gcc/ChangeLog

   1 2011-11-09  Richard Guenther  <rguenther@suse.de>
   2
   3         PR tree-optimization/51039
   4         * tree-cfg.c (verify_gimple_call): Verify that
   5         gimple_call_cannot_inline_p is returning a conservative
   6         correct result according to gimple_check_call_matching_types.
   7         * ipa-inline-analysis.c (estimate_function_body_sizes): Remove
   8         code dealing with un-inlinablility.
   9         * gimple-streamer-in.c (input_gimple_stmt): Update the
  10         non-inlinable flag.
  11
  12 2011-11-09  Richard Guenther  <rguenther@suse.de>
  13
  14         PR tree-optimization/51039
  15         * tree-inline.c (setup_one_parameter): Always perform a
  16         valid gimple type change.
  17         (declare_return_variable): Likewise.
  18
  19 2011-11-09  Jakub Jelinek  <jakub@redhat.com>
  20
  21         * config/rs6000/vector.md (vcondv4sfv4si, vcondv4siv4sf,
  22         vconduv4sfv4si): New patterns.
  23         * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Handle
  24         different dest_mode from comparison mode.
  25
  26 2011-11-09  Richard Guenther  <rguenther@suse.de>
  27
  28         * gimple-fold.c (canonicalize_constructor_val): Mark
  29         address-taken variables addressable.
  30
  31 2011-11-09  Ira Rosen  <ira.rosen@linaro.org>
  32
  33         PR tree-optimization/51015
  34         * tree-vect-loop.c (vect_determine_vectorization_factor): Expect
  35         vectype to be set for pattern def stmts.
  36
  37 2011-11-09  Alan Modra  <amodra@gmail.com>
  38
  39         * function.c (bb_active_p): Delete.
  40         (dup_block_and_redirect, active_insn_between): New functions.
  41         (convert_jumps_to_returns, emit_return_for_exit): New functions,
  42         split out from..
  43         (thread_prologue_and_epilogue_insns): ..here.  Delete
  44         shadowing variables.  Don't do prologue register clobber tests
  45         when shrink wrapping already failed.  Delete all last_bb_active
  46         code.  Instead compute tail block candidates for duplicating
  47         exit path.  Remove these from antic set.  Duplicate tails when
  48         reached from both blocks needing a prologue/epilogue and
  49         blocks not needing such.
  50         * ifcvt.c (dead_or_predicable): Test both flag_shrink_wrap and
  51         HAVE_simple_return.
  52         * bb-reorder.c (get_uncond_jump_length): Make global.
  53         * bb-reorder.h (get_uncond_jump_length): Declare.
  54         * cfgrtl.c (rtl_create_basic_block): Comment typo fix.
  55         (rtl_split_edge): Likewise.  Warning fix.
  56         (rtl_duplicate_bb): New function.
  57         (rtl_cfg_hooks): Enable can_duplicate_block_p and duplicate_block.
  58         * Makefile.in (function.o): Update dependencies.
  59
  60 2011-11-08  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
  61             Georg-Johann Lay  <avr@gjlay.de>
  62
  63         * doc/tm.texi.in (MODE_CODE_BASE_REG_CLASS): Add address space
  64         argument.
  65         (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
  66         * doc/tm.texi: Regenerate.
  67
  68         * config/cris/cris.h (MODE_CODE_BASE_REG_CLASS): Add address
  69         space argument.
  70         (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
  71         * config/bfin/bfin.h (MODE_CODE_BASE_REG_CLASS): Likewise.
  72         (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
  73         * config/avr/avr.h (MODE_CODE_BASE_REG_CLASS): Add address space
  74         argument.
  75         (REGNO_MODE_CODE_OK_FOR_BASE_P): Ditto.
  76         * config/avr/avr-protos.h (avr_mode_code_base_reg_class): Ditto.
  77         (avr_regno_mode_code_ok_for_base_p): Ditto.
  78         * config/avr/avr.c (avr_mode_code_base_reg_class): Ditto.
  79         (avr_regno_mode_code_ok_for_base_p): Ditto.
  80         (avr_reg_ok_for_addr_p): Pass AS down to
  81         avr_regno_mode_code_ok_for_base_p.
  82
  83         * addresses.h (base_reg_class): Add address space argument.
  84         Pass to MODE_CODE_BASE_REG_CLASS.
  85         (ok_for_base_p_1): Add address space argument.  Pass to
  86         REGNO_MODE_CODE_OK_FOR_BASE_P.
  87         (regno_ok_for_base_p): Add address space argument.  Pass to
  88         ok_for_base_p_1.
  89
  90         * regrename.c (scan_rtx_address): Add address space argument.
  91         Pass address space to regno_ok_for_base_p and base_reg_class.
  92         Update recursive calls.
  93         (scan_rtx): Pass address space to scan_rtx_address.
  94         (build_def_use): Likewise.
  95         * regcprop.c (replace_oldest_value_addr): Add address space
  96         argument.  Pass to regno_ok_for_base_p and base_reg_class.
  97         Update recursive calls.
  98         (replace_oldest_value_mem): Pass address space to
  99         replace_oldest_value_addr.
 100         (copyprop_hardreg_forward_1): Likewise.
 101
 102         * reload.c (find_reloads_address_1): Add address space argument.
 103         Pass address space to base_reg_class and regno_ok_for_base_p.
 104         Update recursive calls.
 105         (find_reloads_address): Pass address space to base_reg_class,
 106         regno_ok_for_base_p, and find_reloads_address_1.
 107         (find_reloads): Pass address space to base_reg_class.
 108         (find_reloads_subreg_address): Likewise.
 109
 110         * ira-costs.c (record_reg_classes): Update calls to base_reg_class.
 111         (ok_for_base_p_nonstrict): Add address space argument.  Pass to
 112         ok_for_base_p_1.
 113         (record_address_regs): Add address space argument.  Pass to
 114         base_reg_class and ok_for_base_p_nonstrict.  Update recursive calls.
 115         (record_operand_costs): Pass address space to record_address_regs.
 116         (scan_one_insn): Likewise.
 117
 118         * caller-save.c (init_caller_save): Update call to base_reg_class.
 119         * ira-conflicts.c (ira_build_conflicts): Likewise.
 120         * reload1.c (maybe_fix_stack_asms): Likewise.
 121
 122 2011-11-08  Michael Matz  <matz@suse.de>
 123
 124         * gengtype.c (write_field_root): Avoid out-of-scope access of newv.
 125
 126         * tree-stdarg.c (execute_optimize_stdarg): Accept clobbers.
 127
 128         * tree.h (TREE_CLOBBER_P): New macro.
 129         * gimple.h (gimple_clobber_p): New inline function.
 130         * gimplify.c (gimplify_bind_expr): Add clobbers for all variables
 131         that go out of scope and live in memory.
 132         * tree-ssa-operands.c (get_expr_operands): Transfer volatility also
 133         for constructors.
 134         * cfgexpand.c (decl_to_stack_part): New static variable.
 135         (add_stack_var): Allocate it, and remember mapping.
 136         (fini_vars_expansion): Deallocate it.
 137         (stack_var_conflict_p): Add early outs.
 138         (visit_op, visit_conflict, add_scope_conflicts_1,
 139         add_scope_conflicts): New static functions.
 140         (expand_used_vars_for_block): Don't call add_stack_var_conflict, tidy.
 141         (expand_used_vars): Add scope conflicts.
 142         (expand_gimple_stmt_1): Expand clobbers to nothing.
 143         (expand_debug_expr): Ditto.
 144
 145         * tree-pretty-print.c (dump_generic_node): Dump clobbers nicely.
 146         * tree-ssa-live.c (remove_unused_locals): Remove clobbers that
 147         refer to otherwise unused locals.
 148         * tree-sra.c (build_accesses_from_assign): Ignore clobbers.
 149         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Clobbers of
 150         SSA names aren't necessary.
 151         (propagate_necessity): Accept and ignore constructors on the rhs,
 152         tidy.
 153         * gimple.c (walk_gimple_op): Accept constructors like mem_rhs.
 154         * tree-ssa-structalias.c (find_func_aliases): Clobbers don't store
 155         any known value.
 156         * tree-ssa-sccvn.c (vn_reference_lookup_3): Ditto, in particular they
 157         don't zero-initialize something.
 158         * tree-ssa-phiopt.c (cond_if_else_store_replacement_1): Ignore
 159         clobber RHS, we don't want PHI nodes with those.
 160
 161 2011-11-08  Jakub Jelinek  <jakub@redhat.com>
 162
 163         * config/i386/i386.c (ix86_preferred_simd_mode): Even for TARGET_AVX
 164         if not TARGET_PREFER_AVX128 return 32-byte integer vectors.
 165
 166 2011-11-08  Richard Guenther  <rguenther@suse.de>
 167
 168         PR tree-optimization/51012
 169         * ipa-prop.c (update_indirect_edges_after_inlining): Fixup
 170         non-inlinable state.
 171         * cgraph.c (cgraph_make_edge_direct): Likewise.
 172
 173 2011-11-08  Eric Botcazou  <ebotcazou@adacore.com>
 174
 175         PR rtl-optimization/47698
 176         * ifcvt.c (noce_operand_ok): Move around comment.
 177
 178 2011-11-08  Richard Guenther  <rguenther@suse.de>
 179
 180         PR lto/50999
 181         * lto-opts.c (append_to_collect_gcc_options): Split out from...
 182         (lto_write_options): ... here.  Prepend frontend specific flags.
 183
 184 2011-11-08  Jakub Jelinek  <jakub@redhat.com>
 185
 186         * config/i386/sse.md (*avx_vperm2f128<mode>3_nozero): Emit mask
 187         0x12 and 0x20 as vinsert[fi]128 instead of vperm2[fi]128.
 188
 189         * tree-vect-stmts.c (vectorizable_call): Add SLP_NODE argument.
 190         Handle vectorization of SLP calls.
 191         (vect_analyze_stmt): Adjust caller, add call to it for SLP too.
 192         (vect_transform_stmt): Adjust vectorizable_call caller, remove
 193         assertion.
 194         * tree-vect-slp.c (vect_get_and_check_slp_defs): For calls start
 195         with op_idx 3.
 196         (vect_build_slp_tree): Allow CALL_EXPR.
 197
 198 2011-11-08  Richard Guenther  <rguenther@suse.de>
 199
 200         * gimple-fold.c (canonicalize_constructor_val): Make sure
 201         we have referenced vars setup before adding to them.
 202
 203 2011-11-07  Richard Henderson  <rth@redhat.com>
 204             Aldy Hernandez  <aldyh@redhat.com>
 205             Andrew MacLeod  <amacleod@redhat.com>
 206             Torvald Riegel  <triegel@redhat.com>
 207
 208         Merged from transactional-memory.
 209
 210         * gtm-builtins.def: New file.
 211         * trans-mem.c: New file.
 212         * trans-mem.h: New file.
 213
 214         * opts.c (finish_options): Error out when using -flto and
 215         -fgnu-tm.
 216
 217         * config/i386/i386.c: Define TARGET_VECTORIZE* transactional variants.
 218         (ix86_handle_tm_regparm_attribute, struct bdesc_tm,
 219         ix86_builtin_tm_load, ix86_builtin_tm_store,
 220         ix86_init_tm_builtins): New.
 221         (ix86_init_builtins): Initialize TM builtins.
 222         (struct ix86_attribute_table): Add "*tm regparm".
 223         * config/i386/i386-builtin-types.def (PV2SI): Define.
 224         (PCV2SI): Define.
 225         Define V2SI_FTYPE_PCV2SI.
 226         Define V4SF_FTYPE_PCV4SF.
 227         Define V8SF_FTYPE_PCV8SF.
 228         Define VOID_PV2SI_V2SI.
 229
 230         * doc/invoke.texi (C Dialect Options): Document -fgnu-tm and
 231         tm-max-aggregate-size.
 232         * doc/tm.texi.in: Add TARGET_VECTORIZE_BUILTIN_TM_LOAD and
 233         TARGET_VECTORIZE_BUILTIN_TM_STORE hooks.
 234         * doc/tm.texi: Regenerate.
 235
 236         * attribs.c (apply_tm_attr): New.
 237         (init_attributes): Allow '*' prefix for overrides.
 238         (register_attribute): Likewise.
 239         * builtin-attrs.def (ATTR_TM_TMPURE, ATTR_TM_REGPARM): New.
 240         (ATTR_TM_NOTHROW_LIST, ATTR_TM_TMPURE_NOTHROW_LIST,
 241         ATTR_TM_PURE_TMPURE_NOTHROW_LIST, ATTR_TM_NORETURN_NOTHROW_LIST,
 242         ATTR_TM_CONST_NOTHROW_LIST, ATTR_TMPURE_MALLOC_NOTHROW_LIST,
 243         ATTR_TMPURE_NOTHROW_LIST): New.
 244         * builtin-types.def (BT_FN_I[1248]_VPTR, BT_FN_FLOAT_VPTR,
 245         BT_FN_DOUBLE_VPTR, BT_FN_LDOUBLE_VPTR, BT_FN_VOID_VPTR_I[1248],
 246         BT_FN_VOID_VPTR_FLOAT, BT_FN_VOID_VPTR_DOUBLE,
 247         BT_FN_VOID_VPTR_LDOUBLE, BT_FN_VOID_VPTR_SIZE): New.
 248         * builtins.def: Include gtm-builtins.def. Add comments regarding
 249         transactional memory synchronization.
 250         (DEF_TM_BUILTIN): New.
 251         * c-parser.c (struct c_parser): Add in_transaction.
 252         (c_parser_transaction, c_parser_transaction_expression,
 253         c_parser_transaction_cancel, c_parser_transaction_attributes): New.
 254         (c_parser_attribute_any_word): Split out from c_parser_attributes.
 255         (c_parser_statement_after_labels): Handle RID_TRANSACTION*.
 256         (c_parser_unary_expression): Same.
 257         * c-tree.h (c_finish_transaction): Declare.
 258         * c-typeck.c (c_finish_transaction): New.
 259         (build_function_call_vec): Call tm_malloc_replacement.
 260         * calls.c (is_tm_builtin): New.
 261         (flags_from_decl_or_type): Add ECF_TM_BUILTIN and ECF_TM_PURE.
 262         * cfgbuild.c (make_edges): Add edges for REG_TM notes.
 263         * cfgexpand.c (expand_call_stmt): Call
 264         mark_transaction_restart_calls.
 265         (gimple_expand_cfg): Free the tm_restart map.
 266         (mark_transaction_restart_calls): New.
 267         * cfgrtl.c (purge_dead_edges): Look for REG_TM notes.
 268         * cgraph.c (dump_cgraph_node): Handle tm_clone.
 269         * cgraph.h (struct cgraph_node): Add tm_clone field.
 270         (decl_is_tm_clone): New.
 271         (struct cgraph_local_info): Add tm_may_enter_irr.
 272         (cgraph_copy_node_for_versioning): Declare.
 273         * cgraphunit.c (cgraph_copy_node_for_versioning): Export;
 274         copy analyzed from old version.
 275         * combine.c (distribute_notes): Handle REG_TM notes.
 276         * common.opt: Add -fgnu-tm.
 277         * crtstuff.c (__TMC_LIST__, __TMC_END__): New.
 278         (__do_global_dtors_aux): Deregister clone table.
 279         (frame_dummy): Register clone table.
 280         * emit-rtl.c (try_split): Handle REG_TM. Early return if no function
 281         body.
 282         * gimple-low.c (lower_stmt): Handle GIMPLE_EH_ELSE and
 283         GIMPLE_TRANSACTION.
 284         (gimple_stmt_may_fallthru): Handle GIMPLE_EH_ELSE.
 285         * gimple-pretty-print.c: Include trans-mem.h.
 286         (dump_gimple_fmt): Add %x.
 287         (dump_gimple_call): Dump arguments for calls to _ITM_beginTransaction.
 288         (dump_gimple_eh_else, dump_gimple_transaction): New.
 289         (dump_gimple_stmt): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
 290         * gimple.c (gimple_build_eh_else, gimple_build_transaction): New.
 291         (walk_gimple_seq): Honor removed_stmt. Document usage of removed_stmt
 292         field.
 293         (walk_gimple_op): Handle GIMPLE_TRANSACTION.
 294         (walk_gimple_stmt): Initialize and honor removed_stmt.
 295         Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
 296         (gimple_copy): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
 297         * gimple.def (GIMPLE_TRANSACTION, GIMPLE_EH_ELSE): New.
 298         * gimple.h (struct gimple_statement_eh_else,
 299         gimple_statement_transaction, GTMA_*): New.
 300         (gimple_statement_d): Add gimple_statement_eh_else and
 301         gimple_transaction.
 302         (gimple_build_eh_else, gimple_build_transaction,
 303         gimple_fold_call, diagnose_tm_safe_errors): Declare.
 304         (get_call_expr_in): Remove prototype.
 305         (gimple_has_substatements): Add GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
 306         (gimple_eh_else_n_body, gimple_eh_else_e_body,
 307         gimple_eh_else_set_n_body, gimple_eh_else_set_e_body,
 308         gimple_transaction_body, gimple_transaction_label,
 309         gimple_transaction_label_ptr, gimple_transaction_subcode,
 310         gimple_transaction_set_body, gimple_transaction_set_label,
 311         gimple_transaction_set_subcode): New.
 312         (struct walk_stmt_info): Use BOOL_BITFIELD; add removed_stmt.
 313         * gimplify.c (create_tmp_var_name): Use clean_symbol_name.
 314         (voidify_wrapper_expr): Handle TRANSACTION_EXPR.
 315         (gimplify_transaction): New.
 316         (gimplify_expr): Handle TRANSACTION_EXPR.
 317         * gsstruct.def (GSS_EH_ELSE, GSS_TRANSACTION): New.
 318         * ipa-inline.c (can_inline_edge_p): Do not inline TM safe calling
 319         TM pure functions.
 320         * Makefile.in: Add trans-mem.o and dependencies.
 321         (BUILTINS_DEF): Add gtm-builtins.def.
 322         (gimple-pretty-print.o): Depend on TRANS_MEM_H.
 323         (GTFILES): Add trans-mem.c.
 324         * omp-low.c (WALK_SUBSTMTS): Add GIMPLE_TRANSACTION.
 325         * output.h (record_tm_clone_pair, finish_tm_clone_pairs,
 326         get_tm_clone_pair): Declare.
 327         * params.def (PARAM_TM_MAX_AGGREGATE_SIZE): New.
 328         * passes.c (init_optimization_passes): Place transactional memory
 329         passes.
 330         * print-tree.c (print_node): Dump tm-clone.
 331         * recog.c (peep2_attempt): Handle REG_TM.
 332         * reg-notes.def (TM): New.
 333         * rtlanal.c (alloc_reg_note): Handle REG_TM.
 334         * target.def (builtin_tm_load, builtin_tm_store): New.
 335         * targhooks.c (default_builtin_tm_load_store): New.
 336         * targhooks.h (default_builtin_tm_load_store): Declare.
 337         * timevar.def (TV_TRANS_MEM): New.
 338         * toplev.c (compile_file): Call finish_tm_clone_pairs.
 339         * tree-cfg.c (make_edges): Handle GIMPLE_TRANSACTION.
 340         (cleanup_dead_labels): Handle GIMPLE_TRANSACTION. Avoid unnecessary
 341         writes into the statements to update labels.
 342         (is_ctrl_altering_stmt): Add TM ending statements. Handle
 343         GIMPLE_TRANSACTION.
 344         (verify_gimple_transaction): New.
 345         (verify_gimple_stmt): Handle GIMPLE_TRANSACTION.
 346         (verify_gimple_in_seq_2): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
 347         (gimple_redirect_edge_and_branch): Handle TM_TRANSACTION.
 348         (dump_function_to_file): Display [tm-clone] if applicable.
 349         * tree-eh.c (struct_ptr_eq): Make inline and move to tree.h.
 350         (struct_ptr_hash): Same.
 351         (collect_finally_tree): Handle GIMPLE_EH_ELSE.
 352         (replace_goto_queue_1): Likewise.
 353         (get_eh_else): New.
 354         (honor_protect_cleanup_actions): Handle GIMPLE_EH_ELSE.
 355         (lower_try_finally_nofallthru): Likewise.
 356         (lower_try_finally_onedest): Likewise.
 357         (lower_try_finally_copy): Likewise.
 358         (lower_try_finally_switch): Likewise.
 359         (lower_try_finally): Likewise.
 360         (decide_copy_try_finally): Likewise.
 361         (lower_eh_constructs_2): Likewise.
 362         (refactor_eh_r): Likewise.
 363         * tree-flow.h (struct gimple_df): Add tm_restart field.
 364         Define tm_restart_node.
 365         * tree-inline.c (remap_gimple_stmt): Handle GIMPLE_TRANSACTION.
 366         (estimate_num_insns): Likewise.
 367         (init_inline_once): Init tm_cost.
 368         * tree-inline.h (struct eni_weights_d): Add tm_cost.
 369         * tree-pass.h (pass_diagnose_tm_blocks, pass_lower_tm, pass_tm_init,
 370         pass_tm_mark, pass_tm_memopt, pass_tm_edges, pass_ipa_tm): Declare.
 371         * tree-pretty-print.c (dump_generic_node): Handle TRANSACTION_EXPR.
 372         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
 373         BUILT_IN_TM_MEMSET, BUILT_IN_TM_MEMCPY, BUILT_IN_TM_MEMMOVE.
 374         Add support for TM vector loads.  Add support for TM logging builtins.
 375         (call_may_clobber_ref_p_1): Add support for vector stores.
 376         * tree-ssa-structalias.c (find_func_aliases): Add support for TM
 377         vector stores and loads. Handle BUILT_IN_TM_MEMSET,
 378         BUILT_IN_TM_MEMCPY, BUILT_IN_TM_MEMMOVE.
 379         * tree.c (strip_invariant_refs): Moved from gimple.c to here.
 380         (local_define_builtin): Handle ECF_TM_PURE.
 381         (build_common_builtin_nodes): Set __builtin_eh_pointer to ECF_TM_PURE.
 382         * tree.def (TRANSACTION_EXPR): New.
 383         * tree.h (strip_invariant_refs): Moved from gimple.h to here.
 384         (TRANSACTION_EXPR_BODY, TRANSACTION_EXPR_CHECK,
 385         TRANSACTION_EXPR_OUTER, TRANSACTION_EXPR_RELAXED,
 386         BUILTIN_TM_LOAD_STORE_P, BUILTIN_TM_LOAD_P, BUILTIN_TM_STORE_P,
 387         CASE_BUILT_IN_TM_LOAD, CASE_BUILT_IN_TM_STORE): New.
 388         (ECF_TM_PURE, ECF_TM_BUILTIN): New.
 389         (struct tree_function_decl): Add tm_clone_flag.
 390         (struct_ptr_eq, struct_ptr_hash): New.
 391         (apply_tm_attr): Declare.
 392         (is_tm_safe_or_pure): New.
 393         (build_tm_abort_call, is_tm_safe, is_tm_pure,
 394         is_tm_may_cancel_outer, is_tm_ending_fndecl, record_tm_replacement,
 395         tm_malloc_replacement): Declare.
 396         * varasm.c (tm_clone_hash): New.
 397         (record_tm_clone_pair, finish_tm_clone_pairs, get_tm_clone_pair,
 398         dump_tm_clone_to_vec, dump_tm_clone_pairs, tm_alias_pair_cmp): New.
 399         (struct tm_alias_pair): New.  Declare VEC types for object.
 400
 401 2011-11-07  Richard Henderson  <rth@redhat.com>
 402
 403         * optabs.h (OTI_sync_compare_and_swap, OTI_sync_lock_test_and_set,
 404         OTI_sync_old_add, OTI_sync_old_sub, OTI_sync_old_ior,
 405         OTI_sync_old_and, OTI_sync_old_xor, OTI_sync_old_nand,
 406         OTI_sync_new_add, OTI_sync_new_sub, OTI_sync_new_ior,
 407         OTI_sync_new_and, OTI_sync_new_xor, OTI_sync_new_nand): Move and
 408         rename from the direct_optab_index enum.
 409         (sync_compare_and_swap_optab, sync_lock_test_and_set_optab,
 410         sync_old_add_optab, sync_old_sub_optab, sync_old_ior_optab,
 411         sync_old_and_optab, sync_old_xor_optab, sync_old_nand_optab,
 412         sync_new_add_optab, sync_new_sub_optab, sync_new_ior_optab,
 413         sync_new_and_optab, sync_new_xor_optab, sync_new_nand_optab): Read
 414         from the optab_table, not the direct_optab_table.
 415         (init_sync_libfuncs): Declare.
 416         (can_compare_and_swap_p): Update parameters.
 417         * optabs.c (init_sync_libfuncs_1, init_sync_libfuncs): New.
 418         (can_compare_and_swap_p): Add allow_libcall parameter; if true,
 419         test for the legacy compare-and-swap libcall.
 420         (expand_atomic_exchange): Use the legacy test-and-set libcall.
 421         (expand_atomic_compare_and_swap): Use the legacy CAS libcall.
 422         (struct atomic_op_functions): Update for optab type changes.
 423         (maybe_emit_op): Likewise.
 424         (expand_atomic_fetch_op): Use the legacy fetch-op libcalls.
 425         * builtins.c (fold_builtin_atomic_always_lock_free): Update call
 426         to can_compare_and_swap_p.
 427         * omp-low.c (expand_omp_atomic_fetch_op): Likewise.
 428         (expand_omp_atomic_pipeline): Likewise.
 429         * genopinit.c (optabs): Make sync_old_*_optab, sync_new_*_optab,
 430         sync_compare_and_swap_optab, sync_lock_test_and_set_optab regular
 431         optabs.
 432         * doc/md.texi (sync_compare_and_swap): Update docs for libcalls.
 433
 434 2011-11-07  Jakub Jelinek  <jakub@redhat.com>
 435
 436         * config/i386/i386-bultin-types.def (V8SI_FTYPE_V4DF_V4DF): Add.
 437         * config/i386/i386.c (enum ix86_builtins): Add
 438         IX86_BUILTIN_VEC_PACK_SFIX256.
 439         (bdesc_args): Add __builtin_ia32_vec_pack_sfix256.
 440         (ix86_expand_args_builtin): Handle V8SI_FTYPE_V4DF_V4DF.
 441         (ix86_builtin_vectorized_function): Also vectorize lrint using
 442         256-bit vectors for -mavx.
 443
 444 2011-11-07  Anatoly Sokolov  <aesok@post.ru>
 445
 446         * config/cris/constraints.md: New file.
 447         * config/cris/cris.h (REG_CLASS_FROM_LETTER, CONSTRAINT_LEN,
 448         CRIS_CONST_OK_FOR_LETTER_P, CONST_OK_FOR_CONSTRAINT_P,
 449         CONST_DOUBLE_OK_FOR_LETTER_P, EXTRA_MEMORY_CONSTRAINT,
 450         EXTRA_CONSTRAINT, EXTRA_CONSTRAINT_Q, EXTRA_CONSTRAINT_R,
 451         EXTRA_CONSTRAINT_T, EXTRA_CONSTRAINT_S, EXTRA_CONSTRAINT_U): Remove.
 452         * config/cris/cris.c: Incule "tm-constrs.h".
 453         (cris_print_operand): Use satisfies_constraint_O.
 454         (cris_normal_notice_update_cc, cris_rtx_costs): Use
 455         satisfies_constraint_I.
 456         (cris_address_cost): Use satisfies_constraint_L.
 457         * config/cris/cris.md: Include "constraints.md".
 458         (*mov_side<mode>, *mov_sidesisf, *mov_side<mode>_mem,
 459         *mov_sidesisf_mem, *clear_side<mode>, *ext_sideqihi,
 460         *ext_side<mode>si, *op_side<mode>, *op_swap_side<mode>,
 461         *extopqihi_side, *extop<mode>si_side, *extopqihi_swap_side,
 462         *extop<mode>si_swap_side): Use satisfies_constraint_N and
 463         satisfies_constraint_J.
 464         (moversideqi movemsideqi mover2side peephole2): Use
 465         satisfies_constraint_N and satisfies_constraint_J.
 466         (andu peephole2): Use satisfies_constraint_I and
 467         satisfies_constraint_O.
 468
 469 2011-11-07  Uros Bizjak  <ubizjak@gmail.com>
 470
 471         * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
 472         BUILT_IN_IRINT, BUILT_IN_IRINTF, BUILT_IN_LLRINT and BUILT_IN_LLRINTF.
 473
 474 2011-11-07  Andrew MacLeod  <amacleod@redhat.com>
 475
 476         * optabs.c (get_atomic_op_for_code): Fill in optab table at runtime so
 477         SWITCHABLE_TARGET can change the values during compilation.
 478         (expand_atomic_fetch_op): Handle parameter change ripples for
 479         get_atomic_op_for_code call.
 480
 481 2011-11-07  Andrew MacLeod  <amacleod@redhat.com>
 482
 483         * doc/extend.texi: Docuemnt behaviour change for __atomic_exchange and
 484         __atomic_store.
 485         * optabs.c (expand_atomic_exchange): Expand to __sync_lock_test_and_set
 486         only when originated from that builtin.
 487         (expand_atomic_store): Expand to __sync_lock_release when originated
 488         from that builtin.
 489         * builtins.c (expand_builtin_sync_lock_test_and_set): Add flag that
 490         expand_atomic_exchange call originated from here.
 491         (expand_builtin_sync_lock_release): Add flag that expand_atomic_store
 492         call originated from here.
 493         (expand_builtin_atomic_exchange): Add origination flag.
 494         (expand_builtin_atomic_store): Add origination flag.
 495         * expr.h (expand_atomic_exchange, expand_atomic_store): Add boolean
 496         parameters to indicate implementation fall back options.
 497
 498 2011-11-07  Georg-Johann Lay  <avr@gjlay.de>
 499
 500         * config/avr/avr.c (output_reload_in_const): Can handle CONSTANT_P
 501         now, not only CONST_INT and CONST_DOUBLE.
 502         (output_movqi): Use output_reload_in_const.
 503         (output_reload_inhi): Ditto.
 504         (output_reload_insisf): Move assertion to output_reload_in_const.
 505         (avr_out_reload_inpsi): Ditto.
 506
 507 2011-11-07  Nathan Sidwell  <nathan@acm.org>
 508
 509         * gcov.c (object_summary): Replace with ...
 510         (object_runs): ... this.
 511         (process_file): Remove functions with no data.
 512         (generate_results): Ignore files with no lines.
 513         (release_function): New helper, broken out of ...
 514         (release_structures): ... here.  Use it.
 515         (read_count_file): Adjust for new data file format.
 516         (output_lines): Use object_runs.
 517         * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Obsolete.
 518         (struct gcov_ctr_info): Move definition.
 519         (struct gcov_fn_info): Add key field, use gcov_ctr_info for
 520         trailing array.
 521         (struct gcov_info): Add merge function array, remove mask and
 522         counts.  Trailing array is array of pointers to function info.
 523         * coverage.c (struct function_list): Replace counter numbers with
 524         counter arrays.  Add fndecl field.  GTYify.
 525         (counts_entry): Remove chain workspace.
 526         (functions_head): GTYify.
 527         (prg_n_ctrs): Remove.
 528         (fn_v_ctrs): New.
 529         (tree_ctr_tables): Remove.
 530         (read_counts_file): Cope with blank entries and expect program
 531         summaries before functions.  Don't warn on missing entries.
 532         (coverage_counter_alloc): Allocate individual function arrays.
 533         (tree_coverage_counter_ref, tree_coverage_counter_addr): Adjust
 534         for individual function arrays.
 535         (coverage_end_function): GTYify function list object. Finalize
 536         function's counter arrays.
 537         (build_var): New.  Create a counter-related variable with
 538         appropriate linkage.
 539         (build_fn_info_type): Adjust for new runtime structure.
 540         (build_fn_info_value): Rename to ...
 541         (build_fn_info): ... here.  Build new format data.
 542         (build_ctr_info_type, build_ctr_info_value): Remove.
 543         (build_info_type): New. Build new format data structure.
 544         (build_info): Adjust for new format data.
 545         (create_coverage): Likewise.
 546         * gcov-dump.c (tag_function): Recognize placeholders.
 547
 548 2011-11-07  Georg-Johann Lay  <avr@gjlay.de>
 549
 550         * config/avr/constraints.md (Cm2): New constraint for int -2.
 551         * config/avr/avr.md (addqi3): Use it. New alternatives for +/-2.
 552         (*negqihi2): New insn.
 553
 554 2011-11-07  H.J. Lu  <hongjiu.lu@intel.com>
 555
 556         * dwarf2cfi.c (dwarf2out_frame_debug_expr): Check
 557         HARD_FRAME_POINTER_REGNUM instead of hard_frame_pointer_rtx
 558         in Rule 18.
 559
 560 2011-11-07  Jakub Jelinek  <jakub@redhat.com>
 561
 562         * config/i386/i386.c (ix86_expand_builtin): If gather mask
 563         argument is known to have all high bits set, pass pc_rtx as
 564         second argument to the expander instead of op0.
 565         * config/i386/sse.md (*avx2_gathersi<mode>_2,
 566         *avx2_gatherdi<mode>_2): New patterns.
 567         * config/i386/avx2intrin.h (_mm256_i32gather_pd,
 568         _mm256_i64gather_pd, _mm256_i32gather_ps): Set mask using
 569         _mm256_cmp_pd with zero vector arguments and _CMP_EQ_OQ instead of
 570         _mm256_set1_pd.
 571
 572         PR tree-optimization/50789
 573         * tree-vect-stmts.c (process_use): Add force argument, avoid
 574         exist_non_indexing_operands_for_use_p check if true.
 575         (vect_mark_stmts_to_be_vectorized): Adjust callers.  Handle
 576         STMT_VINFO_GATHER_P.
 577         (gen_perm_mask): New function.
 578         (perm_mask_for_reverse): Use it.
 579         (reverse_vec_element): Rename to...
 580         (permute_vec_elements): ... this.  Add Y and MASK_VEC arguments,
 581         generalize for any permutations.
 582         (vectorizable_load): Adjust caller.  Handle STMT_VINFO_GATHER_P.
 583         * target.def (TARGET_VECTORIZE_BUILTIN_GATHER): New hook.
 584         * doc/tm.texi.in (TARGET_VECTORIZE_BUILTIN_GATHER): Document it.
 585         * doc/tm.texi: Regenerate.
 586         * tree-data-ref.c (initialize_data_dependence_relation,
 587         compute_self_dependence): No longer static.
 588         * tree-data-ref.h (initialize_data_dependence_relation,
 589         compute_self_dependence): New prototypes.
 590         * tree-vect-data-refs.c (vect_check_gather): New function.
 591         (vect_analyze_data_refs): Detect possible gather load data
 592         refs.
 593         * tree-vectorizer.h (struct _stmt_vec_info): Add gather_p field.
 594         (STMT_VINFO_GATHER_P): Define.
 595         (vect_check_gather): New prototype.
 596         * config/i386/i386-builtin-types.def: Add types for alternate
 597         gather builtins.
 598         * config/i386/sse.md (AVXMODE48P_DI): Remove.
 599         (VEC_GATHER_MODE): Rename mode_attr to...
 600         (VEC_GATHER_IDXSI): ... this.
 601         (VEC_GATHER_IDXDI, VEC_GATHER_SRCDI): New mode_attrs.
 602         (avx2_gathersi<mode>, *avx2_gathersi<mode>): Use <VEC_GATHER_IDXSI>
 603         instead of <VEC_GATHER_MODE>.
 604         (avx2_gatherdi<mode>): Use <VEC_GATHER_IDXDI> instead of
 605         <<AVXMODE48P_DI> and <VEC_GATHER_SRCDI> instead of VEC_GATHER_MODE
 606         on src and mask operands.
 607         (*avx2_gatherdi<mode>): Likewise.  Use VEC_GATHER_MODE iterator
 608         instead of AVXMODE48P_DI.
 609         (avx2_gatherdi<mode>256, *avx2_gatherdi<mode>256): Removed.
 610         * config/i386/i386.c (enum ix86_builtins): Add
 611         IX86_BUILTIN_GATHERALTSIV4DF, IX86_BUILTIN_GATHERALTDIV8SF,
 612         IX86_BUILTIN_GATHERALTSIV4DI and IX86_BUILTIN_GATHERALTDIV8SI.
 613         (ix86_init_mmx_sse_builtins): Create those builtins.
 614         (ix86_expand_builtin): Handle those builtins and adjust expansions
 615         of other gather builtins.
 616         (ix86_vectorize_builtin_gather): New function.
 617         (TARGET_VECTORIZE_BUILTIN_GATHER): Define.
 618
 619 2011-11-07  Uros Bizjak  <ubizjak@gmail.com>
 620
 621         * config/i386/f16cintrin: Remove extra _X86INTRIN_H_INCLUDED check.
 622
 623 2011-11-07  Tristan Gingold  <gingold@adacore.com>
 624
 625         * config.gcc (*-*-*vms*): Set c_target_objs, cxx_target_objs.
 626         * config/vms/t-vms: Add vms-c.o rule.
 627         * config/vms/vms-c.c: New file.
 628         * config/vms/vms-protos.h (vms_c_register_pragma): New prototype.
 629         * config/vms/vms.h (REGISTER_TARGET_PRAGMAS): Define.
 630
 631 2011-11-07  Tristan Gingold  <gingold@adacore.com>
 632
 633         * config/alpha/vms.h (TARGET_OBJECT_SUFFIX,
 634         TARGET_EXECUTABLE_SUFFIX, TARGET_OS_CPP_BUILTINS,
 635         TARGET_ABI_OPEN_VMS, LONG_TYPE_SIZE, ADA_LONG_TYPE_SIZE,
 636         POINTER_SIZE, POINTERS_EXTEND_UNSIGNED): Move to config/vms/vms.h
 637         (SUBTARGET_OS_CPP_BUILTINS): Define.
 638         (TARGET_DEFAULT): Tune according to POINTER_SIZE.
 639         (MASK_RETURN_ADDR): Define in 64 bit mode.
 640         * config/ia64/vms.h: Likewise.
 641         * config/vms/vms.h: New file.
 642         * config/vms/vms64.h: New file.
 643         * config/alpha/vms64.h: Removed.
 644         * config/ia64/vms64.h: Removed.
 645         * config.gcc (alpha64-dec-*vms*, alpha*-dec-*vms*, ia64-hp-*vms*):
 646         Adjust for above change.
 647
 648 2011-11-07  Enkovich Ilya  <ilya.enkovich@intel.com>
 649
 650         PR target/50962
 651         * config/i386/i386-protos.h (ix86_use_lea_for_mov): New.
 652         * config/i386/i386.c (ix86_use_lea_for_mov): Likewise.
 653         * config/i386/i386.md (movsi_internal): Emit lea if profitable.
 654         (movdi_internal_rex64): Likewise.
 655
 656 2011-11-07  Sergey Ostanevich  <sergos.gnu@gmail.com>
 657
 658         PR rtl-optimization/47698
 659         * ifcvt.c (noce_operand_ok): Return false for mems with side effects.
 660
 661 2011-11-07  Tristan Gingold  <gingold@adacore.com>
 662
 663         * common/config/alpha/alpha-common.c (alpha_option_init_struct):
 664         New function.
 665         (TARGET_OPTION_INIT_STRUCT): Define.
 666         * config/alpha/alpha.c (TARGET_MIN_ANCHOR_OFFSET)
 667         (TARGET_MAX_ANCHOR_OFFSET)
 668         (TARGET_USE_BLOCKS_FOR_CONSTANT_P): Redefine.
 669
 670 2011-11-06  Quentin Neill  <quentin.neill@amd.com>
 671
 672         Fix r180999, update ChangeLog
 673         * config.gcc: Add f16cintrin.h.
 674         * config/i386/f16cintrin.h: Add missing endif.
 675
 676 2011-11-06  Sebastian Huber  <sebastian.huber@embedded-brains.de>
 677
 678         * config.gcc (arm*-*-rtemseabi*): New target.
 679         * config/arm/rtems-eabi.h: New.
 680         * config/arm/t-rtems-eabi: New.
 681
 682 2011-11-06  David S. Miller  <davem@davemloft.net>
 683
 684         * config/sparc/sparc.md (UNSPEC_SHORT_LOAD): Delete.
 685         (zero_extend_v8qi_vis, zero_extend_v4hi_vis,
 686         *zero_extend_v8qi_<P:mode>_insn,
 687         *zero_extend_v4hi_<P:mode>_insn): Express using vec_merge
 688         and vec_duplicate instead of using an UNSPEC.
 689
 690 2011-11-07  Alan Modra  <amodra@gmail.com>
 691
 692         PR target/30282
 693         * config/rs6000/rs6000.c (rs6000_emit_stack_reset): Always emit
 694         blockage for ABI_V4.
 695
 696 2011-11-06  Dave Korn  <dave.korn.cygwin@gmail.com>
 697
 698         * config/i386/cygwin.h (LIBGCJ_SONAME): Updated to match recent
 699         libgcj version bump.
 700         * config/i386/mingw32.h (LIBGCJ_SONAME): Likewise.
 701
 702 2011-11-06  Eric Botcazou  <ebotcazou@adacore.com>
 703
 704         * gcse.c: Adjust outdated comments throughout.
 705         (struct mem_conflict_info): New structure.
 706         (mems_conflict_for_gcse_p): Use it to communicate with caller.
 707         (load_killed_in_block_p): Pass it to note_stores.
 708         (hash_expr): Remove superfluous line break.
 709         (hash_scan_set): Rename PAT parameter into SET.
 710         (hash_scan_insn): Reorder cases.
 711         (canon_list_insert): Fix long line.
 712         (edge_list): Delete.
 713         (prune_expressions): Rename E local variable into EXPR.
 714         (compute_pre_data): Return struct edge_list * object.
 715         (pre_expr_reaches_here_p_work): Fix formatting.
 716         (process_insert_insn): Move around comment.
 717         (pre_edge_insert): Fix long line.
 718         (pre_insert_copies): Likewise.
 719         (gcse_emit_move_after): Swap SRC and DEST parameters.
 720         (pre_delete): Adjust call to gcse_emit_move_after.
 721         (pre_gcse): Take struct edge_list * parameter.  Fix long line.
 722         (one_pre_gcse_pass): Use flag_gcse_lm condition for all routines.
 723         Use a local list of edges.
 724         (hoist_code): Fix long line.  Adjust call to gcse_emit_move_after.
 725         (pre_ldst_expr_hash): Fix long line.
 726         (free_ldst_mems): Rename into...
 727         (free_ld_motion_mems): ...this.
 728         (first_ls_expr): Delete.
 729         (next_ls_expr): Likewise.
 730         (print_ldst_list): Do not use above two functions.
 731         (simple_mem): Adjust interface.
 732         (compute_ld_motion_mems): Fix formatting.
 733         (update_ld_motion_stores): Reuse local variable.
 734
 735 2011-11-06  Joseph Myers  <joseph@codesourcery.com>
 736
 737         * c-decl.c (shadow_tag_warned, grokdeclarator): Handle _Alignas
 738         specifiers.
 739         (build_null_declspecs): Initialize align_log and alignas_p fields.
 740         (declspecs_add_alignas): New.
 741         * c-parser.c (c_token_starts_declspecs): Handle RID_ALIGNAS.
 742         (c_parser_declspecs): Handle _Alignas specifiers.
 743         (c_parser_alignas_specifier): New.
 744         (c_parser_alignof_expression): Diagnose alignof use for non-C1X.
 745         Diagnose _Alignof (expression).
 746         * c-tree.h (struct c_declspecs): Add align_log and alignas_p fields.
 747         (declspecs_add_alignas): Declare.
 748         * ginclude/stddef.h (max_align_t): Define for C1X and C++11.
 749         * ginclude/stdalign.h: New.
 750         * Makefile.in (USER_H): Add stdalign.h.
 751
 752 2011-11-06  Joern Rennecke  <joern.rennecke@embecosm.com>
 753
 754         * regset.h (fixed_reg_set_regset): Declare.
 755         * dse.c: Include regset.h .
 756         (struct insn_info): Add member fixed_regs_live.
 757         (note_add_store_info): New typedef.
 758         (note_add_store): New function.
 759         (emit_inc_dec_insn_before): Expect arg to be of type insn_info_t.
 760         Use gen_add3_insn / gen_move_insn.
 761         Check new insn for unwanted clobbers before emitting it.
 762         (check_for_inc_dec): Rename to...
 763         (check_for_inc_dec_1:) ... this.  Return bool.  Take insn_info
 764         parameter.  Changed all callers in file.
 765         (check_for_inc_dec, copy_fixed_regs): New functions.
 766         (scan_insn): Set fixed_regs_live field of insn_info.
 767         * rtl.h (check_for_inc_dec): Update prototype.
 768         * postreload.c (reload_cse_simplify): Take new signature of
 769         check_ind_dec into account.
 770         * reginfo.c (fixed_reg_set_regset): New variable.
 771         (init_reg_sets_1): Initialize it.
 772
 773 2011-11-06  Jakub Jelinek  <jakub@redhat.com>
 774
 775         * tree-cfg.c (gimple_can_merge_blocks_p): For -O0 don't remove
 776         any user labels.
 777
 778 2011-11-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
 779
 780         * config/pa/pa.c (pa_hpux_init_libfuncs): Rename to pa_init_libfuncs.
 781         Remove dependence of declaration and target define on definition of
 782         HPUX_LONG_DOUBLE_LIBRARY.  Update implementation.
 783
 784 2011-11-06  Andrew Macleod  <amacleod@redhat.com>
 785             Richard Henderson  <rth@redhat.com>
 786             Aldy Hernandez  <aldyh@redhat.com>
 787
 788         Merged from cxx-mem-model.
 789
 790         * cppbuiltin.c (define__GNUC__): Define __ATOMIC memory models
 791         * coretypes.h (enum memmodel): New.  enumerated memory model type.
 792         * Makefile.in (cppbuiltin.o) Add missing dependency on $(TREE_H)
 793         * genopinit,c (optabs): Add atomic direct optab handlers.
 794         * sync-builtins.def (BUILT_IN_ATOMIC_*): New builtins.
 795         * builtin-types.def (BT_CONST_VOLATILE_PTR,
 796         BT_FN_I{1,2,4,8,16}_CONST_VPTR_INT, BT_FN_VOID_VPTR_INT,
 797         BT_FN_BOOL_VPTR_INT, BT_FN_BOOL_SIZE_CONST_VPTR,
 798         BT_FN_I{1,2,4,8,16}_VPTR_I{1,2,4,8,16}_INT,
 799         BT_FN_VOID_VPTR_I{1,2,4,8,16}_INT, BT_FN_VOID_SIZE_VPTR_PTR_INT,
 800         BT_FN_VOID_SIZE_CONST_VPTR_PTR_INT, BT_FN_VOID_SIZE_VPTR_PTR_PTR_INT,
 801         BT_FN_BOOL_VPTR_PTR_I{1,2,4,8,16}_BOOL_INT_INT): New builtin types.
 802         * expr.h (expand_atomic_*): Add prototypes.
 803         (expand_{bool,val}_compare_and_swap): Remove prototypes.
 804         * c-typeck.c (build_function_call_vec): Don't reprocess __atomic
 805         parameters.
 806         * common.opt (Winvalid-memory-model): New warning flag.
 807         (finline-atomics): New. Flag to disable atomic inlining.
 808         * params.h (ALLOW_LOAD_DATA_RACES): New.
 809         (ALLOW_PACKED_LOAD_DATA_RACES): New.
 810         (ALLOW_PACKED_STORE_DATA_RACES): New.
 811         * params.def (PARAM_ALLOW_LOAD_DATA_RACES): New.
 812         (PARAM_ALLOW_PACKED_LOAD_DATA_RACES): New.
 813         (PARAM_ALLOW_PACKED_STORE_DATA_RACES): New.
 814         * builtins.c (is_builtin_name): Handle __atomic.
 815         (get_memmodel): New.  Extract memory model.
 816         (expand_expr_force_mode): New. Factor out common code for ensuring an
 817         integer argument is in the proper mode.
 818         (expand_builtin_sync_operation): Remove ignore param.  Always call
 819         expand_atomic_fetch_op instead of the old expanders.
 820         (expand_builtin_compare_and_swap,
 821         expand_builtin_sync_lock_test_and_set): Use expand_expr_force_mode,
 822         call atomic expanders instead of sync expanders.
 823         (expand_builtin_sync_lock_release): Call atomic_store expander.
 824         (expand_builtin_atomic_compare_exchange, expand_builtin_atomic_load,
 825         expand_builtin_atomic_store, expand_builtin_atomic_fetch_op): New.
 826         (expand_builtin_atomic_exchange): New.
 827         (fold_builtin_atomic_always_lock_free,
 828         expand_builtin_atomic_always_lock_free,
 829         fold_builtin_atomic_is_lock_free, expand_builtin_atomic_is_lock_free):
 830         New.
 831         (expand_builtin_mem_thread_fence, expand_builtin_atomic_thread_fence,
 832         expand_builtin_atomic_signal_fence): New.
 833         (expand_builtin_mem_signal_fence): New.
 834         (expand_builtin): Add cases for BUILT_IN_ATOMIC_*.
 835         (fold_builtin_2): Add cases for BUILT_IN_ATOMIC_{IS,ALWAYS}_LOCK_FREE.
 836         * optabs.h (DOI_atomic_*): Define new atomics.
 837         (atomic_*_optab): Define.
 838         (can_compare_and_swap_p, expand_atomic_compare_and_swap): New
 839         prototypes.
 840         * optabs.c (expand_sync_operation, expand_sync_fetch_operation): Remove.
 841         (expand_sync_lock_test_and_set): Remove.
 842         (expand_atomic_load, expand_atomic_store): New.
 843         (expand_atomic_exchange): New.
 844         (expand_atomic_compare_and_swap): New.  Implements
 845         atomic_compare_exchange via compare and swap.
 846         (struct atomic_op_functions): Opcode table struct for fetch ops.
 847         (get_atomic_op_for_code): New.  Return an opcode table entry.
 848         (maybe_emit_op): New.  Try to emit a fetch op.
 849         (expand_atomic_fetch_op): New.
 850         (expand_val_compare_and_swap_1): Remove.
 851         (expand_val_compare_and_swap, expand_bool_compare_and_swap): Remove.
 852         (expand_atomic_compare_and_swap): Rename from
 853         expand_atomic_compare_exchange.  Rewrite to return both success and
 854         oldval return values; expand via both atomic and sync optabs.
 855         (can_compare_and_swap_p): New.
 856         (expand_compare_and_swap_loop): Use expand_atomic_compare_and_swap.
 857         (maybe_gen_insn): Handle 7 and 8 operands.
 858         * omp-low.c (expand_omp_atomic_fetch_op): Don't test individual
 859         fetch_op optabs, only test can_compare_and_swap_p.  Use __atomic
 860         builtins instead of __sync builtins.
 861         (expand_omp_atomic_pipeline): Use can_compare_and_swap_p.
 862         * doc/extend.texi: Document __atomic built-in functions.
 863         * doc/invoke.texi: Document data race parameters.
 864         * doc/md.texi: Document atomic patterns.
 865         * config/i386/i386.md (UNSPEC_MOVA): New.
 866         (UNSPECV_CMPXCHG): Split into ...
 867         (UNSPECV_CMPXCHG_1, UNSPECV_CMPXCHG_2,
 868         UNSPECV_CMPXCHG_3, UNSPECV_CMPXCHG_4): New.
 869         * config/i386/sync.md (ATOMIC): New mode iterator.
 870         (atomic_load<ATOMIC>, atomic_store<ATOMIC>): New.
 871         (atomic_loaddi_fpu, atomic_storedi_fpu, movdi_via_fpu): New.
 872         (mem_thread_fence): Rename from memory_barrier.
 873         Handle the added memory model parameter.
 874         (mfence_nosse): Rename from memory_barrier_nosse.
 875         (sync_compare_and_swap<CASMODE>): Split into ...
 876         (atomic_compare_and_swap<SWI124>): this and ...
 877         (atomic_compare_and_swap<CASMODE>): this.  Handle the new parameters.
 878         (atomic_compare_and_swap_single<SWI>): Rename from
 879         sync_compare_and_swap<SWI>; rewrite to use split unspecs.
 880         (atomic_compare_and_swap_double<DCASMODE>): Rename from
 881         sync_double_compare_and_swap<DCASMODE>; rewrite to use split unspecs.
 882         (*atomic_compare_and_swap_doubledi_pic): Rename from
 883         sync_double_compare_and_swapdi_pic; rewrite to use split unspecs.
 884         (atomic_fetch_add<SWI>): Rename from sync_old_add<SWI>; add memory
 885         model parameter.
 886         (*atomic_fetch_add_cmp<SWI>): Similarly.
 887         (atomic_add<SWI>, atomic<any_logic><SWI>): Similarly.
 888         (atomic_sub<SWI>): Similarly.  Use x86_maybe_negate_const_int.
 889         (sync_lock_test_and_set<SWI>): Merge with ...
 890         (atomic_exchange<SWI>): ... this.
 891
 892 2011-11-6  Richard Guenther  <rguenther@suse.de>
 893
 894         * ipa-prop.c (ipa_modify_call_arguments): Re-compute
 895         inlinable flag.
 896
 897 2011-11-06  Ira Rosen  <ira.rosen@linaro.org>
 898
 899         * tree-vectorizer.h (vectorizable_condition): Add argument.
 900         * tree-vect-loop.c (vectorizable_reduction): Fail for condition
 901         in SLP.  Update calls to vectorizable_condition.
 902         * tree-vect-stmts.c (vect_is_simple_cond): Add basic block info to
 903         the arguments.  Pass it to vect_is_simple_use_1.
 904         (vectorizable_condition): Add slp_node to the arguments.  Support
 905         vectorization of basic blocks.  Fail for reduction in SLP.  Update
 906         calls to vect_is_simple_cond and vect_is_simple_use.  Support SLP:
 907         call vect_get_slp_defs to get vector operands.
 908         (vect_analyze_stmt): Update calls to vectorizable_condition.
 909         (vect_transform_stmt): Likewise.
 910         * tree-vect-slp.c (vect_create_new_slp_node): Handle COND_EXPR.
 911         (vect_get_and_check_slp_defs): Handle COND_EXPR.  Allow pattern
 912         def stmts.
 913         (vect_build_slp_tree): Handle COND_EXPR.
 914         (vect_analyze_slp_instance): Push pattern statements to root node.
 915         (vect_get_constant_vectors): Fix comments.  Handle COND_EXPR.
 916
 917 2011-11-05  David S. Miller  <davem@davemloft.net>
 918
 919         * config/sparc/sparc.md (UNSPEC_SHORT_LOAD): New unspec.
 920         (zero-extend_v8qi_vis, zero_extend_v4hi_vis): New expanders.
 921         (*zero_extend_v8qi_<P:mode>_insn,
 922         *zero_extend_v4hi_<P:mode>_insn): New insns.
 923         * config/sparc/sparc.c (vector_init_move_words,
 924         vector_init_prepare_elts, sparc_expand_vector_init_vis2,
 925         sparc_expand_vector_init_vis1): New functions.
 926         (vector_init_bshuffle): Rewrite to handle more cases and make use
 927         of locs[] array prepared by vector_init_prepare_elts.
 928         (vector_init_fpmerge, vector_init_faligndata): Delete.
 929         (sparc_expand_vector_init): Rewrite using new infrastructure.
 930
 931 2011-11-05  Joern Rennecke  <joern.rennecke@embecosm.com>
 932
 933         * config.gcc (epiphany-*-*): New architecture.
 934         (epiphany-*-elf): New configuration.
 935         * config/epiphany, common/config/epiphany : New directories.
 936         * doc/extend.texi (disinterrupt attribute): Add Epiphany.
 937         (interrupt attribute): Add Epiphany.
 938         (long_call, short_call attribute): Add Epiphany.
 939         * doc/invoke.texi (Options): Add Epiphany options.
 940         * doc/md.texi (Machine Constraints): Add Epiphany constraints.
 941         * doc/install.texi (Options specification):
 942         Add --with-stack-offset=@var{num} description.
 943         (host/target specific issues): Add epiphany-*-elf.
 944         * doc/contrib.texi (Contributors): Mention Epiphany port.
 945
 946 2011-11-05  Jakub Jelinek  <jakub@redhat.com>
 947
 948         PR tree-optimization/50693
 949         * tree-cfg.c (gimple_can_merge_blocks_p): Allow merging with
 950         non-forced user labels.
 951         (gimple_merge_blocks): Turn non-forced user labels into
 952         debug bind stmt with the label as first operand and reset value.
 953         (gimple_duplicate_bb): Don't duplicate label debug stmts.
 954         * dwarf2out.c (gen_label_die): Handle NOTE_INSN_DELETED_DEBUG_LABEL.
 955         * final.c (final_scan_insn): Likewise.
 956         (rest_of_clean_state): Don't dump NOTE_INSN_DELETED_DEBUG_LABEL.
 957         * var-tracking.c (debug_label_num): New variable.
 958         (delete_debug_insns): Don't delete DEBUG_INSNs for LABEL_DECLs,
 959         instead turn them into NOTE_INSN_DELETED_DEBUG_LABEL notes.
 960         * cfglayout.c (skip_insns_after_block, duplicate_insn_chain): Handle
 961         NOTE_INSN_DELETED_DEBUG_LABEL.
 962         (duplicate_insn_chain): Don't duplicate LABEL_DECL DEBUG_INSNs.
 963         * insn-notes.def (DELETED_DEBUG_LABEL): New note kind.
 964         * print-rtl.c (print_rtx): Handle NOTE_INSN_DELETED_DEBUG_LABEL.
 965         * gengtype.c (adjust_field_rtx_def): Likewise.
 966         * config/i386/i386.c (ix86_output_function_epilogue): For MachO
 967         clear CODE_LABEL_NUMBER of NOTE_INSN_DELETED_DEBUG_LABEL
 968         if their are at the end of function and nop hasn't been emitted.
 969         * config/rs6000/rs6000.c (rs6000_output_function_epilogue): Likewise.
 970
 971 2011-11-05  Georg-Johann Lay  <avr@gjlay.de>
 972
 973         PR rtl-optimization/50448
 974         * cprop.c (try_replace_reg): Also try to replace uses of FROM that
 975         appear in SET_DEST.
 976
 977 2011-11-05  Peter Dufault  <dufault@hda.com>,
 978             Sebastian Huber  <sebastian.huber@embedded-brains.de>
 979
 980         * config/rs6000/rtems.h (SUBSUBTARGET_OVERRIDE_OPTIONS):
 981         Extend rs6000_spe handling.
 982
 983 2011-11-05  Ralf Corsépius  <ralf.corsepius@rtems.org>
 984
 985         * config/rs6000/t-rtems: Add -mcpu=8540/-mfloat-gprs=double multilib.
 986         Remove -mcpu=601 multilib.
 987         Remove -Dmpc8260 multilib.
 988         * config/rs6000/rtems.h: Allow --float-gprs=... to override grps
 989         on E500 targets.
 990
 991 2011-11-05  Quentin Neill  <quentin.neill@amd.com>
 992
 993         Piledriver f16cintrin.h fix.
 994         * config/i386/f16cintrin.h: Contents moved from immintrin.h.
 995         * config/i386/immintrin.h: Include f16cintrin.h.
 996         * config.gcc (i[34567]86-*-*, x86_64-*-*): Add f16cintrin.h.
 997
 998
 999 2011-11-04  Eric Botcazou  <ebotcazou@adacore.com>
1000
1001         PR c++/50608
1002         * c-parser.c (c_parser_postfix_expression) <RID_OFFSETOF>: Adjust call
1003         to fold_offsetof.
1004         * c-typeck.c (build_unary_op) <ADDR_EXPR>: Call fold_offsetof_1.
1005
1006 2011-11-04  Alan Modra  <amodra@gmail.com>
1007
1008         * reload1.c (gen_reload): Don't use REGNO on SUBREGs.
1009         * print-rtl.c (print_rtx): Don't segfault on negative regno.
1010
1011 2011-11-04  David S. Miller  <davem@davemloft.net>
1012
1013         PR target/49965
1014         * config/sparc/sparc.c (sparc_expand_conditional_move): Handle the
1015         fact that sparc_emit_float_lib_cmp modifies the comparison in
1016         operands[1].
1017
1018 2011-11-04  Ralf Corsépius  <ralf.corsepius@rtems.org>
1019
1020         * config/lm32/t-rtems: New.
1021         * config.gcc (lm32-*-rtems*): Add t-rtems.
1022
1023 2011-11-04  Eric Botcazou  <ebotcazou@adacore.com>
1024
1025         PR target/50979
1026         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=v8.
1027
1028 2011-11-04  Jiangning Liu  <jiangning.liu@arm.com>
1029
1030         PR rtl-optimization/38644
1031         * config/arm/arm.c (thumb1_expand_epilogue): Add memory barrier
1032         for epilogue having stack adjustment.
1033
1034 2011-11-04  Georg-Johann Lay  <avr@gjlay.de>
1035
1036         PR target/50931
1037         * config/avr/avr-modes.def: New file defining PSImode.
1038         * config/avr/avr-c.c (__INT24_MAX__, __INT24_MIN__,
1039         __UINT24_MAX__): New built-in defines.
1040         * config/avr/avr.md (adjust_len): Add tstpsi, mov24,  reload_in24,
1041         ashlpsi, ashrpsi, lshrpsi.
1042         (QISI, QIDI, HISI, HIDI, MPUSH, rotx, rotsmode): Add PSI.
1043         (MOVMODE): New mode iterator.
1044         (movpsi): New expander.
1045         (movqi, movhi, movsi, movsf, movpsi): Write as one using MOVMODE.
1046         (*reload_inpsi, *movpsi): New insns.
1047         (*reload_inpsi): New RTL peephole.
1048         (addpsi3, *addpsi3_zero_extend.qi, *addpsi3_zero_extend.hi,
1049         *addpsi3_sign_extend.hi): New insns.
1050         (subpsi3, *subpsi3_zero_extend.qi, *subpsi3_zero_extend.hi,
1051         *subpsi3_sign_extend.hi): New insns.
1052         (divmodpsi4, udivmodpsi4): New define insn-and-split.
1053         (*divmodpsi4_call, *udivmodpsi4_call): New insns.
1054         (andpsi3, iorpsi3, xorpsi3): New insns.
1055         (*rotlpsi2.1, *rotlpsi2.23): New insns.
1056         (*rotw<mode>): Insn condition only allow even-sized modes.
1057         (*rotb<mode>): Insn condition allows odd-sized modes.
1058         (ashlpsi3, ashrpsi3, lshrpsi3, *addpsi3.lt0): New insns.
1059         (negpsi2, one_cmplpsi2): New insns.
1060         (extendqipsi2, extendhipsi2, extendpsisi2): New insns.
1061         (zero_extendqipsi2, zero_extendhipsi2, zero_extendpsisi2): New
1062         insn-and-splits.
1063         (*cmppsi, *negated_tstpsi, *reversed_tstpsi): New insns.
1064         (cbranchpsi4): New expander.
1065         * config/avr/constraints.md (Ca3, Co3, Cx3): New constraints.
1066         * config/avr/avr-protos.h (avr_out_tstpsi, avr_out_movpsi,
1067         avr_out_ashlpsi3, avr_out_ashrpsi3, avr_out_lshrpsi3,
1068         avr_out_reload_inpsi): New prototypes.
1069
1070         * config/avr/avr.c (TARGET_SCALAR_MODE_SUPPORTED_P): Define to...
1071         (avr_scalar_mode_supported_p): ...this new static function.
1072         (avr_asm_len): Always return "".
1073         (avr_out_load_psi, avr_out_store_psi): New static functions.
1074         (avr_out_movpsi, avr_out_reload_inpsi): New functions.
1075         (avr_out_tstpsi): New function.
1076         (avr_out_ashlpsi3, avr_out_ashrpsi3, avr_out_lshrpsi3): New functions.
1077         (avr_out_plus_1, output_reload_in_const): Handle 3-byte types.
1078         (avr_simplify_comparison_p): Ditto.
1079         (adjust_insn_length): Handle ADJUST_LEN_RELOAD_IN24,
1080         ADJUST_LEN_MOV24, ADJUST_LEN_TSTPSI, ADJUST_LEN_ASHLPSI,
1081         ADJUST_LEN_ASHRPSI, ADJUST_LEN_LSHRPSI.
1082         (avr_rtx_costs_1): Report PSI costs.
1083         (avr_libcall_value): Handle odd-sized parameters.
1084         (avr_init_builtin_int24): New static function to define built-in
1085         24-bit types __int24 and __uint24.
1086         (avr_init_builtins): Use it.
1087
1088 2011-11-04  Thomas Doerfler <thomas.doerfler@embedded-brains.de>
1089
1090         PR target/50989
1091         * config/arm/rtems-elf.h, config/arm/t-rtems: Add optional
1092         support for VFP floating point model.
1093
1094 2011-11-04  Tristan Gingold  <gingold@adacore.com>
1095
1096         * config/alpha/vms.h (ASM_OUTPUT_DEF): Do not switch section.
1097
1098 2011-11-04  Ira Rosen  <ira.rosen@linaro.org>
1099
1100         Unrevert:
1101         2011-10-24  Ira Rosen  <ira.rosen@linaro.org>
1102
1103         PR tree-optimization/50730
1104         * tree-vect-data-refs.c (vect_analyze_data_refs): Stop basic block
1105         analysis if encountered unsupported data-ref.
1106
1107 2011-11-04  Jakub Jelinek  <jakub@redhat.com>
1108
1109         * config/i386/i386.c (ix86_expand_vector_convert_uns_vsivsf): New
1110         function.
1111         * config/i386/i386-protos.h (ix86_expand_vector_convert_uns_vsivsf):
1112         New prototype.
1113         * config/i386/sse.md (floatuns<sseintvecmodelower><mode>2): Use it.
1114         For floatunsv8siv8sf2 require TARGET_AVX2.
1115
1116         * config/i386/i386.c (ix86_expand_adjust_ufix_to_sfix_si): Add
1117         XORP argument.  Subtract 0x1p31 instead of 0x1p32.  Use normal
1118         signalling comparison instead of non-signalling.  Store into
1119         *XORP pseudo holding 0x80000000 integers if 0x1p31 has been
1120         subtracted and 0 otherwise.
1121         * config/i386/i386-protos.h (ix86_expand_adjust_ufix_to_sfix_si):
1122         Adjust prototype.
1123         * config/i386/sse.md (fixuns_trunc<mode><sseintvecmodelower>2): Enable
1124         already for TARGET_SSE2.  Xor in vector initialized by
1125         ix86_expand_adjust_ufix_to_sfix_si at the end.
1126         (vec_pack_ufix_trunc_<mode>): Likewise.
1127
1128         * tree-vect-stmts.c (vectorizable_conversion): Rewritten to handle
1129         not just FLOAT_EXPR and FIX_TRUNC_EXPR, but also CONVERT_EXPR_CODE_P,
1130         WIDEN_MULT_EXPR and WIDEN_LSHIFT_EXPR to handle what
1131         vectorizable_type_demotion and vectorizable_type_promotion did.
1132         Additionally handle FLOAT_EXPR and FIX_TRUNC_EXPR where the integer
1133         is {,un}signed {char,short}.
1134         (vect_create_vectorized_demotion_stmts): Fix comment typo.  For
1135         recursive calls unconditionally use VEC_PACK_TRUNC_EXPR.
1136         Push vec_dest back to the vec_dsts vector at the end.
1137         (vect_create_vectorized_promotion_stmts): Don't recurse, do just
1138         one step.  Removed multi_step_cvt, vec_dsts, slp_node and
1139         prev_stmt_info arguments, add vec_dest argument.  Push always
1140         into vec_tmp, not just when multi_step_cvt != 0, replace *vec_oprdn0
1141         with vec_tmp at the end after freeing old *vec_oprnd0 vector.
1142         (vectorizable_type_demotion, vectorizable_type_promotion): Removed.
1143         (vect_analyze_stmt): Don't call vectorizable_type_demotion and
1144         vectorizable_type_promotion.  Call vectorizable_conversion even
1145         for SLP bb vectorization.
1146         (vect_transform_stmt): Call vectorizable_conversion instead of
1147         vectorizable_type_demotion and vectorizable_type_promotion.
1148         (supportable_widening_operation): Clear *multi_step_cvt first,
1149         simplify c1/c2 computation, free *interm_types vector on failure.
1150         (supportable_narrowing_operation): Clear *multi_step_cvt first,
1151         free *interm_types vector on failure, handle multi-step
1152         FIX_TRUNC_EXPR.
1153
1154 2011-11-04  Tristan Gingold  <gingold@adacore.com>
1155
1156         * config/alpha/alpha.c (alpha_write_linkage): Remove fundecl
1157         argument.  Conditionally generate crash debug info.  Adjust
1158         for alpha_funcs_tree removal.
1159         (machine_function): Add links field.
1160         (alpha_start_function): Conditionally generate crash debug info.
1161         (alpha_end_function): Adjust call to alpha_write_linkage.
1162         (alpha_funcs): Remove.
1163         (links_kind): Remove.
1164         (alpha_links): Remove num, target and lkind field.  Add func field.
1165         (alpha_links_tree): Remove.
1166         (alpha_funcs_tree): Remove.
1167         (alpha_need_linkage): Remove.
1168         (alpha_use_linkage): Change prototype.  Adjust.
1169         (alpha_write_one_linkage): Use ASM_OUTPUT_INTERNAL_LABEL.
1170         Use SYMBOL_REF_EXTERNAL_P and SYMBOL_REF_LOCAL_P macro.
1171         * config/alpha/alpha-protos.h (alpha_use_linkage): Update.
1172         (alpha_need_linkage): Remove.
1173         * config/alpha/alpha.md: Update calls to alpha_use_linkage.
1174         Adjust calls to alpha_need_linkage.
1175
1176 2011-11-03  Uros Bizjak  <ubizjak@gmail.com>
1177
1178         * sched-vis.c (print_value): Handle STRICT_LOW_PART.
1179
1180 2011-11-03  Uros Bizjak  <ubizjak@gmail.com>
1181
1182         * config/i386/i386.md (lround<X87MODEF:mode><SWI248x:mode>2,
1183         rint<mode>2, floor<mode>2, lfloor<MODEF:mode><SWI48:mode>2,
1184         btrunc<mode>2, lwp_lwpval<mode>3): Use operands[N] instead of operandN.
1185
1186 2011-11-03  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1187
1188         PR target/50978
1189         * config/arm/t-bpabi: New file.
1190         * config.gcc (arm*-*-linux*): Add arm/t-bpabi to tmake_file for
1191         arm*-*-linux-*eabi.
1192         (arm*-*-uclinux*): Add arm/t-bpabi to tmake_file for
1193         arm*-*-uclinux*eabi.
1194         (arm*-*-eabi*, arm*-*-symbianelf*): Add arm/t-bpabi to tmake_file
1195         for arm*-*-eabi*.
1196
1197 2011-11-03  Michael Matz  <matz@suse.de>
1198
1199         PR bootstrap/50857
1200         * configure.ac: Check for -fno-exceptions -fno-rtti.
1201         * configure: Regenerate.
1202         * Makefile.in (NOEXCEPTION_FLAGS): New flag.
1203         (ALL_CXXFLAGS): Use it.
1204
1205 2011-11-03  Uros Bizjak  <ubizjak@gmail.com>
1206
1207         * config/i386/i386.md: Use {} for multi-line preparation statements.
1208
1209 2011-11-03  Eric Botcazou  <ebotcazou@adacore.com>
1210
1211         * config/sparc/sparc.md (movtf_insn_sp32_no_fpu): Consolidate into...
1212         (movtf_insn_sp32): ...this.
1213         (movtf_insn_sp64_no_fpu): Consolidate into...
1214         (movtf_insn_sp64): ...this.
1215         (movtf_insn_sp64_hq): Do not test TARGET_FPU.
1216         * config/sparc/sparc.c (sparc_legitimate_address_p): Likewise.
1217
1218 2011-11-03  Tristan Gingold  <gingold@adacore.com>
1219
1220         * config/vms/vms.c (vms_patch_builtins): Fix typo.
1221
1222 2011-11-03  Richard Guenther  <rguenther@suse.de>
1223
1224         PR lto/44965
1225         * lto-opts.c: Re-implement.
1226         * lto-streamer.h (lto_register_user_option): Remove.
1227         (lto_read_file_options): Likewise.
1228         (lto_reissue_options): Likewise.
1229         (lto_clear_user_options): Likewise.
1230         (lto_clear_file_options): Likewise.
1231         * opts-global.c (post_handling_callback): Remove.
1232         (set_default_handlers): Do not set post_handling_callback.
1233         (decode_options): Remove LTO specific code.
1234         * lto-wrapper.c (merge_and_complain): New function.
1235         (run_gcc): Read all input file options and
1236         prepend a merged set before the linker driver options.
1237         * gcc.c (driver_post_handling_callback): Remove.
1238         (set_option_handlers): Do not set post_handling_callback.
1239         * opts-common.c (handle_option): Do not call post_handling_callback.
1240         * opts.h (struct cl_option_handlers): Remove post_handling_callback.
1241
1242 2011-11-03  Richard Guenther  <rguenther@suse.de>
1243
1244         * collect2.c (main): Guard object_nbr variable with TARGET_AIX_VERSION.
1245
1246 2011-11-03  Martin Jambor  <mjambor@suse.cz>
1247
1248         * ipa-prop.c (type_change_info): New fields offset, object,
1249         known_current_type and multiple_types_encountered.
1250         (extr_type_from_vtbl_ptr_store): New function.
1251         (check_stmt_for_type_change): Use it, set multiple_types_encountered if
1252         the result is different from the previous one.
1253         (detect_type_change): Renamed to detect_type_change_1. New parameter
1254         comp_type.  Set up new fields in tci, build known type jump
1255         functions if the new type can be identified.
1256         (detect_type_change): New function.
1257         * tree.h (DECL_CONTEXT): Comment new use.
1258
1259 2011-11-03  Richard Guenther  <rguenther@suse.de>
1260
1261         PR lto/48217
1262         * lto-wrapper.c (get_options_from_collect_gcc_options): Properly
1263         decode an encoded literal '.
1264
1265 2011-11-03  Tristan Gingold  <gingold@adacore.com>
1266
1267         * collect2.c (main): Add support of -f (response file) on AIX.
1268
1269 2011-11-03  Ira Rosen  <ira.rosen@linaro.org>
1270
1271         PR tree-optimization/50912
1272         * tree-vectorizer.h (slp_void_p): New.
1273         (struct _slp_tree): Replace left and right with children.  Update
1274         documentation.
1275         (struct _slp_oprnd_info): New.
1276         (vect_get_vec_defs): Declare.
1277         (vect_get_slp_defs): Update arguments.
1278         * tree-vect-loop.c (vect_create_epilog_for_reduction): Call
1279         vect_get_vec_defs instead of vect_get_slp_defs.
1280         (vectorizable_reduction): Likewise.
1281         * tree-vect-stmts.c (vect_get_vec_defs): Remove static, add argument.
1282         Update call to vect_get_slp_defs.
1283         (vectorizable_conversion): Update call to vect_get_vec_defs.
1284         (vectorizable_assignment, vectorizable_shift,
1285         vectorizable_operation): Likewise.
1286         (vectorizable_type_demotion): Call vect_get_vec_defs instead of
1287         vect_get_slp_defs.
1288         (vectorizable_type_promotion, vectorizable_store): Likewise.
1289         (vect_analyze_stmt): Fix typo.
1290         * tree-vect-slp.c (vect_free_slp_tree): Update SLP tree traversal.
1291         (vect_print_slp_tree, vect_mark_slp_stmts,
1292         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
1293         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
1294         vect_schedule_slp_instance): Likewise.
1295         (vect_create_new_slp_node): New.
1296         (vect_create_oprnd_info, vect_free_oprnd_info): Likewise.
1297         (vect_get_and_check_slp_defs): Pass information about defs using
1298         oprnds_info, allow any number of operands.
1299         (vect_build_slp_tree): Likewise.  Update calls to
1300         vect_get_and_check_slp_defs.  Fix comments.
1301         (vect_analyze_slp_instance): Move node creation to
1302         vect_create_new_slp_node.
1303         (vect_get_slp_defs): Allow any number of operands.
1304
1305 2011-11-02  Peter Bergner  <bergner@vnet.ibm.com>
1306             Iain Sandoe  <iains@gcc.gnu.org>
1307
1308         * config/rs6000/rs6000.c (USE_HIDDEN_LINKONCE): New define.
1309         (get_ppc476_thunk_name): Use it.
1310         (rs6000_code_end): Likewise.
1311         (macho_branch_islands): Fix typo.
1312
1313 2011-11-02  Paolo Carlini  <paolo.carlini@oracle.com>
1314             Jason Merrill  <jason@redhat.com>
1315
1316         PR c++/50810
1317         * configure.ac: Add -Wno-narrowing to warning options.
1318         * doc/invoke.texi ([-Wnarrowing], [-Wc++0x-compat]): Update.
1319
1320 2011-11-02  Eric Botcazou  <ebotcazou@adacore.com>
1321
1322         PR target/50945
1323         * config/sparc/sparc.md (movsf_insn): Reindent constraints.
1324         (movdf_insn_sp32): Likewise.  Remove redundant G constraint.
1325         (movdf_insn_sp64): Likewise.
1326         (DFmode splitter): Do not test TARGET_FPU.
1327         (movtf_insn_sp32): Reindent constraints.
1328         (movtf_insn_sp32_no_fpu): Likewise.
1329         (movtf_insn_sp64): Likewise.
1330         (movtf_insn_sp64_hq): Likewise.
1331         (movtf_insn_sp64_no_fpu): Likewise.
1332
1333 2011-11-02  Paolo Carlini  <paolo.carlini@oracle.com>
1334
1335         PR c++/50956
1336         * builtins.c (fold_builtin_memchr): Fix cast.
1337
1338 2011-11-02  Teresa Johnson  <tejohnson@google.com>
1339
1340         * config/i386/predicates.md (promotable_binary_operator): Add minus
1341         to the list of promotable operators.
1342
1343 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1344
1345         * gthr-single.h, gthr.h: Move to ../libgcc.
1346         * gthr-aix.h: Move to ../libgcc/config/rs6000.
1347         * gthr-dce.h: Move to ../libgcc/config/pa.
1348         * gthr-lynx.h: Move to ../libgcc/config.
1349         * gthr-mipssde.h: Move to ../libgcc/config/mips.
1350         * gthr-posix.h: Move to ../libgcc/config.
1351         * gthr-rtems.h: Likewise.
1352         * gthr-tpf.h: Move to ../libgcc/config/s390.
1353         * gthr-vxworks.h: Move to ../libgcc/config.
1354         * gthr-win32.h: Move to ../libgcc/config/i386.
1355         * configure.ac (gthread_flags): Remove
1356         (gthr-default.h): Don't create.
1357         (thread_file): Don't substitute.
1358         * configure: Regenerate.
1359         * Makefile.in (GCC_THREAD_FILE): Remove.
1360         (GTHREAD_FLAGS): Remove.
1361         (libgcc.mvars): Remove GTHREAD_FLAGS.
1362         * config/t-vxworks (EXTRA_HEADERS): Remove.
1363
1364 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1365             Paolo Bonzini  <bonzini@gnu.org>
1366
1367         * configure.ac (libgcc_tm_file_list, libgcc_tm_include_list): Remove.
1368         * configure: Regenerate.
1369         * Makefile.in (libgcc_tm_file_list, libgcc_tm_include_list): Remove.
1370         (TM_H): Remove libgcc_tm.h, $(libgcc_tm_file_list).
1371         (libgcc_tm.h, cs-libgcc_tm.h): Remove.
1372         (clean): Remove libgcc_tm.h
1373         * mkconfig.sh: Don't include libgcc_tm.h in tm.h.
1374         * config.gcc (libgcc_tm_file): Remove.
1375         (arm*-*-linux*): Remove libgcc_tm_file for arm*-*-linux-*eabi.
1376         (arm*-*-uclinux*): Remove libgcc_tm_file for arm*-*-uclinux*eabi.
1377         (arm*-*-eabi*, arm*-*-symbianelf*): Remove libgcc_tm_file.
1378         (avr-*-rtems*): Likewise.
1379         (avr-*-*): Likewise.
1380         (frv-*-elf): Likewise.
1381         (frv-*-*linux*): Likewise.
1382         (h8300-*-rtems*): Likewise.
1383         (h8300-*-elf*): Likewise.
1384         (i[34567]86-*-darwin*): Likewise.
1385         (x86_64-*-darwin*): Likewise.
1386         (rx-*-elf*): Likewise.
1387         (tic6x-*-elf): Likewise.
1388         (tic6x-*-uclinux): Likewise.
1389         (i[34567]86-*-linux*, x86_64-*-linux*): Likewise.
1390
1391 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1392
1393         * Makefile.in (LIBGCC2_DEBUG_CFLAGS LIBGCC2_CFLAGS)
1394         (LIBGCC2_INCLUDES, TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXTRA)
1395         (LIB2FUNCS_STATIC_EXTRA, LIB2FUNCS_EXCLUDE, T, T_TARGET)
1396         (INCLUDES_FOR_TARGET): Remove.
1397         (LIBGCC2_CFLAGS): Don't export.
1398         (LIB2FUNCS_ST, LIB2_DIVMOD_FUNCS, LIB2ADD, LIB2ADD_ST, srcdirify):
1399         Remove.
1400         (libgcc-support): Remove $(LIB2ADD), $(LIB2ADD_ST) dependencies.
1401         (libgcc.mvars): Likewise.
1402         Don't emit LIB2FUNCS_ST, LIB2FUNCS_EXCLUDE, LIB2ADD, LIB2ADD_ST,
1403         LIB2_SIDITI_CONV_FUNCS, LIB2_DIVMOD_FUNCS, LIBGCC2_CFLAGS,
1404         TARGET_LIBGCC2_CFLAGS.
1405         Emit GTHREAD_FLAGS.
1406         * libgcc2.c, libgcc2.h, gbl-ctors.h, longlong.h: Move to ../libgcc.
1407         * config/darwin-64.c: Move to ../libgcc/config.
1408         * config/divmod.c, config/floatunsidf.c, config/floatunsisf.c,
1409         config/floatunsitf.c, config/floatunsixf.c, config/udivmod.c,
1410         config/udivmodsi4.c: Move to ../libgcc/config.
1411         * config/gthr-posix.c: Move to ../libgcc/config/alpha.
1412         * config/memcmp.c, config/memcpy.c, config/memmove.c,
1413         config/memset.c: Move to ../libgcc/config.
1414         * config/t-darwin (TARGET_LIBGCC2_CFLAGS): Remove.
1415         * config/t-freebsd: Remove.
1416         * config/t-freebsd-thread: Move to ../libgcc/config.
1417         * config/t-libgcc-pic: Move to ../libgcc/config.
1418         * config/t-libunwind (TARGET_LIBGCC2_CFLAGS): Remove.
1419         * config/t-linux: Remove.
1420         * config/t-lynx (TARGET_LIBGCC2_CFLAGS, LIBGCC, INSTALL_LIBGCC):
1421         Remove.
1422         * config/t-openbsd-thread: Move to ../libgcc/config.
1423         * config/t-rtems (LIBGCC2_INCLUDES): Remove.
1424         * config/t-sol2 (TARGET_LIBGCC2_CFLAGS): Remove.
1425         * config/t-svr4: Remove.
1426         * config/t-vxworks (LIBGCC, INSTALL_LIBGCC, TARGET_LIBGCC2_CFLAGS)
1427         (LIBGCC2_DEBUG_CFLAGS, LIB2FUNCS_EXTRA, LIBGCC2_INCLUDES): Remove.
1428         * config/vxlib.c, config/vxlib-tls.c: Move to ../libgcc/config.
1429         * config/alpha/qrnnd.asm: Move to ../libgcc/config/alpha/qrnnd.S.
1430         * config/alpha/t-alpha, config/alpha/t-ieee: Remove.
1431         * config/alpha/t-vms (LIB2FUNCS_EXTRA, LIBGCC, INSTALL_LIBGCC): Remove.
1432         * config/alpha/vms-gcc_shell_handler.c: Move to ../libgcc/config/alpha.
1433         * config/arm/bpabi.c, config/arm/unaligned-funcs.c,
1434         config/arm/fp16.c, config/arm/linux-atomic.c,
1435         config/arm/linux-atomic-64bit.c: Move to ../libgcc/config/arm.
1436         * config/arm/t-arm-elf (LIBGCC, INSTALL_LIBGCC)
1437         (TARGET_LIBGCC2_CFLAGS): Remove.
1438         * config/arm/t-bpabi, config/arm/t-linux: Remove.
1439         * config/arm/t-linux-eabi (TARGET_LIBGCC2_CFLAGS)
1440         (LIB2FUNCS_STATIC_EXTRA): Remove.
1441         * config/arm/t-netbsd: Remove.
1442         * config/arm/t-strongarm-elf (LIBGCC, INSTALL_LIBGCC)
1443         (TARGET_LIBGCC2_CFLAGS): Remove.
1444         * config/arm/t-symbian (LIB2FUNCS_STATIC_EXTRA): Remove.
1445         * config/arm/t-wince-pe (LIBGCC, INSTALL_LIBGCC)
1446         (TARGET_LIBGCC2_CFLAGS): Remove.
1447         * config/avr/t-avr (LIB2FUNCS_EXCLUDE, TARGET_LIBGCC2_CFLAGS)
1448         (LIBGCC, INSTALL_LIBGCC): Remove.
1449         * config/bfin/t-bfin-elf (TARGET_LIBGCC2_CFLAGS): Remove.
1450         * config/bfin/t-bfin-linux: Likewise.
1451         * config/bfin/t-bfin-uclinux: Likewise.
1452         * config/c6x/eqd.c, config/c6x/eqf.c, config/c6x/ged.c,
1453         config/c6x/gef.c, config/c6x/gtd.c, config/c6x/gtf.c,
1454         config/c6x/led.c, config/c6x/lef.c, config/c6x/ltd.c,
1455         config/c6x/ltf.c: Move to ../libgcc/config/c6x.
1456         * config/c6x/t-c6x-elf (LIB2FUNCS_EXCLUDE, LIB2FUNCS_EXTRA): Remove.
1457         * config/c6x/t-c6x-uclinux (TARGET_LIBGCC2_CFLAGS): Remove.
1458         * config/cris/arit.c: Move to ../libgcc/config/cris.
1459         * config/cris/cris_abi_symbol.c: Remove.
1460         * config/cris/cris.h: Remove obsolete comment.
1461         * config/cris/mulsi3.asm: Move to ../libgcc/config/cris/mulsi3.S.
1462         * config/cris/t-cris (LIB2FUNCS_EXTRA, CRIS_LIB1CSRC)
1463         ($(LIB2FUNCS_EXTRA)): Remove.
1464         * config/cris/t-elfmulti (LIB2FUNCS_STATIC_EXTRA, INSTALL_LIBGCC)
1465         (LIBGCC): Remove.
1466         * config/cris/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
1467         * config/fr30/t-fr30: Remove.
1468         * config/frv/cmovd.c, config/frv/cmovh.c, config/frv/cmovw.c,
1469         config/frv/modi.c, config/frv/uitod.c, config/frv/uitof.c,
1470         config/frv/ulltod.c, config/frv/ulltof.c, config/frv/umodi.c: Move
1471         to ../libgcc/config/frv.
1472         * config/frv/t-frv (LIB2FUNCS_EXTRA, TARGET_LIBGCC2_CFLAGS)
1473         (cmovh.c, cmovw.c, cmovd.c, modi.c, umodi.c, uitof.c, uitod.c)
1474         (ulltof.c, LIBGCC, INSTALL_LIBGCC): Remove.
1475         * config/frv/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
1476         * config/h8300/clzhi2.c, config/h8300/ctzhi2.c,
1477         config/h8300/fixunssfsi.c, config/h8300/parityhi2.c,
1478         config/h8300/popcounthi2.c: Move to ../libgcc/config/h8300.
1479         * config/h8300/t-h8300 (LIB2FUNCS_EXTRA, TARGET_LIBGCC2_CFLAGS)
1480         (LIBGCC, INSTALL_LIBGCC): Remove.
1481         * config/i386/gthr-win32.c: Move to ../libgcc/config/i386.
1482         * config/i386/t-cygming (LIBGCC2_INCLUDES): Remove.
1483         * config/i386/t-cygwin: Remove.
1484         * config/i386/t-darwin (LIB2_SIDITI_CONV_FUNCS, LIB2FUNCS_EXTRA)
1485         (LIB2FUNCS_EXCLUDE): Remove.
1486         * config/i386/t-darwin64 (LIB2_SIDITI_CONV_FUNCS, LIB2FUNCS_EXTRA)
1487         (LIBGCC, INSTALL_LIBGCC): Remove.
1488         * config/i386/t-gthr-win32: Move to ../libgcc/config/i386.
1489         * config/i386/t-linux64 (LIBGCC, INSTALL_LIBGCC): Remove.
1490         * config/i386/t-mingw-w32: Likewise.
1491         * config/i386/t-mingw-w64: Likewise.
1492         * config/i386/t-openbsd: Likewise.
1493         * config/i386/t-nto: Remove.
1494         * config/ia64/quadlib.c: Move to ../libgcc/config/ia64.
1495         * config/ia64/t-hpux (LIBGCC, INSTALL_LIBGCC, LIB2FUNCS_EXTRA)
1496         (quadlib.c): Remove.
1497         * config/ia64/t-ia64: Remove comment.
1498         * config/iq2000/lib2extra-funcs.c: Move to
1499         ../libgcc/config/iq2000/lib2funcs.c.
1500         * config/iq2000/t-iq2000: Remove.
1501         * config/m32c/m32c-lib2.c: Move to ../libgcc/config/m32c/lib2funcs.c.
1502         * config/m32c/m32c-lib2-trapv.c: Move to ../libgcc/config/m32c/trapv.c.
1503         * config/m32r/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
1504         * config/m32c/t-m32c (LIB2FUNCS_EXTRA): Remove.
1505         * config/m32r/t-m32r (TARGET_LIBGCC2_CFLAGS, LIBGCC)
1506         (INSTALL_LIBGCC): Remove.
1507         * config/m68k/fpgnulib.c: Move to ../libgcc/config/m68k.
1508         * config/m68k/t-floatlib: Remove.
1509         * config/m68k/t-mlibs (LIBGCC, INSTALL_LIBGCC): Remove.
1510         * config/mcore/t-mcore (TARGET_LIBGCC2_CFLAGS): Remove.
1511         Fix typo.
1512         (LIBGCC, INSTALL_LIBGCC): Remove.
1513         * config/mep/mep-lib2.c: Move to ../libgcc/config/mep/lib2funcs.c.
1514         * config/mep/mep-tramp.c: Move to ../libgcc/config/mep/tramp.c.
1515         * config/mep/t-mep (LIB2FUNCS_EXTRA): Remove.
1516         * config/mips/t-elf (TARGET_LIBGCC2_CFLAGS, LIBGCC)
1517         (INSTALL_LIBGCC): Remove.
1518         * config/mips/t-isa3264: Likewise.
1519         * config/mips/t-mips (LIB2_SIDITI_CONV_FUNCS): Remove.
1520         * config/mips/t-r3900 (TARGET_LIBGCC2_CFLAGS, LIBGCC)
1521         (INSTALL_LIBGCC): Remove.
1522         * config/mips/t-sde (LIBGCC, INSTALL_LIBGCC): Remove.
1523         * config/mips/t-sr71k (TARGET_LIBGCC2_CFLAGS, LIBGCC)
1524         (INSTALL_LIBGCC): Remove.
1525         * config/mips/t-vr (TARGET_LIBGCC2_CFLAGS)
1526         (LIB2FUNCS_STATIC_EXTRA): Remove.
1527         * config/mips/vr4120-div.S: Move to ../libgcc/config/mips.
1528         * config/mmix/t-mmix (TARGET_LIBGCC2_CFLAGS): Remove.
1529         * config/mn10300/t-mn10300 (LIBGCC, INSTALL_LIBGCC): Remove.
1530         * config/pa/fptr.c, config/pa/linux-atomic.c: Move to
1531         ../libgcc/config/pa.
1532         * config/pa/lib2funcs.asm: Move to ../libgcc/config/pa/lib2funcs.S.
1533         * config/pa/quadlib.c: Move to ../libgcc/config/pa.
1534         * config/pa/t-dce-thr (LIBGCC, INSTALL_LIBGCC): Remove.
1535         * config/pa/t-linux, config/pa/t-linux64: Remove.
1536         * config/pa/t-pa-hpux, config/pa/t-pa-hpux10,
1537         config/pa/t-pa-hpux11, config/pa/t-pa64: Remove.
1538         * config/pdp11/t-pdp11 (TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXTRA):
1539         Remove.
1540         * config/picochip/libgccExtras: Move to ../libgcc/config/picochip.
1541         * config/picochip/t-picochip (LIB2FUNCS_EXTRA, RANLIB_FOR_TARGET)
1542         (TARGET_LIBGCC2_CFLAGS, LIBGCC2_DEBUG_CFLAGS): Remove.
1543         * config/rs6000/crtresfpr.asm: Move to
1544         ../libgcc/config/rs6000/crtresfpr.S.
1545         * config/rs6000/crtresgpr.asm: Move to
1546         ../libgcc/config/rs6000/crtresgpr.S.
1547         * config/rs6000/crtresxfpr.asm: Move to
1548         ../libgcc/config/rs6000/crtresxfpr.S.
1549         * config/rs6000/crtresxgpr.asm: Move to
1550         ../libgcc/config/rs6000/crtresxgpr.S.
1551         * config/rs6000/crtsavfpr.asm: Move to
1552         ../libgcc/config/rs6000/crtsavfpr.S.
1553         * config/rs6000/crtsavgpr.asm: Move to
1554         ../libgcc/config/rs6000/crtsavgpr.S.
1555         * config/rs6000/darwin-asm.h: Move to ../libgcc/config/rs6000.
1556         * config/rs6000/darwin-fpsave.asm: Move to
1557         ../libgcc/config/rs6000/darwin-fpsave.S.
1558         * config/rs6000/darwin-gpsave.asm: Move to
1559         ../libgcc/config/rs6000/darwin-gpsave.S.
1560         * config/rs6000/darwin-tramp.asm: Move to
1561         ../libgcc/config/rs6000/darwin-tramp.S.
1562         * config/rs6000/darwin-vecsave.asm: Move to
1563         ../libgcc/config/rs6000/darwin-vecsave.S.
1564         * config/rs6000/darwin-world.asm: Move to
1565         ../libgcc/config/rs6000/darwin-world.S.
1566         * config/rs6000/e500crtres32gpr.asm: Move to
1567         ../libgcc/config/rs6000/e500crtres32gpr.S.
1568         * config/rs6000/e500crtres64gpr.asm: Move to
1569         ../libgcc/config/rs6000/e500crtres64gpr.S.
1570         * config/rs6000/e500crtres64gprctr.asm: Move to
1571         ../libgcc/config/rs6000/e500crtres64gprctr.S.
1572         * config/rs6000/e500crtrest32gpr.asm: Move to
1573         ../libgcc/config/rs6000/e500crtrest32gpr.S.
1574         * config/rs6000/e500crtrest64gpr.asm: Move to
1575         ../libgcc/config/rs6000/e500crtrest64gpr.S.
1576         * config/rs6000/e500crtresx32gpr.asm: Move to
1577         ../libgcc/config/rs6000/e500crtresx32gpr.S.
1578         * config/rs6000/e500crtresx64gpr.asm: Move to
1579         ../libgcc/config/rs6000/e500crtresx64gpr.S.
1580         * config/rs6000/e500crtsav32gpr.asm: Move to
1581         ../libgcc/config/rs6000/e500crtsav32gpr.S.
1582         * config/rs6000/e500crtsav64gpr.asm: Move to
1583         ../libgcc/config/rs6000/e500crtsav64gpr.S.
1584         * config/rs6000/e500crtsav64gprctr.asm: Move to
1585         ../libgcc/config/rs6000/e500crtsav64gprctr.S.
1586         * config/rs6000/e500crtsavg32gpr.asm: Move to
1587         ../libgcc/config/rs6000/e500crtsavg32gpr.S.
1588         * config/rs6000/e500crtsavg64gpr.asm: Move to
1589         ../libgcc/config/rs6000/e500crtsavg64gpr.S.
1590         * config/rs6000/e500crtsavg64gprctr.asm: Move to
1591         ../libgcc/config/rs6000/e500crtsavg64gprctr.S.
1592         * config/rs6000/eabi.asm: Move to ../libgcc/config/rs6000/eabi.S.
1593         * config/rs6000/t-aix43 (LIBGCC, INSTALL_LIBGCC, LIB2FUNCS_EXTRA)
1594         (TARGET_LIBGCC2_CFLAGS): Remove.
1595         * config/rs6000/t-aix52: Likewise.
1596         * config/rs6000/t-darwin: Remove.
1597         * config/rs6000/t-darwin64 (LIB2_SIDITI_CONV_FUNCS)
1598         (LIB2FUNCS_EXTRA): Remove.
1599         * config/rs6000/t-fprules (LIBGCC, INSTALL_LIBGCC): Remove.
1600         * config/rs6000/t-linux64 (TARGET_LIBGCC2_CFLAGS): Remove.
1601         * config/rs6000/t-lynx (LIB2FUNCS_EXTRA, tramp.S, LIBGCC)
1602         (INSTALL_LIBGCC): Remove.
1603         * config/rs6000/t-netbsd (LIB2FUNCS_EXTRA)
1604         (LIB2FUNCS_STATIC_EXTRA, tramp.S, crtsavfpr.S, crtresfpr.S)
1605         (crtsavgpr.S, crtresgpr.S, crtresxfpr.S, crtresxgpr.S, LIBGCC)
1606         (INSTALL_LIBGCC, $(T)crtsavfpr$(objext), $(T)crtresfpr$(objext))
1607         ($(T)crtsavgpr$(objext), $(T)crtresgpr$(objext))
1608         ($(T)crtresxfpr$(objext), $(T)crtresxgpr$(objext)): Remove.
1609         * config/rs6000/t-ppccomm (LIB2FUNCS_EXTRA)
1610         (LIB2FUNCS_STATIC_EXTRA, eabi.S, tramp.S): Remove.
1611         * config/rs6000/t-spe (LIBGCC, INSTALL_LIBGCC): Remove.
1612         * config/rs6000/t-vxworks: Remove comment.
1613         * config/rs6000/tramp.asm: Move to ../libgcc/config/rs6000/tramp.S.
1614         * config/rx/t-rx (LIBGCC, INSTALL_LIBGCC): Remove.
1615         * config/sh/linux-atomic.asm: Move to
1616         ../libgcc/config/sh/linux-atomic.S.
1617         * config/sh/t-linux (LIB2FUNCS_EXTRA): Remove.
1618         * config/sh/t-netbsd: Remove.
1619         * config/sh/t-sh (TARGET_LIBGCC2_CFLAGS, LIBGCC, INSTALL_LIBGCC):
1620         Remove.
1621         * config/sparc/t-elf (LIBGCC, INSTALL_LIBGCC): Remove.
1622         * config/sparc/t-leon: Likewise.
1623         * config/sparc/t-leon3: Likewise.
1624         * config/sparc/t-linux64: Likewise.
1625         * config/sparc/t-netbsd64: Fix typo.
1626         Remove comment.
1627         * config/spu/divmodti4.c, config/spu/divv2df3.c,
1628         config/spu/float_disf.c, config/spu/float_unsdidf.c,
1629         config/spu/float_unsdisf.c, config/spu/float_unssidf.c,
1630         config/spu/mfc_multi_tag_release.c,
1631         config/spu/mfc_multi_tag_reserve.c, config/spu/mfc_tag_release.c,
1632         config/spu/mfc_tag_reserve.c, config/spu/mfc_tag_table.c,
1633         config/spu/multi3.c: Move to ../libgcc/config/spu.
1634         * config/spu/t-spu-elf (TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXCLUDE)
1635         (LIB2FUNCS_STATIC_EXTRA, LIB2_SIDITI_CONV_FUNCS, LIBGCC)
1636         (INSTALL_LIBGCC): Remove.
1637         * config/stormy16/stormy16-lib2.c: Move to
1638         ../libgcc/config/stormy16/lib2.c.
1639         * config/stormy16/stormy16-lib2-ashlsi3.c: Move to
1640         ../libgcc/config/stormy16/ashlsi3.c.
1641         * config/stormy16/stormy16-lib2-ashrsi3.c: Move to
1642         ../libgcc/config/stormy16/ashrsi3.c.
1643         * config/stormy16/stormy16-lib2-clzhi2.c: Move to
1644         ../libgcc/config/stormy16/clzhi2.c.
1645         * config/stormy16/stormy16-lib2-cmpsi2.c: Move to
1646         ../libgcc/config/stormy16/cmpsi2.c.
1647         * config/stormy16/stormy16-lib2-ctzhi2.c: Move to
1648         ../libgcc/config/stormy16/ctzhi2.c.
1649         * config/stormy16/stormy16-lib2-divsi3.c: Move to
1650         ../libgcc/config/stormy16/divsi3.c.
1651         * config/stormy16/stormy16-lib2-ffshi2.c: Move to
1652         ../libgcc/config/stormy16/ffshi2.c.
1653         * config/stormy16/stormy16-lib2-lshrsi3.c: Move to
1654         ../libgcc/config/stormy16/lshrsi3.c.
1655         * config/stormy16/stormy16-lib2-modsi3.c: Move to
1656         ../libgcc/config/stormy16/modsi3.c.
1657         * config/stormy16/stormy16-lib2-parityhi2.c: Move to
1658         ../libgcc/config/stormy16/parityhi2.c.
1659         * config/stormy16/stormy16-lib2-popcounthi2.c: Move to
1660         ../libgcc/config/stormy16/popcounthi2.c.
1661         * config/stormy16/stormy16-lib2-ucmpsi2.c: Move to
1662         ../libgcc/config/stormy16/ucmpsi2.c.
1663         * config/stormy16/stormy16-lib2-udivmodsi4.c: Move to
1664         ../libgcc/config/stormy16/udivmodsi4.c.
1665         * config/stormy16/stormy16-lib2-udivsi3.c: Move to
1666         ../libgcc/config/stormy16/udivsi3.c.
1667         * config/stormy16/stormy16-lib2-umodsi3.c: Move to
1668         ../libgcc/config/stormy16/umodsi3.c.
1669         * config/stormy16/t-stormy16: Move to ../libgcc/config/t-stormy16.
1670         * config/v850/t-v850 (INSTALL_LIBGCC): Remove.
1671         * config/xtensa/lib2funcs.S: Move to ../libgcc/config/xtensa.
1672         * config/xtensa/t-elf: Remove.
1673         * config/xtensa/t-xtensa (LIB2FUNCS_EXTRA): Remove.
1674         * config.gcc (*-*-freebsd*): Remove t-freebsd, t-freebsd-thread
1675         from tmake_file.
1676         (*-*-linux*, frv-*-*linux*, *-*-kfreebsd*-gnu, *-*-knetbsd*-gnu,
1677         *-*-gnu*, *-*-kopensolaris*-gnu): Remove t-linux from tmake_file.
1678         (*-*-netbsd*): Remove t-libgcc-pic from tmake_file.
1679         (*-*-openbsd*): Likewise.
1680         Remove t-openbsd-thread for posix threads.
1681         (alpha*-*-linux*): Remove alpha/t-alpha, alpha/t-ieee from tmake_file.
1682         (alpha*-*-freebsd*): Likewise.
1683         (alpha*-*-netbsd*): Likewise.
1684         (alpha*-*-openbsd*): Likewise.
1685         (alpha64-dec-*vms*): Likewise.
1686         (alpha*-dec-*vms*): Likewise.
1687         (arm*-*-netbsdelf*): Remove arm/t-netbsd from tmake_file.
1688         (arm*-*-linux*): Remove t-linux from tmake_file.
1689         Remove arm/t-bpabi from tmake_file for arm*-*-linux-*eabi.
1690         (arm*-*-uclinux*): Remove arm/t-bpabi from tmake_file for
1691         arm*-*-uclinux*eabi.
1692         (arm*-*-eabi*, arm*-*-symbianelf* ): Remove arm/t-bpabi from
1693         tmake_file for arm*-*-eabi*.
1694         (fr30-*-elf): Remove tmake_file.
1695         (hppa*64*-*-linux*): Remove tmake_file.
1696         (hppa*-*-linux*): Likewise.
1697         (hppa[12]*-*-hpux10*): Remove pa/t-pa-hpux10, pa/t-pa-hpux from
1698         tmake_file.
1699         (hppa*64*-*-hpux11*): Remove pa/t-pa64, pa/t-pa-hpux from tmake_file.
1700         (hppa[12]*-*-hpux11*): Remove pa/t-pa-hpux11, pa/t-pa-hpux from
1701         tmake_file.
1702         (i[34567]86-*-elf*): Remove tmake_file.
1703         (x86_64-*-elf*): Likewise.
1704         (i[34567]86-*-nto-qnx*): Likewise.
1705         (i[34567]86-*-cygwin*): Remove i386/t-cygwin from tmake_file.
1706         (i[34567]86-*-mingw*, x86_64-*-mingw*): Remove i386/t-gthr-win32
1707         from tmake_file if using win32 threads.
1708         (iq2000*-*-elf*): Remove tmake-file.
1709         (microblaze*-linux*): Likewise.
1710         (sh-*-elf*, sh[12346l]*-*-elf*, sh-*-linux*)
1711         (sh[2346lbe]*-*-linux*, sh-*-netbsdelf*, shl*-*-netbsdelf*)
1712         (sh5-*-netbsd*, sh5l*-*-netbsd*, sh64-*-netbsd*)
1713         (sh64l*-*-netbsd*): Remove sh/t-netbsd from tmake_file for
1714         sh5*-*-netbsd*, sh64*-netbsd*, *-*-netbsd.
1715         (xtensa*-*-elf*): Remove tmake_file.
1716
1717 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1718
1719         * Makefile.in (LIB1ASMSRC): Don't export.
1720         (libgcc.mvars): Don't emit LIB1ASMFUNCS, LIB1ASMSRC.
1721         * config/arm/arm.c: Update lib1funcs.asm filename.
1722         * config/arm/linux-eabi.h: Likewise.
1723         * config/arm/bpabi-v6m.S, config/arm/bpabi.S,
1724         config/arm/ieee754-df.S, config/arm/ieee754-sf.S: Move to
1725         ../libgcc/config/arm.
1726         * config/arm/lib1funcs.asm: Move to ../libgcc/config/arm/lib1funcs.S.
1727         * config/arm/t-arm (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1728         * config/arm/t-arm-elf (LIB1ASMFUNCS): Remove.
1729         * config/arm/t-bpabi: Likewise.
1730         * config/arm/t-linux (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1731         * config/arm/t-linux-eabi (LIB1ASMFUNCS): Remove.
1732         * config/arm/t-strongarm-elf: Likewise.
1733         * config/arm/t-symbian: Likewise.
1734         * config/arm/t-vxworks: Likewise.
1735         * config/arm/t-wince-pe: Likewise.
1736         * config/avr/libgcc.S: Move to ../libgcc/config/avr.
1737         * config/avr/t-avr (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1738         * config/bfin/lib1funcs.asm: Move to ../libgcc/config/bfin/lib1funcs.S.
1739         * config/bfin/t-bfin: Remove.
1740         * config/bfin/t-bfin-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1741         * config/bfin/t-bfin-linux: Likewise.
1742         * config/bfin/t-bfin-uclinux: Likewise.
1743         * config/c6x/lib1funcs.asm: Move to ../libgcc/config/c6x/lib1funcs.S.
1744         * config/c6x/t-c6x-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1745         * config/fr30/lib1funcs.asm: Move to ../libgcc/config/fr30/lib1funcs.S.
1746         * config/fr30/t-fr30 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1747         * config/frv/lib1funcs.asm: Move to ../libgcc/config/frv/lib1funcs.S.
1748         * config/frv/t-frv (CROSS_LIBGCC1, LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1749         * config/h8300/fixunssfsi.c: Update lib1funcs.asm filename.
1750         * config/h8300/lib1funcs.asm: Move to
1751         ../libgcc/config/h8300/lib1funcs.S.
1752         * config/h8300/t-h8300 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1753         * config/i386/cygwin.asm: Move to ../libgcc/config/i386/cygwin.S.
1754         * config/i386/t-cygming (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1755         * config/i386/t-interix: Likewise.
1756         * config/ia64/lib1funcs.asm: Move to ../libgcc/config/ia64/lib1funcs.S.
1757         * config/ia64/t-hpux (LIB1ASMFUNCS, LIBGCC1_TEST): Remove.
1758         * config/ia64/t-ia64 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1759         * config/iq2000/t-iq2000 (LIBGCC1, CROSS_LIBGCC1): Remove.
1760         * config/m32c/m32c.c: Update m32c-lib1.S filename.
1761         * config/m32c/m32c-lib1.S: Move to ../libgcc/config/m32c/lib1funcs.S.
1762         * config/m32c/t-m32c (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1763         * config/m32r/t-linux (CROSS_LIBGCC1, LIBGCC1, LIBGCC1_TEST): Remove.
1764         * config/m68k/lb1sf68.asm: Move to ../libgcc/config/m68k/lb1sf68.S.
1765         * config/m68k/t-floatlib (LIB1ASMSRC, LIB1ASMFUNCS): New file.
1766         * config/mcore/lib1.asm: Move to ../libgcc/config/mcore/lib1funcs.S.
1767         * config/mcore/t-mcore (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1768         * config/mep/mep-lib1.asm: Move to ../libgcc/config/mep/lib1funcs.S.
1769         * config/mep/t-mep (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1770         * config/mips/mips16.S: Move to ../libgcc/config/mips.
1771         * config/mips/t-libgcc-mips16: Remove.
1772         * config/mips/t-sr71k (LIBGCC1, CROSS_LIBGCC1): Remove.
1773         * config/pa/milli64.S: Move to ../libgcc/config/pa.
1774         * config/pa/t-linux (LIB1ASMFUNCS, LIB1ASMSRC): Remove.
1775         * config/pa/t-linux64: Likewise.
1776         * config/picochip/libgccExtras/fake_libgcc.asm: Move to
1777         ../libgcc/config/picochip/lib1funcs.S.
1778         * config/picochip/t-picochip (LIB1ASMFUNCS, LIB1ASMSRC): Remove.
1779         * config/sh/lib1funcs.asm: Move to ../libgcc/config/sh/lib1funcs.S.
1780         * config/sh/lib1funcs.h: Move to ../libgcc/config/sh.
1781         * config/sh/sh.h: Update lib1funcs.asm filename.
1782         * config/sh/t-linux (LIB1ASMFUNCS_CACHE): Remove.
1783         * config/sh/t-netbsd: Likewise.
1784         * config/sh/t-sh (LIB1ASMSRC, LIB1ASMFUNCS, LIB1ASMFUNCS_CACHE):
1785         Remove.
1786         * config/sh/t-sh64 (LIB1ASMFUNCS): Remove.
1787         * config/sparc/lb1spc.asm: Move to ../libgcc/config/sparc/lb1spc.S.
1788         * config/sparc/lb1spl.asm: Remove.
1789         * config/sparc/t-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1790         * config/sparc/t-leon: Likewise.
1791         * config/spu/t-spu-elf (LIBGCC1, CROSS_LIBGCC1): Remove.
1792         * config/v850/lib1funcs.asm: Move to ../libgcc/config/v850/lib1funcs.S.
1793         * config/v850/t-v850 (LIB1ASMSRC, LIB1ASMFUNCS): Remove
1794         * config/vax/lib1funcs.asm: Move to ../libgcc/config/vax/lib1funcs.S.
1795         * config/vax/t-linux: Remove.
1796         * config/xtensa/ieee754-df.S, config/xtensa/ieee754-sf.S: Move to
1797         ../libgcc/config/xtensa.
1798         * config/xtensa/lib1funcs.asm: Move to
1799         ../libgcc/config/xtensa/lib1funcs.S.
1800         * config/xtensa/t-xtensa (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
1801         * config.gcc (bfin*-rtems*): Remove bfin/t-bfin from tmake_file.
1802         (bfin*-*): Likewise.
1803         (mips64*-*-linux*, mipsisa64*-*-linux*): Remove
1804         mips/t-libgcc-mips16 from tmake_file.
1805         (mips*-*-linux*): Likewise.
1806         (mips*-sde-elf*): Likewise.
1807         (mipsisa32-*-elf*, mipsisa32el-*-elf*, mipsisa32r2-*-elf*)
1808         (mipsisa32r2el-*-elf*, mipsisa64-*-elf*, mipsisa64el-*-elf*)
1809         (mipsisa64r2-*-elf*, mipsisa64r2el-*-elf*): Likewise.
1810         (mipsisa64sb1-*-elf*, mipsisa64sb1el-*-elf*): Likewise.
1811         (mips-*-elf*, mipsel-*-elf*): Likewise.
1812         (mips64-*-elf*, mips64el-*-elf*): Likewise.
1813         (mips64orion-*-elf*, mips64orionel-*-elf*): Likewise.
1814         (mips*-*-rtems*): Likewise.
1815         (mipstx39-*-elf*, mipstx39el-*-elf*): Likewise.
1816         (vax-*-linux*): Remove vax/t-linux from tmake_file.
1817
1818 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1819
1820         * config.gcc (extra_parts): Remove.
1821         (*-*-freebsd*): Remove extra_parts.
1822         (*-*-linux*, frv-*-*linux*, *-*-kfreebsd*-gnu, *-*-knetbsd*-gnu,
1823         *-*-gnu*, *-*-kopensolaris*-gnu): Likewise.
1824         (*-*-netbsd*): Remove t-libc-ok, t-netbsd from tmake_file.
1825         Remove extra_parts for *-*-netbsd*1.[7-9]*, *-*-netbsd[2-9]*,
1826         *-*-netbsdelf[2-9]*.
1827         (*-*-openbsd*): Remove t-libc-ok from tmake_file.
1828         (alpha*-*-linux*): Remove extra_parts.
1829         (alpha*-*-freebsd*): Likewise.
1830         (bfin*-linux-uclibc*): Likewise.
1831         (fr30-*-elf): Likewise.
1832         (moxie-*-elf): Likewise.
1833         (moxie-*-uclinux*): Likewise.
1834         (h8300-*-rtems*): Remove h8300/t-elf from tmake_file.
1835         (h8300-*-elf*): Likewise.
1836         (hppa*64*-*-hpux11*): Remove extra_parts.
1837         (i[34567]86-*-elf*): Remove i386/t-i386elf, i386/t-crtstuff from
1838         tmake_file.
1839         (x86_64-*-elf*): Likewise.
1840         (i[34567]86-*-freebsd*): Remove tmake_file.
1841         (x86_64-*-freebsd*): Likewise.
1842         (x86_64-*-netbsd*): Likewise.
1843         (i[34567]86-*-openbsd2.*, i[34567]86-*openbsd3.[0123]): Remove
1844         t-libc-ok from tmake_file.
1845         (i[34567]86-*-linux*, i[34567]86-*-kfreebsd*-gnu,
1846         i[34567]86-*-knetbsd*-gnu, i[34567]86-*-gnu*,
1847         i[34567]86-*-kopensolaris*-gnu): Remove i386/t-crtstuff from
1848         tmake_file.
1849         Remove extra_parts.
1850         (x86_64-*-linux*, x86_64-*-kfreebsd*-gnu, x86_64-*-knetbsd*-gnu):
1851         Remove i386/t-crtstuff from tmake_file.
1852         (i[34567]86-*-lynxos*): Likewise.
1853         Remove extra_parts.
1854         (ia64*-*-elf*): Remove extra_parts.
1855         (ia64*-*-freebsd*): Likewise.
1856         (ia64*-*-linux*): Likewise.
1857         (ia64-hp-*vms*): Remove ia64/t-vms from tmake_file.
1858         (m32r-*-elf*): Remove extra_parts.
1859         (m32rle-*-elf*): Likewise.
1860         (m32r-*-rtems*): Likewise.
1861         (m68k-*-elf*, fido-*-elf*): Likewise.
1862         (m68k*-*-openbsd*): Remove t-libc-ok from tmake_file.
1863         (m68k-*-rtems*): Remove extra_parts.
1864         (mep-*-*): Likewise.
1865         (microblaze*-linux*): Likewise.
1866         (mips64*-*-linux*, mipsisa64*-*-linux*): Likewise.
1867         (mips*-*-linux*): Likewise.
1868         (powerpc-*-lynxos*): Likewise.
1869         (s390x-ibm-tpf*): Likewise.
1870         (score-*-elf): Likewise.
1871         Remove tmake_file.
1872         (sh-*-elf*, sh[12346l]*-*-elf*, sh-*-linux*, sh[2346lbe]*-*-linux*,
1873         sh-*-netbsdelf*, shl*-*-netbsdelf*, sh5-*-netbsd*, sh5l*-*-netbsd*,
1874         sh64-*-netbsd*, sh64l*-*-netbsd*): Remove sh/t-elf from tmake_file.
1875         Remove sh/t-superh from tmake_file for sh*-superh-elf.
1876         Remove sh/t-linux64 from tmake_file for sh64*-*-linux*.
1877         (sh-*-rtems*): Remove sh/t-elf from tmake_file.
1878         (sh-wrs-vxworks): Likewise.
1879         (sparc-*-linux*): Remove extra_parts.
1880         (sparc64-*-linux*): Likewise.
1881         (sparc64-*-freebsd*, ultrasparc-*-freebsd*): Likewise.
1882         (xstormy16-*-elf): Likewise.
1883         (xtensa*-*-linux*): Remove xtensa/t-linux from tmake_file.
1884         (am33_2.0-*-linux*): Remove extra_parts.
1885         * configure.ac (extra_parts): Don't substitute.
1886         * configure: Regenerate.
1887         * crtstuff.c: Move to ../libgcc.
1888         * Makefile.in (CRTSTUFF_CFLAGS): Remove.
1889         (EXTRA_PARTS): Remove.
1890         (CRTSTUFF_T_CFLAGS): Remove.
1891         (MOSTLYCLEANFILES): Remove $(EXTRA_PARTS).
1892         (GCC_EXTRA_PARTS): Remove.
1893         (libgcc.mvars): Remove GCC_EXTRA_PARTS, CRTSTUFF_CFLAGS,
1894         CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S.
1895         Emit GCC_CFLAGS, INHIBIT_LIBC_CFLAGS.
1896         ($(T)crtbegin.o, $(T)crtend.o, $(T)crtbeginS.o, $(T)crtendS.o)
1897         ($(T)crtbeginT.o): Remove.
1898         * config/alpha/t-vms (EXTRA_PARTS): Remove.
1899         ($(T)vms-dwarf2.o, $(T)vms-dwarf2eh.o): Remove.
1900         * config/alpha/vms-dwarf2.asm: Move to
1901         ../libgcc/config/alpha/vms-dwarf2.S.
1902         * config/alpha/vms-dwarf2eh.asm: Move to
1903         ../libgcc/config/alpha/vms-dwarf2eh.S.
1904         * config/arm/crti.asm: Move to ../libgcc/config/arm/crti.S.
1905         * config/arm/crtn.asm: Move to ../libgcc/config/arm/crtn.S.
1906         * config/arm/t-arm-elf (EXTRA_MULTILIB_PARTS): Remove.
1907         ($(T)crti.o, $(T)crtn.o): Remove.
1908         * config/arm/t-linux: Remove comment.
1909         * config/arm/t-linux-eabi (EXTRA_MULTILIB_PARTS): Remove.
1910         * config/arm/t-strongarm-elf (EXTRA_MULTILIB_PARTS): Remove.
1911         ($(T)crti.o, $(T)crtn.o): Remove.
1912         * config/arm/t-symbian (EXTRA_MULTILIB_PARTS): Remove.
1913         * config/bfin/crti.s: Move to ../libgcc/config/bfin/crti.S.
1914         * config/bfin/crtn.s: Move to ../libgcc/config/bfin/crtn.S.
1915         * config/bfin/crtlibid.s: Move to ../libgcc/config/bfin/crtlibid.S.
1916         * config/bfin/t-bfin (EXTRA_PARTS): Remove.
1917         ($(T)crti.o, $(T)crtn.o): Remove.
1918         * config/bfin/t-bfin-elf (CRTSTUFF_T_CFLAGS): Remove.
1919         ($(T)crti.o, $(T)crtn.o, $(T)crtlibid.o): Remove
1920         (EXTRA_MULTILIB_PARTS): Remove.
1921         * config/bfin/t-bfin-linux (CRTSTUFF_T_CFLAGS,
1922         EXTRA_MULTILIB_PARTS): Remove.
1923         * config/bfin/t-bfin-uclinux (CRTSTUFF_T_CFLAGS): Remove.
1924         ($(T)crtlibid.o): Remove.
1925         (EXTRA_MULTILIB_PARTS): Remove.
1926         * config/c6x/crti.s: Move to ../libgcc/config/c6x/crti.S.
1927         * config/c6x/crtn.s: Move to ../libgcc/config/c6x/crtn.S.
1928         * config/c6x/t-c6x-elf ($(T)crti.o, $(T)crtn.o): Remove.
1929         (EXTRA_MULTILIB_PARTS): Remove.
1930         (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S): Remove.
1931         * config/c6x/t-c6x-uclinux (CRTSTUFF_T_CFLAGS,
1932         CRTSTUFF_T_CFLAGS_S): Remove.
1933         * config/cris/t-elfmulti (CRTSTUFF_T_CFLAGS): Remove.
1934         * config/cris/t-linux (CRTSTUFF_T_CFLAGS_S): Remove.
1935         * config/fr30/crti.asm: Move to ../libgcc/config/fr30/crti.S.
1936         * config/fr30/crtn.asm: Move to ../libgcc/config/fr30/crtn.S.
1937         * config/fr30/t-fr30 ($(T)crti.o, $(T)crtn.o): Remove.
1938         * config/frv/frvbegin.c, config/frv/frvend.c: Move to
1939         ../libgcc/config/frv.
1940         * config/frv/t-frv (EXTRA_MULTILIB_PARTS): Remove.
1941         (FRVSTUFF_CFLAGS, $(T)frvbegin$(objext), $(T)frvend$(objext)): Remove.
1942         * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Remove.
1943         (CRTSTUFF_T_CFLAGS): Remove.
1944         * config/h8300/crti.asm: Move to ../libgcc/config/h8300/crti.S.
1945         * config/h8300/crtn.asm: Move to ../libgcc/config/h8300/crtn.S.
1946         * config/h8300/t-elf: Remove.
1947         * config/i386/cygming-crtbegin.c, config/i386/cygming-crtend.c:
1948         Move to ../libgcc/config/i386.
1949         * config/i386/t-crtstuff: Remove.
1950         * config/i386/t-i386elf: Remove.
1951         * config/i386/t-linux64 (EXTRA_MULTILIB_PARTS): Remove.
1952         * config/i386/t-nto (CRTSTUFF_T_CFLAGS, EXTRA_PARTS): Remove.
1953         * config/ia64/crtbegin.asm: Move to ../libgcc/config/ia64/crtbegin.S.
1954         * config/ia64/crtend.asm: Move to ../libgcc/config/ia64/crtend.S.
1955         * config/ia64/crti.asm: Move to ../libgcc/config/ia64/crti.S.
1956         * config/ia64/crtn.asm: Move to ../libgcc/config/ia64/crtn.S.
1957         * config/ia64/t-vms: Remove.
1958         * config/ia64/vms-crtinit.asm: Move to
1959         ../libgcc/config/ia64/vms-crtinit.S.
1960         * config/m32c/t-m32c (EXTRA_MULTILIB_PARTS): Remove.
1961         * config/m32r/initfini.c: Move to ../libgcc/config/m32r.
1962         * config/m32r/t-linux (CRTSTUFF_T_CFLAGS_S): Remove.
1963         * config/m32r/t-m32r (CRTSTUFF_T_CFLAGS): Remove.
1964         ($(T)crtinit.o, $(T)crtfini.o): Remove.
1965         (m32rx, m32r2): Remove.
1966         (EXTRA_MULTILIB_PARTS): Remove.
1967         * config/m68k/crti.s: Move to ../libgcc/config/m68k/crti.S.
1968         * config/m68k/crtn.s: Move to ../libgcc/config/m68k/crtn.S.
1969         * config/m68k/t-crtstuff: Remove.
1970         * config/m68k/t-linux (EXTRA_MULTILIB_PARTS): Remove.
1971         * config/m68k/t-m68kelf: Remove.
1972         * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Remove.
1973         * config/mcore/crti.asm: Move to ../libgcc/config/mcore/crti.S.
1974         * config/mcore/crtn.asm: Move to ../libgcc/config/mcore/crtn.S.
1975         * config/mcore/t-mcore ($(T)crti.o, $(T)crtn.o): Remove.
1976         (EXTRA_PARTS, EXTRA_MULTILIB_PARTS): Remove.
1977         * config/mep/t-mep (CRTSTUFF_CFLAGS): Remove.
1978         (EXTRA_MULTILIB_PARTS): Remove.
1979         * config/microblaze/crti.s: Move to ../libgcc/config/microblaze/crti.S.
1980         * config/microblaze/crtn.s: Move to ../libgcc/config/microblaze/crtn.S.
1981         * config/microblaze/t-microblaze (EXTRA_MULTILIB_PARTS,
1982         EXTRA_PARTS): Remove.
1983         ($(T)crti$(objext), $(T)crtn$(objext)): Remove.
1984         * config/mips/crti.asm: Move to ../libgcc/config/mips/crti.S.
1985         * config/mips/crtn.asm: Move to ../libgcc/config/mips/crtn.S.
1986         * config/mips/t-elf (CRTSTUFF_T_CFLAGS): Remove.
1987         ($(T)crti.o, $(T)crtn.o): Remove.
1988         (EXTRA_MULTILIB_PARTS): Remove.
1989         * config/mips/t-isa3264: Likewise.
1990         * config/mips/t-linux64 (EXTRA_MULTILIB_PARTS): Remove.
1991         * config/mips/t-r3900 (EXTRA_MULTILIB_PARTS): Remove.
1992         (CRTSTUFF_T_CFLAGS): Remove.
1993         * config/mips/t-sde (CRTSTUFF_T_CFLAGS): Remove.
1994         ($(T)crti.o, $(T)crtn.o): Remove.
1995         (EXTRA_MULTILIB_PARTS): Remove.
1996         * config/mips/t-sr71k (EXTRA_MULTILIB_PARTS, CRTSTUFF_T_CFLAGS):
1997         Remove.
1998         ($(T)crti.o, $(T)crtn.o): Remove.
1999         * config/mips/t-st (EXTRA_MULTILIB_PARTS): Remove.
2000         * config/mips/t-vr (CRTSTUFF_T_CFLAGS): Remove.
2001         (EXTRA_MULTILIB_PARTS): Remove.
2002         ($(T)crti.o, $(T)crtn.o): Remove.
2003         * config/mmix/crti.asm: Move to ../libgcc/config/crti.S.
2004         * config/mmix/crtn.asm: Move to ../libgcc/config/crtn.S.
2005         * config/mmix/t-mmix (CRTSTUFF_T_CFLAGS): Remove.
2006         * config/moxie/crti.asm, config/moxie/crtn.asm: Remove.
2007         * config/pa/stublib.c: Move to libgcc/config/pa.
2008         * config/pa/t-linux (CRTSTUFF_T_CFLAGS_S): Remove.
2009         * config/pa/t-linux64 (CRTSTUFF_T_CFLAGS_S): Remove.
2010         * config/pa/t-pa-hpux11 (LIBGCCSTUB_OBJS, stublib.c): Remove.
2011         (pthread_default_stacksize_np-stub.o, pthread_mutex_lock-stub.o)
2012         (pthread_mutex_unlock-stub.o, pthread_once-stub.o)
2013         ($(T)libgcc_stub.a): Remove.
2014         * config/pa/t-pa64 (LIBGCCSTUB_OBJS, stublib.c): Remove.
2015         (rfi-stub.o, dfi-stub.o, cxaf-stub.o, jvrc-stub.o)
2016         (pthread_default_stacksize_np-stub.o, pthread_mutex_lock-stub.o)
2017         (pthread_mutex_unlock-stub.o, pthread_once-stub.o)
2018         ($(T)libgcc_stub.a): Remove.
2019         * config/rs6000/eabi-cn.asm: Move to
2020         ../../../libgcc/config/rs6000/eabi-cn.S.
2021         * config/rs6000/eabi-ci.asm: Move to
2022         ../../../libgcc/config/rs6000/eabi-ci.S.
2023         * config/rs6000/sol-ci.asm: Move to
2024         ../../../libgcc/config/rs6000/sol-ci.S.
2025         * config/rs6000/sol-cn.asm: Move to
2026         ../../../libgcc/config/rs6000/sol-cn.S.
2027         * config/rs6000/t-lynx (EXTRA_MULTILIB_PARTS): Remove.
2028         (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S): Remove.
2029         * config/rs6000/t-netbsd (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S):
2030         Remove.
2031         (EXTRA_MULTILIB_PARTS): Remove.
2032         * config/rs6000/t-ppccomm (EXTRA_MULTILIB_PARTS): Remove.
2033         (ecrti.S, ecrtn.S, ncrti.S, ncrtn.S): Remove.
2034         ($(T)ecrti$(objext), $(T)ecrtn$(objext), $(T)ncrti$(objext),
2035         ($(T)ncrtn$(objext)): Remove.
2036         (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S): Remove.
2037         * config/rs6000/t-vxworks (EXTRA_MULTILIB_PARTS): Remove.
2038         * config/rx/t-rx (EXTRA_MULTILIB_PARTS): Remove.
2039         * config/score/crti.asm: Move to ../libgcc/config/score/crti.S.
2040         * config/score/crtn.asm: Move to ../libgcc/config/score/crtn.S.
2041         * config/score/t-score-elf: Remove.
2042         * config/sh/crt1.asm: Move to ../libgcc/config/sh/crt1.S.
2043         * config/sh/crti.asm: Move to ../libgcc/config/sh/crti.S.
2044         * config/sh/crtn.asm: Move to ../libgcc/config/sh/crtn.S.
2045         * config/sh/lib1funcs-4-300.asm: Move to
2046         ../../../libgcc/config/sh/lib1funcs-4-300.S.
2047         * config/sh/lib1funcs-Os-4-200.asm: Move to
2048         ../libgcc/config/sh/lib1funcs-Os-4-200.S.
2049         * config/sh/t-elf: Remove.
2050         * config/sh/t-linux (EXTRA_MULTILIB_PARTS): Remove.
2051         * config/sh/t-linux64: Remove.
2052         * config/sh/t-netbsd (EXTRA_MULTILIB_PARTS): Remove.
2053         * config/sh/t-sh ($(T)crt1.o, $(T)crti.o, $(T)crtn.o): Remove.
2054         (IC_EXTRA_PARTS, OPT_EXTRA_PARTS, EXTRA_MULTILIB_PARTS): Remove.
2055         ($(T)ic_invalidate_array_4-100.o)
2056         ($(T)libic_invalidate_array_4-100.a)
2057         ($(T)ic_invalidate_array_4-200.o)
2058         ($(T)libic_invalidate_array_4-200.a, $(T)ic_invalidate_array_4a.o)
2059         ($(T)libic_invalidate_array_4a.a, $(T)sdivsi3_i4i-Os-4-200.o)
2060         ($(T)udivsi3_i4i-Os-4-200.o, $(T)unwind-dw2-Os-4-200.o)
2061         ($(T)libgcc-Os-4-200.a, $(T)div_table-4-300.o)
2062         ($(T)libgcc-4-300.a): Remove.
2063         * config/sh/t-superh: Remove.
2064         * config/sh/t-vxworks (EXTRA_MULTILIB_PARTS): Remove.
2065         * config/sparc/t-linux64 (CRTSTUFF_T_CFLAGS): Remove.
2066         * config/spu/cache.S: Move to ../libgcc/config/spu.
2067         * config/spu/cachemgr.c: Move to ../libgcc/config/spu.
2068         * config/spu/t-spu-elf (CRTSTUFF_T_CFLAGS): Remove.
2069         (EXTRA_MULTILIB_PARTS): Remove.
2070         ($(T)cachemgr.o, $(T)cachemgr_nonatomic.o, $(T)libgcc_%.a): Remove.
2071         ($(T)cache8k.o, $(T)cache16k.o, $(T)cache32k.o, $(T)cache32k.o)
2072         ($(T)cache64k.o, $(T)cache128k.o): Remove.
2073         * config/t-freebsd (CRTSTUFF_T_CFLAGS_S): Remove.
2074         * config/t-libc-ok: Remove.
2075         * config/t-linux (CRTSTUFF_T_CFLAGS_S): Remove.
2076         * config/t-lynx (CRTSTUFF_T_CFLAGS_S): Remove.
2077         * config/t-netbsd: Remove.
2078         * config/t-svr4 (CRTSTUFF_T_CFLAGS_S): Remove.
2079         * config/t-vxworks (EXTRA_MULTILIB_PARTS): Remove.
2080         * config/vms/t-vms (VMS_EXTRA_PARTS): Remove.
2081         ($(T)vcrt0.o, $(T)pcrt0.o): Remove.
2082         * config/vms/vms-ucrt0.c: Move to ../libgcc/config/vms.
2083         * config/xtensa/crti.asm: Move to ../libgcc/config/xtensa/crti.S.
2084         * config/xtensa/crtn.asm: Move to ../libgcc/config/xtensa/crtn.S.
2085         * config/xtensa/t-elf (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S): Remove.
2086         (EXTRA_MULTILIB_PARTS): Remove.
2087         * config/xtensa/t-linux: Remove.
2088         * config/xtensa/t-xtensa ($(T)crti.o, $(T)crtn.o): Remove.
2089
2090 2011-11-02  Uros Bizjak  <ubizjak@gmail.com>
2091
2092         * config/i386/i386.c (bdesc_args) [IX86_BUILTIN_CVTTPD2DQ256]: Use
2093         CODE_FOR_fix_truncv4dfv4si2, not CODE_FOR_fix_truncv4sfv4si2.
2094
2095 2011-11-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2096
2097         PR translation/45116
2098         * Makefile.in (slibdir): Remove, don't export.
2099         (SHLIB_NM_FLAGS): Remove.
2100         (libgcc.mvars): Don't emit SHLIB_LINK, SHLIB_INSTALL,
2101         SHLIB_DLLDIR, SHLIB_EXT, SHLIB_MKMAP, SHLIB_MKMAP_OPTS,
2102         SHLIB_MAPFILES, SHLIB_NM_FLAGS.
2103         (DRIVER_DEFINES): Test SHLIB instead of SHLIB_LINK.
2104         (gcc.o): Pass SHLIB instead of SHLIB_LINK.
2105         (gccspec.o): Likewise.
2106         (installdirs): Don't create $(DESTDIR)$(slibdir).
2107         * configure.ac (slibdir): Remove.
2108         * configure: Regenerate.
2109         * libgcc-libsystem.ver: Move to ../libgcc/config.
2110         * mkmap-flat.awk, mkmap-symver.awk: Move to ../libgcc.
2111         * config/libgcc-glibc.ver: Move to ../libgcc/config.
2112         * config/t-libunwind (SHLIB_LC): Remove.
2113         * config/t-linux (SHLIB_MAPFILES): Remove.
2114         * config/t-slibgcc-dummy: Rename to config/t-slibgcc.
2115         * config/t-slibgcc-elf-ver: Remove.
2116         * config/t-slibgcc-libgcc, config/t-slibgcc-nolc-override: Move to
2117         ../libgcc/config.
2118         * config/alpha/libgcc-alpha-ldbl.ver, config/alpha/t-linux: Move
2119         to ../libgcc/config/alpha.
2120         * config/alpha/t-vms (shlib_version, SHLIB_EXT, SHLIB_OBJS,
2121         SHLIB_NAME, SHLIB_MULTILIB, SHLIB_INSTALL, SHLIB_SYMVEC,
2122         SHLIB_SYMVECX2, SHLIB_LINK): Remove.
2123         * config/arm/libgcc-bpabi.ver: Move to ../libgcc/config/arm.
2124         * config/arm/t-bpabi (SHLIB_MAPFILES): Remove.
2125         * config/arm/t-netbsd (SHLIB_EXT, SHLIB_NAME, SHLIB_SONAME,
2126         SHLIB_OBJS, SHLIB_LINK, SHLIB_INSTALL): Remove.
2127         * config/arm/t-symbian (SHLIB_LC): Remove.
2128         * config/bfin/libgcc-bfin.ver: Move to
2129         ../libgcc/config/bfin/libgcc-glibc.ver.
2130         * config/bfin/t-bfin-linux (SHLIB_MAPFILES): Remove.
2131         * config/c6x/libgcc-c6xeabi.ver: Move to
2132         ../libgcc/config/c6x/libgcc-eabi.ver.
2133         * config/c6x/t-c6x-elf (SHLIB_MAPFILES): Remove.
2134         * config/cris/libgcc.ver: Move to
2135         ../libgcc/config/cris/libgcc-glibc.ver.
2136         * config/cris/t-linux (SHLIB_MAPFILES): Remove.
2137         * config/frv/libgcc-frv.ver: Move to ../libgcc/config/frv.
2138         * config/frv/t-linux (SHLIB_MAPFILES): Remove.
2139         * config/i386/darwin-libgcc.10.4.ver: Move to
2140         ../libgcc/config/i386/libgcc-darwin.10.4.ver.
2141         * config/i386/darwin-libgcc.10.5.ver: Move to
2142         ../libgcc/config/i386/libgcc-darwin.10.5.ver.
2143         * config/i386/libgcc-glibc.ver: Move to ../libgcc/config/i386.
2144         * config/i386/t-cygming (SHLIB_EXT, SHLIB_IMPLIB, SHLIB_SOVERSION,
2145         SHLIB_SONAME, SHLIB_MAP, SHLIB_OBJS, SHLIB_DIR, SHLIB_SLIBDIR_QUAL)
2146         SHLIB_PTHREAD_CFLAG, SHLIB_PTHREAD_LDFLAG, SHLIB_LINK,
2147         SHLIB_INSTALL, SHLIB_MKMAP, SHLIB_MKMAP_OPTS, SHLIB_MAPFILES): Remove.
2148         * config/i386/t-cygwin (SHLIB_LC, SHLIB_EH_EXTENSION,
2149         SHLIB_IMPLIB, SHLIB_SONAME, SHLIB_MKMAP_OPTS): Remove.
2150         * config/i386/t-dlldir, config/i386/t-dlldir-x: Move to
2151         ../libgcc/config/i386.
2152         * config/i386/t-dw2-eh, config/i386/t-sjlj-eh: Move to
2153         ../libgcc/config/i386.
2154         * config/i386/t-linux: Move to ../libgcc/config/i386.
2155         * config/i386/t-mingw-pthread: Move to ../libgcc/config/i386.
2156         * config/i386/t-mingw-w32 (SHLIB_LC): Remove.
2157         * config/i386/t-mingw-w64: Likewise.
2158         * config/i386/t-mingw32: Remove.
2159         * config/ia64/libgcc-glibc.ver, config/ia64/libgcc-ia64.ver: Move
2160         to ../libgcc/config/ia64.
2161         * config/ia64/t-glibc: Remove.
2162         * config/ia64/t-hpux (SHLIB_EXT, SHLIB_LINK, SHLIB_INSTALL): Remove.
2163         * config/ia64/t-ia64 (SHLIB_MAPFILES): Remove.
2164         * config/ia64/t-vms (shlib_version, SHLIB_EXT, SHLIB_OBJS,
2165         SHLIB_NAME, SHLIB_MULTILIB, SHLIB_INSTALL, SHLIB_LINK): Remove.
2166         * config/ia64/vms_symvec_libgcc_s.opt: Remove.
2167         * config/m32r/libgcc-glibc.ver: Move to ../libgcc/config/m32r.
2168         * config/m32r/t-linux (SHLIB_MAPFILES): Remove.
2169         * config/m68k/t-slibgcc-elf-ver: Move to ../libgcc/config/m68k.
2170         * config/mips/t-libgcc-mips16 (SHLIB_MAPFILES): Remove.
2171         * config/pa/t-hpux-shlib: Move to ../libgcc/config/pa/t-slibgcc-hpux.
2172         * config/pa/t-slibgcc-dwarf-ver, config/pa/t-slibgcc-sjsj-ver:
2173         Move to ../libgcc/config/pa.
2174         * config/rs6000/darwin-libgcc.10.4.ver: Move to
2175         ../libgcc/config/rs6000/libgcc-darwin.10.4.ver.
2176         * config/rs6000/darwin-libgcc.10.5.ver: Move to
2177         ../libgcc/config/rs6000/libgcc-darwin.10.5.ver.
2178         * config/rs6000/t-aix43 (SHLIB_EXT, SHLIB_LINK, SHLIB_INSTALL,
2179         SHLIB_LIBS, SHLIB_MKMAP, SHLIB_NM_FLAGS, AR_FLAGS_FOR_TARGET): Remove.
2180         * config/rs6000/t-aix52: Likewise.
2181         * config/sh/libgcc-excl.ver, config/sh/libgcc-glibc.ver: Move to
2182         ../libgcc/config/sh.
2183         * config/sparc/libgcc-sparc-glibc.ver: Move to
2184         ../libgcc/config/sparc/libgcc-glibc.ver.
2185         * config/sparc/t-linux: Move to ../libgcc/config/sparc.
2186         * config/xtensa/t-linux (SHLIB_MAPFILES): Remove.
2187         * config/xtensa/libgcc-xtensa.ver: Move to
2188         ../libgcc/config/xtensa/libgcc-glibc.ver.
2189         * config.gcc (*-*-freebsd*): Replace t-slibgcc-elf-ver with
2190         t-slibgcc in tmake_file.
2191         Remove t-slibgcc-nolc-override for *-*-freebsd[34],
2192         *-*-freebsd[34].* with pthreads.
2193         (*-*-linux*, frv-*-*linux*, *-*-kfreebsd*-gnu,
2194         *-*-knetbsd*-gnu, *-*-gnu*, *-*-kopensolaris*-gnu): Replace
2195         t-slibgcc-elf-ver with t-slibgcc in tmake_file.
2196         (*-*-netbsd*): Likewise.
2197         (*-*-solaris2*): Replace t-slibgcc-dummy with t-slibgcc in tmake_file.
2198         (*-*-*vms*): Add t-slibgcc to tmake_file.
2199         (alpha*-*-linux*): Remove alpha/t-linux from tmake_file.
2200         (alpha*-dec-osf5.1*): Replace t-slibgcc-dummy with t-slibgcc in
2201         tmake_file.
2202         (arm*-*-linux*): Remove t-slibgcc-libgcc from tmake_file for
2203         arm*-*-linux-*eabi.
2204         (bfin*-linux-uclibc*): Replace t-slibgcc-dummy with t-slibgcc in
2205         tmake_file.
2206         (crisv32-*-linux*, cris-*-linux*): Likewise.
2207         (hppa*-*-linux*): Remove t-slibgcc-libgcc, pa/t-slibgcc-sjlj-ver,
2208         pa/t-slibgcc-dwarf-ver from tmake_file.
2209         (hppa[12]*-*-hpux10*): Replace pa/t-hpux-shlib with t-slibgcc in
2210         tmake_file.
2211         Remove pa/t-slibgcc-sjlj-ver, pa/t-slibgcc-dwarf-ver from tmake_file.
2212         (hppa*64*-*-hpux11*): Likewise.
2213         (hppa[12]*-*-hpux11*): Likewise.
2214         (i[34567]86-*-darwin*): Replace t-slibgcc-dummy in t-slibgcc in
2215         tmake_file.
2216         (x86_64-*-darwin*): Likewise.
2217         (i[34567]86-*-cygwin*): Remove tmake_eh_file, tmake_dlldir_file.
2218         Add t-slibgcc to tmake_file.
2219         (i[34567]86-*-mingw*, x86_64-*-mingw*): Likewise.
2220         Remove i386/t-mingw32 from tmake_file unless x86_64-w64-*,
2221         i[34567]86-w64-*.
2222         Remove i386/t-mingw-pthread from tmake_file.
2223         (ia64*-*-linux*): Remove ia64/t-glibc from tmake_file.
2224         (ia64*-*-hpux*): Add t-slibgcc to tmake_file.
2225         (ia64-hp-*vms*): Likewise.
2226         (m32r-*-linux*): Replace t-slibgcc-elf-ver with t-slibgcc in
2227         tmake_file.
2228         (m32rle-*-linux*): Likewise.
2229         (m68k-*-linux*): Remove m68k/t-slibgcc-elf-ver from tmake_file.
2230         (microblaze*-linux*): Remove t-slibgcc-elf-ver,
2231         t-slibgcc-nolc-override from tmake_file.
2232         (mips-sgi-irix6.5*): Replace t-slibgcc-dummy with t-slibgcc in
2233         tmake_file.
2234         (powerpc-*-darwin*): Likewise.
2235         (powerpc64-*-darwin*): Likewise.
2236         (powerpc-*-freebsd*): Remove t-slibgcc-libgcc from tmake_file.
2237         (powerpc-*-linux*, powerpc64-*-linux*): Likewise.
2238         (rs6000-ibm-aix4.[3456789]*, powerpc-ibm-aix4.[3456789]*): Add
2239         t-slibgcc to tmake_file.
2240         (rs6000-ibm-aix5.1.*, powerpc-ibm-aix5.1.*): Likewise.
2241         (rs6000-ibm-aix5.2.*, powerpc-ibm-aix5.2.*): Likewise.
2242         (rs6000-ibm-aix5.3.*, powerpc-ibm-aix5.3.*): Likewise.
2243         (rs6000-ibm-aix[6789].*, powerpc-ibm-aix[6789].*): Likewise.
2244         (sparc-*-linux*): Remove sparc/t-linux from tmake_file.
2245         (sparc64-*-linux*): Likewise.
2246         (tic6x-*-uclinux): Replace t-slibgcc-elf-ver with t-slibgcc in
2247         tmake_file.
2248         (i[34567]86-*-linux*, x86_64-*-linux*, i[34567]86-*-kfreebsd*-gnu,
2249         x86_64-*-kfreebsd*-gnu, i[34567]86-*-gnu*): Remove i386/t-linux
2250         from tmake_file.
2251
2252 2011-11-02  Richard Guenther  <rguenther@suse.de>
2253
2254         PR tree-optimization/50902
2255         * tree-vect-stmts.c (vectorizable_load): Properly convert
2256         an invariant initializer element.
2257
2258 2010-11-02  Richard Guenther  <rguenther@suse.de>
2259
2260         PR tree-optimization/50890
2261         * gimple.h (gimple_fold_call): Remove.
2262         * gimple-fold.c (fold_stmt_1): Move all call related code to ...
2263         (gimple_fold_call): ... here.  Make static.  Update the
2264         cannot-inline flag on direct calls.
2265         * ipa-inline.c (early_inliner): Copy the cannot-inline flag
2266         from the statements to the edges.
2267
2268 2011-11-01  Ian Lance Taylor  <iant@google.com>
2269
2270         * godump.c (struct macro_hash_value): Define.
2271         (macro_hash_hashval): New static function.
2272         (macro_hash_eq, macro_hash_del): New static functions.
2273         (go_define): Use macro_hash_value to store values in macro_hash.
2274         Replace an old value on a redefinition.  Don't print anything to
2275         go_dump_file.
2276         (go_undef): Delete the entry from the hash table.
2277         (go_output_typedef): For an enum, use macro_hash_value, and don't
2278         print anything to go_dump_file.
2279         (go_print_macro): New static function.
2280         (go_finish): Traverse macro_hash with go_print_macro.
2281         (dump_go_spec_init): Update macro_hash creation for macro_hash_value.
2282
2283 2011-11-02  Alan Modra  <amodra@gmail.com>
2284
2285         * config/rs6000/rs6000.c (rs6000_code_end): Declare ATTRIBUTE_UNUSED.
2286
2287 2011-11-01  Paolo Carlini  <paolo.carlini@oracle.com>
2288
2289         PR c++/44277
2290         * doc/invoke.texi: Document -Wzero-as-null-pointer-constant.
2291
2292 2011-11-01  Andrew Stubbs  <ams@codesourcery.com>
2293
2294         * config/arm/bpabi.h (BE8_LINK_SPEC): Recognize generic-armv7 tuning.
2295
2296 2011-11-01  Uros Bizjak  <ubizjak@gmail.com>
2297
2298         * config/i386/i386.md (splitters for int-float conversion): Use
2299         SUBREG_REG on SUBREGs in splitter constraints.
2300
2301 2011-11-01  Jakub Jelinek  <jakub@redhat.com>
2302
2303         * config/i386/i386-protos.h (ix86_expand_adjust_ufix_to_sfix_si): New
2304         prototype.
2305         * config/i386/i386.c (ix86_expand_adjust_ufix_to_sfix_si): New
2306         function.
2307         * config/i386/sse.md (fixuns_trunc<mode><sseintvecmodelower>2): Use it.
2308         (ssepackfltmode): New mode attr.
2309         (vec_pack_ufix_trunc_<mode>): New expander.
2310
2311 2011-11-01  Uros Bizjak  <ubizjak@gmail.com>
2312
2313         PR target/50940
2314         * config/i386/i386.md (floatsi<mode>2_vector_sse_with_temp splitter):
2315         Compare <ssevecmode>mode to V4SFmode, not V4SImode.
2316
2317 2011-11-01  Peter Bergner  <bergner@vnet.ibm.com>
2318
2319         * config.gcc (powerpc*-*-linux*): Add powerpc*-*-linux*ppc476* variant.
2320         * config/rs6000/476.h: New file.
2321         * config/rs6000/476.opt: Likewise.
2322         * config/rs6000/rs6000.h (TARGET_LINK_STACK): New define.
2323         (SET_TARGET_LINK_STACK): Likewise.
2324         (TARGET_ASM_CODE_END): Define.
2325         * config/rs6000/rs6000.c (rs6000_option_override_internal): Enable
2326         TARGET_LINK_STACK for -mtune=476 and -mtune=476fp.
2327         (rs6000_legitimize_tls_address): Emit the link stack preserving GOT
2328         code if TARGET_LINK_STACK.
2329         (rs6000_emit_load_toc_table): Likewise.
2330         (output_function_profiler): Likewise
2331         (macho_branch_islands): Likewise
2332         (machopic_output_stub): Likewise
2333         (get_ppc476_thunk_name): New function.
2334         (rs6000_code_end): Likewise.
2335         * config/rs6000/rs6000.md (load_toc_v4_PIC_1, load_toc_v4_PIC_1b):
2336         Convert to a define_expand.
2337         (load_toc_v4_PIC_1_normal): New define_insn.
2338         (load_toc_v4_PIC_1_476): Likewise.
2339         (load_toc_v4_PIC_1b_normal): Likewise.
2340         (load_toc_v4_PIC_1b_476): Likewise.
2341
2342 2011-11-01  Georg-Johann Lay  <avr@gjlay.de>
2343
2344         PR target/50910
2345         * config/avr/avr.opt (-mbranch-cost=): New option.
2346         * config/avr/avr.h (BRANCH_COST): Define to avr_branch_cost.
2347         * config/avr/avr.c (avr_rtx_costs_1): Adjust [U]DIV/[U]MOD costs.
2348         * config/avr/avr.md (*addqi3.lt0, *addhi3.lt0, *addsi3.lt0): New insns.
2349         (*addhi3_zero_extend1): Remove % in constraint of operand 1.
2350         (*addhi3.sign_extend1, *subhi3.sign_extend2): New insns.
2351
2352 2011-11-01  Tom de Vries  <tom@codesourcery.com>
2353
2354         PR tree-optimization/50908
2355         * tree-ssa-tail-merge.c (update_vuses): Now that edges are removed
2356         before update_vuses, test for 1 predecessor rather than two.
2357         (delete_block_update_dominator_info): New function, part of it factored
2358         out of ...
2359         (replace_block_by): Use delete_block_update_dominator_info.  Call
2360         update_vuses after deleting bb1 and updating dominator info, instead of
2361         before.
2362
2363 2011-11-01  David S. Miller  <davem@davemloft.net>
2364
2365         * config/sparc/sparc.c (vector_init_faligndata): New function.
2366         (sparc_expand_vector_init): Use it for V4HImode on VIS1.
2367
2368         * config/sparc/sparc.c (sparc_expand_vcond): New function.
2369         * config/sparc/sparc-protos.h (sparc_expand_vcond): Declare it.
2370         * config/sparc/sparc.md (vcond<mode><mode>): New VIS3 expander.
2371         (vconduv8qiv8qi): Likewise.
2372
2373 2011-11-01  Alexandre Oliva  <aoliva@redhat.com>
2374
2375         PR debug/50869
2376         * cselib.c (cfa_base_preserved_regno): Initialize.
2377         (cselib_expand_value_rtx_1): Don't expand it.
2378         * var-tracking.c (vt_expand_var_loc_chain): Initialize depth.
2379         Check it's only zero if result is NULL.
2380
2381 2011-11-01  Jakub Jelinek  <jakub@redhat.com>
2382
2383         * config/i386/sse.md (fixuns_trunc<mode><sseintvecmodelower>2): New
2384         expander.
2385
2386         * config/i386/sse.md (sseintvecmode): Remove duplicate modes.
2387         (sseintvecmodelower): New mode iterator.
2388         (floatv8siv8sf2, floatunsv4siv4sf2): Macroize into...
2389         (float<sseintvecmodelower><mode>2): ... this using VF1 iterator.
2390         (floatunsv4siv4sf2): Macroize into...
2391         (floatuns<sseintvecmodelower><mode>2): ... this using VF1 iterator.
2392
2393 2011-10-31  David S. Miller  <davem@davemloft.net>
2394
2395         * config/sparc/sparc.md (cmask patterns): Allow zero operand.
2396
2397         * dwarf2out.c (cached_next_real_insn): New.
2398         (dwarf2out_end_epilogue): Set it to NULL_RTX.
2399         (dwarf2out_var_location): Remove cached_next_real_insn local static.
2400
2401 2011-10-31  Richard Henderson  <rth@redhat.com>
2402
2403         * config/i386/sse.md (floatv8siv8sf2): Rename from avx_cvtdq2ps256.
2404         (floatv4siv4sf2): Rename from sse2_cvtdq2ps.
2405         (floatunsv4siv4sf2): Rename from sse2_cvtudq2ps.
2406         (fix_truncv8sfv8si2): Rename from avx_cvttps2dq256.
2407         (fix_truncv4sfv4si2): Rename from sse2_cvttps2dq.
2408         (floatv4siv4df2): Rename from avx_cvtdq2pd256.
2409         (fix_truncv4dfv4si2): Rename from avx_cvttpd2dq256.
2410         (vec_unpacku_float_hi_v8si): Update for insn pattern name changes.
2411         * config/i386/i386.md (splitters for int-float conversion): Likewise.
2412         * config/i386/i386.c (ix86_split_convert_uns_si_sse): Likewise.
2413         (bdesc_args): Likewise.
2414         (enum ix86_builtins) [IX86_BUILTIN_CVTUDQ2PS]: Remove.
2415         (ix86_vectorize_builtin_conversion): Remove.
2416         (TARGET_VECTORIZE_BUILTIN_CONVERSION): Remove.
2417
2418 2011-10-31  Martin Jambor  <mjambor@suse.cz>
2419
2420         * ipa-prop.c (mark_modified): Moved up in the file.
2421         (is_parm_modified_before_call): Renamed to
2422         is_parm_modified_before_stmt, moved up in the file.
2423         (load_from_unmodified_param): New function.
2424         (compute_complex_assign_jump_func): Also attempt to create pass
2425         through jump functions for values loaded from (addressable)
2426         parameters.
2427
2428 2011-10-31  Jakub Jelinek  <jakub@redhat.com>
2429
2430         * tree-vect-stmts.c (vectorizable_shift): If op1 is vect_external_def
2431         in a loop and has different type from op0, cast it to op0's type
2432         before the loop first.  For slp give up.  Don't crash if op1_vectype
2433         is NULL.
2434
2435 2011-10-31  Paul Brook  <paul@codesourcery.com>
2436
2437         * cgraphunit.c: Don't mark clones as static constructors.
2438
2439 2011-10-31  David Edelsohn  <dje.gcc@gmail.com>
2440
2441         * gcc-ar: Do not include stdio.h.
2442
2443 2011-10-31  Diego Novillo  <dnovillo@google.com>
2444
2445         * tree-streamer-out.c (pack_ts_base_value_fields): Emit
2446         TYPE_ADDR_SPACE.
2447         * tree-streamer-in.c (unpack_ts_base_value_fields): Read
2448         TYPE_ADDR_SPACE.
2449
2450 2011-10-30  David S. Miller  <davem@davemloft.net>
2451
2452         * config/sparc/sparc.c (vector_init_bshuffle): New function.
2453         (vector_init_fpmerge): New function.
2454         (sparc_expand_vector_init): Use them to improve non-const cases.
2455
2456         * dwarf2out.c (dwarf2out_var_location): When processing several
2457         consecutive location notes, cache the result of next_real_insn().
2458
2459 2011-10-30  Uros Bizjak  <ubizjak@gmail.com>
2460
2461         * config/i386/i386.md (avx2_vec_dup<mode>): Macroize insn from
2462         avx2_vec_dup{v8sf,v4sf} using VF1 mode iterator.
2463         (vec_dupv4sf): Remove expander.
2464         (vec_dupv4sf): Merge from *vec_dupv4sf and *vec_dupv4sf_avx.
2465         (vec_dupv2df): Remove expander.
2466         (vec_dupv2df): Merge from *vec_dupv2df and *vec_dupv2df_sse3.
2467         (*vec_concatv2df): Merge *vec_concatv2df_sse3.
2468         (*vec_dupv4si): Merge *vec_dupv4si_avx.
2469         (*vec_dupv2di): Merge *vec_dupv2di_sse3.
2470
2471 2011-10-30  Dmitry Plotnikov  <dplotnikov@ispras.ru>
2472
2473         * tree-cfg.c (verify_gimple_assign_unary): Allow vector conversions.
2474         * optabs.c (supportable_convert_operation): New function.
2475         * optabs.h (supportable_convert_operation): New prototype.
2476         * tree-vect-stmts.c (vectorizable_conversion): Change condition and
2477         behavior for NONE modifier case.
2478         * tree.h (VECTOR_INTEGER_TYPE_P): New macro.
2479
2480 2011-10-30  Jakub Jelinek  <jakub@redhat.com>
2481
2482         * tree-vectorizer.h (NUM_PATTERNS): Bump to 9.
2483         * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern): New
2484         function.
2485         (vect_vect_recog_func_ptrs): Add it.
2486
2487 2011-10-30  David S. Miller  <davem@davemloft.net>
2488
2489         * reorg.c (label_before_next_insn): New function.
2490         (relax_delay_slots): Use it instead of prev_label.
2491         * rtl.h (prev_label): Delete declaration.
2492         * emit-rtl.c (prev_label): Remove.
2493
2494 2011-10-30  Revital Eres  <revital.eres@linaro.org>
2495
2496         * modulo-sched.c (generate_prolog_epilog): Mark prolog and epilog
2497         as BB_DISABLE_SCHEDULE.
2498         (mark_loop_unsched): New function.
2499         (sms_schedule): Call it.
2500
2501 2011-10-29  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
2502
2503         PR target/50617
2504         * config/pa/protos.h (attr_length_save_restore_dltp): Delete.
2505         (cmpib_comparison_operator): Likewise.
2506         (following_cal, output_and, output_ior, output_move_double,
2507         output_fp_move_double, output_block_move, output_block_clear,
2508         output_cbranch, output_lbranch, output_bb, output_bvb, output_dbra,
2509         output_movb, output_parallel_movb, output_parallel_addb, output_call,
2510         output_indirect_call, output_millicode_call, output_mul_insn,
2511         output_div_insn, output_mod_insn, singlemove_string,
2512         output_arg_descriptor, output_global_address, print_operand,
2513         legitimize_pic_address, hppa_encode_label, symbolic_expression_p,
2514         fmpyaddoperands, fmpysuboperands, emit_bcond_fp, emit_move_sequence,
2515         emit_hpdiv_const, is_function_label_plus_const, jump_in_call_delay,
2516         hppa_fpstore_bypass_p, attr_length_millicode_call, attr_length_call,
2517         attr_length_indirect_call, return_addr_rtx, function_arg_padding,
2518         insn_refs_are_delayed, get_deferred_plabel, ldil_cint_p, zdepi_cint_p,
2519         output_ascii, compute_frame_size, and_mask_p, cint_ok_for_move,
2520         hppa_expand_prologue, hppa_expand_epilogue, ior_mask_p,
2521         compute_zdepdi_operands, output_64bit_and, output_64bit_ior,
2522         reloc_needed, magic_milli, shadd_constant_p): Consistently prefix
2523         exported functions and variables with "pa_".
2524         * config/pa/predicates.md: Likewise.
2525         * config/pa/pa64-hpux.h: likewise.
2526         * config/pa/som.h: Likewise.
2527         * config/pa/elf.h: Likewise.
2528         * config/pa/pa64-linux.h: Likewise.
2529         * config/pa/pa.md: Likewise.
2530         * config/pa/pa.c: Likewise.
2531         * config/pa/pa-linux.h: Likewise.
2532         * config/pa/pa.h: Likewise.
2533         * config/pa/constraints.md: Likewise.
2534
2535 2011-10-29  Uros Bizjak  <ubizjak@gmail.com>
2536
2537         * config/i386/i386.md (xop_sha<mode>3): Rename from xop_ashl<mode>3.
2538         Update all uses.
2539         (xop_shl<mode>3): Rename from xop_lshl<mode>3.  Update all uses.
2540         * config/i386/i386.c: Update all uses.
2541
2542 2011-10-29  Uros Bizjak  <ubizjak@gmail.com>
2543
2544         * config/i386/i386.md (lshlv16qi3): Remove expander.
2545         (lshrv16qi3): New expander.
2546         (<shift_insn>v16qi3): Macroize expander from ashrv16qi3 and lshrv16qi3
2547         using any_shiftrt code iterator. Cleanup.
2548         (ashlv16qi3): Cleanup.
2549         (ashrv2di3): Ditto.
2550
2551 2011-10-29  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
2552
2553         PR target/50691
2554         * config/pa/pa.c (emit_move_sequence): Legitimize TLS symbol
2555         references.
2556         (pa_legitimate_constant_p): Return false for TLS_MODEL_GLOBAL_DYNAMIC
2557         and TLS_MODEL_LOCAL_DYNAMIC symbol references.
2558
2559 2011-10-29  Georg-Johann Lay  <avr@gjlay.de>
2560
2561         PR target/50887
2562         * config/avr/avr.opt (-maccumulate-args): New option.
2563         * config/avr/avr.h (STARTING_FRAME_OFFSET): Redefine to
2564         avr_starting_frame_offset.
2565         (ACCUMULATE_OUTGOING_ARGS): Define to avr_accumulate_outgoing_args.
2566         * config/avr/avr.md (UNSPECV_WRITE_SP_IRQ_ON): Remove.
2567         (UNSPECV_WRITE_SP_IRQ_OFF): Remove.
2568         (UNSPECV_WRITE_SP): New constant.
2569         (*addhi3_sp_R): Rewrite to...
2570         (*addhi3_sp): ...this new insn.
2571         (movhi_sp_r_irq_off, movhi_sp_r_irq_on): Combine to...
2572         (movhi_sp_r): ...this new insn.
2573         * config/avr/avr-protos.h (avr_accumulate_outgoing_args): New.
2574         (avr_starting_frame_offset): New.
2575         * config/avr/avr.c (avr_accumulate_outgoing_args): New function.
2576         (avr_starting_frame_offset): New function.
2577         (avr_outgoing_args_size): New static function.
2578         (avr_initial_elimination_offset): Use it.
2579         (avr_simple_epilogue): Use it.
2580         (avr_asm_function_end_prologue): Use it.
2581         (expand_epilogue): Use it.
2582         (expand_prologue): Use it.  Break out code to...
2583         (avr_prologue_setup_frame): ...this new static function.
2584         (avr_can_eliminate): Allow eliminating to frame pointer if there
2585         is one.
2586         (avr_frame_pointer_required_p): Use frame pointer if target has a
2587         nonlocal label.
2588         * config/avr/constraints.md (R): Remove.
2589         (Csp): New constraint.
2590         * config/avr/predicates.md (avr_sp_immediate_operand): Use it.
2591
2592 2011-10-29  Andi Kleen  <ak@linux.intel.com>
2593
2594         * gcc-ar.c (target_machine): Add.
2595
2596 2011-10-29  Anatoly Sokolov  <aesok@post.ru>
2597
2598         * config/cris/cris.c (reg_ok_for_base_p, reg_ok_for_index_p,
2599         cris_constant_index_p, cris_base_p, cris_index_p,
2600         cris_base_or_autoincr_p, cris_bdap_index_p, cris_biap_index_p,
2601         cris_legitimate_address_p): New functions.
2602         (TARGET_LEGITIMATE_ADDRESS_P): Define.
2603         (cris_pic_symbol_type, cris_valid_pic_const): Change arguments type
2604         from rtx to const_rtx.
2605         (cris_print_operand_address, cris_address_cost,
2606         cris_side_effect_mode_ok):  Use
2607         cris_constant_index_p, cris_base_p, cris_base_or_autoincr_p,
2608         cris_biap_index_p and cris_bdap_index_p.
2609         * config/cris/cris.h (CONSTANT_INDEX_P, BASE_P, BASE_OR_AUTOINCR_P,
2610         BDAP_INDEX_P, BIAP_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2611         REG_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P): Remove.
2612         (EXTRA_CONSTRAINT_Q, EXTRA_CONSTRAINT_R, EXTRA_CONSTRAINT_T): Use
2613         cris_constant_index_p, cris_base_p, cris_base_or_autoincr_p,
2614         cris_biap_index_p and cris_bdap_index_p.
2615         * config/cris/cris.md (moversideqi movemsideqi peephole2): Use
2616         cris_base_p.
2617         * config/cris/cris-protos.h (cris_constant_index_p, cris_base_p,
2618         cris_base_or_autoincr_p, cris_bdap_index_p, cris_biap_index_p): New
2619         prototype.
2620         (cris_pic_symbol_type, cris_valid_pic_const): Update prototype.
2621
2622 2011-10-21  Andi Kleen  <ak@linux.intel.com>
2623
2624         * ggc-page (PAGE_ALIGN): Add.
2625         (alloc_page, ggc_pch_total_size, ggc_pch_this_base, ggc_pch_read):
2626         Replace ROUND_UP with PAGE_ALIGN.
2627
2628 2011-10-20  Andi Kleen  <ak@linux.intel.com>
2629
2630         * ggc-page (alloc_anon): Add check argument.
2631         (alloc_page): Add fallback to 1 page allocation.
2632         Adjust alloc_anon calls to new argument.
2633
2634 2011-10-18  Andi Kleen  <ak@linux.intel.com>
2635
2636         * ggc-page (release_pages): First free large continuous
2637         chunks in the madvise path.
2638
2639 2011-10-18  Andi Kleen  <ak@linux.intel.com>
2640
2641         * ggc-page.c (alloc_pages): Always round up entry_size.
2642
2643 2011-10-19  Andi Kleen  <ak@linux.intel.com>
2644
2645         * Makefile.in (MOSTLYCLEANFILES): Add gcc-ar/nm/ranlib.
2646         (native): Add gcc-ar, gcc-nm, gcc-ranlib.
2647         (AR_LIBS, gcc-ar, gcc-ar.o, gcc-ranlib, gcc-ranlib.o,
2648         gcc-nm, gcc-nm.o, gcc-ranlib.c, gcc-nm.c): Add.
2649         (install): Depend on install-gcc-ar.
2650         (install-gcc-ar): Add.
2651         (uninstall): Uninstall gcc-ar, gcc-nm, gcc-ranlib.
2652         * gcc-ar.c: Add new file.
2653
2654 2011-10-28  Pat Haugen <pthaugen@us.ibm.com>
2655
2656         * config/rs6000/rs6000.md (define_attr "type"): Add vecdouble.
2657         * config/rs6000/vsx.md (VStype_simple, VStype_mul): Use vecdouble
2658         type for V2DF.
2659         (VStype_div): Use vector types for V2DF/V4SF.
2660         (VStype_sqrt): Use *sqrt types.
2661         (VS_spdp_type): Change type to vecdouble.
2662         (*vsx_fmav2df4, *vsx_nfmsv2df4, vsx_xvcvdpsxws, vsx_xvcvdpuxws,
2663         vsx_xvcvuxdsp, vsx_xvcvsxwdp, vsx_xvcvuxwdp, vsx_xvcvspsxds,
2664         vsx_xvcvspuxds): Likewise.
2665         (*vsx_fms<mode>4): Set type via <VStype_mul>.
2666         (*vsx_eq_<mode>_p, *vsx_gt_<mode>_p, *vsx_ge_<mode>_p): Set type via
2667         <VStype_simple>.
2668         * config/rs6000/power7.md (power7-vecstore): Correct VSU pipe.
2669         (power7-fpcompare, power7-sdiv, power7-ddiv, power7-sqrt,
2670         power7-dsqrt): Correct insn latency.
2671         (power7-vecsimple): Add veccmp type and correct dispatch/VSU values.
2672         (power7-veccmp): Delete.
2673         (power7-vecfloat): Correct latency/dispatch/VSU values.
2674         (define_bypass "power7-vecfloat"): Correct latency and types.
2675         (power7-veccomplex, power7-vecperm): Correct dispatch/VSU values.
2676         (power7-vecdouble, power7-vecfdiv, power7-vecdiv): New.
2677
2678 2011-10-28  Uros Bizjak  <ubizjak@gmail.com>
2679
2680         * config/i386/i386.md (shift_insn): Rename code attribute from
2681         shiftrt_insn.  Also handle ashift RTX.
2682         (shift): Rename code attribute from shiftrt.  Also handle ashift RTX.
2683         (vshift): New code attribute.
2684         (<shift_insn>*): Rename from <shiftrt_insn>*. Update asm templates.
2685         (any_lshift): Move and rename code iterator from ...
2686         * config/i386/sse.md (lshift): ... here.
2687         (lshift_insn): Remove code attribute.
2688         (lshift): Remove code attribute.
2689         (vlshr<mode>3): Use lshiftrt RTX.
2690         (vashr<mode>3, ashrv16qi3, ashrv2di3): Use ashiftrt RTX.
2691         (vashl<mode>3, ashlv16qi3): Use ashift RTX.
2692         (avx2_<lshift>v<mode>): Rename from avx2_<shift_insn>v<mode>.  Use
2693         any_lshift code iterator.  Update asm template.
2694         (<shift_insn><mode>3): Macroize insn from lshr<mode>3 and ashl<mode>3
2695         usign any_lshift code iterator.
2696         * config/i386/mmx.md (mmx_<shift_insn><mode>3): Macroize insn from
2697         mmx_lshr<mode>3 and mmx_ashl<mode>3 usign any_lshift code iterator.
2698         * config/i386/i386.c (bdesc_args) <__builtin_ia32_psll>: Update.
2699
2700 2011-10-28  Georg-Johann Lay  <avr@gjlay.de>
2701
2702         PR target/49313
2703         * config/avr/avr.md (parityhi2): Expand allowing pseudos.
2704         (*parityhi2): New pre-reload insn-and-split to map 16-bit parity
2705         to the libgcc insn.
2706         (*parityqihi2): Same for 8-bit parity.
2707
2708 2011-10-28  Julian Brown  <julian@codesourcery.com>
2709
2710         PR rtl-optimization/47918
2711         * reload1.c (set_initial_label_offsets): Use initial offsets
2712         for labels on the nonlocal_goto_handler_labels chain.
2713
2714 2011-10-28  Iain Sandoe  <iains@gcc.gnu.org>
2715
2716         * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA):
2717         Move darwin-fpsave.asm from here to ... LIB2FUNCS_EXTRA.
2718         (LIB2FUNCS_EXTRA):  Add darwin-gpsave.asm.
2719         (TARGET_LIBGCC2_CFLAGS): Ensure that fPIC and -pipe are inherited from
2720         config/t-darwin.
2721         * config/rs6000/darwin.h (FP_SAVE_INLINE): Adjust to enable.
2722         (GP_SAVE_INLINE): Likewise.
2723         (SAVE_FP_PREFIX,  SAVE_FP_SUFFIX, RESTORE_FP_PREFIX,
2724         RESTORE_FP_SUFFIX): Set to empty strings.
2725         * config/rs6000/rs6000.c (rs6000_savres_strategy): Implement for Darwin.
2726         (debug_stack_info): Print savres_strategy.
2727         (rs6000_savres_routine_name): Implement for Darwin.
2728         (rs6000_make_savres_rtx): Adjust used register for Darwin.
2729         (rs6000_emit_prologue): Implement out-of-line saves for Darwin.
2730         (rs6000_output_function_prologue): Don't emit .extern for Mach-O.
2731         (rs6000_emit_epilogue): Implement out-of-line saves for Darwin.
2732         * config/rs6000/darwin-gpsave.asm: New file.
2733
2734 2011-10-28  Jakub Jelinek  <jakub@redhat.com>
2735
2736         * config/i386/sse.md (VI4SD_AVX2): Removed.
2737         (VI48_AVX2, VI128_128, VI48_128, VI48_256): New mode iterators.
2738         (vashl<mode>3): Use VI12_128 iterator instead of VI124_128.
2739         Add another expander using VI48_128 iterator for
2740         TARGET_AVX2 || TARGET_XOP and another using VI48_256 iterator
2741         for TARGET_AVX2.
2742         (vlshr<mode>3): Likewise.  Change register_operand predicate to
2743         nonimmediate_operand on last operand in the VI12_128 expander.
2744         (vashr<mode>3): Use VI128_128 iterator instead of VI124_128.
2745         (vashrv4si3, vashrv8si3): New expanders.
2746         (avx2_ashrvv8si, avx2_ashrvv4si, avx2_<lshift>vv8si,
2747         avx2_<lshift>vv2di): Removed.
2748         (avx2_ashrv<mode>): New insn with VI4_AVX2 iterator.
2749         (avx2_<lshift>v<mode>): Macroize using VI48_AVX2
2750         iterator.  Simplify pattern.
2751
2752 2010-10-28  Richard Guenther  <rguenther@suse.de>
2753
2754         PR driver/50876
2755         * lto-wrapper.c (get_options_from_collect_gcc_options):
2756         Properly count arguments.
2757         (run_gcc): Use an obstack to collect argv, properly separate
2758         switches and their arguments.
2759
2760 2011-10-28  Jakub Jelinek  <jakub@redhat.com>
2761
2762         * tree-vect-stmts.c (vectorizable_shift): Give up if op1 has different
2763         vector mode from vectype's mode.
2764
2765 2011-10-28  Chung-Lin Tang  <cltang@codesourcery.com>
2766
2767         PR rtl-optimization/49720
2768         * simplify-rtx.c (simplify_relational_operation_1): Detect
2769         infinite recursion condition in "(eq/ne (plus x cst1) cst2)
2770         simplifies to (eq/ne x (cst2 - cst1))" case.
2771
2772 2011-10-27  David S. Miller  <davem@davemloft.net>
2773
2774         * config/sparc/sparc.md (snedi_special): Only match when not VIS3.
2775         (*snedi_zero): Likewise.
2776         (*snedi_zero_trunc): Likewise.
2777         (snedi_special_vis3): New expander.
2778         (*snedi_zero_vis3): New insn.
2779         (*snedi_zero_trunc_vis3): Likewise.
2780         (*sltu_insn_vis3): Likewise.
2781         (*sltu_insn_vis3_trunc): Likewise.
2782         (addxc): Likewise.
2783         (*addxc_trunc_sp64_vis3): Likewise.
2784         * config/sparc/sparc.c (emit_scc_insn): When VIS3 use the
2785         gen_snedi_special_vis3 expander, and try GTU/LTU addx based
2786         sequences on DImode values.
2787
2788         * config/sparc/sparc.md (64-bit vector moves): Use 'e' not 'f'
2789         constraint.
2790
2791         * regcprop.c (copyprop_hardreg_forward_1): Reject the
2792         transformation when we narrow the mode on big endian.
2793
2794 2011-10-27  Jakub Jelinek  <jakub@redhat.com>
2795
2796         * config/i386/sse.md (avx_cvtpd2dq256_2, avx_cvttpd2dq256_2,
2797         vec_pack_sfix_trunc_v4df, vec_pack_sfix_v4df): New expanders.
2798         (*avx_cvtpd2dq256_2, *avx_cvttpd2dq256_2): New insns.
2799
2800         * config/i386/i386.c (ix86_print_operand): Handle 'q' and 'x'
2801         overrides for -masm=intel memory.
2802         * config/i386/sse.md (sse2_cvtdq2pd, sse2_cvtps2pd,
2803         sse4_1_<code>v8qiv8hi2, avx2_<code>v8qiv8si2,
2804         sse4_1_<code>v4hiv4si2, avx2_<code>v4hiv4di2,
2805         sse4_1_<code>v2siv2di2): Use %q1 instead of %1 for -masm=intel.
2806         (sse4_1_<code>v4qiv4si2, avx2_<code>v4qiv4di2,
2807         sse4_1_<code>v2hiv2di2): Use %k1 instead of %1 for -masm=intel.
2808         (sse4_1_<code>v2qiv2di2): Use %w1 instead of %1 for -masm=intel.
2809
2810 2011-10-27  Martin Jambor  <mjambor@suse.cz>
2811
2812         * ipa-prop.c (compute_pass_through_member_ptrs): Rename parm_info
2813         to parm_ainfo.
2814         (ipa_compute_jump_functions_for_edge): Likewise.
2815         (ipa_compute_jump_functions): Likewise.
2816         (ipa_analyze_indirect_call_uses): Likewise.
2817         (ipa_analyze_call_uses): Likewise.
2818         (ipa_analyze_params_uses): Likewise.
2819         (ipa_analyze_node): Likewise.
2820
2821 2011-10-27  Uros Bizjak  <ubizjak@gmail.com>
2822
2823         PR target/50875
2824         * config/i386/sse.md (*avx_unpcklpd256): Remove extra insn
2825         constraints.  Change alternative 1 to "x,m,1".
2826
2827 2011-10-27  Jakub Jelinek  <jakub@redhat.com>
2828
2829         * Makefile.in (build/gencheck.o): Depend on tree.def and
2830         c-family/c-common.def.
2831
2832         * tree-ssa-strlen.c: Include expr.h.
2833         (get_stridx): Don't use c_strlen, instead use string_constant
2834         and compute string length from it.
2835         * Makefile.in (tree-ssa-strlen.o): Depend on $(EXPR_H).
2836
2837 2011-10-27  Eric Botcazou  <ebotcazou@adacore.com>
2838
2839         PR rtl-optimization/46603
2840         PR bootstrap/50879
2841         * reload.c (push_reload): In the out case, restore previous behavior
2842         for subregs that don't have word mode.
2843
2844 2011-10-27  Ian Lance Taylor  <iant@google.com>
2845
2846         * cppdefault.c: Undef NATIVE_SYSTEM_HEADER_DIR if
2847         CROSS_DIRECTORY_STRUCTURE is defined and TARGET_SYSTEM_ROOT is not.
2848         (cpp_include_defaults): Only use NATIVE_SYSTEM_HEADER_DIR if it is
2849         defined.
2850
2851 2011-10-27  Richard Henderson  <rth@redhat.com>
2852
2853         * optabs.c (expand_vec_perm): Use the correct mode for scaling the
2854         selector.  Save the qimode constant selector for later use by the
2855         qimode vec_perm pattern.
2856
2857 2011-10-27  Bernd Schmidt  <bernds@codesourcery.com>
2858
2859         * config/c6x/c6x.c (unit_req_imbalance, res_mii): Cast the first arg
2860         to unit_req_factor to the right enum type.
2861         (get_unit_operand_masks, reshuffle_units, try_rename_operands,
2862         hwloop_optimize): Remove unused variables.
2863
2864 2010-10-27  Richard Guenther  <rguenther@suse.de>
2865
2866         PR middle-end/50731
2867         * tree-vect-generic.c (do_binop): Handle scalar operands.
2868
2869 2011-08-27  Uros Bizjak  <ubizjak@gmail.com>
2870
2871         PR target/37191
2872         * config/i386/sse.md (*vec_extract_v4sf_mem): Avoid combining registers
2873         from different units in a single alternative.
2874
2875 2011-10-26  David S. Miller  <davem@davemloft.net>
2876
2877         * config/sparc/sparc.c (emit_scc_insn): Force attempt of v9 sequences
2878         if we're comparing DImode and comparison is other than EQ or NE.
2879
2880         * config/sparc/sparc.c (emit_scc_insn): Do not try v9 sequences until
2881         LEU/LTU/GEU/GTU is attempted.
2882         * config/sparc/sparc.md (*neg_snesi_sign_extend): New 64-bit insn
2883         and split.
2884         (*neg_seqsi_sign_extend): Likewise.
2885         (*sltu_extend_sp64, *neg_sltu_extend_sp64, *sgeu_extend_sp64,
2886         *neg_sgeu_extend_sp64): New insns.
2887
2888         * config/sparc/sparc-protos.h (sparc_expand_conditional_move): Declare.
2889         * config/sparc/sparc.md (mov<I:mode>cc, mov<F:mode>cc): Call it.
2890         (*mov<I:mode>_cc_v9): Normalize to expect operand 0 always in operand 4.
2891         (*mov<I:mode>_cc_reg_sp64): Likewise.
2892         (*movsf_cc_v9): Likewise.
2893         (*movsf_cc_reg_sp64): Likewise.
2894         (*movdf_cc_v9): Likewise.
2895         (*movdf_cc_reg_sp64): Likewise.
2896         (*movtf_cc_hq_v9): Likewise.
2897         (*movtf_cc_reg_hq_sp64): Likewise.
2898         (*movtf_cc_v9): Likewise.
2899         (*movtf_cc_reg_sp64): Likewise.
2900         * config/sparc/sparc.c (sparc_expand_conditional_move): New function.
2901         (sparc_print_operand): Delete 'c' and 'd' handling, no longer used.
2902
2903 2011-10-26  Eric Botcazou  <ebotcazou@adacore.com>
2904
2905         * reload.c (reload_inner_reg_of_subreg): Change type of return value
2906         and type of OUTPUT parameter to bool and adjust.  Document MODE and
2907         OUTPUT parameters.  Use HARD_REGISTER_P.  Reorder final condition
2908         and improve associated comment.
2909         (push_reload): Clarify and update comments about reloading of subregs.
2910         Adjust calls to reload_inner_reg_of_subreg.  Compute the class upfront
2911         for the reloading of subregs in the out case as well.
2912
2913 2011-10-26  Alexandre Oliva  <aoliva@redhat.com>
2914
2915         PR debug/50826
2916         * var-tracking.c (rtx_debug_expr_p): New.
2917         (use_type): Don't use debug exprs to track non-VTA variables.
2918
2919 2011-10-26  Jeff Law  <law@redhat.com>
2920
2921         * doc/invoke.texi (sink-frequency-threshold): Document.
2922         * tree-ssa-sink.c: Include params.h.
2923         (select_best_block): New function.
2924         (statement_sink_location): Use it.
2925         * params.def (SINK_FREQUENCY_THRESHOLD): New PARAM.
2926
2927 2011-10-26  Iain Sandoe  <iains@gcc.gnu.org>
2928
2929         PR target/48108
2930         * config/darwin.c (top level): Amend comments concerning LTO output.
2931         (lto_section_num): New variable.  (darwin_lto_section_e): New GTY.
2932         (LTO_SECTS_SECTION, LTO_INDEX_SECTION): New.
2933         (LTO_NAMES_SECTION): Rename.
2934         (darwin_asm_named_section): Record LTO section counts and switches
2935         in a vec of darwin_lto_section_e.
2936         (darwin_file_start): Remove unused code.
2937         (darwin_file_end): Put an LTO section termination label.  Handle
2938         output of the wrapped LTO sections, index and names table.
2939
2940 2011-10-26  Alan Modra  <amodra@gmail.com>
2941
2942         * config/rs6000/rs6000.c (rs6000_make_savres_rtx): Delete unneeded
2943         declaration.
2944         (rs6000_emit_stack_reset): Only return insn emitted when it adjusts sp.
2945         (rs6000_make_savres_rtx): Rename to rs6000_emit_savres_rtx.  Use
2946         simple_return in pattern, emit instruction, and set jump_label.
2947         (rs6000_emit_prologue): Update for rs6000_emit_savres_rtx.  Use
2948         simple_return rather than return.
2949         (emit_cfa_restores): New function.
2950         (rs6000_emit_epilogue): Emit cfa_restores when flag_shrink_wrap.
2951         Add missing cfa_restores for SAVE_WORLD.  Add missing LR cfa_restore
2952         when using out-of-line gpr restore.  Add missing LR and FP regs
2953         cfa_restores for out-of-line fpr restore.  Consolidate code setting
2954         up cfa_restores.  Formatting.  Use LR_REGNO define.
2955         (rs6000_output_mi_thunk): Use simple_return rather than return.
2956         * config/rs6000/rs6000.md (sibcall*, sibcall_value*): Likewise.
2957         (return_internal*): Likewise.
2958         (any_return, return_pred, return_str): New iterators.
2959         (return, conditional return insns): Provide both return and
2960         simple_return variants.
2961         * config/rs6000/rs6000.h (EARLY_R12, LATE_R12): Define.
2962         (REG_ALLOC_ORDER): Move r12 before call-saved regs when FIXED_R13.
2963         Move r11 and r0 later to suit shrink-wrapping.
2964
2965 2011-10-26  Richard Guenther  <rguenther@suse.de>
2966
2967         * lto-wrapper.c (run_gcc): Properly init/free obstack.
2968
2969 2011-10-26  Jakub Jelinek  <jakub@redhat.com>
2970
2971         * config/i386/i386.md (UNSPEC_VSIBADDR): New.
2972         * config/i386/predicates.md (vsib_address_operand,
2973         vsib_mem_operator): New predicates.
2974         * config/i386/i386.c (ix86_print_operand_address): Handle
2975         UNSPEC_VSIBADDR addresses.
2976         * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>,
2977         avx2_gatherdi<mode>256): Adjust expanders to use MEM with
2978         UNSPEC_VSIBADDR address.
2979         (*avx2_gathersi<mode>, *avx2_gatherdi<mode>, *avx2_gatherdi<mode>256):
2980         Adjust insns to use MEM with UNSPEC_VSIBADDR address.
2981
2982 2011-10-26  Tom de Vries  <tom@codesourcery.com>
2983
2984         PR tree-optimization/50763
2985         * tree-ssa-tail-merge.c (replace_block_by): Update vops if phi_vuse1 or
2986         phi_vuse2 is NULL_TREE only if bb1 dominates or is dominated by bb2.
2987
2988 2011-10-26  Richard Guenther  <rguenther@suse.de>
2989
2990         PR lto/41844
2991         * Makefile.in (lto-wrapper): Depend on and link against opts-common.o.
2992         (lto-wrapper.o): Depend on $(OPTS_H) and $(OPTIONS_H).
2993         * lto-wrapper.c (get_options_from_collect_gcc_options): New function.
2994         (run_gcc): Use it.  Filter out language specific options.
2995
2996 2011-10-26  Andreas Tobler  <andreast@fgznet.ch>
2997
2998         * config/i386/freebsd64.h (LINK_SPEC): Emit the same warning as the
2999         32-bit target does.
3000
3001 2011-10-25  Paolo Carlini  <paolo.carlini@oracle.com>
3002
3003         PR driver/46617
3004         * gcc.c (main): Fix fatal_error string for translation.
3005
3006 2011-10-25  Ian Lance Taylor  <iant@google.com>
3007
3008         * tree-eh.c (do_return_redirection): Remove return_value_p
3009         parameter.  Change all callers.
3010         (lower_try_finally_nofallthru): Remove local return_val.
3011         (lower_try_finally_onedest): Likewise.
3012         (lower_try_finally_copy): Likewise.
3013         (lower_try_finally_switch): Likewise.
3014
3015 2011-10-25  H.J. Lu  <hongjiu.lu@intel.com>
3016
3017         * config/i386/mmx.md (*mmx_maskmovq): Replace :SI with :P and
3018         remove "&& !TARGET_64BIT"
3019         (*mmx_maskmovq_rex): Removed.
3020
3021 2011-10-25  Eric Botcazou  <ebotcazou@adacore.com>
3022
3023         PR rtl-optimization/46603
3024         * reload.c (push_reload): In the out case, reload the subreg as well
3025         as the reg if it has word mode.
3026
3027 2011-10-25  Eric Botcazou  <ebotcazou@adacore.com>
3028
3029         * dwarf2out.c (add_gnat_descriptive_type_attribute): Temporarily
3030         suppress debug info for the parent type.
3031
3032 2011-10-25  Eric Botcazou  <ebotcazou@adacore.com>
3033
3034         * config/ia64/ia64.c (ia64_profile_hook): Fix thinko.
3035
3036 2011-10-25  Richard Henderson  <rth@redhat.com>
3037
3038         * config/i386/sse.md (VEC_EXTRACT_EVENODD_MODE): Remove.
3039         (vec_extract_even<mode>, vec_extract_odd<mode>): Remove.
3040
3041         * config/rs6000/altivec.md (vec_extract_evenv8hi,
3042         vec_extract_evenv16qi, vec_extract_oddv4si,
3043         vec_extract_oddv4sf): Remove.
3044
3045         * config/spu/spu.md (vec_extract_evenv4si, vec_extract_evenv4sf,
3046         vec_extract_evenv8hi, vec_extract_evenv16qi, vec_extract_oddv4si,
3047         vec_extract_oddv4sf, vec_extract_oddv8hi, vec_extract_oddv16qi,
3048         vec_interleave_highv4sf, vec_interleave_lowv4sf,
3049         vec_interleave_highv4si, vec_interleave_lowv4si,
3050         vec_interleave_highv8hi, vec_interleave_lowv8hi,
3051         vec_interleave_highv16qi, vec_interleave_lowv16qi): Remove.
3052
3053         * expr.c (expand_expr_real_2) [VEC_EXTRACT_EVEN_EXPR]: Use binop.
3054         [VEC_EXTRACT_ODD_EXPR, VEC_INTERLEAVE_HIGH_EXPR]: Likewise.
3055         [VEC_INTERLEAVE_LOW_EXPR]: Likewise.
3056         * optabs.c (expand_binop): Implement vec_interleave_high_optab,
3057         vec_interleave_low_optab, vec_extract_even_optab,
3058         vec_extract_odd_optab with expand_vec_perm.
3059         (can_vec_perm_for_code_p): New.
3060         * optabs.h: Update.
3061         * tree-vect-data-refs.c (vect_strided_store_supported): Allow for
3062         fallback via can_vec_perm_for_code_p.
3063         (vect_strided_load_supported): Likewise.
3064         * tree-vect-generic.c (expand_vector_operations_1): Never lower
3065         VEC_INTERLEAVE_HIGH_EXPR, VEC_INTERLEAVE_LOW_EXPR,
3066         VEC_EXTRACT_EVEN_EXPR, VEC_EXTRACT_ODD_EXPR.
3067
3068         * target.def (vec_perm_const_ok): Change parameters to mode and
3069         array of indicies.
3070         * doc/tm.texi: Rebuild.
3071         * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3072         parameters to mode and array of indicies.
3073         * expr.c (expand_expr_real_2) [VEC_PERM_EXPR]: Expand operands here.
3074         * optabs.c (can_vec_perm_p): Rename from can_vec_perm_expr_p.
3075         Change parameters to mode and array of indicies.
3076         (expand_vec_perm_1): Rename from expand_vec_perm_expr_1.
3077         (expand_vec_perm): Rename from expand_vec_perm_expr.  Change
3078         parameters to mode and rtx inputs.  Try lowering to QImode
3079         vec_perm_const before trying fully variable permutation.
3080         * optabs.h: Update decls.
3081         * tree-vect-generic.c (lower_vec_perm): Extract array of indices from
3082         VECTOR_CST to pass to can_vec_perm_p.
3083         * tree-vect-slp.c (vect_get_mask_element): Change mask parameter type
3084         from int pointer to unsigned char pointer.
3085         (vect_transform_slp_perm_load): Update for change to can_vec_perm_p.
3086         * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
3087
3088         * tree.def (VEC_EXTRACT_EVEN_EXPR): Fix typo in text name.
3089         (VEC_EXTRACT_ODD_EXPR, VEC_INTERLEAVE_HIGH_EXPR,
3090         VEC_INTERLEAVE_LOW_EXPR): Likewise.
3091
3092 2011-10-25  Mike Stump  <mikestump@comcast.net>
3093
3094         * reload.c (regno_clobbered_p): Fix typo.
3095
3096 2011-10-25  Dodji Seketeli  <dodji@redhat.com>
3097
3098         * input.c (expand_location): Rewrite using linemap_resolve_location
3099         and linemap_expand_location.  Add a comment.
3100
3101 2011-10-25  Jakub Jelinek  <jakub@redhat.com>
3102
3103         PR tree-optimization/50596
3104         * tree-vect-stmts.c (vect_mark_relevant): Only use
3105         FOR_EACH_IMM_USE_FAST if lhs is SSA_NAME.
3106         (vectorizable_store): If is_pattern_stmt_p look through
3107         VIEW_CONVERT_EXPR on lhs.
3108         * tree-vect-patterns.c (check_bool_pattern, adjust_bool_pattern):
3109         Use unsigned type instead of signed.
3110         (vect_recog_bool_pattern): Optimize also stores into bool memory in
3111         addition to casts from bool to integral types.
3112         (vect_mark_pattern_stmts): If pattern_stmt already has vinfo
3113         created, don't create it again.
3114
3115 2011-10-25  Kai Tietz  <ktietz@redhat.com>
3116
3117         * config/i386/i386.c (ix86_frame_pointer_required): Require
3118         frame-pointer, if setjmp is used for 32-bit ms-abi.
3119
3120 2011-10-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
3121
3122         * builtins.c (set_builtin_user_assembler_name): Remove extra
3123         newline added in October 11th, 2011 change.
3124
3125 2011-10-24  David S. Miller  <davem@davemloft.net>
3126
3127         * config/sparc/little-endian.opt: Delete.
3128         * config.gcc: Remove references to config/sparc/little-endian.opt
3129         * doc/invoke.texi: Remove documentation of -mlittl-endian on sparc.
3130         * config/sparc/linux64.h: Delete references to -mlittle-endian.
3131         * config/sparc/netbsd-elf.h: Likewise.
3132         * config/sparc/openbsd64.h: Likewise.
3133         * config/sparc/sparc.h: Likewise.
3134         * config/sparc/sp64-elf.h: Likewise and delete overrides for
3135         BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
3136         * config/sparc/sparc.c (dump_target_flag_bits): Remove reference
3137         to MASK_LITTLE_ENDIAN.
3138         * config/sparc/sparc.opt (Mask(LITTLE_ENDIAN)): Delete.
3139
3140         * config/sparc/sparc.md: Only use F, G, and C constraints in FP insns.
3141         Only use D, Y, and Z constraints in vector insns.
3142
3143         * config/sparc/sparc.md (cpu_feature, enabled): New attributes.
3144         (*movsi_insn_novis3, *movsi_insn_vis3): Consolidate into one pattern
3145         called *movsi_insn.
3146         (*movdi_insn_sp32_v9_novis3, *movdi_insn_sp32_v9_vis3): Consolidate
3147         into *movdi_insn_sp32.
3148         (*movdi_insn_sp64_novis3, *movdi_insn_sp64_vis3): Consolidate into
3149         one pattern called *movdi_insn_sp64.
3150         (*movsf_insn_novis3, *movsf_insn_vis3, *movsf_insn_no_fpu):
3151         Consolidate into one pattern called *movsf_insn.
3152         (*movdf_insn_sp32_no_fpu, *movdf_insn_sp32_v9_novis3,
3153         *movdf_insn_sp32_v9_vis3, *movdf_insn_sp32_v9_no_fpu): Consolidate
3154         into *movdf_insn_sp32.
3155         (*movdf_insn_sp64_novis3, *movdf_insn_sp64_vis3,
3156         *movdf_insn_sp64_no_fpu): Consolidate into one pattern called
3157         *movdf_insn_sp64.
3158         (*zero_extendsidi2_insn_sp64_novis3,
3159         *zero_extendsidi2_insn_sp64_vis3): Consolidate into one pattern
3160         called *zero_extendsidi2_insn_sp64.
3161         (*sign_extendsidi2_insn_novis3, *sign_extendsidi2_insn_vis3):
3162         Consolidate into one pattern named *sign_extendsidi2_insn.
3163         (*mov<VM32:mode>_insn_novis3, *mov<VM32:mode>_insn_vis3):
3164         Consolidate into one pattern named *mov<VM32:mode>_insn.
3165         (*mov<VM64:mode>_insn_sp64_novis3,
3166         *mov<VM64:mode>_insn_sp64_novis3): Consolidate into one pattern
3167         named *mov<VM64:mode>_insn_sp64.
3168         (*mov<VM64:mode>_insn_sp32_novis3,
3169         *mov<VM64:mode>_insn_sp32_vis3): Consolidate into one pattern
3170         named *mov<VM64:mode>_insn_sp32.
3171
3172 2011-10-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3173
3174         * tree-ssa-strlen.c (get_string_length): Change assertion to STPCPY.
3175         (zero_length_string): Change assertion to accept strinfo without
3176         length but with stmt instead.
3177         Set the endptr pointer also if starting a new chain.
3178         (adjust_related_strinfos): Ignore strinfos marked for delayed
3179         length computation.
3180         (handle_builtin_strcpy): Mark earlier strinfo elements also for
3181         delayed length computation.
3182
3183 2011-10-24  Georg-Johann Lay  <avr@gjlay.de>
3184
3185         PR target/50820
3186         Port from 4.6 branch r180379
3187         * doc/invoke.texi (AVR Options): New subsubsection to explain EIND
3188         handling and indirect jump/calls on devices > 128k.
3189
3190 2011-10-24  Anatoly Sokolov  <aesok@post.ru>
3191             Georg-Johann Lay  <avr@gjlay.de>
3192
3193         PR target/49824
3194         * doc/extend.texi (Declaring Attributes of Functions):
3195         Document OS_main and OS_task attributes.
3196         (Specifying Attributes of Variables): Move up
3197         subsection "AVR Variable Attributes" as of alphabetical order.
3198
3199 2011-10-24  Richard Guenther  <rguenther@suse.de>
3200
3201         * tree-vect-stmts.c (vect_get_vec_def_for_operand): Convert constants
3202         to vector element type.
3203         (vectorizable_assignment): Bail out for non-mode-precision operations.
3204         (vectorizable_shift): Likewise.
3205         (vectorizable_operation): Likewise.
3206         (vectorizable_type_demotion): Likewise.
3207         (vectorizable_type_promotion): Likewise.
3208         (vectorizable_store): Handle non-mode-precision stores.
3209         (vectorizable_load): Handle non-mode-precision loads.
3210         (get_vectype_for_scalar_type_and_size): Return a vector type
3211         for non-mode-precision integers.
3212         * tree-vect-loop.c (vectorizable_reduction): Bail out for
3213         non-mode-precision reductions.
3214
3215 2011-10-24  Julian Brown  <julian@codesourcery.com>
3216
3217         * config/m68k/m68k.c (notice_update_cc): Tighten condition for
3218         setting CC_REVERSED for FP comparisons.
3219
3220 2011-10-24  Richard Guenther  <rguenther@suse.de>
3221
3222         PR tree-optimization/50838
3223         * tree-data-ref.c (dr_analyze_indices): Properly canonicalize
3224         a MEM_REF base if we change it.
3225
3226 2011-10-24  Bernd Schmidt  <bernds@codesourcery.com>
3227
3228         PR bootstrap/50836
3229         * rtlanal.c: Swap includes of "hard-reg-set.h" and "rtl.h".
3230
3231         PR rtl-optimization/50833
3232         * function.c (thread_prologue_and_epilogue_insns): Expect the
3233         return insn optimization only if optimize.
3234
3235 2011-10-24  Georg-Johann Lay  <avr@gjlay.de>
3236
3237         * config/avr/avr.c: Break long lines.
3238         Define target hooks on the fly if applicable.
3239         (TARGET_ASM_FUNCTION_RODATA_SECTION): Remove first definition
3240         overridden later.
3241         (targetm): Move definition to end of file.
3242         (avr_can_eliminate): Make static on the fly.
3243         (avr_frame_pointer_required_p): Ditto.
3244         (avr_hard_regno_scratch_ok): Ditto.
3245         (avr_builtin_setjmp_frame_value): Make static on the fly.
3246         Indent according to coding rules.
3247         (avr_case_values_threshold): Ditto.
3248         (avr_attribute_table): Move down.
3249
3250 2011-10-24  Ira Rosen  <ira.rosen@linaro.org>
3251
3252         PR tree-optimization/50730
3253         * tree-vect-data-refs.c (vect_analyze_data_refs): Stop basic block
3254         analysis if encountered unsupported data-ref.
3255
3256 2011-10-23  David S. Miller  <davem@davemloft.net>
3257
3258         * config/sparc/sparc.c (sparc_option_override): Remove -mv8plus
3259         cpu adjustment.
3260         * config/sparc/linux64.h (CC1_SPEC): When defaulting to 64-bit,
3261         append -mcpu=v9 when -mv8plus is given.
3262
3263         * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): We can move
3264         between float and non-float regs when VIS3.
3265         * config/sparc/sparc.c (eligible_for_restore_insn): We can't
3266         use a restore when the source is a float register.
3267         (sparc_split_regreg_legitimate): When VIS3 allow moves between
3268         float and integer regs.
3269         (sparc_register_move_cost): Adjust to account for VIS3 moves.
3270         (sparc_preferred_reload_class): On 32-bit with VIS3 when moving an
3271         integer reg to a class containing EXTRA_FP_REGS, constrain to FP_REGS.
3272         (sparc_secondary_reload): On 32-bit with VIS3 when moving between
3273         float and integer regs we sometimes need a FP_REGS class
3274         intermediate move to satisfy the reload.  When this happens
3275         specify an extra cost of 2.
3276         (*movsi_insn): Rename to have "_novis3" suffix and add !VIS3 guard.
3277         (*movdi_insn_sp32_v9): Likewise.
3278         (*movdi_insn_sp64): Likewise.
3279         (*movsf_insn): Likewise.
3280         (*movdf_insn_sp32_v9): Likewise.
3281         (*movdf_insn_sp64): Likewise.
3282         (*zero_extendsidi2_insn_sp64): Likewise.
3283         (*sign_extendsidi2_insn): Likewise.
3284         (*movsi_insn_vis3): New insn.
3285         (*movdi_insn_sp32_v9_vis3): New insn.
3286         (*movdi_insn_sp64_vis3): New insn.
3287         (*movsf_insn_vis3): New insn.
3288         (*movdf_insn_sp32_v9_vis3): New insn.
3289         (*movdf_insn_sp64_vis3): New insn.
3290         (*zero_extendsidi2_insn_sp64_vis3): New insn.
3291         (*sign_extendsidi2_insn_vis3): New insn.
3292         (TFmode reg/reg split): Make sure both REG operands are float.
3293         (*mov<VM32:mode>_insn): Add "_novis3" suffix and !VIS3 guard. Remove
3294         easy constant to integer reg alternatives.
3295         (*mov<VM64:mode>_insn_sp64): Likewise.
3296         (*mov<VM64:mode>_insn_sp32_novis3): Likewise.
3297         (*mov<VM32:mode>_insn_vis3): New insn.
3298         (*mov<VM64:mode>_insn_sp64_vis3): New insn.
3299         (*mov<VM64:mode>_insn_sp32_vis3): New insn.
3300         (VM64 reg<-->reg split): New splitter for 32-bit.
3301
3302         * config/sparc/sparc.c (sparc_split_regreg_legitimate): New function.
3303         * config/sparc/sparc-protos.h (sparc_split_regreg_legitimate):
3304         Declare it.
3305         * config/sparc/sparc.md (DImode reg/reg split): Use it.
3306         (DFmode reg/reg split): Likewise.
3307
3308         * config/sparc/sparc.md (*movdi_insn_sp32_v9): Add alternatives for
3309         generating fzero and fone instructions.
3310         (DImode const_int --> reg splitter): Only trigger for integer regs.
3311
3312         * config/sparc/predicates.md (input_operand): Disallow vector
3313         constants other than 0 and -1.
3314         * config/sparc/sparc.c (sparc_preferred_reload_class): Return
3315         NO_REGS for vector constants other than 0 and -1.
3316
3317         * config/sparc/sparc.h (SPARC_FIRST_INT_REG, SPARC_LAST_INT_REG,
3318         SPARC_INT_REG_P): Define.
3319         (HARD_REGNO_NREGS): Use SPARC_INT_REG_P.
3320         (REGNO_OK_FOR_INDEX_P): Likewise.
3321         * config/sparc/sparc.c (gen_df_reg): Likewise.
3322         (eligible_for_return_delay): Likewise.
3323         (eligible_for_sibcall_delay): Likewise.
3324         (sparc_legitimate_address_p): Likewise.
3325         (emit_save_or_restore_regs): Likewise.
3326         (registers_ok_for_ldd_peep): Likewise.
3327         * config/spac/sparc.md (DI mode splitters): Likewise.
3328         (SF&nb