This is an archive of the discontinued LLVM Phabricator instance.

Differential D110707

[OpenMP][OMPT] lightweight tasks and explicit tasks
Needs ReviewPublic

Authored by vladaindjic on Sep 29 2021, 5:58 AM.

Details

Reviewers
protze.joachim
AndreyChurbanov
hbae
jlpeyton
jdoerfert
Summary

__ompt_get_task_info_internal function was extended to support arbitrary
nesting of explicit tasks and serialized parallel regions. In
particular, this fixes problems that occur when a serialized parallel
region is nested inside an explicit task which is at the same time
enclosed by another serialized parallel region.

While forming a serialized parallel region nested inside an explicit task,
a new lightweight task descriptor is created to represent the information
about the explicit task. To indicate whether a lightweight task descriptor
corresponds to an explicit or an implicit task, it now contains td_flags field
of kmp_tasking_flags_t type. Now, the ompt_get_task_info can determine
the right information about the task type.

When a serialized parallel region R2 is nested inside a chain of explicit tasks
belonging to an enclosing serialized parallel region R1, all explicit tasks nested
in R1 and R1’s implicit task share the same ompt_team_info stored inside the lwt
that represents the innermost explicit task in R1. (before nested loop)

The following tests are provided to show that the patch fixes described problems:

  1. lwts_info.c - contains serialized parallel region that encloses an explicit task that encloses another serialized parallel region.
  2. lwts_if_info.c - the same as previous with the difference that if(0) clause is used inside omp parallel and task directives
  3. nested_lwts_info.c - contains three nested serialized parallel regions that enclose three explicit tasks that enclose another three nested serialized parallel regions.
  4. nested_lwts_if0_info.c - the same as previous with the difference that if(0) clause is used inside omp parallel and task directives

Diff Detail

Event Timeline

vladaindjic created this revision.Sep 29 2021, 5:58 AM
Herald added subscribers: guansong, yaxunl. · View Herald TranscriptSep 29 2021, 5:58 AM
vladaindjic requested review of this revision.Sep 29 2021, 5:58 AM
Herald added a reviewer: jdoerfert. · View Herald TranscriptSep 29 2021, 5:58 AM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added subscribers: openmp-commits, sstefan1. · View Herald Transcript
Harbormaster completed remote builds in B126302: Diff 375850.Sep 29 2021, 6:10 AM
vladaindjic updated this revision to Diff 376038.Sep 29 2021, 2:27 PM

Some comments has been clarified.
Also, the diffs in ompt-specific.cpp has been clang-formatted.

Harbormaster completed remote builds in B126442: Diff 376038.Sep 29 2021, 2:46 PM
protze.joachim added a comment.Oct 19 2021, 1:59 AM

Please update the diff with context (-U9999)

vladaindjic updated this revision to Diff 380633.EditedOct 19 2021, 3:31 AM

The diff context should have been provided. @protze.joachim
Also, some comments have been rewritten for clarification.

Harbormaster completed remote builds in B129509: Diff 380633.Oct 19 2021, 3:55 AM
vladaindjic updated this revision to Diff 380638.Oct 19 2021, 4:09 AM
  • [OpenMP][OMPT] clang-formatted diffs of kmp_tasking_flags.h and ompt-specific.cpp
Harbormaster completed remote builds in B129513: Diff 380638.Oct 19 2021, 4:14 AM

Revision Contents

PathSize
openmp/
runtime/
src/
35 lines
34 lines
60 lines
3 lines
96 lines
test/
ompt/
tasks/
121 lines
122 lines
241 lines
241 lines

Diff 380633

openmp/runtime/src/kmp.h

Show All 9 Lines
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// //
//===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===//
#ifndef KMP_H #ifndef KMP_H
#define KMP_H #define KMP_H
#include "kmp_config.h" #include "kmp_config.h"
#include "kmp_tasking_flags.h"
/* #define BUILD_PARALLEL_ORDERED 1 */ /* #define BUILD_PARALLEL_ORDERED 1 */
/* This fix replaces gettimeofday with clock_gettime for better scalability on /* This fix replaces gettimeofday with clock_gettime for better scalability on
the Altix. Requires user code to be linked with -lrt. */ the Altix. Requires user code to be linked with -lrt. */
//#define FIX_SGI_CLOCK //#define FIX_SGI_CLOCK
/* Defines for OpenMP 3.0 tasking and auto scheduling */ /* Defines for OpenMP 3.0 tasking and auto scheduling */
▲ Show 20 Lines • Show All 2,335 Lines • ▼ Show 20 Lines
typedef struct kmp_task_stack { typedef struct kmp_task_stack {
kmp_stack_block_t ts_first_block; // first block of stack entries kmp_stack_block_t ts_first_block; // first block of stack entries
kmp_taskdata_t **ts_top; // pointer to the top of stack kmp_taskdata_t **ts_top; // pointer to the top of stack
kmp_int32 ts_entries; // number of entries on the stack kmp_int32 ts_entries; // number of entries on the stack
} kmp_task_stack_t; } kmp_task_stack_t;
#endif // BUILD_TIED_TASK_STACK #endif // BUILD_TIED_TASK_STACK
typedef struct kmp_tasking_flags { /* Total struct must be exactly 32 bits */
/* Compiler flags */ /* Total compiler flags must be 16 bits */
unsigned tiedness : 1; /* task is either tied (1) or untied (0) */
unsigned final : 1; /* task is final(1) so execute immediately */
unsigned merged_if0 : 1; /* no __kmpc_task_{begin/complete}_if0 calls in if0
code path */
unsigned destructors_thunk : 1; /* set if the compiler creates a thunk to
invoke destructors from the runtime */
unsigned proxy : 1; /* task is a proxy task (it will be executed outside the
context of the RTL) */
unsigned priority_specified : 1; /* set if the compiler provides priority
setting for the task */
unsigned detachable : 1; /* 1 == can detach */
unsigned hidden_helper : 1; /* 1 == hidden helper task */
unsigned reserved : 8; /* reserved for compiler use */
/* Library flags */ /* Total library flags must be 16 bits */
unsigned tasktype : 1; /* task is either explicit(1) or implicit (0) */
unsigned task_serial : 1; // task is executed immediately (1) or deferred (0)
unsigned tasking_ser : 1; // all tasks in team are either executed immediately
// (1) or may be deferred (0)
unsigned team_serial : 1; // entire team is serial (1) [1 thread] or parallel
// (0) [>= 2 threads]
/* If either team_serial or tasking_ser is set, task team may be NULL */
/* Task State Flags: */
unsigned started : 1; /* 1==started, 0==not started */
unsigned executing : 1; /* 1==executing, 0==not executing */
unsigned complete : 1; /* 1==complete, 0==not complete */
unsigned freed : 1; /* 1==freed, 0==allocated */
unsigned native : 1; /* 1==gcc-compiled task, 0==intel */
unsigned reserved31 : 7; /* reserved for library use */
} kmp_tasking_flags_t;
struct kmp_taskdata { /* aligned during dynamic allocation */ struct kmp_taskdata { /* aligned during dynamic allocation */
kmp_int32 td_task_id; /* id, assigned by debugger */ kmp_int32 td_task_id; /* id, assigned by debugger */
kmp_tasking_flags_t td_flags; /* task flags */ kmp_tasking_flags_t td_flags; /* task flags */
kmp_team_t *td_team; /* team for this task */ kmp_team_t *td_team; /* team for this task */
kmp_info_p *td_alloc_thread; /* thread that allocated data structures */ kmp_info_p *td_alloc_thread; /* thread that allocated data structures */
/* Currently not used except for perhaps IDB */ /* Currently not used except for perhaps IDB */
kmp_taskdata_t *td_parent; /* parent task */ kmp_taskdata_t *td_parent; /* parent task */
▲ Show 20 Lines • Show All 1,269 Lines • ▼ Show 20 Lines
extern void __kmp_push_current_task_to_thread(kmp_info_t *this_thr, extern void __kmp_push_current_task_to_thread(kmp_info_t *this_thr,
kmp_team_t *team, int tid); kmp_team_t *team, int tid);
extern void __kmp_pop_current_task_from_thread(kmp_info_t *this_thr); extern void __kmp_pop_current_task_from_thread(kmp_info_t *this_thr);
extern kmp_task_t *__kmp_task_alloc(ident_t *loc_ref, kmp_int32 gtid, extern kmp_task_t *__kmp_task_alloc(ident_t *loc_ref, kmp_int32 gtid,
kmp_tasking_flags_t *flags, kmp_tasking_flags_t *flags,
size_t sizeof_kmp_task_t, size_t sizeof_kmp_task_t,
size_t sizeof_shareds, size_t sizeof_shareds,
kmp_routine_entry_t task_entry); kmp_routine_entry_t task_entry);
void __kmp_init_implicit_task_flags(kmp_taskdata_t *task, kmp_team_t *team);
extern void __kmp_init_implicit_task(ident_t *loc_ref, kmp_info_t *this_thr, extern void __kmp_init_implicit_task(ident_t *loc_ref, kmp_info_t *this_thr,
kmp_team_t *team, int tid, kmp_team_t *team, int tid,
int set_curr_task); int set_curr_task);
extern void __kmp_finish_implicit_task(kmp_info_t *this_thr); extern void __kmp_finish_implicit_task(kmp_info_t *this_thr);
extern void __kmp_free_implicit_task(kmp_info_t *this_thr); extern void __kmp_free_implicit_task(kmp_info_t *this_thr);
extern kmp_event_t *__kmpc_task_allow_completion_event(ident_t *loc_ref, extern kmp_event_t *__kmpc_task_allow_completion_event(ident_t *loc_ref,
int gtid, int gtid,
▲ Show 20 Lines • Show All 683 Lines • Show Last 20 Lines

openmp/runtime/src/kmp_tasking.cpp

Show First 20 Lines • Show All 1,061 Lines • ▼ Show 20 Lines __kmp_task_finish<false>(gtid, task,
NULL); // Not sure how to find task to resume NULL); // Not sure how to find task to resume
KA_TRACE(10, ("__kmpc_omp_task_complete(exit): T#%d loc=%p task=%p\n", gtid, KA_TRACE(10, ("__kmpc_omp_task_complete(exit): T#%d loc=%p task=%p\n", gtid,
loc_ref, KMP_TASK_TO_TASKDATA(task))); loc_ref, KMP_TASK_TO_TASKDATA(task)));
return; return;
} }
#endif // TASK_UNUSED #endif // TASK_UNUSED
// __kmp_init_implicit_task_flags: Initialize td_flags field
// of an implicit task
// task: implicit task which td_flags are initialized
void __kmp_init_implicit_task_flags(kmp_taskdata_t *task, kmp_team_t *team) {
task->td_flags.tiedness = TASK_TIED;
task->td_flags.tasktype = TASK_IMPLICIT;
task->td_flags.proxy = TASK_FULL;
// All implicit tasks are executed immediately, not deferred
task->td_flags.task_serial = 1;
task->td_flags.tasking_ser = (__kmp_tasking_mode == tskm_immediate_exec);
task->td_flags.team_serial = (team->t.t_serialized) ? 1 : 0;
task->td_flags.started = 1;
task->td_flags.executing = 1;
task->td_flags.complete = 0;
task->td_flags.freed = 0;
}
// __kmp_init_implicit_task: Initialize the appropriate fields in the implicit // __kmp_init_implicit_task: Initialize the appropriate fields in the implicit
// task for a given thread // task for a given thread
// //
// loc_ref: reference to source location of parallel region // loc_ref: reference to source location of parallel region
// this_thr: thread data structure corresponding to implicit task // this_thr: thread data structure corresponding to implicit task
// team: team for this_thr // team: team for this_thr
// tid: thread id of given thread within team // tid: thread id of given thread within team
// set_curr_task: TRUE if need to push current task to thread // set_curr_task: TRUE if need to push current task to thread
Show All 13 Linesvoid __kmp_init_implicit_task(ident_t *loc_ref, kmp_info_t *this_thr,
task->td_team = team; task->td_team = team;
// task->td_parent = NULL; // fix for CQ230101 (broken parent task info // task->td_parent = NULL; // fix for CQ230101 (broken parent task info
// in debugger) // in debugger)
task->td_ident = loc_ref; task->td_ident = loc_ref;
task->td_taskwait_ident = NULL; task->td_taskwait_ident = NULL;
task->td_taskwait_counter = 0; task->td_taskwait_counter = 0;
task->td_taskwait_thread = 0; task->td_taskwait_thread = 0;
task->td_flags.tiedness = TASK_TIED; __kmp_init_implicit_task_flags(task, team);
task->td_flags.tasktype = TASK_IMPLICIT;
task->td_flags.proxy = TASK_FULL;
// All implicit tasks are executed immediately, not deferred
task->td_flags.task_serial = 1;
task->td_flags.tasking_ser = (__kmp_tasking_mode == tskm_immediate_exec);
task->td_flags.team_serial = (team->t.t_serialized) ? 1 : 0;
task->td_flags.started = 1;
task->td_flags.executing = 1;
task->td_flags.complete = 0;
task->td_flags.freed = 0;
task->td_depnode = NULL; task->td_depnode = NULL;
task->td_last_tied = task; task->td_last_tied = task;
task->td_allow_completion_event.type = KMP_EVENT_UNINITIALIZED; task->td_allow_completion_event.type = KMP_EVENT_UNINITIALIZED;
if (set_curr_task) { // only do this init first time thread is created if (set_curr_task) { // only do this init first time thread is created
KMP_ATOMIC_ST_REL(&task->td_incomplete_child_tasks, 0); KMP_ATOMIC_ST_REL(&task->td_incomplete_child_tasks, 0);
// Not used: don't need to deallocate implicit task // Not used: don't need to deallocate implicit task
▲ Show 20 Lines • Show All 3,678 Lines • Show Last 20 Lines

openmp/runtime/src/kmp_tasking_flags.h

  • This file was added.
/*
* kmp_tasking_flags.h -- OpenMP task flags
*/
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#ifndef KMP_TASKING_FLAGS_H
#define KMP_TASKING_FLAGS_H
// Set all tasking flags to zero, under assumption
// that kmp_tasking_flags struct is exactly 32 bits long.
#define TASKING_FLAGS_CLEAR(td_flags_ptr) \
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-#define TASKING_FLAGS_CLEAR(td_flags_ptr) \
-  *((uint32_t *) td_flags_ptr) = 0;
+#define TASKING_FLAGS_CLEAR(td_flags_ptr) *((uint32_t *)td_flags_ptr) = 0;
Lint: Pre-merge checks: clang-format: please reformat the code ``` -#define TASKING_FLAGS_CLEAR(td_flags_ptr) \ - *…
*((uint32_t *) td_flags_ptr) = 0;
// In order to avoid circular reference, declare the following data structure
// here in order to be safely included by both ompt-internal.h and kmp.h files.
// It could be declared inside ompt-internal.h instead.
// However, it's not only used by OMPT, so it is better to declare it
// in a separate header file.
typedef struct kmp_tasking_flags { /* Total struct must be exactly 32 bits */
/* Compiler flags */ /* Total compiler flags must be 16 bits */
unsigned tiedness : 1; /* task is either tied (1) or untied (0) */
unsigned final : 1; /* task is final(1) so execute immediately */
unsigned merged_if0 : 1; /* no __kmpc_task_{begin/complete}_if0 calls in if0
code path */
unsigned destructors_thunk : 1; /* set if the compiler creates a thunk to
invoke destructors from the runtime */
unsigned proxy : 1; /* task is a proxy task (it will be executed outside the
context of the RTL) */
unsigned priority_specified : 1; /* set if the compiler provides priority
setting for the task */
unsigned detachable : 1; /* 1 == can detach */
unsigned hidden_helper : 1; /* 1 == hidden helper task */
unsigned reserved : 8; /* reserved for compiler use */
/* Library flags */ /* Total library flags must be 16 bits */
unsigned tasktype : 1; /* task is either explicit(1) or implicit (0) */
unsigned task_serial : 1; // task is executed immediately (1) or deferred (0)
unsigned tasking_ser : 1; // all tasks in team are either executed immediately
// (1) or may be deferred (0)
unsigned team_serial : 1; // entire team is serial (1) [1 thread] or parallel
// (0) [>= 2 threads]
/* If either team_serial or tasking_ser is set, task team may be NULL */
/* Task State Flags: */
unsigned started : 1; /* 1==started, 0==not started */
unsigned executing : 1; /* 1==executing, 0==not executing */
unsigned complete : 1; /* 1==complete, 0==not complete */
unsigned freed : 1; /* 1==freed, 0==allocated */
unsigned native : 1; /* 1==gcc-compiled task, 0==intel */
unsigned reserved31 : 7; /* reserved for library use */
} kmp_tasking_flags_t;
#endif //KMP_TASKING_FLAGS_H
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-#endif //KMP_TASKING_FLAGS_H
+#endif // KMP_TASKING_FLAGS_H
Lint: Pre-merge checks: clang-format: please reformat the code ``` -#endif //KMP_TASKING_FLAGS_H +#endif //…

openmp/runtime/src/ompt-internal.h

/* /*
* ompt-internal.h - header of OMPT internal data structures * ompt-internal.h - header of OMPT internal data structures
*/ */
//===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===//
// //
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information. // See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// //
//===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===//
#ifndef __OMPT_INTERNAL_H__ #ifndef __OMPT_INTERNAL_H__
#define __OMPT_INTERNAL_H__ #define __OMPT_INTERNAL_H__
#include "kmp_tasking_flags.h"
#include "ompt-event-specific.h" #include "ompt-event-specific.h"
#include "omp-tools.h" #include "omp-tools.h"
#define OMPT_VERSION 1 #define OMPT_VERSION 1
#define _OMP_EXTERN extern "C" #define _OMP_EXTERN extern "C"
#define OMPT_INVOKER(x) \ #define OMPT_INVOKER(x) \
▲ Show 20 Lines • Show All 80 Lines • ▼ Show 20 Linestypedef struct {
void *master_return_address; void *master_return_address;
} ompt_team_info_t; } ompt_team_info_t;
typedef struct ompt_lw_taskteam_s { typedef struct ompt_lw_taskteam_s {
ompt_team_info_t ompt_team_info; ompt_team_info_t ompt_team_info;
ompt_task_info_t ompt_task_info; ompt_task_info_t ompt_task_info;
int heap; int heap;
struct ompt_lw_taskteam_s *parent; struct ompt_lw_taskteam_s *parent;
// preserve td_flags of kmp_taskdata_t after linking happens
kmp_tasking_flags_t td_flags;
} ompt_lw_taskteam_t; } ompt_lw_taskteam_t;
typedef struct { typedef struct {
ompt_data_t thread_data; ompt_data_t thread_data;
ompt_data_t task_data; /* stored here from implicit barrier-begin until ompt_data_t task_data; /* stored here from implicit barrier-begin until
implicit-task-end */ implicit-task-end */
void *return_address; /* stored here on entry of runtime */ void *return_address; /* stored here on entry of runtime */
ompt_state_t state; ompt_state_t state;
▲ Show 20 Lines • Show All 49 Lines • Show Last 20 Lines

openmp/runtime/src/ompt-specific.cpp

Show First 20 Lines • Show All 258 Lines • ▼ Show 20 Linesvoid __ompt_lw_taskteam_init(ompt_lw_taskteam_t *lwt, kmp_info_t *thr, int gtid,
lwt->ompt_team_info.parallel_data = *ompt_pid; lwt->ompt_team_info.parallel_data = *ompt_pid;
lwt->ompt_team_info.master_return_address = codeptr; lwt->ompt_team_info.master_return_address = codeptr;
lwt->ompt_task_info.task_data.value = 0; lwt->ompt_task_info.task_data.value = 0;
lwt->ompt_task_info.frame.enter_frame = ompt_data_none; lwt->ompt_task_info.frame.enter_frame = ompt_data_none;
lwt->ompt_task_info.frame.exit_frame = ompt_data_none; lwt->ompt_task_info.frame.exit_frame = ompt_data_none;
lwt->ompt_task_info.scheduling_parent = NULL; lwt->ompt_task_info.scheduling_parent = NULL;
lwt->heap = 0; lwt->heap = 0;
lwt->parent = 0; lwt->parent = 0;
// clear tasking flags
TASKING_FLAGS_CLEAR(&lwt->td_flags);
} }
void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, kmp_info_t *thr, void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, kmp_info_t *thr,
int on_heap, bool always) { int on_heap, bool always) {
ompt_lw_taskteam_t *link_lwt = lwt; ompt_lw_taskteam_t *link_lwt = lwt;
if (always || if (always ||
thr->th.th_team->t.t_serialized > thr->th.th_team->t.t_serialized >
1) { // we already have a team, so link the new team and swap values 1) { // we already have a team, so link the new team and swap values
Show All 17 Lines#if OMPD_SUPPORT
if (ompd_state & OMPD_ENABLE_BP) { if (ompd_state & OMPD_ENABLE_BP) {
ompd_bp_parallel_begin(); ompd_bp_parallel_begin();
} }
#endif #endif
ompt_task_info_t tmp_task = lwt->ompt_task_info; ompt_task_info_t tmp_task = lwt->ompt_task_info;
link_lwt->ompt_task_info = *OMPT_CUR_TASK_INFO(thr); link_lwt->ompt_task_info = *OMPT_CUR_TASK_INFO(thr);
*OMPT_CUR_TASK_INFO(thr) = tmp_task; *OMPT_CUR_TASK_INFO(thr) = tmp_task;
// copy td_flags
// Note: cur_task may belong to the explicit task, so we need
// to preserve the td_flags->tasktype.
kmp_taskdata_t *cur_task = thr->th.th_current_task;
link_lwt->td_flags = cur_task->td_flags;
// linked task isn't executing at the moment
link_lwt->td_flags.executing = 0;
// clear td_flags of cur_task
TASKING_FLAGS_CLEAR(&cur_task->td_flags);
// Since cur_task now represents an implicit task of the serialized
// parallel region, initialize tasking flags (of cur_task) the same way
// it is done for implicit tasks of regular regions.
// Otherwise, td_flags may be inherited from previously linked
// explicit tasks.
__kmp_init_implicit_task_flags(cur_task, thr->th.th_team);
} else { } else {
// this is the first serialized team, so we just store the values in the // this is the first serialized team, so we just store the values in the
// team and drop the taskteam-object // team and drop the taskteam-object
*OMPT_CUR_TEAM_INFO(thr) = lwt->ompt_team_info; *OMPT_CUR_TEAM_INFO(thr) = lwt->ompt_team_info;
#if OMPD_SUPPORT #if OMPD_SUPPORT
if (ompd_state & OMPD_ENABLE_BP) { if (ompd_state & OMPD_ENABLE_BP) {
ompd_bp_parallel_begin(); ompd_bp_parallel_begin();
} }
#endif #endif
*OMPT_CUR_TASK_INFO(thr) = lwt->ompt_task_info; *OMPT_CUR_TASK_INFO(thr) = lwt->ompt_task_info;
} }
} }
void __ompt_lw_taskteam_unlink(kmp_info_t *thr) { void __ompt_lw_taskteam_unlink(kmp_info_t *thr) {
ompt_lw_taskteam_t *lwtask = thr->th.th_team->t.ompt_serialized_team_info; ompt_lw_taskteam_t *lwtask = thr->th.th_team->t.ompt_serialized_team_info;
if (lwtask) { if (lwtask) {
ompt_task_info_t tmp_task = lwtask->ompt_task_info; ompt_task_info_t tmp_task = lwtask->ompt_task_info;
lwtask->ompt_task_info = *OMPT_CUR_TASK_INFO(thr); lwtask->ompt_task_info = *OMPT_CUR_TASK_INFO(thr);
*OMPT_CUR_TASK_INFO(thr) = tmp_task; *OMPT_CUR_TASK_INFO(thr) = tmp_task;
// copy back the td_flags
thr->th.th_current_task->td_flags = lwtask->td_flags;
// unlinked task is executing at the moment
thr->th.th_current_task->td_flags.executing = 1;
#if OMPD_SUPPORT #if OMPD_SUPPORT
if (ompd_state & OMPD_ENABLE_BP) { if (ompd_state & OMPD_ENABLE_BP) {
ompd_bp_parallel_end(); ompd_bp_parallel_end();
} }
#endif #endif
thr->th.th_team->t.ompt_serialized_team_info = lwtask->parent; thr->th.th_team->t.ompt_serialized_team_info = lwtask->parent;
ompt_team_info_t tmp_team = lwtask->ompt_team_info; ompt_team_info_t tmp_team = lwtask->ompt_team_info;
lwtask->ompt_team_info = *OMPT_CUR_TEAM_INFO(thr); lwtask->ompt_team_info = *OMPT_CUR_TEAM_INFO(thr);
*OMPT_CUR_TEAM_INFO(thr) = tmp_team; *OMPT_CUR_TEAM_INFO(thr) = tmp_team;
if (lwtask->heap) { if (lwtask->heap) {
__kmp_free(lwtask); __kmp_free(lwtask);
lwtask = NULL; lwtask = NULL;
} }
} }
// return lwtask; // return lwtask;
} }
//---------------------------------------------------------- //----------------------------------------------------------
// task support // task support
//---------------------------------------------------------- //----------------------------------------------------------
#define OMPT_GET_TASK_FLAGS(task) \
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-#define OMPT_GET_TASK_FLAGS(task) \
-  (task->td_flags.tasktype ? ompt_task_explicit : ompt_task_implicit) | \
-  TASK_TYPE_DETAILS_FORMAT(task)
+#define OMPT_GET_TASK_FLAGS(task)                                              \
+  (task->td_flags.tasktype ? ompt_task_explicit : ompt_task_implicit) |        \
+      TASK_TYPE_DETAILS_FORMAT(task)
Lint: Pre-merge checks: clang-format: please reformat the code ``` -#define OMPT_GET_TASK_FLAGS(task) \ - (task…
(task->td_flags.tasktype ? ompt_task_explicit : ompt_task_implicit) | \
TASK_TYPE_DETAILS_FORMAT(task)
int __ompt_get_task_info_internal(int ancestor_level, int *type, int __ompt_get_task_info_internal(int ancestor_level, int *type,
ompt_data_t **task_data, ompt_data_t **task_data,
ompt_frame_t **task_frame, ompt_frame_t **task_frame,
ompt_data_t **parallel_data, ompt_data_t **parallel_data,
int *thread_num) { int *thread_num) {
if (__kmp_get_gtid() < 0) if (__kmp_get_gtid() < 0)
return 0; return 0;
Show All 12 Lines if (taskdata == NULL)
return 0; return 0;
kmp_team *team = thr->th.th_team, *prev_team = NULL; kmp_team *team = thr->th.th_team, *prev_team = NULL;
if (team == NULL) if (team == NULL)
return 0; return 0;
ompt_lw_taskteam_t *lwt = NULL, ompt_lw_taskteam_t *lwt = NULL,
*next_lwt = LWT_FROM_TEAM(taskdata->td_team), *next_lwt = LWT_FROM_TEAM(taskdata->td_team),
*prev_lwt = NULL; *prev_lwt = NULL;
// This flag indicates that the task T1 at the specified ancestor_level
// is the task that belongs to the serialized parallel region R1.
// Note that the T1 isn't the innermost task of the region R1 at
// the time of the call of this function and it can be either an explicit
// or an implicit task. Let T2 be the innermost explicit task of the R1,
// which is at the same time enclosed by T1 (might not be directly
// enclosed). The T2 contains contains another serialized parallel region
// R2. At the moment of R2's creation, R1's team information is memoized
// inside an lwt, which is put at the beginning of the lightweight tasks
// linked list. The same lwt contains the information about the
// task active at the moment, which is the task T2.
// When providing the information about T1, the team information is read
// from lwt, while the task information is read from the taskdata that
// corresponds to the T1.
bool tasks_share_lwt = false;
while (ancestor_level > 0) { while (ancestor_level > 0) {
// needed for thread_num // needed for thread_num
prev_team = team; prev_team = team;
prev_lwt = lwt; prev_lwt = lwt;
// next lightweight team (if any)
if (lwt) if (lwt) {
kmp_taskdata_t *scheduling_parent =
lwt->ompt_task_info.scheduling_parent;
// At the time lwt was created, the active task was an explicit task.
// Thus, lwt->ompt_task_info represents the information about that
// task. It is possible that the explicit task is nested inside the
// hierarchy of explicit tasks on top of which is the implicit task
// of the enclosing parallel region R. The inner loop iterates over
// this hierarchy by using the scheduling_parent pointer until either
// reaching the requested ancestor_level or encountering the implicit
// task of the enclosing parallel region R. Since this implicit task
// doesn't have the scheduling_parent, search continues from the
// innermost tasks that encloses the region R.
// Note that the lwt->ompt_team_info is going to be shared by all tasks
// nested in the the same serialized parallel region.
while (scheduling_parent && ancestor_level > 0) {
// access to the parent task
taskdata = scheduling_parent;
ancestor_level--;
// get another explicit task, if any
scheduling_parent = taskdata->ompt_task_info.scheduling_parent;
}
if (ancestor_level == 0) {
// The requested ancestor_level is reached, so leave the loop, but
// first mark that the information about the task should be read
// from taskdata instead of the lwt.
tasks_share_lwt = true;
break;
}
// Since the previous loop eventually exhausted all nested tasks that
// belongs to the same serialized parallel region R, try to access the
// lwt that corresponds to the serialized region that encloses the
// region R, if any.
lwt = lwt->parent; lwt = lwt->parent;
}
// next heavyweight team (if any) after // next heavyweight team (if any) after
// lightweight teams are exhausted // lightweight teams are exhausted
if (!lwt && taskdata) { if (!lwt && taskdata) {
// first try scheduling parent (for explicit task scheduling) // first try scheduling parent (for explicit task scheduling)
if (taskdata->ompt_task_info.scheduling_parent) { if (taskdata->ompt_task_info.scheduling_parent) {
taskdata = taskdata->ompt_task_info.scheduling_parent; taskdata = taskdata->ompt_task_info.scheduling_parent;
} else if (next_lwt) { } else if (next_lwt) {
Show All 9 Lines while (ancestor_level > 0) {
next_lwt = LWT_FROM_TEAM(taskdata->td_team); next_lwt = LWT_FROM_TEAM(taskdata->td_team);
} }
} }
} }
ancestor_level--; ancestor_level--;
} }
if (lwt) { if (lwt) {
info = &lwt->ompt_task_info; // Decide whether to read task information from taskdata or lwt.
info = tasks_share_lwt ? &taskdata->ompt_task_info: &lwt->ompt_task_info;
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-      info = tasks_share_lwt ? &taskdata->ompt_task_info: &lwt->ompt_task_info;
+      info = tasks_share_lwt ? &taskdata->ompt_task_info : &lwt->ompt_task_info;
Lint: Pre-merge checks: clang-format: please reformat the code ``` - info = tasks_share_lwt ? &taskdata…
team_info = &lwt->ompt_team_info; team_info = &lwt->ompt_team_info;
if (type) { if (type) {
*type = ompt_task_implicit; // decide whether to read td_flags from taskdata or lwt
// Since macro expands, it is safer to enclose it by parentheses.
*type = tasks_share_lwt ? (OMPT_GET_TASK_FLAGS(taskdata))
: (OMPT_GET_TASK_FLAGS(lwt));
} }
} else if (taskdata) { } else if (taskdata) {
info = &taskdata->ompt_task_info; info = &taskdata->ompt_task_info;
team_info = &team->t.ompt_team_info; team_info = &team->t.ompt_team_info;
if (type) { if (type) {
if (taskdata->td_parent) { if (taskdata->td_parent) {
*type = (taskdata->td_flags.tasktype ? ompt_task_explicit *type = OMPT_GET_TASK_FLAGS(taskdata);
: ompt_task_implicit) |
TASK_TYPE_DETAILS_FORMAT(taskdata);
} else { } else {
*type = ompt_task_initial; *type = ompt_task_initial;
} }
} }
} }
if (task_data) { if (task_data) {
*task_data = info ? &info->task_data : NULL; *task_data = info ? &info->task_data : NULL;
} }
▲ Show 20 Lines • Show All 96 Lines • Show Last 20 Lines

openmp/runtime/test/ompt/tasks/lwts_if0_info.c

  • This file was added.
// RUN: %libomp-compile-and-run | FileCheck %s
// REQUIRES: ompt
#include "callback.h"
#include <omp.h>
__attribute__ ((noinline)) // workaround for bug in icc
void print_task_info_at(int ancestor_level, int id)
{
#pragma omp critical
{
int task_type;
char buffer[2048];
ompt_data_t *parallel_data;
ompt_data_t *task_data;
ompt_get_task_info(ancestor_level, &task_type, &task_data, NULL, &parallel_data, NULL);
format_task_type(task_type, buffer);
printf("%" PRIu64 ": ancestor_level=%d id=%d task_type=%s=%d "
"parallel_id=%" PRIu64 " task_id=%" PRIu64 "\n",
ompt_get_thread_data()->value, ancestor_level, id, buffer,
task_type, parallel_data->value, task_data->value);
}
};
int main()
{
//initial task
print_task_info_at(0, 0);
#pragma omp parallel if(0)
{
// region 0
// outermost lwt
print_task_info_at(0, 1);
#pragma omp task if(0)
{
// task 0
print_task_info_at(0, 2);
// check hierarchy now
print_task_info_at(0, 2);
print_task_info_at(1, 1);
print_task_info_at(2, 0);
#pragma omp parallel if(0)
{
// region 1
print_task_info_at(0, 3);
// check hierarchy now
print_task_info_at(0, 3);
print_task_info_at(1, 2);
print_task_info_at(2, 1);
print_task_info_at(3, 0);
print_task_info_at(0, 3);
}
print_task_info_at(0, 2);
};
print_task_info_at(0, 1);
}
print_task_info_at(0, 0);
// Check if libomp supports the callbacks for this test.
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
// CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
// CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_initial_task_begin: parallel_id=[[PARALLEL_ID_0:[0-9]+]], task_id=[[TASK_ID_0:[0-9]+]], actual_parallelism=1, index=1, flags=1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_0]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_1]], task_id=[[TASK_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_1]]
// CHECK-SAME: new_task_id=[[TASK_ID_2:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// check hierarchy now
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_2]],
// CHESK-SAME: parallel_id=[[PARALLEL_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_2:[0-9]+]], task_id=[[TASK_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// check hierarchy now
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
return 0;
}
No newline at end of file

openmp/runtime/test/ompt/tasks/lwts_info.c

  • This file was added.
// RUN: %libomp-compile-and-run | FileCheck %s
// REQUIRES: ompt
#include "callback.h"
#include <omp.h>
__attribute__ ((noinline)) // workaround for bug in icc
void print_task_info_at(int ancestor_level, int id)
{
#pragma omp critical
{
int task_type;
char buffer[2048];
ompt_data_t *parallel_data;
ompt_data_t *task_data;
ompt_get_task_info(ancestor_level, &task_type, &task_data, NULL, &parallel_data, NULL);
format_task_type(task_type, buffer);
printf("%" PRIu64 ": ancestor_level=%d id=%d task_type=%s=%d "
"parallel_id=%" PRIu64 " task_id=%" PRIu64 "\n",
ompt_get_thread_data()->value, ancestor_level, id, buffer,
task_type, parallel_data->value, task_data->value);
}
};
int main()
{
//initial task
print_task_info_at(0, 0);
#pragma omp parallel num_threads(1)
{
// region 0
// outermost lwt
print_task_info_at(0, 1);
#pragma omp task
{
// task 0
print_task_info_at(0, 2);
// check hierarchy now
print_task_info_at(0, 2);
print_task_info_at(1, 1);
print_task_info_at(2, 0);
#pragma omp parallel num_threads(1)
{
// region 1
print_task_info_at(0, 3);
// check hierarchy now
print_task_info_at(0, 3);
print_task_info_at(1, 2);
print_task_info_at(2, 1);
print_task_info_at(3, 0);
print_task_info_at(0, 3);
}
print_task_info_at(0, 2);
};
print_task_info_at(0, 1);
}
print_task_info_at(0, 0);
// Check if libomp supports the callbacks for this test.
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
// CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
// CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_initial_task_begin: parallel_id=[[PARALLEL_ID_0:[0-9]+]], task_id=[[TASK_ID_0:[0-9]+]], actual_parallelism=1, index=1, flags=1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_0]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_1]], task_id=[[TASK_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_1]]
// CHECK-SAME: new_task_id=[[TASK_ID_2:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// check hierarchy now
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_2]],
// CHESK-SAME: parallel_id=[[PARALLEL_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_2:[0-9]+]], task_id=[[TASK_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// check hierarchy now
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_3]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_2]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
return 0;
}
No newline at end of file

openmp/runtime/test/ompt/tasks/nested_lwts_if0_info.c

  • This file was added.
// RUN: %libomp-compile-and-run | FileCheck %s
// REQUIRES: ompt
#include "callback.h"
#include <omp.h>
__attribute__ ((noinline)) // workaround for bug in icc
void print_task_info_at(int ancestor_level, int id)
{
#pragma omp critical
{
int task_type;
char buffer[2048];
ompt_data_t *parallel_data;
ompt_data_t *task_data;
ompt_get_task_info(ancestor_level, &task_type, &task_data, NULL, &parallel_data, NULL);
format_task_type(task_type, buffer);
printf("%" PRIu64 ": ancestor_level=%d id=%d task_type=%s=%d "
"parallel_id=%" PRIu64 " task_id=%" PRIu64 "\n",
ompt_get_thread_data()->value, ancestor_level, id, buffer,
task_type, parallel_data->value, task_data->value);
}
};
int main()
{
//initial task
print_task_info_at(0, 0);
#pragma omp parallel if(0)
{
// region 0
// outermost lwt
print_task_info_at(0, 1);
#pragma omp parallel if(0)
{
// region 1
print_task_info_at(0, 2);
#pragma omp parallel if(0)
{
// region 2
print_task_info_at(0, 3);
#pragma omp task if(0)
{
// task 0
print_task_info_at(0, 4);
#pragma omp task if(0)
{
// task 1
print_task_info_at(0, 5);
#pragma omp task if(0)
{
// task 2
print_task_info_at(0, 6);
print_task_info_at(0, 6);
print_task_info_at(1, 5);
print_task_info_at(2, 4);
print_task_info_at(3, 3);
print_task_info_at(4, 2);
print_task_info_at(5, 1);
print_task_info_at(6, 0);
#pragma omp parallel if(0)
{
// region 3
print_task_info_at(0, 7);
#pragma omp parallel if(0)
{
// region 4
print_task_info_at(0, 8);
#pragma omp parallel if(0)
{
// region 5
print_task_info_at(0, 9);
print_task_info_at(0, 9);
print_task_info_at(1, 8);
print_task_info_at(2, 7);
print_task_info_at(3, 6);
print_task_info_at(4, 5);
print_task_info_at(5, 4);
print_task_info_at(6, 3);
print_task_info_at(7, 2);
print_task_info_at(8, 1);
print_task_info_at(9, 0);
print_task_info_at(0, 9);
}
print_task_info_at(0, 8);
}
print_task_info_at(0, 7);
}
print_task_info_at(0, 6);
};
print_task_info_at(0, 5);
};
print_task_info_at(0, 4);
};
print_task_info_at(0, 3);
}
print_task_info_at(0, 2);
}
print_task_info_at(0, 1);
}
print_task_info_at(0, 0);
// Check if libomp supports the callbacks for this test.
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
// CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
// CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_initial_task_begin: parallel_id=[[PARALLEL_ID_0:[0-9]+]], task_id=[[TASK_ID_0:[0-9]+]], actual_parallelism=1, index=1, flags=1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_0]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_1]], task_id=[[TASK_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_1]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_2]], task_id=[[TASK_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// region 2
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_2]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_3]], task_id=[[TASK_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_3]]
// CHECK-SAME: new_task_id=[[TASK_ID_4:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// task 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_4]]
// CHECK-SAME: new_task_id=[[TASK_ID_5:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// task 2
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_5]]
// CHECK-SAME: new_task_id=[[TASK_ID_6:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// check hierarchy
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=4 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=5 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=6 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 3
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_6]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_4:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_4]], task_id=[[TASK_ID_7:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// region 4
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_7]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_5:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_5]], task_id=[[TASK_ID_8:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// region 5
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_8]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_6:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_6]], task_id=[[TASK_ID_9:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// check hierarchy
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=4 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=5 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=6 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=7 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=8 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=9 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 5
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// region 4
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// region 3
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// task 2
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// task 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// region 2
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// initial task
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
return 0;
}
No newline at end of file

openmp/runtime/test/ompt/tasks/nested_lwts_info.c

  • This file was added.
// RUN: %libomp-compile-and-run | FileCheck %s
// REQUIRES: ompt
#include "callback.h"
#include <omp.h>
__attribute__ ((noinline)) // workaround for bug in icc
void print_task_info_at(int ancestor_level, int id)
{
#pragma omp critical
{
int task_type;
char buffer[2048];
ompt_data_t *parallel_data;
ompt_data_t *task_data;
ompt_get_task_info(ancestor_level, &task_type, &task_data, NULL, &parallel_data, NULL);
format_task_type(task_type, buffer);
printf("%" PRIu64 ": ancestor_level=%d id=%d task_type=%s=%d "
"parallel_id=%" PRIu64 " task_id=%" PRIu64 "\n",
ompt_get_thread_data()->value, ancestor_level, id, buffer,
task_type, parallel_data->value, task_data->value);
}
};
int main()
{
//initial task
print_task_info_at(0, 0);
#pragma omp parallel num_threads(1)
{
// region 0
// outermost lwt
print_task_info_at(0, 1);
#pragma omp parallel num_threads(1)
{
// region 1
print_task_info_at(0, 2);
#pragma omp parallel num_threads(1)
{
// region 2
print_task_info_at(0, 3);
#pragma omp task
{
// task 0
print_task_info_at(0, 4);
#pragma omp task
{
// task 1
print_task_info_at(0, 5);
#pragma omp task
{
// task 2
print_task_info_at(0, 6);
print_task_info_at(0, 6);
print_task_info_at(1, 5);
print_task_info_at(2, 4);
print_task_info_at(3, 3);
print_task_info_at(4, 2);
print_task_info_at(5, 1);
print_task_info_at(6, 0);
#pragma omp parallel num_threads(1)
{
// region 3
print_task_info_at(0, 7);
#pragma omp parallel num_threads(1)
{
// region 4
print_task_info_at(0, 8);
#pragma omp parallel num_threads(1)
{
// region 5
print_task_info_at(0, 9);
print_task_info_at(0, 9);
print_task_info_at(1, 8);
print_task_info_at(2, 7);
print_task_info_at(3, 6);
print_task_info_at(4, 5);
print_task_info_at(5, 4);
print_task_info_at(6, 3);
print_task_info_at(7, 2);
print_task_info_at(8, 1);
print_task_info_at(9, 0);
print_task_info_at(0, 9);
}
print_task_info_at(0, 8);
}
print_task_info_at(0, 7);
}
print_task_info_at(0, 6);
};
print_task_info_at(0, 5);
};
print_task_info_at(0, 4);
};
print_task_info_at(0, 3);
}
print_task_info_at(0, 2);
}
print_task_info_at(0, 1);
}
print_task_info_at(0, 0);
// Check if libomp supports the callbacks for this test.
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
// CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
// CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
// CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_initial_task_begin: parallel_id=[[PARALLEL_ID_0:[0-9]+]], task_id=[[TASK_ID_0:[0-9]+]], actual_parallelism=1, index=1, flags=1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_0]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_1]], task_id=[[TASK_ID_1:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_1]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_2]], task_id=[[TASK_ID_2:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// region 2
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_2]],
// CHECK-SAME: parallel_id=[[PARALLEL_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_3]], task_id=[[TASK_ID_3:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_3]]
// CHECK-SAME: new_task_id=[[TASK_ID_4:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// task 1
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_4]]
// CHECK-SAME: new_task_id=[[TASK_ID_5:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// task 2
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[TASK_ID_5]]
// CHECK-SAME: new_task_id=[[TASK_ID_6:[0-9]+]]
// CHECK-SAME: task_type=ompt_task_explicit|ompt_task_undeferred=134217732
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// check hierarchy
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=4 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=5 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=6 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 3
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_6]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_4:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_4]], task_id=[[TASK_ID_7:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// region 4
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_7]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_5:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_5]], task_id=[[TASK_ID_8:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// region 5
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[TASK_ID_8]]
// CHECK-SAME: parallel_id=[[PARALLEL_ID_6:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID_6]], task_id=[[TASK_ID_9:[0-9]+]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// check hierarchy
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=1 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=2 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=3 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=4 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=5 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=6 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=7 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=8 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=9 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
// region 5
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=9 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_6]] task_id=[[TASK_ID_9]]
// region 4
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=8 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_5]] task_id=[[TASK_ID_8]]
// region 3
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=7 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_4]] task_id=[[TASK_ID_7]]
// task 2
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=6 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_6]]
// task 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=5 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_5]]
// task 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=4 task_type=ompt_task_explicit|ompt_task_undeferred=134217732 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_4]]
// region 2
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=3 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_3]] task_id=[[TASK_ID_3]]
// region 1
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=2 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_2]] task_id=[[TASK_ID_2]]
// region 0
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730 parallel_id=[[PARALLEL_ID_1]] task_id=[[TASK_ID_1]]
// initial task
// CHECK: {{^}}[[MASTER_ID]]: ancestor_level=0 id=0 task_type=ompt_task_initial=1 parallel_id=[[PARALLEL_ID_0]] task_id=[[TASK_ID_0]]
return 0;
}
No newline at end of file