OpenE2K/gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
3f7c745381
gcc/libgomp/testsuite/libgomp.c/pr26943-4.c

62 lines
1.5 KiB
C
Raw Normal View History

re PR c++/26943 ([gomp] firstprivate + lastprivate uses inefficient barrier) PR c++/26943 * omp-low.c (maybe_lookup_decl_in_outer_ctx): New function. (build_outer_var_ref): Use maybe_lookup_decl_in_outer_ctx to find if var will be a global variable even in the nested context. (omp_copy_decl): Only check for global variable at the end, it might be overridden in outer contexts. (scan_sharing_clauses): For global variables don't create a field. (lower_rec_input_clauses): Do nothing for global shared variables. Emit a barrier at the end of ILIST if there were any decls in both firstprivate and lastprivate clauses. (lower_send_clauses): Do nothing for global variables except for COPYIN. * testsuite/libgomp.c/pr26943-1.c: New test. * testsuite/libgomp.c/pr26943-2.c: New test. * testsuite/libgomp.c/pr26943-3.c: New test. * testsuite/libgomp.c/pr26943-4.c: New test. * testsuite/libgomp.c++/pr27337.C: Remove barrier. * testsuite/libgomp.c++/pr26943.C: New test. From-SVN: r113483
2006-05-02 22:03:38 +02:00
/* PR c++/26943 */
/* { dg-do run } */
extern int omp_set_dynamic (int);
extern int omp_get_thread_num (void);
extern void abort (void);
int a = 8, b = 12, c = 16, d = 20, j = 0, l = 0;
char e[10] = "a", f[10] = "b", g[10] = "c", h[10] = "d";
volatile int k;
int
main (void)
{
int i;
omp_set_dynamic (0);
omp_set_nested (1);
#pragma omp parallel num_threads (2) reduction (+:l) \
firstprivate (a, b, c, d, e, f, g, h, j)
if (k == omp_get_thread_num ())
{
#pragma omp parallel for shared (a, e) firstprivate (b, f) \
lastprivate (c, g) private (d, h) \
schedule (static, 1) num_threads (4) \
reduction (+:j)
for (i = 0; i < 4; i++)
{
if (a != 8 || b != 12 || e[0] != 'a' || f[0] != 'b')
j++;
c-cppbuiltin.c (c_cpp_builtins): Change _OPENMP value to 200805. * c-cppbuiltin.c (c_cpp_builtins): Change _OPENMP value to 200805. * langhooks.h (struct lang_hooks_for_decls): Add omp_finish_clause. Add omp_private_outer_ref hook, add another argument to omp_clause_default_ctor hook. * langhooks-def.h (LANG_HOOKS_OMP_FINISH_CLAUSE): Define. (LANG_HOOKS_OMP_PRIVATE_OUTER_REF): Define. (LANG_HOOKS_OMP_CLAUSE_DEFAULT_CTOR): Change to hook_tree_tree_tree_tree_null. (LANG_HOOKS_DECLS): Add LANG_HOOKS_OMP_FINISH_CLAUSE and LANG_HOOKS_OMP_PRIVATE_OUTER_REF. * hooks.c (hook_tree_tree_tree_tree_null): New function. * hooks.h (hook_tree_tree_tree_tree_null): New prototype. * tree.def (OMP_TASK): New tree code. * tree.h (OMP_TASK_COPYFN, OMP_TASK_ARG_SIZE, OMP_TASK_ARG_ALIGN, OMP_CLAUSE_PRIVATE_OUTER_REF, OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_COLLAPSE_ITERVAR, OMP_CLAUSE_COLLAPSE_COUNT, OMP_TASKREG_CHECK, OMP_TASKREG_BODY, OMP_TASKREG_CLAUSES, OMP_TASKREG_FN, OMP_TASKREG_DATA_ARG, OMP_TASK_BODY, OMP_TASK_CLAUSES, OMP_TASK_FN, OMP_TASK_DATA_ARG, OMP_CLAUSE_COLLAPSE_EXPR): Define. (enum omp_clause_default_kind): Add OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. (OMP_DIRECTIVE_P): Add OMP_TASK. (OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_UNTIED): New clause codes. (OMP_CLAUSE_SCHEDULE_AUTO): New schedule kind. * tree.c (omp_clause_code_name): Add OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED entries. (omp_clause_num_ops): Likewise. Increase OMP_CLAUSE_LASTPRIVATE num_ops to 2. (walk_tree_1): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Walk OMP_CLAUSE_LASTPRIVATE_STMT. * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE_SCHEDULE_AUTO, OMP_CLAUSE_UNTIED, OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. (dump_generic_node): Handle OMP_TASK and collapsed OMP_FOR loops. * c-omp.c (c_finish_omp_for): Allow pointer iterators. Remove warning about unsigned iterators. Change decl/init/cond/incr arguments to TREE_VECs, check arguments for all collapsed loops. (c_finish_omp_taskwait): New function. (c_split_parallel_clauses): Put OMP_CLAUSE_COLLAPSE clause to ws_clauses. * c-parser.c (c_parser_omp_for_loop): Parse collapsed loops. Call default_function_array_conversion on init. Add par_clauses argument. If decl is present in parallel's lastprivate clause, change it to shared and add lastprivate clause for decl to OMP_FOR_CLAUSES. Add clauses argument, on success set OMP_FOR_CLAUSES to it. Look up collapse count in clauses. (c_parser_omp_for, c_parser_omp_parallel): Adjust c_parser_omp_for_loop callers. (OMP_FOR_CLAUSE_MASK): Add 1 << PRAGMA_OMP_CLAUSE_COLLAPSE. (c_parser_pragma): Handle PRAGMA_OMP_TASKWAIT. (c_parser_omp_clause_name): Handle collapse and untied clauses. (c_parser_omp_clause_collapse, c_parser_omp_clause_untied): New functions. (c_parser_omp_clause_schedule): Handle schedule(auto). Include correct location in the error message. (c_parser_omp_all_clauses): Handle PRAGMA_OMP_CLAUSE_COLLAPSE and PRAGMA_OMP_CLAUSE_UNTIED. (OMP_TASK_CLAUSE_MASK): Define. (c_parser_omp_task, c_parser_omp_taskwait): New functions. (c_parser_omp_construct): Handle PRAGMA_OMP_TASK. * tree-nested.c (convert_nonlocal_omp_clauses, convert_local_omp_clauses): Handle OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_REDUCTION_INIT, OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Don't handle TREE_STATIC or DECL_EXTERNAL VAR_DECLs in OMP_CLAUSE_DECL. (conver_nonlocal_reference, convert_local_reference, convert_call_expr): Handle OMP_TASK the same as OMP_PARALLEL. Use OMP_TASKREG_* macros rather than OMP_PARALLEL_*. (walk_omp_for): Adjust for OMP_FOR_{INIT,COND,INCR} changes. * tree-gimple.c (is_gimple_stmt): Handle OMP_TASK. * c-tree.h (c_begin_omp_task, c_finish_omp_task): New prototypes. * c-pragma.h (PRAGMA_OMP_TASK, PRAGMA_OMP_TASKWAIT): New. (PRAGMA_OMP_CLAUSE_COLLAPSE, PRAGMA_OMP_CLAUSE_UNTIED): New. * c-typeck.c (c_begin_omp_task, c_finish_omp_task): New functions. (c_finish_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. * c-pragma.c (init_pragma): Init omp task and omp taskwait pragmas. * c-common.h (c_finish_omp_taskwait): New prototype. * gimple-low.c (lower_stmt): Handle OMP_TASK. * tree-parloops.c (create_parallel_loop): Create 1 entry vectors for OMP_FOR_{INIT,COND,INCR}. * tree-cfg.c (remove_useless_stmts_1): Handle OMP_* containers. (make_edges): Handle OMP_TASK. * tree-ssa-operands.c (get_expr_operands): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. * tree-inline.c (estimate_num_insns_1): Handle OMP_TASK. * builtin-types.def (BT_PTR_ULONGLONG, BT_PTR_FN_VOID_PTR_PTR, BT_FN_BOOL_ULONGLONGPTR_ULONGLONGPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): New. * omp-builtins.def (BUILT_IN_GOMP_TASK, BUILT_IN_GOMP_TASKWAIT, BUILT_IN_GOMP_LOOP_ULL_STATIC_START, BUILT_IN_GOMP_LOOP_ULL_DYNAMIC_START, BUILT_IN_GOMP_LOOP_ULL_GUIDED_START, BUILT_IN_GOMP_LOOP_ULL_RUNTIME_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_STATIC_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_DYNAMIC_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_GUIDED_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_RUNTIME_START, BUILT_IN_GOMP_LOOP_ULL_STATIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_DYNAMIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_GUIDED_NEXT, BUILT_IN_GOMP_LOOP_ULL_RUNTIME_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_STATIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_DYNAMIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_GUIDED_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_RUNTIME_NEXT): New builtins. * gimplify.c (gimplify_omp_for): Allow pointer type for decl, handle POINTER_PLUS_EXPR. If loop counter has been replaced and original iterator is present in lastprivate clause or if collapse > 1, set OMP_CLAUSE_LASTPRIVATE_STMT. Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. (gimplify_expr): Handle OMP_SECTIONS_SWITCH and OMP_TASK. (enum gimplify_omp_var_data): Add GOVD_PRIVATE_OUTER_REF. (omp_notice_variable): Set GOVD_PRIVATE_OUTER_REF if needed, if it is set, lookup var in outer contexts too. Handle OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. Handle vars that are supposed to be implicitly determined firstprivate for task regions. (gimplify_scan_omp_clauses): Set GOVD_PRIVATE_OUTER_REF if needed, if it is set, lookup var in outer contexts too. Set OMP_CLAUSE_PRIVATE_OUTER_REF if GOVD_PRIVATE_OUTER_REF is set. Handle OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Take region_type as last argument instead of in_parallel and in_combined_parallel. (gimplify_omp_parallel, gimplify_omp_for, gimplify_omp_workshare): Adjust callers. (gimplify_adjust_omp_clauses_1): Set OMP_CLAUSE_PRIVATE_OUTER_REF if GOVD_PRIVATE_OUTER_REF is set. Call omp_finish_clause langhook. (new_omp_context): Set default_kind to OMP_CLAUSE_DEFAULT_UNSPECIFIED for OMP_TASK regions. (omp_region_type): New enum. (struct gimplify_omp_ctx): Remove is_parallel and is_combined_parallel fields, add region_type. (new_omp_context): Take region_type as argument instead of is_parallel and is_combined_parallel. (gimple_add_tmp_var, omp_firstprivatize_variable, omp_notice_variable, omp_is_private, omp_check_private): Adjust ctx->is_parallel and ctx->is_combined_parallel checks. (gimplify_omp_task): New function. (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. * omp-low.c (extract_omp_for_data): Use schedule(static) for schedule(auto). Handle pointer and unsigned iterators. Compute fd->iter_type. Handle POINTER_PLUS_EXPR increments. Add loops argument. Extract data for collapsed OMP_FOR loops. (expand_parallel_call): Assert sched_kind isn't auto, map runtime schedule to index 3. (struct omp_for_data_loop): New type. (struct omp_for_data): Remove v, n1, n2, step, cond_code fields. Add loop, loops, collapse and iter_type fields. (workshare_safe_to_combine_p): Disallow combined for if iter_type is unsigned long long. Don't combine collapse > 1 loops unless all bounds and steps are constant. Adjust extract_omp_for_data caller. (expand_omp_for_generic): Handle pointer, unsigned and long long iterators. Handle collapsed OMP_FOR loops. Adjust for struct omp_for_data changes. If libgomp function doesn't return boolean_type_node, add comparison of the return value with 0. (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle pointer, unsigned and long long iterators. Adjust for struct omp_for_data changes. (expand_omp_for): Assert sched_kind isn't auto, map runtime schedule to index 3. Use GOMP_loop_ull*{start,next} if iter_type is unsigned long long. Allocate loops array, pass it to extract_omp_for_data. For collapse > 1 loops use always expand_omp_for_generic. (omp_context): Add sfield_map and srecord_type fields. (is_task_ctx, lookup_sfield): New functions. (use_pointer_for_field): Use is_task_ctx helper. Change first argument's type from const_tree to tree. Clarify comment. In OMP_TASK disallow copy-in/out sharing. (build_sender_ref): Call lookup_sfield instead of lookup_field. (install_var_field): Add mask argument. Populate both record_type and srecord_type if needed. (delete_omp_context): Destroy sfield_map, clear DECL_ABSTRACT_ORIGIN in srecord_type. (fixup_child_record_type): Also remap FIELD_DECL's DECL_SIZE{,_UNIT} and DECL_FIELD_OFFSET. (scan_sharing_clauses): Adjust install_var_field callers. For firstprivate clauses on explicit tasks allocate the var by value in record_type unconditionally, rather than by reference. Handle OMP_CLAUSE_PRIVATE_OUTER_REF. Scan OMP_CLAUSE_LASTPRIVATE_STMT. Use is_taskreg_ctx instead of is_parallel_ctx. Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. (create_omp_child_function_name): Add task_copy argument, use *_omp_cpyfn* names if it is true. (create_omp_child_function): Add task_copy argument, if true create *_omp_cpyfn* helper function. (scan_omp_parallel): Adjust create_omp_child_function callers. Rename parallel_nesting_level to taskreg_nesting_level. (scan_omp_task): New function. (lower_rec_input_clauses): Don't run constructors for firstprivate explicit task vars which are initialized by *_omp_cpyfn*. Pass outer var ref to omp_clause_default_ctor hook if OMP_CLAUSE_PRIVATE_OUTER_REF or OMP_CLAUSE_LASTPRIVATE. Replace OMP_CLAUSE_REDUCTION_PLACEHOLDER decls in OMP_CLAUSE_REDUCTION_INIT. (lower_send_clauses): Clear DECL_ABSTRACT_ORIGIN if in task to avoid duplicate setting of fields. Handle OMP_CLAUSE_PRIVATE_OUTER_REF. (lower_send_shared_vars): Use srecord_type if non-NULL. Don't copy-out if TREE_READONLY, only copy-in. (expand_task_copyfn): New function. (expand_task_call): New function. (struct omp_taskcopy_context): New type. (task_copyfn_copy_decl, task_copyfn_remap_type, create_task_copyfn): New functions. (lower_omp_parallel): Rename to... (lower_omp_taskreg): ... this. Use OMP_TASKREG_* macros where needed. Call create_task_copyfn if srecord_type is needed. Adjust sender_decl type. (task_shared_vars): New variable. (check_omp_nesting_restrictions): Warn if work-sharing, barrier, master or ordered region is closely nested inside OMP_TASK. Add warnings for barrier if closely nested inside of work-sharing, ordered, or master region. (scan_omp_1): Call check_omp_nesting_restrictions even for GOMP_barrier calls. Rename parallel_nesting_level to taskreg_nesting_level. Handle OMP_TASK. (lower_lastprivate_clauses): Even if some lastprivate is found on a work-sharing construct, continue looking for them on parent parallel construct. (lower_omp_for_lastprivate): Add lastprivate clauses to the beginning of dlist rather than end. Adjust for struct omp_for_data changes. (lower_omp_for): Add rec input clauses before OMP_FOR_PRE_BODY, not after it. Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes, adjust extract_omp_for_data caller. (get_ws_args_for): Adjust extract_omp_for_data caller. (scan_omp_for): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. (lower_omp_single_simple): If libgomp function doesn't return boolean_type_node, add comparison of the return value with 0. (diagnose_sb_1, diagnose_sb_2): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. Handle OMP_TASK. (parallel_nesting_level): Rename to... (taskreg_nesting_level): ... this. (is_taskreg_ctx): New function. (build_outer_var_ref, omp_copy_decl): Use is_taskreg_ctx instead of is_parallel_ctx. (execute_lower_omp): Rename parallel_nesting_level to taskreg_nesting_level. (expand_omp_parallel): Rename to... (expand_omp_taskreg): ... this. Use OMP_TASKREG_* macros where needed. Call omp_task_call for OMP_TASK regions. (expand_omp): Adjust caller, handle OMP_TASK. (lower_omp_1): Adjust lower_omp_taskreg caller, handle OMP_TASK. * bitmap.c (bitmap_default_obstack_depth): New variable. (bitmap_obstack_initialize, bitmap_obstack_release): Do nothing if argument is NULL and bitmap_default_obstack is already initialized. * ipa-struct-reorg.c (do_reorg_1): Call bitmap_obstack_release at the end. * matrix-reorg.c (matrix_reorg): Likewise. cp/ * cp-tree.h (cxx_omp_finish_clause, cxx_omp_create_clause_info, dependent_omp_for_p, begin_omp_task, finish_omp_task, finish_omp_taskwait): New prototypes. (cxx_omp_clause_default_ctor): Add outer argument. (finish_omp_for): Add new clauses argument. * cp-gimplify.c (cxx_omp_finish_clause): New function. (cxx_omp_predetermined_sharing): Moved from semantics.c, rewritten. (cxx_omp_clause_default_ctor): Add outer argument. (cp_genericize_r): Walk OMP_CLAUSE_LASTPRIVATE_STMT. * cp-objcp-common.h (LANG_HOOKS_OMP_FINISH_CLAUSE): Define. * parser.c (cp_parser_omp_for_loop): Parse collapsed for loops. Add par_clauses argument. If decl is present in parallel's lastprivate clause, change that clause to shared and add a lastprivate clause for decl to OMP_FOR_CLAUSES. Fix wording of error messages. Adjust finish_omp_for caller. Add clauses argument. Parse loops with random access iterators. (cp_parser_omp_clause_collapse, cp_parser_omp_clause_untied): New functions. (cp_parser_omp_for, cp_parser_omp_parallel): Adjust cp_parser_omp_for_loop callers. (cp_parser_omp_for_cond, cp_parser_omp_for_incr): New helper functions. (cp_parser_omp_clause_name): Handle collapse and untied clauses. (cp_parser_omp_clause_schedule): Handle auto schedule. (cp_parser_omp_all_clauses): Handle PRAGMA_OMP_CLAUSE_COLLAPSE and PRAGMA_OMP_CLAUSE_UNTIED. (OMP_FOR_CLAUSE_MASK): Add PRAGMA_OMP_CLAUSE_COLLAPSE. (OMP_TASK_CLAUSE_MASK): Define. (cp_parser_omp_task, cp_parser_omp_taskwait): New functions. (cp_parser_omp_construct): Handle PRAGMA_OMP_TASK. (cp_parser_pragma): Handle PRAGMA_OMP_TASK and PRAGMA_OMP_TASKWAIT. * pt.c (tsubst_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Handle OMP_CLAUSE_LASTPRIVATE_STMT. (tsubst_omp_for_iterator): New function. (dependent_omp_for_p): New function. (tsubst_expr) <case OMP_FOR>: Use it. Handle collapsed OMP_FOR loops. Adjust finish_omp_for caller. Handle loops with random access iterators. Adjust for OMP_FOR_{INIT,COND,INCR} changes. (tsubst_expr): Handle OMP_TASK. * semantics.c (cxx_omp_create_clause_info): New function. (finish_omp_clauses): Call it. Handle OMP_CLAUSE_UNTIED and OMP_CLAUSE_COLLAPSE. (cxx_omp_predetermined_sharing): Removed. * semantics.c (finish_omp_for): Allow pointer iterators. Use handle_omp_for_class_iterator and dependent_omp_for_p. Handle collapsed for loops. Adjust c_finish_omp_for caller. Add new clauses argument. Fix check for type dependent cond or incr. Set OMP_FOR_CLAUSES to clauses. Use cp_convert instead of fold_convert to convert incr amount to difference_type. Only fold if not in template. If decl is mentioned in lastprivate clause, set OMP_CLAUSE_LASTPRIVATE_STMT. Handle loops with random access iterators. Adjust for OMP_FOR_{INIT,COND,INCR} changes. (finish_omp_threadprivate): Allow static class members of the current class. (handle_omp_for_class_iterator, begin_omp_task, finish_omp_task, finish_omp_taskwait): New functions. * parser.c (cp_parser_binary_expression): Add prec argument. (cp_parser_assignment_expression): Adjust caller. * cp-tree.h (outer_curly_brace_block): New prototype. * decl.c (outer_curly_brace_block): No longer static. fortran/ * scanner.c (skip_free_comments, skip_fixed_comments): Handle tabs. * parse.c (next_free): Allow tab after !$omp. (decode_omp_directive): Handle !$omp task, !$omp taskwait and !$omp end task. (case_executable): Add ST_OMP_TASKWAIT. (case_exec_markers): Add ST_OMP_TASK. (gfc_ascii_statement): Handle ST_OMP_TASK, ST_OMP_END_TASK and ST_OMP_TASKWAIT. (parse_omp_structured_block, parse_executable): Handle ST_OMP_TASK. * gfortran.h (gfc_find_sym_in_expr): New prototype. (gfc_statement): Add ST_OMP_TASK, ST_OMP_END_TASK and ST_OMP_TASKWAIT. (gfc_omp_clauses): Add OMP_SCHED_AUTO to sched_kind, OMP_DEFAULT_FIRSTPRIVATE to default_sharing. Add collapse and untied fields. (gfc_exec_op): Add EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * f95-lang.c (LANG_HOOKS_OMP_CLAUSE_COPY_CTOR, LANG_HOOKS_OMP_CLAUSE_ASSIGN_OP, LANG_HOOKS_OMP_CLAUSE_DTOR, LANG_HOOKS_OMP_PRIVATE_OUTER_REF): Define. * trans.h (gfc_omp_clause_default_ctor): Add another argument. (gfc_omp_clause_copy_ctor, gfc_omp_clause_assign_op, gfc_omp_clause_dtor, gfc_omp_private_outer_ref): New prototypes. * types.def (BT_ULONGLONG, BT_PTR_ULONGLONG, BT_FN_BOOL_ULONGLONGPTR_ULONGLONGPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_VOID_PTR_PTR, BT_PTR_FN_VOID_PTR_PTR, BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): New. (BT_BOOL): Use integer type with BOOL_TYPE_SIZE rather than boolean_type_node. * dump-parse-tree.c (gfc_show_omp_node): Handle EXEC_OMP_TASK, EXEC_OMP_TASKWAIT, OMP_SCHED_AUTO, OMP_DEFAULT_FIRSTPRIVATE, untied and collapse clauses. (gfc_show_code_node): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * trans.c (gfc_trans_code): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * st.c (gfc_free_statement): Likewise. * resolve.c (gfc_resolve_blocks, resolve_code): Likewise. (find_sym_in_expr): Rename to... (gfc_find_sym_in_expr): ... this. No longer static. (resolve_allocate_expr, resolve_ordinary_assign): Adjust caller. * match.h (gfc_match_omp_task, gfc_match_omp_taskwait): New prototypes. * openmp.c (resolve_omp_clauses): Allow allocatable arrays in firstprivate, lastprivate, reduction, copyprivate and copyin clauses. (omp_current_do_code): Made static. (omp_current_do_collapse): New variable. (gfc_resolve_omp_do_blocks): Compute omp_current_do_collapse, clear omp_current_do_code and omp_current_do_collapse on return. (gfc_resolve_do_iterator): Handle collapsed do loops. (resolve_omp_do): Likewise, diagnose errorneous collapsed do loops. (OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_UNTIED): Define. (gfc_match_omp_clauses): Handle default (firstprivate), schedule (auto), untied and collapse (n) clauses. (OMP_DO_CLAUSES): Add OMP_CLAUSE_COLLAPSE. (OMP_TASK_CLAUSES): Define. (gfc_match_omp_task, gfc_match_omp_taskwait): New functions. * trans-openmp.c (gfc_omp_private_outer_ref): New function. (gfc_omp_clause_default_ctor): Add outer argument. For allocatable arrays allocate them with the bounds of the outer var if outer var is allocated. (gfc_omp_clause_copy_ctor, gfc_omp_clause_assign_op, gfc_omp_clause_dtor): New functions. (gfc_trans_omp_array_reduction): If decl is allocatable array, allocate it with outer var's bounds in OMP_CLAUSE_REDUCTION_INIT and deallocate it in OMP_CLAUSE_REDUCTION_MERGE. (gfc_omp_predetermined_sharing): Return OMP_CLAUSE_DEFAULT_SHARED for assumed-size arrays. (gfc_trans_omp_do): Add par_clauses argument. If dovar is present in lastprivate clause and do loop isn't simple, set OMP_CLAUSE_LASTPRIVATE_STMT. If dovar is present in parallel's lastprivate clause, change it to shared and add lastprivate clause to OMP_FOR_CLAUSES. Handle collapsed do loops. (gfc_trans_omp_directive): Adjust gfc_trans_omp_do callers. (gfc_trans_omp_parallel_do): Likewise. Move collapse clause to OMP_FOR from OMP_PARALLEL. (gfc_trans_omp_clauses): Handle OMP_SCHED_AUTO, OMP_DEFAULT_FIRSTPRIVATE, untied and collapse clauses. (gfc_trans_omp_task, gfc_trans_omp_taskwait): New functions. (gfc_trans_omp_directive): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. gcc/testsuite/ * gcc.dg/gomp/collapse-1.c: New test. * gcc.dg/gomp/nesting-1.c: New test. * g++.dg/gomp/task-1.C: New test. * g++.dg/gomp/predetermined-1.C: New test. * g++.dg/gomp/tls-4.C: New test. * gfortran.dg/gomp/collapse1.f90: New test. * gfortran.dg/gomp/sharing-3.f90: New test. * gcc.dg/gomp/pr27499.c (foo): Remove is unsigned dg-warning. * g++.dg/gomp/pr27499.C (foo): Likewise. * g++.dg/gomp/for-16.C (foo): Likewise. * g++.dg/gomp/tls-3.C: Remove dg-error, add S::s definition. * g++.dg/gomp/pr34607.C: Adjust dg-error location. * g++.dg/gomp/for-16.C (foo): Add a new dg-error. * gcc.dg/gomp/appendix-a/a.35.4.c: Add dg-warning. * gcc.dg/gomp/appendix-a/a.35.6.c: Likewise. * gfortran.dg/gomp/appendix-a/a.35.4.f90: Likewise. * gfortran.dg/gomp/appendix-a/a.35.6.f90: Likewise. * gfortran.dg/gomp/omp_parse1.f90: Remove !$omp tab test. * gfortran.dg/gomp/appendix-a/a.33.4.f90: Remove dg-error about allocatable array. * gfortran.dg/gomp/reduction1.f90: Likewise. libgomp/ * configure.ac (LIBGOMP_GNU_SYMBOL_VERSIONING): New AC_DEFINE. Substitute also OMP_*LOCK_25*. * configure: Regenerated. * config.h.in: Regenerated. * Makefile.am (libgomp_la_SOURCES): Add loop_ull.c, iter_ull.c, ptrlock.c and task.c. * Makefile.in: Regenerated. * testsuite/Makefile.in: Regenerated. * task.c: New file. * loop_ull.c: New file. * iter_ull.c: New file. * libgomp.h: Include ptrlock.h. (enum gomp_task_kind): New type. (struct gomp_team): Add task_lock, task_queue, task_count, task_running_count, single_count fields. Add work_share_list_free_lock ifndef HAVE_SYNC_BUILTINS. Remove work_share_lock, generation_mask, oldest_live_gen, num_live_gen and init_work_shares fields, add work work_share_list_alloc, work_share_list_free and work_share_chunk fields. Change work_shares from pointer to pointers into an array. Change ordered_release field into gomp_sem_t ** from flexible array member. Add implicit_task and initial_work_shares fields. Move close to the end of the struct. (struct gomp_team_state): Add single_count, last_work_share, active_level and level fields, remove work_share_generation. (gomp_barrier_handle_tasks): New prototype. (gomp_finish_task): New inline function. (struct gomp_work_share): Move chunk_size, end, incr into transparent union/struct, add chunk_size_ull, end_ll, incr_ll and next_ll fields. Reshuffle fields. Add next_alloc, next_ws, next_free and inline_ordered_team_ids fields, change ordered_team_ids into pointer from flexible array member. Add mode field. Put lock and next into a different cache line from most of the write-once fields. (gomp_iter_ull_static_next, gomp_iter_ull_dynamic_next_locked, gomp_iter_ull_guided_next_locked, gomp_iter_ull_dynamic_next, gomp_iter_ull_guided_next): New prototypes. (gomp_new_icv): New prototype. (struct gomp_thread): Add thread_pool and task fields. (struct gomp_thread_pool): New type. (gomp_new_team): New prototype. (gomp_team_start): Change type of last argument. (gomp_new_work_share): Removed. (gomp_init_work_share, gomp_fini_work_share): New prototypes. (gomp_work_share_init_done): New static inline. (gomp_throttled_spin_count_var, gomp_available_cpus, gomp_managed_threads): New extern decls. (gomp_init_task): New prototype. (gomp_spin_count_var): New extern var decl. (LIBGOMP_GNU_SYMBOL_VERSIONING): Undef if no visibility or no alias support, or if not PIC. (gomp_init_lock_30, gomp_destroy_lock_30, gomp_set_lock_30, gomp_unset_lock_30, gomp_test_lock_30, gomp_init_nest_lock_30, gomp_destroy_nest_lock_30, gomp_set_nest_lock_30, gomp_unset_nest_lock_30, gomp_test_nest_lock_30, gomp_init_lock_25, gomp_destroy_lock_25, gomp_set_lock_25, gomp_unset_lock_25, gomp_test_lock_25, gomp_init_nest_lock_25, gomp_destroy_nest_lock_25, gomp_set_nest_lock_25, gomp_unset_nest_lock_25, gomp_test_nest_lock_25): New prototypes. (omp_lock_symver, strong_alias): Define. (gomp_remaining_threads_count, gomp_remaining_threads_lock): New decls. (gomp_end_task): New. (struct gomp_task_icv, gomp_global_icv): New. (gomp_thread_limit_var, gomp_max_active_levels_var): New. (struct gomp_task): New. (gomp_nthreads_var, gomp_dyn_var, gomp_nest_var, gomp_run_sched_var, gomp_run_sched_chunk): Remove. (gomp_icv): New. (gomp_schedule_type): Reorder enum to match omp_sched_t. * team.c (struct gomp_thread_start_data): Add thread_pool and task fields. (gomp_thread_start): Add gomp_team_barrier_wait call. For non-nested case remove clearing of docked thread thr fields. Use pool fields instead of global gomp_* variables. Use gomp_barrier_wait_last when needed. Initialize ts.active_level. Create tasks for each member thread. (free_team): Only destroy team barrier, task_lock here and free it. (gomp_free_thread): Free last_team if non-NULL. (gomp_team_end): Call gomp_team_barrier_wait instead of gomp_barrier_wait. For nested case call one extra gomp_barrier_wait. Move here some destruction from free_team. Call free_team on pool->last_team if any, rather than freeing current team. Destroy work_share_list_free_lock ifndef HAVE_SYNC_BUILTINS. (gomp_new_icv): New function. (gomp_threads, gomp_threads_size, gomp_threads_used, gomp_threads_dock): Removed. (gomp_thread_destructor): New variable. (gomp_new_thread_pool, gomp_free_pool_helper, gomp_free_thread): New functions. (gomp_team_start): Create new pool if current thread doesn't have one. Use pool fields instead of global gomp_* variables. Initialize thread_pool field for new threads. Clear single_count. Change last argument from ws to team, don't create new team, set ts.work_share to &team->work_shares[0] and clear ts.last_work_share. Don't clear ts.work_share_generation. If number of threads changed, adjust atomically gomp_managed_threads. Use gomp_init_task instead of gomp_new_task, set thr->task to the corresponding implicit_task array entry. Create tasks for each member thread. Initialize ts.level. (initialize_team): Call pthread_key_create on gomp_thread_destructor. (team_destructor): New function. (new_team): Removed. (gomp_new_team): New function. (free_team): Free gomp_work_share blocks chained through next_alloc, instead of freeing work_shares and destroying work_share_lock. (gomp_team_end): Call gomp_fini_work_share. If number of threads changed, adjust atomically gomp_managed_threads. Use gomp_end_task. * barrier.c (GOMP_barrier): Call gomp_team_barrier_wait instead of gomp_barrier_wait. * single.c (GOMP_single_copy_start): Call gomp_team_barrier_wait instead of gomp_barrier_wait. Call gomp_work_share_init_done if gomp_work_share_start returned true. Don't unlock ws->lock. (GOMP_single_copy_end): Call gomp_team_barrier_wait instead of gomp_barrier_wait. (GOMP_single_start): Rewritten if HAVE_SYNC_BUILTINS. Call gomp_work_share_init_done if gomp_work_share_start returned true. Don't unlock ws->lock. * work.c: Include stddef.h. (free_work_share): Use work_share_list_free_lock instead of atomic chaining ifndef HAVE_SYNC_BUILTINS. Add team argument. Call gomp_fini_work_share and then either free ws if orphaned, or put it into work_share_list_free list of the current team. (alloc_work_share, gomp_init_work_share, gomp_fini_work_share): New functions. (gomp_work_share_start, gomp_work_share_end, gomp_work_share_end_nowait): Rewritten. * omp_lib.f90.in Change some tabs to spaces to prevent warnings. (openmp_version): Set to 200805. (omp_sched_kind, omp_sched_static, omp_sched_dynamic, omp_sched_guided, omp_sched_auto): New parameters. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New interfaces. * omp_lib.h.in (openmp_version): Set to 200805. (omp_sched_kind, omp_sched_static, omp_sched_dynamic, omp_sched_guided, omp_sched_auto): New parameters. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New externals. * loop.c: Include limits.h. (GOMP_loop_runtime_next, GOMP_loop_ordered_runtime_next): Handle GFS_AUTO. (GOMP_loop_runtime_start, GOMP_loop_ordered_runtime_start): Likewise. Use gomp_icv. (gomp_loop_static_start, gomp_loop_dynamic_start): Clear ts.static_trip here. (gomp_loop_static_start, gomp_loop_ordered_static_start): Call gomp_work_share_init_done after gomp_loop_init. Don't unlock ws->lock. (gomp_loop_dynamic_start, gomp_loop_guided_start): Call gomp_work_share_init_done after gomp_loop_init. If HAVE_SYNC_BUILTINS, don't unlock ws->lock, otherwise lock it. (gomp_loop_ordered_dynamic_start, gomp_loop_ordered_guided_start): Call gomp_work_share_init_done after gomp_loop_init. Lock ws->lock. (gomp_parallel_loop_start): Call gomp_new_team instead of gomp_new_work_share. Call gomp_loop_init on &team->work_shares[0]. Adjust gomp_team_start caller. Pass 0 as second argument to gomp_resolve_num_threads. (gomp_loop_init): For GFS_DYNAMIC, multiply ws->chunk_size by incr. If adding ws->chunk_size nthreads + 1 times after end won't overflow, set ws->mode to 1. * libgomp_g.h (GOMP_loop_ull_static_start, GOMP_loop_ull_dynamic_start, GOMP_loop_ull_guided_start, GOMP_loop_ull_runtime_start, GOMP_loop_ull_ordered_static_start, GOMP_loop_ull_ordered_dynamic_start, GOMP_loop_ull_ordered_guided_start, GOMP_loop_ull_ordered_runtime_start, GOMP_loop_ull_static_next, GOMP_loop_ull_dynamic_next, GOMP_loop_ull_guided_next, GOMP_loop_ull_runtime_next, GOMP_loop_ull_ordered_static_next, GOMP_loop_ull_ordered_dynamic_next, GOMP_loop_ull_ordered_guided_next, GOMP_loop_ull_ordered_runtime_next, GOMP_task, GOMP_taskwait): New prototypes. * libgomp.map: Export lock routines also @@OMP_2.0. (GOMP_loop_ordered_dynamic_first, GOMP_loop_ordered_guided_first, GOMP_loop_ordered_runtime_first, GOMP_loop_ordered_static_first): Remove. (GOMP_loop_ull_dynamic_next, GOMP_loop_ull_dynamic_start, GOMP_loop_ull_guided_next, GOMP_loop_ull_guided_start, GOMP_loop_ull_ordered_dynamic_next, GOMP_loop_ull_ordered_dynamic_start, GOMP_loop_ull_ordered_guided_next, GOMP_loop_ull_ordered_guided_start, GOMP_loop_ull_ordered_runtime_next, GOMP_loop_ull_ordered_runtime_start, GOMP_loop_ull_ordered_static_next, GOMP_loop_ull_ordered_static_start, GOMP_loop_ull_runtime_next, GOMP_loop_ull_runtime_start, GOMP_loop_ull_static_next, GOMP_loop_ull_static_start, GOMP_task, GOMP_taskwait): Export @@GOMP_2.0. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level, omp_set_schedule_, omp_set_schedule_8_, omp_get_schedule_, omp_get_schedule_8_, omp_get_thread_limit_, omp_set_max_active_levels_, omp_set_max_active_levels_8_, omp_get_max_active_levels_, omp_get_level_, omp_get_ancestor_thread_num_, omp_get_ancestor_thread_num_8_, omp_get_team_size_, omp_get_team_size_8_, omp_get_active_level_): New exports @@OMP_3.0. * omp.h.in (omp_sched_t): New type. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New prototypes. * env.c (gomp_spin_count_var, gomp_throttled_spin_count_var, gomp_available_cpus, gomp_managed_threads, gomp_max_active_levels_var, gomp_thread_limit_var, gomp_remaining_threads_count, gomp_remaining_threads_lock): New variables. (parse_spincount): New function. (initialize_env): Call gomp_init_num_threads unconditionally. Initialize gomp_available_cpus. Call parse_spincount, initialize gomp_{,throttled_}spin_count_var depending on presence and value of OMP_WAIT_POLICY and GOMP_SPINCOUNT env vars. Handle GOMP_BLOCKTIME env var. Handle OMP_WAIT_POLICY, OMP_MAX_ACTIVE_LEVELS, OMP_THREAD_LIMIT, OMP_STACKSIZE env vars. Handle unit specification for GOMP_STACKSIZE. Initialize gomp_remaining_threads_count and gomp_remaining_threads_lock if needed. Use gomp_global_icv. (gomp_nthreads_var, gomp_dyn_var, gomp_nest_var, gomp_run_sched_var, gomp_run_sched_chunk): Remove. (gomp_global_icv): New. (parse_schedule): Use it. Parse "auto". (omp_set_num_threads): Use gomp_icv. (omp_set_dynamic, omp_get_dynamic, omp_set_nested, omp_get_nested): Likewise. (omp_get_max_threads): Move from parallel.c. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels): New functions, add ialias. (parse_stacksize, parse_wait_policy): New functions. * fortran.c: Rewrite lock wrappers, if symbol versioning provide both wrappers for compatibility and new locks. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New ialias_redirect. (omp_set_schedule_, omp_set_schedule_8_, omp_get_schedule_, omp_get_schedule_8_, omp_get_thread_limit_, omp_set_max_active_levels_, omp_set_max_active_levels_8_, omp_get_max_active_levels_, omp_get_level_, omp_get_ancestor_thread_num_, omp_get_ancestor_thread_num_8_, omp_get_team_size_, omp_get_team_size_8_, omp_get_active_level_): New functions. * parallel.c: Include limits.h. (gomp_resolve_num_threads): Add count argument. Rewritten. (GOMP_parallel_start): Call gomp_new_team and pass that as last argument to gomp_team_start. Pass 0 as second argument to gomp_resolve_num_threads. (GOMP_parallel_end): Decrease gomp_remaining_threads_count if gomp_thread_limit_var != ULONG_MAX. (omp_in_parallel): Implement using ts.active_level. (omp_get_max_threads): Move to env.c. (omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New functions, add ialias. * sections.c (GOMP_sections_start): Call gomp_work_share_init_done after gomp_sections_init. If HAVE_SYNC_BUILTINS, call gomp_iter_dynamic_next instead of the _locked variant and don't take lock around it, otherwise acquire it before calling gomp_iter_dynamic_next_locked. (GOMP_sections_next): If HAVE_SYNC_BUILTINS, call gomp_iter_dynamic_next instead of the _locked variant and don't take lock around it. (GOMP_parallel_sections_start): Call gomp_new_team instead of gomp_new_work_share. Call gomp_sections_init on &team->work_shares[0]. Adjust gomp_team_start caller. Pass count as second argument to gomp_resolve_num_threads, don't adjust num_threads after the call. Use gomp_icv. * iter.c (gomp_iter_dynamic_next_locked): Don't multiply ws->chunk_size by incr. (gomp_iter_dynamic_next): Likewise. If ws->mode, use more efficient code. * libgomp_f.h.in (omp_lock_25_arg_t, omp_nest_lock_25_arg_t): New types. (omp_lock_25_arg, omp_nest_lock_25_arg): New macros. (omp_check_defines): Check even the compat defines. * config/linux/ptrlock.c: New file. * config/linux/ptrlock.h: New file. * config/linux/wait.h: New file. * config/posix/ptrlock.c: New file. * config/posix/ptrlock.h: New file. * config/linux/bar.h (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New prototypes. (gomp_team_barrier_set_task_pending, gomp_team_barrier_clear_task_pending, gomp_team_barrier_set_waiting_for_tasks, gomp_team_barrier_waiting_for_tasks, gomp_team_barrier_done): New inlines. (gomp_barrier_t): Rewritten. (gomp_barrier_state_t): New typedef. (gomp_barrier_init, gomp_barrier_reinit, gomp_barrier_destroy, gomp_barrier_wait_start): Rewritten. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_barrier_last_thread, gomp_barrier_wait_last): New static inlines. * config/linux/bar.c: Include wait.h instead of libgomp.h and futex.h. (gomp_barrier_wait_end): Rewritten. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake, gomp_barrier_wait_last): New functions. * config/posix/bar.h (gomp_barrier_t): Add generation field. (gomp_barrier_state_t): New typedef. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New prototypes. (gomp_barrier_wait_start): Or all but low 2 bits from generation into the return value. Return gomp_barrier_state_t. (gomp_team_barrier_set_task_pending, gomp_team_barrier_clear_task_pending, gomp_team_barrier_set_waiting_for_tasks, gomp_team_barrier_waiting_for_tasks, gomp_team_barrier_done): New inlines. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_barrier_last_thread, gomp_barrier_wait_last): New static inlines. * config/posix/bar.c (gomp_barrier_init): Clear generation field. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New functions. * config/linux/mutex.c: Include wait.h instead of libgomp.h and futex.h. (gomp_futex_wake, gomp_futex_wait): New variables. (gomp_mutex_lock_slow): Call do_wait instead of futex_wait. * config/linux/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. Include wait.h instead of libgomp.h and futex.h. (gomp_set_nest_lock_25): Call do_wait instead of futex_wait. * config/posix95/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. * config/posix/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. * config/linux/proc.c (gomp_init_num_threads): Use gomp_global_icv. (get_num_procs, gomp_dynamic_max_threads): Use gomp_icv. * config/posix/proc.c, config/mingw32/proc.c: Similarly. * config/linux/powerpc/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/alpha/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/x86/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/s390/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/ia64/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/sparc/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/sem.c: Include wait.h instead of libgomp.h and futex.h. (gomp_sem_wait_slow): Call do_wait instead of futex_wait. * config/linux/affinity.c: Assume HAVE_SYNC_BUILTINS. * config/linux/omp-lock.h (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_nest_lock_t): Change owner into void *, add lock field. * config/posix95/omp-lock.h: Include semaphore.h. (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_lock_t): Use sem_t instead of mutex if semaphores aren't broken. (omp_nest_lock_t): Likewise. Change owner to void *. * config/posix/omp-lock.h: Include semaphore.h. (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_lock_t): Use sem_t instead of mutex if semaphores aren't broken. (omp_nest_lock_t): Likewise. Add owner field. * testsuite/libgomp.c/collapse-1.c: New test. * testsuite/libgomp.c/collapse-2.c: New test. * testsuite/libgomp.c/collapse-3.c: New test. * testsuite/libgomp.c/icv-1.c: New test. * testsuite/libgomp.c/icv-2.c: New test. * testsuite/libgomp.c/lib-2.c: New test. * testsuite/libgomp.c/lock-1.c: New test. * testsuite/libgomp.c/lock-2.c: New test. * testsuite/libgomp.c/lock-3.c: New test. * testsuite/libgomp.c/loop-4.c: New test. * testsuite/libgomp.c/loop-5.c: New test. * testsuite/libgomp.c/loop-6.c: New test. * testsuite/libgomp.c/loop-7.c: New test. * testsuite/libgomp.c/loop-8.c: New test. * testsuite/libgomp.c/loop-9.c: New test. * testsuite/libgomp.c/nested-3.c: New test. * testsuite/libgomp.c/nestedfn-6.c: New test. * testsuite/libgomp.c/sort-1.c: New test. * testsuite/libgomp.c/task-1.c: New test. * testsuite/libgomp.c/task-2.c: New test. * testsuite/libgomp.c/task-3.c: New test. * testsuite/libgomp.c/task-4.c: New test. * testsuite/libgomp.c++/c++.exp: Add libstdc++-v3 build includes to C++ testsuite default compiler options. * testsuite/libgomp.c++/collapse-1.C: New test. * testsuite/libgomp.c++/collapse-2.C: New test. * testsuite/libgomp.c++/ctor-10.C: New test. * testsuite/libgomp.c++/for-1.C: New test. * testsuite/libgomp.c++/for-2.C: New test. * testsuite/libgomp.c++/for-3.C: New test. * testsuite/libgomp.c++/for-4.C: New test. * testsuite/libgomp.c++/for-5.C: New test. * testsuite/libgomp.c++/loop-8.C: New test. * testsuite/libgomp.c++/loop-9.C: New test. * testsuite/libgomp.c++/loop-10.C: New test. * testsuite/libgomp.c++/task-1.C: New test. * testsuite/libgomp.c++/task-2.C: New test. * testsuite/libgomp.c++/task-3.C: New test. * testsuite/libgomp.c++/task-4.C: New test. * testsuite/libgomp.c++/task-5.C: New test. * testsuite/libgomp.c++/task-6.C: New test. * testsuite/libgomp.fortran/allocatable1.f90: New test. * testsuite/libgomp.fortran/allocatable2.f90: New test. * testsuite/libgomp.fortran/allocatable3.f90: New test. * testsuite/libgomp.fortran/allocatable4.f90: New test. * testsuite/libgomp.fortran/collapse1.f90: New test. * testsuite/libgomp.fortran/collapse2.f90: New test. * testsuite/libgomp.fortran/collapse3.f90: New test. * testsuite/libgomp.fortran/collapse4.f90: New test. * testsuite/libgomp.fortran/lastprivate1.f90: New test. * testsuite/libgomp.fortran/lastprivate2.f90: New test. * testsuite/libgomp.fortran/lib4.f90: New test. * testsuite/libgomp.fortran/lock-1.f90: New test. * testsuite/libgomp.fortran/lock-2.f90: New test. * testsuite/libgomp.fortran/nested1.f90: New test. * testsuite/libgomp.fortran/nestedfn4.f90: New test. * testsuite/libgomp.fortran/strassen.f90: New test. * testsuite/libgomp.fortran/tabs1.f90: New test. * testsuite/libgomp.fortran/tabs2.f: New test. * testsuite/libgomp.fortran/task1.f90: New test. * testsuite/libgomp.fortran/task2.f90: New test. * testsuite/libgomp.fortran/vla4.f90: Add dg-warning. * testsuite/libgomp.fortran/vla5.f90: Likewise. * testsuite/libgomp.c/pr26943-2.c: Likewise. * testsuite/libgomp.c/pr26943-3.c: Likewise. * testsuite/libgomp.c/pr26943-4.c: Likewise. Co-Authored-By: Jakob Blomer <jakob.blomer@ira.uka.de> Co-Authored-By: Richard Henderson <rth@redhat.com> Co-Authored-By: Ulrich Drepper <drepper@redhat.com> From-SVN: r136433
2008-06-06 15:01:54 +02:00
#pragma omp barrier /* { dg-warning "may not be closely nested" } */
re PR c++/26943 ([gomp] firstprivate + lastprivate uses inefficient barrier) PR c++/26943 * omp-low.c (maybe_lookup_decl_in_outer_ctx): New function. (build_outer_var_ref): Use maybe_lookup_decl_in_outer_ctx to find if var will be a global variable even in the nested context. (omp_copy_decl): Only check for global variable at the end, it might be overridden in outer contexts. (scan_sharing_clauses): For global variables don't create a field. (lower_rec_input_clauses): Do nothing for global shared variables. Emit a barrier at the end of ILIST if there were any decls in both firstprivate and lastprivate clauses. (lower_send_clauses): Do nothing for global variables except for COPYIN. * testsuite/libgomp.c/pr26943-1.c: New test. * testsuite/libgomp.c/pr26943-2.c: New test. * testsuite/libgomp.c/pr26943-3.c: New test. * testsuite/libgomp.c/pr26943-4.c: New test. * testsuite/libgomp.c++/pr27337.C: Remove barrier. * testsuite/libgomp.c++/pr26943.C: New test. From-SVN: r113483
2006-05-02 22:03:38 +02:00
#pragma omp atomic
a += i;
b += i;
c = i;
d = i;
#pragma omp atomic
e[0] += i;
f[0] += i;
g[0] = 'g' + i;
h[0] = 'h' + i;
c-cppbuiltin.c (c_cpp_builtins): Change _OPENMP value to 200805. * c-cppbuiltin.c (c_cpp_builtins): Change _OPENMP value to 200805. * langhooks.h (struct lang_hooks_for_decls): Add omp_finish_clause. Add omp_private_outer_ref hook, add another argument to omp_clause_default_ctor hook. * langhooks-def.h (LANG_HOOKS_OMP_FINISH_CLAUSE): Define. (LANG_HOOKS_OMP_PRIVATE_OUTER_REF): Define. (LANG_HOOKS_OMP_CLAUSE_DEFAULT_CTOR): Change to hook_tree_tree_tree_tree_null. (LANG_HOOKS_DECLS): Add LANG_HOOKS_OMP_FINISH_CLAUSE and LANG_HOOKS_OMP_PRIVATE_OUTER_REF. * hooks.c (hook_tree_tree_tree_tree_null): New function. * hooks.h (hook_tree_tree_tree_tree_null): New prototype. * tree.def (OMP_TASK): New tree code. * tree.h (OMP_TASK_COPYFN, OMP_TASK_ARG_SIZE, OMP_TASK_ARG_ALIGN, OMP_CLAUSE_PRIVATE_OUTER_REF, OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_COLLAPSE_ITERVAR, OMP_CLAUSE_COLLAPSE_COUNT, OMP_TASKREG_CHECK, OMP_TASKREG_BODY, OMP_TASKREG_CLAUSES, OMP_TASKREG_FN, OMP_TASKREG_DATA_ARG, OMP_TASK_BODY, OMP_TASK_CLAUSES, OMP_TASK_FN, OMP_TASK_DATA_ARG, OMP_CLAUSE_COLLAPSE_EXPR): Define. (enum omp_clause_default_kind): Add OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. (OMP_DIRECTIVE_P): Add OMP_TASK. (OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_UNTIED): New clause codes. (OMP_CLAUSE_SCHEDULE_AUTO): New schedule kind. * tree.c (omp_clause_code_name): Add OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED entries. (omp_clause_num_ops): Likewise. Increase OMP_CLAUSE_LASTPRIVATE num_ops to 2. (walk_tree_1): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Walk OMP_CLAUSE_LASTPRIVATE_STMT. * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE_SCHEDULE_AUTO, OMP_CLAUSE_UNTIED, OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. (dump_generic_node): Handle OMP_TASK and collapsed OMP_FOR loops. * c-omp.c (c_finish_omp_for): Allow pointer iterators. Remove warning about unsigned iterators. Change decl/init/cond/incr arguments to TREE_VECs, check arguments for all collapsed loops. (c_finish_omp_taskwait): New function. (c_split_parallel_clauses): Put OMP_CLAUSE_COLLAPSE clause to ws_clauses. * c-parser.c (c_parser_omp_for_loop): Parse collapsed loops. Call default_function_array_conversion on init. Add par_clauses argument. If decl is present in parallel's lastprivate clause, change it to shared and add lastprivate clause for decl to OMP_FOR_CLAUSES. Add clauses argument, on success set OMP_FOR_CLAUSES to it. Look up collapse count in clauses. (c_parser_omp_for, c_parser_omp_parallel): Adjust c_parser_omp_for_loop callers. (OMP_FOR_CLAUSE_MASK): Add 1 << PRAGMA_OMP_CLAUSE_COLLAPSE. (c_parser_pragma): Handle PRAGMA_OMP_TASKWAIT. (c_parser_omp_clause_name): Handle collapse and untied clauses. (c_parser_omp_clause_collapse, c_parser_omp_clause_untied): New functions. (c_parser_omp_clause_schedule): Handle schedule(auto). Include correct location in the error message. (c_parser_omp_all_clauses): Handle PRAGMA_OMP_CLAUSE_COLLAPSE and PRAGMA_OMP_CLAUSE_UNTIED. (OMP_TASK_CLAUSE_MASK): Define. (c_parser_omp_task, c_parser_omp_taskwait): New functions. (c_parser_omp_construct): Handle PRAGMA_OMP_TASK. * tree-nested.c (convert_nonlocal_omp_clauses, convert_local_omp_clauses): Handle OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_REDUCTION_INIT, OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Don't handle TREE_STATIC or DECL_EXTERNAL VAR_DECLs in OMP_CLAUSE_DECL. (conver_nonlocal_reference, convert_local_reference, convert_call_expr): Handle OMP_TASK the same as OMP_PARALLEL. Use OMP_TASKREG_* macros rather than OMP_PARALLEL_*. (walk_omp_for): Adjust for OMP_FOR_{INIT,COND,INCR} changes. * tree-gimple.c (is_gimple_stmt): Handle OMP_TASK. * c-tree.h (c_begin_omp_task, c_finish_omp_task): New prototypes. * c-pragma.h (PRAGMA_OMP_TASK, PRAGMA_OMP_TASKWAIT): New. (PRAGMA_OMP_CLAUSE_COLLAPSE, PRAGMA_OMP_CLAUSE_UNTIED): New. * c-typeck.c (c_begin_omp_task, c_finish_omp_task): New functions. (c_finish_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. * c-pragma.c (init_pragma): Init omp task and omp taskwait pragmas. * c-common.h (c_finish_omp_taskwait): New prototype. * gimple-low.c (lower_stmt): Handle OMP_TASK. * tree-parloops.c (create_parallel_loop): Create 1 entry vectors for OMP_FOR_{INIT,COND,INCR}. * tree-cfg.c (remove_useless_stmts_1): Handle OMP_* containers. (make_edges): Handle OMP_TASK. * tree-ssa-operands.c (get_expr_operands): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. * tree-inline.c (estimate_num_insns_1): Handle OMP_TASK. * builtin-types.def (BT_PTR_ULONGLONG, BT_PTR_FN_VOID_PTR_PTR, BT_FN_BOOL_ULONGLONGPTR_ULONGLONGPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): New. * omp-builtins.def (BUILT_IN_GOMP_TASK, BUILT_IN_GOMP_TASKWAIT, BUILT_IN_GOMP_LOOP_ULL_STATIC_START, BUILT_IN_GOMP_LOOP_ULL_DYNAMIC_START, BUILT_IN_GOMP_LOOP_ULL_GUIDED_START, BUILT_IN_GOMP_LOOP_ULL_RUNTIME_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_STATIC_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_DYNAMIC_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_GUIDED_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_RUNTIME_START, BUILT_IN_GOMP_LOOP_ULL_STATIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_DYNAMIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_GUIDED_NEXT, BUILT_IN_GOMP_LOOP_ULL_RUNTIME_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_STATIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_DYNAMIC_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_GUIDED_NEXT, BUILT_IN_GOMP_LOOP_ULL_ORDERED_RUNTIME_NEXT): New builtins. * gimplify.c (gimplify_omp_for): Allow pointer type for decl, handle POINTER_PLUS_EXPR. If loop counter has been replaced and original iterator is present in lastprivate clause or if collapse > 1, set OMP_CLAUSE_LASTPRIVATE_STMT. Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. (gimplify_expr): Handle OMP_SECTIONS_SWITCH and OMP_TASK. (enum gimplify_omp_var_data): Add GOVD_PRIVATE_OUTER_REF. (omp_notice_variable): Set GOVD_PRIVATE_OUTER_REF if needed, if it is set, lookup var in outer contexts too. Handle OMP_CLAUSE_DEFAULT_FIRSTPRIVATE. Handle vars that are supposed to be implicitly determined firstprivate for task regions. (gimplify_scan_omp_clauses): Set GOVD_PRIVATE_OUTER_REF if needed, if it is set, lookup var in outer contexts too. Set OMP_CLAUSE_PRIVATE_OUTER_REF if GOVD_PRIVATE_OUTER_REF is set. Handle OMP_CLAUSE_LASTPRIVATE_STMT, OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Take region_type as last argument instead of in_parallel and in_combined_parallel. (gimplify_omp_parallel, gimplify_omp_for, gimplify_omp_workshare): Adjust callers. (gimplify_adjust_omp_clauses_1): Set OMP_CLAUSE_PRIVATE_OUTER_REF if GOVD_PRIVATE_OUTER_REF is set. Call omp_finish_clause langhook. (new_omp_context): Set default_kind to OMP_CLAUSE_DEFAULT_UNSPECIFIED for OMP_TASK regions. (omp_region_type): New enum. (struct gimplify_omp_ctx): Remove is_parallel and is_combined_parallel fields, add region_type. (new_omp_context): Take region_type as argument instead of is_parallel and is_combined_parallel. (gimple_add_tmp_var, omp_firstprivatize_variable, omp_notice_variable, omp_is_private, omp_check_private): Adjust ctx->is_parallel and ctx->is_combined_parallel checks. (gimplify_omp_task): New function. (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. * omp-low.c (extract_omp_for_data): Use schedule(static) for schedule(auto). Handle pointer and unsigned iterators. Compute fd->iter_type. Handle POINTER_PLUS_EXPR increments. Add loops argument. Extract data for collapsed OMP_FOR loops. (expand_parallel_call): Assert sched_kind isn't auto, map runtime schedule to index 3. (struct omp_for_data_loop): New type. (struct omp_for_data): Remove v, n1, n2, step, cond_code fields. Add loop, loops, collapse and iter_type fields. (workshare_safe_to_combine_p): Disallow combined for if iter_type is unsigned long long. Don't combine collapse > 1 loops unless all bounds and steps are constant. Adjust extract_omp_for_data caller. (expand_omp_for_generic): Handle pointer, unsigned and long long iterators. Handle collapsed OMP_FOR loops. Adjust for struct omp_for_data changes. If libgomp function doesn't return boolean_type_node, add comparison of the return value with 0. (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle pointer, unsigned and long long iterators. Adjust for struct omp_for_data changes. (expand_omp_for): Assert sched_kind isn't auto, map runtime schedule to index 3. Use GOMP_loop_ull*{start,next} if iter_type is unsigned long long. Allocate loops array, pass it to extract_omp_for_data. For collapse > 1 loops use always expand_omp_for_generic. (omp_context): Add sfield_map and srecord_type fields. (is_task_ctx, lookup_sfield): New functions. (use_pointer_for_field): Use is_task_ctx helper. Change first argument's type from const_tree to tree. Clarify comment. In OMP_TASK disallow copy-in/out sharing. (build_sender_ref): Call lookup_sfield instead of lookup_field. (install_var_field): Add mask argument. Populate both record_type and srecord_type if needed. (delete_omp_context): Destroy sfield_map, clear DECL_ABSTRACT_ORIGIN in srecord_type. (fixup_child_record_type): Also remap FIELD_DECL's DECL_SIZE{,_UNIT} and DECL_FIELD_OFFSET. (scan_sharing_clauses): Adjust install_var_field callers. For firstprivate clauses on explicit tasks allocate the var by value in record_type unconditionally, rather than by reference. Handle OMP_CLAUSE_PRIVATE_OUTER_REF. Scan OMP_CLAUSE_LASTPRIVATE_STMT. Use is_taskreg_ctx instead of is_parallel_ctx. Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. (create_omp_child_function_name): Add task_copy argument, use *_omp_cpyfn* names if it is true. (create_omp_child_function): Add task_copy argument, if true create *_omp_cpyfn* helper function. (scan_omp_parallel): Adjust create_omp_child_function callers. Rename parallel_nesting_level to taskreg_nesting_level. (scan_omp_task): New function. (lower_rec_input_clauses): Don't run constructors for firstprivate explicit task vars which are initialized by *_omp_cpyfn*. Pass outer var ref to omp_clause_default_ctor hook if OMP_CLAUSE_PRIVATE_OUTER_REF or OMP_CLAUSE_LASTPRIVATE. Replace OMP_CLAUSE_REDUCTION_PLACEHOLDER decls in OMP_CLAUSE_REDUCTION_INIT. (lower_send_clauses): Clear DECL_ABSTRACT_ORIGIN if in task to avoid duplicate setting of fields. Handle OMP_CLAUSE_PRIVATE_OUTER_REF. (lower_send_shared_vars): Use srecord_type if non-NULL. Don't copy-out if TREE_READONLY, only copy-in. (expand_task_copyfn): New function. (expand_task_call): New function. (struct omp_taskcopy_context): New type. (task_copyfn_copy_decl, task_copyfn_remap_type, create_task_copyfn): New functions. (lower_omp_parallel): Rename to... (lower_omp_taskreg): ... this. Use OMP_TASKREG_* macros where needed. Call create_task_copyfn if srecord_type is needed. Adjust sender_decl type. (task_shared_vars): New variable. (check_omp_nesting_restrictions): Warn if work-sharing, barrier, master or ordered region is closely nested inside OMP_TASK. Add warnings for barrier if closely nested inside of work-sharing, ordered, or master region. (scan_omp_1): Call check_omp_nesting_restrictions even for GOMP_barrier calls. Rename parallel_nesting_level to taskreg_nesting_level. Handle OMP_TASK. (lower_lastprivate_clauses): Even if some lastprivate is found on a work-sharing construct, continue looking for them on parent parallel construct. (lower_omp_for_lastprivate): Add lastprivate clauses to the beginning of dlist rather than end. Adjust for struct omp_for_data changes. (lower_omp_for): Add rec input clauses before OMP_FOR_PRE_BODY, not after it. Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes, adjust extract_omp_for_data caller. (get_ws_args_for): Adjust extract_omp_for_data caller. (scan_omp_for): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. (lower_omp_single_simple): If libgomp function doesn't return boolean_type_node, add comparison of the return value with 0. (diagnose_sb_1, diagnose_sb_2): Handle collapsed OMP_FOR loops, adjust for OMP_FOR_{INIT,COND,INCR} changes. Handle OMP_TASK. (parallel_nesting_level): Rename to... (taskreg_nesting_level): ... this. (is_taskreg_ctx): New function. (build_outer_var_ref, omp_copy_decl): Use is_taskreg_ctx instead of is_parallel_ctx. (execute_lower_omp): Rename parallel_nesting_level to taskreg_nesting_level. (expand_omp_parallel): Rename to... (expand_omp_taskreg): ... this. Use OMP_TASKREG_* macros where needed. Call omp_task_call for OMP_TASK regions. (expand_omp): Adjust caller, handle OMP_TASK. (lower_omp_1): Adjust lower_omp_taskreg caller, handle OMP_TASK. * bitmap.c (bitmap_default_obstack_depth): New variable. (bitmap_obstack_initialize, bitmap_obstack_release): Do nothing if argument is NULL and bitmap_default_obstack is already initialized. * ipa-struct-reorg.c (do_reorg_1): Call bitmap_obstack_release at the end. * matrix-reorg.c (matrix_reorg): Likewise. cp/ * cp-tree.h (cxx_omp_finish_clause, cxx_omp_create_clause_info, dependent_omp_for_p, begin_omp_task, finish_omp_task, finish_omp_taskwait): New prototypes. (cxx_omp_clause_default_ctor): Add outer argument. (finish_omp_for): Add new clauses argument. * cp-gimplify.c (cxx_omp_finish_clause): New function. (cxx_omp_predetermined_sharing): Moved from semantics.c, rewritten. (cxx_omp_clause_default_ctor): Add outer argument. (cp_genericize_r): Walk OMP_CLAUSE_LASTPRIVATE_STMT. * cp-objcp-common.h (LANG_HOOKS_OMP_FINISH_CLAUSE): Define. * parser.c (cp_parser_omp_for_loop): Parse collapsed for loops. Add par_clauses argument. If decl is present in parallel's lastprivate clause, change that clause to shared and add a lastprivate clause for decl to OMP_FOR_CLAUSES. Fix wording of error messages. Adjust finish_omp_for caller. Add clauses argument. Parse loops with random access iterators. (cp_parser_omp_clause_collapse, cp_parser_omp_clause_untied): New functions. (cp_parser_omp_for, cp_parser_omp_parallel): Adjust cp_parser_omp_for_loop callers. (cp_parser_omp_for_cond, cp_parser_omp_for_incr): New helper functions. (cp_parser_omp_clause_name): Handle collapse and untied clauses. (cp_parser_omp_clause_schedule): Handle auto schedule. (cp_parser_omp_all_clauses): Handle PRAGMA_OMP_CLAUSE_COLLAPSE and PRAGMA_OMP_CLAUSE_UNTIED. (OMP_FOR_CLAUSE_MASK): Add PRAGMA_OMP_CLAUSE_COLLAPSE. (OMP_TASK_CLAUSE_MASK): Define. (cp_parser_omp_task, cp_parser_omp_taskwait): New functions. (cp_parser_omp_construct): Handle PRAGMA_OMP_TASK. (cp_parser_pragma): Handle PRAGMA_OMP_TASK and PRAGMA_OMP_TASKWAIT. * pt.c (tsubst_omp_clauses): Handle OMP_CLAUSE_COLLAPSE and OMP_CLAUSE_UNTIED. Handle OMP_CLAUSE_LASTPRIVATE_STMT. (tsubst_omp_for_iterator): New function. (dependent_omp_for_p): New function. (tsubst_expr) <case OMP_FOR>: Use it. Handle collapsed OMP_FOR loops. Adjust finish_omp_for caller. Handle loops with random access iterators. Adjust for OMP_FOR_{INIT,COND,INCR} changes. (tsubst_expr): Handle OMP_TASK. * semantics.c (cxx_omp_create_clause_info): New function. (finish_omp_clauses): Call it. Handle OMP_CLAUSE_UNTIED and OMP_CLAUSE_COLLAPSE. (cxx_omp_predetermined_sharing): Removed. * semantics.c (finish_omp_for): Allow pointer iterators. Use handle_omp_for_class_iterator and dependent_omp_for_p. Handle collapsed for loops. Adjust c_finish_omp_for caller. Add new clauses argument. Fix check for type dependent cond or incr. Set OMP_FOR_CLAUSES to clauses. Use cp_convert instead of fold_convert to convert incr amount to difference_type. Only fold if not in template. If decl is mentioned in lastprivate clause, set OMP_CLAUSE_LASTPRIVATE_STMT. Handle loops with random access iterators. Adjust for OMP_FOR_{INIT,COND,INCR} changes. (finish_omp_threadprivate): Allow static class members of the current class. (handle_omp_for_class_iterator, begin_omp_task, finish_omp_task, finish_omp_taskwait): New functions. * parser.c (cp_parser_binary_expression): Add prec argument. (cp_parser_assignment_expression): Adjust caller. * cp-tree.h (outer_curly_brace_block): New prototype. * decl.c (outer_curly_brace_block): No longer static. fortran/ * scanner.c (skip_free_comments, skip_fixed_comments): Handle tabs. * parse.c (next_free): Allow tab after !$omp. (decode_omp_directive): Handle !$omp task, !$omp taskwait and !$omp end task. (case_executable): Add ST_OMP_TASKWAIT. (case_exec_markers): Add ST_OMP_TASK. (gfc_ascii_statement): Handle ST_OMP_TASK, ST_OMP_END_TASK and ST_OMP_TASKWAIT. (parse_omp_structured_block, parse_executable): Handle ST_OMP_TASK. * gfortran.h (gfc_find_sym_in_expr): New prototype. (gfc_statement): Add ST_OMP_TASK, ST_OMP_END_TASK and ST_OMP_TASKWAIT. (gfc_omp_clauses): Add OMP_SCHED_AUTO to sched_kind, OMP_DEFAULT_FIRSTPRIVATE to default_sharing. Add collapse and untied fields. (gfc_exec_op): Add EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * f95-lang.c (LANG_HOOKS_OMP_CLAUSE_COPY_CTOR, LANG_HOOKS_OMP_CLAUSE_ASSIGN_OP, LANG_HOOKS_OMP_CLAUSE_DTOR, LANG_HOOKS_OMP_PRIVATE_OUTER_REF): Define. * trans.h (gfc_omp_clause_default_ctor): Add another argument. (gfc_omp_clause_copy_ctor, gfc_omp_clause_assign_op, gfc_omp_clause_dtor, gfc_omp_private_outer_ref): New prototypes. * types.def (BT_ULONGLONG, BT_PTR_ULONGLONG, BT_FN_BOOL_ULONGLONGPTR_ULONGLONGPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_BOOL_BOOL_ULL_ULL_ULL_ULL_ULLPTR_ULLPTR, BT_FN_VOID_PTR_PTR, BT_PTR_FN_VOID_PTR_PTR, BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): New. (BT_BOOL): Use integer type with BOOL_TYPE_SIZE rather than boolean_type_node. * dump-parse-tree.c (gfc_show_omp_node): Handle EXEC_OMP_TASK, EXEC_OMP_TASKWAIT, OMP_SCHED_AUTO, OMP_DEFAULT_FIRSTPRIVATE, untied and collapse clauses. (gfc_show_code_node): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * trans.c (gfc_trans_code): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. * st.c (gfc_free_statement): Likewise. * resolve.c (gfc_resolve_blocks, resolve_code): Likewise. (find_sym_in_expr): Rename to... (gfc_find_sym_in_expr): ... this. No longer static. (resolve_allocate_expr, resolve_ordinary_assign): Adjust caller. * match.h (gfc_match_omp_task, gfc_match_omp_taskwait): New prototypes. * openmp.c (resolve_omp_clauses): Allow allocatable arrays in firstprivate, lastprivate, reduction, copyprivate and copyin clauses. (omp_current_do_code): Made static. (omp_current_do_collapse): New variable. (gfc_resolve_omp_do_blocks): Compute omp_current_do_collapse, clear omp_current_do_code and omp_current_do_collapse on return. (gfc_resolve_do_iterator): Handle collapsed do loops. (resolve_omp_do): Likewise, diagnose errorneous collapsed do loops. (OMP_CLAUSE_COLLAPSE, OMP_CLAUSE_UNTIED): Define. (gfc_match_omp_clauses): Handle default (firstprivate), schedule (auto), untied and collapse (n) clauses. (OMP_DO_CLAUSES): Add OMP_CLAUSE_COLLAPSE. (OMP_TASK_CLAUSES): Define. (gfc_match_omp_task, gfc_match_omp_taskwait): New functions. * trans-openmp.c (gfc_omp_private_outer_ref): New function. (gfc_omp_clause_default_ctor): Add outer argument. For allocatable arrays allocate them with the bounds of the outer var if outer var is allocated. (gfc_omp_clause_copy_ctor, gfc_omp_clause_assign_op, gfc_omp_clause_dtor): New functions. (gfc_trans_omp_array_reduction): If decl is allocatable array, allocate it with outer var's bounds in OMP_CLAUSE_REDUCTION_INIT and deallocate it in OMP_CLAUSE_REDUCTION_MERGE. (gfc_omp_predetermined_sharing): Return OMP_CLAUSE_DEFAULT_SHARED for assumed-size arrays. (gfc_trans_omp_do): Add par_clauses argument. If dovar is present in lastprivate clause and do loop isn't simple, set OMP_CLAUSE_LASTPRIVATE_STMT. If dovar is present in parallel's lastprivate clause, change it to shared and add lastprivate clause to OMP_FOR_CLAUSES. Handle collapsed do loops. (gfc_trans_omp_directive): Adjust gfc_trans_omp_do callers. (gfc_trans_omp_parallel_do): Likewise. Move collapse clause to OMP_FOR from OMP_PARALLEL. (gfc_trans_omp_clauses): Handle OMP_SCHED_AUTO, OMP_DEFAULT_FIRSTPRIVATE, untied and collapse clauses. (gfc_trans_omp_task, gfc_trans_omp_taskwait): New functions. (gfc_trans_omp_directive): Handle EXEC_OMP_TASK and EXEC_OMP_TASKWAIT. gcc/testsuite/ * gcc.dg/gomp/collapse-1.c: New test. * gcc.dg/gomp/nesting-1.c: New test. * g++.dg/gomp/task-1.C: New test. * g++.dg/gomp/predetermined-1.C: New test. * g++.dg/gomp/tls-4.C: New test. * gfortran.dg/gomp/collapse1.f90: New test. * gfortran.dg/gomp/sharing-3.f90: New test. * gcc.dg/gomp/pr27499.c (foo): Remove is unsigned dg-warning. * g++.dg/gomp/pr27499.C (foo): Likewise. * g++.dg/gomp/for-16.C (foo): Likewise. * g++.dg/gomp/tls-3.C: Remove dg-error, add S::s definition. * g++.dg/gomp/pr34607.C: Adjust dg-error location. * g++.dg/gomp/for-16.C (foo): Add a new dg-error. * gcc.dg/gomp/appendix-a/a.35.4.c: Add dg-warning. * gcc.dg/gomp/appendix-a/a.35.6.c: Likewise. * gfortran.dg/gomp/appendix-a/a.35.4.f90: Likewise. * gfortran.dg/gomp/appendix-a/a.35.6.f90: Likewise. * gfortran.dg/gomp/omp_parse1.f90: Remove !$omp tab test. * gfortran.dg/gomp/appendix-a/a.33.4.f90: Remove dg-error about allocatable array. * gfortran.dg/gomp/reduction1.f90: Likewise. libgomp/ * configure.ac (LIBGOMP_GNU_SYMBOL_VERSIONING): New AC_DEFINE. Substitute also OMP_*LOCK_25*. * configure: Regenerated. * config.h.in: Regenerated. * Makefile.am (libgomp_la_SOURCES): Add loop_ull.c, iter_ull.c, ptrlock.c and task.c. * Makefile.in: Regenerated. * testsuite/Makefile.in: Regenerated. * task.c: New file. * loop_ull.c: New file. * iter_ull.c: New file. * libgomp.h: Include ptrlock.h. (enum gomp_task_kind): New type. (struct gomp_team): Add task_lock, task_queue, task_count, task_running_count, single_count fields. Add work_share_list_free_lock ifndef HAVE_SYNC_BUILTINS. Remove work_share_lock, generation_mask, oldest_live_gen, num_live_gen and init_work_shares fields, add work work_share_list_alloc, work_share_list_free and work_share_chunk fields. Change work_shares from pointer to pointers into an array. Change ordered_release field into gomp_sem_t ** from flexible array member. Add implicit_task and initial_work_shares fields. Move close to the end of the struct. (struct gomp_team_state): Add single_count, last_work_share, active_level and level fields, remove work_share_generation. (gomp_barrier_handle_tasks): New prototype. (gomp_finish_task): New inline function. (struct gomp_work_share): Move chunk_size, end, incr into transparent union/struct, add chunk_size_ull, end_ll, incr_ll and next_ll fields. Reshuffle fields. Add next_alloc, next_ws, next_free and inline_ordered_team_ids fields, change ordered_team_ids into pointer from flexible array member. Add mode field. Put lock and next into a different cache line from most of the write-once fields. (gomp_iter_ull_static_next, gomp_iter_ull_dynamic_next_locked, gomp_iter_ull_guided_next_locked, gomp_iter_ull_dynamic_next, gomp_iter_ull_guided_next): New prototypes. (gomp_new_icv): New prototype. (struct gomp_thread): Add thread_pool and task fields. (struct gomp_thread_pool): New type. (gomp_new_team): New prototype. (gomp_team_start): Change type of last argument. (gomp_new_work_share): Removed. (gomp_init_work_share, gomp_fini_work_share): New prototypes. (gomp_work_share_init_done): New static inline. (gomp_throttled_spin_count_var, gomp_available_cpus, gomp_managed_threads): New extern decls. (gomp_init_task): New prototype. (gomp_spin_count_var): New extern var decl. (LIBGOMP_GNU_SYMBOL_VERSIONING): Undef if no visibility or no alias support, or if not PIC. (gomp_init_lock_30, gomp_destroy_lock_30, gomp_set_lock_30, gomp_unset_lock_30, gomp_test_lock_30, gomp_init_nest_lock_30, gomp_destroy_nest_lock_30, gomp_set_nest_lock_30, gomp_unset_nest_lock_30, gomp_test_nest_lock_30, gomp_init_lock_25, gomp_destroy_lock_25, gomp_set_lock_25, gomp_unset_lock_25, gomp_test_lock_25, gomp_init_nest_lock_25, gomp_destroy_nest_lock_25, gomp_set_nest_lock_25, gomp_unset_nest_lock_25, gomp_test_nest_lock_25): New prototypes. (omp_lock_symver, strong_alias): Define. (gomp_remaining_threads_count, gomp_remaining_threads_lock): New decls. (gomp_end_task): New. (struct gomp_task_icv, gomp_global_icv): New. (gomp_thread_limit_var, gomp_max_active_levels_var): New. (struct gomp_task): New. (gomp_nthreads_var, gomp_dyn_var, gomp_nest_var, gomp_run_sched_var, gomp_run_sched_chunk): Remove. (gomp_icv): New. (gomp_schedule_type): Reorder enum to match omp_sched_t. * team.c (struct gomp_thread_start_data): Add thread_pool and task fields. (gomp_thread_start): Add gomp_team_barrier_wait call. For non-nested case remove clearing of docked thread thr fields. Use pool fields instead of global gomp_* variables. Use gomp_barrier_wait_last when needed. Initialize ts.active_level. Create tasks for each member thread. (free_team): Only destroy team barrier, task_lock here and free it. (gomp_free_thread): Free last_team if non-NULL. (gomp_team_end): Call gomp_team_barrier_wait instead of gomp_barrier_wait. For nested case call one extra gomp_barrier_wait. Move here some destruction from free_team. Call free_team on pool->last_team if any, rather than freeing current team. Destroy work_share_list_free_lock ifndef HAVE_SYNC_BUILTINS. (gomp_new_icv): New function. (gomp_threads, gomp_threads_size, gomp_threads_used, gomp_threads_dock): Removed. (gomp_thread_destructor): New variable. (gomp_new_thread_pool, gomp_free_pool_helper, gomp_free_thread): New functions. (gomp_team_start): Create new pool if current thread doesn't have one. Use pool fields instead of global gomp_* variables. Initialize thread_pool field for new threads. Clear single_count. Change last argument from ws to team, don't create new team, set ts.work_share to &team->work_shares[0] and clear ts.last_work_share. Don't clear ts.work_share_generation. If number of threads changed, adjust atomically gomp_managed_threads. Use gomp_init_task instead of gomp_new_task, set thr->task to the corresponding implicit_task array entry. Create tasks for each member thread. Initialize ts.level. (initialize_team): Call pthread_key_create on gomp_thread_destructor. (team_destructor): New function. (new_team): Removed. (gomp_new_team): New function. (free_team): Free gomp_work_share blocks chained through next_alloc, instead of freeing work_shares and destroying work_share_lock. (gomp_team_end): Call gomp_fini_work_share. If number of threads changed, adjust atomically gomp_managed_threads. Use gomp_end_task. * barrier.c (GOMP_barrier): Call gomp_team_barrier_wait instead of gomp_barrier_wait. * single.c (GOMP_single_copy_start): Call gomp_team_barrier_wait instead of gomp_barrier_wait. Call gomp_work_share_init_done if gomp_work_share_start returned true. Don't unlock ws->lock. (GOMP_single_copy_end): Call gomp_team_barrier_wait instead of gomp_barrier_wait. (GOMP_single_start): Rewritten if HAVE_SYNC_BUILTINS. Call gomp_work_share_init_done if gomp_work_share_start returned true. Don't unlock ws->lock. * work.c: Include stddef.h. (free_work_share): Use work_share_list_free_lock instead of atomic chaining ifndef HAVE_SYNC_BUILTINS. Add team argument. Call gomp_fini_work_share and then either free ws if orphaned, or put it into work_share_list_free list of the current team. (alloc_work_share, gomp_init_work_share, gomp_fini_work_share): New functions. (gomp_work_share_start, gomp_work_share_end, gomp_work_share_end_nowait): Rewritten. * omp_lib.f90.in Change some tabs to spaces to prevent warnings. (openmp_version): Set to 200805. (omp_sched_kind, omp_sched_static, omp_sched_dynamic, omp_sched_guided, omp_sched_auto): New parameters. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New interfaces. * omp_lib.h.in (openmp_version): Set to 200805. (omp_sched_kind, omp_sched_static, omp_sched_dynamic, omp_sched_guided, omp_sched_auto): New parameters. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New externals. * loop.c: Include limits.h. (GOMP_loop_runtime_next, GOMP_loop_ordered_runtime_next): Handle GFS_AUTO. (GOMP_loop_runtime_start, GOMP_loop_ordered_runtime_start): Likewise. Use gomp_icv. (gomp_loop_static_start, gomp_loop_dynamic_start): Clear ts.static_trip here. (gomp_loop_static_start, gomp_loop_ordered_static_start): Call gomp_work_share_init_done after gomp_loop_init. Don't unlock ws->lock. (gomp_loop_dynamic_start, gomp_loop_guided_start): Call gomp_work_share_init_done after gomp_loop_init. If HAVE_SYNC_BUILTINS, don't unlock ws->lock, otherwise lock it. (gomp_loop_ordered_dynamic_start, gomp_loop_ordered_guided_start): Call gomp_work_share_init_done after gomp_loop_init. Lock ws->lock. (gomp_parallel_loop_start): Call gomp_new_team instead of gomp_new_work_share. Call gomp_loop_init on &team->work_shares[0]. Adjust gomp_team_start caller. Pass 0 as second argument to gomp_resolve_num_threads. (gomp_loop_init): For GFS_DYNAMIC, multiply ws->chunk_size by incr. If adding ws->chunk_size nthreads + 1 times after end won't overflow, set ws->mode to 1. * libgomp_g.h (GOMP_loop_ull_static_start, GOMP_loop_ull_dynamic_start, GOMP_loop_ull_guided_start, GOMP_loop_ull_runtime_start, GOMP_loop_ull_ordered_static_start, GOMP_loop_ull_ordered_dynamic_start, GOMP_loop_ull_ordered_guided_start, GOMP_loop_ull_ordered_runtime_start, GOMP_loop_ull_static_next, GOMP_loop_ull_dynamic_next, GOMP_loop_ull_guided_next, GOMP_loop_ull_runtime_next, GOMP_loop_ull_ordered_static_next, GOMP_loop_ull_ordered_dynamic_next, GOMP_loop_ull_ordered_guided_next, GOMP_loop_ull_ordered_runtime_next, GOMP_task, GOMP_taskwait): New prototypes. * libgomp.map: Export lock routines also @@OMP_2.0. (GOMP_loop_ordered_dynamic_first, GOMP_loop_ordered_guided_first, GOMP_loop_ordered_runtime_first, GOMP_loop_ordered_static_first): Remove. (GOMP_loop_ull_dynamic_next, GOMP_loop_ull_dynamic_start, GOMP_loop_ull_guided_next, GOMP_loop_ull_guided_start, GOMP_loop_ull_ordered_dynamic_next, GOMP_loop_ull_ordered_dynamic_start, GOMP_loop_ull_ordered_guided_next, GOMP_loop_ull_ordered_guided_start, GOMP_loop_ull_ordered_runtime_next, GOMP_loop_ull_ordered_runtime_start, GOMP_loop_ull_ordered_static_next, GOMP_loop_ull_ordered_static_start, GOMP_loop_ull_runtime_next, GOMP_loop_ull_runtime_start, GOMP_loop_ull_static_next, GOMP_loop_ull_static_start, GOMP_task, GOMP_taskwait): Export @@GOMP_2.0. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level, omp_set_schedule_, omp_set_schedule_8_, omp_get_schedule_, omp_get_schedule_8_, omp_get_thread_limit_, omp_set_max_active_levels_, omp_set_max_active_levels_8_, omp_get_max_active_levels_, omp_get_level_, omp_get_ancestor_thread_num_, omp_get_ancestor_thread_num_8_, omp_get_team_size_, omp_get_team_size_8_, omp_get_active_level_): New exports @@OMP_3.0. * omp.h.in (omp_sched_t): New type. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New prototypes. * env.c (gomp_spin_count_var, gomp_throttled_spin_count_var, gomp_available_cpus, gomp_managed_threads, gomp_max_active_levels_var, gomp_thread_limit_var, gomp_remaining_threads_count, gomp_remaining_threads_lock): New variables. (parse_spincount): New function. (initialize_env): Call gomp_init_num_threads unconditionally. Initialize gomp_available_cpus. Call parse_spincount, initialize gomp_{,throttled_}spin_count_var depending on presence and value of OMP_WAIT_POLICY and GOMP_SPINCOUNT env vars. Handle GOMP_BLOCKTIME env var. Handle OMP_WAIT_POLICY, OMP_MAX_ACTIVE_LEVELS, OMP_THREAD_LIMIT, OMP_STACKSIZE env vars. Handle unit specification for GOMP_STACKSIZE. Initialize gomp_remaining_threads_count and gomp_remaining_threads_lock if needed. Use gomp_global_icv. (gomp_nthreads_var, gomp_dyn_var, gomp_nest_var, gomp_run_sched_var, gomp_run_sched_chunk): Remove. (gomp_global_icv): New. (parse_schedule): Use it. Parse "auto". (omp_set_num_threads): Use gomp_icv. (omp_set_dynamic, omp_get_dynamic, omp_set_nested, omp_get_nested): Likewise. (omp_get_max_threads): Move from parallel.c. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels): New functions, add ialias. (parse_stacksize, parse_wait_policy): New functions. * fortran.c: Rewrite lock wrappers, if symbol versioning provide both wrappers for compatibility and new locks. (omp_set_schedule, omp_get_schedule, omp_get_thread_limit, omp_set_max_active_levels, omp_get_max_active_levels, omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New ialias_redirect. (omp_set_schedule_, omp_set_schedule_8_, omp_get_schedule_, omp_get_schedule_8_, omp_get_thread_limit_, omp_set_max_active_levels_, omp_set_max_active_levels_8_, omp_get_max_active_levels_, omp_get_level_, omp_get_ancestor_thread_num_, omp_get_ancestor_thread_num_8_, omp_get_team_size_, omp_get_team_size_8_, omp_get_active_level_): New functions. * parallel.c: Include limits.h. (gomp_resolve_num_threads): Add count argument. Rewritten. (GOMP_parallel_start): Call gomp_new_team and pass that as last argument to gomp_team_start. Pass 0 as second argument to gomp_resolve_num_threads. (GOMP_parallel_end): Decrease gomp_remaining_threads_count if gomp_thread_limit_var != ULONG_MAX. (omp_in_parallel): Implement using ts.active_level. (omp_get_max_threads): Move to env.c. (omp_get_level, omp_get_ancestor_thread_num, omp_get_team_size, omp_get_active_level): New functions, add ialias. * sections.c (GOMP_sections_start): Call gomp_work_share_init_done after gomp_sections_init. If HAVE_SYNC_BUILTINS, call gomp_iter_dynamic_next instead of the _locked variant and don't take lock around it, otherwise acquire it before calling gomp_iter_dynamic_next_locked. (GOMP_sections_next): If HAVE_SYNC_BUILTINS, call gomp_iter_dynamic_next instead of the _locked variant and don't take lock around it. (GOMP_parallel_sections_start): Call gomp_new_team instead of gomp_new_work_share. Call gomp_sections_init on &team->work_shares[0]. Adjust gomp_team_start caller. Pass count as second argument to gomp_resolve_num_threads, don't adjust num_threads after the call. Use gomp_icv. * iter.c (gomp_iter_dynamic_next_locked): Don't multiply ws->chunk_size by incr. (gomp_iter_dynamic_next): Likewise. If ws->mode, use more efficient code. * libgomp_f.h.in (omp_lock_25_arg_t, omp_nest_lock_25_arg_t): New types. (omp_lock_25_arg, omp_nest_lock_25_arg): New macros. (omp_check_defines): Check even the compat defines. * config/linux/ptrlock.c: New file. * config/linux/ptrlock.h: New file. * config/linux/wait.h: New file. * config/posix/ptrlock.c: New file. * config/posix/ptrlock.h: New file. * config/linux/bar.h (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New prototypes. (gomp_team_barrier_set_task_pending, gomp_team_barrier_clear_task_pending, gomp_team_barrier_set_waiting_for_tasks, gomp_team_barrier_waiting_for_tasks, gomp_team_barrier_done): New inlines. (gomp_barrier_t): Rewritten. (gomp_barrier_state_t): New typedef. (gomp_barrier_init, gomp_barrier_reinit, gomp_barrier_destroy, gomp_barrier_wait_start): Rewritten. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_barrier_last_thread, gomp_barrier_wait_last): New static inlines. * config/linux/bar.c: Include wait.h instead of libgomp.h and futex.h. (gomp_barrier_wait_end): Rewritten. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake, gomp_barrier_wait_last): New functions. * config/posix/bar.h (gomp_barrier_t): Add generation field. (gomp_barrier_state_t): New typedef. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New prototypes. (gomp_barrier_wait_start): Or all but low 2 bits from generation into the return value. Return gomp_barrier_state_t. (gomp_team_barrier_set_task_pending, gomp_team_barrier_clear_task_pending, gomp_team_barrier_set_waiting_for_tasks, gomp_team_barrier_waiting_for_tasks, gomp_team_barrier_done): New inlines. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_barrier_last_thread, gomp_barrier_wait_last): New static inlines. * config/posix/bar.c (gomp_barrier_init): Clear generation field. (gomp_barrier_wait_end): Change second argument to gomp_barrier_state_t. (gomp_team_barrier_wait, gomp_team_barrier_wait_end, gomp_team_barrier_wake): New functions. * config/linux/mutex.c: Include wait.h instead of libgomp.h and futex.h. (gomp_futex_wake, gomp_futex_wait): New variables. (gomp_mutex_lock_slow): Call do_wait instead of futex_wait. * config/linux/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. Include wait.h instead of libgomp.h and futex.h. (gomp_set_nest_lock_25): Call do_wait instead of futex_wait. * config/posix95/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. * config/posix/lock.c: Rewrite to make locks task owned, for backwards compatibility provide the old entrypoints if symbol versioning. * config/linux/proc.c (gomp_init_num_threads): Use gomp_global_icv. (get_num_procs, gomp_dynamic_max_threads): Use gomp_icv. * config/posix/proc.c, config/mingw32/proc.c: Similarly. * config/linux/powerpc/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/alpha/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/x86/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/s390/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/ia64/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/sparc/futex.h (FUTEX_WAIT, FUTEX_WAKE): Remove. (sys_futex0): Return error code. (futex_wake, futex_wait): If ENOSYS was returned, clear FUTEX_PRIVATE_FLAG in gomp_futex_wa{ke,it} and retry. (cpu_relax, atomic_write_barrier): New static inlines. * config/linux/sem.c: Include wait.h instead of libgomp.h and futex.h. (gomp_sem_wait_slow): Call do_wait instead of futex_wait. * config/linux/affinity.c: Assume HAVE_SYNC_BUILTINS. * config/linux/omp-lock.h (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_nest_lock_t): Change owner into void *, add lock field. * config/posix95/omp-lock.h: Include semaphore.h. (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_lock_t): Use sem_t instead of mutex if semaphores aren't broken. (omp_nest_lock_t): Likewise. Change owner to void *. * config/posix/omp-lock.h: Include semaphore.h. (omp_lock_25_t, omp_nest_lock_25_t): New types. (omp_lock_t): Use sem_t instead of mutex if semaphores aren't broken. (omp_nest_lock_t): Likewise. Add owner field. * testsuite/libgomp.c/collapse-1.c: New test. * testsuite/libgomp.c/collapse-2.c: New test. * testsuite/libgomp.c/collapse-3.c: New test. * testsuite/libgomp.c/icv-1.c: New test. * testsuite/libgomp.c/icv-2.c: New test. * testsuite/libgomp.c/lib-2.c: New test. * testsuite/libgomp.c/lock-1.c: New test. * testsuite/libgomp.c/lock-2.c: New test. * testsuite/libgomp.c/lock-3.c: New test. * testsuite/libgomp.c/loop-4.c: New test. * testsuite/libgomp.c/loop-5.c: New test. * testsuite/libgomp.c/loop-6.c: New test. * testsuite/libgomp.c/loop-7.c: New test. * testsuite/libgomp.c/loop-8.c: New test. * testsuite/libgomp.c/loop-9.c: New test. * testsuite/libgomp.c/nested-3.c: New test. * testsuite/libgomp.c/nestedfn-6.c: New test. * testsuite/libgomp.c/sort-1.c: New test. * testsuite/libgomp.c/task-1.c: New test. * testsuite/libgomp.c/task-2.c: New test. * testsuite/libgomp.c/task-3.c: New test. * testsuite/libgomp.c/task-4.c: New test. * testsuite/libgomp.c++/c++.exp: Add libstdc++-v3 build includes to C++ testsuite default compiler options. * testsuite/libgomp.c++/collapse-1.C: New test. * testsuite/libgomp.c++/collapse-2.C: New test. * testsuite/libgomp.c++/ctor-10.C: New test. * testsuite/libgomp.c++/for-1.C: New test. * testsuite/libgomp.c++/for-2.C: New test. * testsuite/libgomp.c++/for-3.C: New test. * testsuite/libgomp.c++/for-4.C: New test. * testsuite/libgomp.c++/for-5.C: New test. * testsuite/libgomp.c++/loop-8.C: New test. * testsuite/libgomp.c++/loop-9.C: New test. * testsuite/libgomp.c++/loop-10.C: New test. * testsuite/libgomp.c++/task-1.C: New test. * testsuite/libgomp.c++/task-2.C: New test. * testsuite/libgomp.c++/task-3.C: New test. * testsuite/libgomp.c++/task-4.C: New test. * testsuite/libgomp.c++/task-5.C: New test. * testsuite/libgomp.c++/task-6.C: New test. * testsuite/libgomp.fortran/allocatable1.f90: New test. * testsuite/libgomp.fortran/allocatable2.f90: New test. * testsuite/libgomp.fortran/allocatable3.f90: New test. * testsuite/libgomp.fortran/allocatable4.f90: New test. * testsuite/libgomp.fortran/collapse1.f90: New test. * testsuite/libgomp.fortran/collapse2.f90: New test. * testsuite/libgomp.fortran/collapse3.f90: New test. * testsuite/libgomp.fortran/collapse4.f90: New test. * testsuite/libgomp.fortran/lastprivate1.f90: New test. * testsuite/libgomp.fortran/lastprivate2.f90: New test. * testsuite/libgomp.fortran/lib4.f90: New test. * testsuite/libgomp.fortran/lock-1.f90: New test. * testsuite/libgomp.fortran/lock-2.f90: New test. * testsuite/libgomp.fortran/nested1.f90: New test. * testsuite/libgomp.fortran/nestedfn4.f90: New test. * testsuite/libgomp.fortran/strassen.f90: New test. * testsuite/libgomp.fortran/tabs1.f90: New test. * testsuite/libgomp.fortran/tabs2.f: New test. * testsuite/libgomp.fortran/task1.f90: New test. * testsuite/libgomp.fortran/task2.f90: New test. * testsuite/libgomp.fortran/vla4.f90: Add dg-warning. * testsuite/libgomp.fortran/vla5.f90: Likewise. * testsuite/libgomp.c/pr26943-2.c: Likewise. * testsuite/libgomp.c/pr26943-3.c: Likewise. * testsuite/libgomp.c/pr26943-4.c: Likewise. Co-Authored-By: Jakob Blomer <jakob.blomer@ira.uka.de> Co-Authored-By: Richard Henderson <rth@redhat.com> Co-Authored-By: Ulrich Drepper <drepper@redhat.com> From-SVN: r136433
2008-06-06 15:01:54 +02:00
#pragma omp barrier /* { dg-warning "may not be closely nested" } */
re PR c++/26943 ([gomp] firstprivate + lastprivate uses inefficient barrier) PR c++/26943 * omp-low.c (maybe_lookup_decl_in_outer_ctx): New function. (build_outer_var_ref): Use maybe_lookup_decl_in_outer_ctx to find if var will be a global variable even in the nested context. (omp_copy_decl): Only check for global variable at the end, it might be overridden in outer contexts. (scan_sharing_clauses): For global variables don't create a field. (lower_rec_input_clauses): Do nothing for global shared variables. Emit a barrier at the end of ILIST if there were any decls in both firstprivate and lastprivate clauses. (lower_send_clauses): Do nothing for global variables except for COPYIN. * testsuite/libgomp.c/pr26943-1.c: New test. * testsuite/libgomp.c/pr26943-2.c: New test. * testsuite/libgomp.c/pr26943-3.c: New test. * testsuite/libgomp.c/pr26943-4.c: New test. * testsuite/libgomp.c++/pr27337.C: Remove barrier. * testsuite/libgomp.c++/pr26943.C: New test. From-SVN: r113483
2006-05-02 22:03:38 +02:00
if (a != 8 + 6 || b != 12 + i || c != i || d != i)
j += 8;
if (e[0] != 'a' + 6 || f[0] != 'b' + i || g[0] != 'g' + i)
j += 64;
if (h[0] != 'h' + i)
j += 512;
}
if (j || a != 8 + 6 || b != 12 || c != 3 || d != 20)
++l;
if (e[0] != 'a' + 6 || f[0] != 'b' || g[0] != 'g' + 3 || h[0] != 'd')
l += 8;
}
if (l)
abort ();
if (a != 8 || b != 12 || c != 16 || d != 20)
abort ();
if (e[0] != 'a' || f[0] != 'b' || g[0] != 'c' || h[0] != 'd')
abort ();
return 0;
}
Reference in New Issue Copy Permalink