization/57281
16628         PR rtl-optimization/57300
16629         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16630         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16631         what the other splitter did if the registers are dead.
16632
16633 2013-05-17  Richard Biener  <rguenther@suse.de>
16634
16635         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16636         MEM_REF offsets.
16637
16638 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16639
16640         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16641         linking.
16642
16643 2013-05-17  Marek Polacek  <polacek@redhat.com>
16644
16645         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16646         length when doing non-zero store of storing '\0' to '\0'.
16647
16648 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16649
16650         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16651         vect_external_def oprnd1 with loop_vinfo, try to emit
16652         optional cast, negation and and stmts on the loop preheader
16653         edge instead of into the pattern def seq.
16654
16655         PR tree-optimization/57051
16656         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16657         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16658
16659 2013-05-16  Nick Clifton  <nickc@redhat.com>
16660
16661         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16662         (rl78_is_naked_func): New function.
16663         (rl78_expand_prologue): Skip prologue generation for naked functions.
16664         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16665         * doc/extend.texi (naked): Add RL78 to the list of processors
16666         that supports this attribute.
16667
16668 2013-05-16  Jeff Law  <law@redhat.com>
16669
16670         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16671
16672 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16673
16674         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16675         cache parameters using detect_caches_amd also for CYRIX,
16676         NSC and TM2 signatures.
16677
16678 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16679             Dzianis Kahanovich  <mahatma@eu.by>
16680
16681         PR target/45359
16682         PR target/46396
16683         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16684         VIA/Centaur processors and determine their cache parameters
16685         using detect_caches_amd.
16686
16687 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16688
16689         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16690         (rtl_verify_edges): New function.
16691         (rtl_verify_bb_insns): Ditto.
16692         (rtl_verify_bb_pointers): Ditto.
16693         (rtl_verify_bb_insn_chain): Ditto.
16694         (rtl_verify_fallthru): Ditto.
16695         (rtl_verify_bb_layout): Ditto.
16696         (rtl_verify_flow_info_1): Outline checks into new functions.
16697         (rtl_verify_flow_info): Ditto.
16698
16699 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16700
16701         * cfghooks.c (copy_bbs): Add update_dominance argument.
16702         * cfghooks.h (copy_bbs): Update prototype.
16703         * tree-cfg.c (gimple_duplicate_sese_region):
16704         Add update_dominance argument.
16705         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16706         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16707         gimple_duplicate_sese_region call.
16708         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16709         Update copy_bbs call.
16710         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16711         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16712
16713 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16714
16715         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16716         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
16717         vect_recog_rotate_pattern.
16718         (vect_recog_rotate_pattern): New function.
16719
16720 2013-05-16  Jason Merrill  <jason@redhat.com>
16721
16722         * Makefile.in (LLINKER): New variable.
16723         (mostlyclean): Remove link mutex.
16724         * configure.ac: Handle --enable-link-mutex.
16725         * lock-and-run.sh: New script.
16726
16727 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16728
16729         PR target/19599
16730         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
16731         for NULL decl.
16732
16733 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16734
16735         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
16736
16737 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16738
16739         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
16740         * config/arm/arm.c (next_consecutive_mem): New function.
16741         (gen_movmem_ldrd_strd): Likewise.
16742         * config/arm/arm.md (movmemqi): Update condition and code.
16743         (unaligned_loaddi, unaligned_storedi): New patterns.
16744
16745 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16746
16747         * config.gcc: Obsolete *-*-solaris2.9*.
16748         * doc/install.texi (Specific, *-*-solaris2*): Document it.
16749
16750 2013-05-16  Richard Biener  <rguenther@suse.de>
16751
16752         * passes.c (init_optimization_passes): Move pass_parallelize_loops
16753         earlier, after GRAPHITE transforms and IV canonicalization.
16754
16755 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16756
16757         * omp-low.c (extract_omp_for_data): For collapsed loops,
16758         if at least one of the loops is known at compile time to
16759         iterate zero times, set count to 0.
16760         (expand_omp_regimplify_p): New function.
16761         (expand_omp_for_generic): For collapsed loops, if at least
16762         one of the loops isn't known to iterate at least once,
16763         add runtime check with setting count to 0.
16764         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
16765         For unsigned types if it isn't known at compile time that
16766         the loop will iterate at least once, add runtime check to bypass
16767         the whole loop if initial condition isn't true.
16768
16769 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
16770
16771         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
16772
16773 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
16774
16775         PR middle-end/57286
16776         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
16777         transformations to avoid an infinite loop.
16778
16779 2013-05-16  Marek Polacek  <polacek@redhat.com>
16780
16781         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
16782
16783 2013-05-15  Leif Ekblad  <leif@rdos.net>
16784
16785         * config/i386/i386.c (ix86_decompose_address): Use
16786         DEFAULT_TLS_SEG_REG to access TLS segment register.
16787         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
16788         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
16789         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
16790
16791 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
16792
16793         PR target/57260
16794         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
16795         sibling calls to functions that would normally be lazily bound,
16796         unless $gp is call-clobbered.
16797
16798 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
16799
16800         * config/i386/i386.c (ix86_option_override_internal): Update
16801         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
16802         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
16803         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
16804         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
16805         of TARGET_3DNOW.
16806         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
16807
16808 2013-05-15  Andreas Schwab  <schwab@suse.de>
16809
16810         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
16811         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
16812         third operand.
16813
16814 2013-05-15  Teresa Johnson  <tejohnson@google.com>
16815
16816         * loop-unroll.c (report_unroll_peel): Check decision before
16817         emitting unroll/peel message.
16818
16819 2013-05-15  Teresa Johnson  <tejohnson@google.com>
16820
16821         * function.h (has_bb_partition): New rtl_data flag.
16822         (bb_reorder_complete): Ditto.
16823         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
16824         instead of flag_reorder_blocks_and_partition.
16825         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
16826         with some enhancements.
16827         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
16828         * bb-reorder.c (connect_traces): Check for has_bb_partition
16829         instead of flag_reorder_blocks_and_partition.
16830         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
16831         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
16832         verify_hot_cold_block_grouping.
16833         (partition_hot_cold_basic_blocks): Set has_bb_partition.
16834
16835 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16836
16837         PR target/19599
16838         * config/arm/predicates.md (call_insn_operand): New predicate.
16839         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
16840         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
16841         if insn is not a tail call.
16842         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
16843         registers.
16844         * config/arm/arm.h (enum reg_class): New caller save register class.
16845         (REG_CLASS_NAMES): Likewise.
16846         (REG_CLASS_CONTENTS): Likewise.
16847         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
16848         without decls.
16849
16850 2013-05-15  Richard Biener  <rguenther@suse.de>
16851
16852         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
16853         of MSG_OPTIMIZED_LOCATIONS.
16854         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
16855         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
16856         message.
16857         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
16858         of MSG_OPTIMIZED_LOCATIONS.
16859         (execute_vect_slp): Likewise.
16860         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
16861         (vect_create_cond_for_alias_checks): Likewise.
16862         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
16863         (vect_recog_widen_mult_pattern): Likewise.
16864         (vect_recog_widen_sum_pattern): Likewise.
16865         (vect_recog_over_widening_pattern): Likewise.
16866         (vect_recog_widen_shift_pattern): Likewise.
16867         (vect_recog_vector_vector_shift_pattern): Likewise.
16868         (vect_recog_divmod_pattern): Likewise.
16869         (vect_recog_mixed_size_cond_pattern): Likewise.
16870         (vect_recog_bool_pattern): Likewise.
16871         (vect_pattern_recog_1): Likewise.
16872
16873 2013-05-15  Martin Jambor  <mjambor@suse.cz>
16874
16875         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
16876         non-functions to builtin_unreachable.
16877         * ipa-inline-transform.c (inline_call): Do not assert estimates were
16878         correct when new direct edges were discovered.
16879
16880 2013-05-15  Martin Jambor  <mjambor@suse.cz>
16881
16882         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
16883         header, print symbol order instead of node uid, print more information
16884         about indirect edge targets.
16885         (ipa_make_edge_direct_to_target): Print symbol order instead of node
16886         uids.
16887         (ipa_make_edge_direct_to_target): Likewise.
16888         (remove_described_reference): Likewise.
16889         (propagate_controlled_uses): Likewise.
16890         (ipa_print_node_params): Also print symbol order.
16891         (ipcp_transform_function): Print symbol order instead of node uids.
16892         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
16893         (cgraph_get_create_real_symbol_node): Likewise.
16894         * ipa-cp.c (print_lattice): Likewise.
16895         (print_all_lattices): Likewise.
16896         (determine_versionability): Likewise.
16897         (initialize_node_lattices): Likewise.
16898         (estimate_local_effects): Likewise.
16899         (update_profiling_info): Likewise.
16900         (create_specialized_node): Likewise.
16901         (perhaps_add_new_callers): Likewise.
16902         (decide_about_value): Likewise.
16903         (decide_whether_version_node): Likewise.
16904         (identify_dead_nodes): Likewise.
16905         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
16906         (dump_inline_summary): Likewise.
16907         (estimate_node_size_and_time): Likewise.
16908         (inline_analyze_function): Likewise.
16909         * ipa-inline.c (report_inline_failed_reason): Likewise.
16910         (want_early_inline_function_p): Likewise.
16911         (edge_badness): Likewise.
16912         (update_edge_key): Likewise.
16913         (inline_small_functions): Likewise.  Add dumping of order to two other
16914         dumps.
16915         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
16916         instead of node uids.
16917         (propagate_pure_const): Likewise.
16918         (propagate_pure_const): Likewise.
16919         * ipa-utils.c (dump_cgraph_node_set): Likewise.
16920         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
16921         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
16922         of node uids.
16923         * tree-pretty-print.c (dump_function_header): Likewise.
16924         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
16925         Print symbol order instead of node uids.
16926
16927 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16928
16929         * config/s390/s390.c (s390_register_move_cost): Don't impose the
16930         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
16931
16932 2013-05-15  Richard Biener  <rguenther@suse.de>
16933
16934         PR tree-optimization/57275
16935         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
16936         return value for fail to do runtime alias checks for gather loads.
16937
16938 2013-05-15  Jan Hubicka  <jh@suse.cz>
16939
16940         PR lto/57038
16941         PR lto/47375
16942         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
16943         weakrefs are not external.
16944         (lto_symtab_merge_decls): Fix thinko when dealing with
16945         non-lto_symtab decls.
16946         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
16947         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
16948         * varpool.c (dump_varpool_node): Dump more flags.
16949
16950 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16951
16952         * config/i386/i386.c (processor_alias_table): Add instruction
16953         FSGSBASE for AMD bdver3 architecture.
16954
16955 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
16956
16957         * tree.c (warn_deprecated_use): Print file:line using locus color.
16958         * diagnostic.c (diagnostic_report_current_module): Print file:line
16959         and file:line:column using locus color.
16960
16961 2013-05-14  Mike Stump  <mikestump@comcast.net>
16962
16963         * gdbinit.in: Add __null.
16964
16965 2013-05-14  Mike Stump  <mikestump@comcast.net>
16966
16967         * recog.h: Rename struct recog_data to Recog_data.
16968         * recog.c: Likewise.
16969         * reload.c (can_reload_into): Likewise.
16970         * config/picochip/picochip.c: Likewise.
16971
16972 2013-05-14  Mike Stump  <mikestump@comcast.net>
16973
16974         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
16975
16976 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
16977
16978         * resource.h (struct resources): Remove unch_memory member.
16979         (CLEAR_RESOURCE): Don't clear unch_memory.
16980         * resource.c (mark_referenced_resources): Don't set it.
16981         (mark_set_resources): Likewise.
16982         (mark_target_live_regs): Don't clear it.
16983         (init_resource_info): Likewise.
16984         * reorg.c (resource_conflicts_p): Don't compare it.
16985         (redundant_insn): Don't set it.
16986
16987         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
16988         Remove prototypes.
16989         * emit-rtl.c (next_label): Remove unused function.
16990         (skip_consecutive_labels, link_cc0_insns): Move to ...
16991         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
16992         only place where these functions are used, and make them static.
16993
16994 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
16995
16996         * fold-const.c (fold_negate_expr): Handle vectors.
16997         (fold_truth_not_expr): Make it static.
16998         (fold_invert_truthvalue): New static function.
16999         (invert_truthvalue_loc): Handle vectors. Do not call
17000         fold_truth_not_expr directly.
17001         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17002         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17003         (fold_comparison): Handle vector constants.
17004         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17005         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17006         * tree.h (fold_truth_not_expr): Remove declaration.
17007
17008 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17009
17010         * config/aarch64/aarch64-simd.md
17011         (aarch64_vcond_internal<mode>): Rename to...
17012         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17013         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17014         float modes. Clarify all iterator modes.
17015         (vcond<mode><mode>): Use new name for vcond expanders.
17016         (vcond<v_cmp_result><mode>): Likewise.
17017         (vcondu<mode><mode>: Likewise.
17018         * config/aarch64/iterators.md (VDQF_COND): New.
17019
17020 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17021
17022         PR bootstrap/57266
17023         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17024         variable for the shift amount. Check that we shift by non-negative
17025         amounts.
17026
17027 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17028
17029         PR target/42017
17030         * config/arm/arm.h (EPILOGUE_USES): Only return true
17031         for LR_REGNUM after epilogue_completed.
17032
17033 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17034
17035         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17036         is error_mark_node.
17037
17038 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17039
17040         PR target/57261
17041         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17042         and Solaris 11+/x86 with gld.
17043         * configure: Regenerate.
17044
17045 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17046
17047         * expmed.c (expand_shift_1): Canonicalize rotates by
17048         constant bitsize / 2 to bitsize - 1.
17049         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17050         case ROTATERT>: Likewise.
17051
17052         Revert:
17053         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17054
17055         * config/i386/i386.md (rotateinv): New code attr.
17056         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17057         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17058         roll $31, %eax, etc.
17059
17060 2013-05-14  Richard Biener  <rguenther@suse.de>
17061
17062         PR middle-end/57235
17063         * tree-eh.c (sink_clobbers): Give up for successors with
17064         multiple predecessors and no virtual uses.
17065
17066 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17067
17068         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17069         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17070
17071 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17072
17073         PR middle-end/57251
17074         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17075         the case when both op0 and op1 have VOIDmode.
17076
17077 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17078
17079         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17080         in multiply-accumulate mode.
17081
17082 2013-05-13  Guozhi Wei  <carrot@google.com>
17083
17084         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17085
17086 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17087
17088         PR target/56975
17089         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17090         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17091         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17092         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17093         (ix86_expand_prologue): Likewise.
17094         (ix86_expand_split_stack_prologue): Likewise.
17095         (legitimate_pic_address_disp_p): Likewise.
17096         (legitimize_pic_address): Likewise.
17097         (legitimize_tls_address): Likewise.
17098         (legitimize_pe_coff_symbol): Likewise.
17099         (output_pic_addr_const): Likewise.
17100         (construct_plt_address): Likewise.
17101         (ix86_expand_call): Likewise.
17102         (x86_output_mi_thunk): Likewise.
17103         (x86_function_profiler): Likewise.
17104
17105 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17106
17107         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17108         similar switch cases.
17109         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17110         (aarch64_simd_mov_to_<mode>low): Delete.
17111         (aarch64_simd_mov_to_<mode>high): Delete.
17112         (move_lo_quad_<mode>): Add w<-r alternative.
17113         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17114         (aarch64_simd_mov_from_*): Update type attribute.
17115         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17116         statement.
17117
17118 2013-05-13  Jan Hubicka  <jh@suse.cz>
17119
17120         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17121         * config/i386/i386.c (ix86_compute_frame_layout,
17122         ix86_expand_epilogue, emit_i387_cw_initialization,
17123         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17124         ix86_local_alignment): Fix use of size/speed predicates.
17125
17126 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17127
17128         PR tree-optimization/45216
17129         PR tree-optimization/57157
17130         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17131         the (-Y) & (B - 1) variant if OP is |.
17132         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17133         return shifted.  Use (-op1) & (prec - 1) as other_amount
17134         instead of prec - op1.
17135
17136 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17137
17138         PR middle-end/42371
17139         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17140         (ipa_constant_data): New type.
17141         (ipa_jump_func): Use ipa_constant_data to hold information about
17142         constant jump functions.
17143         (ipa_get_jf_constant): Adjust to jump function type changes.
17144         (ipa_get_jf_constant_rdesc): New function.
17145         (ipa_param_descriptor): New field controlled_uses.
17146         (ipa_get_controlled_uses): New function.
17147         (ipa_set_controlled_uses): Likewise.
17148         * ipa-ref.h (ipa_find_reference): Declare.
17149         * ipa-prop.c (ipa_cst_ref_desc): New type.
17150         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17151         changes.
17152         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17153         New parameter cs.  Adjust all callers.
17154         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17155         (remove_described_reference): New function.
17156         (jfunc_rdesc_usable): Likewise.
17157         (try_make_edge_direct_simple_call): Decrement controlled use count,
17158         attempt to remove reference if it hits zero.
17159         (combine_controlled_uses_counters): New function.
17160         (propagate_controlled_uses): Likewise.
17161         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17162         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17163         (ipa_print_node_params): Print described use counter.
17164         (ipa_write_jump_function): Adjust to jump function type changes.
17165         (ipa_read_jump_function): New parameter CS, pass it to
17166         ipa_set_jf_constant.  Adjust caller.
17167         (ipa_write_node_info): Stream controlled use count
17168         (ipa_read_node_info): Likewise.
17169         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17170         asserting.
17171         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17172         count.  Remove cloning-added reference if it reaches zero.
17173         * ipa-ref.c (ipa_find_reference): New function.
17174
17175 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17176
17177         * config/i386/i386.c (processor_target_table): Modified default
17178         alignment values for AMD BD and BT architectures.
17179
17180 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17181
17182         * tree-vect-generic.c (uniform_vector_p): Move ...
17183         * tree.c (uniform_vector_p): ... here.
17184         * tree.h (uniform_vector_p): Declare it.
17185         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17186         into a scalar.
17187
17188 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17189
17190         PR tree-optimization/57230
17191         * tree-ssa-strlen.c (handle_char_store): Record length for
17192         array store from STRING_CST.
17193
17194         PR tree-optimization/57230
17195         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17196         check.
17197
17198 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17199
17200         * config/epiphany/epiphany.c (epiphany_init): Check size of
17201         NUM_MODES_FOR_MODE_SWITCHING.
17202         (epiphany_expand_prologue):
17203         Remove CONFIG_REGNUM initial value handling code.
17204         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17205         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17206         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17207         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17208         Don't return 1 for FP_MODE_NONE.
17209         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17210         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17211         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17212         * config/epiphany/epiphany.md (save_config): New pattern.
17213
17214 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17215
17216         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17217
17218 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17219
17220         * config/i386/i386.md (memory): Handle sseishft1.
17221         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17222         (*vec_extractv2di_1): Ditto.
17223
17224 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17225
17226         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17227         saved registers.
17228
17229 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17230
17231         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17232         Add mthumb/march=armv7-a multilib.
17233         Add mthumb/march=armv7-r multilib.
17234         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17235
17236 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17237
17238         * config/v850/t-rtems: Add more multilibs.
17239
17240 2013-05-10  Richard Biener  <rguenther@suse.de>
17241
17242         PR tree-optimization/57214
17243         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17244         not propagate from SSA names that occur in abnormal PHI nodes.
17245
17246 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17247
17248         * stor-layout.c (element_precision): New function.
17249         * machmode.h (element_precision): Declare it.
17250         * tree.c (build_minus_one_cst): New function.
17251         (element_precision): Likewise.
17252         * tree.h (build_minus_one_cst): Declare new function.
17253         (element_precision): Likewise.
17254         * fold-const.c (operand_equal_p): Use element_precision.
17255         (fold_binary_loc): Handle vector types.
17256         * convert.c (convert_to_integer): Use element_precision.
17257         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17258         separately.
17259
17260 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17261
17262         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17263         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17264         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17265         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17266         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17267         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17268         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17269         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17270         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17271         (Uuw8): New constraints.
17272         (Usb4): Move into alphabetical order.
17273         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17274         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17275         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17276         previously unnamed patterns.
17277         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17278         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17279         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17280         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17281         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17282         of specific lengths.
17283
17284 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17285
17286         * config/i386/i386.md (rotateinv): New code attr.
17287         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17288         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17289         roll $31, %eax, etc.
17290
17291         PR tree-optimization/45216
17292         PR tree-optimization/57157
17293         * tree-ssa-forwprop.c (simplify_rotate): New function.
17294         (ssa_forward_propagate_and_combine): Call it.
17295
17296 2013-05-10  Richard Biener  <rguenther@suse.de>
17297
17298         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17299         disable peeling when we version for aliasing.
17300         (vector_alignment_reachable_p): Honor explicit user alignment.
17301         (vect_supportable_dr_alignment): Likewise.
17302         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17303         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17304         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17305         then peeling to arrange for the cost-model check to come first.
17306
17307 2013-05-10  Alan Modra  <amodra@gmail.com>
17308
17309         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17310         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17311         * configure: Regenerate.
17312
17313 2013-05-10  Alan Modra  <amodra@gmail.com>
17314
17315         PR target/55033
17316         * varasm.c (default_elf_select_section): Move !DECL_P check..
17317         (get_named_section): ..to here before calling get_section_name.
17318         Adjust assertion.
17319         (default_section_type_flags): Add DECL_P check.
17320         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17321         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17322
17323 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17324
17325         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17326         When using gen_stack_adjust_str with a register offset, add a
17327         REG_FRAME_RELATED_EXPR note.
17328
17329 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17330
17331         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17332         (*vec_extractv4si_zext_mem): Ditto.
17333         (*vec_extractv2di): Add 0->x and x->x alternatives.
17334         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17335         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17336
17337 2013-05-09  Jason Merrill  <jason@redhat.com>
17338
17339         N3639 C++1y VLA support
17340         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17341         DECL_VALUE_EXPR.
17342
17343         * tree.c (build_constructor_va): New.
17344         * tree.h: Declare it.
17345
17346 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17347
17348         PR lto/57084
17349         * gimple-fold.c (canonicalize_constructor_val): Call
17350         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17351
17352 2013-05-09  Jan Hubicka  <jh@suse.cz>
17353             Richard Biener  <rguenther@suse.de>
17354
17355         PR lto/54095
17356         * symtab.c (symtab_make_decl_local): Do not add private names.
17357
17358 2013-05-09  Jan Hubicka  <jh@suse.cz>
17359
17360         PR lto/54095
17361         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17362         (unlink_from_assembler_name_hash): Likewise.
17363         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17364         symtab_unregister_node, symtab_initialize_asm_name_hash,
17365         change_decl_assembler_name): Update.
17366
17367 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17368
17369         * config/aarch64/aarch64.md: New movtf split.
17370         (*movtf_aarch64): Update.
17371         (aarch64_movdi_tilow): Handle TF modes and rename to
17372         aarch64_movdi_<mode>low.
17373         (aarch64_movdi_tihigh): Handle TF modes and rename to
17374         aarch64_movdi_<mode>high
17375         (aarch64_movtihigh_di): Handle TF modes and rename to
17376         aarch64_mov<mode>high_di
17377         (aarch64_movtilow_di): Handle TF modes and rename to
17378         aarch64_mov<mode>low_di
17379         (aarch64_movtilow_tilow): Remove spurious whitespace.
17380         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17381         splits.
17382         (aarch64_print_operand): Update.
17383
17384 2013-05-09  Alan Modra  <amodra@gmail.com>
17385
17386         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17387         powerpc64le.
17388         * configure: Regenerate.
17389
17390 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17391
17392         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17393         splitter preparation statements.
17394         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17395         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17396         adjust_address_nv.
17397
17398 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17399
17400         * gimple-ssa-strength-reduction.c (count_candidates): Change
17401         return value to int.
17402         (analyze_candidates_and_replace): Change type of length to int.
17403
17404 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17405
17406         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17407         (*vec_extract<mode>): Use VI12_128 mode iterator.
17408         (*vec_extract<mode>_mem): Ditto.
17409         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17410         attribute.
17411
17412 2013-05-08  Diego Novillo  <dnovillo@google.com>
17413
17414         PR bootstrap/54659
17415
17416         Revert:
17417         2012-08-17  Diego Novillo  <dnovillo@google.com>
17418
17419         PR bootstrap/54281
17420         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17421         * config.in: Regenerate.
17422         * configure: Regenerate.
17423         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17424
17425 2013-05-08  Jan Hubicka  <jh@suse.cz>
17426
17427         PR lto/54095
17428         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17429         * cgraph.h (symtab_node_base): Add unique_name.
17430         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17431         input_overwrite_node, input_varpool_node): Stream unique_name.
17432         * cgraphclones.c (cgraph_create_virtual_clone,
17433         cgraph_function_versioning): Set unique_name.
17434         * ipa.c (function_and_variable_visibility): Set unique_name.
17435
17436 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17437
17438         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17439         (alloc_cand_and_find_basis): Restrict conditional candidate
17440         processing to CAND_MULTs.
17441
17442 2013-05-08  Jan Hubicka  <jh@suse.cz>
17443
17444         PR lto/54095
17445         lto-symtab.c (lto_symtab_symbol_p): New function.
17446         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17447         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17448         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17449         Skip static symbols.
17450
17451 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17452
17453         PR tree-optimization/57200
17454         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17455         Only call inform if the preceding warning_at returns true.
17456
17457 2013-05-07  Han Shen  <shenhan@google.com>
17458
17459         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17460         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17461         * common.opt (fstack-protector-strong): New option.
17462         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17463         * doc/invoke.texi (Optimization Options): Document
17464         "-fstack-protector-strong".
17465         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17466
17467 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17468
17469         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17470
17471 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17472
17473         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17474         it is not lhs of the insn.
17475         (setup_reg_equiv): Remove insn having equiv memory which it is not
17476         lhs of the insn.
17477         * lra-constraints.c (process_address): Try to improve generation
17478         code for address base + disp.
17479         (lra_constraints): Make correct the code for checking insn setting
17480         up backward equivalence.  Remove insn only if it is in the init
17481         insn list.
17482         * lra-eliminations.c (update_reg_eliminate): Change return value.
17483         (lra_eliminate): Use the result.
17484
17485 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17486
17487         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17488         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17489         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17490         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17491         register target operands.
17492         (*vec_extractv8hi_sse2): New pattern.
17493         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17494         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17495         (*vec_extract<mode>_mem): New insn and split pattern.
17496
17497 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17498
17499         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17500         (TARGET_ASAN_SHADOW_OFFSET): Define.
17501         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17502         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17503
17504 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17505
17506         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17507         (incr_vec_index): Return -1 if increment not found.
17508         (create_add_on_incoming_edge): Assert if increment not found.
17509         (record_increment): Limit number of increments recorded.
17510         (all_phi_incrs_profitable): Return false if an increment not found.
17511         (replace_profitable_candidates): Don't process increments that were
17512         not recorded.
17513         (analyze_candidates_and_replace): Limit size of incr_vec.
17514
17515 2013-05-07  Richard Biener  <rguenther@suse.de>
17516
17517         * calls.c (special_function_p): setjmp-like functions are leaf.
17518         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17519         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17520
17521 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17522
17523         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17524         (aarch64_simd_mov<mode>): New expander.
17525         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17526         (aarch64_simd_mov_to_<mode>high): Likewise.
17527         (aarch64_simd_mov_from_<mode>low): Likewise.
17528         (aarch64_simd_mov_from_<mode>high): Likewise.
17529         (aarch64_dup_lane<mode>): Update.
17530         (aarch64_dup_lanedi): New instruction pattern.
17531         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17532         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17533
17534 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17535
17536         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17537         (replace_mult_candidate): Remove unnecessary argument; remove
17538         unnecessary parameter from call to introduce_cast_before_cand.
17539         (replace_unconditional_candidate): Remove unnecessary parameter
17540         from call to replace_mult_candidate.
17541         (replace_conditional_candidate): Likewise.
17542         (insert_initializers): Use make_temp_ssa_name.
17543         (introduce_cast_before_cand): Remove unnecessary argument; use
17544         make_temp_ssa_name.
17545         (replace_one_candidate): Remove unnecessary argument; remove
17546         unnecessary parameter from calls to introduce_cast_before_cand.
17547         (replace_profitable_candidates): Remove unnecessary parameters
17548         from calls to replace_one_candidate.
17549
17550 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17551
17552         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17553         phi def as possibly hiding a basis for a CAND_ADD whose operands
17554         have been commuted in the analysis.
17555         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17556
17557 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17558
17559         * config/aarch64/aarch64.md
17560         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17561         shift value between 0-4.
17562
17563 2013-05-07  Richard Biener  <rguenther@suse.de>
17564
17565         * double-int.h (rshift): New overload.
17566         * double-int.c (rshift): New function.
17567         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17568         (create_reference_ops_from_ref): Remove.
17569         (vn_reference_insert): Use shared ops for constructing the
17570         reference and copy it.
17571
17572 2013-05-07  Richard Biener  <rguenther@suse.de>
17573
17574         PR middle-end/57190
17575         * tree-eh.c (sink_clobbers): Properly propagate
17576         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17577
17578 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17579
17580         PR tree-optimization/57149
17581         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17582         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17583         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17584         uninit_undefined_value_p instead of ssa_undefined_value_p.
17585
17586         PR debug/57184
17587         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17588         for modifier == EXPAND_INITIALIZER.
17589
17590 2013-05-07  Anton Blanchard  <anton@samba.org>
17591
17592         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17593         for powerpc64 little endian.
17594         * configure: Regenerate.
17595
17596 2013-05-06  Graham Stott  <grahams@btinternet.com>
17597
17598         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17599         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17600         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17601         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17602
17603 2013-05-06  Graham Stott  <grahams@btinternet.com>
17604
17605         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17606         codes which allow non-lvalues.
17607
17608 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17609
17610         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17611         components are all 1s.
17612         (integer_minus_onep): New function.
17613         * tree.h (integer_minus_onep): Declare it.
17614         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17615         integer_minus_onep instead of integer_all_onesp.
17616
17617 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17618
17619         PR target/52933
17620         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17621         variations of these patterns.
17622
17623 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17624
17625         * config/i386/i386.md (isa): Add x64_sse4 member.
17626         (enabled): Handle x64_sse4.
17627         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17628         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17629         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17630         instruction for SSE4_1 targets.  Update insn attributes.
17631         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17632         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17633         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17634         const_1 selector.
17635         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17636         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17637         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17638
17639 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17640
17641         PR target/57108
17642         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17643
17644 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17645
17646         * final.c (do_assembler_dialects): Don't handle curly braces and
17647         vertical bar escaped by % as dialect delimiters.
17648         (output_asm_insn): Print curly braces and vertical bar if escaped
17649         by % and ASSEMBLER_DIALECT defined.
17650         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17651         * doc/tm.texi: Regenerated.
17652
17653 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17654
17655         * config/mips/mips.c: Include tree-pass.h.
17656         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17657         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17658         (pass_mips_machine_reorg2): New machine specific pass.
17659         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17660         (mips_option_override): Register the new pass.
17661         * rtl.h (cleanup_barriers): Remove prototype.
17662         (dbr_schedule): Likewise.
17663         * jump.c (cleanup_barriers): Make static.
17664         * reorg.c (dbr_schedule): Likewise.
17665
17666 2013-05-06  Richard Biener  <rguenther@suse.de>
17667
17668         PR tree-optimization/57185
17669         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17670         SSA names properly.
17671
17672 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17673
17674         PR target/57106
17675         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17676         between "first_arg" and "insn", not "last" and "insn".
17677
17678 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17679
17680         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17681         (find_candidates_in_block): Re-enable slsr_process_phi.
17682         (create_phi_basis): Fix double counting of candidate adjustment.
17683
17684 2013-05-06  Richard Biener  <rguenther@suse.de>
17685
17686         PR middle-end/57147
17687         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17688         the edge is also fallthru, preserve it and just clear the
17689         abnormal flag.
17690         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17691         also complex, preserve that and just clear the fallthru flag.
17692         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17693         update virtual operands.
17694
17695 2013-05-06  Alan Modra  <amodra@gmail.com>
17696
17697         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17698         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17699         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17700         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17701         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17702         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17703
17704 2013-05-06  Alan Modra  <amodra@gmail.com>
17705
17706         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17707         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17708         (DEFAULT_ASM_ENDIAN): Define.
17709         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17710         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17711         Update -K PIC clause from sysv4.h.
17712         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17713         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17714
17715 2013-05-06  Alan Modra  <amodra@gmail.com>
17716
17717         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
17718         twice for little-endian.
17719         (ashrdi3_no_power, ashrdi3): Support little-endian.
17720
17721 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17722
17723         PR target/55303
17724         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
17725         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
17726         related expanders.
17727         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
17728         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
17729         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
17730         New predicates.
17731
17732 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
17733             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17734
17735         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
17736         * config/pa/pa.opt: Make mbig-switch a no-op.
17737         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
17738         (CASE_VECTOR_MODE): Always return SImode.
17739         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
17740         for the !TARGET_BIG_SWITCH case.
17741         * config/pa/pa-linux.h: Likewise.
17742         * config/pa/pa-openbsd.h: Likewise.
17743         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
17744         * config/pa/pa.md (short_jump): Remove define_insn.
17745         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
17746         (casesi0): Remove define_insn.
17747         (type): Remove btable_branch.
17748         (pa_combine_type): Likewise.
17749         (in_nullified_branch_delay): Likewise.
17750         (in_call_delay): Likewise.
17751         (define_delay): Likewise.
17752         (define_insn_reservation "Z3"): Likewise.
17753         (define_insn_reservation "Z4"): Likewise.
17754         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
17755         (pa_adjust_insn_length): Remove adjustment for btable branches.
17756         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
17757         and mno-big-switch
17758
17759 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
17760
17761         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17762         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
17763         Add m->r,x alternatives.
17764         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
17765         splitters using SWI48x mode iterator.
17766         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
17767         TARGET_64BIT.  Add m->x alternative.
17768         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
17769         Add o->x alternative.  Enable for TARGET_SSE.
17770         (sse_storeq): Remove expander.
17771         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
17772         with memory input operand.
17773         (*vec_extractv2di_1 splitter): New.
17774         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
17775         * config/i386/i386.md (ssevecmodelower): New mode attribute.
17776
17777 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
17778
17779         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
17780         (INT_LOWPART): Delete.
17781         (extract_MB): Adjust.
17782         (extract_ME): Adjust.
17783         (print_operand): Adjust.
17784
17785 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
17786
17787         * config/rs6000/predicates.md (reg_or_add_cint_operand,
17788         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
17789         (reg_or_logical_cint_operand, easy_fp_constant,
17790         logical_const_operand): Delete "CONST_DOUBLE" case.
17791         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
17792         "HOST_BITS_PER_WIDE_INT == 64" test.
17793         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
17794         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
17795         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
17796         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
17797         test.
17798         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
17799         CONST_DOUBLE DImode/VOIDmode case.
17800         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
17801         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
17802         CONST_DOUBLE VOIDmode case.
17803         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
17804         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
17805         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
17806         Delete CONST_DOUBLE case.
17807         (splitters for mov FMOVE64 const_double): Delete
17808         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
17809         "HOST_BITS_PER_WIDE_INT >= 64" test.
17810         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
17811         case.
17812         (mov DI const_double): Delete.
17813
17814 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
17815
17816         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
17817         on op shows all bits zero in mode of a lowpart subreg, return zero.
17818
17819 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
17820
17821         PR target/57150
17822         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
17823         to save TFmode registers and DImode to save TImode registers for
17824         caller save operations.
17825         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
17826         mark being partially clobbered since they only use the first
17827         double word.
17828
17829         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
17830         and TDmode only use the upper 64-bits of each VSX register.
17831
17832 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17833
17834         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
17835         (find_candidates_in_block): Disable slsr_process_phi.
17836
17837 2013-05-03  Guozhi Wei  <carrot@google.com>
17838
17839         * coverage.c (coverage_obj_init): Move the construction of gcov
17840         constructor to ...
17841         (build_init_ctor): ... here.
17842
17843 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17844
17845         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
17846         (slsr_cand_d): Redefine def_phi.
17847         (stride_status, phi_adjust_status, count_phis_status): New enums.
17848         (find_phi_def): New.
17849         (find_basis_for_base_expr): New.
17850         (find_basis_for_candidate): Handle hidden bases.
17851         (alloc_cand_and_find_basis): Handle phi candidates.
17852         (slsr_process_phi): New.
17853         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
17854         (create_mul_imm_cand): Likewise.
17855         (create_add_ssa_cand): Exclude phi base candidates.
17856         (create_add_imm_cand): Likewise.
17857         (slsr_process_cast): Likewise.
17858         (slsr_process_copy): Likewise.
17859         (find_candidates_in_block): Handle phi candidates.
17860         (dump_candidate): Likewise.
17861         (unconditional_cands): Delete.
17862         (unconditional_cands_with_known_stride_p): Delete.
17863         (phi_dependent_cand_p): New.
17864         (cand_increment): Handle phi-dependent candidates.
17865         (replace_dependent): Delete.
17866         (replace_mult_candidate): New.
17867         (replace_unconditional_candidate): New.
17868         (incr_vec_index): Move to avoid forward reference.
17869         (create_add_on_incoming_edge): New.
17870         (create_phi_basis): New.
17871         (replace_dependents): Delete.
17872         (replace_conditional_candidate): New.
17873         (phi_add_costs): New.
17874         (replace_uncond_cands_and_profitable_phis): New.
17875         (record_increment): Handle phi adjustments.
17876         (record_phi_increments): New.
17877         (record_increments): Handle phi adjustments.
17878         (phi_incr_cost): New.
17879         (lowest_cost_path): Handle phis.
17880         (total_savings): Likewise.
17881         (analyze_increments): Likewise.
17882         (ncd_with_phi): New.
17883         (ncd_of_cand_and_phis): New.
17884         (nearest_common_dominator_for_cands): Handle phi increments.
17885         (all_phi_incrs_profitable): New.
17886         (replace_profitable_candidates): Handle phi-dependent candidates.
17887         (analyze_candidates_and_replace): Likewise.
17888
17889 2013-05-03  Teresa Johnson  <tejohnson@google.com>
17890
17891         PR bootstrap/57154
17892         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
17893         do not exceed REG_BR_PROB_BASE.
17894
17895 2013-05-03  Jeff Law  <law@redhat.com>
17896
17897         PR tree-optimization/57144
17898         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
17899         operand of the condition will bit into the new type when eliminating
17900         a cast feeding a condition.
17901
17902 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
17903
17904         PR rtl-optimization/57130
17905         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
17906         of COMPARE as in_code to the recursive call if needed.
17907
17908 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
17909
17910         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
17911         (enabled): Handle new members.
17912         * config/i386/sse.md (*vec_concatv2si): Merge from
17913         *vec_concatv2si_sse2 and vec_concatv2si_sse.
17914         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
17915
17916 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
17917
17918         PR tree-optimization/57027
17919         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
17920         for fnms opportunity, check we got the prerequisite kind
17921         of tree / gimple before using accessor functions.
17922
17923 2013-05-03  Richard Biener  <rguenther@suse.de>
17924
17925         * double-int.h (lshift): New overload without precision
17926         and arith argument.
17927         (operator *=, operator +=, operator -=): Move ...
17928         * double-int.c (operator *=, operator +=, operator -=): ... here
17929         and implement more efficiently.
17930         (mul_double_with_sign): Remove.
17931         (lshift_double): Adjust to take unsinged shift argument, push
17932         dispatching code to callers.
17933         (mul_double_wide_with_sign): Add early out for callers that
17934         are not interested in high parts or overflow.
17935         (lshift): New function.
17936         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
17937         dispatch code here.
17938         (lrotate, rrotate): Use logical shifts.
17939         * expr.c (get_inner_reference): Use lshift.
17940         * fixed-value.c (do_fixed_divide): Likewise.
17941         * tree-dfa.c (get_ref_base_and_extent): Likewise.
17942         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
17943         (indirect_refs_may_alias_p): Likewise.
17944         (stmt_kills_ref_p_1): Likewise.
17945
17946 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
17947
17948         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
17949
17950 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
17951
17952         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
17953         scalar form of FABD instruction.
17954
17955 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
17956
17957         * lra-constraints.c (process_alt_operands): Add checking alt
17958         number to choose the best alternative.
17959
17960 2013-05-02  Richard Biener  <rguenther@suse.de>
17961
17962         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
17963         bitmap and its handling.
17964         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
17965
17966 2013-05-02  Richard Biener  <rguenther@suse.de>
17967
17968         PR middle-end/57140
17969         * tree-inline.c (copy_loops): Properly handle removed loops.
17970         (copy_cfg_body): Mark destination loops for fixup if source
17971         loops needed fixup.
17972
17973 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
17974
17975         PR target/56732
17976         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
17977         generating simple_return for naked functions.
17978
17979 2013-05-02  Martin Jambor  <mjambor@suse.cz>
17980
17981         PR middle-end/56988
17982         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
17983         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
17984         flags match.
17985         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
17986         ipa_agg_replacement_value structures.
17987         (known_aggs_to_agg_replacement_list): Likewise.
17988         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
17989         (read_agg_replacement_chain): Likewise.
17990         (ipcp_transform_function): Also check that by_ref flags match.
17991
17992 2013-05-02  Richard Biener  <rguenther@suse.de>
17993
17994         * graphds.h (struct graph): Add obstack member.
17995         * graphds.c (new_graph): Initialize obstack and allocate
17996         vertices from it.
17997         (add_edge): Allocate edge from the obstack.
17998         (free_graph): Free the obstack instead of all edges and vertices.
17999
18000 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18001
18002         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18003         divides.
18004         * cfg.c (update_bb_profile_for_threading): Ditto.
18005         * tree-inline.c (copy_bb): Ditto.
18006         (copy_edges_for_bb): Ditto.
18007         (initialize_cfun): Ditto.
18008         (copy_cfg_body): Ditto.
18009         (expand_call_inline): Ditto.
18010         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18011         (estimate_node_size_and_time): Ditto.
18012         (inline_merge_summary): Ditto.
18013         * cgraphclones.c (cgraph_clone_edge): Ditto.
18014         (cgraph_clone_node): Ditto.
18015         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18016         (compute_trg_info): Ditto.
18017
18018 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18019
18020         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18021         S reg when fp attribute set.
18022         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18023
18024 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18025
18026         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18027         New pattern.
18028         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18029         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18030         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18031
18032 2013-05-02  Richard Biener  <rguenther@suse.de>
18033
18034         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18035         (struct instantiate_cache_entry): New type.
18036         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18037         (struct instantiate_cache_type): New type.
18038         (set_instantiated_value, get_instantiated_value): Remove.
18039         (get_instantiated_value_entry): New function.
18040         (instantiate_scev_name): Use the new cache and adjust.
18041         (instantiate_scev_poly): Adjust.
18042         (instantiate_scev_binary): Likewise.
18043         (instantiate_array_ref): Likewise.
18044         (instantiate_scev_convert): Likewise.
18045         (instantiate_scev_not): Likewise.
18046         (instantiate_scev_3): Likewise.
18047         (instantiate_scev_2): Likewise.
18048         (instantiate_scev_r): Likewise.
18049         (instantiate_scev): Likewise.
18050         (resolve_mixers): Likewise.
18051
18052 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18053
18054         PR target/57091
18055         * lra-constraints.c (best_small_class_operands_num): Remove.
18056         (process_alt_operands): Remove small_class_operands_num.  Take
18057         small classes operands into losers and only if the operand is not
18058         matched.  Modify debugging output.
18059         (curr_insn_transform): Remove best_small_class_operands_num.
18060         Print insn name.
18061
18062 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18063
18064         * config/aarch64/aarch64-builtins.c
18065         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18066         * config/aarch64/aarch64-simd-builtins.def
18067         (reduc_splus_): Add new modes.
18068         (reduc_uplus_): New.
18069         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18070         (reduc_uplus_v4sf): Likewise.
18071         (reduc_splus_v4sf): Likewise.
18072         (aarch64_addv<mode>): Likewise.
18073         (reduc_uplus_<mode>): Likewise.
18074         (reduc_splus_<mode>): Likewise.
18075         (aarch64_addvv2di): Likewise.
18076         (reduc_uplus_v2di): Likewise.
18077         (reduc_splus_v2di): Likewise.
18078         (aarch64_addvv2si): Likewise.
18079         (reduc_uplus_v2si): Likewise.
18080         (reduc_splus_v2si): Likewise.
18081         (reduc_<sur>plus_<mode>): New.
18082         (reduc_<sur>plus_v2di): Likewise.
18083         (reduc_<sur>plus_v2si): Likewise.
18084         (reduc_<sur>plus_v4sf): Likewise.
18085         (aarch64_addpv4sf): Likewise.
18086         * config/aarch64/arm_neon.h
18087         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18088         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18089         add UNSPEC_SADDV, UNSPEC_UADDV.
18090         (SUADDV): New.
18091         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18092
18093 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18094
18095         * config/aarch64/arm_neon.h
18096         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18097
18098 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18099
18100         * config/aarch64/aarch64-builtins
18101         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18102
18103 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18104
18105         * config/aarch64/aarch64-simd-builtins.def
18106         (reduc_smax_): New.
18107         (reduc_smin_): Likewise.
18108         (reduc_umax_): Likewise.
18109         (reduc_umin_): Likewise.
18110         (reduc_smax_nan_): Likewise.
18111         (reduc_smin_nan_): Likewise.
18112         (fmax): Remove.
18113         (fmin): Likewise.
18114         (smax): Update for V2SF, V4SF and V2DF modes.
18115         (smin): Likewise.
18116         (smax_nan): New.
18117         (smin_nan): Likewise.
18118         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18119         (<su><maxmin><mode>3): ...This, refactor.
18120         (s<maxmin><mode>3): New.
18121         (<maxmin_uns><mode>3): Likewise.
18122         (reduc_<maxmin_uns>_<mode>): Refactor.
18123         (reduc_<maxmin_uns>_v4sf): Likewise.
18124         (reduc_<maxmin_uns>_v2si): Likewise.
18125         (aarch64_<fmaxmin><mode>: Remove.
18126         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18127         new builtin names.
18128         (vmin<q>_f<32,64>): Likewise.
18129         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18130         (FMAXMIN): New.
18131         (su): Add mappings for smax, smin, umax, umin.
18132         (maxmin): New.
18133         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18134         (FMAXMIN): Rename as...
18135         (FMAXMIN_UNS): ...This.
18136         (maxminv): Remove.
18137         (fmaxminv): Likewise.
18138         (fmaxmin): Likewise.
18139         (maxmin_uns): New.
18140         (maxmin_uns_op): Likewise.
18141
18142 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18143
18144         * config/aarch64/arm_neon.h
18145         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18146         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18147         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18148
18149 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18150
18151         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18152         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18153
18154 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18155
18156         * config/aarch64/aarch64-simd.md
18157         (vcond<mode>_internal): Handle special cases for constant masks.
18158         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18159         (vcondu<mode><mode>): Likewise.
18160         (vcond<v_cmp_result><mode>): New.
18161
18162 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18163
18164         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18165         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18166         * config/aarch64/aarch64-simd-builtins.def
18167         (cmeq): Update to BUILTIN_VALLDI.
18168         (cmgt): Likewise.
18169         (cmge): Likewise.
18170         (cmle): Likewise.
18171         (cmlt): Likewise.
18172         * config/aarch64/arm_neon.h
18173         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18174         to builtins or C as appropriate.
18175
18176 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18177
18178         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18179         (cmgeu): ...This.
18180         (cmhi): Rename to...
18181         (cmgtu): ...This.
18182         * config/aarch64/aarch64-simd.md
18183         (simd_mode): Add SF.
18184         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18185         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18186         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18187         (cstore<mode>_neg): ...This.
18188         * config/aarch64/iterators.md
18189         (VALLF): new.
18190         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18191         (COMPARISONS): New.
18192         (UCOMPARISONS): Likewise.
18193         (optab): Add missing comparisons.
18194         (n_optab): New.
18195         (cmp_1): Likewise.
18196         (cmp_2): Likewise.
18197         (CMP): Likewise.
18198         (cmp): Remove.
18199         (VCMP_S): Likewise.
18200         (VCMP_U): Likewise.
18201         (V_cmp_result): Add DF, SF modes.
18202         (v_cmp_result): Likewise.
18203         (v): Likewise.
18204         (vmtype): Likewise.
18205         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18206
18207 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18208
18209         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18210         define_insn to define_insn_and_split.
18211         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18212         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18213         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18214         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18215         (thumb2_negscc): Likewise.
18216
18217 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18218
18219         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18220
18221 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18222
18223         * config/arm/thumb2.md: Remove trailing whitespaces.
18224
18225 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18226
18227         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18228         Use gen_int_mode rather than GEN_INT.
18229
18230 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18231
18232         * value-prof.c (stream_in_histogram_value): Remove the strayed
18233         debug_gimple_stmt.
18234
18235 2013-04-30  Richard Biener  <rguenther@suse.de>
18236
18237         PR middle-end/57122
18238         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18239
18240 2013-04-30  Richard Biener  <rguenther@suse.de>
18241
18242         PR middle-end/57107
18243         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18244
18245 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18246
18247         PR rtl-optimization/56957
18248         PR rtl-optimization/57105
18249         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18250         variable.  Use just INSN_UID for determining whether an insn
18251         should be only disconnected from the insn stream.
18252         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18253
18254 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18255
18256         PR tree-optimization/57104
18257         * tsan.c (instrument_expr): Don't instrument accesses to
18258         DECL_HARD_REGISTER VAR_DECLs.
18259
18260 2013-04-30  Richard Biener  <rguenther@suse.de>
18261
18262         * function.h (loops_for_fn): New inline function.
18263         (set_loops_for_fn): Likewise.
18264         * cfgloop.h (place_new_loop): Add struct function parameter.
18265         (get_loop): Likewise.
18266         (get_loops): Likewise.
18267         (number_of_loops): Likewise.
18268         (fel_next): Adjust.
18269         (fel_init): Likewise.
18270         * cfg.c (get_loop_copy): Adjust.
18271         * cfgloop.c (flow_loops_dump): Likewise.
18272         (record_loop_exits): Likewise.
18273         (verify_loop_structure): Likewise.
18274         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18275         (estimate_reg_pressure_cost): Likewise.
18276         (mark_loop_exit_edges): Likewise.
18277         * cfgloopmanip.c (place_new_loop): Likewise.
18278         (add_loop): Likewise.
18279         (duplicate_loop): Likewise.
18280         * graph.c (draw_cfg_nodes): Likewise.
18281         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18282         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18283         (extract_affine_chrec): Likewise.
18284         (build_scop_iteration_domain): Likewise.
18285         * graphite.c (graphite_initialize): Likewise.
18286         * ira-build.c (create_loop_tree_nodes): Likewise.
18287         (more_one_region_p): Likewise.
18288         (rebuild_regno_allocno_maps): Likewise.
18289         (mark_loops_for_removal): Likewise.
18290         (mark_all_loops_for_removal): Likewise.
18291         (remove_unnecessary_regions): Likewise.
18292         (ira_build): Likewise.
18293         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18294         * loop-init.c (fix_loop_structure): Likewise.
18295         (gate_rtl_move_loop_invariants): Likewise.
18296         (gate_rtl_unswitch): Likewise.
18297         (gate_rtl_unroll_and_peel_loops): Likewise.
18298         (rtl_doloop): Likewise.
18299         * lto-streamer-in.c (input_cfg): Likewise.
18300         * lto-streamer-out.c (output_cfg): Likewise.
18301         * modulo-sched.c (sms_schedule): Likewise.
18302         * predict.c (tree_estimate_probability): Likewise.
18303         (tree_estimate_probability_driver): Likewise.
18304         (estimate_loops): Likewise.
18305         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18306         (move_sese_region_to_fn): Likewise.
18307         (debug_loop_num): Likewise.
18308         * tree-chrec.c (chrec_evaluate): Likewise.
18309         (hide_evolution_in_other_loops_than_loop): Likewise.
18310         (chrec_component_in_loop_num): Likewise.
18311         (reset_evolution_in_loop): Likewise.
18312         (evolution_function_is_invariant_rec_p): Likewise.
18313         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18314         * tree-inline.c (copy_loops): Likewise.
18315         (copy_cfg_body): Likewise.
18316         (tree_function_versioning): Likewise.
18317         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18318         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18319         Likewise.
18320         (add_to_evolution_1): Likewise.
18321         (scev_const_prop): Likewise.
18322         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18323         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18324         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18325         (tree_ssa_lim_initialize): Likewise.
18326         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18327         (verify_loop_closed_ssa): Likewise.
18328         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18329         (tree_ssa_loop_im): Likewise.
18330         (tree_ssa_loop_unswitch): Likewise.
18331         (tree_vectorize): Likewise.
18332         (check_data_deps): Likewise.
18333         (tree_ssa_loop_ivcanon): Likewise.
18334         (tree_ssa_loop_bounds): Likewise.
18335         (tree_complete_unroll): Likewise.
18336         (tree_complete_unroll_inner): Likewise.
18337         (tree_parallelize_loops): Likewise.
18338         (tree_ssa_loop_prefetch): Likewise.
18339         (tree_ssa_loop_ivopts): Likewise.
18340         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18341         * tree-vectorizer.c (vectorize_loops): Likewise.
18342
18343 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18344
18345         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18346         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18347         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18348         with EABI_LINK_SPEC.
18349
18350 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18351
18352         PR target/44578
18353         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18354         alternative.
18355
18356 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18357
18358         PR target/57097
18359         * lra-constraints.c (process_alt_operands): Discourage a bit more
18360         using memory for pseudos.  Print cost dump for alternatives.
18361         Modify cost values for conflicts with early clobbers.
18362         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18363
18364 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18365
18366         PR target/57098
18367         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18368
18369 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18370
18371         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18372         from/to S register.
18373         (movdi_aarch64): Support LDR/STR from/to D register.
18374
18375 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18376
18377         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18378         or higher by default.
18379
18380 2013-04-29  Richard Biener  <rguenther@suse.de>
18381
18382         PR middle-end/57075
18383         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18384         even if not adding abnormal edges for calls that can make
18385         abnormal gotos.
18386
18387 2013-04-29  Richard Biener  <rguenther@suse.de>
18388
18389         PR middle-end/57103
18390         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18391         TREE_BLOCK.
18392         (move_stmt_r): Remove redundant checking.
18393
18394 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18395
18396         PR bootstrap/57077
18397         * basic-block.h (apply_scale): New function.
18398         (apply_probability): Use apply_scale.
18399         * gimple-streamer-in.c (input_bb): Ditto.
18400         * lto-streamer-in.c (input_cfg): Ditto.
18401         * lto-cgraph.c (merge_profile_summaries): Ditto.
18402         * tree-optimize.c (execute_fixup_cfg): Ditto.
18403         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18404         (copy_edges_for_bb): Ditto.
18405         (copy_cfg_body): Ditto.
18406
18407 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18408
18409         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18410         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18411         (tail_merge_optimize): Handle current_loops == NULL.
18412
18413 2013-04-26  Jeff Law  <law@redhat.com>
18414
18415         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18416         (simplify_cond_using_ranges): Generalize code to simplify
18417         COND_EXPRs where one argument is a constant and the other
18418         is an SSA_NAME created by an integral type conversion.
18419
18420 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18421
18422         * config/arm/arm.md (store_minmaxsi): Use only when
18423         optimize_insn_for_size_p.
18424
18425 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18426
18427         PR target/57108
18428         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18429
18430 2013-04-29  Richard Biener  <rguenther@suse.de>
18431
18432         PR middle-end/57089
18433         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18434         loop tree make sure to schedule a fixup for the child as well.
18435         (expand_omp_for_generic): Properly add loops.
18436         (expand_omp_for_static_nochunk): Likewise.
18437         (expand_omp_for_static_chunk): Likewise.
18438         (expand_omp_for): For the degenerate case fixup loops.
18439         (expand_omp_sections): Fix default bb placement in loops.
18440         (expand_omp_atomic_pipeline): Properly add loops.
18441
18442 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18443
18444         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18445
18446 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18447
18448         * tree-ssa-tail-merge.c: Update header comment.
18449
18450 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18451
18452         * config/aarch64/arm_neon.h
18453         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18454         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18455         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18456         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18457         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18458         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18459         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18460         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18461
18462 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18463
18464         * config/aarch64/aarch64-simd.md
18465         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18466         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18467         fix_trunc, fixuns_trunc.
18468         (ftrunc<VDQF:mode>2): New.
18469         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18470         (fix_trunc_optab): New.
18471
18472 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18473
18474         * config/aarch64/aarch64-builtins.c
18475         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18476         iceilf, lround, iroundf.
18477
18478 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18479
18480         PR target/54349
18481         * config/i386/i386.h (enum ix86_tune_indices)
18482         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18483         New, split from X86_TUNE_INTER_UNIT_MOVES.
18484         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18485         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18486         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18487         (TARGET_INTER_UNIT_MOVES): Remove.
18488         * config/i386/i386.c (initial_ix86_tune_features): Update.
18489         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18490         (ix86_expand_convert_uns_didf_sse): Use
18491         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18492         (ix86_expand_vector_init_one_nonzero): Ditto.
18493         (ix86_expand_vector_init_interleave): Ditto.
18494         (inline_secondary_memory_needed): Return true for moves from SSE class
18495         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18496         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18497         * config/i386/constraints.md (Yi, Ym): Depend on
18498         TARGET_INTER_UNIT_MOVES_TO_VEC.
18499         (Yj, Yn): New constraints.
18500         * config/i386/i386.md (*movdi_internal): Change constraints of
18501         operand 1 from Yi to Yj and from Ym to Yn.
18502         (*movsi_internal): Ditto.
18503         (*movdf_internal): Ditto.
18504         (*movsf_internal): Ditto.
18505         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18506         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18507         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18508         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18509         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18510         * config/i386/sse.md (movdi_to_sse): Ditto.
18511         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18512         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18513         TARGET_INTER_UNIT_MOVES.
18514         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18515         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18516         instead of TARGET_INTER_UNIT_MOVES.
18517         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18518         operand 1 from Yi to Yj and from Ym to Yn.
18519
18520 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18521
18522         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18523         (float_truncate_hi_): Likewise.
18524         (float_extend_lo_): Likewise.
18525         (float_truncate_lo_): Likewise.
18526         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18527         (aarch64_float_extend_lo_v2df): Likewise.
18528         (vec_unpacks_hi_v4sf): Likewise.
18529         (aarch64_float_truncate_lo_v2sf): Likewise.
18530         (aarch64_float_truncate_hi_v4sf): Likewise.
18531         (vec_pack_trunc_v2df): Likewise.
18532         (vec_pack_trunc_df): Likewise.
18533
18534 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18535
18536         * config/aarch64/aarch64-builtins.c
18537         (aarch64_fold_builtin): Fold float conversions.
18538         * config/aarch64/aarch64-simd-builtins.def
18539         (floatv2si, floatv4si, floatv2di): New.
18540         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18541         * config/aarch64/aarch64-simd.md
18542         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18543         * config/aarch64/iterators.md (FLOATUORS): New.
18544         (optab): Add float, floatuns.
18545         (su_optab): Likewise.
18546
18547 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18548
18549         * config/aarch64/aarch64-builtins.c
18550         (aarch64_builtin_vectorized_function): Use new names for
18551         fcvt builtins.
18552         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18553         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18554         (fcvtzu): Split as...
18555         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18556         (fcvtas): Split as...
18557         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18558         (fcvtau): Split as...
18559         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18560         (fcvtps): Split as...
18561         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18562         (fcvtpu): Split as...
18563         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18564         (fcvtms): Split as...
18565         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18566         (fcvtmu): Split as...
18567         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18568         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18569         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18570         (lfrintnusf, lfrintnudf): Likewise.
18571         * config/aarch64/aarch64-simd.md
18572         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18573         define_insn.
18574         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18575         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18576         (fcvt_pattern): Likewise.
18577
18578 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18579
18580         * config/aarch64/aarch64-simd.md
18581         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18582         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18583
18584 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18585
18586         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18587         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18588         (vrnd<a,m,n,p>_f32): Implement using builtins.
18589         (vrnd<i,x><q>_f<32, 64>): New.
18590
18591 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18592
18593         * config/aarch64/aarch64-builtins.c
18594         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18595         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18596         (frintz): Rename to...
18597         (btrunc): ...this.
18598         (frintp): Rename to...
18599         (ceil): ...this.
18600         (frintm): Rename to...
18601         (floor): ...this.
18602         (frinti): Rename to...
18603         (nearbyint): ...this.
18604         (frintx): Rename to...
18605         (rint): ...this.
18606         (frinta): Rename to...
18607         (round): ...this.
18608         * config/aarch64/aarch64-simd.md
18609         (aarch64_frint<frint_suffix><mode>): Delete.
18610         (<frint_pattern><mode>2): Convert to insn.
18611         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18612         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18613         (frint_pattern): Likewise.
18614         (frint_suffix): Likewise.
18615
18616 2013-04-29  Richard Biener  <rguenther@suse.de>
18617
18618         PR tree-optimization/57081
18619         * loop-init.c: Include tree-flow.h.
18620         (loop_optimizer_finalize): Free number of iteration estimates.
18621         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18622
18623 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18624
18625         PR tree-optimization/57083
18626         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18627         non-singleton shift count range, zero extend low_bound for uns case.
18628
18629         * config/i386/predicates.md (general_vector_operand): New predicate.
18630         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18631         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18632         if they aren't nonimmediate operands.  If their original values
18633         satisfy const_vector_equal_evenodd_p, don't shift them.
18634         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18635         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18636         if not nonimmediate_operand.
18637         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18638         instead of register_operand.
18639         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18640
18641 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18642
18643         * stor-layout.c (finalize_size_functions): Allocate a structure and
18644         reset cfun before dumping the functions.
18645
18646 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18647
18648         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18649
18650         PR target/56866
18651         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18652         use xop_pmacsdqh if uns_p.
18653         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18654         the immediate rotate count.
18655
18656 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18657
18658         * rtl.h (struct rtx_def): Add comment for field jump.
18659         (LRA_SUBREG_P): New macro.
18660         * recog.c (register_operand): Check LRA_SUBREG_P.
18661         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18662         stack frame.
18663         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18664         (lra_final_code_change): Skip subreg change for operators.
18665         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18666         if there are no operand changes.
18667         * lra-constraints.c (curr_insn_set): New.
18668         (match_reload): Set LRA_SUBREG_P.
18669         (emit_spill_move): Ditto.
18670         (check_and_process_move): Use curr_insn_set. Process only single
18671         set insns.  Don't initialize sec_mem_p and change_p.
18672         (simplify_operand_subreg): Use LRA_SUBREG_P.
18673         (reg_in_class_p): New function.
18674         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18675         of #ifdef.  Add code to remove cycling.
18676         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18677         non-null disp.  Reload inner instead of disp when base and index
18678         are null.  Try to put lo_sum into register.
18679         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18680         (check_and_process_move): Move code for move cost check to
18681         simple_move_p.  Remove equiv_substitution.
18682         (simple_move_p): New function.
18683         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18684         curr_insn_set.  Call check_and_process_move only for single set
18685         insns.  Use the new function.  Move call of check_and_process_move
18686         after operand equiv substitution and address process.
18687
18688 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18689
18690         PR go/57045
18691         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18692         with nonlocal goto receivers or returns twice calls, ignore
18693         unininitialized values from abnormal edges to nl goto receiver
18694         or returns twice call.
18695
18696 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18697
18698         PR tree-optimization/57051
18699         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18700         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18701         bitsize.
18702
18703 2013-04-26  Richard Biener  <rguenther@suse.de>
18704
18705         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18706         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18707         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18708         (fixup_loop_arrays_after_move): New function.
18709         (move_sese_region_to_fn): Properly outline the loop tree parts
18710         of the SESE region.
18711
18712 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18713
18714         * config/i386/i386.md (type, unit): Fix long lines.
18715
18716 2013-04-26  Richard Biener  <rguenther@suse.de>
18717
18718         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
18719         (lto-streamer-out.o): Likewise.
18720         * cfgloop.c (init_loops_structure): Export, add struct function
18721         argument and adjust.
18722         (flow_loops_find): Adjust.
18723         * cfgloop.h (enum loop_estimation): Add EST_LAST.
18724         (init_loops_structure): Declare.
18725         * lto-streamer-in.c: Include cfgloop.h.
18726         (input_cfg): Input the loop tree.
18727         * lto-streamer-out.c: Include cfgloop.h.
18728         (output_cfg): Output the loop tree.
18729         (output_struct_function_base): Do not drop PROP_loops.
18730
18731 2013-03-26  Richard Biener  <rguenther@suse.de>
18732
18733         * tree-cfg.c (execute_build_cfg): Build the loop tree.
18734         (pass_build_cfg): Provide PROP_loops.
18735         (move_sese_region_to_fn): Remove loops that are outlined into fn
18736         for now.
18737         * tree-inline.c: Include cfgloop.h.
18738         (initialize_cfun): Do not drop PROP_loops.
18739         (copy_loops): New function.
18740         (copy_cfg_body): Copy loop structure.
18741         (tree_function_versioning): Initialize destination loop tree.
18742         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
18743         (pass_parallelize_loops): Do IL verification.
18744         * loop-init.c (loop_optimizer_init): Fixup loops if required.
18745         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
18746         the CFG make sure we fixup loops as well.
18747         * tree-ssa-tail-merge.c: Include cfgloop.h.
18748         (replace_block_by): When merging loop latches mark loops for fixup.
18749         * lto-streamer-out.c (output_struct_function_base): Drop
18750         PROP_loops for now.
18751         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
18752         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
18753         * ipa-split.c: Include cfgloop.h.
18754         (split_function): Add the new return block to the loop tree root.
18755         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
18756         whether we have removed the forwarder block.
18757         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
18758         * cfgloop.h (place_new_loop): Declare.
18759         * cfgloopmanip.c (place_new_loop): Export.
18760         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
18761         (tree-switch-conversion.o): Likewise.
18762         (tree-complex.o): Likewise.
18763         (tree-inline.o): Likewise.
18764         (tree-ssa-tailmerge.o): Likewise.
18765         (ipa-split.o): Likewise.
18766         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
18767         (tree-ssa-copy.o): Likewise.
18768         * tree-switch-conversion.c: Include cfgloop.h
18769         (process_switch): If we emit a bit-test cascade, schedule loops
18770         for fixup.
18771         * tree-complex.c: Include cfgloop.h.
18772         (expand_complex_div_wide): Properly add new basic-blocks to loops.
18773         * asan.c: Include cfgloop.h.
18774         (create_cond_insert_point): Properly add new basic-blocks to
18775         loops, schedule loop fixup.
18776         * cfgloop.c (verify_loop_structure): Check that looks are not
18777         marked for fixup.
18778         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
18779         to loops.
18780         (expand_omp_for_generic): Likewise.
18781         (expand_omp_sections): Likewise.
18782         (expand_omp_atomic_pipeline): Schedule loops for fixup.
18783         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
18784         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
18785         is initialized, not when loops are present.
18786         * tree-parloops.c (parallelize_loops): Remove checking here.
18787         * passes.c (init_optimization_passes): Schedule a copy-propagation
18788         pass before complete unrolling of inner loops.
18789
18790 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18791
18792         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
18793         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
18794         (colorize_init): Add argument to _WIN32 version.
18795         * toplev.c: Include diagnostic-color.h.
18796         (process_options): Default to -fdiagnostics-color=auto if
18797         GCC_COLORS env var is in the environment.
18798         * common.opt (fdiagnostics-color=): Add Var and Init.
18799         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
18800         env var is in the environment, the default is auto rather than never.
18801
18802         * diagnostic.h (file_name_as_prefix): Add context argument.
18803         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
18804         the string as locus.
18805         * langhooks.c (lhd_print_error_function): Adjust caller.
18806
18807 2013-04-25  Lawrence Crowl  <crowl@google.com>
18808
18809         * var-tracking.c (shared_hash_def::htab):
18810         Change type to hash_table.  Update dependent calls and types.
18811
18812 2013-04-25  Lawrence Crowl  <crowl@google.com>
18813
18814         * Makefile.in: Update as needed below.
18815
18816         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
18817         Move declaration to after the type's method definitons.
18818
18819         * attribs.c (htab_t scoped_attributes::attribute_hash):
18820         Change type to hash_table.  Update dependent calls and types.
18821
18822         * bitmap.c (htab_t bitmap_desc_hash):
18823         Change type to hash_table.  Update dependent calls and types.
18824
18825         * cselib.c (htab_t cselib_hash_table):
18826         Change type to hash_table.  Update dependent calls and types.
18827
18828         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
18829         (hash_string_slot_node): Move implementation into lto-streamer.h
18830         struct string_slot_hasher.
18831         (eq_string_slot_node): Likewise.
18832
18833         * data-streamer-out.c: Update output_block::string_hash_table
18834         dependent calls and types.
18835
18836         * dwarf2cfi.c (htab_t trace_index):
18837         Change type to hash_table.  Update dependent calls and types.
18838
18839         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
18840         Change type to hash_table.  Update dependent calls and types.
18841         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
18842         (htab_t optimize_external_refs::map): Likewise.
18843         (htab_t output_comp_unit::extern_map): Likewise.
18844         (htab_t output_comdat_type_unit::extern_map): Likewise.
18845         (htab_t output_macinfo::macinfo_htab): Likewise.
18846         (htab_t optimize_location_lists::htab): Likewise.
18847         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
18848
18849         * except.c (htab_t ehspec_hash_type):
18850         Change type to hash_table.  Update dependent calls and types.
18851         (assign_filter_values::ttypes): Likewise.
18852         (assign_filter_values::ehspec): Likewise.
18853         (sjlj_assign_call_site_values::ar_hash): Likewise.
18854         (convert_to_eh_region_ranges::ar_hash): Likewise.
18855
18856         * gcse.c (htab_t pre_ldst_table):
18857         Change type to hash_table.  Update dependent calls and types.
18858
18859         * ggc-common.c (htab_t saving_htab):
18860         Change type to hash_table.  Update dependent calls and types.
18861         (htab_t loc_hash): Likewise.
18862         (htab_t ptr_hash): Likewise.
18863         (call_count): Rename ggc_call_count.
18864         (call_alloc): Rename ggc_call_alloc.
18865         (loc_descriptor): Rename make_loc_descriptor.
18866         (add_statistics): Rename ggc_add_statistics.
18867
18868         * ggc-common.c (saving_htab):
18869         Change type to hash_table.  Update dependent calls and types.
18870
18871         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
18872         (push_gimplify_context): Likewise.
18873         (pop_gimplify_context): Likewise.
18874         (struct gimple_temp_hash_elt): Added.
18875         (struct gimplify_hasher): Likewise.
18876         (struct gimplify_ctx.temp_htab):
18877         Change type to hash_table.  Update dependent calls and types.
18878
18879         * gimple-fold.c: Include gimplify-ctx.h.
18880
18881         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
18882         Change type to hash_table.  Update dependent calls and types.
18883         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
18884         avoid potential global name collision.
18885
18886         * gimplify.c: Include gimplify-ctx.h.
18887         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
18888         (htab_t gimplify_ctx::temp_htab):
18889         Update dependent calls and types for new type hash_table.
18890         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
18891         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
18892
18893         * gimplify-ctx.h: New.
18894         (struct gimple_temp_hash_elt): Move from gimplify.c.
18895         (class gimplify_hasher): New.
18896         (struct gimplify_ctx): Move from gimple.h.
18897         (htab_t gimplify_ctx::temp_htab):
18898         Change type to hash_table.  Update dependent calls and types.
18899
18900         * graphite-clast-to-gimple.c: Include graphite-htab.h.
18901         (htab_t ivs_params::newivs_index):
18902         Change type to hash_table.  Update dependent calls and types.
18903         (htab_t ivs_params::params_index): Likewise.
18904         (htab_t print_generated_program::params_index): Likewise.
18905         (htab_t gloog::newivs_index): Likewise.
18906         (htab_t gloog::params_index): Likewise.
18907
18908         * graphite.c: Include graphite-htab.h.
18909         4htab_t graphite_transform_loops::bb_pbb_mapping):
18910         Change type to hash_table.  Update dependent calls and types.
18911
18912         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
18913         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
18914         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
18915
18916         * graphite-dependences.c: Include graphite-htab.h.
18917         (loop_is_parallel_p): Change hash table type of parameter.
18918
18919         * graphite-htab.h: New.
18920         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
18921         (extern find_pbb_via_hash): Move from graphite-poly.h.
18922         (extern loop_is_parallel_p): Move from graphite-poly.h.
18923         (extern get_loop_body_pbbs): Move from graphite-poly.h.
18924
18925         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
18926         (extern loop_is_parallel_p): Move to graphite-htab.h.
18927         (extern get_loop_body_pbbs): Move to graphite-htab.h.
18928
18929         * haifa-sched.c (htab_t delay_htab):
18930         Change type to hash_table.  Update dependent calls and types.
18931         (htab_t delay_htab_i2): Likewise.
18932
18933         * ira-color.c (htab_t allocno_hard_regs_htab):
18934         Change type to hash_table.  Update dependent calls and types.
18935
18936         * ira-costs.c (htab_t cost_classes_htab):
18937         Change type to hash_table.  Update dependent calls and types.
18938
18939         * loop-invariant.c (htab_t merge_identical_invariants::eq):
18940         Change type to hash_table.  Update dependent calls and types.
18941
18942         * loop-iv.c (htab_t bivs):
18943         Change type to hash_table.  Update dependent calls and types.
18944
18945         * loop-unroll.c (htab_t opt_info::insns_to_split):
18946         Change type to hash_table.  Update dependent calls and types.
18947         (htab_t opt_info::insns_with_var_to_expand): Likewise.
18948
18949         * lto-streamer.h (struct string_slot): Move from data-streamer.h
18950         (struct string_slot_hasher): New.
18951         (htab_t output_block::string_hash_table):
18952         Change type to hash_table.  Update dependent calls and types.
18953
18954         * lto-streamer-in.c (freeing_string_slot_hasher): New.
18955         (htab_t file_name_hash_table):
18956         Change type to hash_table.  Update dependent calls and types.
18957
18958         * lto-streamer-out.c: Update output_block::string_hash_table dependent
18959         calls and types.
18960
18961         * lto-streamer.c (htab_t tree_htab):
18962         Change type to hash_table.  Update dependent calls and types.
18963
18964         * omp-low.c: Include gimplify-ctx.h.
18965
18966         * passes.c (htab_t name_to_pass_map):
18967         Change type to hash_table.  Update dependent calls and types.
18968         (pass_traverse): Rename to passes_pass_traverse.
18969
18970         * plugin.c (htab_t event_tab):
18971         Change type to hash_table.  Update dependent calls and types.
18972
18973         * postreload-gcse.c (htab_t expr_table):
18974         Change type to hash_table.  Update dependent calls and types.
18975         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
18976
18977         * sese.c (debug_rename_map_1): Make extern.
18978         (htab_t copy_bb_and_scalar_dependences::rename_map):
18979         Change type to hash_table.  Update dependent calls and types.
18980
18981         * sese.h (extern debug_rename_map): Move to .c file.
18982
18983         * store-motion.c (htab_t store_motion_mems_table):
18984         Change type to hash_table.  Update dependent calls and types.
18985
18986         * trans-mem.c (htab_t tm_new_mem_hash):
18987         Change type to hash_table.  Update dependent calls and types.
18988
18989         * tree-browser.c (htab_t TB_up_ht):
18990         Change type to hash_table.  Update dependent calls and types.
18991
18992         * tree-cfg.c (htab_t discriminator_per_locus):
18993         Change type to hash_table.  Update dependent calls and types.
18994
18995         * tree-complex.c: Include tree-hasher.h
18996         (htab_t complex_variable_components):
18997         Change type to hash_table.  Update dependent calls and types.
18998
18999         * tree-eh.c (htab_t finally_tree):
19000         Change type to hash_table.  Update dependent calls and types.
19001
19002         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19003         struct int_tree_hasher.
19004         (extern int_tree_map_eq): Likewise.
19005         (uid_decl_map_hash): Removed.
19006         (extern decl_tree_map_eq): Likewise.
19007
19008         * tree-hasher.h: New.
19009         (struct int_tree_hasher): New.
19010         (typedef int_tree_htab_type): New.
19011
19012         * tree-inline.c: Include gimplify-ctx.h.
19013
19014         * tree-mudflap.c: Include gimplify-ctx.h.
19015
19016         * tree-parloops.c: Include tree-hasher.h.
19017         (htab_t eliminate_local_variables_stmt::decl_address):
19018         Change type to hash_table.  Update dependent calls and types.
19019         (htab_t separate_decls_in_region::decl_copies): Likewise.
19020
19021         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19022         Change type to hash_table.  Update dependent calls and types.
19023
19024         * tree-sra.c (candidates):
19025         Change type to hash_table.  Update dependent calls and types.
19026
19027         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19028         in tree-flow.h.
19029         (int_tree_map_hash): Likewise.
19030
19031         * tree-ssa-dom.c (htab_t avail_exprs):
19032         Change type to hash_table.  Update dependent calls and types.
19033
19034         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19035         Change type to hash_table.  Update dependent calls and types.
19036
19037         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19038         Change type to hash_table.  Update dependent calls and types.
19039
19040         * tree-ssa-phiopt.c (seen_ssa_names):
19041         Change type to hash_table.  Update dependent calls and types.
19042
19043         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19044         Change type to hash_table.  Update dependent calls and types.
19045
19046         * tree-ssa-uncprop.c (equiv):
19047         Change type to hash_table.  Update dependent calls and types.
19048
19049 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19050
19051         PR rtl-optimization/57003
19052         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19053         call note_stores with kill_clobbered_value callback again after
19054         killing regs_invalidated_by_call.
19055
19056 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19057
19058         * config/aarch64/aarch64-simd.md
19059         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19060         (aarch64_simd_bsl<mode>): Likewise.
19061         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19062
19063 2013-04-25  Marek Polacek  <polacek@redhat.com>
19064
19065         PR tree-optimization/57066
19066         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19067
19068 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19069
19070         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19071
19072 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19073
19074         * config/aarch64/aarch64-builtins.c
19075         (aarch64_fold_builtin): New.
19076         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19077         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19078         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19079         * config/aarch64/arm_neon.h
19080         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19081
19082 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19083             Tejas Belagod  <tejas.belagod@arm.com>
19084
19085         * config/aarch64/aarch64-builtins.c
19086         (aarch64_gimple_fold_builtin): New.
19087         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19088         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19089         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19090         (addvv4sf): Update.
19091         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19092
19093 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19094
19095         * config/aarch64/aarch64.md
19096         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19097
19098 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19099
19100         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19101         (*ngcsi_uxtw): New pattern.
19102
19103 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19104             Julian Brown  <julian@codesourcery.com>
19105
19106         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19107         (TB_DREG): Add T_V4HF.
19108         (v4hf_UP): New macro.
19109         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19110         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19111         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19112         built-ins.
19113         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19114         (arm_vector_mode_supported_p): Handle V4HF.
19115         (arm_mangle_map): Handle V4HFmode.
19116         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19117         * config/arm/arm_neon_builtins.def: Add entries for
19118         vcvtv4hfv4sf, vcvtv4sfv4hf.
19119         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19120         (neon_vcvtv4hfv4sf): Likewise.
19121         * config/arm/neon-gen.ml: Handle half-precision floating point
19122         features.
19123         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19124         * config/arm/arm_neon.h: Regenerate.
19125         * config/arm/neon.ml (type elts): Add F16.
19126         (type vectype): Add T_float16x4, T_floatHF.
19127         (type vecmode): Add V4HF.
19128         (type features): Add Requires_FP_bit feature.
19129         (elt_width): Handle F16.
19130         (elt_class): Likewise.
19131         (elt_of_class_width): Likewise.
19132         (mode_of_elt): Refactor.
19133         (type_for_elt): Handle F16, fix error messages.
19134         (vectype_size): Handle T_float16x4.
19135         (vcvt_sh): New function.
19136         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19137         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19138         (string_of_mode): Handle V4HF.
19139         * doc/arm-neon-intrinsics.texi: Regenerate.
19140
19141 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19142
19143         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19144         format specifier in 'X' case.
19145
19146 2013-04-25  Alan Modra  <amodra@gmail.com>
19147
19148         PR target/57052
19149         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19150         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19151         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19152         Repeat for many other rotate/shift and mask patterns using subregs.
19153         Name lshiftrt insns.
19154         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19155         on WORDS_BIG_ENDIAN.
19156
19157 2013-04-25  Alan Modra  <amodra@gmail.com>
19158
19159         * config.gcc: Support little-endian powerpc-linux targets.
19160         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19161         (LINK_OS_LINUX_SPEC): Define.
19162         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19163         Preserve MASK_LITTLE_ENDIAN.
19164         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19165         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19166         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19167         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19168         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19169         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19170         Correct fp word order for little-endian.  Don't shift toc entries
19171         smaller than a word for little-endian.
19172         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19173         (bswapdi2 splits): Correct low-part subreg for little-endian.
19174         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19175         low/high where such is correct only for be.
19176         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19177         little-endian for -mcall-aixdesc.
19178
19179 2013-04-25  Alan Modra  <amodra@gmail.com>
19180
19181         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19182         replace_equiv_address_nv.
19183
19184 2013-04-25  Alan Modra  <amodra@gmail.com>
19185
19186         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19187
19188 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19189
19190         Revert:
19191         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19192         * rtl.h (struct rtx_def): ...
19193
19194 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19195
19196         PR rtl-optimizations/57046
19197         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19198         for multi-reg splits.
19199
19200 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19201
19202         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19203
19204 2013-04-24  Sterling Augustine  <saugustine@google.com>
19205
19206         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19207         (comp_dir_string, debug_str_dwo_section): New.
19208         (DEBUG_STR_DWO_SECTION): Rename to ...
19209         (DEBUG_DWO_STR_SECTION): ... this.
19210         (DEBUG_NORM_STR_SECTION): Delete.
19211         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19212         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19213         (find_AT_string): Move most logic to ...
19214         (find_AT_string_in_table): ... here.  New.
19215         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19216         add_skeleton_AT_string.  Delete logic.
19217         (output_skeleton_debug_sections): Remove call to
19218         add_top_level_skeleton_die_attrs.
19219         (add_comp_dir_attribute): Move logic to comp_dir_string.
19220         (dwarf2out_init): Initialize debug_str_dwo_section.
19221         (output_indirect_string): Call find_string_form.
19222         (output_indirect_strings): Rewrite.
19223         (prune_unused_types): Empty skeleton_debug_str_hash.
19224         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19225         (dwarf2out_finish):  Call output_indirect_strings.
19226
19227 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19228
19229         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19230
19231 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19232
19233         * rtl.h (struct rtx_def): Add comment for field jump.
19234         (LRA_SUBREG_P): New macro.
19235         * recog.c (register_operand): Check LRA_SUBREG_P.
19236         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19237         stack frame.
19238         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19239         (lra_final_code_change): Skip subreg change for operators.
19240         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19241         if there are no operand changes.
19242         * lra-constraints.c (curr_insn_set): New.
19243         (match_reload): Set LRA_SUBREG_P.
19244         (emit_spill_move): Ditto.
19245         (check_and_process_move): Use curr_insn_set. Process only single
19246         set insns.  Don't initialize sec_mem_p and change_p.
19247         (simplify_operand_subreg): Use LRA_SUBREG_P.
19248         (reg_in_class_p): New function.
19249         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19250         of #ifdef.  Add code to remove cycling.
19251         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19252         non-null disp.  Reload inner instead of disp when base and index
19253         are null.  Try to put lo_sum into register.
19254         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19255         (check_and_process_move): Move code for move cost check to
19256         simple_move_p.  Remove equiv_substitution.
19257         (simple_move_p): New function.
19258         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19259         curr_insn_set.  Call check_and_process_move only for single set
19260         insns.  Use the new function.  Move call of check_and_process_move
19261         after operand equiv substitution and address process.
19262
19263 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19264
19265         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19266         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19267         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19268
19269 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19270
19271         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19272
19273 2013-04-24  Marek Polacek  <polacek@redhat.com>
19274
19275         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19276         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19277         (select_loops_exit_conditions): Likewise.
19278         (number_of_iterations_for_all_loops): Likewise.
19279         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19280         (scev_analysis): Likewise.
19281
19282 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19283             Chao-ying Fu  <fu@mips.com>
19284
19285         * config/mips/micromips.md (jraddiusp): New pattern.
19286         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19287         instruction if possible.
19288
19289 2013-04-24  Alan Modra  <amodra@gmail.com>
19290
19291         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19292
19293 2013-04-24  Julian Brown  <julian@codesourcery.com>
19294             Chung-Lin Tang  <cltang@codesourcery.com>
19295
19296         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19297         dependency behavior in enumeration type DIE generation. Add TODO note
19298         to comments about future DW_FORM_sdata/udata re-work of related code.
19299
19300 2013-04-23  Lawrence Crowl  <crowl@google.com>
19301
19302         * Makefile.in: Update as needed below.
19303
19304         * hash-table.h (class hash_table):
19305         Correct many methods with parameter types compare_type to the correct
19306         value_type.  (Correct code was unlikely to notice the change.)
19307         (hash_table::elements_with_deleted) New.
19308         (class hashtable::iterator): New.
19309         (hashtable::begin()): New.
19310         (hashtable::end()): New.
19311         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19312
19313         * statistics.c (statistics_hashes):
19314         Change type to hash_table.  Update dependent calls and types.
19315
19316         * tree-into-ssa.c (var_infos):
19317         Change type to hash_table.  Update dependent calls and types.
19318
19319         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19320         Change type to hash_table.  Update dependent calls and types.
19321
19322         * tree-ssa-loop-im.c (struct mem_ref.refs):
19323         Change type to hash_table.  Update dependent calls and types.
19324
19325         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19326         Change type to hash_table.  Update dependent calls and types.
19327
19328         * tree-ssa-sccvn.c (vn_tables_s::nary):
19329         Change type to hash_table.  Update dependent calls and types.
19330         (vn_tables_s::phis): Likewise.
19331         (vn_tables_s::references): Likewise.
19332
19333         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19334         (vn_reference_eq): Update parameter and return types.
19335
19336         * tree-ssa-structalias.c (pointer_equiv_class_table):
19337         Change type to hash_table.  Update dependent calls and types.
19338         (location_equiv_class_table): Likewise.
19339
19340         * tree-vect-data-refs.c: Consequential changes for making
19341         peeling a hash_table.
19342
19343         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19344         (destroy_loop_vec_info): Dependent hash_table update.
19345
19346         * tree-vectorizer.h (peeling_htab):
19347         Change type to hash_table.  Update dependent calls and types.
19348
19349 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19350
19351         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19352         to check the register content is equal or not.
19353         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19354         to assign register content record.
19355         * lra-eliminations.c (update_reg_eliminate): Use
19356         lra_update_reg_val_offset to update register content offset.
19357         * lra-int.h (struct lra_reg): Add offset member.
19358         (lra_reg_val_equal_p): New static inline function.
19359         (lra_update_reg_val_offset): New static inline function.
19360         (lra_assign_reg_val): New static inline function.
19361         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19362         to assign register content record.
19363         (initialize_lra_reg_info_element): Initial offset to zero.
19364
19365 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19366
19367         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19368         operands.  Record compression.
19369
19370 2013-04-23  Xinliang David Li  <davidxl@google.com>
19371
19372         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19373
19374 2013-04-23  Richard Biener  <rguenther@suse.de>
19375
19376         PR middle-end/57036
19377         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19378         parameter, only add abnormal goto edges from the copied body
19379         if the call could perform abnormal gotos.
19380         (copy_cfg_body): Adjust.
19381
19382 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19383
19384         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19385
19386 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19387
19388         * coretypes.h (gimple_stmt_iterator): Add struct to make
19389         compatible with C.
19390
19391 2013-04-23  Richard Biener  <rguenther@suse.de>
19392
19393         PR tree-optimization/57026
19394         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19395         from SSA names occuring in abnormal PHI nodes.
19396
19397 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19398
19399         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19400
19401 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19402
19403         * lto/lto.c (print_lto_report_1): Declare early.
19404         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19405
19406 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19407
19408         * common.opt (-flto-report-wpa): Add.
19409         * doc/invoke.texi (-flto-report-wpa): Add.
19410         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19411         (lto_main): dito.
19412
19413 2013-04-22  Xinliang David Li  <davidxl@google.com>
19414
19415         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19416         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19417         * Makefile.in: New dependency
19418
19419             David Daney <ddaney.cavm@gmail.com>
19420
19421         * configure.ac (gcc_cv_as_micromips_support):  Use the
19422         --fatal-warnings option.
19423         * configure: Regenerate.
19424
19425 2013-04-22  Marek Polacek  <polacek@redhat.com>
19426
19427         PR sanitizer/56990
19428         * tsan.c (instrument_expr): Don't instrument expression
19429         in case its size is zero.
19430
19431 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19432
19433         PR target/57032
19434         Revert:
19435         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19436
19437         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19438
19439 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19440
19441         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19442         (gimple_stmt_iterator): New typedef.
19443         * gimple.h (gimple_stmt_iterator): Rename to...
19444         (gimple_stmt_iterator_d): ... This.
19445         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19446         trees be valid for GIMPLE and GENERIC.
19447         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19448         * gimple-fold.c (gimple_fold_call): Call target hook
19449         gimple_fold_builtin.
19450         * hooks.c (hook_bool_gsiptr_false): New.
19451         * hooks.h (hook_bool_gsiptr_false): New.
19452         * target.def (fold_stmt): New.
19453         * doc/tm.texi: Regenerate.
19454
19455 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19456
19457         PR target/57018
19458         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19459         a set sp if no stack realignment.
19460
19461 2013-04-22  Nick Clifton  <nickc@redhat.com>
19462
19463         * config.gcc (tilegx-linux): Extend extra_objs rather than
19464         overwriting it.
19465         (tilepro-linux): Likewise.
19466
19467 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19468
19469         * config/aarch64/aarch64-builtins.c
19470         (CF): Remove.
19471         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19472         (VAR<1-12>): Add MAP parameter.
19473         (BUILTIN_*): Likewise.
19474         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19475         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19476         (aarch64_ushl_n<mode>): Likewise.
19477         (aarch64_sshr_n<mode>): Likewise.
19478         (aarch64_ushr_n<mode>): Likewise.
19479         (aarch64_<maxmin><mode>): Likewise.
19480         (aarch64_sqrt<mode>): Likewise.
19481         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19482         (vshr<q>_n_*): Likewise.
19483
19484 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19485
19486         * config/aarch64/aarch64-builtins.c
19487         (aarch64_simd_builtin_type_mode): Handle SF types.
19488         (sf_UP): Define.
19489         (BUILTIN_GPF): Define.
19490         (aarch64_init_simd_builtins): Handle SF types.
19491         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19492         (frecps): Likewise.
19493         (frecpx): Likewise.
19494         * config/aarch64/aarch64-simd.md
19495         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19496         (aarch64_frecpe<mode>): New.
19497         (aarch64_frecps<mode>): Likewise.
19498         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19499         (v8type): Add frecp<esx>.
19500         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19501         (aarch64_frecps<mode>): Likewise.
19502         * config/aarch64/iterators.md (FRECP): New.
19503         (frecp_suffix): Likewise.
19504         * config/aarch64/arm_neon.h
19505         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19506
19507 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19508
19509         PR target/56995
19510         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19511         (REG_CLASS_NAMES): Idem.
19512         (REG_CLASS_CONTENTS): Idem.
19513         (REGCLASS_HAS_FP_REG): Idem.
19514         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19515         (sh_conditional_register_usage): Idem.
19516
19517 2013-04-21  Jeff Law  <law@redhat.com>
19518
19519         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19520         (ssa_forward_propagate_and_combine): Use it.
19521
19522 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19523
19524         * lra.c: Update the flow chart diagram.
19525
19526 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19527
19528         PR rtl-optimization/56847
19529         * lra-constraints.c (process_alt_operands): Discourage alternative
19530         with non-matche doffsettable memory constraint fro memory with
19531         known offset.
19532
19533 2013-04-19  Richard Biener  <rguenther@suse.de>
19534
19535         PR tree-optimization/56982
19536         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19537         function.
19538         * gimplify.c (gimplify_call_expr): Notice special calls.
19539         (gimplify_modify_expr): Likewise.
19540         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19541         abnormal control flow receivers.
19542         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19543         in the same way as cfun->has_nonlocal_labels.
19544         (gimple_purge_dead_abnormal_call_edges): Likewise.
19545         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19546         receivers start a basic-block.
19547
19548 2013-04-19  Richard Biener  <rguenther@suse.de>
19549
19550         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19551         member ...
19552         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19553         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19554         (SLP_TREE_LOAD_PERMUTATION): Add.
19555         (vect_transform_slp_perm_load): Adjust prototype.
19556         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19557         (vect_free_slp_instance): Likewise.
19558         (vect_create_new_slp_node): Likewise.
19559         (vect_supported_slp_permutation_p): Remove.
19560         (vect_slp_rearrange_stmts): Adjust.
19561         (vect_supported_load_permutation_p): Likewise.  Inline
19562         vect_supported_slp_permutation_p here.
19563         (vect_analyze_slp_instance): Compute load permutations per
19564         slp node instead of per instance.
19565         (vect_get_slp_defs): Adjust.
19566         (vect_transform_slp_perm_load): Likewise.
19567         (vect_schedule_slp_instance): Remove redundant code.
19568         (vect_schedule_slp): Remove hack for PR56270, add it ...
19569         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19570         CSE loads for SLP.  Adjust.
19571
19572 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19573
19574         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19575         spelling in two comments.
19576
19577 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19578
19579         PR target/56797
19580         * config/arm/arm.c (load_multiple_sequence): Require SP
19581         as base register for loads if SP is in the register list.
19582
19583 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19584
19585         PR tree-optimization/56718
19586         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19587         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19588         and made public.  Adjusted all callers.
19589         (ipa_intraprocedural_devirtualization): New function.
19590         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19591         (ipa_intraprocedural_devirtualization): Likewise.
19592         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19593
19594 2013-04-19  Richard Biener  <rguenther@suse.de>
19595
19596         PR tree-optimization/57000
19597         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19598
19599 2013-04-19  Terry Guo  <terry.guo@arm.com>
19600
19601         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19602         Replace with ...
19603         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19604         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19605         (cortex_m4_fmacs): Use new reservations.
19606         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19607
19608 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19609
19610         PR rtl-optimization/56999
19611         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19612         related code.
19613         (lra_coalesce): Remove split_origin_bitmap and related code.
19614         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19615         ranges if necessary.
19616
19617 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19618
19619         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19620         New array.
19621         (ix86_expand_call): Remove clobbered_registers array and use
19622         x86_64_ms_sysv_extra_clobbered_registers instead.
19623         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19624         Declare here.
19625         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19626         predicate.
19627         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19628         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19629         (*call_value_rex64_ms_sysv): Ditto.
19630
19631 2013-04-18  Cary Coutant  <ccoutant@google.com>
19632
19633         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19634         parent instead of die_mark.
19635
19636 2013-04-18  Diego Novillo  <dnovillo@google.com>
19637
19638         * gimple.c (create_gimple_tmp): New.
19639         (get_expr_type): New.
19640         (build_assign): New.
19641         (build_type_cast): New.
19642         * gimple.h (enum ssa_mode): Define.
19643         (gimple_seq_set_location): New.
19644         * asan.c (build_check_stmt): Change some gimple_build_* calls
19645         to use build_assign and build_type_cast.
19646
19647 2013-04-18  Richard Biener  <rguenther@suse.de>
19648
19649         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19650         handle negative step.  Remove redundant checks.
19651         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19652         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19653         for negative step and grouped loads fail to vectorize.
19654
19655 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19656
19657         * emit-rtl.c (reset_insn_used_flags): New function.
19658         (reset_all_used_flags): Use it.
19659         (verify_insn_sharing): New function.
19660         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19661
19662 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19663
19664         PR tree-optimization/56984
19665         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19666         and (x >> M) >= N don't register any assertion if N << M is the
19667         minimum value.
19668
19669 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19670
19671         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19672         do not delete_insn insns that have not yet been emitted, only
19673         unlink them with remove_insn.
19674         * df-scan.c (df_insn_delete): Revert r197492.
19675
19676 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19677
19678         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19679         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19680
19681 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19682
19683         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19684         define_insn_and_split.
19685         (and_scc,ior_scc,negscc): Likewise.
19686         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19687
19688 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19689
19690         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19691         can benefit from using a sequence of LDRD instructions in epilogue
19692         instead of a single LDM instruction.
19693
19694 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19695
19696         PR 45688
19697         * doc/extend.texi: Fix typo.
19698
19699 2013-04-17  Richard Biener  <rguenther@suse.de>
19700
19701         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19702         (vect_build_slp_tree): ... here.
19703         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19704         match.  Remove special-casing of mismatched complex loads.
19705         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19706         re-try the match with swapped commutative operands.
19707         (vect_supported_load_permutation_p): Remove special-casing of
19708         mismatched complex loads.
19709         (vect_analyze_slp_instance): Adjust.
19710
19711 2013-04-17  Richard Biener  <rguenther@suse.de>
19712
19713         PR rtl-optimization/56921
19714         * cfgloop.h (struct loop): Add simple_loop_desc member.
19715         (struct niter_desc): Mark with GTY(()).
19716         (simple_loop_desc): Do not use aux field but simple_loop_desc.
19717         * loop-iv.c (get_simple_loop_desc): Likewise.
19718         (free_simple_loop_desc): Likewise.
19719
19720         Revert
19721         2013-04-16  Richard Biener  <rguenther@suse.de>
19722
19723         PR rtl-optimization/56921
19724         * loop-init.c (pass_rtl_move_loop_invariants): Add
19725         TODO_do_not_ggc_collect to todo_flags_finish.
19726         (pass_rtl_unswitch): Same.
19727         (pass_rtl_unroll_and_peel_loops): Same.
19728         (pass_rtl_doloop): Same.
19729
19730 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
19731
19732         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
19733         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
19734         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
19735         references.
19736         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
19737         * tree-streamer.c (record_common_node): Adjust reference in comment.
19738
19739 2013-04-17  Terry Guo  <terry.guo@arm.com>
19740
19741         * config/arm/cortex-m4.md: Add a new bypass.
19742
19743 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19744
19745         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
19746         New pattern.
19747         (*subs_<optab><mode>_multp2): New pattern.
19748         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19749         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19750
19751 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19752
19753         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
19754         (*subs_mul_imm_<mode>): New pattern.
19755
19756 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
19757
19758         PR target/56948
19759         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
19760         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
19761         (vsx_movti_32bit): Change j->wa to O->wa.
19762
19763 2013-04-16  Richard Biener  <rguenther@suse.de>
19764
19765         PR rtl-optimization/56921
19766         * loop-init.c (pass_rtl_move_loop_invariants): Add
19767         TODO_do_not_ggc_collect to todo_flags_finish.
19768         (pass_rtl_unswitch): Same.
19769         (pass_rtl_unroll_and_peel_loops): Same.
19770         (pass_rtl_doloop): Same.
19771
19772 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
19773
19774         * config/arm/arm.c (emit_multi_reg_push): New declaration
19775         for an existing function.
19776         (arm_emit_strd_push): New function.
19777         (arm_expand_prologue): Used here.
19778         (arm_emit_ldrd_pop): New function.
19779         (arm_expand_epilogue): Used here.
19780         (arm_get_frame_offsets): Update condition.
19781         (arm_emit_multi_reg_pop): Add a special case for load of a single
19782         register with writeback.
19783
19784 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
19785
19786         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
19787         description.
19788
19789 2013-04-16  Richard Biener  <rguenther@suse.de>
19790
19791         PR tree-optimization/56756
19792         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
19793         (first_mem_ref_loc): New.
19794         (execute_sm): Place the load temporarily before a previous
19795         access instead of in the latch edge to ensure its SSA dependencies
19796         are defined at points dominating the load.
19797
19798 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
19799
19800         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
19801         correct fix by moving header and footer insn to the footer of
19802         the merged basic block.  Clear BB_END of the merged-away block.
19803
19804         PR middle-end/43631
19805         * emit-rtl.c (make_note_raw): New function.
19806         (link_insn_into_chain): New static inline function.
19807         (add_insn): Use it.
19808         (add_insn_before, add_insn_after): Factor insn chain linking code...
19809         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
19810         using link_insn_into_chain.
19811         (note_outside_basic_block_p): New helper function for emit_note_after
19812         and emit_note_before.
19813         (emit_note_after): Use nobb variant of add_insn_after if the note
19814         should not be contained in a basic block.
19815         (emit_note_before): Use nobb variant of add_insn_before if the note
19816         should not be contained in a basic block.
19817         (emit_note_copy): Use make_note_raw.
19818         (emit_note): Likewise.
19819         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
19820         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
19821         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
19822         the moved barrier the tail of the basic block it follows.
19823         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
19824
19825 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
19826
19827         PR tree-optimization/56962
19828         * gimple-ssa-strength-reduction.c (record_increment): Only set
19829         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
19830         either rhs1 or rhs2 is equal to c->base_expr.
19831
19832 2013-04-15  Richard Biener  <rguenther@suse.de>
19833
19834         PR tree-optimization/56933
19835         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
19836         member.
19837         (GROUP_READ_WRITE_DEPENDENCE): Remove.
19838         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
19839         * tree-vect-data-refs.c (vect_analyze_group_access): Move
19840         dependence check ...
19841         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
19842         ... here.
19843         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
19844         GROUP_READ_WRITE_DEPENDENCE.
19845
19846 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19847
19848         * emit-rtl.c (reset_all_used_flags): New function.
19849         (verify_rtl_sharing): Call reset_all_used_flags before and after
19850         performing the checks.
19851
19852 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19853
19854         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
19855         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
19856         * config/arm/constraints.md (De): New constraint.
19857         * config/arm/neon.md (anddi3_neon): Delete.
19858         (neon_vand<mode>): Expand to standard anddi3 pattern.
19859         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
19860         Move earlier in the file.
19861         (neon_inv_logic_op2): Likewise.
19862         (arm_anddi_operand_neon): New predicate.
19863
19864 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19865
19866         * configure.ac (gcc_cv_ld_as_needed): Set
19867         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
19868         Use -z ignore, -z record on *-*-solaris2*.
19869         (HAVE_LD_AS_NEEDED): Update comment.
19870         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
19871         * configure: Regenerate.
19872         * config.in: Regenerate.
19873         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
19874         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
19875         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
19876         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
19877         equivalents.  Fix markup.
19878         * doc/tm.texi: Regenerate.
19879
19880 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
19881
19882         * config/i386/i386.opt: New option mstack-protector-guard=.
19883         * config/i386/i386-opts.h: Add enum stack_protector_guard.
19884         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
19885         TARGET_SSP_TLS_GUARD.
19886         * config/i386/i386.c (ix86_option_override_internal): Set
19887         ix86_stack_protector_guard.
19888         * config/i386/i386.md (stack_protect_set): Enable for
19889         TARGET_SSP_TLS_GUARD only.
19890         (stack_protect_set_<mode>): Ditto.
19891         (stack_protect_test): Ditto.
19892         (stack_protect_test_<mode>): Ditto.
19893         * doc/invoke.texi (i386 Option): Document.
19894
19895 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
19896
19897         PR target/56890
19898         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
19899         (S_MODES): Set H_MODE bit.
19900         (SF_MODES): Set only S_MODE and SF_MODE bits.
19901         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
19902         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
19903         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
19904         <MODE_FLOAT>: Likewise.
19905
19906 2013-04-15  Joey Ye  <joey.ye@arm.com>
19907
19908         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
19909
19910 2013-04-15  Joey Ye  <joey.ye@arm.com>
19911
19912         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
19913         for real far jump.
19914         (thumb_far_jump_used_p): Count instruction size and set
19915         far_jump_used.
19916
19917 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
19918
19919         * reorg.c (fill_simple_delay_slots): Reindent block of code.
19920         * resource.c (mark_target_live_regs): Reformat conditional block.
19921
19922 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
19923
19924         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
19925         notes, they are emitted only just before final.
19926         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
19927
19928 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
19929
19930         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
19931         * cfgrtl.c (delete_insn): Call it here instead.
19932         * lra-spills.c (lra_final_code_change): Use delete_insn.
19933         * haifa-sched.c (sched_remove_insn): Likewise.
19934         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
19935         returning to the nop pool.
19936         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
19937         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
19938
19939 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
19940
19941         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
19942         * doc/tm.texi: Regenerated.
19943
19944 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
19945
19946         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
19947         QImode checks.
19948
19949 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
19950
19951         * df-core.c (df_find_def): Compare register numbers.
19952         (df_find_use): Likewise.
19953
19954 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
19955
19956         PR target/56903
19957         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
19958         lra_in_progress for return.
19959
19960 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
19961
19962         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
19963         define_insn into define_insn_and_split and emit movsicc patterns.
19964
19965 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
19966
19967         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
19968
19969 2013-04-12  Richard Biener  <rguenther@suse.de>
19970
19971         * tree-pass.h (TODO_do_not_ggc_collect): New.
19972         * passes.c (execute_one_ipa_transform_pass): Honor
19973         TODO_do_not_ggc_collect.
19974         (execute_one_pass): Likewise.
19975
19976         Revert
19977         2013-04-10  Richard Biener  <rguenther@suse.de>
19978
19979         * passes.c (init_optimization_passes): Remove reload pass.
19980         * ira.c (do_reload): Merge into ...
19981         (ira): ... this.
19982         (rest_of_handle_reload): Remove.
19983         (pass_reload): Likewise.
19984         * config/i386/i386.c (ix86_option_override): Refer to ira instead
19985         of reload for vzeroupper pass placement.
19986
19987 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
19988
19989         PR tree-optimization/56918
19990         PR tree-optimization/56920
19991         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
19992         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
19993         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
19994         use wide_mul_with_sign method.
19995
19996 2013-04-12  Richard Biener  <rguenther@suse.de>
19997
19998         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
19999         not be considered a gimple constant.
20000
20001 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20002
20003         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20004         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20005         operations.
20006
20007 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20008             Jakub Jelinek  <jakub@redhat.com>
20009
20010         * opts.c: Include diagnostic-color.h.
20011         (common_handle_option): Handle OPT_fdiagnostics_color_.
20012         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20013         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20014         (diagnostic-color.o): New.
20015         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20016         (diagnostic_color_rule): New enum.
20017         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20018         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20019         the location string.
20020         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20021         either NULL, or color kind.
20022         * diagnostic-color.c: New file.
20023         * diagnostic-color.h: New file.
20024         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20025         arguments.
20026         * doc/invoke.texi (-fdiagnostics-color): Document.
20027         * pretty-print.h (pp_show_color): Define.
20028         (struct pretty_print_info): Add show_color field.
20029         * diagnostic.c: Include diagnostic-color.h.
20030         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20031         macros.  Colorize error:, warning: etc. strings and also the location
20032         string.
20033         (diagnostic_show_locus): Colorize the caret line.
20034         * pretty-print.c: Include diagnostic-color.h.
20035         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20036         inside of %< %> quotes or quoted through q format modifier.
20037
20038 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20039
20040         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20041
20042 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20043
20044         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20045         code in CC_NZ mode.
20046         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20047         pattern.
20048
20049 2013-04-11  Marek Polacek  <polacek@redhat.com>
20050
20051         PR tree-optimization/48184
20052         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20053
20054 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20055
20056         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20057         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20058         (skip_simple_arithmetic): Tidy up.
20059         * tree.h (skip_simple_constant_arithmetic): Declare.
20060
20061 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20062
20063         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20064
20065 2013-04-11  Richard Biener  <rguenther@suse.de>
20066
20067         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20068         generate vector constants.
20069
20070 2013-04-11  Richard Biener  <rguenther@suse.de>
20071
20072         PR tree-optimization/56878
20073         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20074         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20075         New function.
20076         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20077         Prefer to align the DR with the most invariant base address.
20078
20079 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20080
20081         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20082         comment.
20083
20084 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20085
20086         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20087         floating-point vector comparisons against 0.
20088
20089 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20090
20091         PR tree-optimization/56899
20092         * fold-const.c (extract_muldiv_1): Apply distributive law
20093         only if TYPE_OVERFLOW_WRAPS (ctype).
20094
20095 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20096
20097         PR target/56124
20098         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20099         loading has side effect.
20100
20101 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20102
20103         * config/sparc/sparc.c: Include tree-pass.h.
20104         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20105         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20106         head of file.  Change return type.  Split off gate function.
20107         (sparc_gate_work_around_errata): New function.
20108         (pass_work_around_errata): New pass definition.
20109         (insert_pass_work_around_errata) New pass insert definition to
20110         insert pass_work_around_errata just after delayed-branch scheduling.
20111         (sparc_option_override): Insert the pass.
20112         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20113
20114 2013-04-10  David S. Miller  <davem@davemloft.net>
20115
20116         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20117         or -mcpu=hypersparc.
20118
20119         * target.def (cstore_mode): New hook.
20120         * target.h: Include insn-codes.h
20121         * targhooks.c: Likewise.
20122         (default_cstore_mode): New function.
20123         * targhooks.h: Declare it.
20124         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20125         * doc/tm.texi: Rebuild.
20126         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20127         target hook, rather than inspecting the insn_data.
20128         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20129         (TARGET_CSTORE_MODE): Redefine.
20130         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20131         result patterns.
20132         * config/sparc/predicates.md (cstore_result_operand): New special
20133         predicate.
20134         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20135         Use it for operand 0.
20136         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20137         (*snesi_special): Likewise.
20138         (*snesi_zero): Likewise.
20139         (*seqsi_zero): Likewise.
20140         (*sltu_insn): Likewise.
20141         (*sgeu_insn): Likewise.
20142         (*seqdi_special): Make operand 0 and comparison operation be of
20143         DImode.
20144         (*snedi_special): Likewise.
20145         (*snedi_special_vis3): Likewise.
20146         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20147         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20148         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20149         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20150         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20151         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20152         (*sltu_extend_sp64): Likewise.
20153         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20154         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20155         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20156         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20157         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20158
20159 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20160
20161         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20162         (aarch64_start_file): Use the new function.
20163
20164 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20165             Jason Merrill  <jason@redhat.com>
20166
20167         * common.opt: Add -gdwarf.
20168         * opts.c (common_handle_option): Handle it.
20169         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20170
20171 2013-04-10  Richard Biener  <rguenther@suse.de>
20172
20173         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20174         (execute_one_ipa_transform_pass): But unconditionally here.
20175         (execute_one_pass): And here.
20176         (init_optimization_passes): Remove reload pass.
20177         * tree-pass.h (TODO_ggc_collect): Remove.
20178         (pass_reload): Likewise.
20179         * ira.c (do_reload): Merge into ...
20180         (ira): ... this.
20181         (rest_of_handle_reload): Remove.
20182         (pass_reload): Likewise.
20183         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20184         of reload for vzeroupper pass placement.
20185         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20186         and todo_flags_finish of all passes.
20187
20188 2013-04-10  Richard Biener  <rguenther@suse.de>
20189
20190         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20191         first_const_oprnd field, rename first_def_type to first_op_type.
20192         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20193         (vect_get_and_check_slp_defs): Always use the type of the
20194         operand.  Allow mixed vect_external_def, vect_constant_def types.
20195         (vect_get_constant_vectors): Handle mixed vect_external_def,
20196         vect_constant_def types.
20197
20198 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20199
20200         PR tree-optimization/55524
20201         * tree-ssa-math-opts.c
20202         (convert_mult_to_fma): Don't use an fms construct
20203         when we don't have an fms operation, but fnma, and it looks
20204         likely that we'll be able to use the latter.
20205
20206 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20207
20208         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20209         function.
20210         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20211         inline fail caused by overwritable functions.
20212
20213 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20214
20215         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20216         unnecessary bits in the constant power of two case.
20217
20218 2013-04-10  Richard Biener  <rguenther@suse.de>
20219
20220         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20221         broken code swapping operands.
20222         (vect_build_slp_tree): Do not compute load permutations here.
20223         (vect_analyze_slp_instance): Compute load permutations here,
20224         after building the SLP tree.
20225
20226 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20227
20228         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20229         of next/prev_real_insn.
20230
20231 2013-04-09  Jan Hubicka  <jh@suse.cz>
20232
20233         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20234         Drop aliased parameter.
20235         (function_and_variable_visibility): Do not handle alias pairs.
20236         * cgraph.c (varpool_externally_visible_p): Update prototype.
20237         * varpool.c (varpool_add_new_variable): Update.
20238
20239 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20240
20241         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20242
20243 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20244
20245         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20246
20247         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20248
20249 2013-04-09  Marek Polacek  <polacek@redhat.com>
20250
20251         PR tree-optimization/48762
20252         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20253
20254 2013-04-09  Richard Biener  <rguenther@suse.de>
20255
20256         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20257         dealing with cost.
20258         (vect_build_slp_tree): Likewise.
20259         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20260         calculating the cost of a SLP instance.
20261         (vect_analyze_slp_instance): Use it from here, after building
20262         the SLP tree.
20263
20264 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20265
20266         PR middle-end/56883
20267         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20268         expand_omp_for_static_chunk): Use simple_p = true in
20269         force_gimple_operand_gsi calls when assigning to addressable decls.
20270
20271 2013-04-09  Jeff Law  <law@redhat.com>
20272
20273         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20274         when the boolean was created by converting a wider object which
20275         had a boolean range.
20276
20277 2013-04-09  Richard Biener  <rguenther@suse.de>
20278
20279         * tree-vectorizer.h (slp_void_p): Remove.
20280         (slp_tree): Typedef before _slp_tree declaration.
20281         (struct _slp_tree): Use a vector of slp_tree as children.
20282         (vect_get_place_in_interleaving_chain): Remove.
20283         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20284         Move ...
20285         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20286         and make static.
20287         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20288         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20289         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20290         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20291         Use slp_node instead of slp_void_p and adjust.
20292
20293 2013-04-09  Richard Biener  <rguenther@suse.de>
20294
20295         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20296         work that is not necessary.
20297
20298 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20299
20300         PR tree-optimization/56854
20301         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20302         forward into clobber stmts if it would change MEM_REF lhs into
20303         non-MEM_REF.
20304
20305 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20306
20307         * tree.c (type_hash_lookup, type_hash_add): Make static.
20308         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20309
20310 2013-04-09  Richard Biener  <rguenther@suse.de>
20311
20312         * tree.h (unsave_expr_now): Remove.
20313         * tree-inline.c (mark_local_for_remap_r): Remove.
20314         (unsave_expr_1): Likewise.
20315         (unsave_r): Likewise.
20316         (unsave_expr_now): Likewise.
20317         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20318         (propagate_tree_value): Likewise.
20319
20320 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20321
20322         * doc/rtl.texi (sequence): Rewrite documentation to match the
20323         current use of SEQUENCE rtl objects.
20324         * rtl.def (SEQUENCE): Likewise.
20325
20326         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20327         Update documentation.
20328         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20329         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20330
20331         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20332
20333 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20334
20335         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20336         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20337         methods.
20338         (estimate_edge_size_and_time): Add comment to suggest using rounding
20339         methods.
20340         (estimate_node_size_and_time): Ditto.
20341         (remap_edge_change_prob): Use helper rounding divide methods.
20342         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20343         (gimple_mod_pow2_value_transform): Ditto.
20344         (gimple_mod_subtract_transform): Ditto.
20345         (gimple_ic_transform): Ditto.
20346         (gimple_stringops_transform): Ditto.
20347         * stmt.c (conditional_probability): Ditto.
20348         (emit_case_dispatch_table): Ditto.
20349         * lto-cgraph.c (merge_profile_summaries): Ditto.
20350         * tree-optimize.c (execute_fixup_cfg): Ditto.
20351         * cfgcleanup.c (try_forward_edges): Ditto.
20352         * cfgloopmanip.c (scale_loop_profile): Ditto.
20353         (loopify): Ditto.
20354         (duplicate_loop_to_header_edge): Ditto.
20355         (lv_adjust_loop_entry_edge): Ditto.
20356         * tree-vect-loop.c (vect_transform_loop): Ditto.
20357         * profile.c (compute_branch_probabilities): Ditto.
20358         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20359         * lto-streamer-in.c (input_cfg): Ditto.
20360         * gimple-streamer-in.c (input_bb): Ditto.
20361         * ipa-cp.c (update_profiling_info): Ditto.
20362         (update_specialized_profile): Ditto.
20363         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20364         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20365         rounding methods.
20366         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20367         (compute_trg_info): Ditto.
20368         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20369         (purge_dead_edges): Ditto.
20370         * loop-unswitch.c (unswitch_loop): Ditto.
20371         * cgraphclones.c (cgraph_clone_edge): Ditto.
20372         (cgraph_clone_node): Ditto.
20373         * tree-inline.c (copy_bb): Ditto.
20374         (copy_edges_for_bb): Ditto.
20375         (initialize_cfun): Ditto.
20376         (copy_cfg_body): Ditto.
20377         (expand_call_inline): Ditto.
20378
20379 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20380
20381         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20382         TARGET_CYGWIN64 by TARGET_64BIT.
20383
20384 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20385
20386         * config/epiphany/epiphany.md (GPR_1): New constant.
20387         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20388         * config/epiphany/epiphany.c (gen_compare_reg):
20389         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20390         is already in place.
20391         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20392         Don't require being called during rtl expansion; If y operlaps r0,
20393         return 0.
20394         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20395         (epiphany_expand_epilogue): Likewise.
20396
20397         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20398         Don't use CC_FPmode for ORDERED / UNORDERED.
20399         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20400
20401         * config/epiphany/constraints.md (CnL): New constraint.
20402         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20403         * config/epiphany/predicates.md (add_operand): Allow 1024.
20404
20405         * config/epiphany/epiphany.md (logical_op): New code iterator.
20406         (op_mnc): New code attribute.
20407         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20408         (mov_f+1, mov_f+2): New peephole2 patterns.
20409
20410         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20411         (cstoresi4): Also allow re-use of zero result when doing a NE
20412         comparison to a non-zero operand.
20413         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20414
20415         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20416         Use gen_addsi3_i / gen_subsi3_i.
20417
20418 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20419
20420         PR c++/34949
20421         PR c++/50243
20422         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20423         contain anything but clobbers, at most one __builtin_stack_restore,
20424         optionally debug stmts and final resx, and if it has at least one
20425         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20426         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20427         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20428         which isn't defaut definition, remove them.
20429         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20430         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20431         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20432         with MEM_REF LHS with SSA_NAME address.
20433
20434 2013-04-08  Jeff Law  <law@redhat.com>
20435
20436         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20437
20438 2013-04-08  Richard Biener  <rguenther@suse.de>
20439
20440         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20441         extra newline.
20442         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20443         determined vector type.
20444         (vect_analyze_data_refs): Likewise.
20445         (vect_get_new_vect_var): Adjust.
20446         (vect_create_destination_var): Preserve SSA name versions.
20447         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20448         not dump anything here.
20449
20450 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20451
20452         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20453         Add member lr_slot_known.
20454         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20455         if necessary.
20456         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20457         Remove code that sets lr_slot_offset according to what a previous
20458         version of epiphany_emit_save_restore used to do.
20459         (epiphany_emit_save_restore): When doing an lr save or restore,
20460         set/verify lr_slot_known and lr_slot_offset.
20461
20462 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20463
20464         PR target/54338
20465         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20466         in ALL_REGS.
20467
20468 2013-04-08  Richard Biener  <rguenther@suse.de>
20469
20470         * alias.c (find_base_term): Fix thinko in previous change.
20471
20472 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20473
20474         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20475         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20476         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20477         if possible to compute val.
20478         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20479         For QImode integers don't require anything about precision.  Use
20480         const_with_all_bytes_same to find out if the constant doesn't have
20481         repeated bytes in it.
20482
20483 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20484
20485         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20486         within mode size.
20487
20488 2013-04-08  Marek Polacek  <polacek@redhat.com>
20489
20490         PR rtl-optimization/48182
20491         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20492         value to 1.
20493
20494 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20495
20496         PR target/55487
20497         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20498         nuses, make sure we have a label.
20499
20500 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20501
20502         PR target/56843
20503         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20504         (rs6000_emit_swdiv_low_precision): Remove.
20505         (rs6000_emit_swdiv): Rewrite to handle between one and four
20506         iterations of Newton-Raphson generally; modify required number of
20507         iterations for some cases.
20508         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20509
20510 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20511
20512         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20513         set-but-unused variable.
20514
20515         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20516         basic blocks of released function bodies garbage-collectable.
20517
20518         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20519         (struct rtl_opt_pass): Add TODO_df_finish.
20520
20521         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20522
20523 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20524
20525         * config/arm/constraints.md (q): New constraint.
20526         * config/arm/ldrdstrd.md: New file.
20527         * config/arm/arm.md (ldrdstrd.md) New include.
20528         (arm_movdi): Use "q" instead of "r" constraint
20529         for double-word memory access.
20530         (movdf_soft_insn): Likewise.
20531         * config/arm/vfp.md (movdi_vfp): Likewise.
20532         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20533         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20534         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20535         (mem_ok_for_ldrd_strd): Likewise.
20536         (output_move_double): Update assertion.
20537
20538 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20539
20540         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20541
20542 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20543
20544         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20545         define_insn_and_split.
20546         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20547
20548 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20549
20550         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20551         define_insn_and_split.
20552         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20553         (shiftsi3_compare): New pattern.
20554         (rrx): New pattern.
20555         * config/arm/unspecs.md (UNSPEC_RRX): New.
20556
20557 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20558
20559         * config/arm/arm.md (negdi_extendsidi): New pattern.
20560         (negdi_zero_extendsidi): Likewise.
20561
20562 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20563
20564         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20565         define_insn_and_split.
20566         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20567         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20568
20569 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20570
20571         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20572         define_insn_and_split.
20573         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20574         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20575
20576 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20577
20578         * config/arm/arm.md (subsi3_carryin): New pattern.
20579         (subsi3_carryin_const): Likewise.
20580         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20581         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20582
20583 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20584
20585         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20586
20587 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20588
20589         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20590         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20591
20592 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20593
20594         * config/arm/arm.c (arm_expand_builtin): Change fcode
20595         type to unsigned int.
20596
20597 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20598
20599         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20600
20601 2013-04-04  Ian Lance Taylor  <iant@google.com>
20602
20603         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20604         language standard.
20605
20606 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20607
20608         PR middle-end/56729
20609         * df-scan.c (df_insn_delete): Disable failing assert.
20610
20611 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20612
20613         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20614         New function prototype.
20615         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20616         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20617         (arm_builtin_vectorized_function): New function.
20618
20619 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20620
20621         * config/arm/arm_neon_builtins.def: New file.
20622         * config/arm/arm.c (neon_builtin_data): Move contents to
20623         arm_neon_builtins.def.
20624         (enum arm_builtins): Include neon builtin definitions.
20625         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20626         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20627
20628 2013-04-04  Marek Polacek  <polacek@redhat.com>
20629
20630         PR tree-optimization/48186
20631         * predict.c (maybe_hot_frequency_p): Return false if
20632         HOT_BB_FREQUENCY_FRACTION is 0.
20633         (cgraph_maybe_hot_edge_p): Likewise.
20634
20635 2013-04-04  Richard Biener  <rguenther@suse.de>
20636
20637         PR tree-optimization/56826
20638         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20639         more accurately.
20640
20641 2013-04-04  Richard Biener  <rguenther@suse.de>
20642
20643         PR tree-optimization/56213
20644         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20645         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20646
20647 2013-04-04  Richard Biener  <rguenther@suse.de>
20648
20649         PR tree-optimization/56837
20650         * tree-loop-distribution.c (classify_partition): For non-zero
20651         values require that the value has the same precision as its
20652         mode to be useful as memset value.
20653
20654 2013-04-03  Nick Clifton  <nickc@redhat.com>
20655
20656         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20657         (fmssf4): Use fmsf.s on E3V5 architectures.
20658         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20659         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20660
20661 2013-04-03  Jeff Law  <law@redhat.com>
20662
20663         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20664         (lra-eliminations.o): Likewise.
20665
20666 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20667
20668         * gcov-io.c (compute_working_sets): Moved most of body of old
20669         compute_working_sets here from profile.c.
20670         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20671         (gcov_working_set_t): Moved typedef here from basic-block.h
20672         (compute_working_set): Declare.
20673         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20674         (get_working_sets): Renamed from compute_working_set,
20675         replace most of body with call to new compute_working_sets.
20676         (get_exec_counts): Replace call to compute_working_sets
20677         to get_working_sets.
20678         * profile.h (get_working_sets): Renamed from compute_working_set.
20679         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20680         to get_working_sets.
20681         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20682         * gcov-dump.c (dump_working_sets): New function.
20683
20684 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20685
20686         * hwint.c (sext_hwi, zext_hwi): New functions.
20687         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20688         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20689         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20690         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20691         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20692         (sext_hwi, zext_hwi): New functions.
20693
20694 2013-04-03  Jeff Law  <law@redhat.com>
20695
20696         PR tree-optimization/56799
20697         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20698         back test for widening conversion erroneously dropped in prior change.
20699
20700 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20701
20702         PR target/56809
20703         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20704         instead of next_real_insn.
20705
20706 2013-04-03  Marek Polacek  <polacek@redhat.com>
20707
20708         PR sanitizer/55702
20709         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20710
20711 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20712
20713         PR target/56809
20714         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20715         next_real_insn.
20716         (thumb1_output_casesi): Likewise.
20717         (thumb2_output_casesi): Likewise.
20718
20719 2013-04-03  Richard Biener  <rguenther@suse.de>
20720
20721         PR tree-optimization/56817
20722         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
20723         Split out ...
20724         (tree_unroll_loops_completely_1): ... new function to manually
20725         walk the loop tree, properly defering outer loops of unrolled
20726         loops to later iterations.
20727
20728 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20729
20730         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
20731         (vectorizable_load): Likewise.
20732         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
20733         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
20734
20735 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20736
20737         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
20738         BIT_FIELD_REF.
20739
20740 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
20741
20742         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
20743
20744 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
20745
20746         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
20747
20748 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20749
20750         PR tree-optimization/56790
20751         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
20752         folding.
20753
20754 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20755
20756         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
20757         Handle VEC_MERGE.
20758         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
20759         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
20760         equal arguments.
20761
20762 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
20763
20764         PR c/19449
20765         * tree.h (force_folding_builtin_constant_p): New decl.
20766         * builtins.c (force_folding_builtin_constant_p): New variable.
20767         (fold_builtin_constant_p): Fold immediately also if
20768         force_folding_builtin_constant_p.
20769
20770 2013-04-03  Richard Biener  <rguenther@suse.de>
20771
20772         PR tree-optimization/56812
20773         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
20774         DRs of the same interleaving chain are independent.
20775
20776 2013-04-02  Jason Merrill  <jason@redhat.com>
20777
20778         * gdbinit.in (pbb): Use debug fn.
20779
20780 2013-04-02  Lawrence Crowl  <crowl@google.com>
20781
20782         * sese.h (struct ivtype_map_elt_s): Remove unused.
20783         (extern debug_ivtype_map): Remove unused.
20784         (extern eq_ivtype_map_elts): Remove unused.
20785         * sese.c (debug_ivtype_map): Removed unused.
20786         (debug_ivtype_map_1): Removed unused.
20787         (debug_ivtype_elt): Remove unused.
20788         (eq_ivtype_map_elts): Remove unused.
20789
20790
20791 2013-04-02  Kai Tietz  <ktietz@redhat.com>
20792
20793         PR target/52790
20794         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
20795         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
20796         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
20797         function.
20798         (legitimize_pe_coff_symbol): Likewise.
20799         (is_imported_p): New helper-function.
20800         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
20801         for Windows x64 targets.
20802         (ix86_expand_prologue): Optimize for pe-coff targets.
20803         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
20804         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
20805         medium/large code-model.
20806         (legitimize_pic_address): Likewise.
20807         (legitimize_tls_address): Likewise.
20808         (ix86_expand_call): Likewise.
20809         (x86_output_mi_thunk): Likewise.
20810         (get_dllimport_decl): Add new beimport argument.
20811         (construct_plt_address): Don't assert for x64 pe-coff targets.
20812         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
20813         targets.
20814         (SYMBOL_FLAG_STUBVAR): New macro.
20815         (SYMBOL_REF_STUBVAR_P): Likewise.
20816         * config/i386/winnt.c (stub_list): New structure.
20817         (stub_head): New local variable.
20818         (i386_pe_record_stub): New function.
20819         (i386_pe_file_end): Emit refptr-stubs.
20820
20821 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
20822
20823         PR rtl-optimization/56745
20824         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
20825         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
20826
20827         PR c++/34949
20828         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
20829         and both of them are MEM_REFs, just compare first argument for
20830         equality and attempt to deal even with differing offsets.
20831
20832         PR c++/34949
20833         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
20834         of gimple_clobber_p to be MEM_REF.
20835         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
20836         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
20837         after gimplification.
20838         * asan.c (get_mem_ref_of_assignment): Don't instrument
20839         gimple_clobber_p stmts.
20840         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
20841         gimple_clobber_p stmt if they have MEM_REF lhs and
20842         are dead because of another gimple_clobber_p stmt.
20843         * tree-ssa-live.c (clear_unused_block_pointer): Treat
20844         gimple_clobber_p stmts like debug stmts.
20845         (remove_unused_locals): Remove clobbers with MEM_REF lhs
20846         that refer to unused VAR_DECLs or uninitialized values.
20847         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
20848         gimple_clobber_p stmts if they refer to removed parameters.
20849         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
20850         formatting.
20851
20852 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
20853
20854         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
20855         using SWI48 mode attribute.
20856
20857 2013-04-02  Wei Mi  <wmi@google.com>
20858
20859         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
20860         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
20861         *<rotate_insn><mode>3_mask in i386.md.
20862
20863 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
20864
20865         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
20866
20867 2013-04-02  Richard Biener  <rguenther@suse.de>
20868
20869         PR tree-optimization/56778
20870         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
20871         Runtime alias tests are not supported for gather loads.
20872         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
20873         stmts referenced from SSA operands before updating SSA form.
20874
20875 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
20876             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20877
20878         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
20879         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
20880         * config/arm/cortex-a53.md: New file.
20881         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
20882         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
20883         * config/arm/arm.c (arm_issue_rate): Likewise.
20884         * config/arm/arm-tune.md: Regenerate
20885         * config/arm/arm-tables.opt: Regenerate.
20886         * config/arm/arm-cores.def: Add cortex-a53.
20887
20888 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
20889
20890         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
20891         non-static link.
20892
20893 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
20894
20895         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
20896         scalar load/store operations using B/H registers.
20897         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
20898
20899 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
20900
20901         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
20902         scalar move.
20903         * config/aarch64/aarch64.c
20904         (aarch64_simd_scalar_immediate_valid_for_move): New.
20905         * config/aarch64/aarch64-protos.h
20906         (aarch64_simd_scalar_immediate_valid_for_move): New.
20907         * config/aarch64/constraints.md (Dh, Dq): New.
20908         * config/aarch64/iterators.md (hq): New.
20909
20910 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
20911
20912         * reorg.c (get_branch_condition): Deal with conditional returns.
20913         (fill_simple_delay_slots): Remove dead code dealing with jumps.
20914
20915 2013-04-01  Wei Mi  <wmi@google.com>
20916
20917         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
20918         Truncate operand 2 using %b asm operand modifier.
20919         (*<shift_insn><mode>3_mask): Ditto.
20920         (*<rotate_insn><mode>3_mask): Ditto.
20921
20922 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
20923
20924         PR middle-end/56798
20925         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
20926
20927 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
20928
20929         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
20930         of next_real_insn.
20931         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
20932
20933 2013-03-30  Lawrence Crowl  <crowl@google.com>
20934
20935         * dse.c (clear_alias_sets): Remove never set.
20936         (disqualified_clear_alias_sets): Remove never set.
20937         (clear_alias_mode_pool): Remove never set.
20938         (dse_step0): Remove condition that is never true.
20939         (canon_address): Remove condition that is never true.
20940         (dse_step7): Remove condition that is never true.
20941         (rest_of_handle_dse): Remove condition that is never true.
20942         (rest_of_handle_dse::did_global): Remove never read from above.
20943         (dse_step2_spill): Remove never called from above.
20944         (dse_step5_spill): Remove never called from above.
20945
20946 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
20947
20948         * doc/md.texi (Standard Names) <casesi>: Update documentation for
20949         JUMP_TABLE_DATA changes.
20950         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
20951         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
20952         (Insns) <jump_table_data>: New entry.
20953         * doc/tm.texi: Regenerate.
20954
20955         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
20956
20957         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
20958         for table jump at the end of a basic block using tablejump_p.
20959         * targhooks.c (default_invalid_within_doloop): Likewise.
20960         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
20961         target hook implementation that is identical to the default hook.
20962         (rs6000_invalid_within_doloop): Remove.
20963
20964         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
20965         unused variable from tablejump_p call.
20966
20967         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
20968         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
20969         (INSN_DELETED_P): Likewise.
20970         (emit_jump_table_data): New prototype.
20971         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
20972         after 4th as unused.
20973         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
20974         * sched-vis.c (print_insn): Likewise.
20975         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
20976         insn for compatibility with back ends that use next_active_insn to
20977         identify jump table data.
20978         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
20979         (remove_insn): Likewise.
20980         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
20981         to be emitted.
20982         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
20983         (emit_jump_table_data): New function.
20984
20985         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
20986         basic block, a JUMP_TABLE_DATA never is.
20987         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
20988         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
20989         off from code handling real insns.
20990         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
20991         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
20992         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
20993         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
20994         is not a NONDEBUG_INSN_P.
20995         * ira-costs.c (scan_one_insn): Likewise.
20996         * jump.c (mark_all_labels): Likewise.
20997         (mark_jump_label_1): Likewise.
20998         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
20999         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21000         a basic block.
21001         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21002         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21003         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21004         JUMP_TABLE_DATA_P insns.
21005         (calculate_elim_costs_all_insns): Likewise.
21006         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21007         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21008         (delete_output_reload): Code style fixups.
21009         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21010         insn flags on this non-insn.
21011         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21012         as scheduling barriers, for pre-change compatibility.
21013         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21014         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21015
21016         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21017         redundant JUMP_TABLE_DATA_P test.
21018         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21019         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21020         (frv_for_each_packet): Likewise.
21021         * config/i386/i386.c (min_insn_size): Likewise.
21022         (ix86_avoid_jump_mispredicts): Likewise.
21023         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21024         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21025         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21026         (mips16_insn_length): Robustify.
21027         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21028         (mips16_split_long_branches): Likewise.
21029         * config/pa/pa.c (pa_combine_instructions): Likewise.
21030         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21031         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21032         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21033         as contributing to pool range lengths.
21034         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21035         Remove redundant JUMP_TABLE_DATA_P test.
21036         (sh_loop_align): Likewise.
21037         (split_branches): Likewise.
21038         (sh_insn_length_adjustment): Likewise.
21039         * config/spu/spu.c (get_branch_target): Likewise.
21040
21041 2013-03-29  Jan Hubicka  <jh@suse.cz>
21042
21043         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21044         gcov streaming; stream hot bb threshold to ltrans.
21045         * predict.c (get_hot_bb_threshold): Break out from ....
21046         (maybe_hot_count_p): ... here.
21047         (set_hot_bb_threshold): New function.
21048         * lto-section-in.c (lto_section_name): Add profile.
21049         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21050         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21051         and data-streamer.h
21052         (histogram_entry): New structure.
21053         (histogram, histogram_pool): New global vars.
21054         (histogram_hash): New structure.
21055         (histogram_hash::hash): New method.
21056         (histogram_hash::equal): Likewise.
21057         (account_time_size): New function.
21058         (cmp_counts): New function.
21059         (dump_histogram): New function.
21060         (ipa_profile_generate_summary): New function.
21061         (ipa_profile_write_summary): New function.
21062         (ipa_profile_read_summary): New function.
21063         (ipa_profile): Decide on threshold.
21064         (pass_ipa_profile): Add ipa_profile_write_summary and
21065         ipa_profile_read_summary.
21066         * Makefile.in (ipa.o): Update dependencies.
21067         * lto-streamer.h (LTO_section_ipa_profile): New section.
21068
21069 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21070
21071         * tree.h (VAR_P): New.
21072
21073 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21074
21075         PR lto/56777
21076         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21077
21078 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21079
21080         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21081         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21082         (control_flow_insn_p): Likewise.
21083         * cfgrtl.c (duplicate_insn_chain): Likewise.
21084         * final.c (get_attr_length_1): Likewise.
21085         (shorten_branches): Likewise.
21086         (final_scan_insn): Likewise.
21087         * function.c (instantiate_virtual_regs): Likewise.
21088         * gcse.c (insert_insn_end_basic_block): Likewise.
21089         * ira-costs.c (scan_one_insn): Likewise.
21090         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21091         * lra.c (check_rtl): Likewise.
21092         * reload1.c (elimination_costs_in_insn): Likewise.
21093         * reorg.c (follow_jumps): Likewise.
21094
21095         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21096         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21097         (thumb_far_jump_used_p): Likewise.
21098         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21099         (workaround_speculation): Likewise.
21100         (add_sched_insns_for_speculation): Likewise.
21101         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21102         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21103         (frv_for_each_packet): Likewise.
21104         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21105         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21106         (final_emit_insn_group_barriers): Likewise.
21107         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21108         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21109         (mips16_insn_length): Likewise.
21110         * config/pa/pa.c (pa_reorg): Likewise.
21111         (pa_combine_instructions): Likewise.
21112         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21113         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21114         (sh_reorg): Likewise.
21115         (split_branches): Likewise.
21116         * config/spu/spu.c (get_branch_target): Likewise.
21117
21118         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21119         JUMP_TABLE_DATA_P.
21120
21121 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21122
21123         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21124         Fix declaration name.
21125
21126 2013-03-28  Lawrence Crowl  <crowl@google.com>
21127
21128         * graphds.h (struct graph.indicies): Remove unused.
21129         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21130         (SCOP_ORIGINAL_PDDRS): Remove unused.
21131         * sese.h (extern insert_loop_close_phis): Removed unused.
21132         (extern insert_guard_phis): Removed unused.
21133         (extern ivtype_map_elt_info): Removed unused.
21134         (new_ivtype_map_elt): Removed unused.
21135         * sese.c (ivtype_map_elt_info): Removed unused.
21136
21137 2013-03-28  Lawrence Crowl  <crowl@google.com>
21138
21139         * Makefile.in: Add several missing include dependences.
21140         (DUMPFILE_H): New.
21141         (test-dump.o): New.  This object is not added to any executable,
21142         but is present for ad-hoc testing.
21143         * bitmap.c
21144         (debug (const bitmap_head_def &)): New.
21145         (debug (const bitmap_head_def *)): New.
21146         * bitmap.h
21147         (extern debug (const bitmap_head_def &)): New.
21148         (extern debug (const bitmap_head_def *)): New.
21149         * cfg.c
21150         (debug (edge_def &)): New.
21151         (debug (edge_def *)): New.
21152         * cfghooks.c
21153         (debug (basic_block_def &)): New.
21154         (debug (basic_block_def *)): New.
21155         * dumpfile.h
21156         (dump_node (const_tree, int, FILE *)): Correct source file.
21157         * dwarf2out.c
21158         (debug (die_struct &)): New.
21159         (debug (die_struct *)): New.
21160         * dwarf2out.h
21161         (extern debug (die_struct &)): New.
21162         (extern debug (die_struct *)): New.
21163         * gimple-pretty-print.c
21164         (debug (gimple_statement_d &)): New.
21165         (debug (gimple_statement_d *)): New.
21166         * gimple-pretty-print.h
21167         (extern debug (gimple_statement_d &)): New.
21168         (extern debug (gimple_statement_d *)): New.
21169         * ira-build.c
21170         (debug (ira_allocno_copy &)): New.
21171         (debug (ira_allocno_copy *)): New.
21172         (debug (ira_allocno &)): New.
21173         (debug (ira_allocno *)): New.
21174         * ira-int.h
21175         (extern debug (ira_allocno_copy &)): New.
21176         (extern debug (ira_allocno_copy *)): New.
21177         (extern debug (ira_allocno &)): New.
21178         (extern debug (ira_allocno *)): New.
21179         * ira-lives.c
21180         (debug (live_range &)): New.
21181         (debug (live_range *)): New.
21182         * lra-int.h
21183         (debug (lra_live_range &)): New.
21184         (debug (lra_live_range *)): New.
21185         * lra-lives.c
21186         (debug (lra_live_range &)): New.
21187         (debug (lra_live_range *)): New.
21188         * omega.c
21189         (debug (omega_pb_d &)): New.
21190         (debug (omega_pb_d *)): New.
21191         * omega.h
21192         (extern debug (omega_pb_d &)): New.
21193         (extern debug (omega_pb_d *)): New.
21194         * print-rtl.c
21195         (debug (const rtx_def &)): New.
21196         (debug (const rtx_def *)): New.
21197         * print-tree.c
21198         (debug_tree (tree): Move within file.
21199         (debug_raw (const tree_node &)): New.
21200         (debug_raw (const tree_node *)): New.
21201         (dump_tree_via_hooks (const tree_node *, int)): New.
21202         (debug (const tree_node &)): New.
21203         (debug (const tree_node *)): New.
21204         (debug_verbose (const tree_node &)): New.
21205         (debug_verbose (const tree_node *)): New.
21206         (debug_head (const tree_node &)): New.
21207         (debug_head (const tree_node *)): New.
21208         (debug_body (const tree_node &)): New.
21209         (debug_body (const tree_node *)): New.
21210         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21211         (debug (vec<tree, va_gc> &)): New.
21212         (debug (vec<tree, va_gc> *)): New.
21213         * rtl.h
21214         (extern debug (const rtx_def &)): New.
21215         (extern debug (const rtx_def *)): New.
21216         * sbitmap.c
21217         (debug_raw (simple_bitmap_def &)): New.
21218         (debug_raw (simple_bitmap_def *)): New.
21219         (debug (simple_bitmap_def &)): New.
21220         (debug (simple_bitmap_def *)): New.
21221         * sbitmap.h
21222         (extern debug (simple_bitmap_def &)): New.
21223         (extern debug (simple_bitmap_def *)): New.
21224         (extern debug_raw (simple_bitmap_def &)): New.
21225         (extern debug_raw (simple_bitmap_def *)): New.
21226         * sel-sched-dump.c
21227         (debug (vinsn_def &)): New.
21228         (debug (vinsn_def *)): New.
21229         (debug_verbose (vinsn_def &)): New.
21230         (debug_verbose (vinsn_def *)): New.
21231         (debug (expr_def &)): New.
21232         (debug (expr_def *)): New.
21233         (debug_verbose (expr_def &)): New.
21234         (debug_verbose (expr_def *)): New.
21235         (debug (vec<rtx> &)): New.
21236         (debug (vec<rtx> *)): New.
21237         * sel-sched-dump.h
21238         (extern debug (vinsn_def &)): New.
21239         (extern debug (vinsn_def *)): New.
21240         (extern debug_verbose (vinsn_def &)): New.
21241         (extern debug_verbose (vinsn_def *)): New.
21242         (extern debug (expr_def &)): New.
21243         (extern debug (expr_def *)): New.
21244         (extern debug_verbose (expr_def &)): New.
21245         (extern debug_verbose (expr_def *)): New.
21246         (extern debug (vec<rtx> &)): New.
21247         (extern debug (vec<rtx> *)): New.
21248         * sel-sched-ir.h
21249         (_list_iter_cond_expr): Make inline instead of static.
21250         * sreal.c
21251         (debug (sreal &)): New.
21252         (debug (sreal *)): New.
21253         * sreal.h
21254         (extern debug (sreal &)): New.
21255         (extern debug (sreal *)): New.
21256         * tree.h
21257         (extern debug_raw (const tree_node &)): New.
21258         (extern debug_raw (const tree_node *)): New.
21259         (extern debug (const tree_node &)): New.
21260         (extern debug (const tree_node *)): New.
21261         (extern debug_verbose (const tree_node &)): New.
21262         (extern debug_verbose (const tree_node *)): New.
21263         (extern debug_head (const tree_node &)): New.
21264         (extern debug_head (const tree_node *)): New.
21265         (extern debug_body (const tree_node &)): New.
21266         (extern debug_body (const tree_node *)): New.
21267         (extern debug (vec<tree, va_gc> &)): New.
21268         (extern debug (vec<tree, va_gc> *)): New.
21269         * tree-cfg.c
21270         (debug (struct loop &)): New.
21271         (debug (struct loop *)): New.
21272         (debug_verbose (struct loop &)): New.
21273         (debug_verbose (struct loop *)): New.
21274         * tree-dump.c: Add header dependence.
21275         * tree-flow.h
21276         (extern debug (struct loop &)): New.
21277         (extern debug (struct loop *)): New.
21278         (extern debug_verbose (struct loop &)): New.
21279         (extern debug_verbose (struct loop *)): New.
21280         * tree-data-ref.c
21281         (debug (data_reference &)): New.
21282         (debug (data_reference *)): New.
21283         (debug (vec<data_reference_p> &)): New.
21284         (debug (vec<data_reference_p> *)): New.
21285         (debug (vec<ddr_p> &)): New.
21286         (debug (vec<ddr_p> *)): New.
21287         * tree-data-ref.h
21288         (extern debug (data_reference &)): New.
21289         (extern debug (data_reference *)): New.
21290         (extern debug (vec<data_reference_p> &)): New.
21291         (extern debug (vec<data_reference_p> *)): New.
21292         (extern debug (vec<ddr_p> &)): New.
21293         (extern debug (vec<ddr_p> *)): New.
21294         * tree-ssa-alias.c
21295         (debug (pt_solution &)): New.
21296         (debug (pt_solution *)): New.
21297         * tree-ssa-alias.h
21298         (extern debug (pt_solution &)): New.
21299         (extern debug (pt_solution *)): New.
21300         * tree-ssa-alias.c
21301         (debug (_var_map &)): New.
21302         (debug (_var_map *)): New.
21303         (debug (tree_live_info_d &)): New.
21304         (debug (tree_live_info_d *)): New.
21305         * tree-ssa-alias.h
21306         (extern debug (_var_map &)): New.
21307         (extern debug (_var_map *)): New.
21308         (extern debug (tree_live_info_d &)): New.
21309         (extern debug (tree_live_info_d *)): New.
21310
21311 2013-03-28  Jan Hubicka  <jh@suse.cz>
21312
21313         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21314
21315 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21316
21317         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21318         record only when desired or required.
21319
21320 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21321
21322         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21323         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21324
21325 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21326
21327         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21328         (*andsi3_compare0_uxtw): New pattern.
21329         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21330         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21331
21332 2013-03-28  Jan Hubicka  <jh@suse.cz>
21333
21334         * data-streamer-in.c (streamer_read_gcov_count): New function.
21335         * gimple-streamer-out.c: Include value-prof.h.
21336         (output_gimple_stmt): Output histogram.
21337         (output_bb): Use streamer_write_gcov_count.
21338         * value-prof.c: Include data-streamer.h
21339         (dump_histogram_value): Add HIST_TYPE_MAX.
21340         (stream_out_histogram_value): New function.
21341         (stream_in_histogram_value): New function.
21342         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21343         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21344         * data-streamer-out.c (streamer_write_gcov_count): New function.
21345         (streamer_write_gcov_count_stream): New function.
21346         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21347         (lto_output_node): Likewise.
21348         (input_node, input_edge): Likewise.
21349         * lto-streamer-out.c (output_cfg): Update streaming.
21350         * lto-streamer-in.c (input_cfg): Likewise.
21351         * data-streamer.h (streamer_write_gcov_count,
21352         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21353         * gimple-streamer-in.c: Include value-prof.h
21354         (input_gimple_stmt): Input histograms.
21355         (input_bb): Update profile streaming.
21356
21357 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21358
21359         * genmodes.c (emit_max_int): New function.
21360         (emit_insn_modes_h): Added call to emit_max_function.
21361         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21362         Added doc.
21363         * machmode.def: Fixed comment.
21364
21365 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21366
21367         * combine.c (try_combine): Removed useless assert.
21368         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21369
21370 2013-03-28  Marek Polacek  <polacek@redhat.com>
21371             Richard Biener  <rguenther@suse.de>
21372
21373         PR tree-optimization/56695
21374         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21375         build signed result of a vector comparison.
21376         * tree-cfg.c (verify_gimple_comparison): Check that a result
21377         of a vector comparison has signed type.
21378
21379 2013-03-28  Richard Biener  <rguenther@suse.de>
21380
21381         PR tree-optimization/37021
21382         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21383         do not restrict gaps between groups.
21384         * tree-vect-stmts.c (vectorizable_load): Properly account for
21385         a gap between groups.
21386
21387 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21388
21389         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21390         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21391         is not enabled.
21392
21393 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21394
21395         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21396         * doc/extend.texi (Named Address Spaces): Ditto.
21397         (Variable Attributes): Ditto.
21398
21399 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21400
21401         * config.build: Add support for cygwin x64 target.
21402         * config.gcc: Likewise.
21403         * config.host: Likewise.
21404         * configure.ac: Likewise
21405         * configure: Regenerated.
21406
21407 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21408
21409         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21410         * config/i386/t-cygwin-w64: New file.
21411         * config/i386/cygwin-w64.h: New file.
21412         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21413         and add support for x64-cygwin target.
21414         (CPP_SPEC): Likewise.
21415         (CXX_WRAP_SPEC_LIST): Undefine before define.
21416         (LIBGCJ_SONAME): Use 15 as version.
21417
21418 2013-03-27  Richard Biener  <rguenther@suse.de>
21419
21420         PR tree-optimization/56716
21421         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21422         dumping for ref nodes.
21423
21424 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21425
21426         PR tree-optimization/55334
21427         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21428         restricted pointers to arrays.
21429
21430 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21431
21432         * Makefile.in (.SUFFIXES): Add .cc.
21433         (.c.o): Apply same recipe for implicit rule .cc.o.
21434
21435 2013-03-27  Richard Biener  <rguenther@suse.de>
21436
21437         PR tree-optimization/37021
21438         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21439         REALPART/IMAGPART_EXPRs around the supported refs.
21440         * tree-ssa-structalias.c (find_func_aliases): Assume that
21441         floating-point values are not used to transfer pointers.
21442
21443 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21444
21445         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21446         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21447         * doc/tm.texi: Regenerate.
21448         * targhooks.h (default_has_ifunc_p): New.
21449         * targhooks.c (default_has_ifunc_p): Ditto.
21450         * config/linux-protos.h: New file.
21451         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21452         hook for linux which disables support of indirect functions in android.
21453         * config/linux-android.c: New file.
21454         * config/t-linux-android.c: Ditto.
21455         * config.gcc: Added new object file linux-android.o.
21456         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21457         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21458         * varasm.c (do_assemble_alias): Likewise.
21459         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21460         doesn't support indirect functions.
21461         * configure: Regenerate.
21462
21463 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21464
21465         PR target/56102
21466         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21467         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21468         mult-word mode.
21469
21470 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21471
21472         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21473
21474 2013-03-27  Terry Guo  <terry.guo@arm.com>
21475
21476         * config/arm/arm-cores.def: Added core cortex-r7.
21477         * config/arm/arm-tune.md: Regenerated.
21478         * config/arm/arm-tables.opt: Regenerated.
21479         * doc/invoke.texi: Added entry for core cortex-r7.
21480
21481 2013-03-27  Walter Lee  <walt@tilera.com>
21482
21483         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21484         double-decrement of next_scratch_regno.
21485
21486 2013-03-27  Walter Lee  <walt@tilera.com>
21487
21488         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21489         input operands.
21490         (insn_v1mulus): Ditto.
21491         (insn_v2muls): Ditto.
21492
21493 2013-03-27  Walter Lee  <walt@tilera.com>
21494
21495         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21496         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21497
21498 2013-03-27  Walter Lee  <walt@tilera.com>
21499
21500         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21501         (*sibcall_value): Ditto.
21502
21503 2013-03-27  Walter Lee  <walt@tilera.com>
21504
21505         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21506         (insn_mnz_v8qi): ... this ...
21507         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21508         vector equivalent.
21509         (insn_v<n>mnz): Replaced by ...
21510         (insn_v1mnz): ... this ...
21511         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21512         equivalent.
21513         (insn_mz_<mode>): Replaced by ...
21514         (insn_mz_v8qi): ... this ...
21515         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21516         vector equivalent.
21517         (insn_v<n>mz): Replaced by ...
21518         (insn_v1mz): ... this ...
21519         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21520         equivalent.
21521
21522 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21523
21524         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21525
21526 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21527
21528         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21529         than fprintf with a non-constant, non-format string.
21530
21531 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21532
21533         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21534         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21535         operand 0 predicate.
21536         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21537         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21538         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21539         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21540         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21541         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21542         (mov<mode>_insv_1): Remove expander.  Merge insn with
21543         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21544         Use general_x64nomem_operand as operand 1 predicate.
21545         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21546         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21547         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21548         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21549         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21550         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21551         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21552         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21553         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21554         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21555         (general_x64nomem_operand): Ditto.
21556
21557 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21558
21559         * config/rtems.opt: Add -pthread option.
21560
21561 2013-03-26  Richard Biener  <rguenther@suse.de>
21562
21563         * alias.c (find_base_term): Avoid redundant and not used recursion.
21564         (base_alias_check): Get the initial base term from the caller.
21565         (true_dependence_1): Compute and pass base terms to base_alias_check.
21566         (write_dependence_p): Likewise.
21567         (may_alias_p): Likewise.
21568
21569 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21570
21571         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21572         PC-relative load in SI modes and above only.
21573
21574 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21575
21576         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21577         * config/arm/iwmmxt.md (WCGR0): Update.
21578         (WCGR1, WCGR2, WCGR3): Likewise.
21579
21580 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21581
21582         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21583         Use x64 and nox64 isa attributes.
21584
21585 2013-03-26  Richard Biener  <rguenther@suse.de>
21586
21587         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21588         alignment computations and rely on get_object_alignment_1
21589         for the !TYPE_P case.
21590         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21591
21592 2013-03-26  Walter Lee  <walt@tilera.com>
21593
21594         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21595         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21596
21597 2013-03-25  Jeff Law  <law@redhat.com>
21598
21599         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21600         check for INTEGRAL_TYPE_P that was missing due to checking in
21601         wrong version of prior patch.
21602
21603 2013-03-25  Walter Lee  <walt@tilera.com>
21604
21605         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21606         TILEGX_INSN_SHUFFLEBYTES1.
21607         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21608         shufflebytes1.
21609         (tilegx_builtins): Ditto.
21610         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21611
21612 2013-03-25  Walter Lee  <walt@tilera.com>
21613
21614         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21615         (floatunssisf2): New pattern.
21616         (floatsidf2): New pattern.
21617         (floatunssidf2): New pattern.
21618
21619 2013-03-25  Walter Lee  <walt@tilera.com>
21620
21621         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21622         tests for constraint J, K, N, P.
21623
21624 2013-03-25  Walter Lee  <walt@tilera.com>
21625
21626         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21627         Use indirect/pcrel encoding.
21628         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21629         Ditto.
21630
21631 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21632
21633         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21634         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21635         * config/mips/mips.c (mips_option_override): Set IMADD default.
21636         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21637         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21638         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21639         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21640         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21641
21642 2013-03-25  Jeff Law  <law@redhat.com>
21643
21644         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21645         slightly to avoid creating and folding useless trees.  Simplify
21646         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21647
21648 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21649
21650         * config/i386/i386.md (*zero_extendsidi2): Merge with
21651         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21652         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21653         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21654         targets.  Clarify comment.
21655
21656 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21657
21658         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21659         pass-through jump functions differently.
21660         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21661         up jump functions.
21662
21663 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21664
21665         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21666         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21667         process it.
21668         (ipa_get_indirect_edge_target): New function.
21669         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21670         ipa_get_indirect_edge_target.  Update all callers.
21671         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21672         ipa_get_indirect_edge_target_1 instead of calling
21673         ipa_get_indirect_edge_target.
21674         (create_specialized_node): Pass aggvlas to
21675         ipcp_discover_new_direct_edges.
21676
21677 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21678
21679         * config/arm/arm.md (f_sels, f_seld): New types.
21680         (*cmov<mode>): New pattern.
21681         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21682         predicate.
21683
21684 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21685
21686         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21687         POSIX-printf for mingw-hosted  builds.
21688
21689 2013-03-25  Richard Biener  <rguenther@suse.de>
21690
21691         PR middle-end/56694
21692         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21693         must-not-throw stmt location.
21694
21695 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21696
21697         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21698         Emit load-acquire versions when acq is true.
21699         (arm_emit_store_exclusive): Add rel parameter.
21700         Emit store-release versions when rel is true.
21701         (arm_split_compare_and_swap): Use acquire-release instructions
21702         instead.
21703         of barriers when appropriate.
21704         (arm_split_atomic_op): Likewise.
21705         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21706         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21707         (VUNSPEC_SLX): Likewise.
21708         (VUNSPEC_LDA): Likewise.
21709         (VUNSPEC_STL): Likewise.
21710         * config/arm/sync.md (atomic_load<mode>): New pattern.
21711         (atomic_store<mode>): Likewise.
21712         (arm_load_acquire_exclusive<mode>): Likewise.
21713         (arm_load_acquire_exclusivesi): Likewise.
21714         (arm_load_acquire_exclusivedi): Likewise.
21715         (arm_store_release_exclusive<mode>): Likewise.
21716
21717 2013-03-25  Catherine Moore  <clm@codesourcery.com>
21718
21719         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
21720         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
21721         * config/mip/predicates.md (lwsp_swsp_operand,
21722         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
21723         sb16_operand, db4_operand, db7_operand, ib3_operand,
21724         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
21725         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
21726         andi16_operand): New predicates.
21727         * config/mips/mips.md (compression): New attribute.
21728         (enabled): New attribute.
21729         (length): Consider compression in computing length.
21730         (shift_compression): New code attribute.
21731         (*add<mode>3): New operands. Record compression.
21732         (sub<mode>3): Likewise.
21733         (one_cmpl<mode>2): Likewise.
21734         (*and<mode>3): Likewise.
21735         (*ior<mode>3): Likewise.
21736         (unnamed pattern for xor): Likewise.
21737         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
21738         (*<optab><mode>3): Likewise.
21739         (*mov<mode>_internal: Likewise.
21740         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
21741         (mips_unsigned_immediate_p): New.
21742         (umips_lwsp_swsp_address_p): New.
21743         (m16_based_address_p): New.
21744         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
21745         (mips_unsigned_immediate_p): New prototype.
21746         (lwsp_swsp_address_p): New prototype.
21747         (m16_based_address_p): New prototype.
21748         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
21749         (mips_signed_immediate_p): New function.
21750         (m16_based_address_p): New function.
21751         (lwsp_swsp_address_p): New function.
21752         (mips_print_operand_punctuation): Recognize short delay slot insns
21753         for microMIPS.add<mode>3"
21754
21755 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21756
21757         PR target/56720
21758         * config/arm/iterators.md (v_cmp_result): New mode attribute.
21759         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
21760
21761 2013-03-25  Richard Biener  <rguenther@suse.de>
21762
21763         PR tree-optimization/56689
21764         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
21765         any edge.
21766
21767 2013-03-25  Richard Biener  <rguenther@suse.de>
21768
21769         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
21770         of bitmap.
21771         (memory_references): Likewise.
21772         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
21773         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
21774         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
21775         (gather_mem_refs_in_loops): Fold into ...
21776         (analyze_memory_references): ... this.  Move initialization
21777         to tree_ssa_lim_initialize.
21778         (fill_always_executed_in): Rename to ...
21779         (fill_always_executed_in_1): ... this.
21780         (fill_always_executed_in): Move contains_call computation to
21781         this new function from ...
21782         (tree_ssa_lim_initialize): ... here.
21783         (tree_ssa_lim): Call fill_always_executed_in.
21784
21785 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
21786
21787         * postreload.c (reload_combine): Fix code detecting returns.
21788
21789 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
21790
21791         * function.c (emit_use_return_register_into_block): On cc0 targets,
21792         do not emit the sequence between cc0 setter and user.
21793
21794 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21795
21796         * config/i386/predicates.md (local_symbolic_operand): Interpret
21797         dll-imported symbols as none-local.
21798
21799 2013-03-25  Richard Biener  <rguenther@suse.de>
21800
21801         * tree-ssa-loop-im.c (struct depend): Remove.
21802         (struct lim_aux_data): Make depends a vec of gimples.
21803         (free_lim_aux_data): Adjust.
21804         (add_dependency): Likewise.
21805         (set_level): Likewise.
21806
21807 2013-03-25  Richard Biener  <rguenther@suse.de>
21808
21809         PR middle-end/56434
21810         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
21811         the pointer returned by calls with ECF_MALLOC set.
21812
21813 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
21814
21815         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
21816
21817 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
21818
21819         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
21820         using MMXMODE mode iterator.
21821         (*move<mode>_internal): Merge with *movv2sf_internal and
21822         *movv2sf_internal_rex64 using MMXMODE mode iterator.
21823
21824 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
21825
21826         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
21827         (record_last_mem_set_info): Likewise.
21828
21829         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
21830         of XNEWVEC followed by memset.
21831         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
21832
21833 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
21834
21835         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
21836         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
21837         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
21838         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
21839         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
21840         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
21841         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
21842         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
21843         BARRIER_P instead of GET_CODE.
21844
21845 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21846
21847         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
21848         inaccuracy in the probing code.
21849
21850         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
21851         (ctrapdi4): Likewise.
21852
21853 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21854
21855         * calls.c (expand_call): Add missing guard to code handling return
21856         of non-BLKmode structures in MSB.
21857         * function.c (expand_function_end): Likewise.
21858
21859 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21860
21861         * combine.c (try_combine): Adjust comment.  Do not add the set of
21862         insn #0 if the destination indirectly is set or dies in insn #2.
21863         Tidy up code to distribute a new note.
21864
21865 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
21866
21867         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
21868         also for alternatives 16 and 17.
21869
21870 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
21871
21872         * config/i386/sse.md (*mov<mode>_internal): Merge with
21873         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
21874         Emit insn template depending on type attribute.  Use
21875         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
21876         movd instead of movq mnemonic for interunit moves.  Rewrite mode
21877         attribute calculation.  Remove unit attribute calculation.
21878         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
21879         Set prefix_data16 attribute for DImode ssemov types.
21880         Use Ym instead of y for SSE-MMX conversion alternatives.
21881         Reorder operand constraints.
21882
21883 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
21884
21885         * df.h (df_insn_delete): Adjust prototype.
21886         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
21887         and let it decide whether mark the basic block dirty.
21888         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
21889         * df-scan.c (df_insn_info_delete): New helper function, split
21890         off from df_insn_delete.
21891         (df_scan_free_bb_info): Use it.
21892         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
21893         Likewise.
21894         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
21895         that the insn is actually an insn and it has a non-NULL basic block.
21896         Do not mark basic block dirty if only deleting a DEBUG_INSN.
21897
21898 2013-03-22  Richard Biener  <rguenther@suse.de>
21899
21900         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
21901         dep_ref members.
21902         (mem_ref_alloc): Do not allocate them.
21903         (refs_independent_p): Do not query or maintain a cache.
21904
21905 2013-03-22  Richard Biener  <rguenther@suse.de>
21906
21907         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
21908         (gather_mem_refs_in_loops): Do not compute it.
21909         (analyze_memory_references): Do not allocate it.
21910         (tree_ssa_lim_finalize): Do not free it.
21911         (for_all_locs_in_loop): Do not query all_refs_in_loop.
21912
21913 2013-03-22  Richard Biener  <rguenther@suse.de>
21914
21915         * is-a.h (as_a): Use gcc_checking_assert.
21916
21917 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
21918
21919         * config/aarch64/aarch64.c (aarch64_print_operand): New
21920         format specifier for printing a constant in hex.
21921         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
21922         format specifier for printing second operand.
21923
21924 2013-03-22  Richard Biener  <rguenther@suse.de>
21925
21926         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
21927         bitmaps.
21928         (gather_mem_refs_in_loops): Perform store accumulation here.
21929         (create_vop_ref_mapping_loop): Remove.
21930         (create_vop_ref_mapping): Likewise.
21931         (analyze_memory_references): Initialize refs_stored_in_loop.
21932         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
21933         (record_indep_loop): Remove.
21934         (record_dep_loop): New function.
21935         (ref_indep_loop_p_1): Adjust to only walk over references
21936         in the loop, not its subloops.
21937         (ref_indep_loop_p): Rename to ...
21938         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
21939         maintaining a more fine-grained cache.
21940         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
21941         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
21942
21943 2013-03-22  Richard Biener  <rguenther@suse.de>
21944
21945         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
21946         (struct mem_ref): Make accesses_in_loop a vec of a vec of
21947         aggregate mem_ref_loc.
21948         (free_mem_ref_locs): Inline into ...
21949         (memref_free): ... this and adjust.
21950         (mem_ref_alloc): Adjust.
21951         (mem_ref_locs_alloc): Remove.
21952         (record_mem_ref_loc): Adjust.
21953         (get_all_locs_in_loop): Rewrite into ...
21954         (for_all_locs_in_loop): ... this iterator.
21955         (rewrite_mem_ref_loc): New functor.
21956         (rewrite_mem_refs): Use for_all_locs_in_loop.
21957         (sm_set_flag_if_changed): New functor.
21958         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
21959         (ref_always_accessed): New functor.
21960         (ref_always_accessed_p): Use for_all_locs_in_loop.
21961
21962 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
21963
21964         * tree-pass.h (PROP_gimple_lvec): New.
21965         * passes.c (dump_properties): Handle PROP_gimple_lvec.
21966         (init_optimization_passes): Move pass_lower_vector.
21967         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
21968         PROP_gimple_lvec.
21969         (pass_lower_vector): Provide PROP_gimple_lvec.
21970         (pass_lower_vector_ssa): Likewise.
21971         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
21972
21973 2013-03-21  Mark Wielaard  <mjw@redhat.com>
21974
21975         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
21976
21977 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
21978
21979         * config/i386/i386.md (*movdi_internal): Disparage slightly
21980         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
21981         conversion alternatives.
21982
21983 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
21984
21985         PR middle-end/48087
21986         * diagnostic.def (DK_WERROR): New kind.
21987         * diagnostic.h (werrorcount): Define.
21988         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
21989         promoted to DK_ERROR, increment DK_WERROR counter instead of
21990         DK_ERROR counter.
21991         * toplev.c (toplev_main): Call print_ignored_options even if
21992         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
21993         even if just werrorcount is non-zero.
21994
21995         PR debug/55608
21996         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
21997         on failure.
21998         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
21999         (string_cst_pool_decl): New function.
22000         (optimize_one_addr_into_implicit_ptr): New function.
22001         (resolve_addr_in_expr): Optimize DWARF location expression
22002         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22003         which doesn't live in memory, but has DW_AT_location or
22004         DW_AT_const_value, or refers to a string literal, into
22005         DW_OP_GNU_implicit_pointer.
22006         (optimize_location_into_implicit_ptr): New function.
22007         (resolve_addr): If removing DW_AT_location of a variable because
22008         it was DW_OP_addr of address of the variable, but the variable doesn't
22009         live in memory, try to emit const value attribute for the initializer.
22010
22011 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22012
22013         * tree.h (VECTOR_TYPE_P): New macro.
22014         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22015         TYPE_MODE): Use it.
22016         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22017         VEC_COND_EXPR cannot be lvalues.
22018         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22019
22020 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22021
22022         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22023         Restrict the transformation to equal modes.
22024
22025 2013-03-21  Richard Biener  <rguenther@suse.de>
22026
22027         PR tree-optimization/39326
22028         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22029         (MEM_ANALYZABLE): Adjust.
22030         (record_mem_ref_loc): Move bitmap ops ...
22031         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22032         unanalyzable refs, do not record locations for it.
22033         (analyze_memory_references): Allocate ref zero as shared
22034         unanalyzable ref.
22035         (refs_independent_p): Do not test for unanalyzed mems here.
22036         (ref_indep_loop_p_1): Special-case disambiguation against
22037         the unanalyzed ref.
22038         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22039
22040 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22041
22042         * config/arm/arm-protos.h (tune_params): Add
22043         prefer_neon_for_64bits field.
22044         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22045         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22046         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22047         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22048         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22049         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22050         (arm_option_override): Handle -mneon-for-64bits new option.
22051         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22052         (prefer_neon_for_64bits): Declare new variable.
22053         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22054         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22055         (arch_enabled): Handle new arch types. Remove support for onlya8
22056         and nota8.
22057         (one_cmpldi2): Use new arch names.
22058         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22059         * config/arm/arm.opt (mneon-for-64bits): Add option.
22060         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22061         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22062         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22063         of onlya8.
22064         * doc/invoke.texi (-mneon-for-64bits): Document.
22065
22066 2013-03-21  Richard Biener  <rguenther@suse.de>
22067
22068         PR tree-optimization/39326
22069         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22070         (sort_bbs_in_loop_postorder_cmp): New function.
22071         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22072
22073 2013-03-21  Richard Biener  <rguenther@suse.de>
22074
22075         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22076         (vect_insert_into_interleaving_chain): Likewise.
22077         (vect_drs_dependent_in_basic_block): Inline ...
22078         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22079         split out from ...
22080         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22081         (vect_check_interleaving): Simplify.
22082         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22083         (vect_slp_analyze_data_ref_dependences): ... this new function.
22084         (dr_group_sort_cmp): New function.
22085         (vect_analyze_data_ref_accesses): Compute data-reference groups
22086         here instead of in vect_analyze_data_ref_dependence.  Use
22087         a more efficient algorithm.
22088         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22089         vect_slp_analyze_data_ref_dependences.  Call
22090         vect_analyze_data_ref_accesses earlier.
22091         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22092         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22093         (vect_slp_analyze_data_ref_dependences): New prototype.
22094
22095 2013-03-21  Richard Biener  <rguenther@suse.de>
22096
22097         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22098         ref is stored in the loop.
22099         (find_refs_for_sm): Walk only over all stores.
22100         (store_motion_loop): Allocate from lim_bitmap_obstack.
22101         (store_motion): Likewise.
22102
22103 2013-03-21  Richard Biener  <rguenther@suse.de>
22104
22105         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22106         Update virtual SSA form.
22107
22108 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22109
22110         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22111         * configure: Regenerate.
22112         * config.in: Regenerate.
22113         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22114         if !HAVE_LD_EH_FRAME_CIEV3.
22115
22116 2013-03-21  Richard Biener  <rguenther@suse.de>
22117
22118         * tree-cfg.c (verify_expr_no_block): New function.
22119         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22120         nor DECL_VALUE_EXPR have locations with associated blocks.
22121         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22122         (clear_unused_block_pointer): Remove code dealing with
22123         blocks in DECL_DEBUG_EXPR locations.
22124
22125 2013-03-21  Richard Biener  <rguenther@suse.de>
22126
22127         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22128         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22129         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22130         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22131         instead of DECL_DEBUG_EXPR_IS_FROM.
22132         * gimplify.c (gimplify_modify_expr): Likewise.
22133         * tree-cfg.c (verify_expr_location_1): Likewise.
22134         * tree-complex.c (create_one_component_var): Likewise.
22135         * tree-sra.c (create_access_replacement): Likewise.
22136         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22137         (clear_unused_block_pointer): Likewise.
22138         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22139         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22140         * var-tracking.c (var_debug_decl): Likewise.
22141         (track_expr_p): Likewise.
22142         * tree-inline.c (add_local_variables): Likewise.  Set
22143         DECL_HAS_DEBUG_EXPR_P after copying it.
22144         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22145         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22146
22147 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22148
22149         PR bootstrap/56656
22150         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22151         * configure: Regenerate.
22152         * config.in: Regenerate.
22153         * config/i386/i386.md (*movdf_internal): Use
22154         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22155         movd instead of movq mnemonic for interunit moves.
22156         (*movdi_internal): Ditto.
22157
22158 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22159
22160         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22161         (abd<mode>_3): New pattern.
22162         (aba<mode>_3): New pattern.
22163         (fabd<mode>_3): New pattern.
22164
22165 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22166
22167         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22168         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22169         occurrence of REGISTER_PREFIX as its empty string.
22170
22171 2013-03-20  Jeff Law  <law@redhat.com>
22172
22173         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22174         addititional equivalences for equality comparisons between an SSA_NAME
22175         and a constant where the SSA_NAME was set from a widening conversion.
22176
22177 2013-03-20  Walter Lee  <walt@tilera.com>
22178
22179         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22180
22181 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22182
22183         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22184         depending on type attribute.
22185         (*movti_internal): Ditto.
22186         (*movtf_internal): Ditto.
22187         (*movxf_internal): Ditto.
22188         (*movdf_internal): Ditto.
22189         (*movsf_internal): Ditto.
22190
22191 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22192
22193         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22194         maybe_vex for sselog1 and ssemov types.
22195         (*movdi_internal): Reorder operand constraints.
22196         (*movsi_internal): Ditto.  Set prefix attribute to
22197         maybe_vex for sselog1 and ssemov types.
22198         (*movtf_internal): Set prefix attribute to maybe_vex
22199         for sselog1 and ssemov types.
22200         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22201         DImode ssemov types.  Reorder operand constraints.
22202         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22203         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22204         attribute for SImode ssemov types.  Reorder operand constraints.
22205
22206 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22207
22208         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22209         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22210
22211 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22212
22213         * config/rs6000/predicates.md (indexed_address, update_address_mem
22214         update_indexed_address_mem): New predicates.
22215         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22216         attribute for load/store instructions.
22217         * config/rs6000/dfp.md (movsd_store): Likewise.
22218         (movsd_load): Likewise.
22219         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22220         (unnamed HI->DI extend define_insn): Likewise.
22221         (unnamed SI->DI extend define_insn): Likewise.
22222         (unnamed QI->SI extend define_insn): Likewise.
22223         (unnamed QI->HI extend define_insn): Likewise.
22224         (unnamed HI->SI extend define_insn): Likewise.
22225         (unnamed HI->SI extend define_insn): Likewise.
22226         (extendsfdf2_fpr): Likewise.
22227         (movsi_internal1): Likewise.
22228         (movsi_internal1_single): Likewise.
22229         (movhi_internal): Likewise.
22230         (movqi_internal): Likewise.
22231         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22232         attribute for load/store instructions.
22233         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22234         instructions.
22235         (mov<mode>_softfloat): Likewise.
22236         (mov<mode>_hardfloat32): Likewise.
22237         (mov<mode>_hardfloat64): Likewise.
22238         (mov<mode>_softfloat64): Likewise.
22239         (movdi_internal32): Likewise.
22240         (movdi_internal64): Likewise.
22241         (probe_stack_<mode>): Likewise.
22242
22243 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22244
22245         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22246         floating point, and decimal floating point to reload iterator.
22247
22248         * config/rs6000/constraints.md (wl constraint): New constraints to
22249         return FLOAT_REGS if certain options are used to reduce the number
22250         of separate patterns that exist in the file.
22251         (wx constraint): Likewise.
22252         (wz constraint): Likewise.
22253
22254         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22255         -mdebug=reg, print wg, wl, wx, and wz constraints.
22256         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22257         Initialize the reload functions for 64-bit binary/decimal floating
22258         point types.
22259         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22260         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22261         create the buffer on the stack to overcome not having a 32-bit
22262         load and store.
22263         (rs6000_emit_move): Likewise.
22264         (rs6000_secondary_memory_needed_rtx): Likewise.
22265         (rs6000_alloc_sdmode_stack_slot): Likewise.
22266         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22267         via xxlxor, just like DFmode 0.0.
22268
22269         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22270         define as 1 if we are running on a power7 or newer.
22271         (enum r6000_reg_class_enum): Add new constraints.
22272
22273         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22274         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22275         with other moves by using conditional constraits (wg).  Use LFIWZX
22276         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22277         (movsd splitter): Likewise.
22278         (movsd_hardfloat): Likewise.
22279         (movsd_softfloat): Likewise.
22280
22281         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22282         binary and decimal floating point moves.
22283         (fmove_ok): New attributes to combine binary and decimal floating
22284         point moves, and to combine power6x (mfpgpr) moves along normal
22285         floating moves.
22286         (real_value_to_target): Likewise.
22287         (f32_lr): Likewise.
22288         (f32_lm): Likewise.
22289         (f32_li): Likewise.
22290         (f32_sr): Likewise.
22291         (f32_sm): Likewise.
22292         (f32_si): Likewise.
22293         (movsf): Combine binary and decimal floating point moves.  Combine
22294         power6x (mfpgpr) moves with other moves by using conditional
22295         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22296         (mov<mode> for SFmode/SDmode); Likewise.
22297         (SFmode/SDmode splitters): Likewise.
22298         (movsf_hardfloat): Likewise.
22299         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22300         (movsf_softfloat): Likewise.
22301         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22302
22303         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22304         wx and wz constraints.
22305
22306         * config/rs6000/constraints.md (wg constraint): New constraint to
22307         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22308
22309         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22310         constraint.
22311
22312         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22313         -mdebug=reg, print wg, wl, wx, and wz constraints.
22314         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22315         Initialize the reload functions for 64-bit binary/decimal floating
22316         point types.
22317         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22318         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22319         create the buffer on the stack to overcome not having a 32-bit
22320         load and store.
22321         (rs6000_emit_move): Likewise.
22322         (rs6000_secondary_memory_needed_rtx): Likewise.
22323         (rs6000_alloc_sdmode_stack_slot): Likewise.
22324         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22325         via xxlxor, just like DFmode 0.0.
22326
22327         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22328         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22329         with other moves by using conditional constraits (wg).  Use LFIWZX
22330         and STFIWX for loading SDmode on power7.
22331         (movdd splitters): Likewise.
22332         (movdd_hardfloat32): Likewise.
22333         (movdd_softfloat32): Likewise.
22334         (movdd_hardfloat64_mfpgpr): Likewise.
22335         (movdd_hardfloat64): Likewise.
22336         (movdd_softfloat64): Likewise.
22337
22338         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22339         64-bit binary and decimal floating point moves.
22340         (FMOVE64X): Likewise.
22341         (movdf): Combine 64-bit binary and decimal floating point moves.
22342         Combine power6x (mfpgpr) moves with other moves by using
22343         conditional constraits (wg).
22344         (mov<mode> for DFmode/DDmode): Likewise.
22345         (DFmode/DDmode splitters): Likewise.
22346         (movdf_hardfloat32): Likewise.
22347         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22348         (movdf_softfloat32): Likewise.
22349         (movdf_hardfloat64_mfpgpr): Likewise.
22350         (movdf_hardfloat64): Likewise.
22351         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22352         (movdf_softfloat64): Likewise.
22353         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22354         (reload_<mode>_load): Move to later in the file so they aren't in
22355         the middle of the floating point move insns.
22356         (reload_<mode>_store): Likewise.
22357
22358         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22359         constraint.
22360
22361         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22362         constraint if -mdebug=reg.
22363         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22364         Enable using dd reload support if needed.
22365
22366         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22367         binary and decimal floating point moves in rs6000.md.
22368         (movtd_internal): Likewise.
22369
22370         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22371         decimal floating point moves.
22372         (movtf): Likewise.
22373         (movtf_internal): Likewise.
22374         (mov<mode>_internal, TDmode/TFmode): Likewise.
22375         (movtf_softfloat): Likewise.
22376         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22377
22378         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22379         movdi_internal64, using wg constraint for move direct operations.
22380         (movdi_internal64): Likewise.
22381
22382         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22383         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22384         the various virtual registers. Use GPR/FPR first/last values,
22385         instead of hard coding the register numbers.  Print which modes
22386         have reload functions registered.
22387         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22388         settings before/after setting cpu, target and subtarget settings.
22389         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22390         and for secondary reload failures in rs6000_secondary_reload_inner.
22391         (rs6000_secondary_reload_fail): Likewise.
22392         (rs6000_secondary_reload_inner): Likewise.
22393
22394         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22395         macros for first/last GPR and FPR registers.
22396         (LAST_GPR_REGNO): Likewise.
22397         (FIRST_FPR_REGNO): Likewise.
22398         (LAST_FPR_REGNO): Likewise.
22399
22400         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22401         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22402         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22403         (vcond<mode><mode>): Likewise.
22404         (vcondu<mode><mode>): Likewise.
22405         (vector_gtu<mode>): Likewise.
22406         (vector_gte<mode>): Likewise.
22407         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22408         to prevent the compiler from converting DImode operations to TImode.
22409         (ior<mode>3): Likewise.
22410         (and<mode>3): Likewise.
22411         (one_cmpl<mode>2): Likewise.
22412         (nor<mode>3): Likewise.
22413         (andc<mode>3): Likewise.
22414
22415         * config/rs6000/constraints.md (wt constraint): New constraint
22416         that returns VSX_REGS if TImode is allowed in VSX registers.
22417
22418         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22419         constant under VSX.
22420
22421         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22422         similar to TImode, but it is restricted to being in the GPRs.
22423
22424         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22425         TImode to occupy a single VSX register.
22426
22427         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22428         -mvsx-timode for power7/power8.
22429         (power7 cpu): Likewise.
22430         (power8 cpu): Likewise.
22431
22432         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22433         sure that TFmode/TDmode take up two registers if they are ever
22434         allowed in the upper VSX registers.
22435         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22436         registers.
22437         (rs6000_init_hard_regno_mode_ok): Likewise.
22438         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22439         constraint.  Print if LRA is turned on.
22440         (rs6000_option_override_internal): Give an error if -mvsx-timode
22441         and VSX is not enabled.
22442         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22443         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22444         to reg+offset addressing.  Use PTImode when checking offset
22445         addresses for validity.
22446         (reg_offset_addressing_ok_p): Likewise.
22447         (rs6000_legitimate_offset_address_p): Likewise.
22448         (rs6000_legitimize_address): Likewise.
22449         (rs6000_legitimize_reload_address): Likewise.
22450         (rs6000_legitimate_address_p): Likewise.
22451         (rs6000_eliminate_indexed_memrefs): Likewise.
22452         (rs6000_emit_move): Likewise.
22453         (rs6000_secondary_reload): Likewise.
22454         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22455         reloads to fpr registers to continue to use reg+offset addressing,
22456         but 64-bit reloads to altivec registers need reg+reg addressing.
22457         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22458         it.  Treat LO_SUM like a PLUS operation.
22459         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22460         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22461         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22462         registers to share a register with a smaller sized type, since VSX
22463         puts scalars in the upper 64-bits.
22464         (print_operand): Add support for PTImode.
22465         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22466         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22467         registers, but don't have arithmetic support.
22468         (rs6000_memory_move_cost): Add test for VSX.
22469         (rs6000_opt_masks): Add -mvsx-timode.
22470
22471         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22472         for TImode.
22473         (VSs): Likewise.
22474         (VSr): Use wt constraint for TImode.
22475         (VSv): Drop TImode support.
22476         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22477         (vsx_movti_64bit): Likewise.
22478         (vsx_movti_32bit): Likewise.
22479         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22480         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22481         one '?' on the appropriate output constraint.  Do not allow TImode
22482         logical operations on 32-bit systems.
22483         (vsx_ior<mode>3): Likewise.
22484         (vsx_xor<mode>3): Likewise.
22485         (vsx_one_cmpl<mode>2): Likewise.
22486         (vsx_nor<mode>3): Likewise.
22487         (vsx_andc<mode>3): Likewise.
22488         (vsx_concat_<mode>): Likewise.
22489         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22490
22491         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22492         OPTION_MASK_VSX_TIMODE.
22493         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22494         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22495
22496         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22497         (TI2 iterator): New iterator for TImode, PTImode.
22498         (wd mode attribute): Add values for vector types.
22499         (movti_string): Replace TI move operations with operations for TImode
22500         and PTImode.  Add support for TImode being allowed in VSX registers.
22501         (mov<mode>_string, TImode/PTImode): Likewise.
22502         (movti_ppc64): Likewise.
22503         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22504         (TI mode splitters): Likewise.
22505
22506         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22507         constraint.
22508
22509 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22510
22511         PR tree-optimization/56355
22512         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22513         Also handle integers with undefined overflow.
22514
22515 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22516             Maciej W. Rozycki  <macro@codesourcery.com>
22517             Tom de Vries  <tom@codesourcery.com>
22518             Nathan Sidwell  <nathan@codesourcery.com>
22519             Iain Sandoe  <iain@codesourcery.com>
22520             Nathan Froyd  <froydnj@codesourcery.com>
22521             Chao-ying Fu  <fu@mips.com>
22522
22523         * doc/extend.texi: (micromips, nomicromips, nocompression):
22524         Document new function attributes.
22525         * doc/invoke.texi (minterlink-compressed, mmicromips,
22526         m14k, m14ke, m14kec): Document new options.
22527         (minterlink-mips16): Update documentation.
22528         * doc/md.texi (ZC, ZD): Document new constraints.
22529         * configure.ac (gcc_cv_as_micromips): Check if linker
22530         supports the .set micromips directive.
22531         * configure: Regenerate.
22532         * config.in: Regenerate.
22533         * config/mips/mips-tables.opt: Regenerate.
22534         * config/mips/micromips.md: New file.
22535         * constraints.md (ZC, ZD): New constraints.
22536         * config/mips/predicates.md (movep_src_register): New predicate.
22537         (movep_src_operand): New predicate.
22538         (non_volatile_mem_operand): New predicate.
22539         * config/mips/mips.md (multimem): New type.
22540         (length): Differentiate between 17-bit and 18-bit branch offsets.
22541         (MOVEP1, MOVEP2): New mode iterator.
22542         (mov_<load>l): Use ZC constraint.
22543         (mov_<load>r): Likewise.
22544         (mov_<store>l): Likewise.
22545         (mov_<store>r): Likewise.
22546         (*branch_equality<mode>_inverted): Add microMIPS support.
22547         (*branch_equality<mode>): Likewise.
22548         (*jump_absolute): Likewise.
22549         (indirect_jump_<mode>): Likewise.
22550         (tablejump_<mode>): Likewise.
22551         (<optab>_internal): Likewise.
22552         (sibcall_internal): Likewise.
22553         (sibcall_value_internal): Likewise.
22554         (prefetch): Use constraint ZD.
22555         * config/mips/mips.opt (minterlink-compressed): New option.
22556         (minterlink-mips16): Now an alias for minterlink-compressed.
22557         (mmicromips): New option.
22558         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22559         (compare_and_swap_12): Likewise.
22560         (sync_add<mode>): Likewise.
22561         (sync_<optab>_12): Likewise.
22562         (sync_old_<optab>_12): Likewise.
22563         (sync_new_<optab>_12): Likewise.
22564         (sync_nand_12): Likewise.
22565         (sync_old_nand_12): Likewise.
22566         (sync_new_nand_12): Likewise.
22567         (sync_sub<mode>): Likewise.
22568         (sync_old_add<mode>): Likewise.
22569         (sync_old_sub<mode>): Likewise.
22570         (sync_new_add<mode>): Likewise.
22571         (sync_new_sub<mode>): Likewise.
22572         (sync_<optab><mode>): Likewise.
22573         (sync_old_<optab><mode>): Likewise.
22574         (sync_new_<optab><mode>): Likewise.
22575         (sync_nand<mode>): Likewise.
22576         (sync_old_nand<mode>): Likewise.
22577         (sync_new_nand<mode>): Likewise.
22578         (sync_lock_test_and_set<mode>): Likewise.
22579         (test_and_set_12): Likewise.
22580         (atomic_compare_and_swap<mode>): Likewise.
22581         (atomic_exchange<mode>_llsc): Likewise.
22582         (atomic_fetch_add<mode>_llsc): Likewise.
22583         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22584         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22585         (umips_save_restore_pattern_p): Likewise.
22586         (umips_load_store_pair_p): Likewise.
22587         (umips_output_load_store_pair): Likewise.
22588         (umips_movep_target_p): Likewise.
22589         (umips_12bit_offset_address_p): Likewise.
22590         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22591         (mips_base_mips16): Rename this...
22592         (mips_base_compression_flags): ...to this. Update all uses.
22593         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22594         (mips_mips16_decl_p): Delete.
22595         (mips_nomips16_decl_p): Delete.
22596         (mips_get_compress_on_flags): New function.
22597         (mips_get_compress_off_flags): New function.
22598         (mips_get_compress_mode): New function.
22599         (mips_get_compress_on_name): New function.
22600         (mips_get_compress_off_name): New function.
22601         (mips_insert_attributes): Support multiple compression types.
22602         (mips_merge_decl_attributes): Likewise.
22603         (umips_12bit_offset_address_p): New function.
22604         (mips_start_function_definition): Emit .set micromips directive.
22605         (mips_call_may_need_jalx_p): New function.
22606         (mips_function_ok_for_sibcall): Add microMIPS support.
22607         (mips_print_operand_punctuation): Support short delay slots and
22608         compact jumps.
22609         (umips_swm_mask, umips_swm_encoding): New.
22610         (umips_build_save_restore): New function.
22611         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22612         (was_mips16_p): Remove.
22613         (old_compression_mode): New.
22614         (mips_set_compression_mode): New function.
22615         (mips_set_current_function): Add microMIPS support.
22616         (mips_option_override): Likewise.
22617         (umips_save_restore_pattern_p): New function.
22618         (umips_output_save_restore): New function.
22619         (umips_load_store_pair_p_1): New function.
22620         (umips_load_store_pair_p): New function.
22621         (umips_output_load_store_pair_1): New function.
22622         (umips_output_load_store_pair): New function.
22623         (umips_movep_target_p) New function.
22624         (mips_prepare_pch_save): Add microMIPS support.
22625         * config/mips/mips.h (TARGET_COMPRESSION): New.
22626         (TARGET_CPU_CPP_BUILTINS): Update macro
22627         to use new compression flags and to support microMIPS.
22628         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22629         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22630         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22631         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22632         (ASM_SPEC): Support mmicromips and mno-micromips.
22633         (M16STORE_REG_P): New macro.
22634         (MIPS_CALL): Support TARGET_MICROMIPS.
22635         (MICROMIPS_J): New macro.
22636         (mips_base_mips16): Rename this...
22637         (mips_base_compression_flags): ...to this.
22638         (UMIPS_12BIT_OFFSET_P): New macro.
22639         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22640         (MULTILIB_DIRNAMES): Likewise.
22641 2013-03-20  Richard Biener  <rguenther@suse.de>
22642
22643         PR tree-optimization/56661
22644         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22645         the result does not have to be distinct.
22646
22647 2013-03-20  Richard Biener  <rguenther@suse.de>
22648
22649         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22650         remap_gimple_op_r.
22651
22652 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22653             Steven Bosscher  <steven@gcc.gnu.org>
22654
22655         PR rtl-optimization/56605
22656         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22657
22658 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22659
22660         PR bootstrap/56656
22661         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22662         that require movd instead of movq.
22663
22664 2013-03-20  Richard Biener  <rguenther@suse.de>
22665
22666         * tree-ssa-structalias.c (struct variable_info): Add pointer
22667         to the first field of an aggregate with sub-vars.  Make
22668         this and the pointer to the next subfield its ID.
22669         (vi_next): New function.
22670         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22671         storedanything_id, integer_id): Increment by one.
22672         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22673         get_call_clobber_vi): Adjust.
22674         (solution_set_expand): Simplify and speedup.
22675         (solution_set_add): Inline into ...
22676         (set_union_with_increment): ... this.  Adjust accordingly.
22677         (do_sd_constraint): Likewise.
22678         (do_ds_constraint): Likewise.
22679         (do_complex_constraint): Simplify.
22680         (build_pred_graph): Adjust.
22681         (solve_graph): Likewise.  Simplify and speedup.
22682         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22683         get_constraint_for_component_ref, get_constraint_for_1,
22684         first_vi_for_offset, first_or_preceding_vi_for_offset,
22685         create_function_info_for, create_variable_info_for_1,
22686         create_variable_info_for, intra_create_variable_infos): Adjust.
22687         (init_base_vars): Push NULL for ID zero.
22688         (compute_points_to_sets): Adjust.
22689
22690 2013-03-20  Richard Biener  <rguenther@suse.de>
22691
22692         * cfgloop.c (verify_loop_structure): Streamline and avoid
22693         ICEing on corrupt loop tree.
22694         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22695         loop tree.
22696
22697 2013-03-20  Richard Biener  <rguenther@suse.de>
22698
22699         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22700         check whether an SSA update is needed.
22701
22702 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22703
22704         * config/mips/constraints.md (T): Rename to...
22705         (Yf): ...this.
22706         (U): Rename to...
22707         (Yd): ...this.
22708         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22709         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22710
22711 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22712
22713         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22714         (*subsi3_carryin_uxtw): Likewise.
22715
22716 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22717
22718         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
22719         (*rorsi3_insn_uxtw): Likewise.
22720
22721 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22722
22723         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
22724         (*extrsi5_insn_uxtw): Likewise.
22725
22726 2013-03-19  Richard Biener  <rguenther@suse.de>
22727
22728         PR tree-optimization/56273
22729         * passes.c (init_optimization_passes): Move second VRP after DOM.
22730
22731 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
22732
22733         * config/i386/i386.md (*movti_internal): Merge from
22734         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
22735         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
22736         nox64 isa attributes.
22737
22738 2013-03-18  Richard Biener  <rguenther@suse.de>
22739
22740         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
22741         (unite): Likewise.
22742         (merge_node_constraints): Likewise.
22743         (build_succ_graph): Likewise.
22744         (valid_graph_edge): Inline into single caller.
22745         (unify_nodes): Likewise.  Use bitmap_set_bit return value
22746         and cache varinfo.
22747         (scc_visit): Fix formatting and variable use.
22748         (do_sd_constraint): Use gcc_checking_assert.
22749         (do_ds_constraint): Likewise.
22750         (do_complex_constraint): Likewise.
22751         (condense_visit): Likewise.  Cleanup.
22752         (dump_pred_graph): New function.
22753         (perform_var_substitution): Dump the pred-graph before
22754         variable substitution.
22755         (find_equivalent_node): Use gcc_checking_assert.
22756         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
22757
22758 2013-03-18  Richard Biener  <rguenther@suse.de>
22759
22760         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
22761         Remove cond_expr_stmt_list argument and do not gimplify the
22762         built expression.
22763         (vect_loop_versioning): Adjust.
22764         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
22765         Cleanup to use less temporaries.
22766         (vect_create_data_ref_ptr): Cleanup.
22767
22768 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
22769
22770         PR tree-optimization/56635
22771         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
22772         require types_compatible_p types.
22773
22774 2013-03-18  Nick Clifton  <nickc@redhat.com>
22775
22776         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
22777         spurious backslash.
22778
22779         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
22780         Add missing line to comment describing function.
22781
22782 2013-03-18  Richard Biener  <rguenther@suse.de>
22783
22784         PR tree-optimization/56210
22785         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22786         Handle string / character search functions.
22787         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
22788
22789 2013-03-18  Richard Biener  <rguenther@suse.de>
22790
22791         PR middle-end/56483
22792         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
22793         and implement properly.
22794         * gimple.h (gimple_cond_single_var_p): Remove.
22795
22796 2013-03-18  Richard Biener  <rguenther@suse.de>
22797
22798         * tree-data-ref.h (find_data_references_in_loop): Declare.
22799         * tree-data-ref.c (get_references_in_stmt): Use a stack
22800         vector pre-allocated in the callers.
22801         (find_data_references_in_stmt): Adjust.
22802         (graphite_find_data_references_in_stmt): Likewise.
22803         (create_rdg_vertices): Likewise.
22804         (find_data_references_in_loop): Export.
22805         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
22806         Compute dependences here...
22807         (vect_analyze_data_refs): ...not here.  When we encounter
22808         a non-vectorizable data reference in basic-block vectorization
22809         truncate the data reference vector.  Do not bother to
22810         fixup data-dependence information for gather loads.
22811         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
22812         of data references, as reported.
22813
22814 2013-03-18  Richard Biener  <rguenther@suse.de>
22815
22816         PR tree-optimization/3713
22817         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
22818         has_constants and expr.
22819         (stmt_has_constants): Properly valueize SSA names when deciding
22820         whether the stmt has constants.
22821
22822 2013-03-18  Richard Biener  <rguenther@suse.de>
22823
22824         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
22825         whole function when there is nothing to do.
22826         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
22827         * tree-vectorizer.c (vectorize_loops): Update virtual and
22828         loop-closed SSA once.
22829         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
22830
22831 2013-03-18  Richard Biener  <rguenther@suse.de>
22832
22833         PR middle-end/56113
22834         * domwalk.c (bb_postorder): New global static.
22835         (cmp_bb_postorder): New function.
22836         (walk_dominator_tree): Replace scheme imposing an order for
22837         visiting dominator sons by one sorting them at the time they
22838         are pushed on the stack.
22839
22840 2013-03-18  Richard Biener  <rguenther@suse.de>
22841
22842         PR tree-optimization/39326
22843         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
22844         (struct mem_ref): Replace mem member with ao_ref typed member.
22845         (MEM_ANALYZABLE): Adjust.
22846         (memref_eq): Likewise.
22847         (mem_ref_alloc): Likewise.
22848         (gather_mem_refs_stmt): Likewise.
22849         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
22850         (execute_sm_if_changed_flag_set): Adjust.
22851         (execute_sm): Likewise.
22852         (ref_always_accessed_p): Likewise.
22853         (refs_independent_p): Likewise.
22854         (can_sm_ref_p): Likewise.
22855
22856 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
22857
22858         PR c/56566
22859         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
22860         return 1 even for !unsignedp.
22861
22862 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22863
22864         * config/i386/i386.md (isa): Add x64 and nox64.
22865         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
22866         (*pushtf): Enable *roF alternative for x64 isa only.
22867         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
22868         mode attribute of integer alternatives to DImode for TARGET_64BIT.
22869         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
22870         (*movtf_internal): Merge from *movtf_internal_rex64 and
22871         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
22872         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
22873         nox64 isa attributes.
22874         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
22875         nox64 isa attributes.
22876         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
22877
22878 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22879
22880         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22881
22882 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
22883
22884         PR target/56640
22885         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
22886         class names.  Remove trailing comma after "ALL_REGS".
22887
22888 2013-03-16  Jan Hubicka  <jh@suse.cz>
22889
22890         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
22891         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
22892         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
22893         of cgraph_get_create_node.
22894         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
22895
22896 2013-03-16  Jason Merrill  <jason@redhat.com>
22897
22898         PR debug/49090
22899         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
22900         with DW_AT_default_value.
22901
22902 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
22903
22904         * BASE-VER: Set to 4.9.0.
22905
22906 2013-03-14  Andi Kleen  <ak@linux.intel.com>
22907
22908         PR target/56619
22909         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
22910         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
22911         Document _x* TSX intrinsics.
22912
22913 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
22914             David Holsgrove  <david.holsgrove@xilinx.com>
22915
22916         * configure.ac: Add MicroBlaze TLS support detection.
22917         * configure: Regenerate.
22918         * config/microblaze/microblaze-protos.h
22919         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
22920         symbol_mentioned_p, label_mentioned_p): Add prototypes.
22921         * config/microblaze/microblaze.c (microblaze_address_type): Add
22922         ADDRESS_TLS and tls_reloc address types.
22923         (microblaze_address_info): Add tls_reloc.
22924         (TARGET_HAVE_TLS): Define.
22925         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
22926          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
22927          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
22928         load_tls_operand,  microblaze_call_tls_get_addr,
22929         microblaze_legitimize_tls_address): New functions.
22930         (microblaze_classify_unspec): Handle UNSPEC_TLS.
22931         (get_base_reg): Use microblaze_tls_symbol_p.
22932         (microblaze_classify_address): Handle TLS.
22933         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
22934         label_mentioned_p and microblaze_tls_referenced_p.
22935         (microblaze_legitimize_address): Handle TLS.
22936         (microblaze_address_insns): Handle ADDRESS_TLS.
22937         (pic_address_needs_scratch): Handle TLS.
22938         (print_operand_address): Handle TLS.
22939         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
22940         (microblaze_expand_move): Handle TLS.
22941         (microblaze_legitimate_constant_p): Check
22942         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
22943         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
22944         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
22945         (PIC_OFFSET_TABLE_REGNUM): Set.
22946         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
22947         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
22948         (addsi3, movsi_internal2, movdf_internal): Update constraints
22949         * config/microblaze/predicates.md (arith_plus_operand): Define
22950         (move_operand): Redefine as move_src_operand,
22951         check microblaze_tls_referenced_p.
22952
22953 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
22954
22955         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
22956         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
22957
22958 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
22959
22960         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
22961         CC mode for AND.
22962
22963 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
22964
22965         PR tree-optimization/53265
22966         * common.opt (Waggressive-loop-optimizations): New option.
22967         * tree-ssa-loop-niter.c: Include tree-pass.h.
22968         (do_warn_aggressive_loop_optimizations): New function.
22969         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
22970         if number_of_latch_executions returned constant.
22971         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
22972         early.  If number_of_latch_executions returned constant, set
22973         nb_iterations_upper_bound back to it.
22974         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
22975         field.
22976         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
22977         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
22978
22979         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
22980         (MULTILIB_OSDIRNAMES): Set.
22981         * genmultilib: If defaultosdirname doesn't start with :: , set
22982         defaultosdirname2 instead, clear it and emit two . multilib_raw
22983         entries instead of just one.
22984
22985 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
22986
22987         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
22988         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
22989         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
22990         (SUBTARGET_OVERRIDE_OPTIONS): New.
22991
22992 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
22993
22994         PR target/49880
22995         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
22996         (musermode): Convert to Var(TARGET_USERMODE).
22997         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
22998         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
22999         * config/sh/sh.c (sh_option_override): Use
23000         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23001         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23002         condition.
23003         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23004         TARGET_SH4.
23005         (udivsi3_i4_single, divsi3_i4_single): Use
23006         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23007
23008 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23009
23010         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23011         default setting.
23012
23013 2013-03-13  Richard Biener  <rguenther@suse.de>
23014
23015         PR tree-optimization/56608
23016         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23017         calls when vectorizing basic-blocks.
23018
23019 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23020
23021         PR plugins/45078
23022         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23023         tm_file.
23024
23025 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23026
23027         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23028
23029 2013-03-11  Jan Hubicka  <jh@suse.cz>
23030
23031         PR lto/56557
23032         * lto-streamer-out.c (output_symbol_p): Skip references from
23033         constructors of external variables.
23034
23035 2013-03-11  Jan Hubicka  <jh@suse.cz>
23036
23037         PR middle-end/56571
23038         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23039         from pseudos.
23040         * emit-rtl.c (verify_rtx_sharing): Likewise.
23041         (copy_insn_1): Likewise.
23042         * rtl.c (copy_rtx): Likewise.
23043
23044 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23045
23046         PR target/56591
23047         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23048         output_operand_lossage message.
23049
23050 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23051
23052         PR target/56470
23053         * arm.c (shift_op): Validate RTL pattern on the fly.
23054         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23055         the RTL.
23056
23057 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23058
23059         PR target/56347
23060         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23061         new call patterns that clobber %fr12.
23062         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23063         split and postreload patterns.
23064         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23065         registers %fr12 and %fr12R as call used.
23066
23067 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23068
23069         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23070         (canon_address, record_store, replace_read, check_mem_read_rtx,
23071         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23072         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23073         rest_of_handle_dse): Likewise.
23074
23075 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23076
23077         PR middle-end/56524
23078         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23079         Add base_optabs.
23080         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23081         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23082         (save_optabs_if_changed): Replace with...
23083         (init_tree_optimization_optabs): ...this.
23084         * optabs.c (save_optabs_if_changed): Rename to...
23085         (init_tree_optimization_optabs): ...this.  Take the optimization node
23086         as argument.  Do nothing if the base optabs are already correct.
23087         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23088         to recompute optabs.
23089         * function.h (function): Remove optabs field.
23090         * function.c (invoke_set_current_function_hook): Call
23091         init_tree_optimization_optabs.  Use the result to initialize
23092         this_fn_optabs.
23093
23094 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23095
23096         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23097         if GTMA_HAS_NO_INSTRUMENTATION.
23098         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23099         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23100         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23101         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23102         GTMA_HAS_NO_INSTRUMENTATION.
23103
23104 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23105
23106         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23107         libasan_preinit.o.
23108
23109 2013-03-08  Marek Polacek  <polacek@redhat.com>
23110             Jakub Jelinek  <jakub@redhat.com>
23111
23112         PR tree-optimization/56478
23113         * predict.c (is_comparison_with_loop_invariant_p): Change the
23114         type of loop_step to tree.
23115         (predict_loops): Adjust.
23116         (predict_iv_comparison): Perform the computations on double_ints.
23117
23118 2013-03-08  Richard Biener  <rguenther@suse.de>
23119
23120         PR tree-optimization/56570
23121         * tree-cfg.c (verify_expr_location_1): Verify locations for
23122         DECL_DEBUG_EXPR.
23123         * tree-sra.c (create_access_replacement): Strip locations
23124         from DECL_DEBUG_EXPRs.
23125
23126 2013-03-08  Richard Biener  <rguenther@suse.de>
23127
23128         * tree-inline.c (expand_call_inline): Do not associate
23129         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23130         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23131
23132 2013-03-08  Richard Biener  <rguenther@suse.de>
23133
23134         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23135         or block changes with -Og.  Fix for location / block encoding
23136         changes and PHI arguments with locations.
23137
23138 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23139
23140         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23141         for all counters.
23142         (struct output_info): Likewise.
23143         (register_overhead): Remove bad gcc_assert.
23144         (bitmap_find_bit): If there is only a single bitmap element, do not
23145         count a miss as a search.
23146         (print_statistics): Update for counter type changes.
23147         (dump_bitmap_statistics): Likewise.  Print headers such that they
23148         are properly lined up with the printed counters.
23149
23150 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23151
23152         PR tree-optimization/56559
23153         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23154         check that it has only a single use.
23155
23156 2013-03-07  Richard Biener  <rguenther@suse.de>
23157
23158         * doc/invoke.texi (fwhole-program): Discourage use in combination
23159         with -flto.
23160
23161 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23162
23163         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23164
23165         PR tree-optimization/56539
23166         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23167         instead of GSI_CONTINUE_LINKING as last argument to
23168         force_gimple_operand_gsi.  Adjust function comment.
23169
23170         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23171         aarch64-cores.def.
23172
23173         PR middle-end/56548
23174         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23175         promoted mode, convert the result back to the original mode.
23176
23177 2013-03-06  Richard Biener  <rguenther@suse.de>
23178
23179         PR middle-end/56294
23180         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23181         (insert_updated_phi_nodes_compare_uids): New function.
23182         (update_ssa): Sort symbols_to_rename after UID before
23183         traversing it to insert PHI nodes.
23184
23185 2013-03-06  Richard Biener  <rguenther@suse.de>
23186
23187         PR middle-end/50494
23188         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23189         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23190
23191         Revert
23192         2013-02-13  Richard Biener  <rguenther@suse.de>
23193
23194         PR lto/50494
23195         * varasm.c (output_constant_def_1): Get the decl representing
23196         the constant as argument.
23197         (output_constant_def): Wrap output_constant_def_1.
23198         (make_decl_rtl): Use output_constant_def_1 with the decl
23199         representing the constant.
23200         (build_constant_desc): Optionally re-use a decl already
23201         representing the constant.
23202         (tree_output_constant_def): Adjust.
23203
23204 2013-03-06  Joey Ye  <joey.ye@arm.com>
23205
23206         PR lto/50293
23207         * gcc.c (convert_white_space): New function.
23208         (main): Handles white space in function name.
23209
23210 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23211
23212         PR target/56529
23213         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23214         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23215         to SH_DIV_CALL_TABLE for TARGET_SH2.
23216         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23217         list.
23218         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23219         call-table options.
23220
23221 2013-03-05  Sterling Augustine  <saugustine@google.com>
23222             Cary Coutant  <ccoutant@google.com>
23223
23224         PR debug/55364
23225         * dwarf2out.c (resolve_addr): Don't call
23226         remove_loc_list_addr_table_entries a second time for the same
23227         expression.
23228
23229 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23230
23231         PR debug/56510
23232         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23233         (avoid_complex_debug_insns): New function.
23234         (expand_debug_locations): Call it.
23235
23236         PR rtl-optimization/56484
23237         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23238         lifetimes of hard registers on small register class machines.
23239
23240 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23241
23242         * config/microblaze/microblaze-protos.h: Rename
23243         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23244         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23245         fast_interrupt.
23246         (microblaze_fast_interrupt_function_p): New function.
23247         (microblaze_is_interrupt_handler): Rename to
23248         microblaze_is_interrupt_variant and add fast_interrupt check.
23249         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23250         (save_restore_insns): Likewise.
23251         (compute_frame_size): Likewise.
23252         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23253         (microblaze_globalize_label): Likewise.
23254         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23255         * config/microblaze/microblaze.md: Use wrapper
23256         microblaze_is_interrupt_variant.
23257
23258 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23259
23260         * sdbout.c (sdbout_one_type): Switch to current function's section
23261         supporting cold/hot.
23262
23263 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23264
23265         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23266         -mxl-reorder.
23267
23268 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23269
23270         PR middle-end/56461
23271         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23272         if VALGRIND_GET_VBITS is defined, temporarily make object
23273         memory all defined, and restore previous valgrind addressability
23274         and definability afterwards.  Free this_object at the end.
23275
23276         PR middle-end/56461
23277         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23278         right before calling lra_create_live_ranges, also call it
23279         when clearing live_p.  Only call lra_clear_live_ranges
23280         at the end if live_p.
23281
23282         PR middle-end/56461
23283         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23284
23285 2013-03-05  Richard Biener  <rguenther@suse.de>
23286
23287         PR tree-optimization/56521
23288         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23289         value-id.
23290
23291 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23292
23293         PR c++/55135
23294         * except.h (remove_unreachable_eh_regions): New prototype.
23295         * except.c (remove_eh_handler_splicer): New function, split out
23296         of remove_eh_handler.
23297         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23298         warning about running it on many EH regions one at a time.
23299         (remove_unreachable_eh_regions_worker): New function, walk the
23300         EH tree in depth-first order and remove non-marked regions.
23301         (remove_unreachable_eh_regions): New function.
23302         * tree-eh.c (mark_reachable_handlers): New function, split out
23303         from remove_unreachable_handlers.
23304         (remove_unreachable_handlers): Use mark_reachable_handlers and
23305         remove_unreachable_eh_regions.
23306         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23307         and remove_unreachable_eh_regions.
23308
23309 2013-03-05  Richard Biener  <rguenther@suse.de>
23310
23311         PR middle-end/56525
23312         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23313         not freeing them until the end.
23314
23315 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23316
23317         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23318
23319 2013-03-05  Richard Biener  <rguenther@suse.de>
23320
23321         PR tree-optimization/56270
23322         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23323         of loads after scheduling an SLP instance.
23324
23325 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23326
23327         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23328         tic6x.exp.
23329         (check_gcc_parallelize): Run guality.exp as a separate job from
23330         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23331         struct-layout-1.exp with stackalign.exp.
23332
23333         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23334
23335         PR middle-end/56461
23336         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23337         load_index sbitmap even if some bit in it isn't set.
23338
23339         PR middle-end/56461
23340         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23341         (discover_iteration_bound_by_body_walk): Change queues to
23342         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23343         spelling in comment.  Call safe_push on queues[bound_index] directly.
23344         Release queues[queue_index] in every iteration unconditionally.
23345         Release bounds vector.
23346
23347         PR middle-end/56461
23348         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23349         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23350         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23351         inner_phis vector.
23352
23353 2013-03-05  Richard Biener  <rguenther@suse.de>
23354
23355         PR lto/56515
23356         * tree-inline.c (remap_blocks_to_null): New function.
23357         (expand_call_inline): When expanding a call stmt without
23358         an associated block inline remap all callee blocks to NULL.
23359
23360 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23361
23362         PR rtl-optimization/56494
23363         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23364         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23365         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23366
23367         PR middle-end/56461
23368         * sel-sched-ir.c (free_sched_pools): Release
23369         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23370         if succs_info_pool.max_top isn't -1.
23371
23372         PR bootstrap/56509
23373         * opts.c (opts_obstack, opts_concat): Moved to...
23374         * opts-common.c (opts_obstack, opts_concat): ... here.
23375
23376 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23377
23378         PR middle-end/56461
23379         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23380
23381 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23382
23383         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23384         all appropriate places.
23385
23386 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23387
23388         PR tree-optimization/56424
23389         * ipa-split.c (split_function): Do not set the RSO flag if result is
23390         not by reference and its type is a register type.
23391
23392 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23393
23394         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23395         (microblaze_legitimate_pic_operand): Likewise
23396         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23397         new function microblaze_legitimate_pic_operand
23398         * config/microblaze/microblaze-protos.h
23399         (microblaze_legitimate_pic_operand): Declare.
23400
23401 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23402
23403         * config/microblaze/predicates.md (call_insn_simple_operand):
23404         New predicate for supported rtx code types.
23405         * config/microblaze/microblaze.md (call_internal1): Use
23406         call_insn_simple_operand predicate.
23407
23408 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23409
23410         PR middle-end/56461
23411         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23412         partitions.ordered_remove.
23413
23414         PR middle-end/56461
23415         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23416         vec_oprnds0.create (1) for modifier == NONE.
23417
23418         PR middle-end/56461
23419         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23420         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23421         vec_oprnds1 right before pushing anything to it for
23422         scalar_shift_arg.
23423
23424         PR middle-end/56461
23425         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23426         set nbbs to 0 instead of having separate code path.
23427         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23428         instead of false as last argument if returning NULL.
23429
23430 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23431
23432         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23433         the attribute is now called "target" instead of "option".
23434         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23435         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23436         attribute/pragma name for TARGET_OPTION_VALID_P and
23437         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23438         * doc/tm.texi: Regenerated.
23439
23440 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23441
23442         * config/microblaze/microblaze.c:
23443         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23444         * config/microblaze/microblaze.h: Add -mxl-reorder to
23445         DRIVER_SELF_SPECS.
23446         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23447         instructions emitted if TARGET_REORDER.
23448         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23449         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23450         separately.
23451
23452 2013-03-01  Xinliang David Li  <davidxl@google.com>
23453
23454         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23455         walk length.
23456
23457 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23458
23459         PR middle-end/56461
23460         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23461         vector even when returning true.  Fix up function comment formatting.
23462
23463         PR middle-end/56461
23464         * ira-build.c (ira_loop_nodes_count): New variable.
23465         (create_loop_tree_nodes): Initialize it.
23466         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23467
23468         PR middle-end/56461
23469         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23470         method on dr_chain and result_chain.
23471         * tree-vect-stmts.c (vectorizable_store): Only call
23472         result_chain.create if j == 0.
23473
23474         PR middle-end/56461
23475         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23476         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23477         before overwriting it.
23478
23479 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23480
23481         * doc/extended.texi (C Extensions): Change order in @menu
23482         to match @node.
23483         (Other MIPS Built-in Functions): Move last MIPS entry before
23484         "picoChip Built-in Functions".
23485         (SH Built-in Functions): Move after RX Built-in Functions.
23486         * doc/gcc.texi (Introduction): Change order in @menu
23487         to match @node.
23488         * doc/md.texi (Constraints): Ditto.
23489         * gty.texi (Type Information): Ditto.
23490         (User-provided marking routines for template types): Make
23491         subsection.
23492         * doc/invoke.texi (AArch64 Options): Move before
23493         "Adapteva Epiphany Options".
23494
23495 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23496             Jakub Jelinek  <jakub@redhat.com>
23497
23498         PR sanitizer/56454
23499         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23500         no_address_safety_analysis attribute.
23501         * doc/extend.texi (no_address_safety_attribute): Rename to
23502         no_sanitize_address attribute, mention no_address_safety_analysis
23503         attribute as deprecated alias.
23504
23505 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23506
23507         PR middle-end/56461
23508         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23509         type to vec<vec<tree> > *.
23510         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23511         to be vec<tree> instead of vec<tree> *, set vec_defs
23512         to vNULL and call vec_defs.create (number_of_vects), adjust other
23513         uses of vec_defs.
23514         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23515         vectorizable_condition): Adjust vect_get_slp_defs callers.
23516
23517 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23518
23519         * config/aarch64/aarch64.c
23520         (aarch64_float_const_representable): Remove unused variable.
23521
23522 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23523
23524         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23525
23526 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23527
23528         * config/aarch64/aarch64-builtins.c
23529         (aarch64_init_simd_builtins): Make static.
23530
23531 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23532
23533         * config/aarch64/aarch64.c
23534         (aarch64_simd_make_constant): Make static.
23535
23536 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23537
23538         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23539         with no initialization to the RHS of debug statements.
23540
23541 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23542
23543         PR tree-optimization/56294
23544         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23545         Adjust dumping.
23546         (get_access_replacement): Do not call create_access_replacement.
23547         Assert a replacement exists.
23548         (get_repl_default_def_ssa_name): Create the replacement declaration
23549         itself.
23550
23551 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23552
23553         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23554         final_end_function.
23555
23556 2013-02-28  Marek Polacek  <polacek@redhat.com>
23557
23558         PR rtl-optimization/56466
23559         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23560         if we're changing a loop.
23561         (peel_loops_completely): Likewise.
23562
23563 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23564
23565         PR c++/55813
23566         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23567
23568 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23569
23570         PR target/56445
23571         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23572         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23573         INTX_FTYPE_FX, FX_FTYPE_INTX.
23574         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23575
23576 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23577
23578         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23579         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23580         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23581         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23582         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23583         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23584         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23585         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23586         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23587         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23588         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23589         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23590         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23591         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23592         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23593         (avrxmega6): Increase max flash segments from 5 to 6.
23594         * config/avr/t-multilib: Regenerate.
23595         * config/avr/avr-tables.opt: Regenerate.
23596         * doc/avr-mmcu.texi: Regenerate.
23597
23598 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23599
23600         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23601         (avr_device_to_arch): Rename to avr_device_to_ld.
23602         (avr_device_to_as): New prototype.
23603         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23604         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23605         * config/avr/driver-avr.c (avr_device_to_as): New.
23606         (avr_device_to_arch): Rename to avr_device_to_ld.
23607
23608 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23609
23610         PR middle-end/56461
23611         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23612         method on dr_chain and result_chain.
23613
23614         PR middle-end/56461
23615         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23616         pointer_set_destroy on not_executed_last_iteration.
23617
23618         PR middle-end/56461
23619         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23620
23621         PR middle-end/56461
23622         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23623         FOR_EACH_DEFINED_FUNCTION when freeing state.
23624
23625         PR middle-end/56461
23626         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23627         pool_free.
23628         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23629         overwriting it.
23630
23631         PR middle-end/56461
23632         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23633         known_aggs[i].items and release known_aggs vector.
23634
23635         PR middle-end/56461
23636         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23637
23638 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23639
23640         * config/microblaze/microblaze.c (microblaze_emit_compare):
23641         Use xor for EQ/NE comparisions.
23642         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23643         (cbranchsf4): Adjust operator to comparison_operator.
23644
23645 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23646
23647         PR middle-end/56461
23648         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23649         vector.
23650         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23651         vec_safe_push, always update *slot.
23652         (redirect_edge_var_map_clear): Use vec_free.
23653         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23654         (free_var_map_entry): Use vec_free.
23655         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23656         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23657
23658 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23659
23660         PR middle-end/45472
23661         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23662         when the may_trap_p bit of the exprs being merged differs.
23663         Reorder tests for speculativeness in the logical and operator.
23664
23665 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23666
23667         * incpath.c (add_standard_paths): Use reconcat instead of concat
23668         where appropriate and avoid leaking memory.
23669
23670         * opts.h: Include obstack.h.
23671         (opts_concat): New prototype.
23672         (opts_obstack): New declaration.
23673         * opts.c (opts_concat): New function.
23674         (opts_obstack): New variable.
23675         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23676         (finish_options): Use opts_concat instead of concat
23677         and XOBNEWVEC instead of XNEWVEC.
23678         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23679         generate_option): Likewise.
23680         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23681         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23682
23683         PR target/56455
23684         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23685         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23686
23687 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23688
23689         PR middle-end/56461
23690         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23691
23692 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23693
23694         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23695         (arm_block_move_unaligned_straight): Likewise.
23696         (arm_adjust_block_mem): Likewise.
23697
23698 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23699
23700         PR target/48901
23701         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23702         temp, cond and label.
23703         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23704
23705         PR target/52500
23706         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23707         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23708
23709         PR target/52501
23710         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23711         of prologue/epilogue functions.
23712
23713         PR target/52550
23714         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23715         Remove unused variable cfa_offset.
23716         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
23717
23718         PR target/54639
23719         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
23720         type promotion to unsigned.
23721
23722         PR target/54640
23723         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
23724         for HOST_WIDE_INT of 32 bit / same size as int.
23725         (arm_block_move_unaligned_straight): Likewise.
23726         (arm_adjust_block_mem): Likewise.
23727
23728         PR target/54662
23729         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
23730         ALL_CFLAGS.
23731
23732 2013-02-26  Marek Polacek  <polacek@redhat.com>
23733
23734         PR tree-optimization/56426
23735         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
23736
23737 2013-02-26  Richard Biener  <rguenther@suse.de>
23738
23739         PR target/56444
23740         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
23741         unused variable loops.
23742
23743 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23744
23745         PR tree-optimization/56448
23746         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
23747         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
23748         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
23749         later operands of the references, or even first operand for
23750         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
23751
23752         PR tree-optimization/56443
23753         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
23754         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
23755         to type_for_mode langhook.
23756
23757 2013-02-25  Matt Turner  <mattst88@gmail.com>
23758
23759         * doc/invoke.texi: Document r4700.
23760
23761 2013-02-25  Richard Biener  <rguenther@suse.de>
23762
23763         PR tree-optimization/56175
23764         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
23765         split out from ...
23766         (simplify_bitwise_binary): ... here.  Also guard the conversion
23767         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
23768
23769 2013-02-25  Catherine Moore  <clm@codesourcery.com>
23770
23771         Revert:
23772         2013-02-24  Catherine Moore  <clm@codesourcery.com>
23773             Maciej W. Rozycki  <macro@codesourcery.com>
23774             Tom de Vries  <tom@codesourcery.com>
23775             Nathan Sidwell  <nathan@codesourcery.com>
23776             Iain Sandoe  <iain@codesourcery.com>
23777             Nathan Froyd  <froydnj@codesourcery.com>
23778             Chao-ying Fu  <fu@mips.com>
23779
23780         * doc/extend.texi: (micromips, nomicromips, nocompression):
23781         Document new function attributes.
23782         * doc/invoke.texi (minterlink-compressed, mmicromips,
23783         m14k, m14ke, m14kec): Document new options.
23784         (minterlink-mips16): Update documentation.
23785         * doc/md.texi (ZC, ZD): Document new constraints.
23786         * configure.ac (gcc_cv_as_micromips): Check if linker
23787         supports the .set micromips directive.
23788         * configure: Regenerate.
23789         * config.in: Regenerate.
23790         * config/mips/mips-tables.opt: Regenerate.
23791         * config/mips/micromips.md: New file.
23792         * constraints.md (ZC, AD): New constraints.
23793         * config/mips/predicates.md (movep_src_register): New predicate.
23794         (movep_src_operand): New predicate.
23795         (non_volatile_mem_operand): New predicate.
23796         * config/mips/mips.md (multimem): New type.
23797         (length): Differentiate between 17-bit and 18-bit branch offsets.
23798         (MOVEP1, MOVEP2): New mode iterator.
23799         (mov_<load>l): Use ZC constraint.
23800         (mov_<load>r): Likewise.
23801         (mov_<store>l): Likewise.
23802         (mov_<store>r): Likewise.
23803         (*branch_equality<mode>_inverted): Add microMIPS support.
23804         (*branch_equality<mode>): Likewise.
23805         (*jump_absolute): Likewise.
23806         (indirect_jump_<mode>): Likewise.
23807         (tablejump_<mode>): Likewise.
23808         (<optab>_internal): Likewise.
23809         (sibcall_internal): Likewise.
23810         (sibcall_value_internal): Likewise.
23811         (prefetch): Use constraint ZD.
23812         * config/mips/mips.opt (minterlink-compressed): New option.
23813         (minterlink-mips16): Now an alias for minterlink-compressed.
23814         (mmicromips): New option.
23815         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23816         (compare_and_swap_12): Likewise.
23817         (sync_add<mode>): Likewise.
23818         (sync_<optab>_12): Likewise.
23819         (sync_old_<optab>_12): Likewise.
23820         (sync_new_<optab>_12): Likewise.
23821         (sync_nand_12): Likewise.
23822         (sync_old_nand_12): Likewise.
23823         (sync_new_nand_12): Likewise.
23824         (sync_sub<mode>): Likewise.
23825         (sync_old_add<mode>): Likewise.
23826         (sync_old_sub<mode>): Likewise.
23827         (sync_new_add<mode>): Likewise.
23828         (sync_new_sub<mode>): Likewise.
23829         (sync_<optab><mode>): Likewise.
23830         (sync_old_<optab><mode>): Likewise.
23831         (sync_new_<optab><mode>): Likewise.
23832         (sync_nand<mode>): Likewise.
23833         (sync_old_nand<mode>): Likewise.
23834         (sync_new_nand<mode>): Likewise.
23835         (sync_lock_test_and_set<mode>): Likewise.
23836         (test_and_set_12): Likewise.
23837         (atomic_compare_and_swap<mode>): Likewise.
23838         (atomic_exchange<mode>_llsc): Likewise.
23839         (atomic_fetch_add<mode>_llsc): Likewise.
23840         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23841         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23842         (umips_save_restore_pattern_p): Likewise.
23843         (umips_load_store_pair_p): Likewise.
23844         (umips_output_load_store_pair): Likewise.
23845         (umips_movep_target_p): Likewise.
23846         (umips_12bit_offset_address_p): Likewise.
23847         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23848         (mips_base_mips16): Rename this...
23849         (mips_base_compression_flags): ...to this. Update all uses.
23850         (mips_attribute_table): Add micromips, nomicromips and nocompression.
23851         (mips_mips16_decl_p): Delete.
23852         (mips_nomips16_decl_p): Delete.
23853         (mips_get_compress_on_flags): New function.
23854         (mips_get_compress_off_flags): New function.
23855         (mips_get_compress_mode): New function.
23856         (mips_get_compress_on_name): New function.
23857         (mips_get_compress_off_name): New function.
23858         (mips_insert_attributes): Support multiple compression types.
23859         (mips_merge_decl_attributes): Likewise.
23860         (umips_12bit_offset_address_p): New function.
23861         (mips_start_function_definition): Emit .set micromips directive.
23862         (mips_call_may_need_jalx_p): New function.
23863         (mips_function_ok_for_sibcall): Add microMIPS support.
23864         (mips_print_operand_punctuation): Support short delay slots and
23865         compact jumps.
23866         (umips_swm_mask, umips_swm_encoding): New.
23867         (umips_build_save_restore): New function.
23868         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23869         (was_mips16_p): Remove.
23870         (old_compression_mode): New.
23871         (mips_set_compression_mode): New function.
23872         (mips_set_current_function): Add microMIPS support.
23873         (mips_option_override): Likewise.
23874         (umips_save_restore_pattern_p): New function.
23875         (umips_output_save_restore): New function.
23876         (umips_load_store_pair_p_1): New function.
23877         (umips_load_store_pair_p): New function.
23878         (umips_output_load_store_pair_1): New function.
23879         (umips_output_load_store_pair): New function.
23880         (umips_movep_target_p) New function.
23881         (mips_prepare_pch_save): Add microMIPS support.
23882         * config/mips/mips.h (TARGET_COMPRESSION): New.
23883         (TARGET_CPU_CPP_BUILTINS): Update macro
23884         to use new compression flags and to support microMIPS.
23885         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23886         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23887         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23888         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23889         (ASM_SPEC): Support mmicromips and mno-micromips.
23890         (M16STORE_REG_P): New macro.
23891         (MIPS_CALL): Support TARGET_MICROMIPS.
23892         (MICROMIPS_J): New macro.
23893         (mips_base_mips16): Rename this...
23894         (mips_base_compression_flags): ...to this.
23895         (UMIPS_12BIT_OFFSET_P): New macro.
23896         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23897         (MULTILIB_DIRNAMES): Likewise.
23898
23899 2013-02-25  Tom de Vries  <tom@codesourcery.com>
23900
23901         PR rtl-optimization/56131
23902         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
23903         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
23904         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
23905
23906 2013-02-25  Tobias Burnus  <burnus@net-b.de>
23907
23908         * doc/invoke.texi (-fsanitize=): Move from optimization
23909         to debugging options.
23910
23911 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
23912
23913         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
23914
23915 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
23916             Alexander Monakov  <amonakov@ispras.ru>
23917
23918         PR middle-end/56077
23919         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
23920         flush pending lists also on non-jumps.  Adjust comment.
23921
23922 2013-02-24  Catherine Moore  <clm@codesourcery.com>
23923             Maciej W. Rozycki  <macro@codesourcery.com>
23924             Tom de Vries  <tom@codesourcery.com>
23925             Nathan Sidwell  <nathan@codesourcery.com>
23926             Iain Sandoe  <iain@codesourcery.com>
23927             Nathan Froyd  <froydnj@codesourcery.com>
23928             Chao-ying Fu  <fu@mips.com>
23929
23930         * doc/extend.texi: (micromips, nomicromips, nocompression):
23931         Document new function attributes.
23932         * doc/invoke.texi (minterlink-compressed, mmicromips,
23933         m14k, m14ke, m14kec): Document new options.
23934         (minterlink-mips16): Update documentation.
23935         * doc/md.texi (ZC, ZD): Document new constraints.
23936         * configure.ac (gcc_cv_as_micromips): Check if linker
23937         supports the .set micromips directive.
23938         * configure: Regenerate.
23939         * config.in: Regenerate.
23940         * config/mips/mips-tables.opt: Regenerate.
23941         * config/mips/micromips.md: New file.
23942         * constraints.md (ZC, AD): New constraints.
23943         * config/mips/predicates.md (movep_src_register): New predicate.
23944         (movep_src_operand): New predicate.
23945         (non_volatile_mem_operand): New predicate.
23946         * config/mips/mips.md (multimem): New type.
23947         (length): Differentiate between 17-bit and 18-bit branch offsets.
23948         (MOVEP1, MOVEP2): New mode iterator.
23949         (mov_<load>l): Use ZC constraint.
23950         (mov_<load>r): Likewise.
23951         (mov_<store>l): Likewise.
23952         (mov_<store>r): Likewise.
23953         (*branch_equality<mode>_inverted): Add microMIPS support.
23954         (*branch_equality<mode>): Likewise.
23955         (*jump_absolute): Likewise.
23956         (indirect_jump_<mode>): Likewise.
23957         (tablejump_<mode>): Likewise.
23958         (<optab>_internal): Likewise.
23959         (sibcall_internal): Likewise.
23960         (sibcall_value_internal): Likewise.
23961         (prefetch): Use constraint ZD.
23962         * config/mips/mips.opt (minterlink-compressed): New option.
23963         (minterlink-mips16): Now an alias for minterlink-compressed.
23964         (mmicromips): New option.
23965         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23966         (compare_and_swap_12): Likewise.
23967         (sync_add<mode>): Likewise.
23968         (sync_<optab>_12): Likewise.
23969         (sync_old_<optab>_12): Likewise.
23970         (sync_new_<optab>_12): Likewise.
23971         (sync_nand_12): Likewise.
23972         (sync_old_nand_12): Likewise.
23973         (sync_new_nand_12): Likewise.
23974         (sync_sub<mode>): Likewise.
23975         (sync_old_add<mode>): Likewise.
23976         (sync_old_sub<mode>): Likewise.
23977         (sync_new_add<mode>): Likewise.
23978         (sync_new_sub<mode>): Likewise.
23979         (sync_<optab><mode>): Likewise.
23980         (sync_old_<optab><mode>): Likewise.
23981         (sync_new_<optab><mode>): Likewise.
23982         (sync_nand<mode>): Likewise.
23983         (sync_old_nand<mode>): Likewise.
23984         (sync_new_nand<mode>): Likewise.
23985         (sync_lock_test_and_set<mode>): Likewise.
23986         (test_and_set_12): Likewise.
23987         (atomic_compare_and_swap<mode>): Likewise.
23988         (atomic_exchange<mode>_llsc): Likewise.
23989         (atomic_fetch_add<mode>_llsc): Likewise.
23990         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23991         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23992         (umips_save_restore_pattern_p): Likewise.
23993         (umips_load_store_pair_p): Likewise.
23994         (umips_output_load_store_pair): Likewise.
23995         (umips_movep_target_p): Likewise.
23996         (umips_12bit_offset_address_p): Likewise.
23997         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23998         (mips_base_mips16): Rename this...
23999         (mips_base_compression_flags): ...to this. Update all uses.
24000         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24001         (mips_mips16_decl_p): Delete.
24002         (mips_nomips16_decl_p): Delete.
24003         (mips_get_compress_on_flags): New function.
24004         (mips_get_compress_off_flags): New function.
24005         (mips_get_compress_mode): New function.
24006         (mips_get_compress_on_name): New function.
24007         (mips_get_compress_off_name): New function.
24008         (mips_insert_attributes): Support multiple compression types.
24009         (mips_merge_decl_attributes): Likewise.
24010         (umips_12bit_offset_address_p): New function.
24011         (mips_start_function_definition): Emit .set micromips directive.
24012         (mips_call_may_need_jalx_p): New function.
24013         (mips_function_ok_for_sibcall): Add microMIPS support.
24014         (mips_print_operand_punctuation): Support short delay slots and
24015         compact jumps.
24016         (umips_swm_mask, umips_swm_encoding): New.
24017         (umips_build_save_restore): New function.
24018         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24019         (was_mips16_p): Remove.
24020         (old_compression_mode): New.
24021         (mips_set_compression_mode): New function.
24022         (mips_set_current_function): Add microMIPS support.
24023         (mips_option_override): Likewise.
24024         (umips_save_restore_pattern_p): New function.
24025         (umips_output_save_restore): New function.
24026         (umips_load_store_pair_p_1): New function.
24027         (umips_load_store_pair_p): New function.
24028         (umips_output_load_store_pair_1): New function.
24029         (umips_output_load_store_pair): New function.
24030         (umips_movep_target_p) New function.
24031         (mips_prepare_pch_save): Add microMIPS support.
24032         * config/mips/mips.h (TARGET_COMPRESSION): New.
24033         (TARGET_CPU_CPP_BUILTINS): Update macro
24034         to use new compression flags and to support microMIPS.
24035         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24036         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24037         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24038         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24039         (ASM_SPEC): Support mmicromips and mno-micromips.
24040         (M16STORE_REG_P): New macro.
24041         (MIPS_CALL): Support TARGET_MICROMIPS.
24042         (MICROMIPS_J): New macro.
24043         (mips_base_mips16): Rename this...
24044         (mips_base_compression_flags): ...to this.
24045         (UMIPS_12BIT_OFFSET_P): New macro.
24046         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24047         (MULTILIB_DIRNAMES): Likewise.
24048
24049 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24050
24051         PR target/52555
24052         * target-globals.c (save_target_globals): For init_reg_sets and
24053         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24054
24055 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24056
24057         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24058         * config/aarch64/t-aarch64
24059         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24060
24061 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24062
24063         PR inline-asm/56148
24064         * lra-constraints.c (process_alt_operands): Reload operand
24065         conflicting with earlier clobber only if no more other conflicting
24066         operands.
24067
24068 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24069
24070         PR sanitizer/56393
24071         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24072         if not linking a shared library.
24073
24074 2013-02-22  Seth LaForge  <sethml@google.com>
24075
24076         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24077
24078 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24079
24080         * config/arm/arm.md (split for extendsidi): Update condition.
24081         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24082         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24083         (qhs_zextenddi_cstr): Likewise.
24084
24085 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24086
24087         PR middle-end/56420
24088         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24089         avoid signed wrapping.
24090         (expand_mult): Handle properly multiplication by
24091         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24092         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24093         in the compiler if coeff is HOST_WIDE_INT_MIN.
24094         (expand_divmod): Don't make ext_op1 static, change it's type to
24095         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24096
24097         PR rtl-optimization/50339
24098         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24099         field.
24100         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24101         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24102         into splitting_ashiftrt field.
24103         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24104         ASHIFTRT.
24105         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24106         choices.
24107
24108 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24109
24110         PR middle-end/56108
24111         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24112         are sure to go irrevocable.
24113
24114 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24115
24116         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24117         scalars are valid operands.
24118
24119 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24120
24121         PR tree-optimization/56310
24122         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24123         only matching indices and non-negative final offsets.
24124         (intersect_aggregates_with_edge): Pass src_idx to
24125         agg_replacements_to_vector.  Pass src_idx insstead of index to
24126         intersect_with_agg_replacements.
24127
24128 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24129
24130         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24131         instead of hard-wired defaults.
24132
24133 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24134
24135         * doc/invoke.texi (MIPS Options): Update documentation of the
24136         floating-point multiply-accumulate instruction restrictions.
24137
24138 2013-02-21  Kostya Serebryany  <kcc@google.com>
24139
24140         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24141         asan_shadow_offset on x86_64 linux.
24142
24143 2013-02-21  Richard Biener  <rguenther@suse.de>
24144
24145         PR tree-optimization/56415
24146         Revert
24147         2013-02-11  Richard Biener  <rguenther@suse.de>
24148
24149         PR tree-optimization/56273
24150         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24151         first VRP run.
24152
24153 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24154
24155         PR bootstrap/56258
24156         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24157         instead of @itemx.
24158
24159         PR inline-asm/56405
24160         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24161         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24162
24163 2013-02-20  Jan Hubicka  <jh@suse.cz>
24164
24165         PR tree-optimization/56265
24166         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24167         when target is referenced for first time.
24168
24169 2013-02-20  Richard Biener  <rguenther@suse.de>
24170
24171         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24172         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24173         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24174         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24175         not return anything.
24176         (rename_ssa_copies): Do not remove unused locals.
24177         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24178         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24179         * passes.c (execute_function_todo): Do not schedule unused locals
24180         removal if cleanup_tree_cfg did something.
24181         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24182         about the number of removed locals.
24183
24184 2013-02-20  Richard Biener  <rguenther@suse.de>
24185
24186         PR tree-optimization/56398
24187         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24188
24189 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24190
24191         PR tree-optimization/55334
24192         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24193         restricted pointers to arrays.
24194
24195 2013-02-20  Richard Biener  <rguenther@suse.de>
24196             Jakub Jelinek  <jakub@redhat.com>
24197
24198         PR tree-optimization/56396
24199         * tree-ssa-ccp.c (n_const_val): New static variable.
24200         (get_value): Return NULL for SSA names we don't have a lattice
24201         entry for.
24202         (ccp_initialize): Initialize n_const_val.
24203         * tree-ssa-copy.c (n_copy_of): New static variable.
24204         (init_copy_prop): Initialize n_copy_of.
24205         (get_value): Return NULL_TREE for SSA names we don't have a
24206         lattice entry for.
24207
24208 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24209
24210         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24211
24212 2013-02-20  Richard Biener  <rguenther@suse.de>
24213
24214         * genpreds.c (write_lookup_constraint): Do not compare first
24215         letter of the constraint again.
24216
24217 2013-02-20  Richard Biener  <rguenther@suse.de>
24218
24219         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24220         and ceil_log2.
24221         (get_use_iv_cost): Terminate hashtable walk when coming across
24222         an empty entry.
24223
24224 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24225
24226         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24227         reassociation for avx2 targets.
24228
24229 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24230
24231         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24232         Add version check for v8.10.a to enable microblaze_has_clz
24233         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24234         version and TARGET_PATTERN_COMPARE check
24235         * config/microblaze/microblaze.md: New clzsi2 instruction
24236
24237 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24238
24239         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24240         function before branching.
24241
24242 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24243
24244         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24245         DUMP_INSN_RTX_UID.
24246         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24247
24248 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24249
24250         PR middle-end/55889
24251         * sel-sched.c: Include ira.h.
24252         (implicit_clobber_conflict_p): New function.
24253         (moveup_expr): Use it.
24254         * Makefile.in (sel-sched.o): Depend on ira.h.
24255
24256 2013-02-19  Richard Biener  <rguenther@suse.de>
24257
24258         PR tree-optimization/56384
24259         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24260         (vn_hash_type): Split out from ...
24261         (vn_hash_constant_with_type): ... here.
24262         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24263         (vn_phi_eq): Compare types from vn_phi_s structure.
24264         (vn_phi_lookup): Populate vn_phi_s type.
24265         (vn_phi_insert): Likewise.
24266
24267 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24268
24269         PR tree-optimization/56350
24270         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24271         if haven't found reduction or nested cycle operand, rather than
24272         asserting we must find it.
24273
24274         PR tree-optimization/56381
24275         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24276         to fold_build3.
24277
24278 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24279             Jakub Jelinek  <jakub@redhat.com>
24280
24281         PR target/52555
24282         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24283         (swap_optab_enable): Same.
24284         (init_all_optabs): Use argument instead of global.
24285         * tree.h (struct tree_optimization_option): New field target_optabs.
24286         * expr.h (init_all_optabs): Add argument to prototype.
24287         (TREE_OPTIMIZATION_OPTABS): New.
24288         (save_optabs_if_changed): Protoize.
24289         * optabs.h: Declare this_fn_optabs.
24290         * optabs.c (save_optabs_if_changed): New.
24291         Declare this_fn_optabs.
24292         (init_optabs): Add argument to init_all_optabs() call.
24293         * function.c (invoke_set_current_function_hook): Handle per
24294         function optabs.
24295         * function.h (struct function): New field optabs.
24296         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24297         optimization_current_node has changed.
24298         * target-globals.h (save_target_globals_default_opts): Protoize.
24299         * target-globals.c (save_target_globals_default_opts): New.
24300
24301 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24302
24303         PR target/56347
24304         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24305         registers %fr12 and %fr12R as call used.
24306
24307         PR target/56214
24308         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24309         and HImode, require all displacements to be an integer multiple of
24310         their mode size.
24311         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24312         only allow QImode and HImode when reload is in progress and strict is
24313         true.  Likewise for symbolic addresses.  Use base14_operand to check
24314         displacements in REG+BASE addresses.
24315
24316 2013-02-18  Richard Biener  <rguenther@suse.de>
24317
24318         PR tree-optimization/56366
24319         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24320         handle sign-conversion of outer-loop initial induction value.
24321
24322 2013-02-18  Richard Biener  <rguenther@suse.de>
24323
24324         PR middle-end/56349
24325         * cfghooks.c (merge_blocks): If we merge a latch into another
24326         block adjust references to it.
24327         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24328         (verify_loop_structure): Verify that a recorded latch is in fact
24329         a latch.
24330
24331 2013-02-18  Richard Biener  <rguenther@suse.de>
24332
24333         PR tree-optimization/56321
24334         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24335         order SSA name release and virtual operand unlinking.
24336
24337 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24338
24339         * config/microblaze/microblaze.md (save_stack_block): Define.
24340         (restore_stack_block): Likewise.
24341
24342 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24343
24344         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24345         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24346         * config/microblaze/microblaze.c (microblaze_option_override):
24347         Bail out early for PIC modes when target does not support PIC.
24348
24349 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24350
24351         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24352         Replace with a microblaze version.
24353         (microblaze_trampoline_init): Adapt for microblaze.
24354         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24355         microblaze.
24356
24357 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24358             Dodji Seketeli  <dodji@redhat.com>
24359
24360         PR asan/56330
24361         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24362         (instrument_mem_region_access): Do not forget to always put
24363         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24364         0) statement, even for cases where either 'base' or 'base + len'
24365         are not instrumented -- because they have been previously
24366         instrumented.  Simplify the logic by putting all the statements
24367         instrument 'base + len' inside a sequence, and then insert that
24368         sequence right before the current insertion point.  Then, to
24369         instrument 'base + len', just get an iterator on that statement.
24370         And do not forget to update the pointer to iterator the function
24371         received as argument.
24372
24373 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24374
24375         PR rtl-optimization/56348
24376         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24377
24378 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24379
24380         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24381         (clean_graph_dump_file): Pass base to start_graph_dump.
24382
24383 2013-02-14  Richard Henderson  <rth@redhat.com>
24384
24385         PR target/55941
24386         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24387
24388 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24389
24390         * collect2-aix.h: Define F_LOADONLY.
24391
24392 2013-02-14  Richard Biener  <rguenther@suse.de>
24393
24394         PR lto/50494
24395         * varasm.c (output_constant_def_1): Get the decl representing
24396         the constant as argument.
24397         (output_constant_def): Wrap output_constant_def_1.
24398         (make_decl_rtl): Use output_constant_def_1 with the decl
24399         representing the constant.
24400         (build_constant_desc): Optionally re-use a decl already
24401         representing the constant.
24402         (tree_output_constant_def): Adjust.
24403
24404 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24405
24406         Fix an asan crash
24407         * asan.c (instrument_builtin_call):  Really put the length of the
24408         second source argument into src1_len.
24409
24410 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24411
24412         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24413         argument.  If it is false, don't create edge from then_bb to
24414         fallthru_bb.
24415         (insert_if_then_before_iter): Pass true to it.
24416         (build_check_stmt): Pass false to it.
24417         (transform_statements): Flush hash table only on extended basic
24418         block boundaries, rather than at the beginning of every bb.
24419         Don't flush hash table on nonfreeing_call_p calls.
24420         * tree-flow.h (nonfreeing_call_p): New prototype.
24421         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24422
24423 2013-02-13  David S. Miller  <davem@davemloft.net>
24424
24425         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24426
24427 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24428
24429         PR target/56184
24430         * ira.c (max_regno_before_ira): Move from ...
24431         (ira): ... here.
24432         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24433         vec_safe_length.
24434
24435 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24436
24437         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24438
24439 2013-02-13  Richard Biener  <rguenther@suse.de>
24440
24441         PR lto/56295
24442         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24443         globals in MEM_REFs.
24444
24445 2013-02-13  Richard Biener  <rguenther@suse.de>
24446
24447         * loop-init.c (loop_optimizer_init): Clear loop state when
24448         re-initializing preserved loops.
24449         * loop-unswitch.c (unswitch_single_loop): Return whether
24450         we unswitched the loop.  Do not verify loop state here.
24451         (unswitch_loops): When we unswitched a loop discover new loops.
24452
24453 2013-02-13  Kostya Serebryany  <kcc@google.com>
24454
24455         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24456         on x86_64 linux.
24457         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24458
24459 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24460
24461         Avoid instrumenting duplicated memory access in the same basic block
24462         * Makefile.in (asan.o): Add new dependency on hash-table.h
24463         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24464         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24465         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24466         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24467         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24468         (get_mem_ref_of_assignment): New functions.
24469         (get_mem_refs_of_builtin_call): Extract from
24470         instrument_builtin_call and tweak a little bit to make it fit with
24471         the new signature.
24472         (instrument_builtin_call): Use the new
24473         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24474         of is_gimple_builtin_call.
24475         (instrument_derefs, instrument_mem_region_access): Insert the
24476         instrumented memory reference into the hash table.
24477         (maybe_instrument_assignment): Renamed instrument_assignment into
24478         this, and change it to advance the iterator when instrumentation
24479         actually happened and return true in that case.  This makes it
24480         homogeneous with maybe_instrument_assignment, and thus give a
24481         chance to callers to be more 'regular'.
24482         (transform_statements): Clear the memory reference hash table
24483         whenever we enter a new BB, when we cross a function call, or when
24484         we are done transforming statements.  Use
24485         maybe_instrument_assignment instead of instrumentation.  No more
24486         need to special case maybe_instrument_assignment and advance the
24487         iterator after calling it; it's now handled just like
24488         maybe_instrument_call.  Update comment.
24489
24490 2013-02-13  Richard Biener  <rguenther@suse.de>
24491
24492         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24493         Fix loop discovery code.
24494
24495 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24496
24497         PR inline-asm/56148
24498         * lra-constraints.c (process_alt_operands): Match early clobber
24499         operand with itself.  Check conflicts with earlyclobber only if
24500         the operand is not reloaded.  Prefer to reload conflicting operand
24501         if earlyclobber and matching operands are the same.
24502
24503 2013-02-12  Richard Biener  <rguenther@suse.de>
24504
24505         PR lto/56297
24506         * lto-streamer-out.c (write_symbol): Do not output symbols
24507         for hard register variables.
24508
24509 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24510
24511         PR target/54222
24512         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24513         (umulsidi3_insn, mulsidi3_insn): New insns.
24514
24515 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24516
24517         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24518         (struct tune_params): Add vec_costs field.
24519         * config/arm/arm.c (arm_builtin_vectorization_cost)
24520         (arm_add_stmt_cost): New functions.
24521         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24522         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24523         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24524         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24525         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24526         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24527         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24528
24529 2013-02-12  Richard Biener  <rguenther@suse.de>
24530
24531         PR lto/56295
24532         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24533         decls again if possible.
24534
24535 2013-02-12  Richard Biener  <rguenther@suse.de>
24536
24537         PR middle-end/56288
24538         * tree-ssa.c (verify_ssa_name): Fix check, move
24539         SSA_NAME_IN_FREE_LIST check up.
24540
24541 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24542             Steven Bosscher   <steven@gcc.gnu.org>
24543
24544         PR rtl-optimization/56151
24545         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24546         equal to op0 or op1, and last_insn pattern is CODE operation
24547         with MEM dest and one of the operands matches that MEM.
24548
24549 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24550
24551         * doc/extend.texi: Document Function Multiversioning and "default"
24552         parameter string to target attribute.
24553         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24554         target attribute parameter is "default".
24555         (ix86_compare_version_priority): Remove checks for target attribute.
24556         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24557         Remove check for target attribute equal to NULL. Add assert.
24558         (ix86_generate_version_dispatcher_body): Change error to sorry.
24559
24560 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24561             Jack Howarth  <howarth@bromo.med.uc.edu>
24562             Patrick Marlier  <patrick.marlier@gmail.com>
24563
24564         PR libitm/55693
24565         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24566         define ENDFILE_SPEC as TM_DESTRUCTOR.
24567         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24568
24569 2013-02-11  Alexander Potapenko  <glider@google.com>
24570             Jack Howarth  <howarth@bromo.med.uc.edu>
24571             Jakub Jelinek  <jakub@redhat.com>
24572
24573         PR sanitizer/55617
24574         * config/darwin.c (cdtor_record): Rename ctor_record.
24575         (sort_cdtor_records): Rename sort_ctor_records.
24576         (finalize_dtors): New routine to sort destructors by
24577         priority before use in assemble_integer.
24578         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24579
24580 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24581
24582         PR rtl-optimization/56275
24583         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24584         offset is non-negative and less than cmode size before
24585         calling simplify_subreg.
24586
24587 2013-02-11  Richard Biener  <rguenther@suse.de>
24588
24589         PR tree-optimization/56264
24590         * cfgloop.h (fix_loop_structure): Adjust prototype.
24591         * loop-init.c (fix_loop_structure): Return the number of
24592         newly discovered loops.
24593         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24594         are discovered, do a full loop-closed SSA rewrite.
24595
24596 2013-02-11  Richard Biener  <rguenther@suse.de>
24597
24598         PR tree-optimization/56273
24599         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24600         first VRP run.
24601         (check_array_ref): Fix missing newline in dumps.
24602         (search_for_addr_array): Likewise.
24603
24604 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24605
24606         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24607
24608 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24609
24610         PR target/56256
24611         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24612
24613 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24614
24615         PR rtl-optimization/56246
24616         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24617         reload pseudo.
24618         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24619         constraints are satisfied.
24620
24621 2013-02-08  Jeff Law  <law@redhat.com>
24622
24623         PR debug/53948
24624         * emit-rtl.c (reg_is_parm_p): New function.
24625         * regs.h (reg_is_parm_p): New prototype.
24626         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24627         callee-clobbered registers.
24628
24629 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24630
24631         PR target/56043
24632         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24633         If there is no implicit builtin declaration, just return NULL.
24634
24635 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24636
24637         * config/i386/sse.md (FMAMODEM): New mode iterator.
24638         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24639         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24640
24641 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24642
24643         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24644         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24645         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24646
24647 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24648
24649         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24650         (microblaze*-*-elf): Likewise.
24651         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24652         LINK_SPEC.
24653         * config/microblaze/microblaze-c.c: Add builtin defines for
24654         _LITTLE_ENDIAN and _BIG_ENDIAN.
24655         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24656         add to TARGET_DEFAULT flags.
24657         Expand ASM_SPEC and LINK_SPEC.
24658         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24659         * config/microblaze/microblaze.md: Update extendsidi2 and
24660         movdi_internal instructions to use low-order / high-order reg
24661         print_operands.
24662         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24663         options and inversemask / mask of LITTLE_ENDIAN.
24664         * config/microblaze/t-microblaze: Expand multilib options to
24665         include mlittle-endian (le) and update exceptions patterns.
24666
24667 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24668
24669         PR rtl-optimization/56195
24670         * lra-constraints.c (get_reload_reg): Don't reuse regs
24671         if they have smaller mode than requested, if they have
24672         wider mode than requested, try to return a SUBREG.
24673
24674         PR tree-optimization/56250
24675         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24676         if type is unsigned and code isn't MULT_EXPR.
24677
24678 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24679
24680         PR tree-optimization/56064
24681         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24682         bits according to mode.
24683         * fixed-value.h (fixed_from_double_int)
24684         (const_fixed_from_double_int): Adjust comments.
24685
24686 2013-02-08  Richard Biener  <rguenther@suse.de>
24687
24688         PR lto/56231
24689         * lto-streamer.h (struct data_in): Remove current_file, current_line
24690         and current_col members.
24691         * lto-streamer-out.c (lto_output_location): Stream changed bits
24692         en-block for efficiency.
24693         * lto-streamer-in.c (clear_line_info): Remove.
24694         (lto_input_location): Cache current file, line and column
24695         globally via local statics.  Read changed bits en-block.
24696         (input_function): Do not call clear_line_info.
24697         (lto_read_body): Likewise.
24698         (lto_input_toplevel_asms): Likewise.
24699
24700 2013-02-08  Michael Matz  <matz@suse.de>
24701
24702         PR tree-optimization/52448
24703         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24704         (nt_call_phase): New static.
24705         (add_or_mark_expr): Only mark accesses with newer phase than any
24706         call seen.
24707         (nonfreeing_call_p): New.
24708         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24709         (nt_fini_block): Keep blocks marked as visited.
24710         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24711
24712 2013-02-08  Richard Biener  <rguenther@suse.de>
24713
24714         * ira.c (ira): Free broken dominator information.
24715
24716 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24717
24718         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
24719
24720 2013-02-08  Marek Polacek  <polacek@redhat.com>
24721
24722         * cfgloop.c (verify_loop_structure): Add more checking of headers.
24723
24724 2013-02-08  Richard Biener  <rguenther@suse.de>
24725
24726         PR middle-end/56181
24727         * cfgloop.h (flow_loops_find): Adjust.
24728         (bb_loop_header_p): Declare.
24729         * cfgloop.c (bb_loop_header_p): New function split out from ...
24730         (flow_loops_find): ... here.  Adjust function signature,
24731         support incremental loop structure update.
24732         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
24733         * cfgloopmanip.c (fix_loop_structure): Move ...
24734         * loop-init.c (fix_loop_structure): ... here.
24735         (apply_loop_flags): Split out from ...
24736         (loop_optimizer_init): ... here.
24737         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
24738         in incremental mode, only remove dead loops here.
24739
24740 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24741
24742         PR target/54222
24743         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
24744         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
24745         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
24746         (*round<mode>3.libgcc): New insns for fixed-modes.
24747         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
24748         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
24749         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
24750         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
24751         implementations.  Define to __builtin_avr_absFX,
24752         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
24753         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
24754         __builtin_avr_countlsFX, respectively.
24755         * config/avr/avr-c.c (target.h): Include it.
24756         (enum avr_builtin_id): New enum.
24757         (avr_resolve_overloaded_builtin): New static function.
24758         (avr_register_target_pragmas): Use it to set
24759         targetm.resolve_overloaded_builtin.
24760         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
24761         tree nodes used by DEF_BUILTIN.
24762         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
24763         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
24764         <AVR_BUILTIN_xxBITS>: Same.
24765
24766 2013-02-08  Richard Biener  <rguenther@suse.de>
24767
24768         * cfgloop.c (verify_loop_structure): Properly handle
24769         a loop exiting to another loop header.
24770         * ira-int.h (ira_loops): Remove.
24771         * ira.c (ira_loops): Remove.
24772         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
24773         (do_reload): Use loop_optimizer_finalize.
24774         * ira-build.c (create_loop_tree_nodes): Use get_loops and
24775         number_of_loops to access the loop tree.
24776         (more_one_region_p): Likewise.
24777         (finish_loop_tree_nodes): Likewise.
24778         (rebuild_regno_allocno_maps): Likewise.
24779         (mark_loops_for_removal): Likewise.
24780         (mark_all_loops_for_removal): Likewise.
24781         (remove_unnecessary_regions): Likewise.
24782         (ira_build): Likewise.
24783         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
24784
24785 2013-02-08  Richard Biener  <rguenther@suse.de>
24786
24787         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
24788         * ipa-pure-const.c (analyze_function): Avoid calling
24789         mark_irreducible_loops twice.
24790         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
24791
24792 2013-02-07  David S. Miller  <davem@davemloft.net>
24793
24794         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
24795         on 'reg'.
24796         * var-tracking.c (vt_add_function_parameter): Test the presence of
24797         HAVE_window_save properly and do not remap argument registers when
24798         we have a leaf function.
24799
24800 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
24801
24802         PR bootstrap/56227
24803         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
24804         instead of "ll".
24805         * config/i386/i386.c (ix86_print_operand): Ditto.
24806
24807 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
24808
24809         * lra-constraints.c (process_alt_operands): Fix recently added comment.
24810
24811 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
24812
24813         PR rtl-optimization/56225
24814         * lra-constraints.c (process_alt_operands): Check that reload hard
24815         reg can hold value for strict_low_part.
24816
24817 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
24818
24819         PR debug/56154
24820         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
24821         dwarf2out_end_function.
24822         (in_first_function_p, maybe_at_text_label_p,
24823         first_loclabel_num_not_at_text_label): New variables.
24824         (dwarf2out_var_location): In the first function find out
24825         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
24826         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
24827         functions.
24828
24829 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
24830
24831         PR rtl-optimization/56178
24832         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
24833         SUBREG of a register.  Tidy up related block of code.
24834         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
24835         note if the source is a register or a SUBREG of a register.
24836
24837 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
24838
24839         PR target/56228
24840         * config/rs6000/rs6000.md (ptrm): New mode attr.
24841         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
24842         call_value_indirect_aix<pttrsize>,
24843         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
24844         m in constraints.
24845
24846 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
24847
24848         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
24849         if -bnortl. Convert to strcmp and strncmp.
24850
24851 2013-02-07  Alan Modra  <amodra@gmail.com>
24852
24853         PR target/54009
24854         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
24855         addresses won't wrap when offsetting.
24856         (rs6000_secondary_reload): Provide secondary reloads needed for
24857         wrapping LO_SUM addresses.
24858
24859 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
24860
24861         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
24862         MACH, just __MACH__.
24863
24864 2013-02-06  Richard Biener  <rguenther@suse.de>
24865
24866         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
24867         instead of calling fix_loop_structure.
24868
24869 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
24870
24871         PR middle-end/56217
24872         * omp-low.c (use_pointer_for_field): Return false if
24873         lower_send_shared_vars doesn't generate any copy-out code.
24874
24875 2013-02-06  Tom de Vries  <tom@codesourcery.com>
24876
24877         PR rtl-optimization/56131
24878         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
24879         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
24880         of the label is NULL.  Add comment.
24881
24882 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
24883
24884         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
24885
24886         PR sanitizer/55374
24887         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
24888         (STATIC_LIBTSAN_LIBS): Likewise.
24889         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
24890         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
24891         is defined, don't add anything else beyond that.
24892         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
24893         (LINK_COMMAND_SPEC): Use them.
24894
24895         PR tree-optimization/56205
24896         * tree-stdarg.c (check_all_va_list_escapes): Return true if
24897         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
24898         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
24899
24900 2013-02-05  Richard Biener  <rguenther@suse.de>
24901
24902         PR tree-optimization/53342
24903         PR tree-optimization/53185
24904         * tree-vectorizer.h (vect_check_strided_load): Remove.
24905         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
24906         not disallow peeling for vectorized strided loads.
24907         (vect_check_strided_load): Make static and simplify.
24908         (vect_analyze_data_refs): Adjust.
24909         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
24910         correctly when vectorizing strided loads.
24911
24912 2013-02-05  Richard Biener  <rguenther@suse.de>
24913
24914         * doc/install.texi: Refer to ISL, not PPL.
24915
24916 2013-02-05  Jan Hubicka  <jh@suse.cz>
24917
24918         PR tree-optimization/55789
24919         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
24920
24921 2013-02-05  Jan Hubicka  <jh@suse.cz>
24922
24923         PR tree-optimization/55789
24924         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
24925         the dead call anyway.
24926
24927 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
24928
24929         PR sanitizer/55374
24930         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
24931
24932 2013-02-04  Alexander Potapenko  <glider@google.com>
24933             Jack Howarth  <howarth@bromo.med.uc.edu>
24934             Jakub Jelinek  <jakub@redhat.com>
24935
24936         PR sanitizer/55617
24937         * config/darwin.c (sort_ctor_records): Stabilized qsort
24938         on constructor priority by using original position.
24939         (finalize_ctors): New routine to sort constructors by
24940         priority before use in assemble_integer.
24941         (machopic_asm_out_constructor): Use finalize_ctors if needed.
24942
24943 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
24944
24945         PR libstdc++/54314
24946         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
24947         about visibility on artificial decls.
24948         * config/sol2.c (solaris_assemble_visibility): Likewise.
24949
24950 2013-02-04  Kai Tietz  <ktietz@redhat.com>
24951
24952         PR target/56186
24953         * config/i386/i386.c (function_value_ms_64): Add additional valtype
24954         argument and improve checking of return-argument types for 16-byte
24955         modes.
24956         (ix86_function_value_1): Add additional valtype argument on call
24957         of function_value_64.
24958         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
24959         handling infunction_value_64 function.
24960
24961 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
24962
24963         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
24964
24965 2013-02-04  Richard Biener  <rguenther@suse.de>
24966
24967         PR tree-optimization/56188
24968         * tree-ssa-structalias.c (label_visit): Consider case with
24969         initially non-empty points-to set.
24970         (perform_var_substitution): Dump node mapping and clean up.
24971
24972 2013-02-04  Richard Guenther  <rguenther@suse.de>
24973
24974         PR lto/56168
24975         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
24976         node prevail as last resort.
24977         (lto_symtab_merge_decls): Remove guard on LTRANS here.
24978         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
24979
24980 2013-02-04  Richard Biener  <rguenther@suse.de>
24981
24982         PR tree-optimization/56113
24983         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
24984         Merge into ...
24985         (equiv_class_lookup_or_add): ... this.
24986         (label_visit): Adjust and fix error in previous patch.
24987         (perform_var_substitution): Adjust.
24988
24989 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
24990
24991         * config/sh/divtab.c: Fix formatting and comments throughout the file.
24992         * config/sh/sh4-300.md: Likewise.
24993         * config/sh/sh4a.md: Likewise.
24994         * config/sh/constraints.md: Likewise.
24995         * config/sh/sh.md: Likewise.
24996         * config/sh/netbsd-elf.h: Likewise.
24997         * config/sh/predicates.md: Likewise.
24998         * config/sh/sh-protos.h: Likewise.
24999         * config/sh/ushmedia.h: Likewise.
25000         * config/sh/linux.h: Likewise.
25001         * config/sh/sh.c: Likewise.
25002         * config/sh/superh.h: Likewise.
25003         * config/sh/elf.h: Likewise.
25004         * config/sh/sh4.md: Likewise.
25005         * config/sh/sh.h: Likewise.
25006
25007 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25008
25009         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25010         constraint to match_test floating_point_store_memory_operand().
25011         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25012         (base14_operand): New.
25013         (floating_point_store_memory_operand): New.
25014         (integer_store_memory_operand): Revise to use base14_operand and
25015         reg_plus_base_memory_operand.
25016         (move_dest_operand): Allow symbolic_memory_operands.
25017         (symbolic_memory_operand): Check for LO_SOM.
25018         (symbolic_operand): Change default case to break.
25019         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25020         CONST_DOUBLE values to be reloaded by putting them into memory when
25021         the destination is a floating point register.
25022         (movdf): Remove code to handle CONST_DOUBLE.
25023         (movsf): Likewise.
25024         (reload_indf_r1): New.
25025         (reload_insf_r1): New.
25026         Consistently use "Q" and "T" constraints with integer and floating
25027         point move instructions, respectively.
25028         (movdi): Remove FAIL.
25029         Change predicate for source operand unamed DImode move from
25030         general_operand to move_src_operand.
25031         (umulsidi3): Change predicate for destination operand to
25032         register_operand.
25033         Likewise for similar unamed patterns.
25034         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25035         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25036         (hppa_legitimize_address): Simplify mask calculation.
25037         (pa_emit_move_sequence): Revised handling of secondary reloads from
25038         REG+D addresses for floating point loads and stores.  Directly handle
25039         loading CONST0_RTX (mode) to a floating point register.
25040         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25041         to floating point registers.  Don't restrict secondary reloads to
25042         floating point registers to integer modes.  Revise some comments and
25043         cleanup some code.
25044         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25045         (pa_legitimate_address_p): New.
25046         (pa_legitimize_reload_address): New.
25047         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25048         (STRICT_REG_OK_FOR_BASE_P): New.
25049         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25050         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25051
25052 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25053             Andrew Dixie  <andrewd@gentrack.com>
25054
25055         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25056         flag set.
25057
25058 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25059
25060         * expmed.c (extract_bit_field_1): Pass the full width of the
25061         structure to get_best_reg_extraction_insn.
25062
25063 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25064
25065         PR target/54601
25066         * configure.ac (use_cxa_atexit): Add AIX.
25067         * configure: Regenerate.
25068
25069         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25070
25071 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25072
25073         PR debug/54793
25074         * final.c (need_profile_function): New variable.
25075         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25076         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25077         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25078         notes, targetm.asm_out.function_prologue doesn't emit anything,
25079         HAVE_prologue and profiler should be emitted before prologue,
25080         set need_profile_function instead of emitting it.
25081         (final_scan_insn): If need_profile_function, emit
25082         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25083         NOTE_INSN_FUNCTION_BEG note.
25084
25085 2013-02-01  Richard Henderson  <rth@redhat.com>
25086
25087         * config/rs6000/rs6000.md (smulditi3): New.
25088         (umulditi3): New.
25089
25090         * config/alpha/alpha.md (umulditi3): New.
25091
25092 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25093
25094         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25095         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25096
25097 2013-02-01  Richard Biener  <rguenther@suse.de>
25098
25099         PR tree-optimization/56113
25100         * tree-ssa-structalias.c (label_visit): Reduce work for
25101         single-predecessor nodes.
25102
25103 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25104
25105         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25106         range isn't testing for zero.
25107
25108 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25109
25110         PR middle-end/56113
25111         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25112
25113 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25114             Nick Clifton  <nickc@redhat.com>
25115
25116         * config/v850/constraints.md (Q): Define as a memory constraint.
25117         * config/v850/predicates.md (label_ref_operand): New predicate.
25118         (e3v5_shift_operand): New predicate.
25119         (ior_operator): New predicate.
25120         * config/v850/t-v850: Add e3v5 multilib.
25121         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25122         (v850_gen_movdi): Prototype.
25123         * config/v850/v850.c: Add support for e3v5 architecture.
25124         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25125         TARGET_V850E_UP.
25126         (construct_save_jarl): Add e3v5 long JARL support.
25127         (v850_adjust_insn_length): New function.  Adjust length of call
25128         insns when using e3v5 instructions.
25129         (v850_gen_movdi): New function: Generate instructions to move a
25130         DImode value.
25131         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25132         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25133         (TARGET_USE_FPU): Enable for e3v5.
25134         (CONST_OK_FOR_W): New macro.
25135         (ADJUST_INSN_LENGTH): Define.
25136         * config/v850/v850.md (UNSPEC_LOOP): Define.
25137         (attr cpu): Add v850e3v5.
25138         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25139         (movdi): New pattern.
25140         (movdi_internal): New pattern.
25141         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25142         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25143         (cstoresf4): Likewise.
25144         (cstoredf4): Likewise.
25145         (insv): New pattern.
25146         (rotlso3_a): New pattern.
25147         (rotlsi3_b): New pattern
25148         (rotlsi3_v850e3v5): New pattern.
25149         (doloop_begin): New pattern.
25150         (fix_loop_counter): New pattern.
25151         (doloop_end): New pattern.
25152         (branch_normal): Add e3v5 long branch support.
25153         (branch_invert): Likewise.
25154         (branch_z_normal): Likewise.
25155         (branch_z_invert): Likewise.
25156         (branch_nz_normal): Likewise.
25157         (branch_nz_invert): Likewise.
25158         (call_internal_short): Add e3v5 register-indirect JARL support.
25159         (call_internal_long): Likewise.
25160         (call_value_internal_short): Likewise.
25161         (call_value_internal_long): Likewise.
25162         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25163         (mloop): New option.
25164         * config.gcc: Add support for configuring v840e3v5 target.
25165         * doc/invoke.texi: Document new v850 specific command line options.
25166
25167 2013-01-31  Paul Koning  <ni1d@arrl.net>
25168
25169         PR debug/55059
25170         PR debug/54508
25171         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25172         children if parent is a class.
25173         (prune_unused_types_prune): Don't add DW_AT_declaration.
25174
25175 2013-01-31  Richard Biener  <rguenther@suse.de>
25176
25177         PR tree-optimization/56157
25178         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25179         match up operand with SLP child.
25180
25181 2013-01-31  Jason Merrill  <jason@redhat.com>
25182
25183         PR debug/54410
25184         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25185         parameters the first time.
25186         (gen_scheduled_generic_parms_dies): Check completeness here.
25187
25188 2013-01-31  Richard Biener  <rguenther@suse.de>
25189
25190         PR middle-end/53073
25191         * common.opt (faggressive-loop-optimizations): New flag,
25192         enabled by default.
25193         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25194         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25195         infer_loop_bounds_from_undefined by it.
25196
25197 2013-01-31  Richard Biener  <rguenther@suse.de>
25198
25199         PR tree-optimization/56150
25200         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25201         visit virtual operands.
25202         (find_uses_to_rename_bb): Likewise.
25203
25204 2013-01-31  Richard Biener  <rguenther@suse.de>
25205
25206         PR tree-optimization/56150
25207         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25208         mixed store non-store stmts.
25209
25210 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25211
25212         PR sanitizer/55374
25213         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25214         LIBASAN_EARLY_SPEC is defined.
25215         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25216         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25217         before %o.
25218         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25219
25220         PR c++/55742
25221         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25222         invalid args instead of ICEing on it.
25223         (ix86_valid_target_attribute_tree): Return error_mark_node if
25224         ix86_valid_target_attribute_inner_p failed.
25225         (ix86_valid_target_attribute_p): Return false only if
25226         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25227         target("default") attribute.
25228         (sorted_attr_string): Change argument from const char * to tree,
25229         merge in all target attribute arguments rather than just one.
25230         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25231         instead of free.  Avoid using strcat.
25232         (ix86_mangle_function_version_assembler_name): Mangle
25233         target("default") as if no target attribute is present.  Adjust
25234         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25235         instead of xmalloc and XDELETEVEC instead of free.
25236         (ix86_function_versions): Don't return true if one of the decls
25237         doesn't have target attribute.  If they don't and one of the decls
25238         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25239         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25240         (ix86_supports_function_versions): Remove.
25241         (make_name): Fix up formatting.
25242         (make_dispatcher_decl): Remove resolver_name and its initialization.
25243         Avoid leaking memory.
25244         (is_function_default_version): Return true if there is
25245         target("default") attribute rather than no target attribute at all.
25246         (make_resolver_func): Avoid leaking memory.
25247         (ix86_generate_version_dispatcher_body): Likewise.
25248         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25249         * target.def (supports_function_versions): Remove.
25250         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25251         * doc/tm.texi: Regenerated.
25252
25253 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25254
25255         PR rtl-optimization/56144
25256         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25257         for values with side effects.
25258
25259 2013-01-30  Richard Biener  <rguenther@suse.de>
25260
25261         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25262         (sparseset_pop): Likewise.
25263         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25264         to be able to use quick_push in the worker loop.
25265
25266 2013-01-30  Marek Polacek  <polacek@redhat.com>
25267
25268         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25269
25270 2013-01-30  Richard Biener  <rguenther@suse.de>
25271
25272         PR lto/56147
25273         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25274
25275 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25276
25277         PR tree-optimization/56064
25278         * fixed-value.c (fixed_from_double_int): New function.
25279         * fixed-value.h (fixed_from_double_int): New prototype.
25280         (const_fixed_from_double_int): New static inline function.
25281         * fold-const.c (native_interpret_fixed): New static function.
25282         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25283         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25284         (native_encode_fixed): New static function.
25285         (native_encode_expr) <FIXED_CST>: Use it.
25286         (native_interpret_int): Move double_int worker code to...
25287         * double-int.c (double_int::from_buffer): ...this new static method.
25288         * double-int.h (double_int::from_buffer): Prototype it.
25289
25290 2013-01-30  Richard Biener  <rguenther@suse.de>
25291
25292         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25293         New pointer-map and obstack.
25294         (init_alias_vars): Allocate pointer-map and obstack.
25295         (delete_points_to_sets): Free them.
25296         (find_what_var_points_to): Cache result.
25297         (find_what_p_points_to): Adjust for changed interface of
25298         find_what_var_points_to.
25299         (compute_points_to_sets): Likewise.
25300         (ipa_pta_execute): Likewise.
25301
25302 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25303
25304         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25305         * configure: Regenerate.
25306         * config.in: Regenerate.
25307         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25308         #nobits/#progbits if supported.
25309
25310 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25311
25312         PR target/56121
25313         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25314         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25315         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25316
25317 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25318
25319         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25320         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25321
25322 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25323
25324         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25325         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25326
25327 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25328
25329         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25330         declaration.
25331         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25332         * config/arm/cortex-a7.md: New bypasses using
25333         arm_mac_accumulator_is_result.
25334
25335 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25336
25337         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25338         (cortex_a7_neon_mla): Likewise.
25339         (cortex_a7_fpfmad): New reservation.
25340         (cortex_a7_fpmacs): Use ffmas and update required units.
25341         (cortex_a7_fpmuld): Update required units and latency.
25342         (cortex_a7_fpmacd): Likewise.
25343         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25344         (cortex_a7_neon). Likewise.
25345         (bypass) Update participating units.
25346
25347 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25348
25349         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25350         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25351         from fmac to ffma.
25352         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25353         (vfp_fmul): Use ffmad.
25354         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25355         (cortex_r4_fmacd): Use ffmad.
25356         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25357         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25358         (cortex_a9_fmacd): Use ffmad.
25359         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25360         (cortex_a8_vfp_macd): Use ffmad.
25361         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25362         (cortex_a5_fpmacd): Use ffmad.
25363         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25364         (cortex_a15_vfp_macd): Use ffmad.
25365         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25366
25367 2013-01-29  Jason Merrill  <jason@redhat.com>
25368
25369         PR libstdc++/54314
25370         * varasm.c (default_assemble_visibility): Don't warn about
25371         visibility on artificial decls.
25372
25373 2013-01-29  Richard Biener  <rguenther@suse.de>
25374
25375         PR tree-optimization/56113
25376         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25377         the bitmap leader.
25378         (label_visit): Free duplicate bitmaps and record the leader instead.
25379         (perform_var_substitution): Adjust.
25380
25381 2013-01-29  Richard Biener  <rguenther@suse.de>
25382
25383         PR tree-optimization/55270
25384         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25385         the CFG, schedule loops for fixup.
25386
25387 2013-01-29  Nick Clifton  <nickc@redhat.com>
25388
25389         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25390         SP_REG.
25391
25392 2013-01-28  Leif Ekblad  <leif@rdos.net>
25393
25394         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25395         * config/i386/i386.h (TARGET_RDOS): New macro.
25396         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25397         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25398         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25399         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25400         DEFAULT_LARGE_SECTION_THRESHOLD.
25401         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25402         * config/i386/rdos.h: New file.
25403         * config/i386/rdos64.h: New file.
25404
25405 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25406
25407         PR other/54814
25408         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25409         TEST_HARD_REG_BIT.
25410
25411 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25412
25413         PR rtl-optimization/56117
25414         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25415         call cselib_lookup_from_insn on the MEM before calling
25416         add_insn_mem_dependence.
25417
25418 2013-01-28  Richard Biener  <rguenther@suse.de>
25419
25420         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25421         to a stmt that didn't have one.
25422         (copy_phis_for_bb): Likewise for PHI arguments.
25423         (copy_debug_stmt): Likewise for debug stmts.
25424
25425 2013-01-28  Richard Biener  <rguenther@suse.de>
25426
25427         PR tree-optimization/56034
25428         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25429         (partition_builtin_p): Adjust.
25430         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25431         it is the last partition.
25432         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25433         up the vertex for the definition.
25434         (classify_partition): Classify whether a partition is a
25435         PKIND_REDUCTION, thus has uses outside of the loop.
25436         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25437         Merge all PKIND_REDUCTION partitions into the last partition.
25438         (tree_loop_distribution): Seed partitions from reductions as well.
25439
25440 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25441
25442         PR tree-optimization/56125
25443         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25444         pow(x,c) into sqrt(x) * powi(x, n/2) or
25445         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25446         optimizing for size.
25447         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25448         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25449         integer.
25450
25451         PR tree-optimization/56094
25452         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25453         to UNKNOWN_LOCATION while gimplifying expr.
25454
25455 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25456
25457         PR target/56114
25458         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25459         operand 0 in movabs insn template for -masm=intel asm alternative.
25460         (*movabs<mode>_2): Ditto for operand 1.
25461
25462 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25463
25464         PR target/54663
25465         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25466         of microblaze-c.o
25467
25468 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25469
25470         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25471         tm_file.
25472
25473 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25474
25475         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25476         Undef to avoid warning.
25477
25478 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25479
25480         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25481         * configure: Regenerate.
25482
25483 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25484
25485         PR tree-optimization/56098
25486         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25487         for stmts with volatile ops.
25488         (cond_store_replacement): Don't optimize if assign has volatile ops.
25489         (cond_if_else_store_replacement_1): Don't optimize if either
25490         then_assign or else_assign have volatile ops.
25491         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25492         volatile ops.
25493
25494 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25495
25496         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25497
25498 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25499
25500         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25501         missing ':' in asm example.
25502
25503 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25504
25505         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25506         entries into lane and laneq entries.
25507         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25508         Remove AdvSIMD scalar modes.
25509         (aarch64_sq<r>dmulh_laneq<mode>): New.
25510         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25511         modes.
25512         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25513         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25514         * config/aarch64/iterators.md (VCOND): New.
25515         (VCONQ): New.
25516
25517 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25518
25519         PR target/54222
25520         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25521         Add NULL LIBNAME argument to existing definitions.
25522         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25523         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25524         * config/avr/avr.c (DEF_BUILTIN): Same.
25525         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25526         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25527         implementation is available (DECL_ASSEMBLER_NAME is set).
25528         (avr_fold_absfx): New static function.
25529         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25530         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25531         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25532         AVR_BUILTIN_ABSLLK.
25533         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25534         (abshk, absk, abslk, absllk): Provide as static inline functions.
25535
25536 2013-01-25  Marek Polacek  <polacek@redhat.com>
25537
25538         PR tree-optimization/56035
25539         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25540
25541 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25542
25543         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25544         (*movtf_internal_rex64): Add (!o,C) alternative
25545         (*movxf_internal_rex64): Ditto.
25546         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25547
25548 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25549
25550         * doc/invoke.texi: fix typo.
25551         * doc/objc.texi: fix typo.
25552
25553 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25554
25555         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25556         for the first two alternatives.
25557
25558 2013-01-24  Diego Novillo  <dnovillo@google.com>
25559
25560         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25561         (ggc-zone.o): Remove.
25562         * configure.ac: Remove option --with-gc.
25563         * configure: Re-generate.
25564         * doc/install.texi: Remove documentation for --with-gc.
25565         * gengtype.c (write_enum_defn): Remove.  Update all users.
25566         (write_Types_process_field): Remove generation of gt_e_* argument.
25567         (output_type_enum): Remove.  Update all users.
25568         (write_enum_defn): Remove.  Update all users.
25569         (enum alloc_zone): Remove.  Update all users.
25570         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25571         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25572         Update all callers.
25573         (struct ptr_data): Remove field TYPE.  Update all users.
25574         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25575         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25576         Update all users.
25577         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25578         (struct alloc_zone): Remove.
25579         (ggc_internal_alloc_zone_stat): Remove.
25580         (ggc_internal_cleared_alloc_zone_stat): Remove.
25581         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25582         (ggc_pch_count_object): Remove last argument.  Update all users.
25583         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25584         (struct alloc_zone): Remove.
25585         * ggc-zone.c: Remove.
25586         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25587         (struct alloc_zone): Remove.
25588         (ggc_alloc_typed_stat): Remove.
25589         (ggc_alloc_typed): Remove.
25590         (ggc_splay_alloc): Remove first argument.
25591         (rtl_zone): Remove.  Update all users.
25592         (tree_zone): Remove.  Update all users.
25593         (tree_id_zone): Remove.  Update all users.
25594         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25595         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25596         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25597         * tree-ssanames.c: Remove references to zone allocator in comments.
25598
25599 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25600
25601         * config/avr/avr.c (avr_out_fract): Make register numbers that
25602         might be outside of source operand signed.
25603
25604 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25605
25606         * config/i386/constraints.md (Yf): New constraint.
25607         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25608         of f constraint to conditionaly disable x87 register preferences.
25609         (*movdf_internal): Ditto.
25610         (*movsf_internal): Ditto.
25611
25612 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25613
25614         PR inline-asm/55934
25615         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25616         that have operands with impossible constraints.
25617         Add a FIXME for a speed-up opportunity.
25618         * lra-constraints.c (process_alt_operands): Verify that a class
25619         selected from constraints on asms is valid for the operand mode.
25620         (curr_insn_transform): Remove incorrect comment.
25621
25622 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25623
25624         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25625         TOC operand is a valid symbol ref in the constant pool.
25626
25627 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25628
25629         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25630
25631 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25632
25633         PR target/54222
25634         * config/avr/stdfix.h: New file.
25635         * t-avr (stdfix-gcc.h): New rule to build it.
25636         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25637
25638 2013-01-23  Kostya Serebryany  <kcc@google.com>
25639
25640         * config/darwin.h: remove dependency on
25641         CoreFoundation (asan on Mac OS).
25642
25643 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25644
25645         PR target/49069
25646         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25647         instead of cmpdi_operand for first comparison operand.
25648         Don't assert that comparison operands aren't both constants.
25649
25650 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25651
25652         * doc/install.texi (Downloading the Source): Update references to
25653         downloading separate components.
25654
25655 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25656
25657         * doc/extend.texi (__int128): Improve grammar.
25658
25659 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25660
25661         PR target/56028
25662         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25663         alternative to (o,r).
25664         (*movdi_internal_rex64): Remove (!o,n) alternative.
25665         (DImode immediate->memory splitter): Remove.
25666         (DImode immediate->memory peephole2): Remove.
25667         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25668         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25669         alternative to (!o,*r).
25670         (*movtf_internal_sse): New pattern.
25671         (*movxf_internal_rex64): New pattern.
25672         (*movxf_internal): Disable for TARGET_64BIT.
25673         (*movdf_internal_rex64): Remove (!o,F) alternative.
25674
25675 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25676
25677         PR middle-end/56074
25678         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25679         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25680         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25681         stmt locations where LOCATION_LOCUS of the stmt location is
25682         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25683
25684         PR target/55686
25685         * config/i386/i386.md (UNSPEC_STOS): New.
25686         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25687         *strsetqi_1): Add UNSPEC_STOS.
25688
25689 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25690
25691         PR c++/56067
25692         * doc/invoke.texi: Remove left over -Wsynth example.
25693
25694 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25695
25696         PR tree-optimization/56051
25697         * fold-const.c (fold_binary_loc): Don't fold
25698         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25699         a narrowing conversion, or widening conversion from signed
25700         to unsigned.
25701
25702 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25703
25704         PR rtl-optimization/56023
25705         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25706         dependent on debug instruction.
25707
25708 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25709
25710         PR middle-end/56022
25711         * function.c (allocate_struct_function): Call
25712         invoke_set_current_function_hook earlier.
25713
25714 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25715
25716         * reload1.c (init_reload): Only initialize reload_obstack
25717         during the first call.
25718
25719 2013-01-21  Marek Polacek  <polacek@redhat.com>
25720
25721         * cfgloop.c (verify_loop_structure): Fix up grammar.
25722
25723 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
25724
25725         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
25726         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
25727
25728 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25729
25730         PR target/56058
25731         * config/arm/marvell-pj4.md: Update copyright year.
25732         Fix up use of alu to alu_reg and simple_alu_imm.
25733
25734 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25735
25736         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
25737
25738 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
25739
25740         PR target/55433
25741         * lra-constraints.c (curr_insn_transform): Don't reuse original
25742         insn for secondary memory move when memory mode should be different.
25743
25744 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25745
25746         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
25747         atomic_storedi_1): New patterns.
25748
25749 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
25750
25751         btver2 pipeline descriptions.
25752         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
25753         descriptions.
25754         * config/i386/i386.md (btver2_decode): New type attributes.
25755         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
25756         type attributes.
25757         * config/i386/btver2.md: New file describing btver2 pipelines.
25758
25759 2013-01-19  Andrew Pinski  <apinski@cavium.com>
25760
25761         PR tree-optimization/52631
25762         * tree-ssa-sccvn (visit_use): Before looking up the original
25763         statement, try looking up the simplified expression.
25764
25765 2013-01-19  Anthony Green  <green@moxielogic.com>
25766
25767         * config/moxie/moxie.c (moxie_expand_prologue): Set
25768         current_function_static_stack_size.
25769
25770 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
25771
25772         PR tree-optimization/56029
25773         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
25774         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
25775
25776 2013-01-18  Sharad Singhai  <singhai@google.com>
25777
25778         PR tree-optimization/55995
25779         * dumpfile.c (dump_loc): Print location only if available.
25780         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
25781
25782 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
25783
25784         PR target/55433
25785         * lra-constraints.c (curr_insn_transform): Reuse original insn for
25786         secondary memory move.
25787         (inherit_reload_reg): Use rclass instead of cl for
25788         check_secondary_memory_needed_p.
25789
25790 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
25791
25792         PR middle-end/56015
25793         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
25794         the case where writing real complex part of target modifies op1.
25795
25796 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
25797
25798         * config/aarch64/aarch64-simd.md
25799         (aarch64_vcond_internal<mode>): Handle unordered cases.
25800         * config/aarch64/iterators.md (v_cmp_result): New.
25801
25802 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
25803             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25804
25805         * config/arm/marvell-pj4.md: New file.
25806         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
25807         * config/arm/arm.md (generic_sched): Add marvell_pj4.
25808         (generic_vfp): Likewise.
25809         * config/arm/arm-cores.def: Add marvell-pj4.
25810         * config/arm/arm-tune.md: Regenerate.
25811         * config/arm/arm-tables.opt: Regenerate.
25812         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
25813         * doc/invoke.texi: Document marvell-pj4.
25814
25815 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
25816
25817         * config/aarch64/arm_neon.h: Map scalar types to standard types.
25818
25819 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25820
25821         PR debug/54114
25822         PR debug/54402
25823         PR debug/49888
25824         * var-tracking.c (negative_power_of_two_p): New.
25825         (global_get_addr_cache, local_get_addr_cache): New.
25826         (get_addr_from_global_cache, get_addr_from_local_cache): New.
25827         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
25828         heading comment.
25829         (vt_stack_offset_p): Remove.
25830         (vt_canon_true_dep): Always canonicalize loc's address.
25831         (clobber_overlapping_mems): Make sure we have a MEM.
25832         (local_get_addr_clear_given_value): New.
25833         (val_reset): Clear local cached entries.
25834         (compute_bb_dataflow): Create and release the local cache.
25835         Disable duplicate MEMs clobbering.
25836         (emit_notes_in_bb): Clobber MEMs likewise.
25837         (vt_emit_notes): Create and release the local cache.
25838         (vt_initialize, vt_finalize): Create and release the global
25839         cache, respectively.
25840         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
25841
25842 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25843
25844         PR libmudflap/53359
25845         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
25846         not found in the symtab.
25847
25848 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25849
25850         PR debug/56006
25851         PR rtl-optimization/55547
25852         PR rtl-optimization/53827
25853         PR debug/53671
25854         PR debug/49888
25855         * alias.c (offset_overlap_p): New, factored out of...
25856         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
25857         the conservative special case for symbolic constants.  Don't
25858         adjust zero sizes on alignment.
25859
25860 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
25861
25862         PR rtl-optimization/52573
25863         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
25864         REG_UNUSED for the same register.
25865
25866 2013-01-17  Richard Biener  <rguenther@suse.de>
25867             Marek Polacek  <polacek@redhat.com>
25868
25869         PR rtl-optimization/55833
25870         * loop-unswitch.c (unswitch_loops): Move loop verification...
25871         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
25872         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
25873         Set it to true when we're removing a loop from hierarchy tree in
25874         an irreducible region.
25875         (fix_bb_placements): Adjust caller.
25876         (fix_loop_placements): Likewise.
25877
25878 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
25879
25880         * config/avr/builtins.def (DEF_BUILTIN): Factor out
25881         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
25882         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
25883         Remove ID.  Adjust comments.
25884         * config/avr/avr-c.c (avr_builtin_name): Remove.
25885         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
25886         * config/avr/avr.c (avr_tolower): New static function.
25887         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
25888         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
25889         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
25890         default expansion.
25891
25892 2013-01-17  Jan Hubicka  <jh@suse.cz>
25893
25894         PR tree-optimization/55273
25895         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
25896
25897 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
25898
25899         PR target/55981
25900         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
25901         store through atomic_store<mode>_1.
25902         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
25903
25904 2013-01-17  Martin Jambor  <mjambor@suse.cz>
25905
25906         PR tree-optimizations/55264
25907         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
25908         for virtual methods.
25909         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
25910         virtual methods before inlining is over.
25911         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
25912         virtual functions.
25913         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
25914         non-virtual.
25915
25916 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
25917
25918         PR rtl-optimization/56005
25919         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
25920         pending reads for prefetch.
25921
25922 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
25923
25924         * config/aarch64/aarch64.md
25925         (*cstoresi_neg_uxtw): New pattern.
25926         (*cmovsi_insn_uxtw): New pattern.
25927         (*<optab>si3_uxtw): New pattern.
25928         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
25929         (*<optab>si3_insn_uxtw): New pattern.
25930         (*bswapsi2_uxtw): New pattern.
25931
25932 2013-01-16  Richard Biener  <rguenther@suse.de>
25933
25934         * tree-inline.c (tree_function_versioning): Remove set but
25935         never used variable.
25936
25937 2013-01-16  Richard Biener  <rguenther@suse.de>
25938
25939         PR tree-optimization/55964
25940         * tree-flow.h (rename_variables_in_loop): Remove.
25941         (rename_variables_in_bb): Likewise.
25942         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
25943         (copy_loop_before): Adjust and delete update-ssa status.
25944         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
25945         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
25946         (rename_variables_in_loop): Remove.
25947         (slpeel_update_phis_for_duplicate_loop): Likewise.
25948         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
25949         use available cfg machinery instead of duplicating it.
25950         Update PHI nodes and perform poor-mans SSA update here.
25951         (slpeel_tree_peel_loop_to_edge): Adjust.
25952
25953 2013-01-16  Richard Biener  <rguenther@suse.de>
25954
25955         PR tree-optimization/54767
25956         PR tree-optimization/53465
25957         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
25958         (vrp_visit_phi_node): For PHI arguments coming via backedges
25959         drop all symbolical range information.
25960         (execute_vrp): Compute backedges.
25961
25962 2013-01-16  Richard Biener  <rguenther@suse.de>
25963
25964         * doc/install.texi: Update CLooG and ISL requirements to
25965         0.18.0 and 0.11.1.
25966
25967 2013-01-16  Christian Bruel  <christian.bruel@st.com>
25968
25969         PR target/55301
25970         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
25971         (broken_move): Handle UNSPECV_SP_SWITCH_B.
25972         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
25973
25974 2013-01-16  DJ Delorie  <dj@redhat.com>
25975
25976         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
25977         (UNSPECV_SP_SWITCH_E): New.
25978         (sp_switch_1): Change to an unspec.
25979         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
25980         replace $r15.
25981
25982 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
25983
25984         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
25985         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
25986         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
25987         (expand_mem_thread_fence): Ditto.
25988         (expand_mem_signal_fence): Ditto.
25989         (expand_atomic_load): Ditto.
25990         (expand_atomic_store): Ditto.
25991
25992 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
25993
25994         PR rtl-optimization/55547
25995         PR rtl-optimization/53827
25996         PR debug/53671
25997         PR debug/49888
25998         * alias.c (memrefs_conflict_p): Set sizes to negative after
25999         AND adjustments.
26000
26001 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26002
26003         PR target/55940
26004         * function.c (thread_prologue_and_epilogue_insns): Always
26005         add crtl->drap_reg to set_up_by_prologue.set, even if
26006         stack_realign_drap is false.
26007
26008 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26009
26010         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26011         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26012         *call): Fix indention.
26013
26014 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26015
26016         PR target/55876
26017         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26018         Update comment.
26019
26020 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26021
26022         PR rtl-optimization/55153
26023         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26024
26025 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26026
26027         PR tree-optimization/55920
26028         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26029         accesses as grp_to_be_debug_replaced.
26030
26031 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26032
26033         PR tree-optimization/55920
26034         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26035         there is non-useless type conversion needed from debug rhs to lhs,
26036         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26037
26038 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26039             Mikael Pettersson  <mikpe@it.uu.se>
26040
26041         PR target/43961
26042         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26043         Thumb.
26044         (ASM_OUTPUT_CASE_LABEL): Remove.
26045         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26046         * final.c (shorten_branches): Update alignment of labels before
26047         jump tables if CASE_VECTOR_SHORTEN_MODE.
26048
26049 2013-01-15  Richard Biener  <rguenther@suse.de>
26050
26051         PR bootstrap/55961
26052         * system.h: Do not include gmp.h for building host tools.
26053
26054 2013-01-15  Richard Biener  <rguenther@suse.de>
26055
26056         PR middle-end/55882
26057         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26058         account for bitpos when computing alignment.
26059
26060 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26061
26062         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26063         (ix86_target_macros_internal): Likewise.
26064
26065         * config/i386/i386.c (m_CORE2I7): Removed.
26066         (m_CORE_HASWELL): New macro.
26067         (m_CORE_ALL): Likewise.
26068         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26069         (initial_ix86_arch_features): Likewise.
26070         (processor_target_table): Initializations for Core avx2.
26071         (cpu_names): New names "core-avx2".
26072         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26073         PROCESSOR_CORE_HASWELL.
26074         (ix86_issue_rate): New case.
26075         (ia32_multipass_dfa_lookahead): Likewise.
26076         (ix86_sched_init_global): Likewise.
26077
26078         * config/i386/i386.h (TARGET_HASWELL): New macro.
26079         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26080         (processor_type): New PROCESSOR_HASWELL.
26081
26082 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26083
26084         PR tree-optimization/55955
26085         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26086         *SHIFT_EXPR and *ROTATE_EXPR codes.
26087
26088         PR tree-optimization/48766
26089         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26090         -ftrapv disable -fwrapv.
26091
26092 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26093
26094         PR target/55974
26095         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26096         etc. to 1 and not to __flash.
26097         Use LL suffix for __INT24_MAX__ with -mint8.
26098         Use ULL suffix for __UINT24_MAX__ with -mint8.
26099
26100 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26101
26102         * config/avr/avr-arch.h
26103         (struct base_arch_s): Use typedef avr_arch_t instead.
26104         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26105         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26106         * config/avr/avr.c: Same.
26107         * config/avr/avr-devices.c: Same.
26108         * config/avr/driver-avr.c: Same.
26109         * config/avr/gen-avr-mmcu-texi.c: Same.
26110         * config/avr/avr-mcus.def: Adjust comment.
26111
26112 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26113
26114         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26115         * config/aarch64/iterators.md (VALLDI): New.
26116
26117 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26118             Andi Kleen  <ak@linux.intel.com>
26119
26120         PR target/55948
26121         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26122         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26123         memmodel flag.
26124
26125 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26126
26127         * config/avr/avr-stdint.h: Remove trailing blanks.
26128         * config/avr/avr-log.h: Same.
26129         * config/avr/avr-arch.h: Same.
26130         * config/avr/avr-devices.c: Same.
26131         * config/avr/avr-dimode.md: Same.
26132         * config/avr/predicates.md: Same.
26133         * config/avr/avr-c.c: Same.  And fix typo.
26134
26135         * config/avr/avr-protos.h: Same.  And:
26136         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26137         (init_cumulative_args): Rename to avr_init_cumulative_args.
26138         (expand_prologue): Rename to avr_expand_prologue.
26139         (expand_epilogue): Rename to avr_expand_epilogue.
26140         (adjust_insn_length): Rename to avr_adjust_insn_length.
26141         (notice_update_cc): Rename to avr_notice_update_cc.
26142         (final_prescan_insn): Rename to avr_final_prescan_insn.
26143         * config/avr/avr.c: Same.
26144         * config/avr/avr.h: Same.
26145         * config/avr/avr.md: Remove trailing blanks.
26146         (prologue): Use avr_expand_prologue.
26147         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26148
26149 2013-01-14  Richard Biener  <rguenther@suse.de>
26150
26151         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26152         verify_location, collect_subblocks): New functions.
26153         (verify_gimple_in_cfg): Verify that locations only reference
26154         BLOCKs in the functions BLOCK tree.
26155
26156 2013-01-14  Richard Biener  <rguenther@suse.de>
26157
26158         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26159         PHI argument.
26160         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26161         unshare reference.
26162         (insert_out_of_ssa_copy_on_edge): Likewise.
26163         (rewrite_close_phi_out_of_ssa): Likewise.
26164         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26165         debug expressions.
26166         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26167         propagated constants.
26168         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26169         can not be shared.
26170
26171 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26172
26173         * config/avr/avr-modes.def: Add GPL copyright notice.
26174
26175 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26176
26177         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26178         MEMMODEL_MASK to determine memory model.
26179         (atomic_store<mode>): Ditto from operands[2].
26180         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26181
26182 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26183
26184         PR fortran/55935
26185         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26186         (fold_gimple_assign): Don't call unshare_expr here.
26187         (fold_ctor_reference): Call unshare_expr.
26188
26189 2013-01-13  Terry Guo  <terry.guo@arm.com>
26190
26191         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26192         * doc/fragments.texi: Document MULTILIB_REUSE.
26193         * gcc.c (multilib_reuse): New internal spec.
26194         (set_multilib_dir): Also search multilib from multilib_reuse.
26195         * genmultilib (tmpmultilib3): Refactor code.
26196         (tmpmultilib4): Ditto.
26197         (multilib_reuse): New multilib argument.
26198
26199 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26200
26201         * Makefile.in: Update copyright.
26202
26203 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26204
26205         PR middle-end/55890
26206         * calls.c (expand_call): Check if arg_nr is valid.
26207
26208 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26209
26210         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26211         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26212         documentation.  Add missing '__' in front of
26213         __builtin_ia32_packssdw256.
26214
26215 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26216
26217         PR target/55719
26218         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26219         NO_REGS for larl operands.
26220         (s390_reload_larl_operand): Use s390_load_address instead of
26221         emit_move_insn.
26222
26223 2013-01-11  Richard Biener  <rguenther@suse.de>
26224
26225         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26226         (verify_node_sharing): ... here.
26227         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26228
26229 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26230
26231         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26232         Substitute TREECHECKING.
26233         * configure: Regenerate.
26234         * Makefile.in (TREECHECKING): New.
26235
26236 2013-01-11  Richard Guenther  <rguenther@suse.de>
26237
26238         PR tree-optimization/44061
26239         * tree-vrp.c (extract_range_basic): Compute zero as
26240         value-range for __builtin_constant_p of function parameters.
26241
26242 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26243
26244         Update copyright years.
26245
26246 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26247
26248         PR rtl-optimization/55672
26249         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26250         const to be eliminable.
26251
26252 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26253
26254         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26255         * configure: Regenerate.
26256
26257 2013-01-10  Richard Biener  <rguenther@suse.de>
26258
26259         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26260
26261 2013-01-10  Richard Biener  <rguenther@suse.de>
26262
26263         PR bootstrap/55792
26264         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26265         locations for virtual PHI arguments.
26266         (rewrite_update_phi_arguments): Likewise.
26267
26268 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26269
26270         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26271         on to assembler.
26272
26273 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26274
26275         PR tree-optimization/55921
26276         * tree-complex.c (expand_complex_asm): New function.
26277         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26278
26279 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26280
26281         PR target/55718
26282         * config/s390/s390.c (s390_symref_operand_p)
26283         (s390_loadrelative_operand_p): Merge the two functions.
26284         (s390_check_qrst_address, print_operand_address): Add parameters
26285         to s390_loadrelative_operand_p invokation.
26286         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26287         (s390_reload_larl_operand, s390_secondary_reload): Use
26288         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26289         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26290
26291 2013-01-09  Mike Stump  <mikestump@comcast.net>
26292
26293         * dse.c (record_store): Remove unnecessary assert.
26294
26295 2013-01-09  Jan Hubicka  <jh@suse.cz>
26296
26297         PR tree-optimization/55569
26298         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26299         * cfgloop.h (scale_loop_profile): Likewise.
26300
26301 2013-01-09  Jan Hubicka  <jh@suse.cz>
26302
26303         PR lto/45375
26304         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26305         functions.
26306         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26307
26308 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26309
26310         PR middle-end/55114
26311         * expr.h (maybe_emit_group_store): Declare.
26312         * expr.c (maybe_emit_group_store): New function.
26313         * builtins.c (expand_builtin_int_roundingfn): Call it.
26314         (expand_builtin_int_roundingfn_2): Likewise.
26315
26316 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26317
26318         PR rtl-optimization/55829
26319         * lra-constraints.c (match_reload): Add code for absent output.
26320         (curr_insn_transform): Add code for reloads of matched inputs
26321         without output.
26322
26323 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26324
26325         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26326         attribute of movddup insn to DF.
26327         (*vec_interleave_lowv2df): Ditto.
26328         (vec_dupv2df): Ditto.
26329
26330 2013-01-09  Jan Hubicka  <jh@suse.cz>
26331
26332         PR tree-optimiation/55875
26333         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26334         EVERY_ITERATION parameter.
26335         (number_of_iterations_exit): Check if exit is executed every iteration.
26336         (idx_infer_loop_bounds): Similarly here.
26337         (n_of_executions_at_most): Simplify
26338         to only test for cases where statement is dominated by the
26339         particular bound; handle correctly the "postdominance" test.
26340         (scev_probably_wraps_p): Use max loop iterations info
26341         as a global bound first.
26342
26343 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26344             Nick Clifton  <nickc@redhat.com>
26345
26346         * config/v850/v850.md (cbranchsf4): New pattern.
26347         (cstoresf4): New pattern.
26348         (cbranchdf4): New pattern.
26349         (cstoredf4): New pattern.
26350         (movsicc): Disallow floating point comparisons.
26351         (cmpsf_le_insn): Fix order of operators.
26352         (cmpsf_lt_insn): Likewise.
26353         (cmpsf_eq_insn): Likewise.
26354         (cmpdf_le_insn): Likewise.
26355         (cmpdf_lt_insn): Likewise.
26356         (cmpdf_eq_insn): Likewise.
26357         (cmpsf_ge_insn): Use LE comparison.
26358         (cmpdf_ge_insn): Likewise.
26359         (cmpsf_gt_insn): Use LT comparison.
26360         (cmpdf_gt_insn): Likewise.
26361         (cmpsf_ne_insn): Delete pattern.
26362         (cmpdf_ne_insn): Delete pattern.
26363         * config/v850/v850.c (v850_gen_float_compare): Use
26364         gen_cmpdf_eq_insn for NE comparison.
26365         (v850_float_z_comparison_operator)
26366         (v850_float_nz_comparison_operator): Move from here ...
26367         * config/v850/predicates.md: ... to here.  Move GT and GE
26368         comparisons into v850_float_z_comparison_operator.
26369         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26370         Delete prototype.
26371         (v850_float_nz_comparison_operator): Likewise.
26372
26373 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26374
26375         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26376         with calls to gen_insvsi/gen_insvdi.
26377
26378 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26379
26380         * config/i386/i386.c (initial_ix86_tune_features): Set up
26381         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26382
26383 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26384             Jakub Jelinek  <jakub@redhat.com>
26385
26386         PR tree-optimization/48189
26387         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26388         If nitercst is 0, don't predict the exit edge.
26389
26390 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26391
26392         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26393         in asm_fprintf with reg_names.
26394         (aarch64_print_operand_address): Likewise.
26395         (aarch64_return_addr): Likewise.
26396         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26397
26398 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26399
26400         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26401         (INT_U6_BITS): Likewise.
26402         * config/pa/predicates.md (uint6_operand): New predicate.
26403         (shift5_operand, shift6_operand): Likewise.
26404         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26405         arith32_operand.
26406         (lshrdi3): Use shift6_operand.
26407         (shrpsi4, shrpdi4): New insn patterns.
26408         (extzv): Delete expander.
26409         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26410         predicates in unamed zero extract patterns.  Tighten common constraint.
26411         (extv): Delete expander.
26412         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26413         predicates in unamed sign extract patterns.  Tighten common constraint.
26414         (insv): Delete expander.
26415         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26416         predicates in unamed insert patterns.  Tighten common constraint.
26417         Change uint32_operand predicate to uint6_operand predicate in unamed
26418         DImode pattern to insert constant values of type 1...1xxxx.
26419
26420 2013-01-04  Jan Hubicka  <jh@suse.cz>
26421
26422         PR tree-optimization/55823
26423         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26424         issue.
26425
26426 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26427             Uros Bizjak  <ubizjak@gmail.com>
26428
26429         PR rtl-optimization/55845
26430         * df-problems.c (can_move_insns_across): Stop scanning at
26431         volatile_insn_p source instruction or give up if
26432         across_from .. across_to range contains any volatile_insn_p
26433         instructions.
26434
26435 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26436
26437         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26438         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26439         Declare.
26440         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26441         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26442
26443 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26444
26445         PR fortran/55341
26446         * asan.c (asan_clear_shadow): New function.
26447         (asan_emit_stack_protection): Use it.
26448
26449 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26450
26451         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26452         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26453         with tab instead of space.
26454
26455 2013-01-08  Nick Clifton  <nickc@redhat.com>
26456
26457         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26458         register bank 0 at the start of an interrupt handler.
26459         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26460         MDBH registers.
26461
26462 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26463
26464         * config/aarch64/aarch64-simd.md
26465         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26466         (aarch64_simd_bsl): Likewise.
26467         (aarch64_vcond_internal<mode>): Likewise.
26468         (vcond<mode><mode>): Likewise.
26469         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26470         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26471
26472 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26473
26474         * config/aarch64/aarch64-builtins.c
26475         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26476
26477 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26478
26479         PR debug/55579
26480         * tree-sra.c (analyze_access_subtree): Return true also after
26481         potentially creating a debug-only replacement.
26482
26483 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26484
26485         PR middle-end/55890
26486         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26487
26488         PR tree-optimization/54120
26489         * tree-vrp.c (range_fits_type_p): Don't allow
26490         src_precision < precision from signed vr to unsigned_p
26491         if vr->min or vr->max is negative.
26492         (simplify_float_conversion_using_ranges): Test can_float_p
26493         against CODE_FOR_nothing.
26494
26495 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26496             Richard Biener  <rguenther@suse.de>
26497
26498         PR middle-end/55851
26499         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26500         types instead of just INTEGER_TYPE types.
26501
26502 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26503
26504         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26505         TF_SIZE): Define.
26506
26507 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26508
26509         PR target/42661
26510         * config/mips/mips.opt: Change mad to mmad to match documentation.
26511
26512 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26513
26514         PR target/55897
26515         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26516         .progmemx.data now.
26517
26518 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26519
26520         PR target/55897
26521         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26522         (avr_addrspace_t): Add .section_name field.
26523         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26524         array size.
26525         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26526         NULL entry.  Put __memx into .progmemx.data.
26527         (progmem_section_prefix): Remove.
26528         (avr_asm_init_sections): No need to initialize progmem_section.
26529         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26530         section name prefix.
26531         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26532         retrieve the progmem section.
26533         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26534         boundary to run over avr_addrspace[].
26535         (avr_register_target_pragmas): Ditto.
26536
26537 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26538
26539         * varasm.c (output_constant_def_contents): For asan_protect_global
26540         protected strings, adjust DECL_ALIGN if needed, before testing for
26541         anchored symbols.
26542         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26543         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26544         normal decls.
26545         (output_object_block): For asan protected decls, emit asan padding
26546         after their contents.
26547         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26548         (asan_finish_file): Test it here instead.
26549
26550 2013-01-07  Nick Clifton  <nickc@redhat.com>
26551             Matthias Klose  <doko@debian.org>
26552             Doug Kwan  <dougkwan@google.com>
26553             H.J. Lu  <hongjiu.lu@intel.com>
26554
26555         PR driver/55470
26556         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26557
26558         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26559
26560         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26561
26562         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26563
26564         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26565
26566 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26567
26568         PR target/54461
26569         * doc/install.texi (Cross-Compiler-Specific Options): Document
26570         --with-avrlibc.
26571
26572 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26573
26574         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26575         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26576         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26577         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26578         vqmovun_high_s64): Fix source operand number and update copyright.
26579
26580 2013-01-07  Richard Biener  <rguenther@suse.de>
26581
26582         PR middle-end/55890
26583         * gimple.h (gimple_call_builtin_p): New overload.
26584         * gimple.c (validate_call): New function.
26585         (gimple_call_builtin_p): Likewise.
26586         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26587         Use gimple_call_builtin_p.
26588         (find_func_clobbers): Likewise.
26589         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26590         (strlen_optimize_stmt): Likewise.
26591
26592 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26593
26594         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26595         (vld1q_dup_*): Likewise.
26596         (vld1_*): Likewise.
26597         (vld1q_*): Likewise.
26598         (vld1_lane_*): Likewise.
26599         (vld1q_lane_*): Likewise.
26600
26601 2013-01-07  Richard Biener  <rguenther@suse.de>
26602
26603         * lto-streamer.h (LTO_minor_version): Bump to 2.
26604
26605 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26606
26607         * config/aarch64/aarch64-protos.h
26608         (aarch64_const_double_zero_rtx_p): Rename to...
26609         (aarch64_float_const_zero_rtx_p): ...this.
26610         (aarch64_float_const_representable_p): New.
26611         (aarch64_output_simd_mov_immediate): Likewise.
26612         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26613         move immediate case.
26614         * config/aarch64/aarch64.c
26615         (aarch64_const_double_zero_rtx_p): Rename to...
26616         (aarch64_float_const_zero_rtx_p): ...this.
26617         (aarch64_print_operand): Allow printing of new constants.
26618         (aarch64_valid_floating_const): New.
26619         (aarch64_legitimate_constant_p): Check for valid floating-point
26620         constants.
26621         (aarch64_simd_valid_immediate): Likewise.
26622         (aarch64_vect_float_const_representable_p): New.
26623         (aarch64_float_const_representable_p): Likewise.
26624         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26625         (aarch64_output_simd_mov_immediate): New.
26626         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26627         (*movdf_aarch64): Likewise.
26628         * config/aarch64/constraints.md (Ufc): New.
26629         (Y): call aarch64_float_const_zero_rtx.
26630         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26631
26632 2013-01-07  Richard Biener  <rguenther@suse.de>
26633
26634         PR tree-optimization/55888
26635         PR tree-optimization/55862
26636         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26637         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26638         not if it is contained therein.
26639
26640 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26641
26642         * config/avr/t-avr: Typo.
26643
26644 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26645
26646         PR55243
26647         * config/avr/t-avr: Don't automatically rebuild
26648         $(srcdir)/config/avr/t-multilib
26649         $(srcdir)/config/avr/avr-tables.opt
26650         $(srcdir)/doc/avr-mmcu.texi
26651         (avr-mcus): New phony target to build them on request.
26652         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26653         * avr/avr-mcus.def: Adjust comments.
26654
26655 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26656
26657         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26658
26659 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26660
26661         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26662
26663 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26664
26665         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26666
26667 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26668
26669         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26670         to generate profiling.
26671         * config/rs6000/aix64.h (LIB_SPEC): Same.
26672
26673 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26674
26675         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26676         New function.
26677         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26678
26679 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26680
26681         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26682         unconditionally.
26683         (ix86_expand_move): Ditto.
26684         (ix86_zero_extend_to_Pmode): Ditto.
26685         (ix86_expand_call): Ditto.
26686         (ix86_expand_special_args_builtin): Ditto.
26687         (ix86_expand_builtin): Ditto.
26688
26689 2013-01-04  Richard Biener  <rguenther@suse.de>
26690
26691         PR tree-optimization/55862
26692         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26693         translating them through PHI nodes.
26694
26695 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26696
26697         PR tree-optimization/55755
26698         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26699         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26700
26701 2013-01-04  Marek Polacek  <polacek@redhat.com>
26702
26703         PR middle-end/55859
26704         * opts.c (default_options_optimization): Clarify error message.
26705
26706 2013-01-04  Richard Biener  <rguenther@suse.de>
26707
26708         PR middle-end/55863
26709         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26710         reassociation.
26711
26712 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26713
26714         PR target/53789
26715         * config/pa/pa.md (movsi): Revert previous change.
26716         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
26717         references.
26718
26719 2013-01-03  Richard Henderson  <rth@redhat.com>
26720
26721         * config/i386/i386.c (ix86_expand_move): Always assign to op1
26722         after eliminating TLS symbols.
26723
26724 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
26725
26726         PR bootstrap/50167
26727         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
26728         * graphite-poly.c (debug_gmp_value): Likewise.
26729
26730 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
26731
26732         PR target/55712
26733         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
26734         selected code model, define __code_mode_small__, __code_model_medium__,
26735         __code_model_large__, __code_model_32__ or __code_model_kernel__.
26736         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
26737         xchg temporary register with %k.  Declare temporary register as
26738         early clobbered.
26739         [__x86_64__]: For medium and large code models, preserve %rbx register.
26740
26741 2013-01-03  Richard Biener  <rguenther@suse.de>
26742
26743         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
26744         (dump_subscript): Adjust.
26745         (finalize_ddr_dependent): Do not dump redundant info.
26746         (analyze_siv_subscript): Adjust.
26747         (subscript_dependence_tester): Likewise.
26748         (compute_affine_dependence): Likewise.
26749
26750 2013-01-03  Richard Biener  <rguenther@suse.de>
26751
26752         Revert
26753         2013-01-03  Richard Biener  <rguenther@suse.de>
26754
26755         PR tree-optimization/55857
26756         * tree-vect-stmts.c (vectorizable_load): Do not setup
26757         re-alignment for invariant loads.
26758
26759         2013-01-02  Richard Biener  <rguenther@suse.de>
26760
26761         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26762         invariant load do not generate a vector load from the scalar location.
26763
26764 2013-01-03  Richard Biener  <rguenther@suse.de>
26765
26766         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
26767         for not vectorizing.
26768         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
26769         not build INDIRECT_REFs, call get_name once only.
26770         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
26771         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
26772
26773 2013-01-03  Richard Biener  <rguenther@suse.de>
26774
26775         PR tree-optimization/55857
26776         * tree-vect-stmts.c (vectorizable_load): Do not setup
26777         re-alignment for invariant loads.
26778
26779 2013-01-03  Richard Biener  <rguenther@suse.de>
26780
26781         PR lto/55848
26782         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
26783         prefer a built-in decl.
26784
26785 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26786
26787         * gcc.c (process_command): Update copyright notice dates.
26788         * gcov.c (print_version): Likewise.
26789         * gcov-dump.c (print_version): Likewise.
26790
26791         PR rtl-optimization/55838
26792         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
26793         iv0.step, iv1.step and step.
26794
26795 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26796             Marc Glisse  <marc.glisse@inria.fr>
26797
26798         PR tree-optimization/55832
26799         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
26800         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
26801         integer_{one,zero}_node.
26802
26803 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26804
26805         PR debug/54402
26806         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
26807         * var-tracking.c (reverse_op): Don't add reverse ops to
26808         VALUEs that have already
26809         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
26810
26811 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
26812
26813         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
26814
26815 2013-01-02  Teresa Johnson  <tejohnson@google.com>
26816
26817         * dumpfile.c (dump_loc): Print filename with location.
26818         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
26819         new location_t parameter to emit complete unroll message with
26820         new dump framework.
26821         (canonicalize_loop_induction_variables): Compute loops location
26822         and pass to try_unroll_loop_completely.
26823         * loop-unroll.c (report_unroll_peel): New function.
26824         (peel_loops_completely): Use new dump format with location
26825         for main dumpfile message, and invoke report_unroll_peel on success.
26826         (decide_unrolling_and_peeling): Ditto.
26827         (decide_peel_once_rolling): Remove old dumpfile message subsumed
26828         by report_unroll_peel.
26829         (decide_peel_completely): Ditto.
26830         (decide_unroll_constant_iterations): Ditto.
26831         (decide_unroll_runtime_iterations): Ditto.
26832         (decide_peel_simple): Ditto.
26833         (decide_unroll_stupid): Ditto.
26834         * cfgloop.c (get_loop_location): New function.
26835         * cfgloop.h (get_loop_location): Declare.
26836
26837 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
26838
26839         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
26840         NULL.
26841
26842 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26843
26844         PR middle-end/55198
26845         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
26846         BLKmode objects when EXPAND_MEMORY is specified.
26847
26848 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
26849
26850         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
26851         in loop predicate.
26852         (fold_builtin_cpu): Do not share cpu model decls across statements.
26853
26854 2013-01-02  Jason Merrill  <jason@redhat.com>
26855
26856         PR c++/55804
26857         * tree.c (build_array_type_1): Revert earlier change.
26858
26859 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
26860
26861         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
26862         "cortex-a57".
26863         * config/aarch64/aarch64-tune.md: Re-generate.
26864
26865 2013-01-02  Richard Biener  <rguenther@suse.de>
26866
26867         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26868         invariant load do not generate a vector load from the scalar location.
26869
26870 2013-01-02  Richard Biener  <rguenther@suse.de>
26871
26872         PR bootstrap/55784
26873         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
26874         * configure: Regenerate.
26875
26876 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
26877
26878         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
26879         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
26880         (expand_builtin_int_roundingfn_2): Keep the original target around
26881         for the fallback case.
26882
26883 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
26884
26885         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
26886         to be clear for sign changes.
26887
26888 2013-01-01  Jan Hubicka  <jh@suse.cz>
26889
26890         * ipa-inline-analysis.c: Fix formatting.
26891
26892 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
26893
26894         PR tree-optimization/55831
26895         * tree-vect-loop.c (get_initial_def_for_induction): Use
26896         gsi_after_labels instead of gsi_start_bb.
26897 \f
26898 Copyright (C) 2013 Free Software Foundation, Inc.
26899
26900 Copying and distribution of this file, with or without modification,
26901 are permitted in any medium without royalty provided the copyright
26902 notice and this notice are preserved.