Implementation of OMPT as specified in OpenMP 5.0 Preview 1

[protze.joachim]

Summary

This patch updates the implementation of OMPT to the specification in the upcoming OpenMP 5.0 version.

Half of the touched code adds test cases to test the various OMPT event callbacks. The code is tested to work with latest clang, GNU and Intel compiler. The implementation is optimized for low overhead when no tool is attached shifting the cost to execution with tool attached.

The code is developed in:

https://github.com/OpenMPToolsInterface/LLVM-openmp

The diff can also be viewed at:

https://github.com/OpenMPToolsInterface/LLVM-openmp/pull/2/files

[omalyshe]

runtime/src/kmp_dispatch.cpp

1233–1234

The comment is not relevant any more

2541

Should it be #if OMPT_SUPPORT && OMPT_OPTIONAL in all occurrences below?

runtime/src/kmp_global.cpp

306

LIBOMP_OMPT_SUPPORT -> OMPT_SUPPORT

runtime/src/kmp_gsupport.cpp

34

Should be under OMPT_OPTIONAL?

43

Should be under OMPT_OPTIONAL?

65

Should be under OMPT_OPTIONAL?

75

Should be under OMPT_OPTIONAL?

179

Should be under OMPT_OPTIONAL?

193

Should be under OMPT_OPTIONAL?

216

Should be under OMPT_OPTIONAL and more others below?

252

Should be under OMPT_OPTIONAL?

640

Should be under OMPT_OPTIONAL?

828

Should be under OMPT_OPTIONAL?

[omalyshe]

runtime/src/kmp_sched.cpp

142–146

Remove the comment

194–198

Remove the comment

223–227

Remove the comment

380–384

Remove the comment

runtime/src/kmp_settings.cpp

4355–4377

LIBOMP_OMPT_SUPPORT -> OMPT_SUPPORT

4622

LIBOMP_OMPT_SUPPORT -> OMPT_SUPPORT

runtime/src/kmp_tasking.cpp

538

Remove extra empty line

846

to be removed

852

to be removed

1585–1588

Should it be under OMPT_OPTIONAL?

[hbae]

Just a few small things from me.

runtime/src/exports_so.txt

28

This is not necessary since we have omp_control_tool now.

runtime/src/include/50/ompt.h.var

679–680

Do we still need this function?

runtime/src/kmp_csupport.cpp

2982

Yes, it should be __ompt_get_mutex_impl_type(user_lock)

runtime/src/kmp_tasking.cpp

456

ompt_tool -> ompt_start_tool

460

Remove the commented if condition.

521

task_data is never used.

[omalyshe]

runtime/src/kmp_barrier.cpp

1889

OMPT_CUR_TASK_DATA(this_thr) is defined for this.

runtime/src/kmp_csupport.cpp

486

OMPT_CUR_TASK_INFO () to be used

491

OMPT_CUR_TASK_DATA(this_thr) is defined for this

runtime/src/kmp_ftn_entry.h

359

OMPT_CUR_TASK_INFO() to be used

runtime/src/kmp_runtime.cpp

1207

OMPT_CUR_TASK_INFO() to be used

1392

OMPT_CUR_TASK_DATA(this_thr) is defined for this

7201

OMPT_CUR_TASK_DATA(this_thr) is defined for this

7202

OMPT_CUR_TEAM_DATA to be used

7209

OMPT_CUR_TEAM_INFO() to be used

runtime/src/kmp_tasking.cpp

1625

OMPT_CUR_TEAM_DATA() to be used

runtime/src/kmp_wait_release.h

112

What is pteam?

224–234

OMPT_CUR_TEAM_DATA(this_thr) to be used

225

OMPT_CUR_TASK_DATA(this_thr) to be used

runtime/src/ompt-specific.cpp

266

OMPT_CUR_TEAM_INFO(thr) to be used

267

OMPT_CUR_TEAM_INFO(thr) to be used

271

OMPT_CUR_TASK_INFO() to be used
(and in the line above)

281

OMPT_CUR_TEAM_INFO(thr) to be used

282

OMPT_CUR_TASK_INFO() to be used

293

OMPT_CUR_TEAM_INFO(thr) to be used

298

OMPT_CUR_TASK_INFO() to be used

runtime/src/ompt-specific.h

55

never used

57

never used

59

never used

61

never used

[protze.joachim]

Changes to clang-format factored out into D38837

[protze.joachim]

This diff fixes many of the comments. (I will walk over the comments and mark them as done)
This diff is against master + parent dependent differentals

[protze.joachim]

Marked comments that are solved by latest diff as done.

runtime/src/include/50/ompt.h.var

386

This name is given by the spec. If you really perfer this modification, please create an issue in the OMPT spec repository

482

I agree, this would be better. But we need to change the spec first, before we update this in the implementation.

runtime/test/lit.cfg

114–116

The idea of CHECK-WAIT is to allow testing of a runtime, that does not support "optional" events.
If OMPT_OPTIONAL is off, only CHECK, but non of the optional CHECK-X should be checked

[protze.joachim]

Implemented more requested changes

[protze.joachim]

Applied old clang-format file.

[omalyshe]

LGTM.

[jlpeyton]

A few more comments

runtime/src/kmp_ftn_entry.h

347–349

Remove the xexpand macro here since we aren't matching a version symbol with libgomp

1287

Remove this as well for the same reason

1358

And remove this too.

[protze.joachim]

Addressed latest comments by jlpeyton

[protze.joachim]

runtime/src/kmp_ftn_os.h

54

This is inconsistent with the ifs below: either OMPT_SUPPORT or OMP_50_ENABLED

I would say OMP_50_ENABLED for the user facing functions. The implementation is a stub if OMPT_SUPPORT is off.

[jlpeyton]

runtime/src/kmp_ftn_os.h

54

Yes, can you move this down below OMP_45_ENABLED and wrap it with the OMP_50_ENABLED guard.

420–422

Same with this one.

[Hahnfeld]

Some more comments, there is still a lot of commented code that should be removed.
Another, higher level question: When should frame addresses be set? Currently, that's inconsistently guarded with OMPT_OPTIONAL. However, if I remember correctly this is mandatory?

runtime/CMakeLists.txt

324

Nit: This commit id is probably in the fork.

runtime/src/include/50/ompt.h.var

224

This file might not have been formatted? There are 4 spaces instead of 2 further down...

runtime/src/kmp_gsupport.cpp

155–156

Remove

938

Nothing to do with this patch but an interesting question...

1208–1213

There is IF_OMPT_SUPPORT above, please collapse

runtime/src/kmp_runtime.cpp

1099–1103

This formatting looks weird?

1206–1210

Remove commented coe

1213–1214

Remove

1221–1222

Remove

1473

Remove

1504–1505

Remove

5933–5934

Formatting looks weird

runtime/src/kmp_tasking.cpp

492–494

__ompt_task_finish?

[Hahnfeld]

runtime/src/kmp_tasking.cpp

496–497

Remove this code if that's guaranteed

1458–1459

Remove

1694–1699

Puh, that's ugly. Is this needed for performance reasons?

runtime/src/ompt-event-specific.h

25–27

Remove

runtime/test/ompt/callback.h

59–81

Remove

86–101

Remove

runtime/test/ompt/cancel/cancel_parallel.c

16–19

There is a macro for this!

runtime/test/ompt/loadtool/tool.c

1 ↗

(On Diff #119359)

That's not really a test...

runtime/test/ompt/synchronization/taskwait.c

35–38

Why are these deactivated?

runtime/test/ompt/tasks/task_types.c

112–113

Deactivated?

[omalyshe]

runtime/src/kmp_gsupport.cpp

938

It looks like this can be safely replaced with __kmp_omp_task. The eneral functionality is the same. The only loss would be some statistics (KMP_SET_THREAD_STATE_BLOCK(EXPLICIT_TASK)) that is not default anyway.

[protze.joachim]

runtime/src/include/50/ompt.h.var

224

This file was not formatted.
clang-format has problems with above FOREACH macros and tries to indent each item by another level.

[protze.joachim]

runtime/test/ompt/loadtool/tool.c

1 ↗

(On Diff #119359)

We need the source for the library to be loaded by the other test in this directory (tool_available.c). What is your suggestion for having the code and library available?

We could merge the two files and separate the code by two ifdefs. The test would then compile the same file with two different defines.

[Hahnfeld]

runtime/test/ompt/loadtool/tool.c

1 ↗

(On Diff #119359)

Ah, now I understand the intention. Right, that would be one solution and would keep the code together...
Do we always need a RUN line? What happens if you just remove it?

[protze.joachim]

runtime/test/ompt/loadtool/tool.c

1 ↗

(On Diff #119359)

Without a RUN line, lit complains about the missing RUN line. The lit summary lists the test as UNRESOLVED test.

So, probably merging the two files is the best approach.

[protze.joachim]

Addressing Jonas' comments

[protze.joachim]

runtime/src/kmp_tasking.cpp

1694–1699

The event for this function was identified to incur serious overhead for benchmarks like nqueens/kdtree/fibonacci with if(0) cut-off. Hansang provided this implementation as a performance fix.

[hbae]

runtime/src/kmp_tasking.cpp

1699

I know this is ugly, but we couldn't find a better way to fix ~15% regression in 376.kdtree.

[Hahnfeld]

runtime/src/kmp_tasking.cpp

1699

Duplicating code increases the risk of inconsistencies. Can we use templates so that the compiler can do the work? Something like the following:

template <bool ompt>
kmp_int32 __kmpc_omp_taskwait_temp(ident_t *loc_ref, kmp_int32 gtid) {
  ...
}

kmp_int32 __kmpc_omp_taskwait(ident_t *loc_ref, kmp_int32 gtid) {
#if OMPT_SUPPORT && OMPT_OPTIONAL
  if (UNLIKELY(ompt_enabled.enabled)) {
    return __kmpc_omp_taskwait_temp<true>(loc_ref, gtid);
  }
#endif
  return __kmpc_omp_taskwait_temp<false>(loc_ref, gtid);
}

[hbae]

runtime/src/kmp_tasking.cpp

1699

Yes, we can try this, but I suggest to do it in a separate patch since it requires performance validation.

[Hahnfeld]

runtime/src/kmp_tasking.cpp

1699

But this solution is a maintenance nightmare. If we need time to do performance tests, then the "unoptimized" version should go in at first. So only one function that has all the instrumentation in it

[Hahnfeld]

runtime/src/kmp_tasking.cpp

1699

For reference: In some offline tests with Joachim, I was able to convince the compiler to generate the exact same code. Slightly modified sketch that allows inlining of version without OMPT:

template <bool ompt>
static kmp_int32 __kmpc_omp_taskwait_temp(ident_t *loc_ref, kmp_int32 gtid, void *frame_address, void *return_address) {
  ...
}

OMPT_NOINLINE
static kmp_int32 __kmpc_omp_taskwait_ompt(ident *loc_ref, kmp_int32 gtid, void *frame_address, void *return_address) {
  return __kmpc_omp_taskwait_temp<true>(loc_ref, gtid, frame_address, return_address);
}

kmp_int32 __kmpc_omp_taskwait(ident_t *loc_ref, kmp_int32 gtid) {
#if OMPT_SUPPORT && OMPT_OPTIONAL
  if (UNLIKELY(ompt_enabled.enabled)) {
    return __kmpc_omp_taskwait_temp<true>(loc_ref, gtid, OMPT_GET_FRAME_ADDRESS(1), OMPT_GET_RETURN_ADDRESS(0));
  }
#endif
  return __kmpc_omp_taskwait_temp<false>(loc_ref, gtid, NULL, NULL);
}

Clang 5.0.0 even optimizes out the two NULL parameters.

[protze.joachim]

runtime/src/kmp_tasking.cpp

1699

I agree with Jonas, that this code duplication is a maintenance nightmare. Looking at the object dump, the latest version looks very promising.

We will push an implementation of this modification (also applied for the task_if0 functions) to a new branch of the OMPT github repository later this week. This way you will have the chance to run performance tests.

[hbae]

runtime/src/kmp_tasking.cpp

1699

Sounds good to me.

[protze.joachim]

runtime/src/kmp_tasking.cpp

1699

The modifications are available at:
https://github.com/OpenMPToolsInterface/LLVM-openmp/tree/template-versioning-inline

Comparison of objects dumps for current implementation and templated implementation shows no changes in instructions when built with clang compiler (ignoring changed addresses/names). Therefore I tend to update this patch.

[protze.joachim]

The latest update replaces things like "REQUIRES taskgroup" by explicit "XFAIL gcc" for testcases that are expected to fail because either gcc or the gomp interface of libomp don't support the construct or the construct has no runtime call. (see D38880)

Replaced the two function versions of __kmpc_omp_task_begin_if0, __kmpc_omp_task_complete_if0, __kmpc_omp_taskwait by a templated implementation.

[hbae]

runtime/src/kmp_tasking.cpp

569

Return type is void.

591

Return type is void.

1707

I think the intention here was to call the outlined function __kmpc_omp_taskwait_ompt.

[Hahnfeld]

I think we are getting there :-) however, the current base for the diff looks old, can you update to the changes to .clang-format by @jlpeyton?

runtime/src/kmp_tasking.cpp

1707

Oops, I must have missed that when I pushed my experimental changes to our internal GitLab...

runtime/src/ompt-general.cpp

12

Do we need this include? D39182 removes it again

123

next_tool = (ompt_start_tool_t)dlsym(RTLD_NEXT, "ompt_start_tool"); looks much more readable IMO

[protze.joachim]

Since prerequisites are already in master, now the diff against master. (As Jonas pointed out, the last diff missed the clang-format in the base)

[protze.joachim]

Fixed the return of void. Run clang-format again.

[protze.joachim]

Avoid inlining for the ompt_enabled case as Hansang pointed out

[Hahnfeld]

I think I went over all comments and marked all that are not done yet.

One additional thing that should be doable with a global search-and-replace: This patch uses if (ompt.enabled), __builtin_expect and UNLIKELY inconsistently. At least for the latter two, we should agree on one. I prefer the macro because this makes it easier to no-op should there be a platform that doesn't support __builtin_expect (for whatever reason)

runtime/src/kmp_gsupport.cpp

1233–1234

OMPT_STORE_RETURN_ADDRESS has if (ompt_enabled.enabled) built-in

runtime/src/kmp_tasking.cpp

538

That's not done...

846

Not done yet

runtime/src/ompt-general.cpp

57–61

I think I asked this some days ago: Aren't these defined in ompt.h?

EDIT: Ah, in the TR6 patch!

123

ping

219

Hasn't this been parsed to __kmp_tool_libraries?

228

start_tool = (ompt_start_tool_t)dlsym(RTLD_NEXT, "ompt_start_tool"); as above

runtime/src/ompt-internal.h

34

Why is this here?

[protze.joachim]

Addressing the last remaining comments.

[hbae]

runtime/src/ompt-general.cpp

123

I am fine with the change.

219

__kmp_tool_libraries is defined when the runtime reads environment variables, but ompt_pre_init() is called before that happens, so I think __kmp_tool_libraries is just used for printing the values of OMP_TOOL_LIBRARIES.

228

Looks good to me as above.

[protze.joachim]

Addressing Jonas latest comments.

runtime/src/ompt-general.cpp

219

This function is reached before __kmp_env_initialize, so before __kmp_tool_librariesis set,

runtime/src/ompt-internal.h

34

Below in line 47, this is used to store the pointer for the scheduling task, which is needed to return the frame information.

[protze.joachim]

runtime/src/kmp_tasking.cpp

846

Just changed this locally, waiting for comments before updating the diff again,

[Hahnfeld]

Two minor comments and UNLIKELY vs __builtin_expect

runtime/src/kmp_tasking.cpp

846

Still

runtime/src/ompt-internal.h

34

Sure, shouldn't this live in kmp.h is it's not already there?

[hbae]

I also prefer UNLIKELY over __builtin_expect.

[protze.joachim]

runtime/src/ompt-internal.h

34

Do you suggest to move all ompt_*_info_t to kmp.h?

Or should I just move the include of ompt-internal.h after the forward declarations in kmp.h?

[Hahnfeld]

runtime/src/ompt-internal.h

34

Ah, now I understand why you need this here, didn't get this before.

In this case, I'm fine with this (please add a comment that this file is included before the forward declarations) or you could replace the single use of kmp_taskdata_t with struct kmp_taskdata...

[protze.joachim]

Remove the forward declaration in ompt-internal.h.
Remove unnecessary includes of ompt-internal.h.
Replace all _builtin_expect by the UNLIKELY macro.

[Hahnfeld]

LGTM. Sorry for my continued nit-picking but I think a patch of this size deserves good reviews

[hbae]

LGTM.

[Hahnfeld]

@protze.joachim have you been granted commit access?

[hans]

I believe this broke some tests on Windows (full output below). The failures look pretty basic, like not finding pthread.h. Does the runtime not have test coverage on buildbots?

I'll disable openmp on windows for the 6.0 release and snapshots in the meantime.

[1/2] Running all regression tests
llvm-lit.py: C:/src/llvm\utils\lit\lit\llvm\config.py:334: note: using clang: c:\src\llvm\build.release\bin\clang.EXE
llvm-lit.py: C:/src/llvm/build.release/utils/lit/tests/lit.cfg:61: warning: Could not import psutil. Some tests will be skipped and the --timeout command line argument will not work.
-- Testing: 38733 tests, 32 threads --
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90
FAIL: libomp :: ompt/misc/control_tool_no_ompt_support.c (37124 of 38733)
******************** TEST 'libomp :: ompt/misc/control_tool_no_ompt_support.c' FAILED ********************
Script:
--
C:/src/llvm/build.release/./bin/clang.exe -fopenmp  -I C:/src/llvm/projects/openmp/runtime/test -I C:/src/llvm/build.release/projects/openmp/runtime/src -L C:/src/llvm/build.release/bin  C:\src\llvm\projects\openmp\runtime\test\ompt\misc\control_tool_no_ompt_support.c -o C:\src\llvm\build.release\projects\openmp\runtime\test\ompt\misc\Output\control_tool_no_ompt_support.c.tmp && C:\src\llvm\build.release\projects\openmp\runtime\test\ompt\misc\Output\control_tool_no_ompt_support.c.tmp
--
Exit Code: 1120

Command Output (stdout):
--
$ "C:/src/llvm/build.release/./bin/clang.exe" "-fopenmp" "-I" "C:/src/llvm/projects/openmp/runtime/test" "-I" "C:/src/llvm/build.release/projects/openmp/runtime/src" "-L" "C:/src/llvm/build.release/bin" "C:\src\llvm\projects\openmp\runtime\test\ompt\misc\control_tool_no_ompt_support.c" "-o" "C:\src\llvm\build.release\projects\openmp\runtime\test\ompt\misc\Output\control_tool_no_ompt_support.c.tmp"
# command output:
control_tool_no_ompt_support-3fc04d.o : error LNK2019: unresolved external symbol omp_control_tool referenced in function .omp_outlined.
C:\src\llvm\build.release\projects\openmp\runtime\test\ompt\misc\Output\control_tool_no_ompt_support.c.tmp : fatal error LNK1120: 1 unresolved externals

# command stderr:
clang.exe: error: linker command failed with exit code 1120 (use -v to see invocation)

error: command failed with exit status: 1120

--

********************
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90
FAIL: libomp :: misc_bugs/omp_foreign_thread_team_reuse.c (37143 of 38733)
******************** TEST 'libomp :: misc_bugs/omp_foreign_thread_team_reuse.c' FAILED ********************
Script:
--
C:/src/llvm/build.release/./bin/clang.exe -fopenmp  -I C:/src/llvm/projects/openmp/runtime/test -I C:/src/llvm/build.release/projects/openmp/runtime/src -L C:/src/llvm/build.release/bin  C:\src\llvm\projects\openmp\runtime\test\misc_bugs\omp_foreign_thread_team_reuse.c -o C:\src\llvm\build.release\projects\openmp\runtime\test\misc_bugs\Output\omp_foreign_thread_team_reuse.c.tmp -lpthread && C:\src\llvm\build.release\projects\openmp\runtime\test\misc_bugs\Output\omp_foreign_thread_team_reuse.c.tmp
--
Exit Code: 1181

Command Output (stdout):
--
$ "C:/src/llvm/build.release/./bin/clang.exe" "-fopenmp" "-I" "C:/src/llvm/projects/openmp/runtime/test" "-I" "C:/src/llvm/build.release/projects/openmp/runtime/src" "-L" "C:/src/llvm/build.release/bin" "C:\src\llvm\projects\openmp\runtime\test\misc_bugs\omp_foreign_thread_team_reuse.c" "-o" "C:\src\llvm\build.release\projects\openmp\runtime\test\misc_bugs\Output\omp_foreign_thread_team_reuse.c.tmp" "-lpthread"
# command output:
LINK : fatal error LNK1181: cannot open input file 'pthread.lib'

# command stderr:
In file included from C:\src\llvm\projects\openmp\runtime\test\misc_bugs\omp_foreign_thread_team_reuse.c:3:
C:/src/llvm/projects/openmp/runtime/test\omp_testsuite.h:53:60: warning: format specifies type 'unsigned int' but the argument has type 'DWORD' (aka 'unsigned long') [-Wformat]
    fprintf(stderr, "CreateThread() failed: Error #%u.\n", GetLastError());
                                                   ~~      ^~~~~~~~~~~~~~
                                                   %lu
C:/src/llvm/projects/openmp/runtime/test\omp_testsuite.h:65:13: warning: format specifies type 'unsigned int' but the argument has type 'DWORD' (aka 'unsigned long') [-Wformat]
            GetLastError());
            ^~~~~~~~~~~~~~
C:/src/llvm/projects/openmp/runtime/test\omp_testsuite.h:70:59: warning: format specifies type 'unsigned int' but the argument has type 'DWORD' (aka 'unsigned long') [-Wformat]
    fprintf(stderr, "CloseHandle() failed: Error #%u.\n", GetLastError());
                                                  ~~      ^~~~~~~~~~~~~~
                                                  %lu
C:\src\llvm\projects\openmp\runtime\test\misc_bugs\omp_foreign_thread_team_reuse.c:40:1: warning: control reaches end of non-void function [-Wreturn-type]
}
^
4 warnings generated.
clang.exe: error: linker command failed with exit code 1181 (use -v to see invocation)

error: command failed with exit status: 1181

--

********************
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90
FAIL: libomp :: tasking/bug_nested_proxy_task.c (37154 of 38733)
******************** TEST 'libomp :: tasking/bug_nested_proxy_task.c' FAILED ********************
Script:
--
C:/src/llvm/build.release/./bin/clang.exe -fopenmp  -I C:/src/llvm/projects/openmp/runtime/test -I C:/src/llvm/build.release/projects/openmp/runtime/src -L C:/src/llvm/build.release/bin  C:\src\llvm\projects\openmp\runtime\test\tasking\bug_nested_proxy_task.c -o C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_nested_proxy_task.c.tmp -lpthread && C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_nested_proxy_task.c.tmp
--
Exit Code: 1

Command Output (stdout):
--
$ "C:/src/llvm/build.release/./bin/clang.exe" "-fopenmp" "-I" "C:/src/llvm/projects/openmp/runtime/test" "-I" "C:/src/llvm/build.release/projects/openmp/runtime/src" "-L" "C:/src/llvm/build.release/bin" "C:\src\llvm\projects\openmp\runtime\test\tasking\bug_nested_proxy_task.c" "-o" "C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_nested_proxy_task.c.tmp" "-lpthread"
# command stderr:
C:\src\llvm\projects\openmp\runtime\test\tasking\bug_nested_proxy_task.c:7:10: fatal error: 'pthread.h' file not found
#include <pthread.h>
         ^~~~~~~~~~~
1 error generated.

error: command failed with exit status: 1

--

********************
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90
FAIL: libomp :: tasking/bug_proxy_task_dep_waiting.c (37156 of 38733)
******************** TEST 'libomp :: tasking/bug_proxy_task_dep_waiting.c' FAILED ********************
Script:
--
C:/src/llvm/build.release/./bin/clang.exe -fopenmp  -I C:/src/llvm/projects/openmp/runtime/test -I C:/src/llvm/build.release/projects/openmp/runtime/src -L C:/src/llvm/build.release/bin  C:\src\llvm\projects\openmp\runtime\test\tasking\bug_proxy_task_dep_waiting.c -o C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_proxy_task_dep_waiting.c.tmp -lpthread && C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_proxy_task_dep_waiting.c.tmp
--
Exit Code: 1

Command Output (stdout):
--
$ "C:/src/llvm/build.release/./bin/clang.exe" "-fopenmp" "-I" "C:/src/llvm/projects/openmp/runtime/test" "-I" "C:/src/llvm/build.release/projects/openmp/runtime/src" "-L" "C:/src/llvm/build.release/bin" "C:\src\llvm\projects\openmp\runtime\test\tasking\bug_proxy_task_dep_waiting.c" "-o" "C:\src\llvm\build.release\projects\openmp\runtime\test\tasking\Output\bug_proxy_task_dep_waiting.c.tmp" "-lpthread"
# command stderr:
C:\src\llvm\projects\openmp\runtime\test\tasking\bug_proxy_task_dep_waiting.c:7:10: fatal error: 'pthread.h' file not found
#include <pthread.h>
         ^~~~~~~~~~~
1 error generated.

error: command failed with exit status: 1

--

********************
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90..
Testing Time: 269.12s
********************
Failing Tests (4):
    libomp :: misc_bugs/omp_foreign_thread_team_reuse.c
    libomp :: ompt/misc/control_tool_no_ompt_support.c
    libomp :: tasking/bug_nested_proxy_task.c
    libomp :: tasking/bug_proxy_task_dep_waiting.c

  Expected Passes    : 37046
  Expected Failures  : 243
  Unsupported Tests  : 1440
  Unexpected Failures: 4

[hans]

Also, r317085 doesn't seem to have gone to any mailing list, which is unfortunate. Maybe because the committer was not subscribed and it got stuck in moderation.

[protze.joachim]

In D38185#985047, @hans wrote:

Also, r317085 doesn't seem to have gone to any mailing list, which is unfortunate. Maybe because the committer was not subscribed and it got stuck in moderation.

I think at the time of the commit, I was not subscribed with the matching mail address :( And I guess the size of the diff might also be a reason for the mailing list to put the mail on wait.

Of the failing tests, we only added one with the OMPT commit:

libomp :: misc_bugs/omp_foreign_thread_team_reuse.c

https://reviews.llvm.org/D23084

libomp :: ompt/misc/control_tool_no_ompt_support.c

by this commit.

libomp :: tasking/bug_nested_proxy_task.c

https://reviews.llvm.org/D23115

libomp :: tasking/bug_proxy_task_dep_waiting.c

https://reviews.llvm.org/D23116

I uploaded a diff, that should address all the broken test cases:
https://reviews.llvm.org/D42427

[hans]

In D38185#985231, @protze.joachim wrote:

In D38185#985047, @hans wrote:

Also, r317085 doesn't seem to have gone to any mailing list, which is unfortunate. Maybe because the committer was not subscribed and it got stuck in moderation.

I think at the time of the commit, I was not subscribed with the matching mail address :( And I guess the size of the diff might also be a reason for the mailing list to put the mail on wait.

Of the failing tests, we only added one with the OMPT commit:

libomp :: misc_bugs/omp_foreign_thread_team_reuse.c

https://reviews.llvm.org/D23084

libomp :: ompt/misc/control_tool_no_ompt_support.c

by this commit.

libomp :: tasking/bug_nested_proxy_task.c

https://reviews.llvm.org/D23115

libomp :: tasking/bug_proxy_task_dep_waiting.c

https://reviews.llvm.org/D23116

I uploaded a diff, that should address all the broken test cases:
https://reviews.llvm.org/D42427

Thanks!

openmp/trunk/runtime/src/ompt-general.cpp

118 ↗

(On Diff #121108)

This fails on Windows obviously. When building with clang it looks like this:

[903/3933] Building CXX object projects\openmp\runtime\src\CMakeFiles\omp.dir\ompt-general.cpp.obj
FAILED: projects/openmp/runtime/src/CMakeFiles/omp.dir/ompt-general.cpp.obj
\src\llvm\build.release\bin\clang-cl.exe  /nologo -TP -D_CRT_NONSTDC_NO_DEPRECATE -D_CRT_NONSTDC_NO_WARNINGS -D_CRT_SECURE_NO_DEPRECATE -D_CRT_SECURE_NO_WARNINGS -D_DEBUG_POINTER_IMPL="" -D_SCL_SECURE_NO_DEPRECATE -D_SCL_SECURE_NO_WARNINGS -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -Domp_EXPORTS -Iprojects\openmp\runtime\src -I..\projects\openmp\runtime\src -Iinclude -I..\include -I..\projects\openmp\runtime\src\i18n -I..\projects\openmp\runtime\src\include\50 -I..\projects\openmp\runtime\src\thirdparty\ittnotify /DWIN32 /D_WINDOWS   /Zc:inline /Zc:strictStrings /Oi /Zc:rvalueCast /Brepro /W4  -W -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -Wcovered-switch-default -Wdelete-non-virtual-dtor -Wstring-conversion /MT /O2 /Ob2   -UNDEBUG -D _CRT_SECURE_NO_WARNINGS -D _CRT_SECURE_NO_DEPRECATE -D _WINDOWS -D _WINNT -D _WIN32_WINNT=0x0501 -D _USRDLL -Wno-sign-compare -Wno-unused-function -Wno-unused-local-typedef -Wno-unused-value -Wno-unused-variable -Wno-switch -Wno-covered-switch-default -Wno-deprecated-register -Wno-gnu-anonymous-struct -Wno-unknown-pragmas -Wno-missing-field-initializers -Wno-missing-braces -Wno-comment -Wno-self-assign -Wno-vla-extension -Wno-format-pedantic /GS /EHsc /showIncludes /Foprojects\openmp\runtime\src\CMakeFiles\omp.dir\ompt-general.cpp.obj /Fdprojects\openmp\runtime\src\CMakeFiles\omp.dir\ -c ..\projects\openmp\runtime\src\ompt-general.cpp
..\projects\openmp\runtime\src\ompt-general.cpp(131,32):  error: use of undeclared identifier 'RTLD_NEXT'
      (ompt_start_tool_t)dlsym(RTLD_NEXT, "ompt_start_tool");
                               ^

But for some reason I don't see that when building with MSVC.

[Hahnfeld]

In D38185#985231, @protze.joachim wrote:

Of the failing tests, we only added one with the OMPT commit:

libomp :: misc_bugs/omp_foreign_thread_team_reuse.c

https://reviews.llvm.org/D23084

libomp :: ompt/misc/control_tool_no_ompt_support.c

by this commit.

libomp :: tasking/bug_nested_proxy_task.c

https://reviews.llvm.org/D23115

libomp :: tasking/bug_proxy_task_dep_waiting.c

https://reviews.llvm.org/D23116

Except for the OMPT test all of those were in August 2016 (!) and should be in 4.0 and 5.0...

Does the runtime not have test coverage on buildbots?

Not on Windows and not as good as I would like it to be on Linux. We (IT Center, RWTH Aachen University) are looking into improving the latter. Meanwhile, Intel is saying that they do extensive testing of the runtime - whatever that means on Windows...

[Hahnfeld]

openmp/trunk/runtime/src/ompt-general.cpp

118 ↗

(On Diff #121108)

Because it's #if KMP_OS_UNIX. If that's not set with clang-cl on Windows then there is definitely something wrong!

[omalyshe]

Intel is saying that they do extensive testing of the runtime - whatever that means on Windows...

Currently we have limited testing of OMPT features on Windows.

runtime/CMakeLists.txt

 if(LIBOMP_USE_VERSION_SYMBOLS AND (NOT LIBOMP_ENABLE_SHARED) )
   message(STATUS "Version symbols not supported for static libraries - forcing Version symbols functionality off")
   set (LIBOMP_USE_VERSION_SYMBOLS FALSE)
 endif()
 
 # OMPT-support
 set(LIBOMP_OMPT_DEBUG FALSE CACHE BOOL
   "Trace OMPT initialization?")
+#after testing: turn on ompt support by default for OpenMP 5.0 and higher
 set(LIBOMP_OMPT_SUPPORT FALSE CACHE BOOL
   "OMPT-support?")
-set(LIBOMP_OMPT_BLAME TRUE CACHE BOOL
-  "OMPT-blame?")
-set(LIBOMP_OMPT_TRACE TRUE CACHE BOOL
-  "OMPT-trace?")
+set(LIBOMP_OMPT_OPTIONAL TRUE CACHE BOOL
+  "OMPT-optional?")
 if(LIBOMP_OMPT_SUPPORT AND (NOT LIBOMP_HAVE_OMPT_SUPPORT))
   libomp_error_say("OpenMP Tools Interface requested but not available in this implementation")
 endif()
 if(LIBOMP_OMPT_SUPPORT AND (${LIBOMP_OMP_VERSION} LESS 50))
   libomp_error_say("OpenMP Tools Interface only available with OpenMP 5.0, LIBOMP_OMP_VERSION is ${LIBOMP_OMP_VERSION}")
 endif()
 
 # TSAN-support
 set(LIBOMP_TSAN_SUPPORT FALSE CACHE BOOL
else()
     set(LIBOMP_BUILD ${LIBOMP_VERSION_BUILD})
   endif()
   libomp_say("Build                -- ${LIBOMP_BUILD}")
   libomp_say("Use Stats-gathering  -- ${LIBOMP_STATS}")
   libomp_say("Use Debugger-support -- ${LIBOMP_USE_DEBUGGER}")
   libomp_say("Use ITT notify       -- ${LIBOMP_USE_ITT_NOTIFY}")
   libomp_say("Use OMPT-support     -- ${LIBOMP_OMPT_SUPPORT}")
   if(${LIBOMP_OMPT_SUPPORT})
-    libomp_say("Use OMPT-blame       -- ${LIBOMP_OMPT_BLAME}")
-    libomp_say("Use OMPT-trace       -- ${LIBOMP_OMPT_TRACE}")
+    libomp_say("Use OMPT-optional  -- ${LIBOMP_OMPT_OPTIONAL}")
   endif()
   libomp_say("Use Adaptive locks   -- ${LIBOMP_USE_ADAPTIVE_LOCKS}")
   libomp_say("Use quad precision   -- ${LIBOMP_USE_QUAD_PRECISION}")
   libomp_say("Use TSAN-support     -- ${LIBOMP_TSAN_SUPPORT}")
   libomp_say("Use Hwloc library    -- ${LIBOMP_USE_HWLOC}")
 endif()
 
 add_subdirectory(src)
 add_subdirectory(test)

runtime/src/exports_so.txt

global: # Exported symbols.
         #
         # "Normal" symbols.
         #
         omp_*;     # Standard OpenMP functions.
 
         #
         # OMPT API
         #
-        ompt_tool;           # OMPT initialization interface
-        ompt_control;        # OMPT control interface
+        ompt_start_tool;     # OMPT start interface
 
         # icc drops weak attribute at linking step without the following line:
         Annotate*;           # TSAN annotation
 
         #
         # OMPT state placeholders
         #
         ompt_idle;

 OMP_2.0 {
 } OMP_1.0;
 OMP_3.0 {
 } OMP_2.0;
 OMP_3.1 {
 } OMP_3.0;
 OMP_4.0 {
 } OMP_3.1;
 
 # sets up GCC GOMP_ version dependency chain
 GOMP_1.0 {
 };
 GOMP_2.0 {
 } GOMP_1.0;
 GOMP_3.0 {
 } GOMP_2.0;
 GOMP_4.0 {
 } GOMP_3.0;
 
 # end of file #

runtime/src/include/50/omp.h.var

#   include <stdlib.h>
     extern void * __KAI_KMPC_CONVENTION  kmp_aligned_malloc  (size_t, size_t);
     extern void * __KAI_KMPC_CONVENTION  kmp_calloc  (size_t, size_t);
     extern void * __KAI_KMPC_CONVENTION  kmp_realloc (void *, size_t);
     extern void   __KAI_KMPC_CONVENTION  kmp_free    (void *);
 
     extern void   __KAI_KMPC_CONVENTION  kmp_set_warnings_on(void);
     extern void   __KAI_KMPC_CONVENTION  kmp_set_warnings_off(void);
 
+    /* OpenMP 5.0 Tool Control */
+    typedef enum omp_control_tool_result_t {
+        omp_control_tool_notool = -2,
+        omp_control_tool_nocallback = -1,
+        omp_control_tool_success = 0,
+        omp_control_tool_ignored = 1
+    } omp_control_tool_result_t;
+
+    typedef enum omp_control_tool_t {
+        omp_control_tool_start = 1,
+        omp_control_tool_pause = 2,
+        omp_control_tool_flush = 3,
+        omp_control_tool_end = 4
+    } omp_control_tool_t;
+    
+    extern int __KAI_KMPC_CONVENTION omp_control_tool(int, int, void*);
+
 #   undef __KAI_KMPC_CONVENTION
 
     /* Warning:
        The following typedefs are not standard, deprecated and will be removed in a future release.
     */
     typedef int     omp_int_t;
     typedef double  omp_wtime_t;
 
 #   ifdef __cplusplus
     }
 #   endif
 
 #endif /* __OMP_H */
 

runtime/src/include/50/omp_lib.h.var

!DIR$ fixedformlinesize:132
       integer, parameter :: omp_lock_kind          = int_ptr_kind()
       integer, parameter :: omp_nest_lock_kind     = int_ptr_kind()
       integer, parameter :: omp_sched_kind         = omp_integer_kind
       integer, parameter :: omp_proc_bind_kind     = omp_integer_kind
       integer, parameter :: kmp_pointer_kind       = int_ptr_kind()
       integer, parameter :: kmp_size_t_kind        = int_ptr_kind()
       integer, parameter :: kmp_affinity_mask_kind = int_ptr_kind()
       integer, parameter :: omp_lock_hint_kind     = omp_integer_kind
+      integer, parameter :: omp_control_tool_kind  = omp_integer_kind
+      integer, parameter :: omp_control_tool_result_kind = omp_integer_kind
 
       integer (kind=omp_integer_kind), parameter :: openmp_version    = @LIBOMP_OMP_YEAR_MONTH@
       integer (kind=omp_integer_kind), parameter :: kmp_version_major = @LIBOMP_VERSION_MAJOR@
       integer (kind=omp_integer_kind), parameter :: kmp_version_minor = @LIBOMP_VERSION_MINOR@
       integer (kind=omp_integer_kind), parameter :: kmp_version_build = @LIBOMP_VERSION_BUILD@
       character(*)               kmp_build_date
       parameter( kmp_build_date = '@LIBOMP_BUILD_DATE@' )
 
!DIR$ fixedformlinesize:132
       integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_uncontended    = 1
       integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_contended      = 2
       integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_nonspeculative = 4
       integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_speculative    = 8
       integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_hle            = 65536
       integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_rtm            = 131072
       integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_adaptive       = 262144
 
+      integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_start = 1
+      integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_pause = 2
+      integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_flush = 3
+      integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_end = 4
+
+      integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_notool = -2
+      integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_nocallback = -1
+      integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_success = 0
+      integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_ignored = 1
+
       interface
 
 !       ***
 !       *** omp_* entry points
 !       ***
 
         subroutine omp_set_num_threads(num_threads) bind(c)
           import
!       ***
         end subroutine omp_init_lock_with_hint
 
         subroutine omp_init_nest_lock_with_hint(nvar, hint) bind(c)
           import
           integer (kind=omp_nest_lock_kind) nvar
           integer (kind=omp_lock_hint_kind), value :: hint
         end subroutine omp_init_nest_lock_with_hint
 
+        function omp_control_tool(command, modifier) bind(c)
+          import
+          integer (kind=omp_integer_kind) omp_control_tool
+          integer (kind=omp_control_tool_kind), value :: command
+          integer (kind=omp_control_tool_kind), value :: modifier
+        end function omp_control_tool
+
       end interface
 
 !DIR$ IF DEFINED (__INTEL_OFFLOAD)
 !DIR$ ATTRIBUTES OFFLOAD:MIC :: omp_set_num_threads
 !DIR$ ATTRIBUTES OFFLOAD:MIC :: omp_set_dynamic
 !DIR$ ATTRIBUTES OFFLOAD:MIC :: omp_set_nested
 !DIR$ ATTRIBUTES OFFLOAD:MIC :: omp_get_num_threads
 !DIR$ ATTRIBUTES OFFLOAD:MIC :: omp_get_max_threads

runtime/src/include/50/omp_lib.f.var

!dec$ fixedformlinesize:132
         integer, parameter :: omp_nest_lock_kind     = int_ptr_kind()
         integer, parameter :: omp_sched_kind         = omp_integer_kind
         integer, parameter :: omp_proc_bind_kind     = omp_integer_kind
         integer, parameter :: kmp_pointer_kind       = int_ptr_kind()
         integer, parameter :: kmp_size_t_kind        = int_ptr_kind()
         integer, parameter :: kmp_affinity_mask_kind = int_ptr_kind()
         integer, parameter :: kmp_cancel_kind        = omp_integer_kind
         integer, parameter :: omp_lock_hint_kind     = omp_integer_kind
+        integer, parameter :: omp_control_tool_kind  = omp_integer_kind
+        integer, parameter :: omp_control_tool_result_kind = omp_integer_kind
 
       end module omp_lib_kinds
 
       module omp_lib
 
         use omp_lib_kinds
 
         integer (kind=omp_integer_kind), parameter :: kmp_version_major = @LIBOMP_VERSION_MAJOR@
!         ***
           end subroutine omp_init_lock_with_hint
 
           subroutine omp_init_nest_lock_with_hint(nvar, hint)
             use omp_lib_kinds
             integer (kind=omp_nest_lock_kind) nvar
             integer (kind=omp_lock_hint_kind) hint
           end subroutine omp_init_nest_lock_with_hint
 
+          function omp_control_tool(command, modifier)
+            use omp_lib_kinds
+            integer (kind=omp_integer_kind) omp_control_tool
+            integer (kind=omp_control_tool_kind) command
+            integer (kind=omp_control_tool_kind) modifier
+          end function omp_control_tool
+
         end interface
 
 !dec$ if defined(_WIN32)
 !dec$   if defined(_WIN64) .or. defined(_M_AMD64)
 
 !***
 !*** The Fortran entry points must be in uppercase, even if the /Qlowercase
 !*** option is specified.  The alias attribute ensures that the specified

 !dec$ attributes alias:'OMP_GET_DEFAULT_DEVICE' :: omp_get_default_device
 !dec$ attributes alias:'OMP_SET_DEFAULT_DEVICE' :: omp_set_default_device
 !dec$ attributes alias:'OMP_GET_NUM_DEVICES' :: omp_get_num_devices
 !dec$ attributes alias:'OMP_GET_NUM_TEAMS' :: omp_get_num_teams
 !dec$ attributes alias:'OMP_GET_TEAM_NUM' :: omp_get_team_num
 !dec$ attributes alias:'OMP_GET_CANCELLATION' :: omp_get_cancellation
 !dec$ attributes alias:'OMP_IS_INITIAL_DEVICE' :: omp_is_initial_device
 !dec$ attributes alias:'OMP_GET_MAX_TASK_PRIORITY' :: omp_get_max_task_priority
+!dec$ attributes alias:'OMP_CONTROL_TOOL' :: omp_control_tool
 
 !dec$ attributes alias:'omp_init_lock' :: omp_init_lock
 !dec$ attributes alias:'omp_init_lock_with_hint' :: omp_init_lock_with_hint
 !dec$ attributes alias:'omp_destroy_lock' :: omp_destroy_lock
 !dec$ attributes alias:'omp_set_lock' :: omp_set_lock
 !dec$ attributes alias:'omp_unset_lock' :: omp_unset_lock
 !dec$ attributes alias:'omp_test_lock' :: omp_test_lock
 !dec$ attributes alias:'omp_init_nest_lock' :: omp_init_nest_lock

 !dec$ attributes alias:'KMP_REALLOC'::kmp_realloc
 !dec$ attributes alias:'KMP_FREE'::kmp_free
 
 !dec$ attributes alias:'KMP_SET_WARNINGS_ON'::kmp_set_warnings_on
 !dec$ attributes alias:'KMP_SET_WARNINGS_OFF'::kmp_set_warnings_off
 
 !dec$ attributes alias:'KMP_GET_CANCELLATION_STATUS' :: kmp_get_cancellation_status
 
 !dec$   else
 
 !***
 !*** On Windows* OS IA-32 architecture, the Fortran entry points have an underscore prepended.
 !***
 
 !dec$ attributes alias:'_OMP_SET_NUM_THREADS' :: omp_set_num_threads
 !dec$ attributes alias:'_OMP_SET_DYNAMIC' :: omp_set_dynamic
 !dec$ attributes alias:'_OMP_SET_NESTED' :: omp_set_nested

 !dec$ attributes alias:'_OMP_GET_DEFAULT_DEVICE' :: omp_get_default_device
 !dec$ attributes alias:'_OMP_SET_DEFAULT_DEVICE' :: omp_set_default_device
 !dec$ attributes alias:'_OMP_GET_NUM_DEVICES' :: omp_get_num_devices
 !dec$ attributes alias:'_OMP_GET_NUM_TEAMS' :: omp_get_num_teams
 !dec$ attributes alias:'_OMP_GET_TEAM_NUM' :: omp_get_team_num
 !dec$ attributes alias:'_OMP_GET_CANCELLATION' :: omp_get_cancellation
 !dec$ attributes alias:'_OMP_IS_INITIAL_DEVICE' :: omp_is_initial_device
 !dec$ attributes alias:'_OMP_GET_MAX_TASK_PRIORTY' :: omp_get_max_task_priority
+!dec$ attributes alias:'_OMP_CONTROL_TOOL' :: omp_control_tool
 
 !dec$ attributes alias:'_omp_init_lock' :: omp_init_lock
 !dec$ attributes alias:'_omp_init_lock_with_hint' :: omp_init_lock_with_hint
 !dec$ attributes alias:'_omp_destroy_lock' :: omp_destroy_lock
 !dec$ attributes alias:'_omp_set_lock' :: omp_set_lock
 !dec$ attributes alias:'_omp_unset_lock' :: omp_unset_lock
 !dec$ attributes alias:'_omp_test_lock' :: omp_test_lock
 !dec$ attributes alias:'_omp_init_nest_lock' :: omp_init_nest_lock

 !dec$ attributes alias:'_KMP_REALLOC'::kmp_realloc
 !dec$ attributes alias:'_KMP_FREE'::kmp_free
 
 !dec$ attributes alias:'_KMP_SET_WARNINGS_ON'::kmp_set_warnings_on
 !dec$ attributes alias:'_KMP_SET_WARNINGS_OFF'::kmp_set_warnings_off
 
 !dec$ attributes alias:'_KMP_GET_CANCELLATION_STATUS' :: kmp_get_cancellation_status
 
 !dec$   endif
 !dec$ endif
 
 !dec$ if defined(__linux)
 
 !***
 !*** The Linux* OS entry points are in lowercase, with an underscore appended.
 !***
 

 !dec$ attributes alias:'omp_unset_lock_'::omp_unset_lock
 !dec$ attributes alias:'omp_test_lock_'::omp_test_lock
 !dec$ attributes alias:'omp_init_nest_lock_'::omp_init_nest_lock
 !dec$ attributes alias:'omp_init_nest_lock_with_hint_'::omp_init_nest_lock_with_hint
 !dec$ attributes alias:'omp_destroy_nest_lock_'::omp_destroy_nest_lock
 !dec$ attributes alias:'omp_set_nest_lock_'::omp_set_nest_lock
 !dec$ attributes alias:'omp_unset_nest_lock_'::omp_unset_nest_lock
 !dec$ attributes alias:'omp_test_nest_lock_'::omp_test_nest_lock
+!dec$ attributes alias:'omp_control_tool_'::omp_control_tool
 
 !dec$ attributes alias:'kmp_set_stacksize_'::kmp_set_stacksize
 !dec$ attributes alias:'kmp_set_stacksize_s_'::kmp_set_stacksize_s
 !dec$ attributes alias:'kmp_set_blocktime_'::kmp_set_blocktime
 !dec$ attributes alias:'kmp_set_library_serial_'::kmp_set_library_serial
 !dec$ attributes alias:'kmp_set_library_turnaround_'::kmp_set_library_turnaround
 !dec$ attributes alias:'kmp_set_library_throughput_'::kmp_set_library_throughput
 !dec$ attributes alias:'kmp_set_library_'::kmp_set_library

 !dec$ attributes alias:'kmp_calloc_'::kmp_calloc
 !dec$ attributes alias:'kmp_realloc_'::kmp_realloc
 !dec$ attributes alias:'kmp_free_'::kmp_free
 
 !dec$ attributes alias:'kmp_set_warnings_on_'::kmp_set_warnings_on
 !dec$ attributes alias:'kmp_set_warnings_off_'::kmp_set_warnings_off
 !dec$ attributes alias:'kmp_get_cancellation_status_'::kmp_get_cancellation_status
 
 !dec$ endif
 
 !dec$ if defined(__APPLE__)
 
 !***
 !*** The Mac entry points are in lowercase, with an both an underscore
 !*** appended and an underscore prepended.
 !***
 

 !dec$ attributes alias:'_omp_unset_lock_'::omp_unset_lock
 !dec$ attributes alias:'_omp_test_lock_'::omp_test_lock
 !dec$ attributes alias:'_omp_init_nest_lock_'::omp_init_nest_lock
 !dec$ attributes alias:'_omp_init_nest_lock_with_hint_'::omp_init_nest_lock_with_hint
 !dec$ attributes alias:'_omp_destroy_nest_lock_'::omp_destroy_nest_lock
 !dec$ attributes alias:'_omp_set_nest_lock_'::omp_set_nest_lock
 !dec$ attributes alias:'_omp_unset_nest_lock_'::omp_unset_nest_lock
 !dec$ attributes alias:'_omp_test_nest_lock_'::omp_test_nest_lock
+!dec$ attributes alias:'_omp_control_tool_'::omp_control_tool
 
 !dec$ attributes alias:'_kmp_set_stacksize_'::kmp_set_stacksize
 !dec$ attributes alias:'_kmp_set_stacksize_s_'::kmp_set_stacksize_s
 !dec$ attributes alias:'_kmp_set_blocktime_'::kmp_set_blocktime
 !dec$ attributes alias:'_kmp_set_library_serial_'::kmp_set_library_serial
 !dec$ attributes alias:'_kmp_set_library_turnaround_'::kmp_set_library_turnaround
 !dec$ attributes alias:'_kmp_set_library_throughput_'::kmp_set_library_throughput
 !dec$ attributes alias:'_kmp_set_library_'::kmp_set_library

 !dec$ attributes alias:'_kmp_realloc_'::kmp_realloc
 !dec$ attributes alias:'_kmp_free_'::kmp_free
 
 !dec$ attributes alias:'_kmp_set_warnings_on_'::kmp_set_warnings_on
 !dec$ attributes alias:'_kmp_set_warnings_off_'::kmp_set_warnings_off
 
 !dec$ attributes alias:'_kmp_get_cancellation_status_'::kmp_get_cancellation_status
 
 !dec$ endif
 
       end module omp_lib
 

runtime/src/include/50/omp_lib.f90.var

!
         integer, parameter :: omp_nest_lock_kind     = c_intptr_t
         integer, parameter :: omp_sched_kind         = omp_integer_kind
         integer, parameter :: omp_proc_bind_kind     = omp_integer_kind
         integer, parameter :: kmp_pointer_kind       = c_intptr_t
         integer, parameter :: kmp_size_t_kind        = c_size_t
         integer, parameter :: kmp_affinity_mask_kind = c_intptr_t
         integer, parameter :: kmp_cancel_kind        = omp_integer_kind
         integer, parameter :: omp_lock_hint_kind     = omp_integer_kind
+        integer, parameter :: omp_control_tool_kind  = omp_integer_kind
+        integer, parameter :: omp_control_tool_result_kind = omp_integer_kind
 
       end module omp_lib_kinds
 
       module omp_lib
 
         use omp_lib_kinds
 
         integer (kind=omp_integer_kind), parameter :: openmp_version    = @LIBOMP_OMP_YEAR_MONTH@
!
         integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_uncontended    = 1
         integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_contended      = 2
         integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_nonspeculative = 4
         integer (kind=omp_lock_hint_kind), parameter :: omp_lock_hint_speculative    = 8
         integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_hle            = 65536
         integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_rtm            = 131072
         integer (kind=omp_lock_hint_kind), parameter :: kmp_lock_hint_adaptive       = 262144
 
+        integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_start = 1
+        integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_pause = 2
+        integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_flush = 3
+        integer (kind=omp_control_tool_kind), parameter :: omp_control_tool_end = 4
+
+        integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_notool = -2
+        integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_nocallback = -1
+        integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_success = 0
+        integer (kind=omp_control_tool_result_kind), parameter :: omp_control_tool_ignored = 1
+
         interface
 
 !         ***
 !         *** omp_* entry points
 !         ***
 
           subroutine omp_set_num_threads(num_threads) bind(c)
             use omp_lib_kinds
!         ***
           end subroutine omp_init_lock_with_hint
 
           subroutine omp_init_nest_lock_with_hint(nvar, hint) bind(c)
             use omp_lib_kinds
             integer (kind=omp_nest_lock_kind) nvar
             integer (kind=omp_lock_hint_kind), value :: hint
           end subroutine omp_init_nest_lock_with_hint
 
+          function omp_control_tool(command, modifier) bind(c)
+            use omp_lib_kinds
+            integer (kind=omp_integer_kind) omp_control_tool
+            integer (kind=omp_control_tool_kind), value :: command
+            integer (kind=omp_control_tool_kind), value :: modifier
+          end function omp_control_tool
+
         end interface
 
       end module omp_lib

runtime/src/include/50/ompt.h.var

 /*
  * include/50/ompt.h.var
  */
 
 #ifndef __OMPT__
 #define __OMPT__
 
 /*****************************************************************************
  * system include files
  *****************************************************************************/
 
 #include <stdint.h>
+#include <stddef.h>
 
 
 
 /*****************************************************************************
  * iteration macros
  *****************************************************************************/
 
-#define FOREACH_OMPT_INQUIRY_FN(macro)  \
-    macro (ompt_enumerate_state)        \
+#define FOREACH_OMPT_INQUIRY_FN(macro)      \
+    macro (ompt_enumerate_states)           \
+    macro (ompt_enumerate_mutex_impls)      \
                                             \
-    macro (ompt_set_callback)           \
-    macro (ompt_get_callback)           \
+    macro (ompt_set_callback)               \
+    macro (ompt_get_callback)               \
                                             \
-    macro (ompt_get_idle_frame)         \
-    macro (ompt_get_task_frame)         \
-                                        \
-    macro (ompt_get_state)              \
+    macro (ompt_get_state)                  \
                                             \
-    macro (ompt_get_parallel_id)        \
-    macro (ompt_get_parallel_team_size) \
-    macro (ompt_get_task_id)            \
-    macro (ompt_get_thread_id)
+    macro (ompt_get_parallel_info)          \
+    macro (ompt_get_task_info)              \
+    macro (ompt_get_thread_data)            \
+    macro (ompt_get_unique_id)              \
+                                            \
+    macro(ompt_get_num_places)              \
+    macro(ompt_get_place_proc_ids)          \
+    macro(ompt_get_place_num)               \
+    macro(ompt_get_partition_place_nums)    \
+    macro(ompt_get_proc_id)                 \
+                                            \
+    macro(ompt_get_target_info)             \
+    macro(ompt_get_num_devices)
 
 #define FOREACH_OMPT_PLACEHOLDER_FN(macro)  \
     macro (ompt_idle)                       \
     macro (ompt_overhead)                   \
     macro (ompt_barrier_wait)               \
     macro (ompt_task_wait)                  \
     macro (ompt_mutex_wait)
 
-#define FOREACH_OMPT_STATE(macro)                                                               \
+#define FOREACH_OMP_STATE(macro)                                                                \
                                                                                                 \
-    /* first */                                                                                 \
-    macro (ompt_state_first, 0x71)          /* initial enumeration state */                     \
+    /* first available state */                                                                 \
+    macro (omp_state_undefined, 0x102)      /* undefined thread state */                        \
                                                                                                 \
     /* work states (0..15) */                                                                   \
-    macro (ompt_state_work_serial, 0x00)    /* working outside parallel */                      \
-    macro (ompt_state_work_parallel, 0x01)  /* working within parallel */                       \
-    macro (ompt_state_work_reduction, 0x02) /* performing a reduction */                        \
-                                                                                                \
-    /* idle (16..31) */                                                                         \
-    macro (ompt_state_idle, 0x10)            /* waiting for work */                             \
-                                                                                                \
-    /* overhead states (32..63) */                                                              \
-    macro (ompt_state_overhead, 0x20)        /* overhead excluding wait states */               \
-                                                                                                \
-    /* barrier wait states (64..79) */                                                          \
-    macro (ompt_state_wait_barrier, 0x40)    /* waiting at a barrier */                         \
-    macro (ompt_state_wait_barrier_implicit, 0x41)    /* implicit barrier */                    \
-    macro (ompt_state_wait_barrier_explicit, 0x42)    /* explicit barrier */                    \
-                                                                                                \
-    /* task wait states (80..95) */                                                             \
-    macro (ompt_state_wait_taskwait, 0x50)   /* waiting at a taskwait */                        \
-    macro (ompt_state_wait_taskgroup, 0x51)  /* waiting at a taskgroup */                       \
-                                                                                                \
-    /* mutex wait states (96..111) */                                                           \
-    macro (ompt_state_wait_lock, 0x60)       /* waiting for lock */                             \
-    macro (ompt_state_wait_nest_lock, 0x61)  /* waiting for nest lock */                        \
-    macro (ompt_state_wait_critical, 0x62)   /* waiting for critical */                         \
-    macro (ompt_state_wait_atomic, 0x63)     /* waiting for atomic */                           \
-    macro (ompt_state_wait_ordered, 0x64)    /* waiting for ordered */                          \
-    macro (ompt_state_wait_single, 0x6F)     /* waiting for single region (non-standard!) */    \
-                                                                                                \
-    /* misc (112..127) */                                                                       \
-    macro (ompt_state_undefined, 0x70)       /* undefined thread state */
-
+    macro (omp_state_work_serial, 0x000)    /* working outside parallel */                      \
+    macro (omp_state_work_parallel, 0x001)  /* working within parallel */                       \
+    macro (omp_state_work_reduction, 0x002) /* performing a reduction */                        \
+                                                                                                \
+    /* barrier wait states (16..31) */                                                          \
+    macro (omp_state_wait_barrier, 0x010)   /* waiting at a barrier */                          \
+    macro (omp_state_wait_barrier_implicit_parallel, 0x011)                                     \
+                                            /* implicit barrier at the end of parallel region */\
+    macro (omp_state_wait_barrier_implicit_workshare, 0x012)                                    \
+                                            /* implicit barrier at the end of worksharing */    \
+    macro (omp_state_wait_barrier_implicit, 0x013)  /* implicit barrier */                      \
+    macro (omp_state_wait_barrier_explicit, 0x014)  /* explicit barrier */                      \
+                                                                                                \
+    /* task wait states (32..63) */                                                             \
+    macro (omp_state_wait_taskwait, 0x020)  /* waiting at a taskwait */                         \
+    macro (omp_state_wait_taskgroup, 0x021) /* waiting at a taskgroup */                        \
+                                                                                                \
+    /* mutex wait states (64..127) */                                                           \
+    macro (omp_state_wait_mutex, 0x040)                                                         \
+    macro (omp_state_wait_lock, 0x041)      /* waiting for lock */                              \
+    macro (omp_state_wait_critical, 0x042)  /* waiting for critical */                          \
+    macro (omp_state_wait_atomic, 0x043)    /* waiting for atomic */                            \
+    macro (omp_state_wait_ordered, 0x044)   /* waiting for ordered */                           \
+                                                                                                \
+    /* target wait states (128..255) */                                                         \
+    macro (omp_state_wait_target, 0x080)        /* waiting for target region */                 \
+    macro (omp_state_wait_target_map, 0x081)    /* waiting for target data mapping operation */ \
+    macro (omp_state_wait_target_update, 0x082) /* waiting for target update operation */       \
+                                                                                                \
+    /* misc (256..511) */                                                                       \
+    macro (omp_state_idle, 0x100)           /* waiting for work */                              \
+    macro (omp_state_overhead, 0x101)       /* overhead excluding wait states */                \
+                                                                                                \
+    /* implementation-specific states (512..) */
+
+
+#define FOREACH_OMPT_MUTEX_IMPL(macro)                                                \
+    macro (ompt_mutex_impl_unknown, 0)      /* unknown implementatin */               \
+    macro (ompt_mutex_impl_spin, 1)         /* based on spin */                       \
+    macro (ompt_mutex_impl_queuing, 2)      /* based on some fair policy */           \
+    macro (ompt_mutex_impl_speculative, 3)  /* based on HW-supported speculation */
 
 #define FOREACH_OMPT_EVENT(macro)                                                                               \
                                                                                                                 \
     /*--- Mandatory Events ---*/                                                                                \
-    macro (ompt_event_parallel_begin,           ompt_new_parallel_callback_t,   1) /* parallel begin */         \
-    macro (ompt_event_parallel_end,             ompt_end_parallel_callback_t,   2) /* parallel end */           \
-                                                                                                                \
-    macro (ompt_event_task_begin,               ompt_new_task_callback_t,       3) /* task begin */             \
-    macro (ompt_event_task_end,                 ompt_task_callback_t,           4) /* task destroy */           \
-                                                                                                                \
-    macro (ompt_event_thread_begin,             ompt_thread_type_callback_t,    5) /* thread begin */           \
-    macro (ompt_event_thread_end,               ompt_thread_type_callback_t,    6) /* thread end */             \
-                                                                                                                \
-    macro (ompt_event_control,                  ompt_control_callback_t,        7) /* support control calls */  \
+    macro (ompt_callback_thread_begin,          ompt_callback_thread_begin_t,   1) /* thread begin */           \
+    macro (ompt_callback_thread_end,            ompt_callback_thread_end_t,     2) /* thread end */             \
                                                                                                                 \
-    macro (ompt_event_runtime_shutdown,         ompt_callback_t,                8) /* runtime shutdown */       \
+    macro (ompt_callback_parallel_begin,        ompt_callback_parallel_begin_t, 3) /* parallel begin */         \
+    macro (ompt_callback_parallel_end,          ompt_callback_parallel_end_t,   4) /* parallel end */           \
                                                                                                                 \
-    /*--- Optional Events (blame shifting, ompt_event_unimplemented) ---*/                                      \
-    macro (ompt_event_idle_begin,               ompt_thread_callback_t,         9) /* begin idle state */       \
-    macro (ompt_event_idle_end,                 ompt_thread_callback_t,        10) /* end idle state */         \
+    macro (ompt_callback_task_create,           ompt_callback_task_create_t,    5) /* task begin */             \
+    macro (ompt_callback_task_schedule,         ompt_callback_task_schedule_t,  6) /* task schedule */          \
+    macro (ompt_callback_implicit_task,         ompt_callback_implicit_task_t,  7) /* implicit task   */        \
                                                                                                                 \
-    macro (ompt_event_wait_barrier_begin,       ompt_parallel_callback_t,      11) /* begin wait at barrier */  \
-    macro (ompt_event_wait_barrier_end,         ompt_parallel_callback_t,      12) /* end wait at barrier */    \
+    macro (ompt_callback_target,                ompt_callback_target_t,         8) /* target */                 \
+    macro (ompt_callback_target_data_op,        ompt_callback_target_data_op_t, 9) /* target data op*/          \
+    macro (ompt_callback_target_submit,         ompt_callback_target_submit_t, 10) /* target  submit*/          \
                                                                                                                 \
-    macro (ompt_event_wait_taskwait_begin,      ompt_parallel_callback_t,      13) /* begin wait at taskwait */ \
-    macro (ompt_event_wait_taskwait_end,        ompt_parallel_callback_t,      14) /* end wait at taskwait */   \
+    macro (ompt_callback_control_tool,          ompt_callback_control_tool_t,  11) /* control tool */           \
                                                                                                                 \
-    macro (ompt_event_wait_taskgroup_begin,     ompt_parallel_callback_t,      15) /* begin wait at taskgroup */\
-    macro (ompt_event_wait_taskgroup_end,       ompt_parallel_callback_t,      16) /* end wait at taskgroup */  \
+    macro (ompt_callback_device_initialize,     ompt_callback_device_initialize_t, 12) /* device initialize */  \
+    macro (ompt_callback_device_finalize,       ompt_callback_device_finalize_t, 13)   /* device finalize   */  \
                                                                                                                 \
-    macro (ompt_event_release_lock,             ompt_wait_callback_t,          17) /* lock release */           \
-    macro (ompt_event_release_nest_lock_last,   ompt_wait_callback_t,          18) /* last nest lock release */ \
-    macro (ompt_event_release_critical,         ompt_wait_callback_t,          19) /* critical release */       \
+    /*--- Optional Events (blame shifting, ompt_event_unimplemented) ---*/                                      \
                                                                                                                 \
-    macro (ompt_event_release_atomic,           ompt_wait_callback_t,          20) /* atomic release */         \
+    macro (ompt_callback_sync_region_wait,      ompt_callback_sync_region_t,   14) /* sync region wait begin or end*/  \
                                                                                                                 \
-    macro (ompt_event_release_ordered,          ompt_wait_callback_t,          21) /* ordered release */        \
+    macro (ompt_callback_mutex_released,        ompt_callback_mutex_t,         15) /* mutex released */         \
                                                                                                                 \
     /*--- Optional Events (synchronous events, ompt_event_unimplemented) --- */                                 \
-    macro (ompt_event_implicit_task_begin,      ompt_parallel_callback_t,      22) /* implicit task begin   */  \
-    macro (ompt_event_implicit_task_end,        ompt_parallel_callback_t,      23) /* implicit task end  */     \
-                                                                                                                \
-    macro (ompt_event_initial_task_begin,       ompt_parallel_callback_t,      24) /* initial task begin   */   \
-    macro (ompt_event_initial_task_end,         ompt_parallel_callback_t,      25) /* initial task end  */      \
-                                                                                                                \
-    macro (ompt_event_task_switch,              ompt_task_pair_callback_t,     26) /* task switch */            \
-                                                                                                                \
-    macro (ompt_event_loop_begin,               ompt_new_workshare_callback_t, 27) /* task at loop begin */     \
-    macro (ompt_event_loop_end,                 ompt_parallel_callback_t,      28) /* task at loop end */       \
-                                                                                                                \
-    macro (ompt_event_sections_begin,           ompt_new_workshare_callback_t, 29) /* task at sections begin  */\
-    macro (ompt_event_sections_end,             ompt_parallel_callback_t,      30) /* task at sections end */   \
-                                                                                                                \
-    macro (ompt_event_single_in_block_begin,    ompt_new_workshare_callback_t, 31) /* task at single begin*/    \
-    macro (ompt_event_single_in_block_end,      ompt_parallel_callback_t,      32) /* task at single end */     \
                                                                                                                 \
-    macro (ompt_event_single_others_begin,      ompt_parallel_callback_t,      33) /* task at single begin */   \
-    macro (ompt_event_single_others_end,        ompt_parallel_callback_t,      34) /* task at single end */     \
+    macro (ompt_callback_task_dependences,      ompt_callback_task_dependences_t, 16) /* report task dependences  */\
+    macro (ompt_callback_task_dependence,       ompt_callback_task_dependence_t, 17) /* report task dependence  */\
                                                                                                                 \
-    macro (ompt_event_workshare_begin,          ompt_new_workshare_callback_t, 35) /* task at workshare begin */\
-    macro (ompt_event_workshare_end,            ompt_parallel_callback_t,      36) /* task at workshare end */  \
+    macro (ompt_callback_work,                  ompt_callback_work_t,          18) /* task at work begin or end*/\
                                                                                                                 \
-    macro (ompt_event_master_begin,             ompt_parallel_callback_t,      37) /* task at master begin */   \
-    macro (ompt_event_master_end,               ompt_parallel_callback_t,      38) /* task at master end */     \
+    macro (ompt_callback_master,                ompt_callback_master_t,        19) /* task at master begin or end */\
                                                                                                                 \
-    macro (ompt_event_barrier_begin,            ompt_parallel_callback_t,      39) /* task at barrier begin  */ \
-    macro (ompt_event_barrier_end,              ompt_parallel_callback_t,      40) /* task at barrier end */    \
+    macro (ompt_callback_target_map,            ompt_callback_target_map_t,    20) /* target map */             \
                                                                                                                 \
-    macro (ompt_event_taskwait_begin,           ompt_parallel_callback_t,      41) /* task at taskwait begin */ \
-    macro (ompt_event_taskwait_end,             ompt_parallel_callback_t,      42) /* task at task wait end */  \
+    macro (ompt_callback_sync_region,           ompt_callback_sync_region_t,   21) /* sync region begin or end */ \
                                                                                                                 \
-    macro (ompt_event_taskgroup_begin,          ompt_parallel_callback_t,      43) /* task at taskgroup begin */\
-    macro (ompt_event_taskgroup_end,            ompt_parallel_callback_t,      44) /* task at taskgroup end */  \
+    macro (ompt_callback_lock_init,             ompt_callback_mutex_acquire_t, 22) /* lock init */              \
+    macro (ompt_callback_lock_destroy,          ompt_callback_mutex_t,         23) /* lock destroy */           \
                                                                                                                 \
-    macro (ompt_event_release_nest_lock_prev,   ompt_wait_callback_t,          45) /* prev nest lock release */ \
+    macro (ompt_callback_mutex_acquire,         ompt_callback_mutex_acquire_t, 24) /* mutex acquire */          \
+    macro (ompt_callback_mutex_acquired,        ompt_callback_mutex_t,         25) /* mutex acquired */         \
                                                                                                                 \
-    macro (ompt_event_wait_lock,                ompt_wait_callback_t,          46) /* lock wait */              \
-    macro (ompt_event_wait_nest_lock,           ompt_wait_callback_t,          47) /* nest lock wait */         \
-    macro (ompt_event_wait_critical,            ompt_wait_callback_t,          48) /* critical wait */          \
-    macro (ompt_event_wait_atomic,              ompt_wait_callback_t,          49) /* atomic wait */            \
-    macro (ompt_event_wait_ordered,             ompt_wait_callback_t,          50) /* ordered wait */           \
+    macro (ompt_callback_nest_lock,             ompt_callback_nest_lock_t,     26) /* nest lock */              \
                                                                                                                 \
-    macro (ompt_event_acquired_lock,            ompt_wait_callback_t,          51) /* lock acquired */          \
-    macro (ompt_event_acquired_nest_lock_first, ompt_wait_callback_t,          52) /* 1st nest lock acquired */ \
-    macro (ompt_event_acquired_nest_lock_next,  ompt_wait_callback_t,          53) /* next nest lock acquired*/ \
-    macro (ompt_event_acquired_critical,        ompt_wait_callback_t,          54) /* critical acquired */      \
-    macro (ompt_event_acquired_atomic,          ompt_wait_callback_t,          55) /* atomic acquired */        \
-    macro (ompt_event_acquired_ordered,         ompt_wait_callback_t,          56) /* ordered acquired */       \
+    macro (ompt_callback_flush,                 ompt_callback_flush_t,         27) /* after executing flush */  \
                                                                                                                 \
-    macro (ompt_event_init_lock,                ompt_wait_callback_t,          57) /* lock init */              \
-    macro (ompt_event_init_nest_lock,           ompt_wait_callback_t,          58) /* nest lock init */         \
-                                                                                                                \
-    macro (ompt_event_destroy_lock,             ompt_wait_callback_t,          59) /* lock destruction */       \
-    macro (ompt_event_destroy_nest_lock,        ompt_wait_callback_t,          60) /* nest lock destruction */  \
-                                                                                                                \
-    macro (ompt_event_flush,                    ompt_callback_t,               61) /* after executing flush */  \
-                                                                                                                \
-    macro (ompt_event_task_dependences,         ompt_task_dependences_callback_t, 69) /* report task dependences  */\
-    macro (ompt_event_task_dependence_pair,     ompt_task_pair_callback_t,     70) /* report task dependence pair */
+    macro (ompt_callback_cancel,                ompt_callback_cancel_t,        28) /*cancel innermost binding region*/\
+    macro (ompt_callback_idle,                  ompt_callback_idle_t,          29) /* begin or end idle state */\
 
 
 
 /*****************************************************************************
  * data types
  *****************************************************************************/
 
 /*---------------------
  * identifiers
  *---------------------*/
 
-typedef uint64_t ompt_thread_id_t;
-#define ompt_thread_id_none ((ompt_thread_id_t) 0)     /* non-standard */
+typedef uint64_t ompt_id_t;
+#define ompt_id_none 0
 
-typedef uint64_t ompt_task_id_t;
-#define ompt_task_id_none ((ompt_task_id_t) 0)         /* non-standard */
+typedef union ompt_data_u {
+  uint64_t value; /* data initialized by runtime to unique id */
+  void *ptr;      /* pointer under tool control */
+} ompt_data_t;
 
-typedef uint64_t ompt_parallel_id_t;
-#define ompt_parallel_id_none ((ompt_parallel_id_t) 0) /* non-standard */
+static const ompt_data_t ompt_data_none = {0};
 
 typedef uint64_t ompt_wait_id_t;
-#define ompt_wait_id_none ((ompt_wait_id_t) 0)         /* non-standard */
+static const ompt_wait_id_t ompt_wait_id_none = 0;
 
+typedef void ompt_device_t;
 
 /*---------------------
  * ompt_frame_t
  *---------------------*/
 
 typedef struct ompt_frame_s {
     void *exit_runtime_frame;    /* next frame is user code     */
     void *reenter_runtime_frame; /* previous frame is user code */

  * enumerations for thread states and runtime events
  *****************************************************************************/
 
 /*---------------------
  * runtime states
  *---------------------*/
 
 typedef enum {
-#define ompt_state_macro(state, code) state = code,
-    FOREACH_OMPT_STATE(ompt_state_macro)
-#undef ompt_state_macro
-} ompt_state_t;
+#define omp_state_macro(state, code) state = code,
+    FOREACH_OMP_STATE(omp_state_macro)
+#undef omp_state_macro
+} omp_state_t;
 
 
 /*---------------------
  * runtime events
  *---------------------*/
 
-typedef enum {
+typedef enum ompt_callbacks_e{
 #define ompt_event_macro(event, callback, eventid) event = eventid,
     FOREACH_OMPT_EVENT(ompt_event_macro)
 #undef ompt_event_macro
-} ompt_event_t;
+} ompt_callbacks_t;
 
 
 /*---------------------
  * set callback results
  *---------------------*/
-typedef enum {
-    ompt_set_result_registration_error              = 0,
-    ompt_set_result_event_may_occur_no_callback     = 1,
-    ompt_set_result_event_never_occurs              = 2,
-    ompt_set_result_event_may_occur_callback_some   = 3,
-    ompt_set_result_event_may_occur_callback_always = 4,
+typedef enum ompt_set_result_e {
+    ompt_set_error = 0,
+    ompt_set_never = 1,
+    ompt_set_sometimes = 2,
+    ompt_set_sometimes_paired = 3,
+    ompt_set_always = 4
 } ompt_set_result_t;
 
 
+/*----------------------
+ * mutex implementations
+ *----------------------*/
+typedef enum ompt_mutex_impl_e {
+#define ompt_mutex_impl_macro(impl, code) impl = code,
+    FOREACH_OMPT_MUTEX_IMPL(ompt_mutex_impl_macro)
+#undef ompt_mutex_impl_macro
+} ompt_mutex_impl_t;
+
 
 /*****************************************************************************
  * callback signatures
  *****************************************************************************/
 
 /* initialization */
 typedef void (*ompt_interface_fn_t)(void);
 
 typedef ompt_interface_fn_t (*ompt_function_lookup_t)(
-    const char *                      /* entry point to look up       */
+    const char *                          /* entry point to look up              */
 );
 
 /* threads */
-typedef void (*ompt_thread_callback_t) (
-    ompt_thread_id_t thread_id        /* ID of thread                 */
-);
-
 typedef enum {
     ompt_thread_initial = 1, // start the enumeration at 1
     ompt_thread_worker  = 2,
     ompt_thread_other   = 3
 } ompt_thread_type_t;
 
 typedef enum {
-    ompt_invoker_program = 0,         /* program invokes master task  */
-    ompt_invoker_runtime = 1          /* runtime invokes master task  */
+    ompt_invoker_program = 1,             /* program invokes master task         */
+    ompt_invoker_runtime = 2              /* runtime invokes master task         */
 } ompt_invoker_t;
 
-typedef void (*ompt_thread_type_callback_t) (
-    ompt_thread_type_t thread_type,   /* type of thread               */
-    ompt_thread_id_t thread_id        /* ID of thread                 */
+typedef void (*ompt_callback_thread_begin_t) (
+    ompt_thread_type_t thread_type,       /* type of thread                      */
+    ompt_data_t *thread_data              /* data of thread                      */
+);
+
+typedef void (*ompt_callback_thread_end_t) (
+    ompt_data_t *thread_data              /* data of thread                      */
 );
 
 typedef void (*ompt_wait_callback_t) (
-    ompt_wait_id_t wait_id            /* wait id                      */
+    ompt_wait_id_t wait_id                /* wait data                           */
 );
 
 /* parallel and workshares */
-typedef void (*ompt_parallel_callback_t) (
-    ompt_parallel_id_t parallel_id,    /* id of parallel region       */
-    ompt_task_id_t task_id             /* id of task                  */
-);
-
-typedef void (*ompt_new_workshare_callback_t) (
-    ompt_parallel_id_t parallel_id,   /* id of parallel region        */
-    ompt_task_id_t parent_task_id,    /* id of parent task            */
-    void *workshare_function          /* pointer to outlined function */
-);
-
-typedef void (*ompt_new_parallel_callback_t) (
-    ompt_task_id_t parent_task_id,    /* id of parent task            */
-    ompt_frame_t *parent_task_frame,  /* frame data of parent task    */
-    ompt_parallel_id_t parallel_id,   /* id of parallel region        */
-    uint32_t requested_team_size,     /* number of threads in team    */
-    void *parallel_function,          /* pointer to outlined function */
-    ompt_invoker_t invoker            /* who invokes master task?     */
-);
-
-typedef void (*ompt_end_parallel_callback_t) (
-    ompt_parallel_id_t parallel_id,   /* id of parallel region       */
-    ompt_task_id_t task_id,           /* id of task                  */
-    ompt_invoker_t invoker            /* who invokes master task?    */
+typedef enum ompt_scope_endpoint_e {
+    ompt_scope_begin = 1,
+    ompt_scope_end = 2
+} ompt_scope_endpoint_t;
+
+
+/* implicit task */
+typedef void (*ompt_callback_implicit_task_t) (
+    ompt_scope_endpoint_t endpoint,       /* endpoint of implicit task           */
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    ompt_data_t *task_data,               /* data of implicit task               */
+    unsigned int team_size,               /* team size                           */
+    unsigned int thread_num               /* thread number of calling thread     */
+);
+
+typedef void (*ompt_callback_parallel_begin_t) (
+    ompt_data_t *parent_task_data,        /* data of parent task                 */
+    const ompt_frame_t *parent_frame,     /* frame data of parent task           */
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    unsigned int requested_team_size,     /* requested number of threads in team */
+    ompt_invoker_t invoker,               /* invoker of master task              */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_parallel_end_t) (
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    ompt_data_t *task_data,               /* data of task                        */
+    ompt_invoker_t invoker,               /* invoker of master task              */ 
+    const void *codeptr_ra                /* return address of runtime call      */
 );
 
 /* tasks */
-typedef void (*ompt_task_callback_t) (
-    ompt_task_id_t task_id            /* id of task                   */
+typedef enum ompt_task_type_e {
+    ompt_task_initial    = 0x1,
+    ompt_task_implicit   = 0x2,
+    ompt_task_explicit   = 0x4,
+    ompt_task_target     = 0x8,
+    ompt_task_undeferred = 0x8000000,
+    ompt_task_untied     = 0x10000000,
+    ompt_task_final      = 0x20000000,
+    ompt_task_mergeable  = 0x40000000,
+    ompt_task_merged     = 0x80000000
+} ompt_task_type_t;
+
+typedef enum ompt_task_status_e {
+    ompt_task_complete = 1,
+    ompt_task_yield    = 2,
+    ompt_task_cancel   = 3,
+    ompt_task_others   = 4
+} ompt_task_status_t;
+
+typedef void (*ompt_callback_task_schedule_t) (
+    ompt_data_t *prior_task_data,         /* data of prior task                  */
+    ompt_task_status_t prior_task_status, /* status of prior task                */
+    ompt_data_t *next_task_data           /* data of next task                   */
+);
+
+typedef void (*ompt_callback_task_create_t) (
+    ompt_data_t *parent_task_data,        /* data of parent task                 */
+    const ompt_frame_t *parent_frame,     /* frame data for parent task          */
+    ompt_data_t *new_task_data,           /* data of created task                */
+    int type,                             /* type of created task                */
+    int has_dependences,                  /* created task has dependences        */
+    const void *codeptr_ra                /* return address of runtime call      */
 );
 
-typedef void (*ompt_task_pair_callback_t) (
-    ompt_task_id_t first_task_id,
-    ompt_task_id_t second_task_id
-);
-
-typedef void (*ompt_new_task_callback_t) (
-    ompt_task_id_t parent_task_id,    /* id of parent task            */
-    ompt_frame_t *parent_task_frame,  /* frame data for parent task   */
-    ompt_task_id_t  new_task_id,      /* id of created task           */
-    void *task_function               /* pointer to outlined function */
+/* task dependences */
+typedef void (*ompt_callback_task_dependences_t) (
+    ompt_data_t *task_data,               /* data of task                        */
+    const ompt_task_dependence_t *deps,   /* dependences of task                 */
+    int ndeps                             /* dependences count of task           */
+);
+
+typedef void (*ompt_callback_task_dependence_t) (
+    ompt_data_t *src_task_data,           /* data of source task                 */
+    ompt_data_t *sink_task_data           /* data of sink task                   */
+);
+
+/* target and device */
+typedef enum ompt_target_type_e {
+    ompt_target = 1,
+    ompt_target_enter_data = 2,
+    ompt_target_exit_data = 3,
+    ompt_target_update = 4
+} ompt_target_type_t;
+
+typedef void (*ompt_callback_target_t) (
+    ompt_target_type_t kind,
+    ompt_scope_endpoint_t endpoint,
+    uint64_t device_num,
+    ompt_data_t *task_data,
+    ompt_id_t target_id,
+    const void *codeptr_ra
+);
+
+typedef enum ompt_target_data_op_e {
+    ompt_target_data_alloc = 1,
+    ompt_target_data_transfer_to_dev = 2,
+    ompt_target_data_transfer_from_dev = 3,
+    ompt_target_data_delete = 4
+} ompt_target_data_op_t;
+
+typedef void (*ompt_callback_target_data_op_t) (
+    ompt_id_t target_id,
+    ompt_id_t host_op_id,
+    ompt_target_data_op_t optype,
+    void *host_addr,
+    void *device_addr,
+    size_t bytes
+);
+
+typedef void (*ompt_callback_target_submit_t) (
+    ompt_id_t target_id,
+    ompt_id_t host_op_id
+);
+
+typedef void (*ompt_callback_target_map_t) (
+    ompt_id_t target_id,
+    unsigned int nitems,
+    void **host_addr,
+    void **device_addr,
+    size_t *bytes,
+    unsigned int *mapping_flags
+);
+
+typedef void (*ompt_callback_device_initialize_t) (
+    uint64_t device_num,
+    const char *type,
+    ompt_device_t *device,
+    ompt_function_lookup_t lookup,
+    const char *documentation
 );
 
-/* task dependences */
-typedef void (*ompt_task_dependences_callback_t) (
-    ompt_task_id_t task_id,            /* ID of task with dependences */
-    const ompt_task_dependence_t *deps,/* vector of task dependences  */
-    int ndeps                          /* number of dependences       */
+typedef void (*ompt_callback_device_finalize_t) (
+    uint64_t device_num
 );
 
-/* program */
-typedef void (*ompt_control_callback_t) (
-    uint64_t command,                 /* command of control call      */
-    uint64_t modifier                 /* modifier of control call     */
+/* control_tool */
+typedef int (*ompt_callback_control_tool_t) (
+    uint64_t command,                     /* command of control call             */
+    uint64_t modifier,                    /* modifier of control call            */
+    void *arg,                            /* argument of control call            */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef enum ompt_mutex_kind_e {
+    ompt_mutex = 0x10,
+    ompt_mutex_lock = 0x11,
+    ompt_mutex_nest_lock = 0x12,
+    ompt_mutex_critical = 0x13,
+    ompt_mutex_atomic = 0x14,
+    ompt_mutex_ordered = 0x20
+} ompt_mutex_kind_t;
+
+typedef void (*ompt_callback_mutex_acquire_t) (
+    ompt_mutex_kind_t kind,               /* mutex kind                          */
+    unsigned int hint,                    /* mutex hint                          */
+    unsigned int impl,                    /* mutex implementation                */
+    ompt_wait_id_t wait_id,               /* id of object being awaited          */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_mutex_t) (
+    ompt_mutex_kind_t kind,               /* mutex kind                          */
+    ompt_wait_id_t wait_id,               /* id of object being awaited          */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_nest_lock_t) (
+    ompt_scope_endpoint_t endpoint,       /* endpoint of nested lock             */
+    ompt_wait_id_t wait_id,               /* id of object being awaited          */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_master_t) (
+    ompt_scope_endpoint_t endpoint,       /* endpoint of master region           */
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    ompt_data_t *task_data,               /* data of task                        */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_idle_t) (
+    ompt_scope_endpoint_t endpoint        /* endpoint of idle time               */
+);
+
+typedef enum ompt_work_type_e {
+    ompt_work_loop = 1,
+    ompt_work_sections = 2,
+    ompt_work_single_executor = 3,
+    ompt_work_single_other = 4,
+    ompt_work_workshare = 5,
+    ompt_work_distribute = 6,
+    ompt_work_taskloop = 7
+} ompt_work_type_t;
+
+typedef void (*ompt_callback_work_t) (
+    ompt_work_type_t wstype,              /* type of work region                 */
+    ompt_scope_endpoint_t endpoint,       /* endpoint of work region             */
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    ompt_data_t *task_data,               /* data of task                        */
+    uint64_t count,                       /* quantity of work                    */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef enum ompt_sync_region_kind_e {
+    ompt_sync_region_barrier = 1,
+    ompt_sync_region_taskwait = 2,
+    ompt_sync_region_taskgroup = 3
+} ompt_sync_region_kind_t;
+
+typedef void (*ompt_callback_sync_region_t) (
+    ompt_sync_region_kind_t kind,         /* kind of sync region                 */
+    ompt_scope_endpoint_t endpoint,       /* endpoint of sync region             */
+    ompt_data_t *parallel_data,           /* data of parallel region             */
+    ompt_data_t *task_data,               /* data of task                        */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef enum ompt_cancel_flag_e {
+    ompt_cancel_parallel       = 0x1,
+    ompt_cancel_sections       = 0x2,
+    ompt_cancel_do             = 0x4,
+    ompt_cancel_taskgroup      = 0x8,
+    ompt_cancel_activated      = 0x10,
+    ompt_cancel_detected       = 0x20,
+    ompt_cancel_discarded_task = 0x40
+} ompt_cancel_flag_t;
+
+typedef void (*ompt_callback_cancel_t) (
+    ompt_data_t *task_data,               /* data of task                        */
+    int flags,                            /* cancel flags                        */
+    const void *codeptr_ra                /* return address of runtime call      */
+);
+
+typedef void (*ompt_callback_flush_t) (
+    ompt_data_t *thread_data,             /* data of thread                      */
+    const void *codeptr_ra                /* return address of runtime call      */
 );
 
-typedef void (*ompt_callback_t)(void);
-
-
 /****************************************************************************
  * ompt API
  ***************************************************************************/
 
 #ifdef  __cplusplus
 extern "C" {
 #endif
 
 #define OMPT_API_FNTYPE(fn) fn##_t
 
 #define OMPT_API_FUNCTION(return_type, fn, args)  \
     typedef return_type (*OMPT_API_FNTYPE(fn)) args
 
 
 
 /****************************************************************************
  * INQUIRY FUNCTIONS
  ***************************************************************************/
 
 /* state */
-OMPT_API_FUNCTION(ompt_state_t, ompt_get_state, (
-    ompt_wait_id_t *ompt_wait_id
+OMPT_API_FUNCTION(omp_state_t, ompt_get_state, (
+    ompt_wait_id_t *wait_id
 ));
 
 /* thread */
-OMPT_API_FUNCTION(ompt_thread_id_t, ompt_get_thread_id, (void));
-
-OMPT_API_FUNCTION(void *, ompt_get_idle_frame, (void));
+OMPT_API_FUNCTION(ompt_data_t*, ompt_get_thread_data, (void));
 
 /* parallel region */
-OMPT_API_FUNCTION(ompt_parallel_id_t, ompt_get_parallel_id, (
-    int ancestor_level
+OMPT_API_FUNCTION(int, ompt_get_parallel_info, (
+    int ancestor_level,
+    ompt_data_t **parallel_data,
+    int *team_size
 ));
 
-OMPT_API_FUNCTION(int, ompt_get_parallel_team_size, (
-    int ancestor_level
+/* task */
+OMPT_API_FUNCTION(int, ompt_get_task_info, (
+    int ancestor_level,
+    int *type,
+    ompt_data_t **task_data,
+    ompt_frame_t **task_frame,
+    ompt_data_t **parallel_data,
+    int *thread_num
 ));
 
-/* task */
-OMPT_API_FUNCTION(ompt_task_id_t, ompt_get_task_id, (
-    int depth
+/* places */
+OMPT_API_FUNCTION(int, ompt_get_num_places, (void));
+
+OMPT_API_FUNCTION(int, ompt_get_place_proc_ids, (
+    int place_num,
+    int ids_size,
+    int *ids
 ));
 
-OMPT_API_FUNCTION(ompt_frame_t *, ompt_get_task_frame, (
-    int depth
+OMPT_API_FUNCTION(int, ompt_get_place_num, (void));
+
+OMPT_API_FUNCTION(int, ompt_get_partition_place_nums, (
+    int place_nums_size,
+    int *place_nums
 ));
 
+/* proc_id */
+OMPT_API_FUNCTION(int, ompt_get_proc_id, (void));
 
 
 /****************************************************************************
  * PLACEHOLDERS FOR PERFORMANCE REPORTING
  ***************************************************************************/
 
 /* idle */
 OMPT_API_FUNCTION(void, ompt_idle, (

 ));
 
 
 
 /****************************************************************************
  * INITIALIZATION FUNCTIONS
  ***************************************************************************/
 
-OMPT_API_FUNCTION(void, ompt_initialize, (
+typedef struct ompt_fns_t ompt_fns_t;
+
+OMPT_API_FUNCTION(int, ompt_initialize, (
     ompt_function_lookup_t ompt_fn_lookup,
-    const char *runtime_version,
-    unsigned int ompt_version
+    ompt_fns_t *fns
 ));
 
+OMPT_API_FUNCTION(void, ompt_finalize, (
+    ompt_fns_t *fns
+));
+
+struct ompt_fns_t {
+    ompt_initialize_t initialize;
+    ompt_finalize_t finalize;
+};
 
 /* initialization interface to be defined by tool */
-ompt_initialize_t ompt_tool(void);
+#ifdef _WIN32
+__declspec(dllexport)
+#endif
+ompt_fns_t * ompt_start_tool(
+    unsigned int omp_version, 
+    const char * runtime_version
+);
 
-typedef enum opt_init_mode_e {
-    ompt_init_mode_never  = 0,
-    ompt_init_mode_false  = 1,
-    ompt_init_mode_true   = 2,
-    ompt_init_mode_always = 3
-} ompt_init_mode_t;
+typedef void (*ompt_callback_t)(void);
 
 OMPT_API_FUNCTION(int, ompt_set_callback, (
-    ompt_event_t event,
+    ompt_callbacks_t which,
     ompt_callback_t callback
 ));
 
 typedef enum ompt_set_callback_rc_e {  /* non-standard */
     ompt_set_callback_error      = 0,
     ompt_has_event_no_callback   = 1,
     ompt_no_event_no_callback    = 2,
     ompt_has_event_may_callback  = 3,
     ompt_has_event_must_callback = 4,
 } ompt_set_callback_rc_t;
 
 
 OMPT_API_FUNCTION(int, ompt_get_callback, (
-    ompt_event_t event,
+    ompt_callbacks_t which,
     ompt_callback_t *callback
 ));
 
 
 
 /****************************************************************************
  * MISCELLANEOUS FUNCTIONS
  ***************************************************************************/
 
-/* control */
-// FIXME: remove workaround for clang
-#if !defined(__clang__) && defined(_OPENMP) && (_OPENMP >= 201307)
-#pragma omp declare target
-#endif
-void ompt_control(
-    uint64_t command,
-    uint64_t modifier
-);
-#if !defined(__clang__) && defined(_OPENMP) && (_OPENMP >= 201307)
-#pragma omp end declare target
-#endif
-
 /* state enumeration */
-OMPT_API_FUNCTION(int, ompt_enumerate_state, (
+OMPT_API_FUNCTION(int, ompt_enumerate_states, (
     int current_state,
     int *next_state,
     const char **next_state_name
 ));
 
+/* mutex implementation enumeration */
+OMPT_API_FUNCTION(int, ompt_enumerate_mutex_impls, (
+    int current_impl,
+    int *next_impl,
+    const char **next_impl_name
+));
+
+/* get_unique_id */
+OMPT_API_FUNCTION(uint64_t, ompt_get_unique_id, (void));
+
 #ifdef  __cplusplus
 };
 #endif
 
-#endif
+/****************************************************************************
+ * TARGET
+ ***************************************************************************/
+
+ OMPT_API_FUNCTION(int, ompt_get_target_info, (
+    uint64_t *device_num,
+    ompt_id_t *target_id,
+    ompt_id_t *host_op_id
+));
+
+ OMPT_API_FUNCTION(int, ompt_get_num_devices, (void));
 
+#endif /* __OMPT__ */

runtime/src/kmp.h

 #define KMP_IDENT_BARRIER_IMPL 0x0040
 #define KMP_IDENT_BARRIER_IMPL_MASK 0x01C0
 #define KMP_IDENT_BARRIER_IMPL_FOR 0x0040
 #define KMP_IDENT_BARRIER_IMPL_SECTIONS 0x00C0
 
 #define KMP_IDENT_BARRIER_IMPL_SINGLE 0x0140
 #define KMP_IDENT_BARRIER_IMPL_WORKSHARE 0x01C0
 
+#define KMP_IDENT_WORK_LOOP 0x200 // static loop
+#define KMP_IDENT_WORK_SECTIONS 0x400 // sections
+#define KMP_IDENT_WORK_DISTRIBUTE 0x800 // distribute
+
 /*!
  * The ident structure that describes a source location.
  */
 typedef struct ident {
   kmp_int32 reserved_1; /**<  might be used in Fortran; see above  */
   kmp_int32 flags; /**<  also f.flags; KMP_IDENT_xxx flags; KMP_IDENT_KMPC
                       identifies this union member  */
   kmp_int32 reserved_2; /**<  not really used in Fortran any more; see above */

 extern kmp_hws_item_t __kmp_hws_socket;
 extern kmp_hws_item_t __kmp_hws_node;
 extern kmp_hws_item_t __kmp_hws_tile;
 extern kmp_hws_item_t __kmp_hws_core;
 extern kmp_hws_item_t __kmp_hws_proc;
 extern int __kmp_hws_requested;
 extern int __kmp_hws_abs_flag; // absolute or per-item number requested
 
+#if OMP_50_ENABLED && LIBOMP_OMPT_SUPPORT
+extern char const *__kmp_tool_libraries;
+#endif // OMP_50_ENABLED && LIBOMP_OMPT_SUPPORT
+
 /* ------------------------------------------------------------------------ */
 
 #define KMP_PAD(type, sz)                                                      \
   (sizeof(type) + (sz - ((sizeof(type) - 1) % (sz)) - 1))
 
 // We need to avoid using -1 as a GTID as +1 is added to the gtid
 // when storing it in a lock, and the value 0 is reserved.
 #define KMP_GTID_DNE (-2) /* Does not exist */

 extern void __kmp_suspend_uninitialize_thread(kmp_info_t *th);
 
 extern kmp_info_t *__kmp_allocate_thread(kmp_root_t *root, kmp_team_t *team,
                                          int tid);
 #if OMP_40_ENABLED
 extern kmp_team_t *
 __kmp_allocate_team(kmp_root_t *root, int new_nproc, int max_nproc,
 #if OMPT_SUPPORT
-                    ompt_parallel_id_t ompt_parallel_id,
+                    ompt_data_t ompt_parallel_data,
 #endif
                     kmp_proc_bind_t proc_bind, kmp_internal_control_t *new_icvs,
                     int argc USE_NESTED_HOT_ARG(kmp_info_t *thr));
 #else
 extern kmp_team_t *
 __kmp_allocate_team(kmp_root_t *root, int new_nproc, int max_nproc,
 #if OMPT_SUPPORT
-                    ompt_parallel_id_t ompt_parallel_id,
+                    ompt_id_t ompt_parallel_id,
 #endif
                     kmp_internal_control_t *new_icvs,
                     int argc USE_NESTED_HOT_ARG(kmp_info_t *thr));
 #endif // OMP_40_ENABLED
 extern void __kmp_free_thread(kmp_info_t *);
 extern void __kmp_free_team(kmp_root_t *,
                             kmp_team_t *USE_NESTED_HOT_ARG(kmp_info_t *));
 extern kmp_team_t *__kmp_reap_team(kmp_team_t *);

 enum fork_context_e {
   fork_context_gnu, /**< Called from GNU generated code, so must not invoke the
                        microtask internally. */
   fork_context_intel, /**< Called from Intel generated code.  */
   fork_context_last
 };
 extern int __kmp_fork_call(ident_t *loc, int gtid,
                            enum fork_context_e fork_context, kmp_int32 argc,
-#if OMPT_SUPPORT
-                           void *unwrapped_task,
-#endif
                            microtask_t microtask, launch_t invoker,
 /* TODO: revert workaround for Intel(R) 64 tracker #96 */
 #if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && KMP_OS_LINUX
                            va_list *ap
 #else
                            va_list ap
 #endif
                            );

runtime/src/kmp_atomic.h

 
 extern int __kmp_atomic_mode;
 
 // Atomic locks can easily become contended, so we use queuing locks for them.
 typedef kmp_queuing_lock_t kmp_atomic_lock_t;
 
 static inline void __kmp_acquire_atomic_lock(kmp_atomic_lock_t *lck,
                                              kmp_int32 gtid) {
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_wait_atomic)) {
-    ompt_callbacks.ompt_callback(ompt_event_wait_atomic)((ompt_wait_id_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_atomic, 0, ompt_mutex_impl_queuing, (ompt_wait_id_t)lck,
+        OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 
   __kmp_acquire_queuing_lock(lck, gtid);
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_acquired_atomic)) {
-    ompt_callbacks.ompt_callback(ompt_event_acquired_atomic)(
-        (ompt_wait_id_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_mutex_acquired) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+        ompt_mutex_atomic, (ompt_wait_id_t)lck, OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 }
 
 static inline int __kmp_test_atomic_lock(kmp_atomic_lock_t *lck,
                                          kmp_int32 gtid) {
   return __kmp_test_queuing_lock(lck, gtid);
 }
 
 static inline void __kmp_release_atomic_lock(kmp_atomic_lock_t *lck,
                                              kmp_int32 gtid) {
   __kmp_release_queuing_lock(lck, gtid);
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_release_atomic)) {
-    ompt_callbacks.ompt_callback(ompt_event_release_atomic)(
-        (ompt_wait_id_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_mutex_released) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+        ompt_mutex_atomic, (ompt_wait_id_t)lck, OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 }
 
 static inline void __kmp_init_atomic_lock(kmp_atomic_lock_t *lck) {
   __kmp_init_queuing_lock(lck);
 }
 

runtime/src/kmp_barrier.cpp

 //
 //===----------------------------------------------------------------------===//
 
 #include "kmp.h"
 #include "kmp_wait_release.h"
 #include "kmp_itt.h"
 #include "kmp_os.h"
 #include "kmp_stats.h"
+#if OMPT_SUPPORT
+#include "ompt-specific.h"
+#endif
 
 #if KMP_MIC
 #include <immintrin.h>
 #define USE_NGO_STORES 1
 #endif // KMP_MIC
 
 #include "tsan_annotations.h"
 
int __kmp_barrier(enum barrier_type bt, int gtid, int is_split,
   KMP_TIME_PARTITIONED_BLOCK(OMP_plain_barrier);
   KMP_SET_THREAD_STATE_BLOCK(PLAIN_BARRIER);
   int tid = __kmp_tid_from_gtid(gtid);
   kmp_info_t *this_thr = __kmp_threads[gtid];
   kmp_team_t *team = this_thr->th.th_team;
   int status = 0;
   ident_t *loc = __kmp_threads[gtid]->th.th_ident;
 #if OMPT_SUPPORT
-  ompt_task_id_t my_task_id;
-  ompt_parallel_id_t my_parallel_id;
+  ompt_data_t *my_task_data;
+  ompt_data_t *my_parallel_data;
+  void *return_address;
 #endif
 
   KA_TRACE(15, ("__kmp_barrier: T#%d(%d:%d) has arrived\n", gtid,
                 __kmp_team_from_gtid(gtid)->t.t_id, __kmp_tid_from_gtid(gtid)));
 
   ANNOTATE_BARRIER_BEGIN(&team->t.t_bar);
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-#if OMPT_BLAME
-    my_task_id = team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id;
-    my_parallel_id = team->t.ompt_team_info.parallel_id;
-
-#if OMPT_TRACE
-    if (this_thr->th.ompt_thread_info.state == ompt_state_wait_single) {
-      if (ompt_callbacks.ompt_callback(ompt_event_single_others_end)) {
-        ompt_callbacks.ompt_callback(ompt_event_single_others_end)(
-            my_parallel_id, my_task_id);
-      }
-    }
-#endif
-    if (ompt_callbacks.ompt_callback(ompt_event_barrier_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_barrier_begin)(my_parallel_id,
-                                                             my_task_id);
+  if (ompt_enabled.enabled) {
+#if OMPT_OPTIONAL
+    my_task_data = OMPT_CUR_TASK_DATA(this_thr);
+    my_parallel_data = OMPT_CUR_TEAM_DATA(this_thr);
+    return_address = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (ompt_enabled.ompt_callback_sync_region) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+          ompt_sync_region_barrier, ompt_scope_begin, my_parallel_data,
+          my_task_data, return_address);
+    }
+    if (ompt_enabled.ompt_callback_sync_region_wait) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_barrier, ompt_scope_begin, my_parallel_data,
+          my_task_data, return_address);
     }
 #endif
     // It is OK to report the barrier state after the barrier begin callback.
     // According to the OMPT specification, a compliant implementation may
     // even delay reporting this state until the barrier begins to wait.
-    this_thr->th.ompt_thread_info.state = ompt_state_wait_barrier;
+    this_thr->th.ompt_thread_info.state = omp_state_wait_barrier;
   }
 #endif
 
   if (!team->t.t_serialized) {
 #if USE_ITT_BUILD
     // This value will be used in itt notify events below.
     void *itt_sync_obj = NULL;
 #if USE_ITT_NOTIFY

 #endif
     }
   }
   KA_TRACE(15, ("__kmp_barrier: T#%d(%d:%d) is leaving with return value %d\n",
                 gtid, __kmp_team_from_gtid(gtid)->t.t_id,
                 __kmp_tid_from_gtid(gtid), status));
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-#if OMPT_BLAME
-    if (ompt_callbacks.ompt_callback(ompt_event_barrier_end)) {
-      ompt_callbacks.ompt_callback(ompt_event_barrier_end)(my_parallel_id,
-                                                           my_task_id);
+  if (ompt_enabled.enabled) {
+#if OMPT_OPTIONAL
+    if (ompt_enabled.ompt_callback_sync_region_wait) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_barrier, ompt_scope_end, my_parallel_data,
+          my_task_data, return_address);
+    }
+    if (ompt_enabled.ompt_callback_sync_region) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+          ompt_sync_region_barrier, ompt_scope_end, my_parallel_data,
+          my_task_data, return_address);
     }
 #endif
-    this_thr->th.ompt_thread_info.state = ompt_state_work_parallel;
+    this_thr->th.ompt_thread_info.state = omp_state_work_parallel;
   }
 #endif
   ANNOTATE_BARRIER_END(&team->t.t_bar);
 
   return status;
 }
 
 void __kmp_end_split_barrier(enum barrier_type bt, int gtid) {
#endif /* KMP_DEBUG */
   KMP_DEBUG_ASSERT(TCR_PTR(this_thr->th.th_team));
   KMP_DEBUG_ASSERT(TCR_PTR(this_thr->th.th_root));
   KMP_DEBUG_ASSERT(this_thr == team->t.t_threads[tid]);
   KA_TRACE(10, ("__kmp_join_barrier: T#%d(%d:%d) arrived at join barrier\n",
                 gtid, team_id, tid));
 
   ANNOTATE_BARRIER_BEGIN(&team->t.t_bar);
 #if OMPT_SUPPORT
-#if OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_barrier_begin)) {
-    ompt_callbacks.ompt_callback(ompt_event_barrier_begin)(
-        team->t.ompt_team_info.parallel_id,
-        team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
+  ompt_data_t *my_task_data;
+  ompt_data_t *my_parallel_data;
+  if (ompt_enabled.enabled) {
+#if OMPT_OPTIONAL
+    void *codeptr = NULL;
+    int ds_tid = this_thr->th.th_info.ds.ds_tid;
+    if (KMP_MASTER_TID(ds_tid) &&
+        (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) ||
+         ompt_callbacks.ompt_callback(ompt_callback_sync_region)))
+      codeptr = team->t.ompt_team_info.master_return_address;
+    my_task_data = OMPT_CUR_TASK_DATA(this_thr);
+    my_parallel_data = OMPT_CUR_TEAM_DATA(this_thr);
+    if (ompt_enabled.ompt_callback_sync_region) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+          ompt_sync_region_barrier, ompt_scope_begin, my_parallel_data,
+          my_task_data, codeptr);
+    }
+    if (ompt_enabled.ompt_callback_sync_region_wait) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_barrier, ompt_scope_begin, my_parallel_data,
+          my_task_data, codeptr);
     }
 #endif
-  this_thr->th.ompt_thread_info.state = ompt_state_wait_barrier;
+    this_thr->th.ompt_thread_info.state = omp_state_wait_barrier_implicit;
+  }
 #endif
 
   if (__kmp_tasking_mode == tskm_extra_barrier) {
     __kmp_tasking_barrier(team, this_thr, gtid);
     KA_TRACE(10, ("__kmp_join_barrier: T#%d(%d:%d) past taking barrier\n", gtid,
                   team_id, tid));
   }
 #ifdef KMP_DEBUG
#if KMP_DEBUG
   }
 #endif /* KMP_DEBUG */
 
   // TODO now, mark worker threads as done so they may be disbanded
   KMP_MB(); // Flush all pending memory write invalidates.
   KA_TRACE(10,
            ("__kmp_join_barrier: T#%d(%d:%d) leaving\n", gtid, team_id, tid));
 
-#if OMPT_SUPPORT
-  if (ompt_enabled) {
-#if OMPT_BLAME
-    if (ompt_callbacks.ompt_callback(ompt_event_barrier_end)) {
-      ompt_callbacks.ompt_callback(ompt_event_barrier_end)(
-          team->t.ompt_team_info.parallel_id,
-          team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
-    }
-#endif
-
-    // return to default state
-    this_thr->th.ompt_thread_info.state = ompt_state_overhead;
-  }
-#endif
   ANNOTATE_BARRIER_END(&team->t.t_bar);
 }
 
 // TODO release worker threads' fork barriers as we are ready instead of all at
 // once
 void __kmp_fork_barrier(int gtid, int tid) {
   KMP_TIME_PARTITIONED_BLOCK(OMP_fork_barrier);
   KMP_SET_THREAD_STATE_BLOCK(FORK_JOIN_BARRIER);
case bp_tree_bar: {
     break;
   }
   default: {
     __kmp_linear_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid,
                                  TRUE USE_ITT_BUILD_ARG(itt_sync_obj));
   }
   }
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    if (this_thr->th.ompt_thread_info.state ==
+        omp_state_wait_barrier_implicit) {
+      int ds_tid = this_thr->th.th_info.ds.ds_tid;
+      ompt_data_t *tId = (team) ? OMPT_CUR_TASK_DATA(this_thr)
+                                : &(this_thr->th.ompt_thread_info.task_data);
+      this_thr->th.ompt_thread_info.state = omp_state_overhead;
+#if OMPT_OPTIONAL
+      void *codeptr = NULL;
+      if (KMP_MASTER_TID(ds_tid) &&
+          (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) ||
+           ompt_callbacks.ompt_callback(ompt_callback_sync_region)))
+        codeptr = team->t.ompt_team_info.master_return_address;
+      if (ompt_enabled.ompt_callback_sync_region_wait) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+            ompt_sync_region_barrier, ompt_scope_end, NULL, tId, codeptr);
+      }
+      if (ompt_enabled.ompt_callback_sync_region) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+            ompt_sync_region_barrier, ompt_scope_end, NULL, tId, codeptr);
+      }
+#endif
+      if (!KMP_MASTER_TID(ds_tid) && ompt_enabled.ompt_callback_implicit_task) {
+        ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+            ompt_scope_end, NULL, tId, 0, ds_tid);
+      }
+      // return to idle state
+      this_thr->th.ompt_thread_info.state = omp_state_overhead;
+    }
+  }
+#endif
+
   // Early exit for reaping threads releasing forkjoin barrier
   if (TCR_4(__kmp_global.g.g_done)) {
     this_thr->th.th_task_team = NULL;
 
 #if USE_ITT_BUILD && USE_ITT_NOTIFY
     if (__itt_sync_create_ptr || KMP_ITT_DEBUG) {
       if (!KMP_MASTER_TID(tid)) {
         itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier);

runtime/src/kmp_cancel.cpp

 
 //===----------------------------------------------------------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is dual licensed under the MIT and the University of Illinois Open
 // Source Licenses. See LICENSE.txt for details.
 //
 //===----------------------------------------------------------------------===//
 
 #include "kmp.h"
 #include "kmp_i18n.h"
 #include "kmp_io.h"
 #include "kmp_str.h"
+#if OMPT_SUPPORT
+#include "ompt-specific.h"
+#endif
 
 #if OMP_40_ENABLED
 
 /*!
 @ingroup CANCELLATION
 @param loc_ref location of the original task directive
 @param gtid Global thread ID of encountering thread
 @param cncl_kind Cancellation kind (parallel, for, sections, taskgroup)
case cancel_sections:
       // cancellation requests for parallel and worksharing constructs
       // are handled through the team structure
       {
         kmp_team_t *this_team = this_thr->th.th_team;
         KMP_DEBUG_ASSERT(this_team);
         kmp_int32 old = KMP_COMPARE_AND_STORE_RET32(
             &(this_team->t.t_cancel_request), cancel_noreq, cncl_kind);
         if (old == cancel_noreq || old == cncl_kind) {
-          // printf("__kmpc_cancel: this_team->t.t_cancel_request=%d @ %p\n",
-          //       this_team->t.t_cancel_request,
-          //       &(this_team->t.t_cancel_request));
-          // we do not have a cancellation request in this team or we do have
-          // one that matches the current request -> cancel
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+          if (ompt_enabled.ompt_callback_cancel) {
+            ompt_data_t *task_data;
+            __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL,
+                                          NULL);
+            ompt_cancel_flag_t type = ompt_cancel_parallel;
+            if (cncl_kind == cancel_parallel)
+              type = ompt_cancel_parallel;
+            else if (cncl_kind == cancel_loop)
+              type = ompt_cancel_do;
+            else if (cncl_kind == cancel_sections)
+              type = ompt_cancel_sections;
+            ompt_callbacks.ompt_callback(ompt_callback_cancel)(
+                task_data, type | ompt_cancel_activated,
+                OMPT_GET_RETURN_ADDRESS(0));
+          }
+#endif
           return 1 /* true */;
         }
         break;
       }
     case cancel_taskgroup:
       // cancellation requests for a task group
       // are handled through the taskgroup structure
       {
         kmp_taskdata_t *task;
         kmp_taskgroup_t *taskgroup;
 
         task = this_thr->th.th_current_task;
         KMP_DEBUG_ASSERT(task);
 
         taskgroup = task->td_taskgroup;
         if (taskgroup) {
           kmp_int32 old = KMP_COMPARE_AND_STORE_RET32(
               &(taskgroup->cancel_request), cancel_noreq, cncl_kind);
           if (old == cancel_noreq || old == cncl_kind) {
-            // we do not have a cancellation request in this taskgroup or we do
-            // have one that matches the current request -> cancel
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+            if (ompt_enabled.ompt_callback_cancel) {
+              ompt_data_t *task_data;
+              __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL,
+                                            NULL);
+              ompt_callbacks.ompt_callback(ompt_callback_cancel)(
+                  task_data, ompt_cancel_taskgroup | ompt_cancel_activated,
+                  OMPT_GET_RETURN_ADDRESS(0));
+            }
+#endif
             return 1 /* true */;
           }
         } else {
           // TODO: what needs to happen here?
           // the specification disallows cancellation w/o taskgroups
           // so we might do anything here, let's abort for now
           KMP_ASSERT(0 /* false */);
         }
if (__kmp_omp_cancellation) {
     case cancel_sections:
       // cancellation requests for parallel and worksharing constructs
       // are handled through the team structure
       {
         kmp_team_t *this_team = this_thr->th.th_team;
         KMP_DEBUG_ASSERT(this_team);
         if (this_team->t.t_cancel_request) {
           if (cncl_kind == this_team->t.t_cancel_request) {
-            // the request in the team structure matches the type of
-            // cancellation point so we can cancel
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+            if (ompt_enabled.ompt_callback_cancel) {
+              ompt_data_t *task_data;
+              __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL,
+                                            NULL);
+              ompt_cancel_flag_t type = ompt_cancel_parallel;
+              if (cncl_kind == cancel_parallel)
+                type = ompt_cancel_parallel;
+              else if (cncl_kind == cancel_loop)
+                type = ompt_cancel_do;
+              else if (cncl_kind == cancel_sections)
+                type = ompt_cancel_sections;
+              ompt_callbacks.ompt_callback(ompt_callback_cancel)(
+                  task_data, type | ompt_cancel_detected,
+                  OMPT_GET_RETURN_ADDRESS(0));
+            }
+#endif
             return 1 /* true */;
           }
           KMP_ASSERT(0 /* false */);
         } else {
           // we do not have a cancellation request pending, so we just
           // ignore this cancellation point
           return 0;
         }
         break;
       }
     case cancel_taskgroup:
       // cancellation requests for a task group
       // are handled through the taskgroup structure
       {
         kmp_taskdata_t *task;
         kmp_taskgroup_t *taskgroup;
 
         task = this_thr->th.th_current_task;
         KMP_DEBUG_ASSERT(task);
 
         taskgroup = task->td_taskgroup;
         if (taskgroup) {
-          // return the current status of cancellation for the taskgroup
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+          if (ompt_enabled.ompt_callback_cancel &&
+              !!taskgroup->cancel_request) {
+            ompt_data_t *task_data;
+            __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL,
+                                          NULL);
+            ompt_callbacks.ompt_callback(ompt_callback_cancel)(
+                task_data, ompt_cancel_taskgroup | ompt_cancel_detected,
+                OMPT_GET_RETURN_ADDRESS(0));
+          }
+#endif
           return !!taskgroup->cancel_request;
         } else {
           // if a cancellation point is encountered by a task that does not
           // belong to a taskgroup, it is OK to ignore it
           return 0 /* false */;
         }
       }
     default:

runtime/src/kmp_config.h.cmake

 #cmakedefine01 LIBOMP_HAVE___RDTSC
 #define KMP_HAVE___RDTSC LIBOMP_HAVE___RDTSC
 #cmakedefine01 LIBOMP_USE_DEBUGGER
 #define USE_DEBUGGER LIBOMP_USE_DEBUGGER
 #cmakedefine01 LIBOMP_OMPT_DEBUG
 #define OMPT_DEBUG LIBOMP_OMPT_DEBUG
 #cmakedefine01 LIBOMP_OMPT_SUPPORT
 #define OMPT_SUPPORT LIBOMP_OMPT_SUPPORT
-#cmakedefine01 LIBOMP_OMPT_BLAME
-#define OMPT_BLAME LIBOMP_OMPT_BLAME
-#cmakedefine01 LIBOMP_OMPT_TRACE
-#define OMPT_TRACE LIBOMP_OMPT_TRACE
+#cmakedefine01 LIBOMP_OMPT_OPTIONAL
+#define OMPT_OPTIONAL LIBOMP_OMPT_OPTIONAL
 #cmakedefine01 LIBOMP_USE_ADAPTIVE_LOCKS
 #define KMP_USE_ADAPTIVE_LOCKS LIBOMP_USE_ADAPTIVE_LOCKS
 #define KMP_DEBUG_ADAPTIVE_LOCKS 0
 #cmakedefine01 LIBOMP_USE_INTERNODE_ALIGNMENT
 #define KMP_USE_INTERNODE_ALIGNMENT LIBOMP_USE_INTERNODE_ALIGNMENT
 #cmakedefine01 LIBOMP_ENABLE_ASSERTIONS
 #define KMP_USE_ASSERT LIBOMP_ENABLE_ASSERTIONS
 #cmakedefine01 STUBS_LIBRARY

runtime/src/kmp_csupport.cpp

#endif
 
   // maybe to save thr_state is enough here
   {
     va_list ap;
     va_start(ap, microtask);
 
 #if OMPT_SUPPORT
     ompt_frame_t *ompt_frame;
-    if (ompt_enabled) {
+    if (ompt_enabled.enabled) {
       kmp_info_t *master_th = __kmp_threads[gtid];
       kmp_team_t *parent_team = master_th->th.th_team;
       ompt_lw_taskteam_t *lwt = parent_team->t.ompt_serialized_team_info;
       if (lwt)
         ompt_frame = &(lwt->ompt_task_info.frame);
       else {
         int tid = __kmp_tid_from_gtid(gtid);
         ompt_frame = &(
             parent_team->t.t_implicit_task_taskdata[tid].ompt_task_info.frame);
       }
-      ompt_frame->reenter_runtime_frame = __builtin_frame_address(1);
+      ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(gtid);
     }
 #endif
 
 #if INCLUDE_SSC_MARKS
     SSC_MARK_FORKING();
 #endif
     __kmp_fork_call(loc, gtid, fork_context_intel, argc,
-#if OMPT_SUPPORT
-                    VOLATILE_CAST(void *) microtask, // "unwrapped" task
-#endif
                     VOLATILE_CAST(microtask_t) microtask, // "wrapped" task
                     VOLATILE_CAST(launch_t) __kmp_invoke_task_func,
 /* TODO: revert workaround for Intel(R) 64 tracker #96 */
 #if (KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64) && KMP_OS_LINUX
                     &ap
 #else
                     ap
 #endif
                     );
 #if INCLUDE_SSC_MARKS
     SSC_MARK_JOINING();
 #endif
     __kmp_join_call(loc, gtid
 #if OMPT_SUPPORT
                     ,
                     fork_context_intel
 #endif
                     );
 
     va_end(ap);
   }
void __kmpc_fork_teams(ident_t *loc, kmp_int32 argc, kmpc_micro microtask,
   // remember teams entry point and nesting level
   this_thr->th.th_teams_microtask = microtask;
   this_thr->th.th_teams_level =
       this_thr->th.th_team->t.t_level; // AC: can be >0 on host
 
 #if OMPT_SUPPORT
   kmp_team_t *parent_team = this_thr->th.th_team;
   int tid = __kmp_tid_from_gtid(gtid);
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     parent_team->t.t_implicit_task_taskdata[tid]
-        .ompt_task_info.frame.reenter_runtime_frame =
-        __builtin_frame_address(1);
+        .ompt_task_info.frame.reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
   }
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 #endif
 
   // check if __kmpc_push_num_teams called, set default number of teams
   // otherwise
   if (this_thr->th.th_teams_size.nteams == 0) {
     __kmp_push_num_teams(loc, gtid, 0, 0);
   }
   KMP_DEBUG_ASSERT(this_thr->th.th_set_nproc >= 1);
   KMP_DEBUG_ASSERT(this_thr->th.th_teams_size.nteams >= 1);
   KMP_DEBUG_ASSERT(this_thr->th.th_teams_size.nth >= 1);
 
   __kmp_fork_call(loc, gtid, fork_context_intel, argc,
-#if OMPT_SUPPORT
-                  VOLATILE_CAST(void *) microtask, // "unwrapped" task
-#endif
                   VOLATILE_CAST(microtask_t)
                       __kmp_teams_master, // "wrapped" task
                   VOLATILE_CAST(launch_t) __kmp_invoke_teams_master,
 #if (KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64) && KMP_OS_LINUX
                   &ap
 #else
                   ap
 #endif

 Enter a serialized parallel construct. This interface is used to handle a
 conditional parallel region, like this,
 @code
 #pragma omp parallel if (condition)
 @endcode
 when the condition is false.
 */
 void __kmpc_serialized_parallel(ident_t *loc, kmp_int32 global_tid) {
-  // The implementation is now in kmp_runtime.cpp so that it can share static
-  // functions with kmp_fork_call since the tasks to be done are similar in
-  // each case.
+#if OMPT_SUPPORT
+  OMPT_STORE_RETURN_ADDRESS(global_tid);
+#endif
   __kmp_serialized_parallel(loc, global_tid);
 }
 
 /*!
 @ingroup PARALLEL
 @param loc  source location information
 @param global_tid  global thread number
 
#endif
   KMP_MB();
   KMP_DEBUG_ASSERT(serial_team);
   KMP_ASSERT(serial_team->t.t_serialized);
   KMP_DEBUG_ASSERT(this_thr->th.th_team == serial_team);
   KMP_DEBUG_ASSERT(serial_team != this_thr->th.th_root->r.r_root_team);
   KMP_DEBUG_ASSERT(serial_team->t.t_threads);
   KMP_DEBUG_ASSERT(serial_team->t.t_threads[0] == this_thr);
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled &&
+      this_thr->th.ompt_thread_info.state != omp_state_overhead) {
+    OMPT_CUR_TASK_INFO(this_thr)->frame.exit_runtime_frame = NULL;
+    if (ompt_enabled.ompt_callback_implicit_task) {
+      ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+          ompt_scope_end, NULL, OMPT_CUR_TASK_DATA(this_thr), 1,
+          __kmp_tid_from_gtid(global_tid));
+    }
+
+    // reset clear the task id only after unlinking the task
+    ompt_data_t *parent_task_data;
+    __ompt_get_task_info_internal(1, NULL, &parent_task_data, NULL, NULL, NULL);
+
+    if (ompt_enabled.ompt_callback_parallel_end) {
+      ompt_callbacks.ompt_callback(ompt_callback_parallel_end)(
+          &(serial_team->t.ompt_team_info.parallel_data), parent_task_data,
+          ompt_invoker_program, OMPT_LOAD_RETURN_ADDRESS(global_tid));
+    }
+    __ompt_lw_taskteam_unlink(this_thr);
+    this_thr->th.ompt_thread_info.state = omp_state_overhead;
+  }
+#endif
+
   /* If necessary, pop the internal control stack values and replace the team
    * values */
   top = serial_team->t.t_control_stack_top;
   if (top && top->serial_nesting_level == serial_team->t.t_serialized) {
     copy_icvs(&serial_team->t.t_threads[0]->th.th_current_task->td_icvs, top);
     serial_team->t.t_control_stack_top = top->next;
     __kmp_free(top);
   }
if (__kmp_tasking_mode != tskm_immediate_exec) {
       KA_TRACE(20, ("__kmpc_end_serialized_parallel: T#%d decreasing nesting "
                     "depth of serial team %p to %d\n",
                     global_tid, serial_team, serial_team->t.t_serialized));
     }
   }
 
   if (__kmp_env_consistency_check)
     __kmp_pop_parallel(global_tid, NULL);
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled)
+    this_thr->th.ompt_thread_info.state =
+        ((this_thr->th.th_team_serialized) ? omp_state_work_serial
+                                           : omp_state_work_parallel);
+#endif
 }
 
 /*!
 @ingroup SYNCHRONIZATION
 @param loc  source location information.
 
 Execute <tt>flush</tt>. This is implemented as a full memory fence. (Though
 depending on the memory ordering convention obeyed by the compiler
#if KMP_OS_CNK
   // }
   // and adding the yield here is good for at least a 10x speedup
   // when running >2 threads per core (on the NAS LU benchmark).
   __kmp_yield(TRUE);
 #endif
 #else
 #error Unknown or unsupported architecture
 #endif
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_flush) {
+    ompt_callbacks.ompt_callback(ompt_callback_flush)(
+        __ompt_get_thread_data_internal(), OMPT_GET_RETURN_ADDRESS(0));
+  }
+#endif
 }
 
 /* -------------------------------------------------------------------------- */
 /*!
 @ingroup SYNCHRONIZATION
 @param loc source location information
 @param global_tid thread id.
 
void __kmpc_barrier(ident_t *loc, kmp_int32 global_tid) {
   if (__kmp_env_consistency_check) {
     if (loc == 0) {
       KMP_WARNING(ConstructIdentInvalid); // ??? What does it mean for the user?
     }
 
     __kmp_check_barrier(global_tid, ct_barrier, loc);
   }
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT
   ompt_frame_t *ompt_frame;
-  if (ompt_enabled) {
-    ompt_frame = __ompt_get_task_frame_internal(0);
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
     if (ompt_frame->reenter_runtime_frame == NULL)
-      ompt_frame->reenter_runtime_frame = __builtin_frame_address(1);
+      ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(global_tid);
   }
 #endif
   __kmp_threads[global_tid]->th.th_ident = loc;
   // TODO: explicit barrier_wait_id:
   //   this function is called when 'barrier' directive is present or
   //   implicit barrier at the end of a worksharing construct.
   // 1) better to add a per-thread barrier counter to a thread data structure
   // 2) set to 0 when a new team is created
   // 4) no sync is required
 
   __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL);
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
     ompt_frame->reenter_runtime_frame = NULL;
   }
 #endif
 }
 
 /* The BARRIER for a MASTER section is always explicit   */
 /*!
 @ingroup WORK_SHARING
if (!TCR_4(__kmp_init_parallel))
     __kmp_parallel_initialize();
 
   if (KMP_MASTER_GTID(global_tid)) {
     KMP_COUNT_BLOCK(OMP_MASTER);
     KMP_PUSH_PARTITIONED_TIMER(OMP_master);
     status = 1;
   }
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   if (status) {
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_master_begin)) {
+    if (ompt_enabled.ompt_callback_master) {
       kmp_info_t *this_thr = __kmp_threads[global_tid];
       kmp_team_t *team = this_thr->th.th_team;
 
       int tid = __kmp_tid_from_gtid(global_tid);
-      ompt_callbacks.ompt_callback(ompt_event_master_begin)(
-          team->t.ompt_team_info.parallel_id,
-          team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
+      ompt_callbacks.ompt_callback(ompt_callback_master)(
+          ompt_scope_begin, &(team->t.ompt_team_info.parallel_data),
+          &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+          OMPT_GET_RETURN_ADDRESS(0));
     }
   }
 #endif
 
   if (__kmp_env_consistency_check) {
 #if KMP_USE_DYNAMIC_LOCK
     if (status)
       __kmp_push_sync(global_tid, ct_master, loc, NULL, 0);

 thread that executes the <tt>master</tt> region.
 */
 void __kmpc_end_master(ident_t *loc, kmp_int32 global_tid) {
   KC_TRACE(10, ("__kmpc_end_master: called T#%d\n", global_tid));
 
   KMP_DEBUG_ASSERT(KMP_MASTER_GTID(global_tid));
   KMP_POP_PARTITIONED_TIMER();
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   kmp_info_t *this_thr = __kmp_threads[global_tid];
   kmp_team_t *team = this_thr->th.th_team;
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_master_end)) {
+  if (ompt_enabled.ompt_callback_master) {
     int tid = __kmp_tid_from_gtid(global_tid);
-    ompt_callbacks.ompt_callback(ompt_event_master_end)(
-        team->t.ompt_team_info.parallel_id,
-        team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
+    ompt_callbacks.ompt_callback(ompt_callback_master)(
+        ompt_scope_end, &(team->t.ompt_team_info.parallel_data),
+        &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+        OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 
   if (__kmp_env_consistency_check) {
     if (global_tid < 0)
       KMP_WARNING(ThreadIdentInvalid);
 
     if (KMP_MASTER_GTID(global_tid))

 
 #if USE_ITT_BUILD
   __kmp_itt_ordered_prep(gtid);
 // TODO: ordered_wait_id
 #endif /* USE_ITT_BUILD */
 
   th = __kmp_threads[gtid];
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  kmp_team_t *team;
+  ompt_wait_id_t lck;
+  void *codeptr_ra;
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+    team = __kmp_team_from_gtid(gtid);
+    lck = (ompt_wait_id_t)&team->t.t_ordered.dt.t_value;
     /* OMPT state update */
-    th->th.ompt_thread_info.wait_id = (uint64_t)loc;
-    th->th.ompt_thread_info.state = ompt_state_wait_ordered;
+    th->th.ompt_thread_info.wait_id = lck;
+    th->th.ompt_thread_info.state = omp_state_wait_ordered;
 
     /* OMPT event callback */
-    if (ompt_callbacks.ompt_callback(ompt_event_wait_ordered)) {
-      ompt_callbacks.ompt_callback(ompt_event_wait_ordered)(
-          th->th.ompt_thread_info.wait_id);
+    codeptr_ra = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (ompt_enabled.ompt_callback_mutex_acquire) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+          ompt_mutex_ordered, omp_lock_hint_none, ompt_mutex_impl_spin,
+          (ompt_wait_id_t)lck, codeptr_ra);
     }
   }
 #endif
 
   if (th->th.th_dispatch->th_deo_fcn != 0)
     (*th->th.th_dispatch->th_deo_fcn)(&gtid, &cid, loc);
   else
     __kmp_parallel_deo(&gtid, &cid, loc);
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
     /* OMPT state update */
-    th->th.ompt_thread_info.state = ompt_state_work_parallel;
+    th->th.ompt_thread_info.state = omp_state_work_parallel;
     th->th.ompt_thread_info.wait_id = 0;
 
     /* OMPT event callback */
-    if (ompt_callbacks.ompt_callback(ompt_event_acquired_ordered)) {
-      ompt_callbacks.ompt_callback(ompt_event_acquired_ordered)(
-          th->th.ompt_thread_info.wait_id);
+    if (ompt_enabled.ompt_callback_mutex_acquired) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+          ompt_mutex_ordered, (ompt_wait_id_t)lck, codeptr_ra);
     }
   }
 #endif
 
 #if USE_ITT_BUILD
   __kmp_itt_ordered_start(gtid);
 #endif /* USE_ITT_BUILD */
 }
#endif /* USE_ITT_BUILD */
 
   th = __kmp_threads[gtid];
 
   if (th->th.th_dispatch->th_dxo_fcn != 0)
     (*th->th.th_dispatch->th_dxo_fcn)(&gtid, &cid, loc);
   else
     __kmp_parallel_dxo(&gtid, &cid, loc);
 
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_release_ordered)) {
-    ompt_callbacks.ompt_callback(ompt_event_release_ordered)(
-        th->th.ompt_thread_info.wait_id);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+  if (ompt_enabled.ompt_callback_mutex_released) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+        ompt_mutex_ordered,
+        (ompt_wait_id_t)&__kmp_team_from_gtid(gtid)->t.t_ordered.dt.t_value,
+        OMPT_LOAD_RETURN_ADDRESS(gtid));
   }
 #endif
 }
 
 #if KMP_USE_DYNAMIC_LOCK
 
 static __forceinline void
 __kmp_init_indirect_csptr(kmp_critical_name *crit, ident_t const *loc,

 associated with the critical section, or some other suitably unique value.
 
 Enter code protected by a `critical` construct.
 This function blocks until the executing thread can enter the critical section.
 */
 void __kmpc_critical(ident_t *loc, kmp_int32 global_tid,
                      kmp_critical_name *crit) {
 #if KMP_USE_DYNAMIC_LOCK
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(global_tid);
+#endif // OMPT_SUPPORT
   __kmpc_critical_with_hint(loc, global_tid, crit, omp_lock_hint_none);
 #else
   KMP_COUNT_BLOCK(OMP_CRITICAL);
   KMP_TIME_PARTITIONED_BLOCK(
       OMP_critical_wait); /* Time spent waiting to enter the critical section */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  omp_state_t prev_state = omp_state_undefined;
+  ompt_thread_info_t ti;
+#endif
   kmp_user_lock_p lck;
 
   KC_TRACE(10, ("__kmpc_critical: called T#%d\n", global_tid));
 
   // TODO: add THR_OVHD_STATE
 
   KMP_CHECK_USER_LOCK_INIT();
 

 // since the critical directive binds to all threads, not just the current
 // team we have to check this even if we are in a serialized team.
 // also, even if we are the uber thread, we still have to conduct the lock,
 // as we have to contend with sibling threads.
 
 #if USE_ITT_BUILD
   __kmp_itt_critical_acquiring(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+  void *codeptr_ra = NULL;
+  if (ompt_enabled.enabled) {
+    ti = __kmp_threads[global_tid]->th.ompt_thread_info;
+    /* OMPT state update */
+    prev_state = ti.state;
+    ti.wait_id = (ompt_wait_id_t)lck;
+    ti.state = omp_state_wait_critical;
+
+    /* OMPT event callback */
+    codeptr_ra = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (ompt_enabled.ompt_callback_mutex_acquire) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+          ompt_mutex_critical, omp_lock_hint_none, __ompt_get_mutex_impl_type(),
+          (ompt_wait_id_t)crit, codeptr_ra);
+    }
+  }
+#endif
   // Value of 'crit' should be good for using as a critical_id of the critical
   // section directive.
   __kmp_acquire_user_lock_with_checks(lck, global_tid);
 
 #if USE_ITT_BUILD
   __kmp_itt_critical_acquired(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    /* OMPT state update */
+    ti.state = prev_state;
+    ti.wait_id = 0;
+
+    /* OMPT event callback */
+    if (ompt_enabled.ompt_callback_mutex_acquired) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+          ompt_mutex_critical, (ompt_wait_id_t)crit, codeptr_ra);
+    }
+  }
+#endif
 
   KMP_START_EXPLICIT_TIMER(OMP_critical);
   KA_TRACE(15, ("__kmpc_critical: done T#%d\n", global_tid));
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 #if KMP_USE_DYNAMIC_LOCK
 
#endif
 
   // HLE lock for speculation
   if (hint & omp_lock_hint_speculative)
     return KMP_TSX_LOCK(hle);
 
   return __kmp_user_lock_seq;
 }
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+static ompt_mutex_impl_t
+__ompt_get_mutex_impl_type(void *user_lock, kmp_indirect_lock_t *ilock = 0) {
+  if (user_lock) {
+    switch (KMP_EXTRACT_D_TAG(user_lock)) {
+    case 0:
+      break;
+#if KMP_USE_FUTEX
+    case locktag_futex:
+      return ompt_mutex_impl_queuing;
+#endif
+    case locktag_tas:
+      return ompt_mutex_impl_spin;
+#if KMP_USE_TSX
+    case locktag_hle:
+      return ompt_mutex_impl_speculative;
+#endif
+    default:
+      return ompt_mutex_impl_unknown;
+    }
+    ilock = KMP_LOOKUP_I_LOCK(user_lock);
+  }
+  KMP_ASSERT(ilock);
+  switch (ilock->type) {
+#if KMP_USE_TSX
+  case locktag_adaptive:
+  case locktag_rtm:
+    return ompt_mutex_impl_speculative;
+#endif
+  case locktag_nested_tas:
+    return ompt_mutex_impl_spin;
+#if KMP_USE_FUTEX
+  case locktag_nested_futex:
+#endif
+  case locktag_ticket:
+  case locktag_queuing:
+  case locktag_drdpa:
+  case locktag_nested_ticket:
+  case locktag_nested_queuing:
+  case locktag_nested_drdpa:
+    return ompt_mutex_impl_queuing;
+  default:
+    return ompt_mutex_impl_unknown;
+  }
+}
+
+// For locks without dynamic binding
+static ompt_mutex_impl_t __ompt_get_mutex_impl_type() {
+  switch (__kmp_user_lock_kind) {
+  case lk_tas:
+    return ompt_mutex_impl_spin;
+#if KMP_USE_FUTEX
+  case lk_futex:
+#endif
+  case lk_ticket:
+  case lk_queuing:
+  case lk_drdpa:
+    return ompt_mutex_impl_queuing;
+#if KMP_USE_TSX
+  case lk_hle:
+  case lk_rtm:
+  case lk_adaptive:
+    return ompt_mutex_impl_speculative;
+#endif
+  default:
+    return ompt_mutex_impl_unknown;
+  }
+}
+#endif
+
 /*!
 @ingroup WORK_SHARING
 @param loc  source location information.
 @param global_tid  global thread number.
 @param crit identity of the critical section. This could be a pointer to a lock
 associated with the critical section, or some other suitably unique value.
 @param hint the lock hint.
 
 Enter code protected by a `critical` construct with a hint. The hint value is
 used to suggest a lock implementation. This function blocks until the executing
 thread can enter the critical section unless the hint suggests use of
 speculative execution and the hardware supports it.
 */
 void __kmpc_critical_with_hint(ident_t *loc, kmp_int32 global_tid,
                                kmp_critical_name *crit, uintptr_t hint) {
   KMP_COUNT_BLOCK(OMP_CRITICAL);
   kmp_user_lock_p lck;
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  omp_state_t prev_state = omp_state_undefined;
+  ompt_thread_info_t ti;
+  // This is the case, if called from __kmpc_critical:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(global_tid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+#endif
 
   KC_TRACE(10, ("__kmpc_critical: called T#%d\n", global_tid));
 
   kmp_dyna_lock_t *lk = (kmp_dyna_lock_t *)crit;
   // Check if it is initialized.
   if (*lk == 0) {
     kmp_dyna_lockseq_t lckseq = __kmp_map_hint_to_lock(hint);
     if (KMP_IS_D_LOCK(lckseq)) {
if (KMP_EXTRACT_D_TAG(lk) != 0) {
     lck = (kmp_user_lock_p)lk;
     if (__kmp_env_consistency_check) {
       __kmp_push_sync(global_tid, ct_critical, loc, lck,
                       __kmp_map_hint_to_lock(hint));
     }
 #if USE_ITT_BUILD
     __kmp_itt_critical_acquiring(lck);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ti = __kmp_threads[global_tid]->th.ompt_thread_info;
+      /* OMPT state update */
+      prev_state = ti.state;
+      ti.wait_id = (ompt_wait_id_t)lck;
+      ti.state = omp_state_wait_critical;
+
+      /* OMPT event callback */
+      if (ompt_enabled.ompt_callback_mutex_acquire) {
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+            ompt_mutex_critical, (unsigned int)hint,
+            __ompt_get_mutex_impl_type(crit), (ompt_wait_id_t)crit, codeptr);
+      }
+    }
+#endif
 #if KMP_USE_INLINED_TAS
     if (__kmp_user_lock_seq == lockseq_tas && !__kmp_env_consistency_check) {
       KMP_ACQUIRE_TAS_LOCK(lck, global_tid);
     } else
 #elif KMP_USE_INLINED_FUTEX
     if (__kmp_user_lock_seq == lockseq_futex && !__kmp_env_consistency_check) {
       KMP_ACQUIRE_FUTEX_LOCK(lck, global_tid);
     } else
 #endif
     {
       KMP_D_LOCK_FUNC(lk, set)(lk, global_tid);
     }
   } else {
     kmp_indirect_lock_t *ilk = *((kmp_indirect_lock_t **)lk);
     lck = ilk->lock;
     if (__kmp_env_consistency_check) {
       __kmp_push_sync(global_tid, ct_critical, loc, lck,
                       __kmp_map_hint_to_lock(hint));
     }
 #if USE_ITT_BUILD
     __kmp_itt_critical_acquiring(lck);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ti = __kmp_threads[global_tid]->th.ompt_thread_info;
+      /* OMPT state update */
+      prev_state = ti.state;
+      ti.wait_id = (ompt_wait_id_t)lck;
+      ti.state = omp_state_wait_critical;
+
+      /* OMPT event callback */
+      if (ompt_enabled.ompt_callback_mutex_acquire) {
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+            ompt_mutex_critical, (unsigned int)hint,
+            __ompt_get_mutex_impl_type(0, ilk), (ompt_wait_id_t)crit, codeptr);
+      }
+    }
+#endif
     KMP_I_LOCK_FUNC(ilk, set)(lck, global_tid);
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_critical_acquired(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    /* OMPT state update */
+    ti.state = prev_state;
+    ti.wait_id = 0;
+
+    /* OMPT event callback */
+    if (ompt_enabled.ompt_callback_mutex_acquired) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+          ompt_mutex_critical, (ompt_wait_id_t)crit, codeptr);
+    }
+  }
+#endif
 
   KMP_PUSH_PARTITIONED_TIMER(OMP_critical);
   KA_TRACE(15, ("__kmpc_critical: done T#%d\n", global_tid));
 } // __kmpc_critical_with_hint
 
 #endif // KMP_USE_DYNAMIC_LOCK
 
 /*!

 
 #if USE_ITT_BUILD
   __kmp_itt_critical_releasing(lck);
 #endif /* USE_ITT_BUILD */
   // Value of 'crit' should be good for using as a critical_id of the critical
   // section directive.
   __kmp_release_user_lock_with_checks(lck, global_tid);
 
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_release_critical)) {
-    ompt_callbacks.ompt_callback(ompt_event_release_critical)((uint64_t)lck);
+#endif // KMP_USE_DYNAMIC_LOCK
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  /* OMPT release event triggers after lock is released; place here to trigger
+   * for all #if branches */
+  OMPT_STORE_RETURN_ADDRESS(global_tid);
+  if (ompt_enabled.ompt_callback_mutex_released) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+        ompt_mutex_critical, (ompt_wait_id_t)crit, OMPT_LOAD_RETURN_ADDRESS(0));
   }
 #endif
 
-#endif // KMP_USE_DYNAMIC_LOCK
   KMP_POP_PARTITIONED_TIMER();
   KA_TRACE(15, ("__kmpc_end_critical: done T#%d\n", global_tid));
 }
 
 /*!
 @ingroup SYNCHRONIZATION
 @param loc source location information
 @param global_tid thread id.
 @return one if the thread should execute the master block, zero otherwise
 
 Start execution of a combined barrier and master. The barrier is executed inside
 this function.
 */
 kmp_int32 __kmpc_barrier_master(ident_t *loc, kmp_int32 global_tid) {
   int status;
 
   KC_TRACE(10, ("__kmpc_barrier_master: called T#%d\n", global_tid));
 
   if (!TCR_4(__kmp_init_parallel))
     __kmp_parallel_initialize();
 
   if (__kmp_env_consistency_check)
     __kmp_check_barrier(global_tid, ct_barrier, loc);
 
+#if OMPT_SUPPORT
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    if (ompt_frame->reenter_runtime_frame == NULL)
+      ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(global_tid);
+  }
+#endif
 #if USE_ITT_NOTIFY
   __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
   status = __kmp_barrier(bs_plain_barrier, global_tid, TRUE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 
   return (status != 0) ? 0 : 1;
 }
 
 /*!
 @ingroup SYNCHRONIZATION
 @param loc source location information
 @param global_tid thread id.
kmp_int32 __kmpc_barrier_master_nowait(ident_t *loc, kmp_int32 global_tid) {
 
   if (__kmp_env_consistency_check) {
     if (loc == 0) {
       KMP_WARNING(ConstructIdentInvalid); // ??? What does it mean for the user?
     }
     __kmp_check_barrier(global_tid, ct_barrier, loc);
   }
 
+#if OMPT_SUPPORT
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    if (ompt_frame->reenter_runtime_frame == NULL)
+      ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(global_tid);
+  }
+#endif
 #if USE_ITT_NOTIFY
   __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
   __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 
   ret = __kmpc_master(loc, global_tid);
 
   if (__kmp_env_consistency_check) {
     /*  there's no __kmpc_end_master called; so the (stats) */
     /*  actions of __kmpc_end_master are done here          */
 
     if (global_tid < 0) {
kmp_int32 __kmpc_single(ident_t *loc, kmp_int32 global_tid) {
   kmp_int32 rc = __kmp_enter_single(global_tid, loc, TRUE);
 
   if (rc) {
     // We are going to execute the single statement, so we should count it.
     KMP_COUNT_BLOCK(OMP_SINGLE);
     KMP_PUSH_PARTITIONED_TIMER(OMP_single);
   }
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   kmp_info_t *this_thr = __kmp_threads[global_tid];
   kmp_team_t *team = this_thr->th.th_team;
   int tid = __kmp_tid_from_gtid(global_tid);
 
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     if (rc) {
-      if (ompt_callbacks.ompt_callback(ompt_event_single_in_block_begin)) {
-        ompt_callbacks.ompt_callback(ompt_event_single_in_block_begin)(
-            team->t.ompt_team_info.parallel_id,
-            team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id,
-            team->t.ompt_team_info.microtask);
+      if (ompt_enabled.ompt_callback_work) {
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_executor, ompt_scope_begin,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_work) {
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_other, ompt_scope_begin,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_other, ompt_scope_end,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
       }
-    } else {
-      if (ompt_callbacks.ompt_callback(ompt_event_single_others_begin)) {
-        ompt_callbacks.ompt_callback(ompt_event_single_others_begin)(
-            team->t.ompt_team_info.parallel_id,
-            team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
-      }
-      this_thr->th.ompt_thread_info.state = ompt_state_wait_single;
     }
   }
 #endif
 
   return rc;
 }
 
 /*!
 @ingroup WORK_SHARING
 @param loc  source location information
 @param global_tid  global thread number
 
 Mark the end of a <tt>single</tt> construct.  This function should
 only be called by the thread that executed the block of code protected
 by the `single` construct.
 */
 void __kmpc_end_single(ident_t *loc, kmp_int32 global_tid) {
   __kmp_exit_single(global_tid);
   KMP_POP_PARTITIONED_TIMER();
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   kmp_info_t *this_thr = __kmp_threads[global_tid];
   kmp_team_t *team = this_thr->th.th_team;
   int tid = __kmp_tid_from_gtid(global_tid);
 
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_single_in_block_end)) {
-    ompt_callbacks.ompt_callback(ompt_event_single_in_block_end)(
-        team->t.ompt_team_info.parallel_id,
-        team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id);
+  if (ompt_enabled.ompt_callback_work) {
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_single_executor, ompt_scope_end,
+        &(team->t.ompt_team_info.parallel_data),
+        &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data), 1,
+        OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 }
 
 /*!
 @ingroup WORK_SHARING
 @param loc Source location
 @param global_tid Global thread id
 
 Mark the end of a statically scheduled loop.
 */
 void __kmpc_for_static_fini(ident_t *loc, kmp_int32 global_tid) {
   KE_TRACE(10, ("__kmpc_for_static_fini called T#%d\n", global_tid));
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_end)) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_work) {
+    ompt_work_type_t ompt_work_type;
     ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-    ompt_callbacks.ompt_callback(ompt_event_loop_end)(team_info->parallel_id,
-                                                      task_info->task_id);
+    ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
+    // Determine workshare type
+    if (loc != NULL) {
+      if ((loc->flags & KMP_IDENT_WORK_LOOP) != 0) {
+        ompt_work_type = ompt_work_loop;
+      } else if ((loc->flags & KMP_IDENT_WORK_SECTIONS) != 0) {
+        ompt_work_type = ompt_work_sections;
+      } else if ((loc->flags & KMP_IDENT_WORK_DISTRIBUTE) != 0) {
+        ompt_work_type = ompt_work_distribute;
+      } else {
+        KMP_ASSERT2(0,
+                    "__kmpc_for_static_fini: can't determine workshare type");
+      }
+      KMP_DEBUG_ASSERT(ompt_work_type);
+    }
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_type, ompt_scope_end, &(team_info->parallel_data),
+        &(task_info->task_data), 0, OMPT_GET_RETURN_ADDRESS(0));
   }
 #endif
 
   if (__kmp_env_consistency_check)
     __kmp_pop_workshare(global_tid, ct_pdo, loc);
 }
 
 // User routines which take C-style arguments (call by value)
if (__kmp_env_consistency_check) {
     }
   }
 
   // ToDo: Optimize the following two barriers into some kind of split barrier
 
   if (didit)
     *data_ptr = cpy_data;
 
+#if OMPT_SUPPORT
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    if (ompt_frame->reenter_runtime_frame == NULL)
+      ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
 /* This barrier is not a barrier region boundary */
 #if USE_ITT_NOTIFY
   __kmp_threads[gtid]->th.th_ident = loc;
 #endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
 
   if (!didit)
     (*cpy_func)(cpy_data, *data_ptr);
 
 // Consider next barrier a user-visible barrier for barrier region boundaries
 // Nesting checks are already handled by the single construct checks
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
 #if USE_ITT_NOTIFY
   __kmp_threads[gtid]->th.th_ident = loc; // TODO: check if it is needed (e.g.
 // tasks can overwrite the location)
 #endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 }
 
 /* -------------------------------------------------------------------------- */
 
 #define INIT_LOCK __kmp_init_user_lock_with_checks
 #define INIT_NESTED_LOCK __kmp_init_nested_user_lock_with_checks
 #define ACQUIRE_LOCK __kmp_acquire_user_lock_with_checks
 #define ACQUIRE_LOCK_TIMED __kmp_acquire_user_lock_with_checks_timed

 void __kmpc_init_lock_with_hint(ident_t *loc, kmp_int32 gtid, void **user_lock,
                                 uintptr_t hint) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
   if (__kmp_env_consistency_check && user_lock == NULL) {
     KMP_FATAL(LockIsUninitialized, "omp_init_lock_with_hint");
   }
 
   __kmp_init_lock_with_hint(loc, user_lock, __kmp_map_hint_to_lock(hint));
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_lock, (omp_lock_hint_t)hint,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
 }
 
 /* initialize the lock with a hint */
 void __kmpc_init_nest_lock_with_hint(ident_t *loc, kmp_int32 gtid,
                                      void **user_lock, uintptr_t hint) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
   if (__kmp_env_consistency_check && user_lock == NULL) {
     KMP_FATAL(LockIsUninitialized, "omp_init_nest_lock_with_hint");
   }
 
   __kmp_init_nest_lock_with_hint(loc, user_lock, __kmp_map_hint_to_lock(hint));
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_nest_lock, (omp_lock_hint_t)hint,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
 }
 
 #endif // KMP_USE_DYNAMIC_LOCK
 
 /* initialize the lock */
 void __kmpc_init_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
   KMP_DEBUG_ASSERT(__kmp_init_serial);
   if (__kmp_env_consistency_check && user_lock == NULL) {
     KMP_FATAL(LockIsUninitialized, "omp_init_lock");
   }
   __kmp_init_lock_with_hint(loc, user_lock, __kmp_user_lock_seq);
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_lock, omp_lock_hint_none,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
+
 #else // KMP_USE_DYNAMIC_LOCK
 
   static char const *const func = "omp_init_lock";
   kmp_user_lock_p lck;
   KMP_DEBUG_ASSERT(__kmp_init_serial);
 
   if (__kmp_env_consistency_check) {
     if (user_lock == NULL) {
#if KMP_USE_FUTEX
   }
 #endif
   else {
     lck = __kmp_user_lock_allocate(user_lock, gtid, 0);
   }
   INIT_LOCK(lck);
   __kmp_set_user_lock_location(lck, loc);
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_init_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_init_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(),
+        (ompt_wait_id_t)user_lock, codeptr);
   }
 #endif
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_creating(lck);
 #endif /* USE_ITT_BUILD */
 
 #endif // KMP_USE_DYNAMIC_LOCK
 } // __kmpc_init_lock
 
 /* initialize the lock */
 void __kmpc_init_nest_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
   KMP_DEBUG_ASSERT(__kmp_init_serial);
   if (__kmp_env_consistency_check && user_lock == NULL) {
     KMP_FATAL(LockIsUninitialized, "omp_init_nest_lock");
   }
   __kmp_init_nest_lock_with_hint(loc, user_lock, __kmp_user_lock_seq);
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_nest_lock, omp_lock_hint_none,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
+
 #else // KMP_USE_DYNAMIC_LOCK
 
   static char const *const func = "omp_init_nest_lock";
   kmp_user_lock_p lck;
   KMP_DEBUG_ASSERT(__kmp_init_serial);
 
   if (__kmp_env_consistency_check) {
     if (user_lock == NULL) {

 #endif
   else {
     lck = __kmp_user_lock_allocate(user_lock, gtid, 0);
   }
 
   INIT_NESTED_LOCK(lck);
   __kmp_set_user_lock_location(lck, loc);
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_init_nest_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_init_nest_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_init) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_init)(
+        ompt_mutex_nest_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(),
+        (ompt_wait_id_t)user_lock, codeptr);
   }
 #endif
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_creating(lck);
 #endif /* USE_ITT_BUILD */
 
 #endif // KMP_USE_DYNAMIC_LOCK
 } // __kmpc_init_nest_lock
 
 void __kmpc_destroy_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
 #if USE_ITT_BUILD
   kmp_user_lock_p lck;
   if (KMP_EXTRACT_D_TAG(user_lock) == 0) {
     lck = ((kmp_indirect_lock_t *)KMP_LOOKUP_I_LOCK(user_lock))->lock;
   } else {
     lck = (kmp_user_lock_p)user_lock;
   }
   __kmp_itt_lock_destroyed(lck);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_destroy) {
+    kmp_user_lock_p lck;
+    if (KMP_EXTRACT_D_TAG(user_lock) == 0) {
+      lck = ((kmp_indirect_lock_t *)KMP_LOOKUP_I_LOCK(user_lock))->lock;
+    } else {
+      lck = (kmp_user_lock_p)user_lock;
+    }
+    ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)(
+        ompt_mutex_lock, (ompt_wait_id_t)user_lock, codeptr);
+  }
+#endif
   KMP_D_LOCK_FUNC(user_lock, destroy)((kmp_dyna_lock_t *)user_lock);
 #else
   kmp_user_lock_p lck;
 
   if ((__kmp_user_lock_kind == lk_tas) &&
       (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #if KMP_USE_FUTEX
   else if ((__kmp_user_lock_kind == lk_futex) &&
            (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_destroy_lock");
   }
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_destroy_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_destroy_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_destroy) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)(
+        ompt_mutex_lock, (ompt_wait_id_t)user_lock, codeptr);
   }
 #endif
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_destroyed(lck);
 #endif /* USE_ITT_BUILD */
   DESTROY_LOCK(lck);
 

 /* destroy the lock */
 void __kmpc_destroy_nest_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
 #if USE_ITT_BUILD
   kmp_indirect_lock_t *ilk = KMP_LOOKUP_I_LOCK(user_lock);
   __kmp_itt_lock_destroyed(ilk->lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_destroy) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)(
+        ompt_mutex_nest_lock, (ompt_wait_id_t)user_lock, codeptr);
+  }
+#endif
   KMP_D_LOCK_FUNC(user_lock, destroy)((kmp_dyna_lock_t *)user_lock);
 
 #else // KMP_USE_DYNAMIC_LOCK
 
   kmp_user_lock_p lck;
 
   if ((__kmp_user_lock_kind == lk_tas) &&
       (sizeof(lck->tas.lk.poll) + sizeof(lck->tas.lk.depth_locked) <=
        OMP_NEST_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #if KMP_USE_FUTEX
   else if ((__kmp_user_lock_kind == lk_futex) &&
            (sizeof(lck->futex.lk.poll) + sizeof(lck->futex.lk.depth_locked) <=
             OMP_NEST_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_destroy_nest_lock");
   }
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_destroy_nest_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_destroy_nest_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_lock_destroy) {
+    ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)(
+        ompt_mutex_nest_lock, (ompt_wait_id_t)user_lock, codeptr);
   }
 #endif
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_destroyed(lck);
 #endif /* USE_ITT_BUILD */
 
   DESTROY_NESTED_LOCK(lck);
void __kmpc_set_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
   KMP_COUNT_BLOCK(OMP_set_lock);
 #if KMP_USE_DYNAMIC_LOCK
   int tag = KMP_EXTRACT_D_TAG(user_lock);
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring(
       (kmp_user_lock_p)
           user_lock); // itt function will get to the right lock object.
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_lock, omp_lock_hint_none,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
 #if KMP_USE_INLINED_TAS
   if (tag == locktag_tas && !__kmp_env_consistency_check) {
     KMP_ACQUIRE_TAS_LOCK(user_lock, gtid);
   } else
 #elif KMP_USE_INLINED_FUTEX
   if (tag == locktag_futex && !__kmp_env_consistency_check) {
     KMP_ACQUIRE_FUTEX_LOCK(user_lock, gtid);
   } else
 #endif
   {
     __kmp_direct_set[tag]((kmp_dyna_lock_t *)user_lock, gtid);
   }
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquired((kmp_user_lock_p)user_lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_mutex_acquired) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+        ompt_mutex_lock, (ompt_wait_id_t)user_lock, codeptr);
+  }
+#endif
 
 #else // KMP_USE_DYNAMIC_LOCK
 
   kmp_user_lock_p lck;
 
   if ((__kmp_user_lock_kind == lk_tas) &&
       (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #if KMP_USE_FUTEX
   else if ((__kmp_user_lock_kind == lk_futex) &&
            (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_set_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(),
+        (ompt_wait_id_t)lck, codeptr);
+  }
+#endif
 
   ACQUIRE_LOCK(lck, gtid);
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquired(lck);
 #endif /* USE_ITT_BUILD */
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_acquired_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_acquired_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_mutex_acquired) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+        ompt_mutex_lock, (ompt_wait_id_t)lck, codeptr);
   }
 #endif
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 void __kmpc_set_nest_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring((kmp_user_lock_p)user_lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.enabled) {
+    if (ompt_enabled.ompt_callback_mutex_acquire) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+          ompt_mutex_nest_lock, omp_lock_hint_none,
+          __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+          codeptr);
+    }
+  }
+#endif
+  int acquire_status =
       KMP_D_LOCK_FUNC(user_lock, set)((kmp_dyna_lock_t *)user_lock, gtid);
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquired((kmp_user_lock_p)user_lock);
 #endif
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
-    // missing support here: need to know whether acquired first or not
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    if (acquire_status == KMP_LOCK_ACQUIRED_FIRST) {
+      if (ompt_enabled.ompt_callback_mutex_acquired) {
+        // lock_first
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)user_lock, codeptr);
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_nest_lock) {
+        // lock_next
+        ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+            ompt_scope_begin, (ompt_wait_id_t)user_lock, codeptr);
+      }
+    }
   }
 #endif
 
 #else // KMP_USE_DYNAMIC_LOCK
   int acquire_status;
   kmp_user_lock_p lck;
 
   if ((__kmp_user_lock_kind == lk_tas) &&

 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_set_nest_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.enabled) {
+    if (ompt_enabled.ompt_callback_mutex_acquire) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+          ompt_mutex_nest_lock, omp_lock_hint_none,
+          __ompt_get_mutex_impl_type(), (ompt_wait_id_t)lck, codeptr);
+    }
+  }
+#endif
 
   ACQUIRE_NESTED_LOCK(lck, gtid, &acquire_status);
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquired(lck);
 #endif /* USE_ITT_BUILD */
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
     if (acquire_status == KMP_LOCK_ACQUIRED_FIRST) {
-      if (ompt_callbacks.ompt_callback(ompt_event_acquired_nest_lock_first))
-        ompt_callbacks.ompt_callback(ompt_event_acquired_nest_lock_first)(
-            (uint64_t)lck);
-    } else {
-      if (ompt_callbacks.ompt_callback(ompt_event_acquired_nest_lock_next))
-        ompt_callbacks.ompt_callback(ompt_event_acquired_nest_lock_next)(
-            (uint64_t)lck);
+      if (ompt_enabled.ompt_callback_mutex_acquired) {
+        // lock_first
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)lck, codeptr);
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_nest_lock) {
+        // lock_next
+        ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+            ompt_scope_begin, (ompt_wait_id_t)lck, codeptr);
+      }
     }
   }
 #endif
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 void __kmpc_unset_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
#elif KMP_USE_INLINED_FUTEX
   if (tag == locktag_futex && !__kmp_env_consistency_check) {
     KMP_RELEASE_FUTEX_LOCK(user_lock, gtid);
   } else
 #endif
   {
     __kmp_direct_unset[tag]((kmp_dyna_lock_t *)user_lock, gtid);
   }
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_released) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+        ompt_mutex_lock, (ompt_wait_id_t)user_lock, codeptr);
+  }
+#endif
+
 #else // KMP_USE_DYNAMIC_LOCK
 
   kmp_user_lock_p lck;
 
   /* Can't use serial interval since not block structured */
   /* release the lock */
 
   if ((__kmp_user_lock_kind == lk_tas) &&
       (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) {
 #if KMP_OS_LINUX &&                                                            \
     (KMP_ARCH_X86 || KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64)
 // "fast" path implemented to fix customer performance issue
 #if USE_ITT_BUILD
     __kmp_itt_lock_releasing((kmp_user_lock_p)user_lock);
 #endif /* USE_ITT_BUILD */
     TCW_4(((kmp_user_lock_p)user_lock)->tas.lk.poll, 0);
     KMP_MB();
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    // This is the case, if called from omp_init_lock_with_hint:
+    void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (!codeptr)
+      codeptr = OMPT_GET_RETURN_ADDRESS(0);
+    if (ompt_enabled.ompt_callback_mutex_released) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+          ompt_mutex_lock, (ompt_wait_id_t)lck, codeptr);
+    }
+#endif
+
     return;
 #else
     lck = (kmp_user_lock_p)user_lock;
 #endif
   }
 #if KMP_USE_FUTEX
   else if ((__kmp_user_lock_kind == lk_futex) &&
            (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_unset_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_releasing(lck);
 #endif /* USE_ITT_BUILD */
 
   RELEASE_LOCK(lck, gtid);
 
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_release_lock)) {
-    ompt_callbacks.ompt_callback(ompt_event_release_lock)((uint64_t)lck);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_released) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+        ompt_mutex_lock, (ompt_wait_id_t)lck, codeptr);
   }
 #endif
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 /* release the lock */
 void __kmpc_unset_nest_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_releasing((kmp_user_lock_p)user_lock);
 #endif
+  int release_status =
       KMP_D_LOCK_FUNC(user_lock, unset)((kmp_dyna_lock_t *)user_lock, gtid);
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.enabled) {
+    if (release_status == KMP_LOCK_RELEASED) {
+      if (ompt_enabled.ompt_callback_mutex_released) {
+        // release_lock_last
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)user_lock, codeptr);
+      }
+    } else if (ompt_enabled.ompt_callback_nest_lock) {
+      // release_lock_prev
+      ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+          ompt_scope_end, (ompt_wait_id_t)user_lock, codeptr);
+    }
+  }
+#endif
+
 #else // KMP_USE_DYNAMIC_LOCK
 
   kmp_user_lock_p lck;
 
   /* Can't use serial interval since not block structured */
 
   if ((__kmp_user_lock_kind == lk_tas) &&
       (sizeof(lck->tas.lk.poll) + sizeof(lck->tas.lk.depth_locked) <=
        OMP_NEST_LOCK_T_SIZE)) {
 #if KMP_OS_LINUX &&                                                            \
     (KMP_ARCH_X86 || KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64)
     // "fast" path implemented to fix customer performance issue
     kmp_tas_lock_t *tl = (kmp_tas_lock_t *)user_lock;
 #if USE_ITT_BUILD
     __kmp_itt_lock_releasing((kmp_user_lock_p)user_lock);
 #endif /* USE_ITT_BUILD */
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    int release_status = KMP_LOCK_STILL_HELD;
+#endif
+
     if (--(tl->lk.depth_locked) == 0) {
       TCW_4(tl->lk.poll, 0);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+      release_status = KMP_LOCK_RELEASED;
+#endif
     }
     KMP_MB();
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    // This is the case, if called from omp_init_lock_with_hint:
+    void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (!codeptr)
+      codeptr = OMPT_GET_RETURN_ADDRESS(0);
+    if (ompt_enabled.enabled) {
+      if (release_status == KMP_LOCK_RELEASED) {
+        if (ompt_enabled.ompt_callback_mutex_released) {
+          // release_lock_last
+          ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+              ompt_mutex_nest_lock, (ompt_wait_id_t)lck, codeptr);
+        }
+      } else if (ompt_enabled.ompt_callback_nest_lock) {
+        // release_lock_previous
+        ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+            ompt_mutex_scope_end, (ompt_wait_id_t)lck, codeptr);
+      }
+    }
+#endif
+
     return;
 #else
     lck = (kmp_user_lock_p)user_lock;
 #endif
   }
 #if KMP_USE_FUTEX
   else if ((__kmp_user_lock_kind == lk_futex) &&
            (sizeof(lck->futex.lk.poll) + sizeof(lck->futex.lk.depth_locked) <=
             OMP_NEST_LOCK_T_SIZE)) {
     lck = (kmp_user_lock_p)user_lock;
   }
 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_unset_nest_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_releasing(lck);
 #endif /* USE_ITT_BUILD */
 
   int release_status;
   release_status = RELEASE_NESTED_LOCK(lck, gtid);
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.enabled) {
     if (release_status == KMP_LOCK_RELEASED) {
-      if (ompt_callbacks.ompt_callback(ompt_event_release_nest_lock_last)) {
-        ompt_callbacks.ompt_callback(ompt_event_release_nest_lock_last)(
-            (uint64_t)lck);
-      }
-    } else if (ompt_callbacks.ompt_callback(
-                   ompt_event_release_nest_lock_prev)) {
-      ompt_callbacks.ompt_callback(ompt_event_release_nest_lock_prev)(
-          (uint64_t)lck);
+      if (ompt_enabled.ompt_callback_mutex_released) {
+        // release_lock_last
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)lck, codeptr);
+      }
+    } else if (ompt_enabled.ompt_callback_nest_lock) {
+      // release_lock_previous
+      ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+          ompt_mutex_scope_end, (ompt_wait_id_t)lck, codeptr);
     }
   }
 #endif
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 /* try to acquire the lock */
 int __kmpc_test_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
   KMP_COUNT_BLOCK(OMP_test_lock);
 
 #if KMP_USE_DYNAMIC_LOCK
   int rc;
   int tag = KMP_EXTRACT_D_TAG(user_lock);
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring((kmp_user_lock_p)user_lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_lock, omp_lock_hint_none,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
 #if KMP_USE_INLINED_TAS
   if (tag == locktag_tas && !__kmp_env_consistency_check) {
     KMP_TEST_TAS_LOCK(user_lock, gtid, rc);
   } else
 #elif KMP_USE_INLINED_FUTEX
   if (tag == locktag_futex && !__kmp_env_consistency_check) {
     KMP_TEST_FUTEX_LOCK(user_lock, gtid, rc);
   } else
 #endif
   {
     rc = __kmp_direct_test[tag]((kmp_dyna_lock_t *)user_lock, gtid);
   }
   if (rc) {
 #if USE_ITT_BUILD
     __kmp_itt_lock_acquired((kmp_user_lock_p)user_lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.ompt_callback_mutex_acquired) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+          ompt_mutex_lock, (ompt_wait_id_t)user_lock, codeptr);
+    }
+#endif
     return FTN_TRUE;
   } else {
 #if USE_ITT_BUILD
     __kmp_itt_lock_cancelled((kmp_user_lock_p)user_lock);
 #endif
     return FTN_FALSE;
   }
 

 #endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_test_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring(lck);
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(),
+        (ompt_wait_id_t)lck, codeptr);
+  }
+#endif
 
   rc = TEST_LOCK(lck, gtid);
 #if USE_ITT_BUILD
   if (rc) {
     __kmp_itt_lock_acquired(lck);
   } else {
     __kmp_itt_lock_cancelled(lck);
   }
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (rc && ompt_enabled.ompt_callback_mutex_acquired) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+        ompt_mutex_lock, (ompt_wait_id_t)lck, codeptr);
+  }
+#endif
+
   return (rc ? FTN_TRUE : FTN_FALSE);
 
 /* Can't use serial interval since not block structured */
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 /* try to acquire the lock */
 int __kmpc_test_nest_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) {
 #if KMP_USE_DYNAMIC_LOCK
   int rc;
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring((kmp_user_lock_p)user_lock);
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.ompt_callback_mutex_acquire) {
+    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+        ompt_mutex_nest_lock, omp_lock_hint_none,
+        __ompt_get_mutex_impl_type(user_lock), (ompt_wait_id_t)user_lock,
+        codeptr);
+  }
+#endif
   rc = KMP_D_LOCK_FUNC(user_lock, test)((kmp_dyna_lock_t *)user_lock, gtid);
 #if USE_ITT_BUILD
   if (rc) {
     __kmp_itt_lock_acquired((kmp_user_lock_p)user_lock);
   } else {
     __kmp_itt_lock_cancelled((kmp_user_lock_p)user_lock);
   }
 #endif
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled && rc) {
+    if (rc == 1) {
+      if (ompt_enabled.ompt_callback_mutex_acquired) {
+        // lock_first
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)user_lock, codeptr);
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_nest_lock) {
+        // lock_next
+        ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+            ompt_scope_begin, (ompt_wait_id_t)user_lock, codeptr);
+      }
+    }
+  }
+#endif
   return rc;
 
 #else // KMP_USE_DYNAMIC_LOCK
 
   kmp_user_lock_p lck;
   int rc;
 
   if ((__kmp_user_lock_kind == lk_tas) &&
#endif
   else {
     lck = __kmp_lookup_user_lock(user_lock, "omp_test_nest_lock");
   }
 
 #if USE_ITT_BUILD
   __kmp_itt_lock_acquiring(lck);
 #endif /* USE_ITT_BUILD */
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  // This is the case, if called from omp_init_lock_with_hint:
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+  if (!codeptr)
+    codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  if (ompt_enabled.enabled) &&
+        ompt_enabled.ompt_callback_mutex_acquire) {
+      ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
+          ompt_mutex_nest_lock, omp_lock_hint_none,
+          __ompt_get_mutex_impl_type(), (ompt_wait_id_t)lck, codeptr);
+    }
+#endif
+
   rc = TEST_NESTED_LOCK(lck, gtid);
 #if USE_ITT_BUILD
   if (rc) {
     __kmp_itt_lock_acquired(lck);
   } else {
     __kmp_itt_lock_cancelled(lck);
   }
 #endif /* USE_ITT_BUILD */
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled && rc) {
+    if (rc == 1) {
+      if (ompt_enabled.ompt_callback_mutex_acquired) {
+        // lock_first
+        ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
+            ompt_mutex_nest_lock, (ompt_wait_id_t)lck, codeptr);
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_nest_lock) {
+        // lock_next
+        ompt_callbacks.ompt_callback(ompt_callback_nest_lock)(
+            ompt_mutex_scope_begin, (ompt_wait_id_t)lck, codeptr);
+      }
+    }
+  }
+#endif
   return rc;
 
 /* Can't use serial interval since not block structured */
 
 #endif // KMP_USE_DYNAMIC_LOCK
 }
 
 // Interface to fast scalable reduce methods routines

 // be confusing to a customer)
 // AT: another implementation of *barrier_gather*nowait() (or some other design)
 // might go faster and be more in line with sense of NOWAIT
 // AT: TO DO: do epcc test and compare times
 
 // this barrier should be invisible to a customer and to the threading profile
 // tool (it's neither a terminating barrier nor customer's code, it's
 // used for an internal purpose)
+#if OMPT_SUPPORT
+    // JP: can this barrier potentially leed to task scheduling?
+    // JP: as long as there is a barrier in the implementation, OMPT should and
+    // will provide the barrier events
+    //         so we set-up the necessary frame/return addresses.
+    ompt_frame_t *ompt_frame;
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+      if (ompt_frame->reenter_runtime_frame == NULL)
+        ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(global_tid);
+    }
+#endif
 #if USE_ITT_NOTIFY
     __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
     retval =
         __kmp_barrier(UNPACK_REDUCTION_BARRIER(packed_reduction_method),
                       global_tid, FALSE, reduce_size, reduce_data, reduce_func);
     retval = (retval != 0) ? (0) : (1);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ompt_frame->reenter_runtime_frame = NULL;
+    }
+#endif
 
     // all other workers except master should do this pop here
     //     ( none of other workers will get to __kmpc_end_reduce_nowait() )
     if (__kmp_env_consistency_check) {
       if (retval == 0) {
         __kmp_pop_sync(global_tid, ct_reduce, loc);
       }
     }
if (packed_reduction_method == critical_reduce_block) {
     retval = 2;
 
   } else if (TEST_REDUCTION_METHOD(packed_reduction_method,
                                    tree_reduce_block)) {
 
 // case tree_reduce_block:
 // this barrier should be visible to a customer and to the threading profile
 // tool (it's a terminating barrier on constructs if NOWAIT not specified)
+#if OMPT_SUPPORT
+    ompt_frame_t *ompt_frame;
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+      if (ompt_frame->reenter_runtime_frame == NULL)
+        ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(global_tid);
+    }
+#endif
 #if USE_ITT_NOTIFY
     __kmp_threads[global_tid]->th.th_ident =
         loc; // needed for correct notification of frames
 #endif
     retval =
         __kmp_barrier(UNPACK_REDUCTION_BARRIER(packed_reduction_method),
                       global_tid, TRUE, reduce_size, reduce_data, reduce_func);
     retval = (retval != 0) ? (0) : (1);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ompt_frame->reenter_runtime_frame = NULL;
+    }
+#endif
 
     // all other workers except master should do this pop here
     // ( none of other workers except master will enter __kmpc_end_reduce() )
     if (__kmp_env_consistency_check) {
       if (retval == 0) { // 0: all other workers; 1: master
         __kmp_pop_sync(global_tid, ct_reduce, loc);
       }
     }
void __kmpc_end_reduce(ident_t *loc, kmp_int32 global_tid,
   // this barrier should be visible to a customer and to the threading profile
   // tool (it's a terminating barrier on constructs if NOWAIT not specified)
 
   if (packed_reduction_method == critical_reduce_block) {
 
     __kmp_end_critical_section_reduce_block(loc, global_tid, lck);
 
 // TODO: implicit barrier: should be exposed
+#if OMPT_SUPPORT
+    ompt_frame_t *ompt_frame;
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+      if (ompt_frame->reenter_runtime_frame == NULL)
+        ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(global_tid);
+    }
+#endif
 #if USE_ITT_NOTIFY
     __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
     __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ompt_frame->reenter_runtime_frame = NULL;
+    }
+#endif
 
   } else if (packed_reduction_method == empty_reduce_block) {
 
 // usage: if team size==1, no synchronization is required (Intel platforms only)
 
 // TODO: implicit barrier: should be exposed
+#if OMPT_SUPPORT
+    ompt_frame_t *ompt_frame;
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+      if (ompt_frame->reenter_runtime_frame == NULL)
+        ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(global_tid);
+    }
+#endif
 #if USE_ITT_NOTIFY
     __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
     __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ompt_frame->reenter_runtime_frame = NULL;
+    }
+#endif
 
   } else if (packed_reduction_method == atomic_reduce_block) {
 
+#if OMPT_SUPPORT
+    ompt_frame_t *ompt_frame;
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+      if (ompt_frame->reenter_runtime_frame == NULL)
+        ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+      OMPT_STORE_RETURN_ADDRESS(global_tid);
+    }
+#endif
 // TODO: implicit barrier: should be exposed
 #if USE_ITT_NOTIFY
     __kmp_threads[global_tid]->th.th_ident = loc;
 #endif
     __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.enabled) {
+      ompt_frame->reenter_runtime_frame = NULL;
+    }
+#endif
 
   } else if (TEST_REDUCTION_METHOD(packed_reduction_method,
                                    tree_reduce_block)) {
 
     // only master executes here (master releases all other workers)
     __kmp_end_split_barrier(UNPACK_REDUCTION_BARRIER(packed_reduction_method),
                             global_tid);
 

runtime/src/kmp_dispatch.cpp

if (schedule == kmp_sch_static_steal) {
     // Other threads will inspect this variable when searching for a victim.
     // This is a flag showing that other threads may steal from this thread
     // since then.
     volatile T *p = &pr->u.p.static_steal_counter;
     *p = *p + 1;
   }
 #endif // ( KMP_STATIC_STEAL_ENABLED )
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_begin)) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_work) {
     ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-    ompt_callbacks.ompt_callback(ompt_event_loop_begin)(
-        team_info->parallel_id, task_info->task_id, team_info->microtask);
+    ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
+    kmp_info_t *thr = __kmp_threads[gtid];
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_loop, ompt_scope_begin, &(team_info->parallel_data),
+        &(task_info->task_data),
+        tc, 
+        OMPT_LOAD_RETURN_ADDRESS(gtid));
   }
 #endif
 }
 
 /* For ordered loops, either __kmp_dispatch_finish() should be called after
  * every iteration, or __kmp_dispatch_finish_chunk() should be called after
  * every chunk of iterations.  If the ordered section(s) were not executed
  * for this iteration (or every iteration in this chunk), we need to set the
#endif
   KD_TRACE(100, ("__kmp_dispatch_finish_chunk: T#%d returned\n", gtid));
 }
 
 #endif /* KMP_GOMP_COMPAT */
 
 /* Define a macro for exiting __kmp_dispatch_next(). If status is 0 (no more
    work), then tell OMPT the loop is over. In some cases kmp_dispatch_fini()
    is not called. */
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
 #define OMPT_LOOP_END                                                          \
   if (status == 0) {                                                           \
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_end)) {   \
+    if (ompt_enabled.ompt_callback_work) {                                     \
       ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);              \
-      ompt_task_info_t *task_info = __ompt_get_taskinfo(0);                    \
-      ompt_callbacks.ompt_callback(ompt_event_loop_end)(                       \
-          team_info->parallel_id, task_info->task_id);                         \
+      ompt_task_info_t *task_info = __ompt_get_task_info_object(0);            \
+      ompt_callbacks.ompt_callback(ompt_callback_work)(                        \
+          ompt_work_loop, ompt_scope_end, &(team_info->parallel_data),         \
+          &(task_info->task_data), 0, codeptr);                                \
     }                                                                          \
   }
+// TODO: implement count
 #else
 #define OMPT_LOOP_END // no-op
 #endif
 
 template <typename T>
 static int __kmp_dispatch_next(ident_t *loc, int gtid, kmp_int32 *p_last,
                                T *p_lb, T *p_ub,
-                               typename traits_t<T>::signed_t *p_st) {
+                               typename traits_t<T>::signed_t *p_st
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                               ,
+                               void *codeptr
+#endif
+                               ) {
 
   typedef typename traits_t<T>::unsigned_t UT;
   typedef typename traits_t<T>::signed_t ST;
   typedef typename traits_t<T>::floating_t DBL;
 
   // This is potentially slightly misleading, schedule(runtime) will appear here
   // even if the actual runtme schedule is static. (Which points out a
   // disadavantage of schedule(runtime): even when static scheduling is used it

 saving the loop arguments.
 These functions are all identical apart from the types of the arguments.
 */
 
 void __kmpc_dispatch_init_4(ident_t *loc, kmp_int32 gtid,
                             enum sched_type schedule, kmp_int32 lb,
                             kmp_int32 ub, kmp_int32 st, kmp_int32 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dispatch_init<kmp_int32>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 /*!
 See @ref __kmpc_dispatch_init_4
 */
 void __kmpc_dispatch_init_4u(ident_t *loc, kmp_int32 gtid,
                              enum sched_type schedule, kmp_uint32 lb,
                              kmp_uint32 ub, kmp_int32 st, kmp_int32 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dispatch_init<kmp_uint32>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 /*!
 See @ref __kmpc_dispatch_init_4
 */
 void __kmpc_dispatch_init_8(ident_t *loc, kmp_int32 gtid,
                             enum sched_type schedule, kmp_int64 lb,
                             kmp_int64 ub, kmp_int64 st, kmp_int64 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dispatch_init<kmp_int64>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 /*!
 See @ref __kmpc_dispatch_init_4
 */
 void __kmpc_dispatch_init_8u(ident_t *loc, kmp_int32 gtid,
                              enum sched_type schedule, kmp_uint64 lb,
                              kmp_uint64 ub, kmp_int64 st, kmp_int64 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dispatch_init<kmp_uint64>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 /*!
 See @ref __kmpc_dispatch_init_4
 
 Difference from __kmpc_dispatch_init set of functions is these functions
 are called for composite distribute parallel for construct. Thus before
 regular iterations dispatching we need to calc per-team iteration space.
 
 These functions are all identical apart from the types of the arguments.
 */
 void __kmpc_dist_dispatch_init_4(ident_t *loc, kmp_int32 gtid,
                                  enum sched_type schedule, kmp_int32 *p_last,
                                  kmp_int32 lb, kmp_int32 ub, kmp_int32 st,
                                  kmp_int32 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dist_get_bounds<kmp_int32>(loc, gtid, p_last, &lb, &ub, st);
   __kmp_dispatch_init<kmp_int32>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 void __kmpc_dist_dispatch_init_4u(ident_t *loc, kmp_int32 gtid,
                                   enum sched_type schedule, kmp_int32 *p_last,
                                   kmp_uint32 lb, kmp_uint32 ub, kmp_int32 st,
                                   kmp_int32 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dist_get_bounds<kmp_uint32>(loc, gtid, p_last, &lb, &ub, st);
   __kmp_dispatch_init<kmp_uint32>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 void __kmpc_dist_dispatch_init_8(ident_t *loc, kmp_int32 gtid,
                                  enum sched_type schedule, kmp_int32 *p_last,
                                  kmp_int64 lb, kmp_int64 ub, kmp_int64 st,
                                  kmp_int64 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dist_get_bounds<kmp_int64>(loc, gtid, p_last, &lb, &ub, st);
   __kmp_dispatch_init<kmp_int64>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 void __kmpc_dist_dispatch_init_8u(ident_t *loc, kmp_int32 gtid,
                                   enum sched_type schedule, kmp_int32 *p_last,
                                   kmp_uint64 lb, kmp_uint64 ub, kmp_int64 st,
                                   kmp_int64 chunk) {
   KMP_DEBUG_ASSERT(__kmp_init_serial);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmp_dist_get_bounds<kmp_uint64>(loc, gtid, p_last, &lb, &ub, st);
   __kmp_dispatch_init<kmp_uint64>(loc, gtid, schedule, lb, ub, st, chunk, true);
 }
 
 /*!
 @param loc Source code location
 @param gtid Global thread id
 @param p_last Pointer to a flag set to one if this is the last chunk or zero
 otherwise
 @param p_lb   Pointer to the lower bound for the next chunk of work
 @param p_ub   Pointer to the upper bound for the next chunk of work
 @param p_st   Pointer to the stride for the next chunk of work
 @return one if there is work to be done, zero otherwise
 
 Get the next dynamically allocated chunk of work for this thread.
 If there is no more work, then the lb,ub and stride need not be modified.
 */
 int __kmpc_dispatch_next_4(ident_t *loc, kmp_int32 gtid, kmp_int32 *p_last,
                            kmp_int32 *p_lb, kmp_int32 *p_ub, kmp_int32 *p_st) {
-  return __kmp_dispatch_next<kmp_int32>(loc, gtid, p_last, p_lb, p_ub, p_st);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmp_dispatch_next<kmp_int32>(loc, gtid, p_last, p_lb, p_ub, p_st
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                        ,
+                                        OMPT_LOAD_RETURN_ADDRESS(gtid)
+#endif
+                                            );
 }
 
 /*!
 See @ref __kmpc_dispatch_next_4
 */
 int __kmpc_dispatch_next_4u(ident_t *loc, kmp_int32 gtid, kmp_int32 *p_last,
                             kmp_uint32 *p_lb, kmp_uint32 *p_ub,
                             kmp_int32 *p_st) {
-  return __kmp_dispatch_next<kmp_uint32>(loc, gtid, p_last, p_lb, p_ub, p_st);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmp_dispatch_next<kmp_uint32>(loc, gtid, p_last, p_lb, p_ub, p_st
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                         ,
+                                         OMPT_LOAD_RETURN_ADDRESS(gtid)
+#endif
+                                             );
 }
 
 /*!
 See @ref __kmpc_dispatch_next_4
 */
 int __kmpc_dispatch_next_8(ident_t *loc, kmp_int32 gtid, kmp_int32 *p_last,
                            kmp_int64 *p_lb, kmp_int64 *p_ub, kmp_int64 *p_st) {
-  return __kmp_dispatch_next<kmp_int64>(loc, gtid, p_last, p_lb, p_ub, p_st);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmp_dispatch_next<kmp_int64>(loc, gtid, p_last, p_lb, p_ub, p_st
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                        ,
+                                        OMPT_LOAD_RETURN_ADDRESS(gtid)
+#endif
+                                            );
 }
 
 /*!
 See @ref __kmpc_dispatch_next_4
 */
 int __kmpc_dispatch_next_8u(ident_t *loc, kmp_int32 gtid, kmp_int32 *p_last,
                             kmp_uint64 *p_lb, kmp_uint64 *p_ub,
                             kmp_int64 *p_st) {
-  return __kmp_dispatch_next<kmp_uint64>(loc, gtid, p_last, p_lb, p_ub, p_st);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmp_dispatch_next<kmp_uint64>(loc, gtid, p_last, p_lb, p_ub, p_st
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                         ,
+                                         OMPT_LOAD_RETURN_ADDRESS(gtid)
+#endif
+                                             );
 }
 
 /*!
 @param loc Source code location
 @param gtid Global thread id
 
 Mark the end of a dynamic loop.
 */

runtime/src/kmp_ftn_entry.h

 #endif
 
 #ifdef KMP_STUB
 #include "kmp_stub.h"
 #endif
 
 #include "kmp_i18n.h"
 
+#if OMPT_SUPPORT
+#include "ompt-specific.h"
+#endif
+
 #ifdef __cplusplus
 extern "C" {
 #endif // __cplusplus
 
 /* For compatibility with the Gnu/MS Open MP codegen, omp_set_num_threads(),
  * omp_set_nested(), and omp_set_dynamic() [in lowercase on MS, and w/o
  * a trailing underscore on Linux* OS] take call by value integer arguments.
  * + omp_set_max_active_levels()
#else
   gtid = __kmp_entry_gtid();
   thread = __kmp_threads[gtid];
   // return thread -> th.th_team -> t.t_current_task[
   // thread->th.th_info.ds.ds_tid ] -> icvs.nproc;
   return thread->th.th_current_task->td_icvs.nproc;
 #endif
 }
 
+#if OMP_50_ENABLED
+int FTN_STDCALL FTN_CONTROL_TOOL(uint64_t command, uint64_t modifier,
+                                 void *arg) {
+#if defined(KMP_STUB) || !OMPT_SUPPORT
+  return -2;
+#else
+  OMPT_STORE_RETURN_ADDRESS(__kmp_entry_gtid());
+  if (!TCR_4(__kmp_init_middle)) {
+    return -2;
+  }
+  kmp_info_t *this_thr = __kmp_threads[__kmp_entry_gtid()];
+  ompt_task_info_t *parent_task_info = OMPT_CUR_TASK_INFO(this_thr);
+  parent_task_info->frame.reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+  int ret = __kmp_control_tool(command, modifier, arg);
+  parent_task_info->frame.reenter_runtime_frame = 0;
+  return ret;
+#endif
+}
+#endif
+
 int FTN_STDCALL xexpand(FTN_GET_THREAD_NUM)(void) {
 #ifdef KMP_STUB
   return 0;
 #else
   int gtid;
 
 #if KMP_OS_DARWIN || KMP_OS_FREEBSD || KMP_OS_NETBSD
   gtid = __kmp_entry_gtid();

 #endif /* KMP_STUB */
 
 #if KMP_USE_DYNAMIC_LOCK
 void FTN_STDCALL FTN_INIT_LOCK_WITH_HINT(void **user_lock,
                                          uintptr_t KMP_DEREF hint) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNLOCKED;
 #else
-  __kmpc_init_lock_with_hint(NULL, __kmp_entry_gtid(), user_lock,
-                             KMP_DEREF hint);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_init_lock_with_hint(NULL, gtid, user_lock, KMP_DEREF hint);
 #endif
 }
 
 void FTN_STDCALL FTN_INIT_NEST_LOCK_WITH_HINT(void **user_lock,
                                               uintptr_t KMP_DEREF hint) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNLOCKED;
 #else
-  __kmpc_init_nest_lock_with_hint(NULL, __kmp_entry_gtid(), user_lock,
-                                  KMP_DEREF hint);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_init_nest_lock_with_hint(NULL, gtid, user_lock, KMP_DEREF hint);
 #endif
 }
 #endif
 
 /* initialize the lock */
 void FTN_STDCALL xexpand(FTN_INIT_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNLOCKED;
 #else
-  __kmpc_init_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_init_lock(NULL, gtid, user_lock);
 #endif
 }
 
 /* initialize the lock */
 void FTN_STDCALL xexpand(FTN_INIT_NEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNLOCKED;
 #else
-  __kmpc_init_nest_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_init_nest_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_DESTROY_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNINIT;
 #else
-  __kmpc_destroy_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_destroy_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_DESTROY_NEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   *((kmp_stub_lock_t *)user_lock) = UNINIT;
 #else
-  __kmpc_destroy_nest_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_destroy_nest_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_SET_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   if (*((kmp_stub_lock_t *)user_lock) != UNLOCKED) {
     // TODO: Issue an error.
   }
   *((kmp_stub_lock_t *)user_lock) = LOCKED;
 #else
-  __kmpc_set_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_set_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_SET_NEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   (*((int *)user_lock))++;
 #else
-  __kmpc_set_nest_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_set_nest_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_UNSET_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   if (*((kmp_stub_lock_t *)user_lock) == UNLOCKED) {
     // TODO: Issue an error.
   }
   *((kmp_stub_lock_t *)user_lock) = UNLOCKED;
 #else
-  __kmpc_unset_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_unset_lock(NULL, gtid, user_lock);
 #endif
 }
 
 void FTN_STDCALL xexpand(FTN_UNSET_NEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   if (*((kmp_stub_lock_t *)user_lock) == UNLOCKED) {
     // TODO: Issue an error.
   }
   (*((int *)user_lock))--;
 #else
-  __kmpc_unset_nest_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  __kmpc_unset_nest_lock(NULL, gtid, user_lock);
 #endif
 }
 
 int FTN_STDCALL xexpand(FTN_TEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   if (*((kmp_stub_lock_t *)user_lock) == LOCKED) {
     return 0;
   }
   *((kmp_stub_lock_t *)user_lock) = LOCKED;
   return 1;
 #else
-  return __kmpc_test_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmpc_test_lock(NULL, gtid, user_lock);
 #endif
 }
 
 int FTN_STDCALL xexpand(FTN_TEST_NEST_LOCK)(void **user_lock) {
 #ifdef KMP_STUB
   if (*((kmp_stub_lock_t *)user_lock) == UNINIT) {
     // TODO: Issue an error.
   }
   return ++(*((int *)user_lock));
 #else
-  return __kmpc_test_nest_lock(NULL, __kmp_entry_gtid(), user_lock);
+  int gtid = __kmp_entry_gtid();
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+  return __kmpc_test_nest_lock(NULL, gtid, user_lock);
 #endif
 }
 
 double FTN_STDCALL xexpand(FTN_GET_WTIME)(void) {
 #ifdef KMP_STUB
   return __kmps_get_wtime();
 #else
   double data;

 #endif /* OMP_40_ENABLED */
 
 #if OMP_45_ENABLED
 // OMP_4.5 aliases
 #endif
 
 #if OMP_50_ENABLED
 // OMP_5.0 aliases
 #endif
 
 // OMP_1.0 versioned symbols
 xversionify(FTN_SET_NUM_THREADS, 10, "OMP_1.0");
 xversionify(FTN_GET_NUM_THREADS, 10, "OMP_1.0");
 xversionify(FTN_GET_MAX_THREADS, 10, "OMP_1.0");
 xversionify(FTN_GET_THREAD_NUM, 10, "OMP_1.0");
 xversionify(FTN_GET_NUM_PROCS, 10, "OMP_1.0");
 xversionify(FTN_IN_PARALLEL, 10, "OMP_1.0");

 #endif /* OMP_40_ENABLED */
 
 #if OMP_45_ENABLED
 // OMP_4.5 versioned symbols
 #endif
 
 #if OMP_50_ENABLED
 // OMP_5.0 versioned symbols
 #endif
 
 #endif // KMP_USE_VERSION_SYMBOLS
 
 #ifdef __cplusplus
 } // extern "C"
 #endif // __cplusplus
 
 // end of file //

runtime/src/kmp_ftn_os.h

 #define FTN_MALLOC kmp_malloc
 #define FTN_ALIGNED_MALLOC kmp_aligned_malloc
 #define FTN_CALLOC kmp_calloc
 #define FTN_REALLOC kmp_realloc
 #define FTN_FREE kmp_free
 
 #define FTN_GET_NUM_KNOWN_THREADS kmp_get_num_known_threads
 
 #define FTN_SET_NUM_THREADS omp_set_num_threads
 #define FTN_GET_NUM_THREADS omp_get_num_threads
 #define FTN_GET_MAX_THREADS omp_get_max_threads
 #define FTN_GET_THREAD_NUM omp_get_thread_num
 #define FTN_GET_NUM_PROCS omp_get_num_procs
 #define FTN_SET_DYNAMIC omp_set_dynamic
 #define FTN_GET_DYNAMIC omp_get_dynamic
 #define FTN_SET_NESTED omp_set_nested
 #define FTN_GET_NESTED omp_get_nested

 #define FTN_TARGET_IS_PRESENT omp_target_is_present
 #define FTN_TARGET_MEMCPY omp_target_memcpy
 #define FTN_TARGET_MEMCPY_RECT omp_target_memcpy_rect
 #define FTN_TARGET_ASSOCIATE_PTR omp_target_associate_ptr
 #define FTN_TARGET_DISASSOCIATE_PTR omp_target_disassociate_ptr
 #endif
 #endif
 
+#if OMP_50_ENABLED
+#define FTN_CONTROL_TOOL omp_control_tool
+#endif
+
 #endif /* KMP_FTN_PLAIN */
 
 /* ------------------------------------------------------------------------ */
 
 #if KMP_FTN_ENTRIES == KMP_FTN_APPEND
 
 #define FTN_SET_STACKSIZE kmp_set_stacksize_
 #define FTN_SET_STACKSIZE_S kmp_set_stacksize_s_

 #define FTN_TARGET_IS_PRESENT omp_target_is_present_
 #define FTN_TARGET_MEMCPY omp_target_memcpy_
 #define FTN_TARGET_MEMCPY_RECT omp_target_memcpy_rect_
 #define FTN_TARGET_ASSOCIATE_PTR omp_target_associate_ptr_
 #define FTN_TARGET_DISASSOCIATE_PTR omp_target_disassociate_ptr_
 #endif
 #endif
 
+#if OMP_50_ENABLED
+#define FTN_CONTROL_TOOL OMP_CONTROL_TOOL
+#endif
+
 #endif /* KMP_FTN_APPEND */
 
 /* ------------------------------------------------------------------------ */
 
 #if KMP_FTN_ENTRIES == KMP_FTN_UPPER
 
 #define FTN_SET_STACKSIZE KMP_SET_STACKSIZE
 #define FTN_SET_STACKSIZE_S KMP_SET_STACKSIZE_S

 #define FTN_TARGET_IS_PRESENT OMP_TARGET_IS_PRESENT
 #define FTN_TARGET_MEMCPY OMP_TARGET_MEMCPY
 #define FTN_TARGET_MEMCPY_RECT OMP_TARGET_MEMCPY_RECT
 #define FTN_TARGET_ASSOCIATE_PTR OMP_TARGET_ASSOCIATE_PTR
 #define FTN_TARGET_DISASSOCIATE_PTR OMP_TARGET_DISASSOCIATE_PTR
 #endif
 #endif
 
+#if OMP_50_ENABLED
+#define FTN_CONTROL_TOOL OMP_CONTROL_TOOL
+#endif
+
 #endif /* KMP_FTN_UPPER */
 
 /* ------------------------------------------------------------------------ */
 
 #if KMP_FTN_ENTRIES == KMP_FTN_UAPPEND
 
 #define FTN_SET_STACKSIZE KMP_SET_STACKSIZE_
 #define FTN_SET_STACKSIZE_S KMP_SET_STACKSIZE_S_

 #define FTN_MALLOC KMP_MALLOC_
 #define FTN_ALIGNED_MALLOC KMP_ALIGNED_MALLOC_
 #define FTN_CALLOC KMP_CALLOC_
 #define FTN_REALLOC KMP_REALLOC_
 #define FTN_FREE KMP_FREE_
 
 #define FTN_GET_NUM_KNOWN_THREADS KMP_GET_NUM_KNOWN_THREADS_
 
 #define FTN_SET_NUM_THREADS OMP_SET_NUM_THREADS_
 #define FTN_GET_NUM_THREADS OMP_GET_NUM_THREADS_
 #define FTN_GET_MAX_THREADS OMP_GET_MAX_THREADS_
 #define FTN_GET_THREAD_NUM OMP_GET_THREAD_NUM_
 #define FTN_GET_NUM_PROCS OMP_GET_NUM_PROCS_
 #define FTN_SET_DYNAMIC OMP_SET_DYNAMIC_
 #define FTN_GET_DYNAMIC OMP_GET_DYNAMIC_
 #define FTN_SET_NESTED OMP_SET_NESTED_
 #define FTN_GET_NESTED OMP_GET_NESTED_
 #define FTN_IN_PARALLEL OMP_IN_PARALLEL_
 #define FTN_GET_THREAD_LIMIT OMP_GET_THREAD_LIMIT_

 #define FTN_TARGET_IS_PRESENT OMP_TARGET_IS_PRESENT_
 #define FTN_TARGET_MEMCPY OMP_TARGET_MEMCPY_
 #define FTN_TARGET_MEMCPY_RECT OMP_TARGET_MEMCPY_RECT_
 #define FTN_TARGET_ASSOCIATE_PTR OMP_TARGET_ASSOCIATE_PTR_
 #define FTN_TARGET_DISASSOCIATE_PTR OMP_TARGET_DISASSOCIATE_PTR_
 #endif
 #endif
 
+#if OMP_50_ENABLED
+#define FTN_CONTROL_TOOL OMP_CONTROL_TOOL_
+#endif
+
 #endif /* KMP_FTN_UAPPEND */
 
 /* -------------------------- GOMP API NAMES ------------------------ */
 // All GOMP_1.0 symbols
 #define KMP_API_NAME_GOMP_ATOMIC_END GOMP_atomic_end
 #define KMP_API_NAME_GOMP_ATOMIC_START GOMP_atomic_start
 #define KMP_API_NAME_GOMP_BARRIER GOMP_barrier
 #define KMP_API_NAME_GOMP_CRITICAL_END GOMP_critical_end

runtime/src/kmp_global.cpp

 #endif
 
 kmp_tasking_mode_t __kmp_tasking_mode = tskm_task_teams;
 #if OMP_45_ENABLED
 kmp_int32 __kmp_max_task_priority = 0;
 kmp_uint64 __kmp_taskloop_min_tasks = 0;
 #endif
 
+#if OMP_50_ENABLED && OMPT_SUPPORT
+char const *__kmp_tool_libraries = NULL;
+#endif
+
 /* This check ensures that the compiler is passing the correct data type for the
    flags formal parameter of the function kmpc_omp_task_alloc(). If the type is
    not a 4-byte type, then give an error message about a non-positive length
    array pointing here.  If that happens, the kmp_tasking_flags_t structure must
    be redefined to have exactly 32 bits. */
 KMP_BUILD_ASSERT(sizeof(kmp_tasking_flags_t) == 4);
 
 kmp_int32 __kmp_task_stealing_constraint =

runtime/src/kmp_gsupport.cpp

#define MKLOC(loc, routine)                                                    \
   static ident_t(loc) = {0, KMP_IDENT_KMPC, 0, 0, ";unknown;unknown;0;0;;"};
 
 #include "kmp_ftn_os.h"
 
 void xexpand(KMP_API_NAME_GOMP_BARRIER)(void) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_barrier");
   KA_TRACE(20, ("GOMP_barrier: T#%d\n", gtid));
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   ompt_frame_t *ompt_frame;
-  if (ompt_enabled) {
-    ompt_frame = __ompt_get_task_frame_internal(0);
-    ompt_frame->reenter_runtime_frame = __builtin_frame_address(1);
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
   }
 #endif
   __kmpc_barrier(&loc, gtid);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 }
 
 // Mutual exclusion
 
 // The symbol that icc/ifort generates for unnamed for unnamed critical sections
 // - .gomp_critical_user_ - is defined using .comm in any objects reference it.
 // We can't reference it directly here in C code, as the symbol contains a ".".
 //
 // The RTL contains an assembly language definition of .gomp_critical_user_
 // with another symbol __kmp_unnamed_critical_addr initialized with it's
 // address.
 extern kmp_critical_name *__kmp_unnamed_critical_addr;
 
 void xexpand(KMP_API_NAME_GOMP_CRITICAL_START)(void) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_critical_start");
   KA_TRACE(20, ("GOMP_critical_start: T#%d\n", gtid));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmpc_critical(&loc, gtid, __kmp_unnamed_critical_addr);
 }
 
 void xexpand(KMP_API_NAME_GOMP_CRITICAL_END)(void) {
   int gtid = __kmp_get_gtid();
   MKLOC(loc, "GOMP_critical_end");
   KA_TRACE(20, ("GOMP_critical_end: T#%d\n", gtid));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmpc_end_critical(&loc, gtid, __kmp_unnamed_critical_addr);
 }
 
 void xexpand(KMP_API_NAME_GOMP_CRITICAL_NAME_START)(void **pptr) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_critical_name_start");
   KA_TRACE(20, ("GOMP_critical_name_start: T#%d\n", gtid));
   __kmpc_critical(&loc, gtid, (kmp_critical_name *)pptr);
int xexpand(KMP_API_NAME_GOMP_SINGLE_START)(void) {
   KA_TRACE(20, ("GOMP_single_start: T#%d\n", gtid));
 
   if (!TCR_4(__kmp_init_parallel))
     __kmp_parallel_initialize();
 
   // 3rd parameter == FALSE prevents kmp_enter_single from pushing a
   // workshare when USE_CHECKS is defined.  We need to avoid the push,
   // as there is no corresponding GOMP_single_end() call.
-  return __kmp_enter_single(gtid, &loc, FALSE);
+  kmp_int32 rc = __kmp_enter_single(gtid, &loc, FALSE);
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  kmp_info_t *this_thr = __kmp_threads[gtid];
+  kmp_team_t *team = this_thr->th.th_team;
+  int tid = __kmp_tid_from_gtid(gtid);
+
+  if (ompt_enabled.enabled) {
+    if (rc) {
+      if (ompt_enabled.ompt_callback_work) {
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_executor, ompt_scope_begin,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
+      }
+    } else {
+      if (ompt_enabled.ompt_callback_work) {
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_other, ompt_scope_begin,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
+        ompt_callbacks.ompt_callback(ompt_callback_work)(
+            ompt_work_single_other, ompt_scope_end,
+            &(team->t.ompt_team_info.parallel_data),
+            &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data),
+            1, OMPT_GET_RETURN_ADDRESS(0));
+      }
+    }
+  }
+#endif
+
+  return rc;
 }
 
 void *xexpand(KMP_API_NAME_GOMP_SINGLE_COPY_START)(void) {
   void *retval;
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_single_copy_start");
   KA_TRACE(20, ("GOMP_single_copy_start: T#%d\n", gtid));
 
   if (!TCR_4(__kmp_init_parallel))
     __kmp_parallel_initialize();
 
   // If this is the first thread to enter, return NULL.  The generated code will
   // then call GOMP_single_copy_end() for this thread only, with the
   // copyprivate data pointer as an argument.
   if (__kmp_enter_single(gtid, &loc, FALSE))
     return NULL;
 
-  // Wait for the first thread to set the copyprivate data pointer,
-  // and for all other threads to reach this point.
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
 
   // Retrieve the value of the copyprivate data point, and wait for all
   // threads to do likewise, then return.
   retval = __kmp_team_from_gtid(gtid)->t.t_copypriv_data;
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
   return retval;
 }
 
 void xexpand(KMP_API_NAME_GOMP_SINGLE_COPY_END)(void *data) {
   int gtid = __kmp_get_gtid();
   KA_TRACE(20, ("GOMP_single_copy_end: T#%d\n", gtid));
 
   // Set the copyprivate data pointer fo the team, then hit the barrier so that
   // the other threads will continue on and read it.  Hit another barrier before
   // continuing, so that the know that the copyprivate data pointer has been
   // propagated to all threads before trying to reuse the t_copypriv_data field.
   __kmp_team_from_gtid(gtid)->t.t_copypriv_data = data;
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 }
 
 void xexpand(KMP_API_NAME_GOMP_ORDERED_START)(void) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_ordered_start");
   KA_TRACE(20, ("GOMP_ordered_start: T#%d\n", gtid));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmpc_ordered(&loc, gtid);
 }
 
 void xexpand(KMP_API_NAME_GOMP_ORDERED_END)(void) {
   int gtid = __kmp_get_gtid();
   MKLOC(loc, "GOMP_ordered_end");
   KA_TRACE(20, ("GOMP_ordered_start: T#%d\n", gtid));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
   __kmpc_end_ordered(&loc, gtid);
 }
 
 // Dispatch macro defs
 //
 // They come in two flavors: 64-bit unsigned, and either 32-bit signed
 // (IA-32 architecture) or 64-bit signed (Intel(R) 64).
 

 static
 #endif /* KMP_DEBUG */
     void
     __kmp_GOMP_microtask_wrapper(int *gtid, int *npr, void (*task)(void *),
                                  void *data) {
 #if OMPT_SUPPORT
   kmp_info_t *thr;
   ompt_frame_t *ompt_frame;
-  ompt_state_t enclosing_state;
+  omp_state_t enclosing_state;
 
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     // get pointer to thread data structure
     thr = __kmp_threads[*gtid];
 
     // save enclosing task state; set current state for task
     enclosing_state = thr->th.ompt_thread_info.state;
-    thr->th.ompt_thread_info.state = ompt_state_work_parallel;
+    thr->th.ompt_thread_info.state = omp_state_work_parallel;
 
     // set task frame
-    ompt_frame = __ompt_get_task_frame_internal(0);
-    ompt_frame->exit_runtime_frame = __builtin_frame_address(0);
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(0);
   }
 #endif
 
   task(data);
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     // clear task frame
     ompt_frame->exit_runtime_frame = NULL;
 
     // restore enclosing state
     thr->th.ompt_thread_info.state = enclosing_state;
   }
 #endif
 }
 
 #ifndef KMP_DEBUG
 static
 #endif /* KMP_DEBUG */
     void
     __kmp_GOMP_parallel_microtask_wrapper(int *gtid, int *npr,
                                           void (*task)(void *), void *data,
                                           unsigned num_threads, ident_t *loc,
                                           enum sched_type schedule, long start,
                                           long end, long incr,
                                           long chunk_size) {
-  // Intialize the loop worksharing construct.
+
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled)
+    OMPT_STORE_RETURN_ADDRESS(*gtid);
+#endif
   KMP_DISPATCH_INIT(loc, *gtid, schedule, start, end, incr, chunk_size,
                     schedule != kmp_sch_static);
 
 #if OMPT_SUPPORT
   kmp_info_t *thr;
   ompt_frame_t *ompt_frame;
-  ompt_state_t enclosing_state;
+  omp_state_t enclosing_state;
 
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     thr = __kmp_threads[*gtid];
     // save enclosing task state; set current state for task
     enclosing_state = thr->th.ompt_thread_info.state;
-    thr->th.ompt_thread_info.state = ompt_state_work_parallel;
+    thr->th.ompt_thread_info.state = omp_state_work_parallel;
 
     // set task frame
-    ompt_frame = __ompt_get_task_frame_internal(0);
-    ompt_frame->exit_runtime_frame = __builtin_frame_address(0);
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(0);
   }
 #endif
 
   // Now invoke the microtask.
   task(data);
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     // clear task frame
     ompt_frame->exit_runtime_frame = NULL;
 
     // reset enclosing state
     thr->th.ompt_thread_info.state = enclosing_state;
   }
 #endif
 }
 
 #ifndef KMP_DEBUG
 static
 #endif /* KMP_DEBUG */
     void
     __kmp_GOMP_fork_call(ident_t *loc, int gtid, void (*unwrapped_task)(void *),
                          microtask_t wrapper, int argc, ...) {
   int rc;
   kmp_info_t *thr = __kmp_threads[gtid];
   kmp_team_t *team = thr->th.th_team;
   int tid = __kmp_tid_from_gtid(gtid);
 
   va_list ap;
   va_start(ap, argc);
 
-  rc = __kmp_fork_call(loc, gtid, fork_context_gnu, argc,
-#if OMPT_SUPPORT
-                       VOLATILE_CAST(void *) unwrapped_task,
-#endif
-                       wrapper, __kmp_invoke_task_func,
+  rc = __kmp_fork_call(loc, gtid, fork_context_gnu, argc, wrapper,
+                       __kmp_invoke_task_func,
 #if (KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64) && KMP_OS_LINUX
                        &ap
 #else
                        ap
 #endif
                        );
 
   va_end(ap);
 
   if (rc) {
     __kmp_run_before_invoked_task(gtid, tid, thr, team);
   }
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-#if OMPT_TRACE
+  int ompt_team_size;
+  if (ompt_enabled.enabled) {
     ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
+    ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
 
     // implicit task callback
-    if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(
-          team_info->parallel_id, task_info->task_id);
+    if (ompt_enabled.ompt_callback_implicit_task) {
+      ompt_team_size = __kmp_team_from_gtid(gtid)->t.t_nproc;
+      ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+          ompt_scope_begin, &(team_info->parallel_data),
+          &(task_info->task_data), ompt_team_size, __kmp_tid_from_gtid(gtid));
     }
-#endif
-    thr->th.ompt_thread_info.state = ompt_state_work_parallel;
+    thr->th.ompt_thread_info.state = omp_state_work_parallel;
   }
 #endif
 }
 
 static void __kmp_GOMP_serialized_parallel(ident_t *loc, kmp_int32 gtid,
                                            void (*task)(void *)) {
 #if OMPT_SUPPORT
-  ompt_parallel_id_t ompt_parallel_id;
-  if (ompt_enabled) {
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-
-    ompt_parallel_id = __ompt_parallel_id_new(gtid);
-
-    // parallel region callback
-    if (ompt_callbacks.ompt_callback(ompt_event_parallel_begin)) {
-      int team_size = 1;
-      ompt_callbacks.ompt_callback(ompt_event_parallel_begin)(
-          task_info->task_id, &task_info->frame, ompt_parallel_id, team_size,
-          (void *)task, OMPT_INVOKER(fork_context_gnu));
-    }
-  }
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 #endif
-
   __kmp_serialized_parallel(loc, gtid);
-
-#if OMPT_SUPPORT
-  if (ompt_enabled) {
-    kmp_info_t *thr = __kmp_threads[gtid];
-
-    ompt_task_id_t my_ompt_task_id = __ompt_task_id_new(gtid);
-
-    // set up lightweight task
-    ompt_lw_taskteam_t *lwt =
-        (ompt_lw_taskteam_t *)__kmp_allocate(sizeof(ompt_lw_taskteam_t));
-    __ompt_lw_taskteam_init(lwt, thr, gtid, (void *)task, ompt_parallel_id);
-    lwt->ompt_task_info.task_id = my_ompt_task_id;
-    __ompt_lw_taskteam_link(lwt, thr);
-
-#if OMPT_TRACE
-    // implicit task callback
-    if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(
-          ompt_parallel_id, my_ompt_task_id);
-    }
-    thr->th.ompt_thread_info.state = ompt_state_work_parallel;
-#endif
-  }
-#endif
 }
 
 void xexpand(KMP_API_NAME_GOMP_PARALLEL_START)(void (*task)(void *), void *data,
                                                unsigned num_threads) {
   int gtid = __kmp_entry_gtid();
 
 #if OMPT_SUPPORT
   ompt_frame_t *parent_frame, *frame;
 
-  if (ompt_enabled) {
-    parent_frame = __ompt_get_task_frame_internal(0);
-    parent_frame->reenter_runtime_frame = __builtin_frame_address(1);
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &parent_frame, NULL, NULL);
+    parent_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
   }
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 #endif
 
   MKLOC(loc, "GOMP_parallel_start");
   KA_TRACE(20, ("GOMP_parallel_start: T#%d\n", gtid));
 
   if (__kmpc_ok_to_fork(&loc) && (num_threads != 1)) {
     if (num_threads != 0) {
       __kmp_push_num_threads(&loc, gtid, num_threads);
     }
     __kmp_GOMP_fork_call(&loc, gtid, task,
                          (microtask_t)__kmp_GOMP_microtask_wrapper, 2, task,
                          data);
   } else {
     __kmp_GOMP_serialized_parallel(&loc, gtid, task);
   }
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    frame = __ompt_get_task_frame_internal(0);
-    frame->exit_runtime_frame = __builtin_frame_address(1);
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &frame, NULL, NULL);
+    frame->exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
   }
 #endif
 }
 
 void xexpand(KMP_API_NAME_GOMP_PARALLEL_END)(void) {
   int gtid = __kmp_get_gtid();
   kmp_info_t *thr;
+  int ompt_team_size = __kmp_team_from_gtid(gtid)->t.t_nproc;
 
   thr = __kmp_threads[gtid];
 
   MKLOC(loc, "GOMP_parallel_end");
   KA_TRACE(20, ("GOMP_parallel_end: T#%d\n", gtid));
 
-#if OMPT_SUPPORT
-  ompt_parallel_id_t parallel_id;
-  ompt_task_id_t serialized_task_id;
-  ompt_frame_t *ompt_frame = NULL;
-
-  if (ompt_enabled) {
-    ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);
-    parallel_id = team_info->parallel_id;
-
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-    serialized_task_id = task_info->task_id;
-
-    // unlink if necessary. no-op if there is not a lightweight task.
-    ompt_lw_taskteam_t *lwt = __ompt_lw_taskteam_unlink(thr);
-    // GOMP allocates/frees lwt since it can't be kept on the stack
-    if (lwt) {
-      __kmp_free(lwt);
-    }
-  }
-#endif
-
   if (!thr->th.th_team->t.t_serialized) {
     __kmp_run_after_invoked_task(gtid, __kmp_tid_from_gtid(gtid), thr,
                                  thr->th.th_team);
 
 #if OMPT_SUPPORT
-    if (ompt_enabled) {
+    if (ompt_enabled.enabled) {
       // Implicit task is finished here, in the barrier we might schedule
       // deferred tasks,
       // these don't see the implicit task on the stack
-      ompt_frame = __ompt_get_task_frame_internal(0);
-      ompt_frame->exit_runtime_frame = NULL;
+      OMPT_CUR_TASK_INFO(thr)->frame.exit_runtime_frame = NULL;
     }
 #endif
 
     __kmp_join_call(&loc, gtid
 #if OMPT_SUPPORT
                     ,
                     fork_context_gnu
 #endif
                     );
   } else {
-#if OMPT_SUPPORT && OMPT_TRACE
-    if (ompt_enabled &&
-        ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-      ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-          parallel_id, serialized_task_id);
-    }
-#endif
-
     __kmpc_end_serialized_parallel(&loc, gtid);
-
-#if OMPT_SUPPORT
-    if (ompt_enabled) {
-      // Record that we re-entered the runtime system in the frame that
-      // created the parallel region.
-      ompt_task_info_t *parent_task_info = __ompt_get_taskinfo(0);
-
-      if (ompt_callbacks.ompt_callback(ompt_event_parallel_end)) {
-        ompt_callbacks.ompt_callback(ompt_event_parallel_end)(
-            parallel_id, parent_task_info->task_id,
-            OMPT_INVOKER(fork_context_gnu));
-      }
-
-      parent_task_info->frame.reenter_runtime_frame = NULL;
-
-      thr->th.ompt_thread_info.state =
-          (((thr->th.th_team)->t.t_serialized) ? ompt_state_work_serial
-                                               : ompt_state_work_parallel);
-    }
-#endif
   }
 }
 
 // Loop worksharing constructs
 
 // The Gnu codegen passes in an exclusive upper bound for the overall range,
 // but the libguide dispatch code expects an inclusive upper bound, hence the
 // "end - incr" 5th argument to KMP_DISPATCH_INIT (and the " ub - str" 11th

 //
 // The profile lib cannot support construct checking of unordered loops that
 // are predetermined by the compiler to be statically scheduled, as the gcc
 // codegen will not always emit calls to GOMP_loop_static_next() to get the
 // next iteration.  Instead, it emits inline code to call omp_get_thread_num()
 // num and calculate the iteration space using the result.  It doesn't do this
 // with ordered static loop, so they can be checked.
 
+#if OMPT_SUPPORT
+#define IF_OMPT_SUPPORT(code) code
+#else
+#define IF_OMPT_SUPPORT(code)
+#endif
+
 #define LOOP_START(func, schedule)                                             \
   int func(long lb, long ub, long str, long chunk_sz, long *p_lb,              \
            long *p_ub) {                                                       \
     int status;                                                                \
     long stride;                                                               \
     int gtid = __kmp_entry_gtid();                                             \
     MKLOC(loc, #func);                                                         \
     KA_TRACE(20,                                                               \
              (#func ": T#%d, lb 0x%lx, ub 0x%lx, str 0x%lx, chunk_sz 0x%lx\n", \
               gtid, lb, ub, str, chunk_sz));                                   \
                                                                                \
     if ((str > 0) ? (lb < ub) : (lb > ub)) {                                   \
+      IF_OMPT_SUPPORT(OMPT_STORE_RETURN_ADDRESS(gtid);)                        \
       KMP_DISPATCH_INIT(&loc, gtid, (schedule), lb,                            \
                         (str > 0) ? (ub - 1) : (ub + 1), str, chunk_sz,        \
                         (schedule) != kmp_sch_static);                         \
+      IF_OMPT_SUPPORT(OMPT_STORE_RETURN_ADDRESS(gtid);)                        \
       status = KMP_DISPATCH_NEXT(&loc, gtid, NULL, (kmp_int *)p_lb,            \
                                  (kmp_int *)p_ub, (kmp_int *)&stride);         \
       if (status) {                                                            \
         KMP_DEBUG_ASSERT(stride == str);                                       \
         *p_ub += (str > 0) ? 1 : -1;                                           \
       }                                                                        \
     } else {                                                                   \
       status = 0;                                                              \
int func(long lb, long ub, long str, long *p_lb, long *p_ub) {               \
     long chunk_sz = 0;                                                         \
     int gtid = __kmp_entry_gtid();                                             \
     MKLOC(loc, #func);                                                         \
     KA_TRACE(20,                                                               \
              (#func ": T#%d, lb 0x%lx, ub 0x%lx, str 0x%lx, chunk_sz %d\n",    \
               gtid, lb, ub, str, chunk_sz));                                   \
                                                                                \
     if ((str > 0) ? (lb < ub) : (lb > ub)) {                                   \
+      IF_OMPT_SUPPORT(OMPT_STORE_RETURN_ADDRESS(gtid);)                        \
       KMP_DISPATCH_INIT(&loc, gtid, (schedule), lb,                            \
                         (str > 0) ? (ub - 1) : (ub + 1), str, chunk_sz, TRUE); \
+      IF_OMPT_SUPPORT(OMPT_STORE_RETURN_ADDRESS(gtid);)                        \
       status = KMP_DISPATCH_NEXT(&loc, gtid, NULL, (kmp_int *)p_lb,            \
                                  (kmp_int *)p_ub, (kmp_int *)&stride);         \
       if (status) {                                                            \
         KMP_DEBUG_ASSERT(stride == str);                                       \
         *p_ub += (str > 0) ? 1 : -1;                                           \
       }                                                                        \
     } else {                                                                   \
       status = 0;                                                              \
     }                                                                          \
                                                                                \
     KA_TRACE(20,                                                               \
              (#func " exit: T#%d, *p_lb 0x%lx, *p_ub 0x%lx, returning %d\n",   \
               gtid, *p_lb, *p_ub, status));                                    \
     return status;                                                             \
   }
 
 #define LOOP_NEXT(func, fini_code)                                             \
   int func(long *p_lb, long *p_ub) {                                           \
     int status;                                                                \
     long stride;                                                               \
     int gtid = __kmp_get_gtid();                                               \
     MKLOC(loc, #func);                                                         \
     KA_TRACE(20, (#func ": T#%d\n", gtid));                                    \
                                                                                \
+    IF_OMPT_SUPPORT(OMPT_STORE_RETURN_ADDRESS(gtid);)                          \
     fini_code status = KMP_DISPATCH_NEXT(&loc, gtid, NULL, (kmp_int *)p_lb,    \
                                          (kmp_int *)p_ub, (kmp_int *)&stride); \
     if (status) {                                                              \
       *p_ub += (stride > 0) ? 1 : -1;                                          \
     }                                                                          \
                                                                                \
     KA_TRACE(20,                                                               \
              (#func " exit: T#%d, *p_lb 0x%lx, *p_ub 0x%lx, stride 0x%lx, "    \
LOOP_RUNTIME_START(xexpand(KMP_API_NAME_GOMP_LOOP_ORDERED_RUNTIME_START),
                    kmp_ord_runtime)
 LOOP_NEXT(xexpand(KMP_API_NAME_GOMP_LOOP_ORDERED_RUNTIME_NEXT),
           { KMP_DISPATCH_FINI_CHUNK(&loc, gtid); })
 
 void xexpand(KMP_API_NAME_GOMP_LOOP_END)(void) {
   int gtid = __kmp_get_gtid();
   KA_TRACE(20, ("GOMP_loop_end: T#%d\n", gtid))
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 
   KA_TRACE(20, ("GOMP_loop_end exit: T#%d\n", gtid))
 }
 
 void xexpand(KMP_API_NAME_GOMP_LOOP_END_NOWAIT)(void) {
   KA_TRACE(20, ("GOMP_loop_end_nowait: T#%d\n", __kmp_get_gtid()))
 }
 
KMP_DISPATCH_INIT(&loc, gtid, (schedule), lb,                              \
                       (str > 0) ? (ub - 1) : (ub + 1), str, chunk_sz,          \
                       (schedule) != kmp_sch_static);                           \
                                                                                \
     ompt_post();                                                               \
                                                                                \
     KA_TRACE(20, (#func " exit: T#%d\n", gtid));                               \
   }
 
-#if OMPT_SUPPORT
+#if OMPT_SUPPORT && OMPT_OPTIONAL
 
 #define OMPT_LOOP_PRE()                                                        \
   ompt_frame_t *parent_frame;                                                  \
-  if (ompt_enabled) {                                                          \
-    parent_frame = __ompt_get_task_frame_internal(0);                          \
-    parent_frame->reenter_runtime_frame = __builtin_frame_address(1);          \
-  }
+  if (ompt_enabled.enabled) {                                                  \
+    __ompt_get_task_info_internal(0, NULL, NULL, &parent_frame, NULL, NULL);   \
+    parent_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);           \
+  }                                                                            \
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 
 #define OMPT_LOOP_POST()                                                       \
-  if (ompt_enabled) {                                                          \
+  if (ompt_enabled.enabled) {                                                  \
     parent_frame->reenter_runtime_frame = NULL;                                \
   }
 
 #else
 
 #define OMPT_LOOP_PRE()
 
 #define OMPT_LOOP_POST()
if (arg_size > 0) {
 
     if (copy_func) {
       (*copy_func)(task->shareds, data);
     } else {
       KMP_MEMCPY(task->shareds, data, arg_size);
     }
   }
 
+#if OMPT_SUPPORT
+  kmp_taskdata_t *current_task;
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+    current_task = __kmp_threads[gtid]->th.th_current_task;
+    current_task->ompt_task_info.frame.reenter_runtime_frame =
+        OMPT_GET_FRAME_ADDRESS(1);
+  }
+#endif
+
   if (if_cond) {
 #if OMP_40_ENABLED
     if (gomp_flags & 8) {
       KMP_ASSERT(depend);
       const size_t ndeps = (kmp_intptr_t)depend[0];
       const size_t nout = (kmp_intptr_t)depend[1];
       kmp_depend_info_t dep_list[ndeps];
 
       for (size_t i = 0U; i < ndeps; i++) {
         dep_list[i].base_addr = (kmp_intptr_t)depend[2U + i];
         dep_list[i].len = 0U;
         dep_list[i].flags.in = 1;
         dep_list[i].flags.out = (i < nout);
       }
       __kmpc_omp_task_with_deps(&loc, gtid, task, ndeps, dep_list, 0, NULL);
-    } else
+    } else {
 #endif
       __kmpc_omp_task(&loc, gtid, task);
+    }
   } else {
 #if OMPT_SUPPORT
     ompt_thread_info_t oldInfo;
     kmp_info_t *thread;
     kmp_taskdata_t *taskdata;
-    if (ompt_enabled) {
+    kmp_taskdata_t *current_task;
+    if (ompt_enabled.enabled) {
       // Store the threads states and restore them after the task
       thread = __kmp_threads[gtid];
       taskdata = KMP_TASK_TO_TASKDATA(task);
       oldInfo = thread->th.ompt_thread_info;
       thread->th.ompt_thread_info.wait_id = 0;
-      thread->th.ompt_thread_info.state = ompt_state_work_parallel;
+      thread->th.ompt_thread_info.state = omp_state_work_parallel;
       taskdata->ompt_task_info.frame.exit_runtime_frame =
-          __builtin_frame_address(0);
+          OMPT_GET_FRAME_ADDRESS(0);
+      OMPT_STORE_RETURN_ADDRESS(gtid);
     }
 #endif
 
     __kmpc_omp_task_begin_if0(&loc, gtid, task);
     func(data);
     __kmpc_omp_task_complete_if0(&loc, gtid, task);
 
 #if OMPT_SUPPORT
-    if (ompt_enabled) {
+    if (ompt_enabled.enabled) {
       thread->th.ompt_thread_info = oldInfo;
       taskdata->ompt_task_info.frame.exit_runtime_frame = NULL;
     }
 #endif
   }
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    current_task->ompt_task_info.frame.reenter_runtime_frame = NULL;
+  }
+#endif
 
   KA_TRACE(20, ("GOMP_task exit: T#%d\n", gtid));
 }
 
 void xexpand(KMP_API_NAME_GOMP_TASKWAIT)(void) {
   MKLOC(loc, "GOMP_taskwait");
   int gtid = __kmp_entry_gtid();
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled)
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+
   KA_TRACE(20, ("GOMP_taskwait: T#%d\n", gtid));
 
   __kmpc_omp_taskwait(&loc, gtid);
 
   KA_TRACE(20, ("GOMP_taskwait exit: T#%d\n", gtid));
 }
 
 // Sections worksharing constructs
void xexpand(KMP_API_NAME_GOMP_PARALLEL_SECTIONS_START)(void (*task)(void *),
                                                         void *data,
                                                         unsigned num_threads,
                                                         unsigned count) {
   int gtid = __kmp_entry_gtid();
 
 #if OMPT_SUPPORT
   ompt_frame_t *parent_frame;
 
-  if (ompt_enabled) {
-    parent_frame = __ompt_get_task_frame_internal(0);
-    parent_frame->reenter_runtime_frame = __builtin_frame_address(1);
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &parent_frame, NULL, NULL);
+    parent_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
   }
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 #endif
 
   MKLOC(loc, "GOMP_parallel_sections_start");
   KA_TRACE(20, ("GOMP_parallel_sections_start: T#%d\n", gtid));
 
   if (__kmpc_ok_to_fork(&loc) && (num_threads != 1)) {
     if (num_threads != 0) {
       __kmp_push_num_threads(&loc, gtid, num_threads);
     }
     __kmp_GOMP_fork_call(&loc, gtid, task,
                          (microtask_t)__kmp_GOMP_parallel_microtask_wrapper, 9,
                          task, data, num_threads, &loc, kmp_nm_dynamic_chunked,
                          (kmp_int)1, (kmp_int)count, (kmp_int)1, (kmp_int)1);
   } else {
     __kmp_GOMP_serialized_parallel(&loc, gtid, task);
   }
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     parent_frame->reenter_runtime_frame = NULL;
   }
 #endif
 
   KMP_DISPATCH_INIT(&loc, gtid, kmp_nm_dynamic_chunked, 1, count, 1, 1, TRUE);
 
   KA_TRACE(20, ("GOMP_parallel_sections_start exit: T#%d\n", gtid));
 }
 
 void xexpand(KMP_API_NAME_GOMP_SECTIONS_END)(void) {
   int gtid = __kmp_get_gtid();
   KA_TRACE(20, ("GOMP_sections_end: T#%d\n", gtid))
 
+#if OMPT_SUPPORT
+  ompt_frame_t *ompt_frame;
+  if (ompt_enabled.enabled) {
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   __kmp_barrier(bs_plain_barrier, gtid, FALSE, 0, NULL, NULL);
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
 
   KA_TRACE(20, ("GOMP_sections_end exit: T#%d\n", gtid))
 }
 
 void xexpand(KMP_API_NAME_GOMP_SECTIONS_END_NOWAIT)(void) {
   KA_TRACE(20, ("GOMP_sections_end_nowait: T#%d\n", __kmp_get_gtid()))
 }
 
void xexpand(KMP_API_NAME_GOMP_PARALLEL)(void (*task)(void *), void *data,
                                          unsigned num_threads,
                                          unsigned int flags) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_parallel");
   KA_TRACE(20, ("GOMP_parallel: T#%d\n", gtid));
 
 #if OMPT_SUPPORT
   ompt_task_info_t *parent_task_info, *task_info;
-  if (ompt_enabled) {
-    parent_task_info = __ompt_get_taskinfo(0);
-    parent_task_info->frame.reenter_runtime_frame = __builtin_frame_address(1);
+  if (ompt_enabled.enabled) {
+    parent_task_info = __ompt_get_task_info_object(0);
+    parent_task_info->frame.reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
   }
+  OMPT_STORE_RETURN_ADDRESS(gtid);
 #endif
   if (__kmpc_ok_to_fork(&loc) && (num_threads != 1)) {
     if (num_threads != 0) {
       __kmp_push_num_threads(&loc, gtid, num_threads);
     }
     if (flags != 0) {
       __kmp_push_proc_bind(&loc, gtid, (kmp_proc_bind_t)flags);
     }
     __kmp_GOMP_fork_call(&loc, gtid, task,
                          (microtask_t)__kmp_GOMP_microtask_wrapper, 2, task,
                          data);
   } else {
     __kmp_GOMP_serialized_parallel(&loc, gtid, task);
   }
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    task_info = __ompt_get_taskinfo(0);
-    task_info->frame.exit_runtime_frame = __builtin_frame_address(0);
+  if (ompt_enabled.enabled) {
+    task_info = __ompt_get_task_info_object(0);
+    task_info->frame.exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(0);
   }
 #endif
   task(data);
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+  }
+#endif
   xexpand(KMP_API_NAME_GOMP_PARALLEL_END)();
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     task_info->frame.exit_runtime_frame = NULL;
     parent_task_info->frame.reenter_runtime_frame = NULL;
   }
 #endif
 }
 
 void xexpand(KMP_API_NAME_GOMP_PARALLEL_SECTIONS)(void (*task)(void *),
                                                   void *data,
                                                   unsigned num_threads,
                                                   unsigned count,
                                                   unsigned flags) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_parallel_sections");
   KA_TRACE(20, ("GOMP_parallel_sections: T#%d\n", gtid));
 
+#if OMPT_SUPPORT
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+
   if (__kmpc_ok_to_fork(&loc) && (num_threads != 1)) {
     if (num_threads != 0) {
       __kmp_push_num_threads(&loc, gtid, num_threads);
     }
     if (flags != 0) {
       __kmp_push_proc_bind(&loc, gtid, (kmp_proc_bind_t)flags);
     }
     __kmp_GOMP_fork_call(&loc, gtid, task,
                          (microtask_t)__kmp_GOMP_parallel_microtask_wrapper, 9,
                          task, data, num_threads, &loc, kmp_nm_dynamic_chunked,
                          (kmp_int)1, (kmp_int)count, (kmp_int)1, (kmp_int)1);
   } else {
     __kmp_GOMP_serialized_parallel(&loc, gtid, task);
   }
 
   KMP_DISPATCH_INIT(&loc, gtid, kmp_nm_dynamic_chunked, 1, count, 1, 1, TRUE);
 
   task(data);
   xexpand(KMP_API_NAME_GOMP_PARALLEL_END)();
   KA_TRACE(20, ("GOMP_parallel_sections exit: T#%d\n", gtid));
 }
 
 #define PARALLEL_LOOP(func, schedule, ompt_pre, ompt_post)                     \
   void func(void (*task)(void *), void *data, unsigned num_threads, long lb,   \
             long ub, long str, long chunk_sz, unsigned flags) {                \
     int gtid = __kmp_entry_gtid();                                             \
     MKLOC(loc, #func);                                                         \
     KA_TRACE(20,                                                               \
              (#func ": T#%d, lb 0x%lx, ub 0x%lx, str 0x%lx, chunk_sz 0x%lx\n", \
               gtid, lb, ub, str, chunk_sz));                                   \
                                                                                \
     ompt_pre();                                                                \
     if (__kmpc_ok_to_fork(&loc) && (num_threads != 1)) {                       \
       if (num_threads != 0) {                                                  \
         __kmp_push_num_threads(&loc, gtid, num_threads);                       \
       }                                                                        \
       if (flags != 0) {                                                        \
         __kmp_push_proc_bind(&loc, gtid, (kmp_proc_bind_t)flags);              \
       }                                                                        \
       __kmp_GOMP_fork_call(&loc, gtid, task,                                   \
                            (microtask_t)__kmp_GOMP_parallel_microtask_wrapper, \
                            9, task, data, num_threads, &loc, (schedule), lb,   \
                            (str > 0) ? (ub - 1) : (ub + 1), str, chunk_sz);    \
     } else {                                                                   \
       __kmp_GOMP_serialized_parallel(&loc, gtid, task);                        \
     }                                                                          \
                                                                                \
+    IF_OMPT_SUPPORT(if (ompt_enabled.enabled)                          \
+                                OMPT_STORE_RETURN_ADDRESS(gtid);)              \
     KMP_DISPATCH_INIT(&loc, gtid, (schedule), lb,                              \
                       (str > 0) ? (ub - 1) : (ub + 1), str, chunk_sz,          \
                       (schedule) != kmp_sch_static);                           \
     task(data);                                                                \
     xexpand(KMP_API_NAME_GOMP_PARALLEL_END)();                                 \
     ompt_post();                                                               \
                                                                                \
     KA_TRACE(20, (#func " exit: T#%d\n", gtid));                               \
   }
 
 PARALLEL_LOOP(xexpand(KMP_API_NAME_GOMP_PARALLEL_LOOP_STATIC), kmp_sch_static,
               OMPT_LOOP_PRE, OMPT_LOOP_POST)
 PARALLEL_LOOP(xexpand(KMP_API_NAME_GOMP_PARALLEL_LOOP_DYNAMIC),
               kmp_sch_dynamic_chunked, OMPT_LOOP_PRE, OMPT_LOOP_POST)
 PARALLEL_LOOP(xexpand(KMP_API_NAME_GOMP_PARALLEL_LOOP_GUIDED),
               kmp_sch_guided_chunked, OMPT_LOOP_PRE, OMPT_LOOP_POST)
 PARALLEL_LOOP(xexpand(KMP_API_NAME_GOMP_PARALLEL_LOOP_RUNTIME), kmp_sch_runtime,
               OMPT_LOOP_PRE, OMPT_LOOP_POST)
 
 void xexpand(KMP_API_NAME_GOMP_TASKGROUP_START)(void) {
   int gtid = __kmp_entry_gtid();
   MKLOC(loc, "GOMP_taskgroup_start");
   KA_TRACE(20, ("GOMP_taskgroup_start: T#%d\n", gtid));
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled)
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+
   __kmpc_taskgroup(&loc, gtid);
 
   return;
 }
 
 void xexpand(KMP_API_NAME_GOMP_TASKGROUP_END)(void) {
   int gtid = __kmp_get_gtid();
   MKLOC(loc, "GOMP_taskgroup_end");
   KA_TRACE(20, ("GOMP_taskgroup_end: T#%d\n", gtid));
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled)
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
+
   __kmpc_end_taskgroup(&loc, gtid);
 
   return;
 }
 
 #ifndef KMP_DEBUG
 static
 #endif /* KMP_DEBUG */

runtime/src/kmp_lock.h

typedef struct {
   kmp_user_lock_p lock;
   kmp_indirect_locktag_t type;
 } kmp_indirect_lock_t;
 
 // Function tables for direct locks. Set/unset/test differentiate functions
 // with/without consistency checking.
 extern void (*__kmp_direct_init[])(kmp_dyna_lock_t *, kmp_dyna_lockseq_t);
 extern void (*__kmp_direct_destroy[])(kmp_dyna_lock_t *);
-extern void (*(*__kmp_direct_set))(kmp_dyna_lock_t *, kmp_int32);
+extern int (*(*__kmp_direct_set))(kmp_dyna_lock_t *, kmp_int32);
 extern int (*(*__kmp_direct_unset))(kmp_dyna_lock_t *, kmp_int32);
 extern int (*(*__kmp_direct_test))(kmp_dyna_lock_t *, kmp_int32);
 
 // Function tables for indirect locks. Set/unset/test differentiate functions
 // with/withuot consistency checking.
 extern void (*__kmp_indirect_init[])(kmp_user_lock_p);
 extern void (*__kmp_indirect_destroy[])(kmp_user_lock_p);
-extern void (*(*__kmp_indirect_set))(kmp_user_lock_p, kmp_int32);
+extern int (*(*__kmp_indirect_set))(kmp_user_lock_p, kmp_int32);
 extern int (*(*__kmp_indirect_unset))(kmp_user_lock_p, kmp_int32);
 extern int (*(*__kmp_indirect_test))(kmp_user_lock_p, kmp_int32);
 
 // Extracts direct lock tag from a user lock pointer
 #define KMP_EXTRACT_D_TAG(l)                                                   \
   (*((kmp_dyna_lock_t *)(l)) & ((1 << KMP_LOCK_SHIFT) - 1) &                   \
    -(*((kmp_dyna_lock_t *)(l)) & 1))
 

runtime/src/kmp_lock.cpp

__kmp_acquire_queuing_lock_timed_template(kmp_queuing_lock_t *lck,
                                           kmp_int32 gtid) {
   kmp_info_t *this_thr = __kmp_thread_from_gtid(gtid);
   volatile kmp_int32 *head_id_p = &lck->lk.head_id;
   volatile kmp_int32 *tail_id_p = &lck->lk.tail_id;
   volatile kmp_uint32 *spin_here_p;
   kmp_int32 need_mf = 1;
 
 #if OMPT_SUPPORT
-  ompt_state_t prev_state = ompt_state_undefined;
+  omp_state_t prev_state = omp_state_undefined;
 #endif
 
   KA_TRACE(1000,
            ("__kmp_acquire_queuing_lock: lck:%p, T#%d entering\n", lck, gtid));
 
   KMP_FSYNC_PREPARE(lck);
   KMP_DEBUG_ASSERT(this_thr != NULL);
   spin_here_p = &this_thr->th.th_spin_here;
#endif
             1000,
             ("__kmp_acquire_queuing_lock: lck:%p, T#%d exiting: no queuing\n",
              lck, gtid));
 #ifdef DEBUG_QUEUING_LOCKS
         TRACE_LOCK_HT(gtid + 1, "acq exit: ", head, 0);
 #endif
 
 #if OMPT_SUPPORT
-        if (ompt_enabled && prev_state != ompt_state_undefined) {
+        if (ompt_enabled.enabled && prev_state != omp_state_undefined) {
           /* change the state before clearing wait_id */
           this_thr->th.ompt_thread_info.state = prev_state;
           this_thr->th.ompt_thread_info.wait_id = 0;
         }
 #endif
 
         KMP_FSYNC_ACQUIRED(lck);
         return KMP_LOCK_ACQUIRED_FIRST; /* lock holder cannot be on queue */
       }
       enqueued = FALSE;
     } break;
     }
 
 #if OMPT_SUPPORT
-    if (ompt_enabled && prev_state == ompt_state_undefined) {
+    if (ompt_enabled.enabled && prev_state == omp_state_undefined) {
       /* this thread will spin; set wait_id before entering wait state */
       prev_state = this_thr->th.ompt_thread_info.state;
       this_thr->th.ompt_thread_info.wait_id = (uint64_t)lck;
-      this_thr->th.ompt_thread_info.state = ompt_state_wait_lock;
+      this_thr->th.ompt_thread_info.state = omp_state_wait_lock;
     }
 #endif
 
     if (enqueued) {
       if (tail > 0) {
         kmp_info_t *tail_thr = __kmp_thread_from_gtid(tail - 1);
         KMP_ASSERT(tail_thr != NULL);
         tail_thr->th.th_next_waiting = gtid + 1;

 }
 
 #endif // KMP_USE_TSX
 
 // Entry functions for indirect locks (first element of direct lock jump tables)
 static void __kmp_init_indirect_lock(kmp_dyna_lock_t *l,
                                      kmp_dyna_lockseq_t tag);
 static void __kmp_destroy_indirect_lock(kmp_dyna_lock_t *lock);
-static void __kmp_set_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32);
+static int __kmp_set_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32);
 static int __kmp_unset_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32);
 static int __kmp_test_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32);
-static void __kmp_set_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
-                                                kmp_int32);
+static int __kmp_set_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
 static int __kmp_unset_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
                                                  kmp_int32);
 static int __kmp_test_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
                                                 kmp_int32);
 
 // Jump tables for the indirect lock functions
 // Only fill in the odd entries, that avoids the need to shift out the low bit
 
 // init functions
 #define expand(l, op) 0, __kmp_init_direct_lock,
 void (*__kmp_direct_init[])(kmp_dyna_lock_t *, kmp_dyna_lockseq_t) = {
     __kmp_init_indirect_lock, 0, KMP_FOREACH_D_LOCK(expand, init)};
 #undef expand
 
 // destroy functions
 #define expand(l, op) 0, (void (*)(kmp_dyna_lock_t *))__kmp_##op##_##l##_lock,
 void (*__kmp_direct_destroy[])(kmp_dyna_lock_t *) = {
     __kmp_destroy_indirect_lock, 0, KMP_FOREACH_D_LOCK(expand, destroy)};
 #undef expand
 
 // set/acquire functions
 #define expand(l, op)                                                          \
-  0, (void (*)(kmp_dyna_lock_t *, kmp_int32))__kmp_##op##_##l##_lock,
-static void (*direct_set[])(kmp_dyna_lock_t *, kmp_int32) = {
+  0, (int (*)(kmp_dyna_lock_t *, kmp_int32))__kmp_##op##_##l##_lock,
+static int (*direct_set[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_set_indirect_lock, 0, KMP_FOREACH_D_LOCK(expand, acquire)};
 #undef expand
 #define expand(l, op)                                                          \
-  0, (void (*)(kmp_dyna_lock_t *,                                              \
-               kmp_int32))__kmp_##op##_##l##_lock_with_checks,
-static void (*direct_set_check[])(kmp_dyna_lock_t *, kmp_int32) = {
+  0, (int (*)(kmp_dyna_lock_t *, kmp_int32))__kmp_##op##_##l##_lock_with_checks,
+static int (*direct_set_check[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_set_indirect_lock_with_checks, 0,
     KMP_FOREACH_D_LOCK(expand, acquire)};
 #undef expand
 
 // unset/release and test functions
 #define expand(l, op)                                                          \
   0, (int (*)(kmp_dyna_lock_t *, kmp_int32))__kmp_##op##_##l##_lock,
 static int (*direct_unset[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_unset_indirect_lock, 0, KMP_FOREACH_D_LOCK(expand, release)};
 static int (*direct_test[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_test_indirect_lock, 0, KMP_FOREACH_D_LOCK(expand, test)};
 #undef expand
 #define expand(l, op)                                                          \
   0, (int (*)(kmp_dyna_lock_t *, kmp_int32))__kmp_##op##_##l##_lock_with_checks,
 static int (*direct_unset_check[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_unset_indirect_lock_with_checks, 0,
     KMP_FOREACH_D_LOCK(expand, release)};
 static int (*direct_test_check[])(kmp_dyna_lock_t *, kmp_int32) = {
     __kmp_test_indirect_lock_with_checks, 0, KMP_FOREACH_D_LOCK(expand, test)};
 #undef expand
 
 // Exposes only one set of jump tables (*lock or *lock_with_checks).
-void (*(*__kmp_direct_set))(kmp_dyna_lock_t *, kmp_int32) = 0;
+int (*(*__kmp_direct_set))(kmp_dyna_lock_t *, kmp_int32) = 0;
 int (*(*__kmp_direct_unset))(kmp_dyna_lock_t *, kmp_int32) = 0;
 int (*(*__kmp_direct_test))(kmp_dyna_lock_t *, kmp_int32) = 0;
 
 // Jump tables for the indirect lock functions
 #define expand(l, op) (void (*)(kmp_user_lock_p)) __kmp_##op##_##l##_##lock,
 void (*__kmp_indirect_init[])(kmp_user_lock_p) = {
     KMP_FOREACH_I_LOCK(expand, init)};
 void (*__kmp_indirect_destroy[])(kmp_user_lock_p) = {
     KMP_FOREACH_I_LOCK(expand, destroy)};
 #undef expand
 
 // set/acquire functions
 #define expand(l, op)                                                          \
-  (void (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock,
-static void (*indirect_set[])(kmp_user_lock_p, kmp_int32) = {
-    KMP_FOREACH_I_LOCK(expand, acquire)};
+  (int (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock,
+static int (*indirect_set[])(kmp_user_lock_p,
+                             kmp_int32) = {KMP_FOREACH_I_LOCK(expand, acquire)};
 #undef expand
 #define expand(l, op)                                                          \
-  (void (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock_with_checks,
-static void (*indirect_set_check[])(kmp_user_lock_p, kmp_int32) = {
+  (int (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock_with_checks,
+static int (*indirect_set_check[])(kmp_user_lock_p, kmp_int32) = {
     KMP_FOREACH_I_LOCK(expand, acquire)};
 #undef expand
 
 // unset/release and test functions
 #define expand(l, op)                                                          \
   (int (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock,
 static int (*indirect_unset[])(kmp_user_lock_p, kmp_int32) = {
     KMP_FOREACH_I_LOCK(expand, release)};
 static int (*indirect_test[])(kmp_user_lock_p,
                               kmp_int32) = {KMP_FOREACH_I_LOCK(expand, test)};
 #undef expand
 #define expand(l, op)                                                          \
   (int (*)(kmp_user_lock_p, kmp_int32)) __kmp_##op##_##l##_##lock_with_checks,
 static int (*indirect_unset_check[])(kmp_user_lock_p, kmp_int32) = {
     KMP_FOREACH_I_LOCK(expand, release)};
 static int (*indirect_test_check[])(kmp_user_lock_p, kmp_int32) = {
     KMP_FOREACH_I_LOCK(expand, test)};
 #undef expand
 
 // Exposes only one jump tables (*lock or *lock_with_checks).
-void (*(*__kmp_indirect_set))(kmp_user_lock_p, kmp_int32) = 0;
+int (*(*__kmp_indirect_set))(kmp_user_lock_p, kmp_int32) = 0;
 int (*(*__kmp_indirect_unset))(kmp_user_lock_p, kmp_int32) = 0;
 int (*(*__kmp_indirect_test))(kmp_user_lock_p, kmp_int32) = 0;
 
 // Lock index table.
 kmp_indirect_lock_table_t __kmp_i_lock_table;
 
 // Size of indirect locks.
 static kmp_uint32 __kmp_indirect_lock_size[KMP_NUM_I_LOCKS] = {0};
static void __kmp_destroy_indirect_lock(kmp_dyna_lock_t *lock) {
   if (OMP_LOCK_T_SIZE < sizeof(void *)) {
     l->lock->pool.index = KMP_EXTRACT_I_INDEX(lock);
   }
   __kmp_indirect_lock_pool[tag] = l;
 
   __kmp_release_lock(&__kmp_global_lock, gtid);
 }
 
-static void __kmp_set_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) {
+static int __kmp_set_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) {
   kmp_indirect_lock_t *l = KMP_LOOKUP_I_LOCK(lock);
-  KMP_I_LOCK_FUNC(l, set)(l->lock, gtid);
+  return KMP_I_LOCK_FUNC(l, set)(l->lock, gtid);
 }
 
 static int __kmp_unset_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) {
   kmp_indirect_lock_t *l = KMP_LOOKUP_I_LOCK(lock);
   return KMP_I_LOCK_FUNC(l, unset)(l->lock, gtid);
 }
 
 static int __kmp_test_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) {
   kmp_indirect_lock_t *l = KMP_LOOKUP_I_LOCK(lock);
   return KMP_I_LOCK_FUNC(l, test)(l->lock, gtid);
 }
 
-static void __kmp_set_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
-                                                kmp_int32 gtid) {
+static int __kmp_set_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
   kmp_indirect_lock_t *l =
       __kmp_lookup_indirect_lock((void **)lock, "omp_set_lock");
-  KMP_I_LOCK_FUNC(l, set)(l->lock, gtid);
+  return KMP_I_LOCK_FUNC(l, set)(l->lock, gtid);
 }
 
 static int __kmp_unset_indirect_lock_with_checks(kmp_dyna_lock_t *lock,
                                                  kmp_int32 gtid) {
   kmp_indirect_lock_t *l =
       __kmp_lookup_indirect_lock((void **)lock, "omp_unset_lock");
   return KMP_I_LOCK_FUNC(l, unset)(l->lock, gtid);
 }

runtime/src/kmp_runtime.cpp

 #ifdef BUILD_PARALLEL_ORDERED
   if (!team->t.t_serialized) {
     KMP_MB(); /* Flush all pending memory write invalidates.  */
 
     /* use the tid of the next thread in this team */
     /* TODO replace with general release procedure */
     team->t.t_ordered.dt.t_value = ((tid + 1) % team->t.t_nproc);
 
-#if OMPT_SUPPORT && OMPT_BLAME
-    if (ompt_enabled &&
-        ompt_callbacks.ompt_callback(ompt_event_release_ordered)) {
-      /* accept blame for "ordered" waiting */
-      kmp_info_t *this_thread = __kmp_threads[gtid];
-      ompt_callbacks.ompt_callback(ompt_event_release_ordered)(
-          this_thread->th.ompt_thread_info.wait_id);
-    }
-#endif
-
     KMP_MB(); /* Flush all pending memory write invalidates.  */
   }
 #endif /* BUILD_PARALLEL_ORDERED */
 }
 
 /* ------------------------------------------------------------------------ */
 /* The BARRIER for a SINGLE process section is always explicit   */
 
if (__kmp_inherit_fp_control) {
     kmp_uint32 mxcsr;
 
     // Get master values of FPU control flags (both X87 and vector)
     __kmp_store_x87_fpu_control_word(&x87_fpu_control_word);
     __kmp_store_mxcsr(&mxcsr);
     mxcsr &= KMP_X86_MXCSR_MASK;
 
     // There is no point looking at t_fp_control_saved here.
     // If it is TRUE, we still have to update the values if they are different
     // from those we now have. If it is FALSE we didn't save anything yet, but
     // our objective is the same. We have to ensure that the values in the team
     // are the same as those we have.
     // So, this code achieves what we need whether or not t_fp_control_saved is
     // true. By checking whether the value needs updating we avoid unnecessary
     // writes that would put the cache-line into a written state, causing all
     // threads in the team to have to read it again.
     KMP_CHECK_UPDATE(team->t.t_x87_fpu_control_word, x87_fpu_control_word);
     KMP_CHECK_UPDATE(team->t.t_mxcsr, mxcsr);
     // Although we don't use this value, other code in the runtime wants to know
     // whether it should restore them. So we must ensure it is correct.
     KMP_CHECK_UPDATE(team->t.t_fp_control_saved, TRUE);
if (this_thr->th.th_current_task->td_icvs.proc_bind == proc_bind_false) {
     // No proc_bind clause was specified, so use the current value
     // of proc-bind-var for this parallel region.
     proc_bind = this_thr->th.th_current_task->td_icvs.proc_bind;
   }
   // Reset for next parallel region
   this_thr->th.th_set_proc_bind = proc_bind_default;
 #endif /* OMP_40_ENABLED */
 
+#if OMPT_SUPPORT
+  ompt_data_t ompt_parallel_data;
+  ompt_parallel_data.ptr = NULL;
+  ompt_data_t *implicit_task_data;
+  void *codeptr = OMPT_LOAD_RETURN_ADDRESS(global_tid);
+  if (ompt_enabled.enabled &&
+      this_thr->th.ompt_thread_info.state != omp_state_overhead) {
+
+    ompt_task_info_t *parent_task_info;
+    parent_task_info = OMPT_CUR_TASK_INFO(this_thr);
+
+    parent_task_info->frame.reenter_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+    if (ompt_enabled.ompt_callback_parallel_begin) {
+      int team_size = 1;
+
+      ompt_callbacks.ompt_callback(ompt_callback_parallel_begin)(
+          &(parent_task_info->task_data), &(parent_task_info->frame),
+          &ompt_parallel_data, team_size,
+          ompt_invoker_program, codeptr);
+    }
+  }
+#endif // OMPT_SUPPORT
+
   if (this_thr->th.th_team != serial_team) {
     // Nested level will be an index in the nested nthreads array
     int level = this_thr->th.th_team->t.t_level;
 
     if (serial_team->t.t_serialized) {
       /* this serial team was already used
          TODO increase performance by making this locks more specific */
       kmp_team_t *new_team;
 
       __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock);
 
-#if OMPT_SUPPORT
-      ompt_parallel_id_t ompt_parallel_id = __ompt_parallel_id_new(global_tid);
-#endif
-
       new_team = __kmp_allocate_team(this_thr->th.th_root, 1, 1,
 #if OMPT_SUPPORT
-                                     ompt_parallel_id,
+                                     ompt_parallel_data,
 #endif
 #if OMP_40_ENABLED
                                      proc_bind,
 #endif
                                      &this_thr->th.th_current_task->td_icvs,
                                      0 USE_NESTED_HOT_ARG(NULL));
       __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock);
       KMP_ASSERT(new_team);
#endif
     KMP_DEBUG_ASSERT(serial_team->t.t_dispatch);
     if (!serial_team->t.t_dispatch->th_disp_buffer) {
       serial_team->t.t_dispatch->th_disp_buffer =
           (dispatch_private_info_t *)__kmp_allocate(
               sizeof(dispatch_private_info_t));
     }
     this_thr->th.th_dispatch = serial_team->t.t_dispatch;
 
-#if OMPT_SUPPORT
-    ompt_parallel_id_t ompt_parallel_id = __ompt_parallel_id_new(global_tid);
-    __ompt_team_assign_id(serial_team, ompt_parallel_id);
-#endif
-
     KMP_MB();
 
   } else {
     /* this serialized team is already being used,
      * that's fine, just add another nested level */
     KMP_DEBUG_ASSERT(this_thr->th.th_team == serial_team);
     KMP_DEBUG_ASSERT(serial_team->t.t_threads);
     KMP_DEBUG_ASSERT(serial_team->t.t_threads[0] == this_thr);
#endif
     KMP_MB();
   }
 #if OMP_40_ENABLED
   KMP_CHECK_UPDATE(serial_team->t.t_cancel_request, cancel_noreq);
 #endif
 
   if (__kmp_env_consistency_check)
     __kmp_push_parallel(global_tid, NULL);
+#if OMPT_SUPPORT
+  serial_team->t.ompt_team_info.master_return_address = codeptr;
+  if (ompt_enabled.enabled &&
+      this_thr->th.ompt_thread_info.state != omp_state_overhead) {
+    OMPT_CUR_TASK_INFO(this_thr)
+        ->frame.exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+
+    ompt_lw_taskteam_t lw_taskteam;
+    __ompt_lw_taskteam_init(&lw_taskteam, this_thr, global_tid,
+                            &ompt_parallel_data, codeptr);
+
+    __ompt_lw_taskteam_link(&lw_taskteam, this_thr, 1);
+    // don't use lw_taskteam after linking. content was swaped
+
+    /* OMPT implicit task begin */
+    implicit_task_data = OMPT_CUR_TASK_DATA(this_thr);
+    if (ompt_enabled.ompt_callback_implicit_task) {
+      ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+          ompt_scope_begin, OMPT_CUR_TEAM_DATA(this_thr),
+          OMPT_CUR_TASK_DATA(this_thr), 1, __kmp_tid_from_gtid(global_tid));
+    }
+
+    /* OMPT state */
+    this_thr->th.ompt_thread_info.state = omp_state_work_parallel;
+    OMPT_CUR_TASK_INFO(this_thr)
+        ->frame.exit_runtime_frame = OMPT_GET_FRAME_ADDRESS(1);
+  }
+#endif
 }
 
 /* most of the work for a fork */
 /* return true if we really went parallel, false if serialized */
 int __kmp_fork_call(ident_t *loc, int gtid,
                     enum fork_context_e call_context, // Intel, GNU, ...
-                    kmp_int32 argc,
-#if OMPT_SUPPORT
-                    void *unwrapped_task,
-#endif
-                    microtask_t microtask, launch_t invoker,
+                    kmp_int32 argc, microtask_t microtask, launch_t invoker,
 /* TODO: revert workaround for Intel(R) 64 tracker #96 */
 #if (KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64) && KMP_OS_LINUX
                     va_list *ap
 #else
                     va_list ap
 #endif
                     ) {
   void **argv;
#endif
     parent_team = master_th->th.th_team;
     master_tid = master_th->th.th_info.ds.ds_tid;
     master_this_cons = master_th->th.th_local.this_construct;
     root = master_th->th.th_root;
     master_active = root->r.r_active;
     master_set_numthreads = master_th->th.th_set_nproc;
 
 #if OMPT_SUPPORT
-    ompt_parallel_id_t ompt_parallel_id;
-    ompt_task_id_t ompt_task_id;
+    ompt_data_t ompt_parallel_data;
+    ompt_parallel_data.ptr = NULL;
+    ompt_data_t *parent_task_data;
     ompt_frame_t *ompt_frame;
-    ompt_task_id_t my_task_id;
-    ompt_parallel_id_t my_parallel_id;
+    ompt_data_t *implicit_task_data;
+    void *return_address = NULL;
 
-    if (ompt_enabled) {
-      ompt_parallel_id = __ompt_parallel_id_new(gtid);
-      ompt_task_id = __ompt_get_task_id_internal(0);
-      ompt_frame = __ompt_get_task_frame_internal(0);
+    if (ompt_enabled.enabled) {
+      __ompt_get_task_info_internal(0, NULL, &parent_task_data, &ompt_frame,
+                                    NULL, NULL);
+      return_address = OMPT_LOAD_RETURN_ADDRESS(gtid);
     }
 #endif
 
     // Nested level will be an index in the nested nthreads array
     level = parent_team->t.t_level;
     // used to launch non-serial teams even if nested is not allowed
     active_level = parent_team->t.t_active_level;
 #if OMP_40_ENABLED
     // needed to check nesting inside the teams
     teams_level = master_th->th.th_teams_level;
 #endif
 #if KMP_NESTED_HOT_TEAMS
     p_hot_teams = &master_th->th.th_hot_teams;
     if (*p_hot_teams == NULL && __kmp_hot_teams_max_level > 0) {
       *p_hot_teams = (kmp_hot_team_ptr_t *)__kmp_allocate(
           sizeof(kmp_hot_team_ptr_t) * __kmp_hot_teams_max_level);
       (*p_hot_teams)[0].hot_team = root->r.r_hot_team;
       // it is either actual or not needed (when active_level > 0)
       (*p_hot_teams)[0].hot_team_nth = 1;
     }
 #endif
 
 #if OMPT_SUPPORT
-    if (ompt_enabled &&
-        ompt_callbacks.ompt_callback(ompt_event_parallel_begin)) {
-      int team_size = master_set_numthreads;
-
-      ompt_callbacks.ompt_callback(ompt_event_parallel_begin)(
-          ompt_task_id, ompt_frame, ompt_parallel_id, team_size, unwrapped_task,
-          OMPT_INVOKER(call_context));
+    if (ompt_enabled.enabled) {
+      if (ompt_enabled.ompt_callback_parallel_begin) {
+        int team_size = master_set_numthreads
+                            ? master_set_numthreads
+                            : get__nproc_2(parent_team, master_tid);
+        ompt_callbacks.ompt_callback(ompt_callback_parallel_begin)(
+            parent_task_data, ompt_frame, &ompt_parallel_data, team_size,
+            OMPT_INVOKER(call_context), return_address);
+      }
+      master_th->th.ompt_thread_info.state = omp_state_overhead;
     }
 #endif
 
     master_th->th.th_ident = loc;
 
 #if OMP_40_ENABLED
     if (master_th->th.th_teams_microtask && ap &&
         microtask != (microtask_t)__kmp_teams_master && level == teams_level) {
#endif
         // correctly, will restore at join time
         parent_team->t.t_serialized--;
 #if OMPT_SUPPORT
         void *dummy;
         void **exit_runtime_p;
 
         ompt_lw_taskteam_t lw_taskteam;
 
-        if (ompt_enabled) {
-          __ompt_lw_taskteam_init(&lw_taskteam, master_th, gtid, unwrapped_task,
-                                  ompt_parallel_id);
-          lw_taskteam.ompt_task_info.task_id = __ompt_task_id_new(gtid);
+        if (ompt_enabled.enabled) {
+          __ompt_lw_taskteam_init(&lw_taskteam, master_th, gtid,
+                                  &ompt_parallel_data, return_address);
           exit_runtime_p =
               &(lw_taskteam.ompt_task_info.frame.exit_runtime_frame);
 
-          __ompt_lw_taskteam_link(&lw_taskteam, master_th);
+          __ompt_lw_taskteam_link(&lw_taskteam, master_th, 0);
+          // don't use lw_taskteam after linking. content was swaped
 
-#if OMPT_TRACE
           /* OMPT implicit task begin */
-          my_task_id = lw_taskteam.ompt_task_info.task_id;
-          my_parallel_id = parent_team->t.ompt_team_info.parallel_id;
-          if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-            ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(
-                my_parallel_id, my_task_id);
+          implicit_task_data = OMPT_CUR_TASK_DATA(master_th);
+          if (ompt_enabled.ompt_callback_implicit_task) {
+            ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                ompt_scope_begin, OMPT_CUR_TEAM_DATA(master_th),
+                implicit_task_data, 1, __kmp_tid_from_gtid(gtid));
           }
-#endif
 
           /* OMPT state */
-          master_th->th.ompt_thread_info.state = ompt_state_work_parallel;
+          master_th->th.ompt_thread_info.state = omp_state_work_parallel;
         } else {
           exit_runtime_p = &dummy;
         }
 #endif
 
         {
           KMP_TIME_PARTITIONED_BLOCK(OMP_parallel);
           KMP_SET_THREAD_STATE_BLOCK(IMPLICIT_TASK);
           __kmp_invoke_microtask(microtask, gtid, 0, argc, parent_team->t.t_argv
 #if OMPT_SUPPORT
                                  ,
                                  exit_runtime_p
 #endif
                                  );
         }
 
 #if OMPT_SUPPORT
         *exit_runtime_p = NULL;
-        if (ompt_enabled) {
-#if OMPT_TRACE
-          lw_taskteam.ompt_task_info.frame.exit_runtime_frame = NULL;
-
-          if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-            ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-                ompt_parallel_id, ompt_task_id);
+        if (ompt_enabled.enabled) {
+          OMPT_CUR_TASK_INFO(master_th)->frame.exit_runtime_frame = NULL;
+          if (ompt_enabled.ompt_callback_implicit_task) {
+            ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                ompt_scope_end, NULL, implicit_task_data, 1,
+                __kmp_tid_from_gtid(gtid));
           }
-
           __ompt_lw_taskteam_unlink(master_th);
-          // reset clear the task id only after unlinking the task
-          lw_taskteam.ompt_task_info.task_id = ompt_task_id_none;
-#endif
 
-          if (ompt_callbacks.ompt_callback(ompt_event_parallel_end)) {
-            ompt_callbacks.ompt_callback(ompt_event_parallel_end)(
-                ompt_parallel_id, ompt_task_id, OMPT_INVOKER(call_context));
+          if (ompt_enabled.ompt_callback_parallel_end) {
+            ompt_callbacks.ompt_callback(ompt_callback_parallel_end)(
+                OMPT_CUR_TEAM_DATA(master_th), OMPT_CUR_TASK_DATA(master_th),
+                OMPT_INVOKER(call_context), return_address);
           }
-          master_th->th.ompt_thread_info.state = ompt_state_overhead;
+          master_th->th.ompt_thread_info.state = omp_state_overhead;
         }
 #endif
         return TRUE;
       }
 
       parent_team->t.t_pkfn = microtask;
-#if OMPT_SUPPORT
-      parent_team->t.ompt_team_info.microtask = unwrapped_task;
-#endif
       parent_team->t.t_invoke = invoker;
       KMP_TEST_THEN_INC32((kmp_int32 *)&root->r.r_in_parallel);
       parent_team->t.t_active_level++;
       parent_team->t.t_level++;
 
       /* Change number of threads in the team if requested */
       if (master_set_numthreads) { // The parallel has num_threads clause
         if (master_set_numthreads < master_th->th.th_teams_size.nth) {
#if OMP_40_ENABLED
         if (!ap) {
           // revert change made in __kmpc_serialized_parallel()
           master_th->th.th_serial_team->t.t_level--;
 // Get args from parent team for teams construct
 
 #if OMPT_SUPPORT
           void *dummy;
           void **exit_runtime_p;
+          ompt_task_info_t *task_info;
 
           ompt_lw_taskteam_t lw_taskteam;
 
-          if (ompt_enabled) {
+          if (ompt_enabled.enabled) {
             __ompt_lw_taskteam_init(&lw_taskteam, master_th, gtid,
-                                    unwrapped_task, ompt_parallel_id);
-            lw_taskteam.ompt_task_info.task_id = __ompt_task_id_new(gtid);
-            exit_runtime_p =
-                &(lw_taskteam.ompt_task_info.frame.exit_runtime_frame);
+                                    &ompt_parallel_data, return_address);
 
-            __ompt_lw_taskteam_link(&lw_taskteam, master_th);
+            __ompt_lw_taskteam_link(&lw_taskteam, master_th, 0);
+            // don't use lw_taskteam after linking. content was swaped
 
-#if OMPT_TRACE
-            my_task_id = lw_taskteam.ompt_task_info.task_id;
-            if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-              ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(
-                  ompt_parallel_id, my_task_id);
+            task_info = OMPT_CUR_TASK_INFO(master_th);
+            exit_runtime_p = &(task_info->frame.exit_runtime_frame);
+            if (ompt_enabled.ompt_callback_implicit_task) {
+              ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                  ompt_scope_begin, OMPT_CUR_TEAM_DATA(master_th),
+                  &(task_info->task_data), 1, __kmp_tid_from_gtid(gtid));
             }
-#endif
 
             /* OMPT state */
-            master_th->th.ompt_thread_info.state = ompt_state_work_parallel;
+            master_th->th.ompt_thread_info.state = omp_state_work_parallel;
           } else {
             exit_runtime_p = &dummy;
           }
 #endif
 
           {
             KMP_TIME_PARTITIONED_BLOCK(OMP_parallel);
             KMP_SET_THREAD_STATE_BLOCK(IMPLICIT_TASK);
             __kmp_invoke_microtask(microtask, gtid, 0, argc,
                                    parent_team->t.t_argv
 #if OMPT_SUPPORT
                                    ,
                                    exit_runtime_p
 #endif
                                    );
           }
 
 #if OMPT_SUPPORT
-          *exit_runtime_p = NULL;
-          if (ompt_enabled) {
-            lw_taskteam.ompt_task_info.frame.exit_runtime_frame = NULL;
-
-#if OMPT_TRACE
-            if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-              ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-                  ompt_parallel_id, ompt_task_id);
+          if (ompt_enabled.enabled) {
+            exit_runtime_p = NULL;
+            if (ompt_enabled.ompt_callback_implicit_task) {
+              ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                  ompt_scope_end, NULL, &(task_info->task_data), 1,
+                  __kmp_tid_from_gtid(gtid));
             }
-#endif
 
             __ompt_lw_taskteam_unlink(master_th);
-            // reset clear the task id only after unlinking the task
-            lw_taskteam.ompt_task_info.task_id = ompt_task_id_none;
-
-            if (ompt_callbacks.ompt_callback(ompt_event_parallel_end)) {
-              ompt_callbacks.ompt_callback(ompt_event_parallel_end)(
-                  ompt_parallel_id, ompt_task_id, OMPT_INVOKER(call_context));
+            if (ompt_enabled.ompt_callback_parallel_end) {
+              ompt_callbacks.ompt_callback(ompt_callback_parallel_end)(
+                  OMPT_CUR_TEAM_DATA(master_th), parent_task_data,
+                  OMPT_INVOKER(call_context), return_address);
             }
-            master_th->th.ompt_thread_info.state = ompt_state_overhead;
+            master_th->th.ompt_thread_info.state = omp_state_overhead;
           }
 #endif
         } else if (microtask == (microtask_t)__kmp_teams_master) {
           KMP_DEBUG_ASSERT(master_th->th.th_team ==
                            master_th->th.th_serial_team);
           team = master_th->th.th_team;
           // team->t.t_pkfn = microtask;
           team->t.t_invoke = invoker;

 #else
           *argv++ = va_arg(ap, void *);
 #endif
           KMP_MB();
 
 #if OMPT_SUPPORT
           void *dummy;
           void **exit_runtime_p;
+          ompt_task_info_t *task_info;
 
           ompt_lw_taskteam_t lw_taskteam;
 
-          if (ompt_enabled) {
+          if (ompt_enabled.enabled) {
             __ompt_lw_taskteam_init(&lw_taskteam, master_th, gtid,
-                                    unwrapped_task, ompt_parallel_id);
-            lw_taskteam.ompt_task_info.task_id = __ompt_task_id_new(gtid);
-            exit_runtime_p =
-                &(lw_taskteam.ompt_task_info.frame.exit_runtime_frame);
-
-            __ompt_lw_taskteam_link(&lw_taskteam, master_th);
+                                    &ompt_parallel_data, return_address);
+            __ompt_lw_taskteam_link(&lw_taskteam, master_th, 0);
+            // don't use lw_taskteam after linking. content was swaped
+            task_info = OMPT_CUR_TASK_INFO(master_th);
+            exit_runtime_p = &(task_info->frame.exit_runtime_frame);
 
-#if OMPT_TRACE
             /* OMPT implicit task begin */
-            my_task_id = lw_taskteam.ompt_task_info.task_id;
-            my_parallel_id = ompt_parallel_id;
-            if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-              ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(
-                  my_parallel_id, my_task_id);
+            implicit_task_data = OMPT_CUR_TASK_DATA(master_th);
+            if (ompt_enabled.ompt_callback_implicit_task) {
+              ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                  ompt_scope_begin, OMPT_CUR_TEAM_DATA(master_th),
+                  implicit_task_data, 1, __kmp_tid_from_gtid(gtid));
             }
-#endif
 
             /* OMPT state */
-            master_th->th.ompt_thread_info.state = ompt_state_work_parallel;
+            master_th->th.ompt_thread_info.state = omp_state_work_parallel;
           } else {
             exit_runtime_p = &dummy;
           }
 #endif
 
           {
             KMP_TIME_PARTITIONED_BLOCK(OMP_parallel);
             KMP_SET_THREAD_STATE_BLOCK(IMPLICIT_TASK);
             __kmp_invoke_microtask(microtask, gtid, 0, argc, args
 #if OMPT_SUPPORT
                                    ,
                                    exit_runtime_p
 #endif
                                    );
           }
 
 #if OMPT_SUPPORT
+          if (ompt_enabled.enabled) {
             *exit_runtime_p = NULL;
-          if (ompt_enabled) {
-#if OMPT_TRACE
-            lw_taskteam.ompt_task_info.frame.exit_runtime_frame = NULL;
-
-            if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-              ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-                  my_parallel_id, my_task_id);
+            if (ompt_enabled.ompt_callback_implicit_task) {
+              ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+                  ompt_scope_end, NULL, &(task_info->task_data), 1,
+                  __kmp_tid_from_gtid(gtid));
             }
-#endif
 
+            ompt_parallel_data = *OMPT_CUR_TEAM_DATA(master_th);
             __ompt_lw_taskteam_unlink(master_th);
-            // reset clear the task id only after unlinking the task
-            lw_taskteam.ompt_task_info.task_id = ompt_task_id_none;
-
-            if (ompt_callbacks.ompt_callback(ompt_event_parallel_end)) {
-              ompt_callbacks.ompt_callback(ompt_event_parallel_end)(
-                  ompt_parallel_id, ompt_task_id, OMPT_INVOKER(call_context));
+            if (ompt_enabled.ompt_callback_parallel_end) {
+              ompt_callbacks.ompt_callback(ompt_callback_parallel_end)(
+                  &ompt_parallel_data, parent_task_data,
+                  OMPT_INVOKER(call_context), return_address);
             }
-            master_th->th.ompt_thread_info.state = ompt_state_overhead;
+            master_th->th.ompt_thread_info.state = omp_state_overhead;
           }
 #endif
 #if OMP_40_ENABLED
         }
 #endif /* OMP_40_ENABLED */
       } else if (call_context == fork_context_gnu) {
 #if OMPT_SUPPORT
-        ompt_lw_taskteam_t *lwt =
-            (ompt_lw_taskteam_t *)__kmp_allocate(sizeof(ompt_lw_taskteam_t));
-        __ompt_lw_taskteam_init(lwt, master_th, gtid, unwrapped_task,
-                                ompt_parallel_id);
-
-        lwt->ompt_task_info.task_id = __ompt_task_id_new(gtid);
-        lwt->ompt_task_info.frame.exit_runtime_frame = NULL;
-        __ompt_lw_taskteam_link(lwt, master_th);
+        ompt_lw_taskteam_t lwt;
+        __ompt_lw_taskteam_init(&lwt, master_th, gtid, &ompt_parallel_data,
+                                return_address);
+
+        lwt.ompt_task_info.frame.exit_runtime_frame = NULL;
+        __ompt_lw_taskteam_link(&lwt, master_th, 1);
+// don't use lw_taskteam after linking. content was swaped
 #endif
 
         // we were called from GNU native code
         KA_TRACE(20, ("__kmp_fork_call: T#%d serial exit\n", gtid));
         return FALSE;
       } else {
         KMP_ASSERT2(call_context < fork_context_last,
                     "__kmp_fork_call: unknown fork_context parameter");
#if OMP_40_ENABLED
         new_icvs.proc_bind = proc_bind_icv;
       }
 #endif /* OMP_40_ENABLED */
 
       /* allocate a new parallel team */
       KF_TRACE(10, ("__kmp_fork_call: before __kmp_allocate_team\n"));
       team = __kmp_allocate_team(root, nthreads, nthreads,
 #if OMPT_SUPPORT
-                                 ompt_parallel_id,
+                                 ompt_parallel_data,
 #endif
 #if OMP_40_ENABLED
                                  proc_bind,
 #endif
                                  &new_icvs, argc USE_NESTED_HOT_ARG(master_th));
     } else {
       /* allocate a new parallel team */
       KF_TRACE(10, ("__kmp_fork_call: before __kmp_allocate_team\n"));
       team = __kmp_allocate_team(root, nthreads, nthreads,
 #if OMPT_SUPPORT
-                                 ompt_parallel_id,
+                                 ompt_parallel_data,
 #endif
 #if OMP_40_ENABLED
                                  proc_bind,
 #endif
                                  &master_th->th.th_current_task->td_icvs,
                                  argc USE_NESTED_HOT_ARG(master_th));
     }
     KF_TRACE(
         10, ("__kmp_fork_call: after __kmp_allocate_team - team = %p\n", team));
 
     /* setup the new team */
     KMP_CHECK_UPDATE(team->t.t_master_tid, master_tid);
     KMP_CHECK_UPDATE(team->t.t_master_this_cons, master_this_cons);
     KMP_CHECK_UPDATE(team->t.t_ident, loc);
     KMP_CHECK_UPDATE(team->t.t_parent, parent_team);
     KMP_CHECK_UPDATE_SYNC(team->t.t_pkfn, microtask);
 #if OMPT_SUPPORT
-    KMP_CHECK_UPDATE_SYNC(team->t.ompt_team_info.microtask, unwrapped_task);
+    KMP_CHECK_UPDATE_SYNC(team->t.ompt_team_info.master_return_address,
+                          return_address);
 #endif
     KMP_CHECK_UPDATE(team->t.t_invoke, invoker); // TODO move to root, maybe
 // TODO: parent_team->t.t_level == INT_MAX ???
 #if OMP_40_ENABLED
     if (!master_th->th.th_teams_microtask || level > teams_level) {
 #endif /* OMP_40_ENABLED */
       int new_level = parent_team->t.t_level + 1;
       KMP_CHECK_UPDATE(team->t.t_level, new_level);
#endif /* OMP_40_ENABLED */
     if (!root->r.r_active) // Only do assignment if it prevents cache ping-pong
       root->r.r_active = TRUE;
 
     __kmp_fork_team_threads(root, team, master_th, gtid);
     __kmp_setup_icv_copy(team, nthreads,
                          &master_th->th.th_current_task->td_icvs, loc);
 
 #if OMPT_SUPPORT
-    master_th->th.ompt_thread_info.state = ompt_state_work_parallel;
+    master_th->th.ompt_thread_info.state = omp_state_work_parallel;
 #endif
 
     __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock);
 
 #if USE_ITT_BUILD
     if (team->t.t_active_level == 1 // only report frames at level 1
 #if OMP_40_ENABLED
         && !master_th->th.th_teams_microtask // not in teams construct
#endif /* OMP_40_ENABLED */
   }
   KA_TRACE(20, ("__kmp_fork_call: T#%d(%d:0) done microtask = %p\n", gtid,
                 team->t.t_id, team->t.t_pkfn));
   KMP_MB(); /* Flush all pending memory write invalidates.  */
 
   KA_TRACE(20, ("__kmp_fork_call: parallel exit T#%d\n", gtid));
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    master_th->th.ompt_thread_info.state = ompt_state_overhead;
+  if (ompt_enabled.enabled) {
+    master_th->th.ompt_thread_info.state = omp_state_overhead;
   }
 #endif
 
   return TRUE;
 }
 
 #if OMPT_SUPPORT
 static inline void __kmp_join_restore_state(kmp_info_t *thread,
                                             kmp_team_t *team) {
   // restore state outside the region
   thread->th.ompt_thread_info.state =
-      ((team->t.t_serialized) ? ompt_state_work_serial
-                              : ompt_state_work_parallel);
+      ((team->t.t_serialized) ? omp_state_work_serial
+                              : omp_state_work_parallel);
 }
 
-static inline void __kmp_join_ompt(kmp_info_t *thread, kmp_team_t *team,
-                                   ompt_parallel_id_t parallel_id,
-                                   fork_context_e fork_context) {
-  ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-  if (ompt_callbacks.ompt_callback(ompt_event_parallel_end)) {
-    ompt_callbacks.ompt_callback(ompt_event_parallel_end)(
-        parallel_id, task_info->task_id, OMPT_INVOKER(fork_context));
+static inline void __kmp_join_ompt(int gtid, kmp_info_t *thread,
+                                   kmp_team_t *team, ompt_data_t *parallel_data,
+                                   fork_context_e fork_context, void *codeptr) {
+  ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
+  if (ompt_enabled.ompt_callback_parallel_end) {
+    ompt_callbacks.ompt_callback(ompt_callback_parallel_end)(
+        parallel_data, &(task_info->task_data), OMPT_INVOKER(fork_context),
+        codeptr);
   }
 
   task_info->frame.reenter_runtime_frame = NULL;
   __kmp_join_restore_state(thread, team);
 }
 #endif
 
 void __kmp_join_call(ident_t *loc, int gtid
#endif /* OMP_40_ENABLED */
   master_th = __kmp_threads[gtid];
   root = master_th->th.th_root;
   team = master_th->th.th_team;
   parent_team = team->t.t_parent;
 
   master_th->th.th_ident = loc;
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    master_th->th.ompt_thread_info.state = ompt_state_overhead;
+  if (ompt_enabled.enabled) {
+    master_th->th.ompt_thread_info.state = omp_state_overhead;
   }
 #endif
 
 #if KMP_DEBUG
   if (__kmp_tasking_mode != tskm_immediate_exec && !exit_teams) {
     KA_TRACE(20, ("__kmp_join_call: T#%d, old team = %p old task_team = %p, "
                   "th_task_team = %p\n",
                   __kmp_gtid_from_thread(master_th), team,
if (master_th->th.th_teams_microtask) {
         // __kmpc_end_serialized_parallel
         team->t.t_serialized++;
       }
     }
 #endif /* OMP_40_ENABLED */
     __kmpc_end_serialized_parallel(loc, gtid);
 
 #if OMPT_SUPPORT
-    if (ompt_enabled) {
+    if (ompt_enabled.enabled) {
       __kmp_join_restore_state(master_th, parent_team);
     }
 #endif
 
     return;
   }
 
   master_active = team->t.t_master_active;
else {
     master_th->th.th_task_state =
         0; // AC: no tasking in teams (out of any parallel)
   }
 #endif /* OMP_40_ENABLED */
 
   KMP_MB();
 
 #if OMPT_SUPPORT
-  ompt_parallel_id_t parallel_id = team->t.ompt_team_info.parallel_id;
+  ompt_data_t *parallel_data = &(team->t.ompt_team_info.parallel_data);
+  void *codeptr = team->t.ompt_team_info.master_return_address;
 #endif
 
 #if USE_ITT_BUILD
   if (__itt_stack_caller_create_ptr) {
     __kmp_itt_stack_caller_destroy(
         (__itt_caller)team->t
             .t_stack_id); // destroy the stack stitching id after join barrier
   }
#endif
         if (__kmp_tasking_mode != tskm_immediate_exec) {
           // Synchronize thread's task state
           other_threads[i]->th.th_task_state = master_th->th.th_task_state;
         }
       }
     }
 
 #if OMPT_SUPPORT
-    if (ompt_enabled) {
-      __kmp_join_ompt(master_th, parent_team, parallel_id, fork_context);
+    if (ompt_enabled.enabled) {
+      __kmp_join_ompt(gtid, master_th, parent_team, parallel_data, fork_context,
+                      codeptr);
     }
 #endif
 
     return;
   }
 #endif /* OMP_40_ENABLED */
 
   /* do cleanup and restore the parent team */
if (!master_th->th.th_teams_microtask ||
       team->t.t_level > master_th->th.th_teams_level)
 #endif /* OMP_40_ENABLED */
   {
     /* Decrement our nested depth level */
     KMP_TEST_THEN_DEC32((kmp_int32 *)&root->r.r_in_parallel);
   }
   KMP_DEBUG_ASSERT(root->r.r_in_parallel >= 0);
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled) {
-    ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
-    if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-      ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-          parallel_id, task_info->task_id);
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
+    if (ompt_enabled.ompt_callback_implicit_task) {
+      int ompt_team_size = team->t.t_nproc;
+      ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+          ompt_scope_end, NULL, &(task_info->task_data), ompt_team_size,
+          __kmp_tid_from_gtid(gtid));
     }
+
     task_info->frame.exit_runtime_frame = NULL;
-    task_info->task_id = 0;
+    task_info->task_data = ompt_data_none;
   }
 #endif
 
   KF_TRACE(10, ("__kmp_join_call1: T#%d, this_thread=%p team=%p\n", 0,
                 master_th, team));
   __kmp_pop_current_task_from_thread(master_th);
 
 #if OMP_40_ENABLED && KMP_AFFINITY_SUPPORTED
#endif /* OMP_40_ENABLED */
   // TODO: GEH - cannot do this assertion because root thread not set up as
   // executing
   // KMP_ASSERT( master_th->th.th_current_task->td_flags.executing == 0 );
   master_th->th.th_current_task->td_flags.executing = 1;
 
   __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock);
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    __kmp_join_ompt(master_th, parent_team, parallel_id, fork_context);
+  if (ompt_enabled.enabled) {
+    __kmp_join_ompt(gtid, master_th, parent_team, parallel_data, fork_context,
+                    codeptr);
   }
 #endif
 
   KMP_MB();
   KA_TRACE(20, ("__kmp_join_call: exit T#%d\n", gtid));
 }
 
 /* Check whether we should push an internal control record onto the
static void __kmp_initialize_root(kmp_root_t *root) {
   /* allocate the root team structure */
   KF_TRACE(10, ("__kmp_initialize_root: before root_team\n"));
 
   root_team =
       __kmp_allocate_team(root,
                           1, // new_nproc
                           1, // max_nproc
 #if OMPT_SUPPORT
-                          0, // root parallel id
+                          ompt_data_none, // root parallel id
 #endif
 #if OMP_40_ENABLED
                           __kmp_nested_proc_bind.bind_types[0],
 #endif
                           &r_icvs,
                           0 // argc
                           USE_NESTED_HOT_ARG(NULL) // master thread is unknown
                           );
#endif
   /* allocate the hot team structure */
   KF_TRACE(10, ("__kmp_initialize_root: before hot_team\n"));
 
   hot_team =
       __kmp_allocate_team(root,
                           1, // new_nproc
                           __kmp_dflt_team_nth_ub * 2, // max_nproc
 #if OMPT_SUPPORT
-                          0, // root parallel id
+                          ompt_data_none, // root parallel id
 #endif
 #if OMP_40_ENABLED
                           __kmp_nested_proc_bind.bind_types[0],
 #endif
                           &r_icvs,
                           0 // argc
                           USE_NESTED_HOT_ARG(NULL) // master thread is unknown
                           );
#endif
   if (root->r.r_uber_thread) {
     root_thread = root->r.r_uber_thread;
   } else {
     root_thread = (kmp_info_t *)__kmp_allocate(sizeof(kmp_info_t));
     if (__kmp_storage_map) {
       __kmp_print_thread_storage_map(root_thread, gtid);
     }
     root_thread->th.th_info.ds.ds_gtid = gtid;
+#if OMPT_SUPPORT
+    root_thread->th.ompt_thread_info.thread_data.ptr = NULL;
+#endif
     root_thread->th.th_root = root;
     if (__kmp_env_consistency_check) {
       root_thread->th.th_cons = __kmp_allocate_cons_stack(gtid);
     }
 #if USE_FAST_MEMORY
     __kmp_initialize_fast_memory(root_thread);
 #endif /* USE_FAST_MEMORY */
 
 #if KMP_USE_BGET
     KMP_DEBUG_ASSERT(root_thread->th.th_local.bget_data == NULL);
     __kmp_initialize_bget(root_thread);
 #endif
     __kmp_init_random(root_thread); // Initialize random number generator
   }
 
   /* setup the serial team held in reserve by the root thread */
   if (!root_thread->th.th_serial_team) {
     kmp_internal_control_t r_icvs = __kmp_get_global_icvs();
     KF_TRACE(10, ("__kmp_register_root: before serial_team\n"));
     root_thread->th.th_serial_team =
         __kmp_allocate_team(root, 1, 1,
 #if OMPT_SUPPORT
-                            0, // root parallel id
+                            ompt_data_none, // root parallel id
 #endif
 #if OMP_40_ENABLED
                             proc_bind_default,
 #endif
                             &r_icvs, 0 USE_NESTED_HOT_ARG(NULL));
   }
   KMP_ASSERT(root_thread->th.th_serial_team);
   KF_TRACE(10, ("__kmp_register_root: after serial_team = %p\n",
#endif
 
   if (TCR_4(__kmp_init_middle)) {
     __kmp_affinity_set_init_mask(gtid, TRUE);
   }
 #endif /* KMP_AFFINITY_SUPPORTED */
 
   __kmp_root_counter++;
 
+#if OMPT_SUPPORT
+  if (!initial_thread && ompt_enabled.enabled) {
+
+    ompt_thread_t *root_thread = ompt_get_thread();
+
+    ompt_set_thread_state(root_thread, omp_state_overhead);
+
+    if (ompt_enabled.ompt_callback_thread_begin) {
+      ompt_callbacks.ompt_callback(ompt_callback_thread_begin)(
+          ompt_thread_initial, __ompt_get_thread_data_internal());
+    }
+    ompt_data_t *task_data;
+    __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL, NULL);
+    if (ompt_enabled.ompt_callback_task_create) {
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          NULL, NULL, task_data, ompt_task_initial, 0, NULL);
+      // initial task has nothing to return to
+    }
+
+    ompt_set_thread_state(root_thread, omp_state_work_serial);
+  }
+#endif
+
   KMP_MB();
   __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock);
 
   return gtid;
 }
 
 #if KMP_NESTED_HOT_TEAMS
 static int __kmp_free_hot_teams(kmp_root_t *root, kmp_info_t *thr, int level,
KA_TRACE(
       10, ("__kmp_reset_root: free handle, th = %p, handle = %" KMP_UINTPTR_SPEC
            "\n",
            (LPVOID) & (root->r.r_uber_thread->th),
            root->r.r_uber_thread->th.th_info.ds.ds_thread));
   __kmp_free_handle(root->r.r_uber_thread->th.th_info.ds.ds_thread);
 #endif /* KMP_OS_WINDOWS */
 
 #if OMPT_SUPPORT
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_thread_end)) {
-    int gtid = __kmp_get_gtid();
-    __ompt_thread_end(ompt_thread_initial, gtid);
+  if (ompt_enabled.ompt_callback_thread_end) {
+    ompt_callbacks.ompt_callback(ompt_callback_thread_end)(
+        &(root->r.r_uber_thread->th.ompt_thread_info.thread_data));
   }
 #endif
 
   TCW_4(__kmp_nth,
         __kmp_nth - 1); // __kmp_reap_thread will decrement __kmp_all_nth.
   root->r.r_cg_nthreads--;
 
   __kmp_reap_thread(root->r.r_uber_thread, 1);
#if OMP_45_ENABLED
   kmp_info_t *thread = __kmp_threads[gtid];
   kmp_team_t *team = thread->th.th_team;
   kmp_task_team_t *task_team = thread->th.th_task_team;
 
   // we need to wait for the proxy tasks before finishing the thread
   if (task_team != NULL && task_team->tt.tt_found_proxy_tasks) {
 #if OMPT_SUPPORT
     // the runtime is shutting down so we won't report any events
-    thread->th.ompt_thread_info.state = ompt_state_undefined;
+    thread->th.ompt_thread_info.state = omp_state_undefined;
 #endif
     __kmp_task_team_wait(thread, team USE_ITT_BUILD_ARG(NULL));
   }
 #endif
 
   __kmp_reset_root(gtid, root);
 
   /* free up this thread slot */
#endif
 
   // add the reserve serialized team, initialized from the team's master thread
   {
     kmp_internal_control_t r_icvs = __kmp_get_x_global_icvs(team);
     KF_TRACE(10, ("__kmp_allocate_thread: before th_serial/serial_team\n"));
     new_thr->th.th_serial_team = serial_team =
         (kmp_team_t *)__kmp_allocate_team(root, 1, 1,
 #if OMPT_SUPPORT
-                                          0, // root parallel id
+                                          ompt_data_none, // root parallel id
 #endif
 #if OMP_40_ENABLED
                                           proc_bind_default,
 #endif
                                           &r_icvs, 0 USE_NESTED_HOT_ARG(NULL));
   }
   KMP_ASSERT(serial_team);
   serial_team->t.t_serialized = 0; // AC: the team created in reserve, not for

 
 #endif /* OMP_40_ENABLED && KMP_AFFINITY_SUPPORTED */
 
 /* allocate a new team data structure to use.  take one off of the free pool if
    available */
 kmp_team_t *
 __kmp_allocate_team(kmp_root_t *root, int new_nproc, int max_nproc,
 #if OMPT_SUPPORT
-                    ompt_parallel_id_t ompt_parallel_id,
+                    ompt_data_t ompt_parallel_data,
 #endif
 #if OMP_40_ENABLED
                     kmp_proc_bind_t new_proc_bind,
 #endif
                     kmp_internal_control_t *new_icvs,
                     int argc USE_NESTED_HOT_ARG(kmp_info_t *master)) {
   KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_allocate_team);
   int f;
#if KMP_DEBUG
     if (__kmp_tasking_mode != tskm_immediate_exec) {
       KA_TRACE(20, ("__kmp_allocate_team: hot team task_team[0] = %p "
                     "task_team[1] = %p after reinit\n",
                     team->t.t_task_team[0], team->t.t_task_team[1]));
     }
 #endif
 
 #if OMPT_SUPPORT
-    __ompt_team_assign_id(team, ompt_parallel_id);
+    __ompt_team_assign_id(team, ompt_parallel_data);
 #endif
 
     KMP_MB();
 
     return team;
   }
 
   /* next, let's try to take one from the team pool */

 #if OMP_40_ENABLED
       team->t.t_proc_bind = new_proc_bind;
 #endif
 
       KA_TRACE(20, ("__kmp_allocate_team: using team from pool %d.\n",
                     team->t.t_id));
 
 #if OMPT_SUPPORT
-      __ompt_team_assign_id(team, ompt_parallel_id);
+      __ompt_team_assign_id(team, ompt_parallel_data);
 #endif
 
       KMP_MB();
 
       return team;
     }
 
     /* reap team if it is too small, then loop back and check the next one */
#endif
     }
   }
 
 #if OMP_40_ENABLED
   team->t.t_proc_bind = new_proc_bind;
 #endif
 
 #if OMPT_SUPPORT
-  __ompt_team_assign_id(team, ompt_parallel_id);
+  __ompt_team_assign_id(team, ompt_parallel_data);
   team->t.ompt_serialized_team_info = NULL;
 #endif
 
   KMP_MB();
 
   KA_TRACE(20, ("__kmp_allocate_team: done creating a new team %d.\n",
                 team->t.t_id));
 
void *__kmp_launch_thread(kmp_info_t *this_thr) {
   KMP_MB();
   KA_TRACE(10, ("__kmp_launch_thread: T#%d start\n", gtid));
 
   if (__kmp_env_consistency_check) {
     this_thr->th.th_cons = __kmp_allocate_cons_stack(gtid); // ATT: Memory leak?
   }
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
-    this_thr->th.ompt_thread_info.state = ompt_state_overhead;
+  ompt_data_t *thread_data;
+  if (ompt_enabled.enabled) {
+    thread_data = &(this_thr->th.ompt_thread_info.thread_data);
+    thread_data->ptr = NULL;
+
+    this_thr->th.ompt_thread_info.state = omp_state_overhead;
     this_thr->th.ompt_thread_info.wait_id = 0;
-    this_thr->th.ompt_thread_info.idle_frame = __builtin_frame_address(0);
-    if (ompt_callbacks.ompt_callback(ompt_event_thread_begin)) {
-      __ompt_thread_begin(ompt_thread_worker, gtid);
+    this_thr->th.ompt_thread_info.idle_frame = OMPT_GET_FRAME_ADDRESS(0);
+    if (ompt_enabled.ompt_callback_thread_begin) {
+      ompt_callbacks.ompt_callback(ompt_callback_thread_begin)(
+          ompt_thread_worker, thread_data);
     }
   }
 #endif
 
+#if OMPT_SUPPORT
+  if (ompt_enabled.enabled) {
+    this_thr->th.ompt_thread_info.state = omp_state_idle;
+  }
+#endif
   /* This is the place where threads wait for work */
   while (!TCR_4(__kmp_global.g.g_done)) {
     KMP_DEBUG_ASSERT(this_thr == __kmp_threads[gtid]);
     KMP_MB();
 
     /* wait for work to do */
     KA_TRACE(20, ("__kmp_launch_thread: T#%d waiting for work\n", gtid));
 
-#if OMPT_SUPPORT
-    if (ompt_enabled) {
-      this_thr->th.ompt_thread_info.state = ompt_state_idle;
-    }
-#endif
-
     /* No tid yet since not part of a team */
     __kmp_fork_barrier(gtid, KMP_GTID_DNE);
 
 #if OMPT_SUPPORT
-    if (ompt_enabled) {
-      this_thr->th.ompt_thread_info.state = ompt_state_overhead;
+    if (ompt_enabled.enabled) {
+      this_thr->th.ompt_thread_info.state = omp_state_overhead;
     }
 #endif
 
     pteam = (kmp_team_t * (*))(&this_thr->th.th_team);
 
     /* have we been allocated? */
     if (TCR_SYNC_PTR(*pteam) && !TCR_4(__kmp_global.g.g_done)) {
-#if OMPT_SUPPORT
-      ompt_task_info_t *task_info;
-      ompt_parallel_id_t my_parallel_id;
-      if (ompt_enabled) {
-        task_info = __ompt_get_taskinfo(0);
-        my_parallel_id = (*pteam)->t.ompt_team_info.parallel_id;
-      }
-#endif
       /* we were just woken up, so run our new task */
       if (TCR_SYNC_PTR((*pteam)->t.t_pkfn) != NULL) {
         int rc;
         KA_TRACE(20,
                  ("__kmp_launch_thread: T#%d(%d:%d) invoke microtask = %p\n",
                   gtid, (*pteam)->t.t_id, __kmp_tid_from_gtid(gtid),
                   (*pteam)->t.t_pkfn));
 
         updateHWFPControl(*pteam);
 
 #if OMPT_SUPPORT
-        if (ompt_enabled) {
-          this_thr->th.ompt_thread_info.state = ompt_state_work_parallel;
-          // Initialize OMPT task id for implicit task.
-          int tid = __kmp_tid_from_gtid(gtid);
-          task_info->task_id = __ompt_task_id_new(tid);
+        if (ompt_enabled.enabled) {
+          this_thr->th.ompt_thread_info.state = omp_state_work_parallel;
         }
 #endif
 
         {
           KMP_TIME_PARTITIONED_BLOCK(OMP_parallel);
           KMP_SET_THREAD_STATE_BLOCK(IMPLICIT_TASK);
           rc = (*pteam)->t.t_invoke(gtid);
         }
         KMP_ASSERT(rc);
 
-#if OMPT_SUPPORT
-        if (ompt_enabled) {
-          /* no frame set while outside task */
-          task_info->frame.exit_runtime_frame = NULL;
-
-          this_thr->th.ompt_thread_info.state = ompt_state_overhead;
-        }
-#endif
         KMP_MB();
         KA_TRACE(20, ("__kmp_launch_thread: T#%d(%d:%d) done microtask = %p\n",
                       gtid, (*pteam)->t.t_id, __kmp_tid_from_gtid(gtid),
                       (*pteam)->t.t_pkfn));
       }
-      /* join barrier after parallel region */
-      __kmp_join_barrier(gtid);
-#if OMPT_SUPPORT && OMPT_TRACE
-      if (ompt_enabled) {
-        if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
-          // don't access *pteam here: it may have already been freed
-          // by the master thread behind the barrier (possible race)
-          ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
-              my_parallel_id, task_info->task_id);
-        }
-        task_info->frame.exit_runtime_frame = NULL;
-        task_info->task_id = 0;
+#if OMPT_SUPPORT
+      if (ompt_enabled.enabled) {
+        /* no frame set while outside task */
+        __ompt_get_task_info_object(0)->frame.exit_runtime_frame = NULL;
+
+        this_thr->th.ompt_thread_info.state = omp_state_overhead;
+        this_thr->th.ompt_thread_info.task_data = *OMPT_CUR_TASK_DATA(this_thr);
       }
 #endif
+      /* join barrier after parallel region */
+      __kmp_join_barrier(gtid);
     }
   }
   TCR_SYNC_PTR((intptr_t)__kmp_global.g.g_done);
 
 #if OMPT_SUPPORT
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_thread_end)) {
-    __ompt_thread_end(ompt_thread_worker, gtid);
+  if (ompt_enabled.ompt_callback_thread_end) {
+    ompt_callbacks.ompt_callback(ompt_callback_thread_end)(thread_data);
   }
 #endif
 
   this_thr->th.th_task_team = NULL;
   /* run the destructors for the threadprivate data for this thread */
   __kmp_common_destroy_gtid(gtid);
 
   KA_TRACE(10, ("__kmp_launch_thread: T#%d done\n", gtid));
#endif
   if (i < __kmp_threads_capacity) {
 #if KMP_USE_MONITOR
     // 2009-09-08 (lev): Other alive roots found. Why do we kill the monitor??
     KMP_MB(); /* Flush all pending memory write invalidates.  */
 
     // Need to check that monitor was initialized before reaping it. If we are
     // called form __kmp_atfork_child (which sets __kmp_init_parallel = 0), then
     // __kmp_monitor will appear to contain valid data, but it is only valid in
     // the parent process, not the child.
     // New behavior (201008): instead of keying off of the flag
     // __kmp_init_parallel, the monitor thread creation is keyed off
     // of the new flag __kmp_init_monitor.
     __kmp_acquire_bootstrap_lock(&__kmp_monitor_lock);
     if (TCR_4(__kmp_init_monitor)) {
       __kmp_reap_monitor(&__kmp_monitor);
       TCW_4(__kmp_init_monitor, 0);
     }
     __kmp_release_bootstrap_lock(&__kmp_monitor_lock);

 #endif /* USE_ITT_BUILD */
 #if INCLUDE_SSC_MARKS
   SSC_MARK_INVOKING();
 #endif
 
 #if OMPT_SUPPORT
   void *dummy;
   void **exit_runtime_p;
-  ompt_task_id_t my_task_id;
-  ompt_parallel_id_t my_parallel_id;
+  ompt_data_t *my_task_data;
+  ompt_data_t *my_parallel_data;
+  int ompt_team_size;
 
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     exit_runtime_p = &(team->t.t_implicit_task_taskdata[tid]
                            .ompt_task_info.frame.exit_runtime_frame);
   } else {
     exit_runtime_p = &dummy;
   }
 
-#if OMPT_TRACE
-  my_task_id = team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id;
-  my_parallel_id = team->t.ompt_team_info.parallel_id;
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)) {
-    ompt_callbacks.ompt_callback(ompt_event_implicit_task_begin)(my_parallel_id,
-                                                                 my_task_id);
+  my_task_data =
+      &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data);
+  my_parallel_data = &(team->t.ompt_team_info.parallel_data);
+  if (ompt_enabled.ompt_callback_implicit_task) {
+    ompt_team_size = team->t.t_nproc;
+    ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+        ompt_scope_begin, my_parallel_data, my_task_data, ompt_team_size,
+        __kmp_tid_from_gtid(gtid));
   }
 #endif
-#endif
 
   {
     KMP_TIME_PARTITIONED_BLOCK(OMP_parallel);
     KMP_SET_THREAD_STATE_BLOCK(IMPLICIT_TASK);
     rc =
         __kmp_invoke_microtask((microtask_t)TCR_SYNC_PTR(team->t.t_pkfn), gtid,
                                tid, (int)team->t.t_argc, (void **)team->t.t_argv
 #if OMPT_SUPPORT
void __kmp_teams_master(int gtid) {
   KA_TRACE(20, ("__kmp_teams_master: T#%d, Tid %d, microtask %p\n", gtid,
                 __kmp_tid_from_gtid(gtid), thr->th.th_teams_microtask));
 // Launch league of teams now, but not let workers execute
 // (they hang on fork barrier until next parallel)
 #if INCLUDE_SSC_MARKS
   SSC_MARK_FORKING();
 #endif
   __kmp_fork_call(loc, gtid, fork_context_intel, team->t.t_argc,
-#if OMPT_SUPPORT
-                  (void *)thr->th.th_teams_microtask, // "unwrapped" task
-#endif
                   (microtask_t)thr->th.th_teams_microtask, // "wrapped" task
                   VOLATILE_CAST(launch_t) __kmp_invoke_task_func, NULL);
 #if INCLUDE_SSC_MARKS
   SSC_MARK_JOINING();
 #endif
 
   // AC: last parameter "1" eliminates join barrier which won't work because
   // worker threads are in a fork barrier waiting for more parallel regions
__kmp_printf("__kmp_threads[%d]->th.th_team_nproc=%d, TEAM: %p, "
                  team->t.t_nproc);
     __kmp_print_structure();
   }
   KMP_DEBUG_ASSERT(__kmp_threads[gtid] &&
                    __kmp_threads[gtid]->th.th_team_nproc == team->t.t_nproc);
 #endif /* KMP_DEBUG */
 
   __kmp_join_barrier(gtid); /* wait for everyone */
+#if OMPT_SUPPORT
+  int ds_tid = this_thr->th.th_info.ds.ds_tid;
+  if (this_thr->th.ompt_thread_info.state == omp_state_wait_barrier_implicit) {
+    ompt_data_t *tId = OMPT_CUR_TASK_DATA(this_thr);
+    ompt_data_t *pId = OMPT_CUR_TEAM_DATA(this_thr);
+    this_thr->th.ompt_thread_info.state = omp_state_overhead;
+#if OMPT_OPTIONAL
+    void *codeptr = NULL;
+    if (KMP_MASTER_TID(ds_tid) &&
+        (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) ||
+         ompt_callbacks.ompt_callback(ompt_callback_sync_region)))
+      codeptr = OMPT_CUR_TEAM_INFO(this_thr)->master_return_address;
+
+    if (ompt_enabled.ompt_callback_sync_region_wait) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_barrier, ompt_scope_end, pId, tId, codeptr);
+    }
+    if (ompt_enabled.ompt_callback_sync_region) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+          ompt_sync_region_barrier, ompt_scope_end, pId, tId, codeptr);
+    }
+#endif
+    if (!KMP_MASTER_TID(ds_tid) && ompt_enabled.ompt_callback_implicit_task) {
+      ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+          ompt_scope_end, NULL, tId, 0, ds_tid);
+    }
+    // return to idle state
+    this_thr->th.ompt_thread_info.state = omp_state_overhead;
+  }
+#endif
 
   KMP_MB(); /* Flush all pending memory write invalidates.  */
   KMP_ASSERT(this_thr->th.th_team == team);
 }
 
 /* ------------------------------------------------------------------------ */
 
 #ifdef USE_LOAD_BALANCE

runtime/src/kmp_sched.cpp

 #endif
 
 template <typename T>
 static void __kmp_for_static_init(ident_t *loc, kmp_int32 global_tid,
                                   kmp_int32 schedtype, kmp_int32 *plastiter,
                                   T *plower, T *pupper,
                                   typename traits_t<T>::signed_t *pstride,
                                   typename traits_t<T>::signed_t incr,
-                                  typename traits_t<T>::signed_t chunk) {
+                                  typename traits_t<T>::signed_t chunk
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                  ,
+                                  void *codeptr
+#endif
+                                  ) {
   KMP_COUNT_BLOCK(OMP_FOR_static);
   KMP_TIME_PARTITIONED_BLOCK(FOR_static_scheduling);
 
   typedef typename traits_t<T>::unsigned_t UT;
   typedef typename traits_t<T>::signed_t ST;
   /*  this all has to be changed back to TID and such.. */
   kmp_int32 gtid = global_tid;
   kmp_uint32 tid;
   kmp_uint32 nth;
   UT trip_count;
   kmp_team_t *team;
   kmp_info_t *th = __kmp_threads[gtid];
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT && OMPT_OPTIONAL
   ompt_team_info_t *team_info = NULL;
   ompt_task_info_t *task_info = NULL;
+  ompt_work_type_t ompt_work_type;
 
-  if (ompt_enabled) {
+  if (ompt_enabled.enabled) {
     // Only fully initialize variables needed by OMPT if OMPT is enabled.
     team_info = __ompt_get_teaminfo(0, NULL);
-    task_info = __ompt_get_taskinfo(0);
+    task_info = __ompt_get_task_info_object(0);
+    // Determine workshare type
+    if (loc != NULL) {
+      if ((loc->flags & KMP_IDENT_WORK_LOOP) != 0) {
+        ompt_work_type = ompt_work_loop;
+      } else if ((loc->flags & KMP_IDENT_WORK_SECTIONS) != 0) {
+        ompt_work_type = ompt_work_sections;
+      } else if ((loc->flags & KMP_IDENT_WORK_DISTRIBUTE) != 0) {
+        ompt_work_type = ompt_work_distribute;
+      } else {
+        KMP_ASSERT2(0,
+                    "__kmpc_for_static_init: can't determine workshare type");
+      }
+      KMP_DEBUG_ASSERT(ompt_work_type);
+    }
   }
 #endif
 
   KMP_DEBUG_ASSERT(plastiter && plower && pupper && pstride);
   KE_TRACE(10, ("__kmpc_for_static_init called (%d)\n", global_tid));
 #ifdef KMP_DEBUG
   {
     const char *buff;
#ifdef KMP_DEBUG
                               traits_t<ST>::spec, traits_t<T>::spec);
       KD_TRACE(100,
                (buff, *plastiter, *plower, *pupper, *pstride, loc->psource));
       __kmp_str_free(&buff);
     }
 #endif
     KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid));
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_loop_begin)(
-          team_info->parallel_id, task_info->task_id, team_info->microtask);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.ompt_callback_work) {
+      ompt_callbacks.ompt_callback(ompt_callback_work)(
+          ompt_work_type, ompt_scope_begin, &(team_info->parallel_data),
+          &(task_info->task_data), 0, codeptr);
     }
 #endif
     KMP_COUNT_VALUE(FOR_static_iterations, 0);
     return;
   }
 
 #if OMP_40_ENABLED
   // Although there are schedule enumerations above kmp_ord_upper which are not
#ifdef KMP_DEBUG
                               traits_t<T>::spec, traits_t<T>::spec,
                               traits_t<ST>::spec);
       KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride));
       __kmp_str_free(&buff);
     }
 #endif
     KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid));
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_loop_begin)(
-          team_info->parallel_id, task_info->task_id, team_info->microtask);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.ompt_callback_work) {
+      ompt_callbacks.ompt_callback(ompt_callback_work)(
+          ompt_work_type, ompt_scope_begin, &(team_info->parallel_data),
+          &(task_info->task_data), *pstride, codeptr);
     }
 #endif
     return;
   }
   nth = team->t.t_nproc;
   if (nth == 1) {
     if (plastiter != NULL)
       *plastiter = TRUE;
     *pstride =
         (incr > 0) ? (*pupper - *plower + 1) : (-(*plower - *pupper + 1));
 #ifdef KMP_DEBUG
     {
       const char *buff;
       // create format specifiers before the debug output
       buff = __kmp_str_format("__kmpc_for_static_init: (serial) liter=%%d "
                               "lower=%%%s upper=%%%s stride = %%%s\n",
                               traits_t<T>::spec, traits_t<T>::spec,
                               traits_t<ST>::spec);
       KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride));
       __kmp_str_free(&buff);
     }
 #endif
     KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid));
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_loop_begin)(
-          team_info->parallel_id, task_info->task_id, team_info->microtask);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (ompt_enabled.ompt_callback_work) {
+      ompt_callbacks.ompt_callback(ompt_callback_work)(
+          ompt_work_type, ompt_scope_begin, &(team_info->parallel_data),
+          &(task_info->task_data), *pstride, codeptr);
     }
 #endif
     return;
   }
 
   /* compute trip count */
   if (incr == 1) {
     trip_count = *pupper - *plower + 1;
buff = __kmp_str_format("__kmpc_for_static_init: liter=%%d lower=%%%s "
                             traits_t<T>::spec, traits_t<T>::spec,
                             traits_t<ST>::spec, traits_t<T>::spec);
     KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride));
     __kmp_str_free(&buff);
   }
 #endif
   KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid));
 
-#if OMPT_SUPPORT && OMPT_TRACE
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_loop_begin)) {
-    ompt_callbacks.ompt_callback(ompt_event_loop_begin)(
-        team_info->parallel_id, task_info->task_id, team_info->microtask);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_work) {
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_type, ompt_scope_begin, &(team_info->parallel_data),
+        &(task_info->task_data), trip_count, codeptr);
   }
 #endif
 
   return;
 }
 
 template <typename T>
 static void __kmp_dist_for_static_init(ident_t *loc, kmp_int32 gtid,

 
 @{
 */
 void __kmpc_for_static_init_4(ident_t *loc, kmp_int32 gtid, kmp_int32 schedtype,
                               kmp_int32 *plastiter, kmp_int32 *plower,
                               kmp_int32 *pupper, kmp_int32 *pstride,
                               kmp_int32 incr, kmp_int32 chunk) {
   __kmp_for_static_init<kmp_int32>(loc, gtid, schedtype, plastiter, plower,
-                                   pupper, pstride, incr, chunk);
+                                   pupper, pstride, incr, chunk
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                   ,
+                                   OMPT_GET_RETURN_ADDRESS(0)
+#endif
+                                       );
 }
 
 /*!
  See @ref __kmpc_for_static_init_4
  */
 void __kmpc_for_static_init_4u(ident_t *loc, kmp_int32 gtid,
                                kmp_int32 schedtype, kmp_int32 *plastiter,
                                kmp_uint32 *plower, kmp_uint32 *pupper,
                                kmp_int32 *pstride, kmp_int32 incr,
                                kmp_int32 chunk) {
   __kmp_for_static_init<kmp_uint32>(loc, gtid, schedtype, plastiter, plower,
-                                    pupper, pstride, incr, chunk);
+                                    pupper, pstride, incr, chunk
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                    ,
+                                    OMPT_GET_RETURN_ADDRESS(0)
+#endif
+                                        );
 }
 
 /*!
  See @ref __kmpc_for_static_init_4
  */
 void __kmpc_for_static_init_8(ident_t *loc, kmp_int32 gtid, kmp_int32 schedtype,
                               kmp_int32 *plastiter, kmp_int64 *plower,
                               kmp_int64 *pupper, kmp_int64 *pstride,
                               kmp_int64 incr, kmp_int64 chunk) {
   __kmp_for_static_init<kmp_int64>(loc, gtid, schedtype, plastiter, plower,
-                                   pupper, pstride, incr, chunk);
+                                   pupper, pstride, incr, chunk
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                   ,
+                                   OMPT_GET_RETURN_ADDRESS(0)
+#endif
+                                       );
 }
 
 /*!
  See @ref __kmpc_for_static_init_4
  */
 void __kmpc_for_static_init_8u(ident_t *loc, kmp_int32 gtid,
                                kmp_int32 schedtype, kmp_int32 *plastiter,
                                kmp_uint64 *plower, kmp_uint64 *pupper,
                                kmp_int64 *pstride, kmp_int64 incr,
                                kmp_int64 chunk) {
   __kmp_for_static_init<kmp_uint64>(loc, gtid, schedtype, plastiter, plower,
-                                    pupper, pstride, incr, chunk);
+                                    pupper, pstride, incr, chunk
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+                                    ,
+                                    OMPT_GET_RETURN_ADDRESS(0)
+#endif
+                                        );
 }
 /*!
 @}
 */
 
 /*!
 @ingroup WORK_SHARING
 @param    loc       Source code location

runtime/src/kmp_settings.cpp

if (msg != NULL) {
       __kmp_str_buf_print_size(&buf, *out);
       KMP_WARNING(ParseSizeIntWarn, name, value, msg);
       KMP_INFORM(Using_str_Value, name, buf.str);
       __kmp_str_buf_free(&buf);
     }
   }
 } // __kmp_stg_parse_size
 
-#if KMP_AFFINITY_SUPPORTED
 static void __kmp_stg_parse_str(char const *name, char const *value,
                                 char const **out) {
   __kmp_str_free(out);
   *out = __kmp_str_format("%s", value);
 } // __kmp_stg_parse_str
-#endif
 
 static void __kmp_stg_parse_int(
     char const
         *name, // I: Name of environment variable (used in warning messages).
     char const *value, // I: Value of environment variable to parse.
     int min, // I: Miminal allowed value.
     int max, // I: Maximum allowed value.
     int *out // O: Output (parsed) value.

 
 static void __kmp_stg_print_omp_cancellation(kmp_str_buf_t *buffer,
                                              char const *name, void *data) {
   __kmp_stg_print_bool(buffer, name, __kmp_omp_cancellation);
 } // __kmp_stg_print_omp_cancellation
 
 #endif
 
-// -----------------------------------------------------------------------------
+#if OMP_50_ENABLED && OMPT_SUPPORT
+
+static void __kmp_stg_parse_omp_tool_libraries(char const *name,
+                                               char const *value, void *data) {
+  __kmp_stg_parse_str(name, value, &__kmp_tool_libraries);
+} // __kmp_stg_parse_omp_tool_libraries
+
+static void __kmp_stg_print_omp_tool_libraries(kmp_str_buf_t *buffer,
+                                               char const *name, void *data) {
+  if (__kmp_tool_libraries)
+    __kmp_stg_print_str(buffer, name, __kmp_tool_libraries);
+  else {
+    if (__kmp_env_format) {
+      KMP_STR_BUF_PRINT_NAME;
+    } else {
+      __kmp_str_buf_print(buffer, "   %s", name);
+    }
+    __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined));
+  }
+} // __kmp_stg_print_omp_tool_libraries
+
+#endif
+
 // Table.
 
 static kmp_setting_t __kmp_stg_table[] = {
 
     {"KMP_ALL_THREADS", __kmp_stg_parse_device_thread_limit, NULL, NULL, 0, 0},
     {"KMP_BLOCKTIME", __kmp_stg_parse_blocktime, __kmp_stg_print_blocktime,
      NULL, 0, 0},
     {"KMP_DUPLICATE_LIB_OK", __kmp_stg_parse_duplicate_lib_ok,

 #endif
 
 #if OMP_40_ENABLED
     {"OMP_DISPLAY_ENV", __kmp_stg_parse_omp_display_env,
      __kmp_stg_print_omp_display_env, NULL, 0, 0},
     {"OMP_CANCELLATION", __kmp_stg_parse_omp_cancellation,
      __kmp_stg_print_omp_cancellation, NULL, 0, 0},
 #endif
+
+#if OMP_50_ENABLED && OMPT_SUPPORT
+    {"OMP_TOOL_LIBRARIES", __kmp_stg_parse_omp_tool_libraries,
+     __kmp_stg_print_omp_tool_libraries, NULL, 0, 0},
+#endif
+
     {"", NULL, NULL, NULL, 0, 0}}; // settings
 
 static int const __kmp_stg_count =
     sizeof(__kmp_stg_table) / sizeof(kmp_setting_t);
 
 static inline kmp_setting_t *__kmp_stg_find(char const *name) {
 
   int i;

runtime/src/kmp_taskdeps.cpp

 //
 //===----------------------------------------------------------------------===//
 
 //#define KMP_SUPPORT_GRAPH_OUTPUT 1
 
 #include "kmp.h"
 #include "kmp_io.h"
 #include "kmp_wait_release.h"
+#if OMPT_SUPPORT
+#include "ompt-specific.h"
+#endif
 
 #if OMP_40_ENABLED
 
 // TODO: Improve memory allocation? keep a list of pre-allocated structures?
 // allocate in blocks? re-use list finished list entries?
 // TODO: don't use atomic ref counters for stack-allocated nodes.
 // TODO: find an alternate to atomic refs for heap-allocated nodes?
 // TODO: Finish graph output support
#ifdef KMP_SUPPORT_GRAPH_OUTPUT
   // do not use sink->dn.task as that is only filled after the dependencies
   // are already processed!
   kmp_taskdata_t *task_sink = KMP_TASK_TO_TASKDATA(sink_task);
 
   __kmp_printf("%d(%s) -> %d(%s)\n", source->dn.id,
                task_source->td_ident->psource, sink->dn.id,
                task_sink->td_ident->psource);
 #endif
-#if OMPT_SUPPORT && OMPT_TRACE
-  // OMPT tracks dependences between task (a=source, b=sink) in which
-  // task a blocks the execution of b through the ompt_new_dependence_callback
-  if (ompt_enabled &&
-      ompt_callbacks.ompt_callback(ompt_event_task_dependence_pair)) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  /* OMPT tracks dependences between task (a=source, b=sink) in which
+     task a blocks the execution of b through the ompt_new_dependence_callback
+     */
+  if (ompt_enabled.ompt_callback_task_dependence) {
     kmp_taskdata_t *task_source = KMP_TASK_TO_TASKDATA(source->dn.task);
     kmp_taskdata_t *task_sink = KMP_TASK_TO_TASKDATA(sink_task);
 
-    ompt_callbacks.ompt_callback(ompt_event_task_dependence_pair)(
-        task_source->ompt_task_info.task_id, task_sink->ompt_task_info.task_id);
+    ompt_callbacks.ompt_callback(ompt_callback_task_dependence)(
+        &(task_source->ompt_task_info.task_data),
+        &(task_sink->ompt_task_info.task_data));
   }
-#endif /* OMPT_SUPPORT && OMPT_TRACE */
+#endif /* OMPT_SUPPORT && OMPT_OPTIONAL */
 }
 
 template <bool filter>
 static inline kmp_int32
 __kmp_process_deps(kmp_int32 gtid, kmp_depnode_t *node, kmp_dephash_t *hash,
                    bool dep_barrier, kmp_int32 ndeps,
                    kmp_depend_info_t *dep_list, kmp_task_t *task) {
   KA_TRACE(30, ("__kmp_process_deps<%d>: T#%d processing %d dependencies : "
kmp_int32 __kmpc_omp_task_with_deps(ident_t *loc_ref, kmp_int32 gtid,
 
   kmp_taskdata_t *new_taskdata = KMP_TASK_TO_TASKDATA(new_task);
   KA_TRACE(10, ("__kmpc_omp_task_with_deps(enter): T#%d loc=%p task=%p\n", gtid,
                 loc_ref, new_taskdata));
 
   kmp_info_t *thread = __kmp_threads[gtid];
   kmp_taskdata_t *current_task = thread->th.th_current_task;
 
-#if OMPT_SUPPORT && OMPT_TRACE
+#if OMPT_SUPPORT
+  OMPT_STORE_RETURN_ADDRESS(gtid);
+
+  if (ompt_enabled.enabled) {
+    if (ompt_enabled.ompt_callback_task_create) {
+      kmp_taskdata_t *parent = new_taskdata->td_parent;
+      ompt_data_t task_data = ompt_data_none;
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          parent ? &(parent->ompt_task_info.task_data) : &task_data,
+          parent ? &(parent->ompt_task_info.frame) : NULL,
+          &(new_taskdata->ompt_task_info.task_data),
+          ompt_task_explicit | TASK_TYPE_DETAILS_FORMAT(new_taskdata), 1,
+          OMPT_LOAD_RETURN_ADDRESS(gtid));
+    }
+
+    new_taskdata->ompt_task_info.frame.reenter_runtime_frame =
+        OMPT_GET_FRAME_ADDRESS(0);
+  }
+
+#if OMPT_OPTIONAL
   /* OMPT grab all dependences if requested by the tool */
-  if (ompt_enabled && ndeps + ndeps_noalias > 0 &&
-      ompt_callbacks.ompt_callback(ompt_event_task_dependences)) {
+  if (ndeps + ndeps_noalias > 0 &&
+      ompt_enabled.ompt_callback_task_dependences) {
     kmp_int32 i;
 
     new_taskdata->ompt_task_info.ndeps = ndeps + ndeps_noalias;
     new_taskdata->ompt_task_info.deps =
         (ompt_task_dependence_t *)KMP_OMPT_DEPS_ALLOC(
             thread, (ndeps + ndeps_noalias) * sizeof(ompt_task_dependence_t));
 
     KMP_ASSERT(new_taskdata->ompt_task_info.deps != NULL);
for (i = 0; i < ndeps_noalias; i++) {
             ompt_task_dependence_type_inout;
       else if (noalias_dep_list[i].flags.out)
         new_taskdata->ompt_task_info.deps[ndeps + i].dependence_flags =
             ompt_task_dependence_type_out;
       else if (noalias_dep_list[i].flags.in)
         new_taskdata->ompt_task_info.deps[ndeps + i].dependence_flags =
             ompt_task_dependence_type_in;
     }
+    ompt_callbacks.ompt_callback(ompt_callback_task_dependences)(
+        &(new_taskdata->ompt_task_info.task_data),
+        new_taskdata->ompt_task_info.deps, new_taskdata->ompt_task_info.ndeps);
+    /* We can now free the allocated memory for the dependencies */
+    /* For OMPD we might want to delay the free until task_end */
+    KMP_OMPT_DEPS_FREE(thread, new_taskdata->ompt_task_info.deps);
+    new_taskdata->ompt_task_info.deps = NULL;
+    new_taskdata->ompt_task_info.ndeps = 0;
   }
-#endif /* OMPT_SUPPORT && OMPT_TRACE */
+#endif /* OMPT_OPTIONAL */
+#endif /* OMPT_SUPPORT */
 
   bool serial = current_task->td_flags.team_serial ||
                 current_task->td_flags.tasking_ser ||
                 current_task->td_flags.final;
 #if OMP_45_ENABLED
   kmp_task_team_t *task_team = thread->th.th_task_team;
   serial = serial && !(task_team && task_team->tt.tt_found_proxy_tasks);
 #endif
KA_TRACE(10, ("__kmpc_omp_task_with_deps(exit): T#%d ignored dependencies "
                   gtid, loc_ref, new_taskdata));
   }
 
   KA_TRACE(10, ("__kmpc_omp_task_with_deps(exit): T#%d task had no blocking "
                 "dependencies : "
                 "loc=%p task=%p, transferring to __kmpc_omp_task\n",
                 gtid, loc_ref, new_taskdata));
 
-  return __kmpc_omp_task(loc_ref, gtid, new_task);
+  return __kmp_omp_task(gtid, new_task, true);
 }
 
 /*!
 @ingroup TASKING
 @param loc_ref location of the original task directive
 @param gtid Global Thread ID of encountering thread
 @param ndeps Number of depend items with possible aliasing
 @param dep_list List of depend items with possible aliasing

runtime/src/kmp_tasking.cpp

#endif /* BUILD_TIED_TASK_STACK */
 
   // GEH TODO: shouldn't we pass some sort of location identifier here?
   // APT: yes, we will pass location here.
   // need to store current thread state (in a thread or taskdata structure)
   // before setting work_state, otherwise wrong state is set after end of task
 
   KA_TRACE(10, ("__kmp_task_start(exit): T#%d task=%p\n", gtid, taskdata));
 
+  return;
+}
+
 #if OMPT_SUPPORT
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_task_begin)) {
-    kmp_taskdata_t *parent = taskdata->td_parent;
-    ompt_callbacks.ompt_callback(ompt_event_task_begin)(
-        parent ? parent->ompt_task_info.task_id : ompt_task_id_none,
-        parent ? &(parent->ompt_task_info.frame) : NULL,
-        taskdata->ompt_task_info.task_id, taskdata->ompt_task_info.function);
-  }
-#endif
-#if OMP_40_ENABLED && OMPT_SUPPORT && OMPT_TRACE
-  /* OMPT emit all dependences if requested by the tool */
-  if (ompt_enabled && taskdata->ompt_task_info.ndeps > 0 &&
-      ompt_callbacks.ompt_callback(ompt_event_task_dependences)) {
-    ompt_callbacks.ompt_callback(ompt_event_task_dependences)(
-        taskdata->ompt_task_info.task_id, taskdata->ompt_task_info.deps,
-        taskdata->ompt_task_info.ndeps);
-    /* We can now free the allocated memory for the dependencies */
-    KMP_OMPT_DEPS_FREE(thread, taskdata->ompt_task_info.deps);
-    taskdata->ompt_task_info.deps = NULL;
-    taskdata->ompt_task_info.ndeps = 0;
-  }
-#endif /* OMP_40_ENABLED && OMPT_SUPPORT && OMPT_TRACE */
+//------------------------------------------------------------------------------
+// __ompt_task_init:
+//   Initialize OMPT fields maintained by a task. This will only be called after
+//   ompt_start_tool, so we already know whether ompt is enabled or not.
+
+static inline void __ompt_task_init(kmp_taskdata_t *task, int tid) {
+  // The calls to __ompt_task_init already have the ompt_enabled condition.
+  task->ompt_task_info.task_data.value = 0;
+  task->ompt_task_info.frame.exit_runtime_frame = NULL;
+  task->ompt_task_info.frame.reenter_runtime_frame = NULL;
+#if OMP_40_ENABLED
+  task->ompt_task_info.ndeps = 0;
+  task->ompt_task_info.deps = NULL;
+#endif /* OMP_40_ENABLED */
+
+// __ompt_task_start:
+//   Build and trigger task-begin event
+static inline void __ompt_task_start(kmp_task_t *task,
+                                     kmp_taskdata_t *current_task,
+                                     kmp_int32 gtid) {
+  kmp_taskdata_t *taskdata = KMP_TASK_TO_TASKDATA(task);
+  ompt_task_status_t status = ompt_task_others;
+  if (__kmp_threads[gtid]->th.ompt_thread_info.ompt_task_yielded) {
+    status = ompt_task_yield;
+    __kmp_threads[gtid]->th.ompt_thread_info.ompt_task_yielded = 0;
+  }
+  /* let OMPT know that we're about to run this task */
+  if (ompt_enabled.ompt_callback_task_schedule) {
+    ompt_callbacks.ompt_callback(ompt_callback_task_schedule)(
+        &(current_task->ompt_task_info.task_data), status,
+        &(taskdata->ompt_task_info.task_data));
+  }
+  taskdata->ompt_task_info.scheduling_parent = current_task;
 
-  return;
+// __ompt_task_finish:
+//   Build and trigger final task-schedule event
+static inline void __ompt_task_finish(kmp_task_t *task,
+                                      kmp_taskdata_t *resumed_task) {
+  kmp_taskdata_t *taskdata = KMP_TASK_TO_TASKDATA(task);
+  ompt_task_status_t status = ompt_task_complete;
+  if (taskdata->td_flags.tiedness == TASK_UNTIED &&
+      KMP_TEST_THEN_ADD32(&(taskdata->td_untied_count), 0) > 1)
+    status = ompt_task_others;
+  if (__kmp_omp_cancellation && taskdata->td_taskgroup &&
+      taskdata->td_taskgroup->cancel_request == cancel_taskgroup) {
+    status = ompt_task_cancel;
   }
 
-// __kmpc_omp_task_begin_if0: report that a given serialized task has started
-// execution
-//
-// loc_ref: source location information; points to beginning of task block.
-// gtid: global thread number.
-// task: task thunk for the started task.
-void __kmpc_omp_task_begin_if0(ident_t *loc_ref, kmp_int32 gtid,
-                               kmp_task_t *task) {
+  /* let OMPT know that we're returning to the callee task */
+  if (ompt_enabled.ompt_callback_task_schedule) {
+    ompt_callbacks.ompt_callback(ompt_callback_task_schedule)(
+        &(taskdata->ompt_task_info.task_data), status,
+        &((resumed_task ? resumed_task
+                        : (taskdata->ompt_task_info.scheduling_parent
+                               ? taskdata->ompt_task_info.scheduling_parent
+                               : taskdata->td_parent))
+              ->ompt_task_info.task_data));
+  }
+}
+#endif
+
+template<bool ompt>
+static void __kmpc_omp_task_begin_if0_template(ident_t *loc_ref, kmp_int32 gtid,
+                                          kmp_task_t *task, void *frame_address,
+                                          void *return_address) {
   kmp_taskdata_t *taskdata = KMP_TASK_TO_TASKDATA(task);
   kmp_taskdata_t *current_task = __kmp_threads[gtid]->th.th_current_task;
 
   KA_TRACE(10, ("__kmpc_omp_task_begin_if0(enter): T#%d loc=%p task=%p "
                 "current_task=%p\n",
                 gtid, loc_ref, taskdata, current_task));
 
   if (taskdata->td_flags.tiedness == TASK_UNTIED) {
     // untied task needs to increment counter so that the task structure is not
     // freed prematurely
     kmp_int32 counter = 1 + KMP_TEST_THEN_INC32(&taskdata->td_untied_count);
     KA_TRACE(20, ("__kmpc_omp_task_begin_if0: T#%d untied_count (%d) "
                   "incremented for task %p\n",
                   gtid, counter, taskdata));
   }
 
   taskdata->td_flags.task_serial =
       1; // Execute this task immediately, not deferred.
+
   __kmp_task_start(gtid, task, current_task);
 
+#if OMPT_SUPPORT
+  if(ompt)
+  {
+    if (current_task->ompt_task_info.frame.reenter_runtime_frame == NULL) {
+      current_task->ompt_task_info.frame.reenter_runtime_frame =
+          taskdata->ompt_task_info.frame.exit_runtime_frame = frame_address;
+    }
+    if (ompt_enabled.ompt_callback_task_create) {
+      ompt_task_info_t *parent_info = &(current_task->ompt_task_info);
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          &(parent_info->task_data), &(parent_info->frame),
+          &(taskdata->ompt_task_info.task_data),
+          ompt_task_explicit | TASK_TYPE_DETAILS_FORMAT(taskdata), 0,
+          return_address);
+    }
+    __ompt_task_start(task, current_task, gtid);
+  }
+#endif // OMPT_SUPPORT
+
   KA_TRACE(10, ("__kmpc_omp_task_begin_if0(exit): T#%d loc=%p task=%p,\n", gtid,
                 loc_ref, taskdata));
+}
+
+#if OMPT_SUPPORT
+OMPT_NOINLINE
+static void __kmpc_omp_task_begin_if0_ompt(ident_t *loc_ref, kmp_int32 gtid,
+                                          kmp_task_t *task, void *frame_address,
+                                          void *return_address) {
+  return __kmpc_omp_task_begin_if0_template<true>(loc_ref, gtid, task,
+                                                 frame_address, return_address);
+}
+#endif // OMPT_SUPPORT
 
+// __kmpc_omp_task_begin_if0: report that a given serialized task has started
+// execution
+//
+// loc_ref: source location information; points to beginning of task block.
+// gtid: global thread number.
+// task: task thunk for the started task.
+void __kmpc_omp_task_begin_if0(ident_t *loc_ref, kmp_int32 gtid,
+                               kmp_task_t *task) {
+#if OMPT_SUPPORT
+  if (UNLIKELY(ompt_enabled.enabled)) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+    __kmpc_omp_task_begin_if0_ompt(loc_ref, gtid, task,
+                                  OMPT_GET_FRAME_ADDRESS(1),
+                                  OMPT_LOAD_RETURN_ADDRESS(gtid));
     return;
   }
+#endif
+  return __kmpc_omp_task_begin_if0_template<false>(loc_ref, gtid, task,
+                                                  NULL, NULL);
+}
 
 #ifdef TASK_UNUSED
 // __kmpc_omp_task_begin: report that a given task has started execution
 // NEVER GENERATED BY COMPILER, DEPRECATED!!!
 void __kmpc_omp_task_begin(ident_t *loc_ref, kmp_int32 gtid, kmp_task_t *task) {
   kmp_taskdata_t *current_task = __kmp_threads[gtid]->th.th_current_task;
 
   KA_TRACE(

 static void __kmp_task_finish(kmp_int32 gtid, kmp_task_t *task,
                               kmp_taskdata_t *resumed_task) {
   kmp_taskdata_t *taskdata = KMP_TASK_TO_TASKDATA(task);
   kmp_info_t *thread = __kmp_threads[gtid];
   kmp_task_team_t *task_team =
       thread->th.th_task_team; // might be NULL for serial teams...
   kmp_int32 children = 0;
 
-#if OMPT_SUPPORT
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_task_end)) {
-    kmp_taskdata_t *parent = taskdata->td_parent;
-    ompt_callbacks.ompt_callback(ompt_event_task_end)(
-        taskdata->ompt_task_info.task_id);
-  }
-#endif
-
   KA_TRACE(10, ("__kmp_task_finish(enter): T#%d finishing task %p and resuming "
                 "task %p\n",
                 gtid, taskdata, resumed_task));
 
   KMP_DEBUG_ASSERT(taskdata->td_flags.tasktype == TASK_EXPLICIT);
 
 // Pop task from stack if tied
 #ifdef BUILD_TIED_TASK_STACK
#endif
 
   KA_TRACE(
       10, ("__kmp_task_finish(exit): T#%d finished task %p, resuming task %p\n",
            gtid, taskdata, resumed_task));
 
   return;
 }
 
-// __kmpc_omp_task_complete_if0: report that a task has completed execution
-//
-// loc_ref: source location information; points to end of task block.
-// gtid: global thread number.
-// task: task thunk for the completed task.
-void __kmpc_omp_task_complete_if0(ident_t *loc_ref, kmp_int32 gtid,
+template<bool ompt>
+static void __kmpc_omp_task_complete_if0_template(ident_t *loc_ref, kmp_int32 gtid,
                                              kmp_task_t *task) {
   KA_TRACE(10, ("__kmpc_omp_task_complete_if0(enter): T#%d loc=%p task=%p\n",
                 gtid, loc_ref, KMP_TASK_TO_TASKDATA(task)));
+
   // this routine will provide task to resume
   __kmp_task_finish(gtid, task, NULL);
 
   KA_TRACE(10, ("__kmpc_omp_task_complete_if0(exit): T#%d loc=%p task=%p\n",
                 gtid, loc_ref, KMP_TASK_TO_TASKDATA(task)));
+
+#if OMPT_SUPPORT
+  if(ompt)
+  {
+    __ompt_task_finish(task, NULL);
+    ompt_frame_t *ompt_frame;
+    __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL);
+    ompt_frame->reenter_runtime_frame = NULL;
+  }
+#endif
+
   return;
 }
 
+#if OMPT_SUPPORT
+OMPT_NOINLINE
+void __kmpc_omp_task_complete_if0_ompt(ident_t *loc_ref, kmp_int32 gtid,
+                                  kmp_task_t *task) {
+  __kmpc_omp_task_complete_if0_template<true>(loc_ref, gtid, task);
+}
+#endif // OMPT_SUPPORT
+
+// __kmpc_omp_task_complete_if0: report that a task has completed execution
+//
+// loc_ref: source location information; points to end of task block.
+// gtid: global thread number.
+// task: task thunk for the completed task.
+void __kmpc_omp_task_complete_if0(ident_t *loc_ref, kmp_int32 gtid,
+                                  kmp_task_t *task) {
+#if OMPT_SUPPORT
+  if (UNLIKELY(ompt_enabled.enabled)) {
+    __kmpc_omp_task_complete_if0_ompt(loc_ref, gtid, task);
+    return;
+  }
+#endif
+  __kmpc_omp_task_complete_if0_template<false>(loc_ref, gtid, task);
+}
+
 #ifdef TASK_UNUSED
 // __kmpc_omp_task_complete: report that a task has completed execution
 // NEVER GENERATED BY COMPILER, DEPRECATED!!!
 void __kmpc_omp_task_complete(ident_t *loc_ref, kmp_int32 gtid,
                               kmp_task_t *task) {
   KA_TRACE(10, ("__kmpc_omp_task_complete(enter): T#%d loc=%p task=%p\n", gtid,
                 loc_ref, KMP_TASK_TO_TASKDATA(task)));
 
   __kmp_task_finish(gtid, task, 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,
                 loc_ref, KMP_TASK_TO_TASKDATA(task)));
   return;
 }
 #endif // TASK_UNUSED
 
-#if OMPT_SUPPORT
-// __kmp_task_init_ompt: Initialize OMPT fields maintained by a task. This will
-//  only be called after ompt_tool, so we already know whether ompt is enabled
-// or not.
-static inline void __kmp_task_init_ompt(kmp_taskdata_t *task, int tid,
-                                        void *function) {
-  if (ompt_enabled) {
-    task->ompt_task_info.task_id = __ompt_task_id_new(tid);
-    task->ompt_task_info.function = function;
-    task->ompt_task_info.frame.exit_runtime_frame = NULL;
-    task->ompt_task_info.frame.reenter_runtime_frame = NULL;
-#if OMP_40_ENABLED
-    task->ompt_task_info.ndeps = 0;
-    task->ompt_task_info.deps = NULL;
-#endif /* OMP_40_ENABLED */
-  }
-}
-#endif
-
 // __kmp_init_implicit_task: Initialize the appropriate fields in the implicit
 // task for a given thread
 //
 // loc_ref:  reference to source location of parallel region
 // this_thr:  thread data structure corresponding to implicit task
 // team: team for this_thr
 // tid: thread id of given thread within team
 // set_curr_task: TRUE if need to push current task to thread

 #endif
     __kmp_push_current_task_to_thread(this_thr, team, tid);
   } else {
     KMP_DEBUG_ASSERT(task->td_incomplete_child_tasks == 0);
     KMP_DEBUG_ASSERT(task->td_allocated_child_tasks == 0);
   }
 
 #if OMPT_SUPPORT
-  __kmp_task_init_ompt(task, tid, NULL);
+  if (__builtin_expect(ompt_enabled.enabled, 0))
+    __ompt_task_init(task, tid);
 #endif
 
   KF_TRACE(10, ("__kmp_init_implicit_task(exit): T#:%d team=%p task=%p\n", tid,
                 team, task));
 }
 
 // __kmp_finish_implicit_task: Release resources associated to implicit tasks
 // at the end of parallel regions. Some resources are kept for reuse in the next
#endif
     }
   }
 
   KA_TRACE(20, ("__kmp_task_alloc(exit): T#%d created task %p parent=%p\n",
                 gtid, taskdata, taskdata->td_parent));
   ANNOTATE_HAPPENS_BEFORE(task);
 
 #if OMPT_SUPPORT
-  __kmp_task_init_ompt(taskdata, gtid, (void *)task_entry);
+  if (__builtin_expect(ompt_enabled.enabled, 0))
+    __ompt_task_init(taskdata, gtid);
 #endif
 
   return task;
 }
 
 kmp_task_t *__kmpc_omp_task_alloc(ident_t *loc_ref, kmp_int32 gtid,
                                   kmp_int32 flags, size_t sizeof_kmp_task_t,
                                   size_t sizeof_shareds,
#if USE_ITT_BUILD && USE_ITT_NOTIFY
   }
 #endif
 
 #if OMP_45_ENABLED
   // Proxy tasks are not handled by the runtime
   if (taskdata->td_flags.proxy != TASK_PROXY) {
 #endif
     ANNOTATE_HAPPENS_AFTER(task);
-    __kmp_task_start(gtid, task, current_task);
+    __kmp_task_start(gtid, task, current_task); // OMPT only if not discarded
 #if OMP_45_ENABLED
   }
 #endif
 
 #if OMPT_SUPPORT
   ompt_thread_info_t oldInfo;
   kmp_info_t *thread;
-  if (ompt_enabled) {
+  if (__builtin_expect(ompt_enabled.enabled, 0)) {
     // Store the threads states and restore them after the task
     thread = __kmp_threads[gtid];
     oldInfo = thread->th.ompt_thread_info;
     thread->th.ompt_thread_info.wait_id = 0;
-    thread->th.ompt_thread_info.state = ompt_state_work_parallel;
+    thread->th.ompt_thread_info.state = (thread->th.th_team_serialized)
+                                            ? omp_state_work_serial
+                                            : omp_state_work_parallel;
     taskdata->ompt_task_info.frame.exit_runtime_frame =
-        __builtin_frame_address(0);
+        OMPT_GET_FRAME_ADDRESS(0);
   }
 #endif
 
 #if OMP_40_ENABLED
   // TODO: cancel tasks if the parallel region has also been cancelled
   // TODO: check if this sequence can be hoisted above __kmp_task_start
   // if cancellation has been enabled for this run ...
   if (__kmp_omp_cancellation) {
     kmp_info_t *this_thr = __kmp_threads[gtid];
     kmp_team_t *this_team = this_thr->th.th_team;
     kmp_taskgroup_t *taskgroup = taskdata->td_taskgroup;
     if ((taskgroup && taskgroup->cancel_request) ||
         (this_team->t.t_cancel_request == cancel_parallel)) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+      ompt_data_t *task_data;
+      if (__builtin_expect(ompt_enabled.ompt_callback_cancel, 0)) {
+        __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL, NULL);
+        ompt_callbacks.ompt_callback(ompt_callback_cancel)(
+            task_data,
+            ((taskgroup && taskgroup->cancel_request) ? ompt_cancel_taskgroup
+                                                      : ompt_cancel_parallel) |
+                ompt_cancel_discarded_task,
+            NULL);
+      }
+#endif
       KMP_COUNT_BLOCK(TASK_cancelled);
       // this task belongs to a task group and we need to cancel it
       discard = 1 /* true */;
     }
   }
 
   // Invoke the task routine and pass in relevant data.
   // Thunks generated by gcc take a different argument list.
case TASKGROUP:
       break;
     default:
       KMP_PUSH_PARTITIONED_TIMER(OMP_task_immediate);
       break;
     }
 #endif // KMP_STATS_ENABLED
 #endif // OMP_40_ENABLED
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    /* let OMPT know that we're about to run this task */
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_task_switch)) {
-      ompt_callbacks.ompt_callback(ompt_event_task_switch)(
-          current_task->ompt_task_info.task_id,
-          taskdata->ompt_task_info.task_id);
-    }
+// OMPT task begin
+#if OMPT_SUPPORT
+    if (__builtin_expect(ompt_enabled.enabled, 0))
+      __ompt_task_start(task, current_task, gtid);
 #endif
 
 #ifdef KMP_GOMP_COMPAT
     if (taskdata->td_flags.native) {
       ((void (*)(void *))(*(task->routine)))(task->shareds);
     } else
 #endif /* KMP_GOMP_COMPAT */
     {
       (*(task->routine))(gtid, task);
     }
     KMP_POP_PARTITIONED_TIMER();
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    /* let OMPT know that we're returning to the callee task */
-    if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_task_switch)) {
-      ompt_callbacks.ompt_callback(ompt_event_task_switch)(
-          taskdata->ompt_task_info.task_id,
-          current_task->ompt_task_info.task_id);
-    }
+#if OMPT_SUPPORT
+    if (__builtin_expect(ompt_enabled.enabled, 0))
+      __ompt_task_finish(task, current_task);
 #endif
-
 #if OMP_40_ENABLED
   }
 #endif // OMP_40_ENABLED
 
 #if OMPT_SUPPORT
-  if (ompt_enabled) {
+  if (__builtin_expect(ompt_enabled.enabled, 0)) {
     thread->th.ompt_thread_info = oldInfo;
     taskdata->ompt_task_info.frame.exit_runtime_frame = NULL;
   }
 #endif
 
 #if OMP_45_ENABLED
   // Proxy tasks are not handled by the runtime
   if (taskdata->td_flags.proxy != TASK_PROXY) {
 #endif
     ANNOTATE_HAPPENS_BEFORE(taskdata->td_parent);
-    __kmp_task_finish(gtid, task, current_task);
+    __kmp_task_finish(gtid, task, current_task); // OMPT only if not discarded
 #if OMP_45_ENABLED
   }
 #endif
 
 #if USE_ITT_BUILD && USE_ITT_NOTIFY
   // Barrier imbalance - correct arrive time after the task finished
   if (__kmp_forkjoin_frames_mode == 3) {
     kmp_info_t *this_thr = __kmp_threads[gtid];

 //    resumed later.
 kmp_int32 __kmpc_omp_task_parts(ident_t *loc_ref, kmp_int32 gtid,
                                 kmp_task_t *new_task) {
   kmp_taskdata_t *new_taskdata = KMP_TASK_TO_TASKDATA(new_task);
 
   KA_TRACE(10, ("__kmpc_omp_task_parts(enter): T#%d loc=%p task=%p\n", gtid,
                 loc_ref, new_taskdata));
 
+#if OMPT_SUPPORT
+  kmp_taskdata_t *parent;
+  if (__builtin_expect(ompt_enabled.enabled, 0)) {
+    parent = new_taskdata->td_parent;
+    if (ompt_enabled.ompt_callback_task_create) {
+      ompt_data_t task_data = ompt_data_none;
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          parent ? &(parent->ompt_task_info.task_data) : &task_data,
+          parent ? &(parent->ompt_task_info.frame) : NULL,
+          &(new_taskdata->ompt_task_info.task_data), ompt_task_explicit, 0,
+          OMPT_GET_RETURN_ADDRESS(0));
+    }
+  }
+#endif
+
   /* Should we execute the new task or queue it? For now, let's just always try
      to queue it.  If the queue fills up, then we'll execute it.  */
 
   if (__kmp_push_task(gtid, new_task) == TASK_NOT_PUSHED) // if cannot defer
   { // Execute this task immediately
     kmp_taskdata_t *current_task = __kmp_threads[gtid]->th.th_current_task;
     new_taskdata->td_flags.task_serial = 1;
     __kmp_invoke_task(gtid, new_task, current_task);
   }
 
   KA_TRACE(
       10,
       ("__kmpc_omp_task_parts(exit): T#%d returning TASK_CURRENT_NOT_QUEUED: "
        "loc=%p task=%p, return: TASK_CURRENT_NOT_QUEUED\n",
        gtid, loc_ref, new_taskdata));
 
   ANNOTATE_HAPPENS_BEFORE(new_task);
+#if OMPT_SUPPORT
+  if (__builtin_expect(ompt_enabled.enabled, 0)) {
+    parent->ompt_task_info.frame.reenter_runtime_frame = NULL;
+  }
+#endif
   return TASK_CURRENT_NOT_QUEUED;
 }
 
 // __kmp_omp_task: Schedule a non-thread-switchable task for execution
 //
 // gtid: Global Thread ID of encountering thread
 // new_task:non-thread-switchable task thunk allocated by __kmp_omp_task_alloc()
 // serialize_immediate: if TRUE then if the task is executed immediately its
 // execution will be serialized
 // Returns:
 //    TASK_CURRENT_NOT_QUEUED (0) if did not suspend and queue current task to
 //    be resumed later.
 //    TASK_CURRENT_QUEUED (1) if suspended and queued the current task to be
 //    resumed later.
 kmp_int32 __kmp_omp_task(kmp_int32 gtid, kmp_task_t *new_task,
                          bool serialize_immediate) {
   kmp_taskdata_t *new_taskdata = KMP_TASK_TO_TASKDATA(new_task);
 
-#if OMPT_SUPPORT
-  if (ompt_enabled) {
-    new_taskdata->ompt_task_info.frame.reenter_runtime_frame =
-        __builtin_frame_address(1);
-  }
-#endif
-
 /* Should we execute the new task or queue it? For now, let's just always try to
    queue it.  If the queue fills up, then we'll execute it.  */
 #if OMP_45_ENABLED
   if (new_taskdata->td_flags.proxy == TASK_PROXY ||
       __kmp_push_task(gtid, new_task) == TASK_NOT_PUSHED) // if cannot defer
 #else
   if (__kmp_push_task(gtid, new_task) == TASK_NOT_PUSHED) // if cannot defer
 #endif
   { // Execute this task immediately
     kmp_taskdata_t *current_task = __kmp_threads[gtid]->th.th_current_task;
     if (serialize_immediate)
       new_taskdata->td_flags.task_serial = 1;
     __kmp_invoke_task(gtid, new_task, current_task);
   }
 
-#if OMPT_SUPPORT
-  if (ompt_enabled) {
-    new_taskdata->ompt_task_info.frame.reenter_runtime_frame = NULL;
-  }
-#endif
-
   ANNOTATE_HAPPENS_BEFORE(new_task);
   return TASK_CURRENT_NOT_QUEUED;
 }
 
 // __kmpc_omp_task: Wrapper around __kmp_omp_task to schedule a
 // non-thread-switchable task from the parent thread only!
 //
 // loc_ref: location of original task pragma (ignored)
 // gtid: Global Thread ID of encountering thread
 // new_task: non-thread-switchable task thunk allocated by
 // __kmp_omp_task_alloc()
 // Returns:
 //    TASK_CURRENT_NOT_QUEUED (0) if did not suspend and queue current task to
 //    be resumed later.
 //    TASK_CURRENT_QUEUED (1) if suspended and queued the current task to be
 //    resumed later.
 kmp_int32 __kmpc_omp_task(ident_t *loc_ref, kmp_int32 gtid,
                           kmp_task_t *new_task) {
   kmp_int32 res;
   KMP_SET_THREAD_STATE_BLOCK(EXPLICIT_TASK);
 
-#if KMP_DEBUG
+#if KMP_DEBUG || OMPT_SUPPORT
   kmp_taskdata_t *new_taskdata = KMP_TASK_TO_TASKDATA(new_task);
 #endif
   KA_TRACE(10, ("__kmpc_omp_task(enter): T#%d loc=%p task=%p\n", gtid, loc_ref,
                 new_taskdata));
 
+#if OMPT_SUPPORT
+  kmp_taskdata_t *parent = NULL;
+  if (__builtin_expect(ompt_enabled.enabled && !new_taskdata->td_flags.started,
+                       0)) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+    parent = new_taskdata->td_parent;
+    if (!parent->ompt_task_info.frame.reenter_runtime_frame)
+      parent->ompt_task_info.frame.reenter_runtime_frame =
+          OMPT_GET_FRAME_ADDRESS(1);
+    if (ompt_enabled.ompt_callback_task_create) {
+      ompt_data_t task_data = ompt_data_none;
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          parent ? &(parent->ompt_task_info.task_data) : &task_data,
+          parent ? &(parent->ompt_task_info.frame) : NULL,
+          &(new_taskdata->ompt_task_info.task_data),
+          ompt_task_explicit | TASK_TYPE_DETAILS_FORMAT(new_taskdata), 0,
+          OMPT_LOAD_RETURN_ADDRESS(gtid));
+    }
+  }
+#endif
+
   res = __kmp_omp_task(gtid, new_task, true);
 
   KA_TRACE(10, ("__kmpc_omp_task(exit): T#%d returning "
                 "TASK_CURRENT_NOT_QUEUED: loc=%p task=%p\n",
                 gtid, loc_ref, new_taskdata));
+#if OMPT_SUPPORT
+  if (__builtin_expect(ompt_enabled.enabled && parent != NULL, 0)) {
+    parent->ompt_task_info.frame.reenter_runtime_frame = NULL;
+  }
+#endif
   return res;
 }
 
-// __kmpc_omp_taskwait: Wait until all tasks generated by the current task are
-// complete
-kmp_int32 __kmpc_omp_taskwait(ident_t *loc_ref, kmp_int32 gtid) {
+template <bool ompt>
+static kmp_int32 __kmpc_omp_taskwait_template(ident_t *loc_ref, kmp_int32 gtid,
+                                         void *frame_address,
+                                         void *return_address) {
   kmp_taskdata_t *taskdata;
   kmp_info_t *thread;
   int thread_finished = FALSE;
   KMP_SET_THREAD_STATE_BLOCK(TASKWAIT);
 
   KA_TRACE(10, ("__kmpc_omp_taskwait(enter): T#%d loc=%p\n", gtid, loc_ref));
 
   if (__kmp_tasking_mode != tskm_immediate_exec) {
     thread = __kmp_threads[gtid];
     taskdata = thread->th.th_current_task;
-#if OMPT_SUPPORT && OMPT_TRACE
-    ompt_task_id_t my_task_id;
-    ompt_parallel_id_t my_parallel_id;
 
-    if (ompt_enabled) {
-      kmp_team_t *team = thread->th.th_team;
-      my_task_id = taskdata->ompt_task_info.task_id;
-      my_parallel_id = team->t.ompt_team_info.parallel_id;
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    ompt_data_t *my_task_data;
+    ompt_data_t *my_parallel_data;
+
+    if(ompt){
+      my_task_data = &(taskdata->ompt_task_info.task_data);
+      my_parallel_data = OMPT_CUR_TEAM_DATA(thread);
+
+      taskdata->ompt_task_info.frame.reenter_runtime_frame = frame_address;
 
-      taskdata->ompt_task_info.frame.reenter_runtime_frame =
-          __builtin_frame_address(1);
-      if (ompt_callbacks.ompt_callback(ompt_event_taskwait_begin)) {
-        ompt_callbacks.ompt_callback(ompt_event_taskwait_begin)(my_parallel_id,
-                                                                my_task_id);
+      if (ompt_enabled.ompt_callback_sync_region) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+            ompt_sync_region_taskwait, ompt_scope_begin, my_parallel_data,
+            my_task_data, return_address);
       }
+
+      if (ompt_enabled.ompt_callback_sync_region_wait) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+            ompt_sync_region_taskwait, ompt_scope_begin, my_parallel_data,
+            my_task_data, return_address);
       }
-#endif
+    }
+#endif // OMPT_SUPPORT && OMPT_OPTIONAL
 
 // Debugger: The taskwait is active. Store location and thread encountered the
 // taskwait.
 #if USE_ITT_BUILD
 // Note: These values are used by ITT events as well.
 #endif /* USE_ITT_BUILD */
     taskdata->td_taskwait_counter += 1;
     taskdata->td_taskwait_ident = loc_ref;
     taskdata->td_taskwait_thread = gtid + 1;
 
 #if USE_ITT_BUILD
#if USE_ITT_BUILD
     if (itt_sync_obj != NULL)
       __kmp_itt_taskwait_finished(gtid, itt_sync_obj);
 #endif /* USE_ITT_BUILD */
 
     // Debugger:  The taskwait is completed. Location remains, but thread is
     // negated.
     taskdata->td_taskwait_thread = -taskdata->td_taskwait_thread;
 
-#if OMPT_SUPPORT && OMPT_TRACE
-    if (ompt_enabled) {
-      if (ompt_callbacks.ompt_callback(ompt_event_taskwait_end)) {
-        ompt_callbacks.ompt_callback(ompt_event_taskwait_end)(my_parallel_id,
-                                                              my_task_id);
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if(ompt){
+      if (ompt_enabled.ompt_callback_sync_region_wait) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+            ompt_sync_region_taskwait, ompt_scope_end, my_parallel_data,
+            my_task_data, return_address);
+      }
+      if (ompt_enabled.ompt_callback_sync_region) {
+        ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+            ompt_sync_region_taskwait, ompt_scope_end, my_parallel_data,
+            my_task_data, return_address);
       }
       taskdata->ompt_task_info.frame.reenter_runtime_frame = NULL;
     }
-#endif
+#endif // OMPT_SUPPORT && OMPT_OPTIONAL
+
     ANNOTATE_HAPPENS_AFTER(taskdata);
   }
 
   KA_TRACE(10, ("__kmpc_omp_taskwait(exit): T#%d task %p finished waiting, "
                 "returning TASK_CURRENT_NOT_QUEUED\n",
                 gtid, taskdata));
 
   return TASK_CURRENT_NOT_QUEUED;
 }
 
+#if OMPT_SUPPORT
+OMPT_NOINLINE
+static kmp_int32 __kmpc_omp_taskwait_ompt(ident_t *loc_ref, kmp_int32 gtid,
+                                          void *frame_address,
+                                          void *return_address) {
+  return __kmpc_omp_taskwait_template<true>(loc_ref, gtid, frame_address,
+                                            return_address);
+}
+#endif // OMPT_SUPPORT
+
+// __kmpc_omp_taskwait: Wait until all tasks generated by the current task are
+// complete
+kmp_int32 __kmpc_omp_taskwait(ident_t *loc_ref, kmp_int32 gtid) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (UNLIKELY(ompt_enabled.enabled)) {
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+    return __kmpc_omp_taskwait_template<true>(loc_ref, gtid, OMPT_GET_FRAME_ADDRESS(1),
+                                   OMPT_LOAD_RETURN_ADDRESS(gtid));
+  }
+#endif
+  return __kmpc_omp_taskwait_template<false>(loc_ref, gtid, NULL,
+                                 NULL);
+}
+
 // __kmpc_omp_taskyield: switch to a different task
 kmp_int32 __kmpc_omp_taskyield(ident_t *loc_ref, kmp_int32 gtid, int end_part) {
   kmp_taskdata_t *taskdata;
   kmp_info_t *thread;
   int thread_finished = FALSE;
 
   KMP_COUNT_BLOCK(OMP_TASKYIELD);
   KMP_SET_THREAD_STATE_BLOCK(TASKYIELD);
#if USE_ITT_BUILD
     void *itt_sync_obj = __kmp_itt_taskwait_object(gtid);
     if (itt_sync_obj != NULL)
       __kmp_itt_taskwait_starting(gtid, itt_sync_obj);
 #endif /* USE_ITT_BUILD */
     if (!taskdata->td_flags.team_serial) {
       kmp_task_team_t *task_team = thread->th.th_task_team;
       if (task_team != NULL) {
         if (KMP_TASKING_ENABLED(task_team)) {
+#if OMPT_SUPPORT
+          if (__builtin_expect(ompt_enabled.enabled, 0))
+            thread->th.ompt_thread_info.ompt_task_yielded = 1;
+#endif
           __kmp_execute_tasks_32(
               thread, gtid, NULL, FALSE,
               &thread_finished USE_ITT_BUILD_ARG(itt_sync_obj),
               __kmp_task_stealing_constraint);
+#if OMPT_SUPPORT
+          if (__builtin_expect(ompt_enabled.enabled, 0))
+            thread->th.ompt_thread_info.ompt_task_yielded = 0;
+#endif
         }
       }
     }
 #if USE_ITT_BUILD
     if (itt_sync_obj != NULL)
       __kmp_itt_taskwait_finished(gtid, itt_sync_obj);
 #endif /* USE_ITT_BUILD */
 
void __kmpc_taskgroup(ident_t *loc, int gtid) {
   tg_new->cancel_request = cancel_noreq;
   tg_new->parent = taskdata->td_taskgroup;
 // TODO: change to OMP_50_ENABLED, need to change build tools for this to work
 #if OMP_45_ENABLED
   tg_new->reduce_data = NULL;
   tg_new->reduce_num_data = 0;
 #endif
   taskdata->td_taskgroup = tg_new;
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (__builtin_expect(ompt_enabled.ompt_callback_sync_region, 0)) {
+    void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (!codeptr)
+      codeptr = OMPT_GET_RETURN_ADDRESS(0);
+    kmp_team_t *team = thread->th.th_team;
+    ompt_data_t my_task_data = taskdata->ompt_task_info.task_data;
+    // FIXME: I think this is wrong for lwt!
+    ompt_data_t my_parallel_data = team->t.ompt_team_info.parallel_data;
+
+    ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+        ompt_sync_region_taskgroup, ompt_scope_begin, &(my_parallel_data),
+        &(my_task_data), codeptr);
+  }
+#endif
 }
 
 // __kmpc_end_taskgroup: Wait until all tasks generated by the current task
 //                       and its descendants are complete
 void __kmpc_end_taskgroup(ident_t *loc, int gtid) {
   kmp_info_t *thread = __kmp_threads[gtid];
   kmp_taskdata_t *taskdata = thread->th.th_current_task;
   kmp_taskgroup_t *taskgroup = taskdata->td_taskgroup;
   int thread_finished = FALSE;
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  kmp_team_t *team;
+  ompt_data_t my_task_data;
+  ompt_data_t my_parallel_data;
+  void *codeptr;
+  if (__builtin_expect(ompt_enabled.enabled, 0)) {
+    team = thread->th.th_team;
+    my_task_data = taskdata->ompt_task_info.task_data;
+    // FIXME: I think this is wrong for lwt!
+    my_parallel_data = team->t.ompt_team_info.parallel_data;
+    codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid);
+    if (!codeptr)
+      codeptr = OMPT_GET_RETURN_ADDRESS(0);
+  }
+#endif
+
   KA_TRACE(10, ("__kmpc_end_taskgroup(enter): T#%d loc=%p\n", gtid, loc));
   KMP_DEBUG_ASSERT(taskgroup != NULL);
   KMP_SET_THREAD_STATE_BLOCK(TASKGROUP);
 
   if (__kmp_tasking_mode != tskm_immediate_exec) {
 #if USE_ITT_BUILD
     // For ITT the taskgroup wait is similar to taskwait until we need to
     // distinguish them
     void *itt_sync_obj = __kmp_itt_taskwait_object(gtid);
     if (itt_sync_obj != NULL)
       __kmp_itt_taskwait_starting(gtid, itt_sync_obj);
 #endif /* USE_ITT_BUILD */
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (__builtin_expect(ompt_enabled.ompt_callback_sync_region_wait, 0)) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_taskgroup, ompt_scope_begin, &(my_parallel_data),
+          &(my_task_data), codeptr);
+    }
+#endif
+
 #if OMP_45_ENABLED
     if (!taskdata->td_flags.team_serial ||
         (thread->th.th_task_team != NULL &&
          thread->th.th_task_team->tt.tt_found_proxy_tasks))
 #else
     if (!taskdata->td_flags.team_serial)
 #endif
     {
       kmp_flag_32 flag(RCAST(kmp_uint32 *, &taskgroup->count), 0U);
       while (TCR_4(taskgroup->count) != 0) {
         flag.execute_tasks(thread, gtid, FALSE,
                            &thread_finished USE_ITT_BUILD_ARG(itt_sync_obj),
                            __kmp_task_stealing_constraint);
       }
     }
 
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    if (__builtin_expect(ompt_enabled.ompt_callback_sync_region_wait, 0)) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_taskgroup, ompt_scope_end, &(my_parallel_data),
+          &(my_task_data), codeptr);
+    }
+#endif
+
 #if USE_ITT_BUILD
     if (itt_sync_obj != NULL)
       __kmp_itt_taskwait_finished(gtid, itt_sync_obj);
 #endif /* USE_ITT_BUILD */
   }
   KMP_DEBUG_ASSERT(taskgroup->count == 0);
 
 // TODO: change to OMP_50_ENABLED, need to change build tools for this to work
 #if OMP_45_ENABLED
   if (taskgroup->reduce_data != NULL) // need to reduce?
     __kmp_task_reduction_fini(thread, taskgroup);
 #endif
   // Restore parent taskgroup for the current task
   taskdata->td_taskgroup = taskgroup->parent;
   __kmp_thread_free(thread, taskgroup);
 
   KA_TRACE(10, ("__kmpc_end_taskgroup(exit): T#%d task %p finished waiting\n",
                 gtid, taskdata));
   ANNOTATE_HAPPENS_AFTER(taskdata);
+
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (__builtin_expect(ompt_enabled.ompt_callback_sync_region, 0)) {
+    ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+        ompt_sync_region_taskgroup, ompt_scope_end, &(my_parallel_data),
+        &(my_task_data), codeptr);
+  }
+#endif
 }
 #endif
 
 // __kmp_remove_my_task: remove a task from my own deque
 static kmp_task_t *__kmp_remove_my_task(kmp_info_t *thread, kmp_int32 gtid,
                                         kmp_task_team_t *task_team,
                                         kmp_int32 is_constrained) {
   kmp_task_t *task;
if (!(taskdata->td_flags.team_serial || taskdata->td_flags.tasking_ser)) {
     if (taskdata->td_parent->td_flags.tasktype == TASK_EXPLICIT)
       KMP_TEST_THEN_INC32(&taskdata->td_parent->td_allocated_child_tasks);
   }
 
   KA_TRACE(20,
            ("__kmp_task_dup_alloc(exit): Th %p, created task %p, parent=%p\n",
             thread, taskdata, taskdata->td_parent));
 #if OMPT_SUPPORT
-  __kmp_task_init_ompt(taskdata, thread->th.th_info.ds.ds_gtid,
-                       (void *)task->routine);
+  if (__builtin_expect(ompt_enabled.enabled, 0))
+    __ompt_task_init(taskdata, thread->th.th_info.ds.ds_gtid);
 #endif
   return task;
 }
 
 // Routine optionally generated by the compiler for setting the lastprivate flag
 // and calling needed constructors for private/firstprivate objects
 // (used to form taskloop tasks from pattern task)
 // Parameters: dest task, src task, lastprivate flag.
void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int if_val,
                      int sched, kmp_uint64 grainsize, void *task_dup) {
   kmp_taskdata_t *taskdata = KMP_TASK_TO_TASKDATA(task);
   KMP_DEBUG_ASSERT(task != NULL);
 
   KA_TRACE(20, ("__kmpc_taskloop: T#%d, task %p, lb %lld, ub %lld, st %lld, "
                 "grain %llu(%d), dup %p\n",
                 gtid, taskdata, *lb, *ub, st, grainsize, sched, task_dup));
 
-  if (nogroup == 0)
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL);
+  ompt_task_info_t *task_info = __ompt_get_task_info_object(0);
+  if (ompt_enabled.ompt_callback_work) {
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_taskloop, ompt_scope_begin, &(team_info->parallel_data),
+        &(task_info->task_data),
+        0,
+        OMPT_GET_RETURN_ADDRESS(0));
+  }
+#endif
+
+  if (nogroup == 0) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
     __kmpc_taskgroup(loc, gtid);
+  }
 
   // =========================================================================
   // calculate loop parameters
   kmp_uint64 tc;
   kmp_uint64 lower = *lb; // compiler provides global bounds here
   kmp_uint64 upper = *ub;
   kmp_uint64 ub_glob = upper; // global upper used to calc lastprivate flag
   kmp_uint64 num_tasks = 0, extras = 0;
#endif
   KMP_DEBUG_ASSERT(num_tasks > extras);
   KMP_DEBUG_ASSERT(num_tasks > 0);
   // =========================================================================
 
   // check if clause value first
   if (if_val == 0) { // if(0) specified, mark task as serial
     taskdata->td_flags.task_serial = 1;
     taskdata->td_flags.tiedness = TASK_TIED; // AC: serial task cannot be untied
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
     // always start serial tasks linearly
     __kmp_taskloop_linear(loc, gtid, task, lb, ub, st, ub_glob, num_tasks,
                           grainsize, extras, tc, task_dup);
   } else if (num_tasks > num_tasks_min) {
     KA_TRACE(20, ("__kmpc_taskloop: T#%d, go recursive: tc %llu, #tasks %llu"
                   "(%lld), grain %llu, extras %llu\n",
                   gtid, tc, num_tasks, num_tasks_min, grainsize, extras));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
     __kmp_taskloop_recur(loc, gtid, task, lb, ub, st, ub_glob, num_tasks,
                          grainsize, extras, tc, num_tasks_min, task_dup);
   } else {
     KA_TRACE(20, ("__kmpc_taskloop: T#%d, go linear: tc %llu, #tasks %llu"
                   "(%lld), grain %llu, extras %llu\n",
                   gtid, tc, num_tasks, num_tasks_min, grainsize, extras));
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
     __kmp_taskloop_linear(loc, gtid, task, lb, ub, st, ub_glob, num_tasks,
                           grainsize, extras, tc, task_dup);
   }
 
-  if (nogroup == 0)
+  if (nogroup == 0) {
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+    OMPT_STORE_RETURN_ADDRESS(gtid);
+#endif
     __kmpc_end_taskgroup(loc, gtid);
+  }
+#if OMPT_SUPPORT && OMPT_OPTIONAL
+  if (ompt_enabled.ompt_callback_work) {
+    ompt_callbacks.ompt_callback(ompt_callback_work)(
+        ompt_work_taskloop, ompt_scope_end, &(team_info->parallel_data),
+        &(task_info->task_data),
+        0,
+        OMPT_GET_RETURN_ADDRESS(0));
+  }
+#endif
   KA_TRACE(20, ("__kmpc_taskloop(exit): T#%d\n", gtid));
 }
 
 #endif

runtime/src/kmp_wait_release.h

 //===----------------------------------------------------------------------===//
 
 #ifndef KMP_WAIT_RELEASE_H
 #define KMP_WAIT_RELEASE_H
 
 #include "kmp.h"
 #include "kmp_itt.h"
 #include "kmp_stats.h"
+#if OMPT_SUPPORT
+#include "ompt-specific.h"
+#endif
 
 /*!
 @defgroup WAIT_RELEASE Wait/Release operations
 
 The definitions and functions here implement the lowest level thread
 synchronizations of suspending a thread and awaking it. They are used to build
 higher level operations such as barriers and fork/join.
 */
public:
   bool is_sleeping_val(P old_loc);
   int execute_tasks(kmp_info_t *this_thr, kmp_int32 gtid, int final_spin,
                     int *thread_finished
                     USE_ITT_BUILD_ARG(void * itt_sync_obj), kmp_int32
                     is_constrained);
   */
 };
 
+#if OMPT_SUPPORT
+static inline void __ompt_implicit_task_end(kmp_info_t *this_thr,
+                                            omp_state_t omp_state,
+                                            ompt_data_t *tId,
+                                            ompt_data_t *pId) {
+  int ds_tid = this_thr->th.th_info.ds.ds_tid;
+  if (omp_state == omp_state_wait_barrier_implicit) {
+    this_thr->th.ompt_thread_info.state = omp_state_overhead;
+#if OMPT_OPTIONAL
+    void *codeptr = NULL;
+    if (ompt_enabled.ompt_callback_sync_region_wait) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)(
+          ompt_sync_region_barrier, ompt_scope_end, NULL, tId, codeptr);
+    }
+    if (ompt_enabled.ompt_callback_sync_region) {
+      ompt_callbacks.ompt_callback(ompt_callback_sync_region)(
+          ompt_sync_region_barrier, ompt_scope_end, NULL, tId, codeptr);
+    }
+#endif
+    if (!KMP_MASTER_TID(ds_tid)) {
+      if (ompt_enabled.ompt_callback_implicit_task) {
+        ompt_callbacks.ompt_callback(ompt_callback_implicit_task)(
+            ompt_scope_end, NULL, tId, 0, ds_tid);
+      }
+#if OMPT_OPTIONAL
+      if (ompt_enabled.ompt_callback_idle) {
+        ompt_callbacks.ompt_callback(ompt_callback_idle)(ompt_scope_begin);
+      }
+#endif
+      // return to idle state
+      this_thr->th.ompt_thread_info.state = omp_state_idle;
+    } else {
+      this_thr->th.ompt_thread_info.state = omp_state_overhead;
+    }
+  }
+}
+#endif
+
 /* Spin wait loop that first does pause, then yield, then sleep. A thread that
    calls __kmp_wait_*  must make certain that another thread calls __kmp_release
    to wake it back up to prevent deadlocks!  */
 template <class C>
 static inline void
 __kmp_wait_template(kmp_info_t *this_thr, C *flag,
                     int final_spin USE_ITT_BUILD_ARG(void *itt_sync_obj)) {
   // NOTE: We may not belong to a team at this point.
#endif
   }
   th_gtid = this_thr->th.th_info.ds.ds_gtid;
   KA_TRACE(20,
            ("__kmp_wait_sleep: T#%d waiting for flag(%p)\n", th_gtid, flag));
 #if KMP_STATS_ENABLED
   stats_state_e thread_state = KMP_GET_THREAD_STATE();
 #endif
 
-#if OMPT_SUPPORT && OMPT_BLAME
-  ompt_state_t ompt_state = this_thr->th.ompt_thread_info.state;
-  if (ompt_enabled && ompt_state != ompt_state_undefined) {
-    if (ompt_state == ompt_state_idle) {
-      if (ompt_callbacks.ompt_callback(ompt_event_idle_begin)) {
-        ompt_callbacks.ompt_callback(ompt_event_idle_begin)(th_gtid + 1);
-      }
-    } else if (ompt_callbacks.ompt_callback(ompt_event_wait_barrier_begin)) {
-      KMP_DEBUG_ASSERT(ompt_state == ompt_state_wait_barrier ||
-                       ompt_state == ompt_state_wait_barrier_implicit ||
-                       ompt_state == ompt_state_wait_barrier_explicit);
+/* OMPT Behavior:
+THIS function is called from
+  __kmp_barrier (2 times)  (implicit or explicit barrier in parallel regions)
+            these have join / fork behavior
+
+       In these cases, we don't change the state or trigger events in THIS
+function.
+       Events are triggered in the calling code (__kmp_barrier):
+
+                state := omp_state_overhead
+            barrier-begin
+            barrier-wait-begin
+                state := omp_state_wait_barrier
+          call join-barrier-implementation (finally arrive here)
+          {}
+          call fork-barrier-implementation (finally arrive here)
+          {}
+                state := omp_state_overhead
+            barrier-wait-end
+            barrier-end
+                state := omp_state_work_parallel
+
+
+  __kmp_fork_barrier  (after thread creation, before executing implicit task)
+          call fork-barrier-implementation (finally arrive here)
+          {} // worker arrive here with state = omp_state_idle
+
+
+  __kmp_join_barrier  (implicit barrier at end of parallel region)
+                state := omp_state_barrier_implicit
+            barrier-begin
+            barrier-wait-begin
+          call join-barrier-implementation (finally arrive here
+final_spin=FALSE)
+          {
+          }
+  __kmp_fork_barrier  (implicit barrier at end of parallel region)
+          call fork-barrier-implementation (finally arrive here final_spin=TRUE)
 
+       Worker after task-team is finished:
+            barrier-wait-end
+            barrier-end
+            implicit-task-end
+            idle-begin
+                state := omp_state_idle
+
+       Before leaving, if state = omp_state_idle
+            idle-end
+                state := omp_state_overhead
+*/
+#if OMPT_SUPPORT
+  omp_state_t ompt_entry_state;
+  ompt_data_t *pId = NULL;
+  ompt_data_t *tId;
+  if (ompt_enabled.enabled) {
+    ompt_entry_state = this_thr->th.ompt_thread_info.state;
+    if (!final_spin || ompt_entry_state != omp_state_wait_barrier_implicit ||
+        KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid)) {
       ompt_lw_taskteam_t *team =
           this_thr->th.th_team->t.ompt_serialized_team_info;
-      ompt_parallel_id_t pId;
-      ompt_task_id_t tId;
       if (team) {
-        pId = team->ompt_team_info.parallel_id;
-        tId = team->ompt_task_info.task_id;
-      } else {
-        pId = this_thr->th.th_team->t.ompt_team_info.parallel_id;
-        tId = this_thr->th.th_current_task->ompt_task_info.task_id;
-      }
-      ompt_callbacks.ompt_callback(ompt_event_wait_barrier_begin)(pId, tId);
+        pId = &(team->ompt_team_info.parallel_data);
+        tId = &(team->ompt_task_info.task_data);
+      } else {
+        pId = OMPT_CUR_TEAM_DATA(this_thr);
+        tId = OMPT_CUR_TASK_DATA(this_thr);
+      }
+      pId = NULL;
+      tId = &(this_thr->th.ompt_thread_info.task_data);
+#if OMPT_OPTIONAL
+    if (ompt_entry_state == omp_state_idle) {
+      if (ompt_enabled.ompt_callback_idle) {
+        ompt_callbacks.ompt_callback(ompt_callback_idle)(ompt_scope_begin);
+      }
+    } else
+#endif
+        if (final_spin && (__kmp_tasking_mode == tskm_immediate_exec ||
+                           this_thr->th.th_task_team == NULL)) {
+      // implicit task is done. Either no taskqueue, or task-team finished
+      __ompt_implicit_task_end(this_thr, ompt_entry_state, tId, pId);
     }
   }
 #endif
 
   // Setup for waiting
   KMP_INIT_YIELD(spins);
 
   if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) {
if (__kmp_tasking_mode != tskm_immediate_exec) {
           if (KMP_TASKING_ENABLED(task_team))
             flag->execute_tasks(
                 this_thr, th_gtid, final_spin,
                 &tasks_completed USE_ITT_BUILD_ARG(itt_sync_obj), 0);
           else
             this_thr->th.th_reap_state = KMP_SAFE_TO_REAP;
         } else {
           KMP_DEBUG_ASSERT(!KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid));
+#if OMPT_SUPPORT
+          // task-team is done now, other cases should be catched above
+          if (final_spin && ompt_enabled.enabled)
+            __ompt_implicit_task_end(this_thr, ompt_entry_state, tId, pId);
+#endif
           this_thr->th.th_task_team = NULL;
           this_thr->th.th_reap_state = KMP_SAFE_TO_REAP;
         }
       } else {
         this_thr->th.th_reap_state = KMP_SAFE_TO_REAP;
       } // if
     } // if
 
if (TCR_4(__kmp_global.g.g_done)) {
       break;
     } else if (__kmp_tasking_mode != tskm_immediate_exec &&
                this_thr->th.th_reap_state == KMP_SAFE_TO_REAP) {
       this_thr->th.th_reap_state = KMP_NOT_SAFE_TO_REAP;
     }
     // TODO: If thread is done with work and times out, disband/free
   }
 
-#if OMPT_SUPPORT && OMPT_BLAME
-  if (ompt_enabled && ompt_state != ompt_state_undefined) {
-    if (ompt_state == ompt_state_idle) {
-      if (ompt_callbacks.ompt_callback(ompt_event_idle_end)) {
-        ompt_callbacks.ompt_callback(ompt_event_idle_end)(th_gtid + 1);
-      }
-    } else if (ompt_callbacks.ompt_callback(ompt_event_wait_barrier_end)) {
-      KMP_DEBUG_ASSERT(ompt_state == ompt_state_wait_barrier ||
-                       ompt_state == ompt_state_wait_barrier_implicit ||
-                       ompt_state == ompt_state_wait_barrier_explicit);
-
-      ompt_lw_taskteam_t *team =
-          this_thr->th.th_team->t.ompt_serialized_team_info;
-      ompt_parallel_id_t pId;
-      ompt_task_id_t tId;
-      if (team) {
-        pId = team->ompt_team_info.parallel_id;
-        tId = team->ompt_task_info.task_id;
-      } else {
-        pId = this_thr->th.th_team->t.ompt_team_info.parallel_id;
-        tId = this_thr->th.th_current_task->ompt_task_info.task_id;
+#if OMPT_SUPPORT
+  omp_state_t ompt_exit_state = this_thr->th.ompt_thread_info.state;
+  if (ompt_enabled.enabled && ompt_exit_state != omp_state_undefined) {
+#if OMPT_OPTIONAL
+    if (final_spin) {
+      __ompt_implicit_task_end(this_thr, ompt_exit_state, tId, pId);
+      ompt_exit_state = this_thr->th.ompt_thread_info.state;
+    }
+#endif
+    if (ompt_exit_state == omp_state_idle) {
+#if OMPT_OPTIONAL
+      if (ompt_enabled.ompt_callback_idle) {
+        ompt_callbacks.ompt_callback(ompt_callback_idle)(ompt_scope_end);
       }
-      ompt_callbacks.ompt_callback(ompt_event_wait_barrier_end)(pId, tId);
+#endif
+      this_thr->th.ompt_thread_info.state = omp_state_overhead;
     }
   }
 #endif
 #if KMP_STATS_ENABLED
   // If we were put into idle state, pop that off the state stack
   if (KMP_GET_THREAD_STATE() == IDLE) {
     KMP_POP_PARTITIONED_TIMER();
     KMP_SET_THREAD_STATE(thread_state);

runtime/src/ompt-event-specific.h

 /*----------------------------------------------------------------------------
  | Specify whether an event may occur or not, and whether event callbacks
  | never, sometimes, or always occur.
  |
  | The values for these constants are defined in section 6.1.2 of
  | the OMPT TR. They are exposed to tools through ompt_set_callback.
  +--------------------------------------------------------------------------*/
 
-#define ompt_event_NEVER ompt_set_result_event_never_occurs
-#define ompt_event_UNIMPLEMENTED ompt_set_result_event_may_occur_no_callback
-#define ompt_event_MAY_CONVENIENT ompt_set_result_event_may_occur_callback_some
-#define ompt_event_MAY_ALWAYS ompt_set_result_event_may_occur_callback_always
+#define ompt_event_UNIMPLEMENTED ompt_set_never
+#define ompt_event_MAY_CONVENIENT ompt_set_sometimes
+#define ompt_event_MAY_ALWAYS ompt_set_always
 
-#if OMPT_TRACE
-#define ompt_event_MAY_ALWAYS_TRACE ompt_event_MAY_ALWAYS
+#if OMPT_OPTIONAL
+#define ompt_event_MAY_ALWAYS_OPTIONAL ompt_event_MAY_ALWAYS
 #else
-#define ompt_event_MAY_ALWAYS_TRACE ompt_event_UNIMPLEMENTED
-#endif
-
-#if OMPT_BLAME
-#define ompt_event_MAY_ALWAYS_BLAME ompt_event_MAY_ALWAYS
-#else
-#define ompt_event_MAY_ALWAYS_BLAME ompt_event_UNIMPLEMENTED
+#define ompt_event_MAY_ALWAYS_OPTIONAL ompt_event_UNIMPLEMENTED
 #endif
 
 /*----------------------------------------------------------------------------
  | Mandatory Events
  +--------------------------------------------------------------------------*/
 
-#define ompt_event_parallel_begin_implemented ompt_event_MAY_ALWAYS
-#define ompt_event_parallel_end_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_parallel_begin_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_parallel_end_implemented ompt_event_MAY_ALWAYS
 
-#define ompt_event_task_begin_implemented ompt_event_MAY_ALWAYS
-#define ompt_event_task_end_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_task_create_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_task_schedule_implemented ompt_event_MAY_ALWAYS
 
-#define ompt_event_thread_begin_implemented ompt_event_MAY_ALWAYS
-#define ompt_event_thread_end_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_thread_begin_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_thread_end_implemented ompt_event_MAY_ALWAYS
 
-#define ompt_event_control_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_control_tool_implemented ompt_event_MAY_ALWAYS
 
-#define ompt_event_runtime_shutdown_implemented ompt_event_MAY_ALWAYS
+#define ompt_callback_implicit_task_implemented ompt_event_MAY_ALWAYS
 
 /*----------------------------------------------------------------------------
- | Optional Events (blame shifting)
+ | Target Related Events (not yet implemented)
  +--------------------------------------------------------------------------*/
 
-#define ompt_event_idle_begin_implemented ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_idle_end_implemented ompt_event_MAY_ALWAYS_BLAME
+#define ompt_callback_target_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_target_data_op_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_target_submit_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_device_initialize_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_device_finalize_implemented ompt_event_UNIMPLEMENTED
 
-#define ompt_event_wait_barrier_begin_implemented ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_wait_barrier_end_implemented ompt_event_MAY_ALWAYS_BLAME
-
-#define ompt_event_wait_taskwait_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_wait_taskwait_end_implemented ompt_event_UNIMPLEMENTED
-
-#define ompt_event_wait_taskgroup_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_wait_taskgroup_end_implemented ompt_event_UNIMPLEMENTED
-
-#define ompt_event_release_lock_implemented ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_release_nest_lock_last_implemented                          \
-  ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_release_critical_implemented ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_release_atomic_implemented ompt_event_MAY_ALWAYS_BLAME
-#define ompt_event_release_ordered_implemented ompt_event_MAY_ALWAYS_BLAME
+#define ompt_callback_target_map_implemented ompt_event_UNIMPLEMENTED
 
 /*----------------------------------------------------------------------------
- | Optional Events (synchronous events)
+ | Optional Events (blame shifting)
  +--------------------------------------------------------------------------*/
 
-#define ompt_event_implicit_task_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_implicit_task_end_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_idle_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_initial_task_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_initial_task_end_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_sync_region_wait_implemented                             \
+  ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_task_switch_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_mutex_released_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_loop_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_loop_end_implemented ompt_event_MAY_ALWAYS_TRACE
-
-#define ompt_event_sections_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_sections_end_implemented ompt_event_UNIMPLEMENTED
-
-#define ompt_event_single_in_block_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_single_in_block_end_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_single_others_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_single_others_end_implemented ompt_event_MAY_ALWAYS_TRACE
-
-#define ompt_event_workshare_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_workshare_end_implemented ompt_event_UNIMPLEMENTED
-
-#define ompt_event_master_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_master_end_implemented ompt_event_MAY_ALWAYS_TRACE
-
-#define ompt_event_barrier_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_barrier_end_implemented ompt_event_MAY_ALWAYS_TRACE
+/*----------------------------------------------------------------------------
+ | Optional Events (synchronous events)
+ +--------------------------------------------------------------------------*/
 
-#define ompt_event_taskwait_begin_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_taskwait_end_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_work_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_taskgroup_begin_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_taskgroup_end_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_master_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_release_nest_lock_prev_implemented                          \
-  ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_wait_lock_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_wait_nest_lock_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_wait_critical_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_wait_atomic_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_wait_ordered_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_sync_region_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_acquired_lock_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_acquired_nest_lock_first_implemented                        \
-  ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_acquired_nest_lock_next_implemented                         \
-  ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_acquired_critical_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_acquired_atomic_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_acquired_ordered_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_mutex_acquire_implemented ompt_event_MAY_ALWAYS_OPTIONAL
+#define ompt_callback_mutex_acquired_implemented ompt_event_MAY_ALWAYS_OPTIONAL
+#define ompt_callback_nest_lock_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_init_lock_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_init_nest_lock_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_lock_init_implemented ompt_event_MAY_ALWAYS_OPTIONAL
+#define ompt_callback_lock_destroy_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_destroy_lock_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_destroy_nest_lock_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_flush_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
-#define ompt_event_flush_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_cancel_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 
 #if OMP_40_ENABLED
-#define ompt_event_task_dependences_implemented ompt_event_MAY_ALWAYS_TRACE
-#define ompt_event_task_dependence_pair_implemented ompt_event_MAY_ALWAYS_TRACE
+#define ompt_callback_task_dependences_implemented                             \
+  ompt_event_MAY_ALWAYS_OPTIONAL
+#define ompt_callback_task_dependence_implemented ompt_event_MAY_ALWAYS_OPTIONAL
 #else
-#define ompt_event_task_dependences_implemented ompt_event_UNIMPLEMENTED
-#define ompt_event_task_dependence_pair_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_task_dependences_implemented ompt_event_UNIMPLEMENTED
+#define ompt_callback_task_dependence_implemented ompt_event_UNIMPLEMENTED
 #endif /* OMP_40_ENABLED */
 
 #endif

runtime/src/ompt-general.cpp

 /*****************************************************************************
  * system include files
  ****************************************************************************/
 
 #include <assert.h>
 
 #include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#if KMP_OS_UNIX
+#include <dlfcn.h>
+#endif
 
 /*****************************************************************************
  * ompt include files
  ****************************************************************************/
 
 #include "ompt-specific.cpp"
 
 /*****************************************************************************

 #endif
 
 /*****************************************************************************
  * types
  ****************************************************************************/
 
 typedef struct {
   const char *state_name;
-  ompt_state_t state_id;
-} ompt_state_info_t;
+  omp_state_t state_id;
+} omp_state_info_t;
+
+typedef struct {
+  const char *name;
+  ompt_mutex_impl_t id;
+} ompt_mutex_impl_info_t;
 
 enum tool_setting_e {
   omp_tool_error,
   omp_tool_unset,
   omp_tool_disabled,
   omp_tool_enabled
 };
 
-typedef void (*ompt_initialize_t)(ompt_function_lookup_t ompt_fn_lookup,
-                                  const char *version,
-                                  unsigned int ompt_version);
+typedef int (*ompt_initialize_t)(ompt_function_lookup_t lookup,
+                                 struct ompt_fns_t *fns);
+
+typedef void (*ompt_finalize_t)(struct ompt_fns_t *fns);
 
 /*****************************************************************************
  * global variables
  ****************************************************************************/
 
-int ompt_enabled = 0;
+ompt_callbacks_active_t ompt_enabled;
+
+omp_state_info_t omp_state_info[] = {
+#define omp_state_macro(state, code) {#state, state},
+    FOREACH_OMP_STATE(omp_state_macro)
+#undef omp_state_macro
+};
 
-ompt_state_info_t ompt_state_info[] = {
-#define ompt_state_macro(state, code) {#state, state},
-    FOREACH_OMPT_STATE(ompt_state_macro)
-#undef ompt_state_macro
+ompt_mutex_impl_info_t ompt_mutex_impl_info[] = {
+#define ompt_mutex_impl_macro(name, id) {#name, name},
+    FOREACH_OMPT_MUTEX_IMPL(ompt_mutex_impl_macro)
+#undef ompt_mutex_impl_macro
 };
 
-ompt_callbacks_t ompt_callbacks;
+ompt_callbacks_internal_t ompt_callbacks;
 
-static ompt_initialize_t ompt_initialize_fn = NULL;
+static ompt_fns_t *ompt_fns = NULL;
 
 /*****************************************************************************
  * forward declarations
  ****************************************************************************/
 
 static ompt_interface_fn_t ompt_fn_lookup(const char *s);
 
-OMPT_API_ROUTINE ompt_thread_id_t ompt_get_thread_id(void);
+OMPT_API_ROUTINE ompt_data_t *ompt_get_thread_data(void);
 
 /*****************************************************************************
  * initialization and finalization (private operations)
  ****************************************************************************/
 
 /* On Unix-like systems that support weak symbols the following implementation
- * of ompt_tool() will be used in case no tool-supplied implementation of
+ * of ompt_start_tool() will be used in case no tool-supplied implementation of
  * this function is present in the address space of a process.
  *
  * On Windows, the ompt_tool_windows function is used to find the
  * ompt_tool symbol across all modules loaded by a process. If ompt_tool is
  * found, ompt_tool's return value is used to initialize the tool. Otherwise,
  * NULL is returned and OMPT won't be enabled */
+
+typedef ompt_fns_t *(*ompt_start_tool_t)(unsigned int, const char *);
+
+#if KMP_OS_UNIX
+
 #if OMPT_HAVE_WEAK_ATTRIBUTE
+_OMP_EXTERN __attribute__((weak))
+#elif defined KMP_DYNAMIC_LIB
 _OMP_EXTERN
-__attribute__((weak)) ompt_initialize_t ompt_tool() {
+#warning Activation of OMPT is might fail for tools statically linked into the application.
+#else
+#error Activation of OMPT is not supported on this platform.
+#endif
+ompt_fns_t *
+ompt_start_tool(unsigned int omp_version, const char *runtime_version) {
+#ifdef KMP_DYNAMIC_LIB
+  ompt_fns_t *ret = NULL;
+  // Try next symbol in the address space
+  ompt_start_tool_t next_tool = NULL;
+  *(void **)(&next_tool) = dlsym(RTLD_NEXT, "ompt_start_tool");
+  if (next_tool)
+    ret = (next_tool)(omp_version, runtime_version);
+  return ret;
+#else
 #if OMPT_DEBUG
-  printf("ompt_tool() is called from the RTL\n");
+  printf("ompt_start_tool() is called from the RTL\n");
 #endif
   return NULL;
+#endif
 }
 
 #elif OMPT_HAVE_PSAPI
 
 #include <psapi.h>
 #pragma comment(lib, "psapi.lib")
-#define ompt_tool ompt_tool_windows
+#define ompt_start_tool ompt_tool_windows
 
 // The number of loaded modules to start enumeration with EnumProcessModules()
 #define NUM_MODULES 128
 
-static ompt_initialize_t ompt_tool_windows() {
+static ompt_fns_t *ompt_tool_windows(unsigned int omp_version,
+                                     const char *runtime_version) {
   int i;
   DWORD needed, new_size;
   HMODULE *modules;
   HANDLE process = GetCurrentProcess();
   modules = (HMODULE *)malloc(NUM_MODULES * sizeof(HMODULE));
-  ompt_initialize_t (*ompt_tool_p)() = NULL;
+  ompt_start_tool_t ompt_tool_p = NULL;
 
 #if OMPT_DEBUG
-  printf("ompt_tool_windows(): looking for ompt_tool\n");
+  printf("ompt_tool_windows(): looking for ompt_start_tool\n");
 #endif
   if (!EnumProcessModules(process, modules, NUM_MODULES * sizeof(HMODULE),
                           &needed)) {
     // Regardless of the error reason use the stub initialization function
     free(modules);
     return NULL;
   }
   // Check if NUM_MODULES is enough to list all modules
   new_size = needed / sizeof(HMODULE);
   if (new_size > NUM_MODULES) {
 #if OMPT_DEBUG
     printf("ompt_tool_windows(): resize buffer to %d bytes\n", needed);
 #endif
     modules = (HMODULE *)realloc(modules, needed);
     // If resizing failed use the stub function.
     if (!EnumProcessModules(process, modules, needed, &needed)) {
       free(modules);
       return NULL;
     }
   }
   for (i = 0; i < new_size; ++i) {
-    (FARPROC &)ompt_tool_p = GetProcAddress(modules[i], "ompt_tool");
+    (FARPROC &)ompt_tool_p = GetProcAddress(modules[i], "ompt_start_tool");
     if (ompt_tool_p) {
 #if OMPT_DEBUG
       TCHAR modName[MAX_PATH];
       if (GetModuleFileName(modules[i], modName, MAX_PATH))
-        printf("ompt_tool_windows(): ompt_tool found in module %s\n", modName);
+        printf("ompt_tool_windows(): ompt_start_tool found in module %s\n",
+               modName);
 #endif
       free(modules);
-      return ompt_tool_p();
+      return (*ompt_tool_p)(omp_version, runtime_version);
     }
 #if OMPT_DEBUG
     else {
       TCHAR modName[MAX_PATH];
       if (GetModuleFileName(modules[i], modName, MAX_PATH))
-        printf("ompt_tool_windows(): ompt_tool not found in module %s\n",
+        printf("ompt_tool_windows(): ompt_start_tool not found in module %s\n",
                modName);
     }
 #endif
   }
   free(modules);
   return NULL;
 }
 #else
 #error Either __attribute__((weak)) or psapi.dll are required for OMPT support
 #endif // OMPT_HAVE_WEAK_ATTRIBUTE
 
+static ompt_fns_t *ompt_try_start_tool(unsigned int omp_version,
+                                       const char *runtime_version) {
+  ompt_fns_t *ret = NULL;
+  ompt_start_tool_t start_tool = NULL;
+#if KMP_OS_WINDOWS
+  // Cannot use colon to describe a list of absolute paths on Windows
+  const char *sep = ";";
+#else
+  const char *sep = ":";
+#endif
+
+  // Try in the current address space
+  if ((ret = ompt_start_tool(omp_version, runtime_version)))
+    return ret;
+
+  // Try tool-libraries-var ICV
+  const char *tool_libs = getenv("OMP_TOOL_LIBRARIES");
+  if (tool_libs) {
+    const char *libs = __kmp_str_format("%s", tool_libs);
+    char *buf;
+    char *fname = __kmp_str_token(CCAST(char *, libs), sep, &buf);
+    while (fname) {
+#if KMP_OS_UNIX
+      void *h = dlopen(fname, RTLD_LAZY);
+      if (h) {
+        *(void **)(&start_tool) = dlsym(h, "ompt_start_tool");
+#elif KMP_OS_WINDOWS
+      HMODULE h = LoadLibrary(fname);
+      if (h) {
+        start_tool = (ompt_start_tool_t)GetProcAddress(h, "ompt_start_tool");
+#else
+#error Activation of OMPT is not supported on this platform.
+#endif
+        if (start_tool && (ret = (*start_tool)(omp_version, runtime_version)))
+          break;
+      }
+      fname = __kmp_str_token(NULL, sep, &buf);
+    }
+    __kmp_str_free(&libs);
+  }
+  return ret;
+}
+
 void ompt_pre_init() {
   //--------------------------------------------------
   // Execute the pre-initialization logic only once.
   //--------------------------------------------------
   static int ompt_pre_initialized = 0;
 
   if (ompt_pre_initialized)
     return;
#if OMPT_DEBUG
   printf("ompt_pre_init(): tool_setting = %d\n", tool_setting);
 #endif
   switch (tool_setting) {
   case omp_tool_disabled:
     break;
 
   case omp_tool_unset:
   case omp_tool_enabled:
-    ompt_initialize_fn = ompt_tool();
-    if (ompt_initialize_fn) {
-      ompt_enabled = 1;
-    }
+
+    //--------------------------------------------------
+    // Load tool iff specified in environment variable
+    //--------------------------------------------------
+    ompt_fns =
+        ompt_try_start_tool(__kmp_openmp_version, ompt_get_runtime_version());
+
+    memset(&ompt_enabled, 0, sizeof(ompt_enabled));
     break;
 
   case omp_tool_error:
     fprintf(stderr, "Warning: OMP_TOOL has invalid value \"%s\".\n"
                     "  legal values are (NULL,\"\",\"disabled\","
                     "\"enabled\").\n",
             ompt_env_var);
     break;
void ompt_post_init() {
   if (ompt_post_initialized)
     return;
 
   ompt_post_initialized = 1;
 
   //--------------------------------------------------
   // Initialize the tool if so indicated.
   //--------------------------------------------------
-  if (ompt_enabled) {
-    ompt_initialize_fn(ompt_fn_lookup, ompt_get_runtime_version(),
-                       OMPT_VERSION);
+  if (ompt_fns) {
+    ompt_enabled.enabled = !!ompt_fns->initialize(ompt_fn_lookup, ompt_fns);
 
     ompt_thread_t *root_thread = ompt_get_thread();
 
-    ompt_set_thread_state(root_thread, ompt_state_overhead);
+    ompt_set_thread_state(root_thread, omp_state_overhead);
 
-    if (ompt_callbacks.ompt_callback(ompt_event_thread_begin)) {
-      ompt_callbacks.ompt_callback(ompt_event_thread_begin)(
-          ompt_thread_initial, ompt_get_thread_id());
+    if (ompt_enabled.ompt_callback_thread_begin) {
+      ompt_callbacks.ompt_callback(ompt_callback_thread_begin)(
+          ompt_thread_initial, __ompt_get_thread_data_internal());
+    }
+    ompt_data_t *task_data;
+    __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL, NULL);
+    if (ompt_enabled.ompt_callback_task_create) {
+      ompt_callbacks.ompt_callback(ompt_callback_task_create)(
+          NULL, NULL, task_data, ompt_task_initial, 0, NULL);
     }
 
-    ompt_set_thread_state(root_thread, ompt_state_work_serial);
+    ompt_set_thread_state(root_thread, omp_state_work_serial);
   }
 }
 
 void ompt_fini() {
-  if (ompt_enabled) {
-    if (ompt_callbacks.ompt_callback(ompt_event_runtime_shutdown)) {
-      ompt_callbacks.ompt_callback(ompt_event_runtime_shutdown)();
-    }
+  if (ompt_enabled.enabled) {
+    ompt_fns->finalize(ompt_fns);
   }
 
-  ompt_enabled = 0;
+  memset(&ompt_enabled, 0, sizeof(ompt_enabled));
 }
 
 /*****************************************************************************
  * interface operations
  ****************************************************************************/
 
 /*****************************************************************************
  * state
  ****************************************************************************/
 
-OMPT_API_ROUTINE int ompt_enumerate_state(int current_state, int *next_state,
+OMPT_API_ROUTINE int ompt_enumerate_states(int current_state, int *next_state,
                                            const char **next_state_name) {
-  const static int len = sizeof(ompt_state_info) / sizeof(ompt_state_info_t);
+  const static int len = sizeof(omp_state_info) / sizeof(omp_state_info_t);
   int i = 0;
 
   for (i = 0; i < len - 1; i++) {
-    if (ompt_state_info[i].state_id == current_state) {
-      *next_state = ompt_state_info[i + 1].state_id;
-      *next_state_name = ompt_state_info[i + 1].state_name;
+    if (omp_state_info[i].state_id == current_state) {
+      *next_state = omp_state_info[i + 1].state_id;
+      *next_state_name = omp_state_info[i + 1].state_name;
       return 1;
     }
   }
 
   return 0;
 }
 
+OMPT_API_ROUTINE int ompt_enumerate_mutex_impls(int current_impl,
+                                                int *next_impl,
+                                                const char **next_impl_name) {
+  const static int len =
+      sizeof(ompt_mutex_impl_info) / sizeof(ompt_mutex_impl_info_t);
+  int i = 0;
+  for (i = 0; i < len - 1; i++) {
+    if (ompt_mutex_impl_info[i].id != current_impl)
+      continue;
+    *next_impl = ompt_mutex_impl_info[i + 1].id;
+    *next_impl_name = ompt_mutex_impl_info[i + 1].name;
+    return 1;
+  }
+  return 0;
+}
+
 /*****************************************************************************
  * callbacks
  ****************************************************************************/
 
-OMPT_API_ROUTINE int ompt_set_callback(ompt_event_t evid, ompt_callback_t cb) {
-  switch (evid) {
+OMPT_API_ROUTINE int ompt_set_callback(ompt_callbacks_t which,
+                                       ompt_callback_t callback) {
+  switch (which) {
 
 #define ompt_event_macro(event_name, callback_type, event_id)                  \
   case event_name:                                                             \
     if (ompt_event_implementation_status(event_name)) {                        \
-      ompt_callbacks.ompt_callback(event_name) = (callback_type)cb;            \
+      ompt_callbacks.ompt_callback(event_name) = (callback_type)callback;      \
+      ompt_enabled.event_name = 1;                                             \
     }                                                                          \
     return ompt_event_implementation_status(event_name);
 
     FOREACH_OMPT_EVENT(ompt_event_macro)
 
 #undef ompt_event_macro
 
   default:
-    return ompt_set_result_registration_error;
+    return ompt_set_error;
   }
 }
 
-OMPT_API_ROUTINE int ompt_get_callback(ompt_event_t evid, ompt_callback_t *cb) {
-  switch (evid) {
+OMPT_API_ROUTINE int ompt_get_callback(ompt_callbacks_t which,
+                                       ompt_callback_t *callback) {
+  switch (which) {
 
 #define ompt_event_macro(event_name, callback_type, event_id)                  \
   case event_name:                                                             \
     if (ompt_event_implementation_status(event_name)) {                        \
       ompt_callback_t mycb =                                                   \
           (ompt_callback_t)ompt_callbacks.ompt_callback(event_name);           \
       if (mycb) {                                                              \
-        *cb = mycb;                                                            \
+        *callback = mycb;                                                      \
         return ompt_get_callback_success;                                      \
       }                                                                        \
     }                                                                          \
     return ompt_get_callback_failure;
 
     FOREACH_OMPT_EVENT(ompt_event_macro)
 
 #undef ompt_event_macro
 
   default:
     return ompt_get_callback_failure;
   }
 }
 
 /*****************************************************************************
  * parallel regions
  ****************************************************************************/
 
-OMPT_API_ROUTINE ompt_parallel_id_t ompt_get_parallel_id(int ancestor_level) {
-  return __ompt_get_parallel_id_internal(ancestor_level);
+OMPT_API_ROUTINE int ompt_get_parallel_info(int ancestor_level,
+                                            ompt_data_t **parallel_data,
+                                            int *team_size) {
+  return __ompt_get_parallel_info_internal(ancestor_level, parallel_data,
+                                           team_size);
 }
 
-OMPT_API_ROUTINE int ompt_get_parallel_team_size(int ancestor_level) {
-  return __ompt_get_parallel_team_size_internal(ancestor_level);
-}
+OMPT_API_ROUTINE omp_state_t ompt_get_state(ompt_wait_id_t *wait_id) {
+  omp_state_t thread_state = __ompt_get_state_internal(wait_id);
 
-OMPT_API_ROUTINE void *ompt_get_parallel_function(int ancestor_level) {
-  return __ompt_get_parallel_function_internal(ancestor_level);
-}
-
-OMPT_API_ROUTINE ompt_state_t ompt_get_state(ompt_wait_id_t *ompt_wait_id) {
-  ompt_state_t thread_state = __ompt_get_state_internal(ompt_wait_id);
-
-  if (thread_state == ompt_state_undefined) {
-    thread_state = ompt_state_work_serial;
+  if (thread_state == omp_state_undefined) {
+    thread_state = omp_state_work_serial;
   }
 
   return thread_state;
 }
 
 /*****************************************************************************
- * threads
+ * tasks
  ****************************************************************************/
 
-OMPT_API_ROUTINE void *ompt_get_idle_frame() {
-  return __ompt_get_idle_frame_internal();
+OMPT_API_ROUTINE ompt_data_t *ompt_get_thread_data(void) {
+  return __ompt_get_thread_data_internal();
+}
+
+OMPT_API_ROUTINE int ompt_get_task_info(int ancestor_level, int *type,
+                                        ompt_data_t **task_data,
+                                        ompt_frame_t **task_frame,
+                                        ompt_data_t **parallel_data,
+                                        int *thread_num) {
+  return __ompt_get_task_info_internal(ancestor_level, type, task_data,
+                                       task_frame, parallel_data, thread_num);
 }
 
 /*****************************************************************************
- * tasks
+ * places
  ****************************************************************************/
 
-OMPT_API_ROUTINE ompt_thread_id_t ompt_get_thread_id(void) {
-  return __ompt_get_thread_id_internal();
+OMPT_API_ROUTINE int ompt_get_num_places(void) {
+// copied from kmp_ftn_entry.h (but modified)
+#if !KMP_AFFINITY_SUPPORTED
+  return 0;
+#else
+  if (!KMP_AFFINITY_CAPABLE())
+    return 0;
+  return __kmp_affinity_num_masks;
+#endif
+}
+
+OMPT_API_ROUTINE int ompt_get_place_proc_ids(int place_num, int ids_size,
+                                             int *ids) {
+// copied from kmp_ftn_entry.h (but modified)
+#if !KMP_AFFINITY_SUPPORTED
+  return 0;
+#else
+  int i, count;
+  int tmp_ids[ids_size];
+  if (!KMP_AFFINITY_CAPABLE())
+    return 0;
+  if (place_num < 0 || place_num >= (int)__kmp_affinity_num_masks)
+    return 0;
+  /* TODO: Is this safe for asynchronous call from signal handler during runtime
+   * shutdown? */
+  kmp_affin_mask_t *mask = KMP_CPU_INDEX(__kmp_affinity_masks, place_num);
+  count = 0;
+  KMP_CPU_SET_ITERATE(i, mask) {
+    if ((!KMP_CPU_ISSET(i, __kmp_affin_fullMask)) ||
+        (!KMP_CPU_ISSET(i, mask))) {
+      continue;
+    }
+    if (count < ids_size)
+      tmp_ids[count] = i;
+    count++;
+  }
+  if (ids_size >= count) {
+    for (i = 0; i < count; i++) {
+      ids[i] = tmp_ids[i];
+    }
+  }
+  return count;
+#endif
 }
 
-OMPT_API_ROUTINE ompt_task_id_t ompt_get_task_id(int depth) {
-  return __ompt_get_task_id_internal(depth);
+OMPT_API_ROUTINE int ompt_get_place_num(void) {
+// copied from kmp_ftn_entry.h (but modified)
+#if !KMP_AFFINITY_SUPPORTED
+  return -1;
+#else
+  int gtid;
+  kmp_info_t *thread;
+  if (!KMP_AFFINITY_CAPABLE())
+    return -1;
+  gtid = __kmp_entry_gtid();
+  thread = __kmp_thread_from_gtid(gtid);
+  if (thread == NULL || thread->th.th_current_place < 0)
+    return -1;
+  return thread->th.th_current_place;
+#endif
 }
 
-OMPT_API_ROUTINE ompt_frame_t *ompt_get_task_frame(int depth) {
-  return __ompt_get_task_frame_internal(depth);
+OMPT_API_ROUTINE int ompt_get_partition_place_nums(int place_nums_size,
+                                                   int *place_nums) {
+// copied from kmp_ftn_entry.h (but modified)
+#if !KMP_AFFINITY_SUPPORTED
+  return 0;
+#else
+  int i, gtid, place_num, first_place, last_place, start, end;
+  kmp_info_t *thread;
+  if (!KMP_AFFINITY_CAPABLE())
+    return 0;
+  gtid = __kmp_entry_gtid();
+  thread = __kmp_thread_from_gtid(gtid);
+  if (thread == NULL)
+    return 0;
+  first_place = thread->th.th_first_place;
+  last_place = thread->th.th_last_place;
+  if (first_place < 0 || last_place < 0)
+    return 0;
+  if (first_place <= last_place) {
+    start = first_place;
+    end = last_place;
+  } else {
+    start = last_place;
+    end = first_place;
+  }
+  if (end - start <= place_nums_size)
+    for (i = 0, place_num = start; place_num <= end; ++place_num, ++i) {
+      place_nums[i] = place_num;
+    }
+  return end - start;
+#endif
 }
 
-OMPT_API_ROUTINE void *ompt_get_task_function(int depth) {
-  return __ompt_get_task_function_internal(depth);
+/*****************************************************************************
+ * places
+ ****************************************************************************/
+
+OMPT_API_ROUTINE int ompt_get_proc_id(void) {
+#if KMP_OS_LINUX
+  return sched_getcpu();
+#else
+  return -1;
+#endif
 }
 
 /*****************************************************************************
  * placeholders
  ****************************************************************************/
 
 // Don't define this as static. The loader may choose to eliminate the symbol
 // even though it is needed by tools.

 
 /*****************************************************************************
  * compatability
  ****************************************************************************/
 
 OMPT_API_ROUTINE int ompt_get_ompt_version() { return OMPT_VERSION; }
 
 /*****************************************************************************
- * application-facing API
  ****************************************************************************/
 
 /*----------------------------------------------------------------------------
  | control
  ---------------------------------------------------------------------------*/
 
-_OMP_EXTERN void ompt_control(uint64_t command, uint64_t modifier) {
-  if (ompt_enabled && ompt_callbacks.ompt_callback(ompt_event_control)) {
-    ompt_callbacks.ompt_callback(ompt_event_control)(command, modifier);
-  }
+int __kmp_control_tool(uint64_t command, uint64_t modifier, void *arg) {
+
+  if (ompt_enabled.enabled) {
+    if (ompt_enabled.ompt_callback_control_tool) {
+      return ompt_callbacks.ompt_callback(ompt_callback_control_tool)(
+          command, modifier, arg, OMPT_LOAD_RETURN_ADDRESS(__kmp_entry_gtid()));
+    } else {
+      return -1;
+    }
+  } else {
+    return -2;
+  }
+}
+
+/*****************************************************************************
+ * misc
+ ****************************************************************************/
+
+OMPT_API_ROUTINE uint64_t ompt_get_unique_id(void) {
+  return __ompt_get_unique_id_internal();
+
+/*****************************************************************************
+ * Target
+ ****************************************************************************/
+
+OMPT_API_ROUTINE int ompt_get_target_info(uint64_t *device_num,
+                                          ompt_id_t *target_id,
+                                          ompt_id_t *host_op_id) {
+  return 0; // thread is not in a target region
+}
+
+OMPT_API_ROUTINE int ompt_get_num_devices(void) {
+  return 1; // only one device (the current device) is available
 }
 
 /*****************************************************************************
  * API inquiry for tool
  ****************************************************************************/
 
 static ompt_interface_fn_t ompt_fn_lookup(const char *s) {
 
 #define ompt_interface_fn(fn)                                                  \
+  fn##_t fn##_f = fn;                                                          \
   if (strcmp(s, #fn) == 0)                                                     \
-    return (ompt_interface_fn_t)fn;
+    return (ompt_interface_fn_t)fn##_f;
 
   FOREACH_OMPT_INQUIRY_FN(ompt_interface_fn)
 
   FOREACH_OMPT_PLACEHOLDER_FN(ompt_interface_fn)
 
   return (ompt_interface_fn_t)0;
 }

runtime/src/ompt-internal.h

 #ifndef __OMPT_INTERNAL_H__
 #define __OMPT_INTERNAL_H__
 
 #include "ompt-event-specific.h"
 #include "ompt.h"
 
 #define OMPT_VERSION 1
 
 #define _OMP_EXTERN extern "C"
 
 #define OMPT_INVOKER(x)                                                        \
   ((x == fork_context_gnu) ? ompt_invoker_program : ompt_invoker_runtime)
 
 #define ompt_callback(e) e##_callback
 
-typedef struct ompt_callbacks_s {
+typedef struct ompt_callbacks_internal_s {
 #define ompt_event_macro(event, callback, eventid)                             \
   callback ompt_callback(event);
 
   FOREACH_OMPT_EVENT(ompt_event_macro)
 
 #undef ompt_event_macro
-} ompt_callbacks_t;
+} ompt_callbacks_internal_t;
+
+typedef struct ompt_callbacks_active_s {
+  unsigned int enabled : 1;
+#define ompt_event_macro(event, callback, eventid) unsigned int event : 1;
+
+  FOREACH_OMPT_EVENT(ompt_event_macro)
+
+#undef ompt_event_macro
+} ompt_callbacks_active_t;
+
+typedef struct kmp_taskdata kmp_taskdata_t;
+
+#define TASK_TYPE_DETAILS_FORMAT(info)                                         \
+  ((info->td_flags.task_serial || info->td_flags.tasking_ser)                  \
+       ? ompt_task_undeferred                                                  \
+       : 0x0) |                                                                \
+      ((!(info->td_flags.tiedness)) ? ompt_task_untied : 0x0) |                \
+      (info->td_flags.final ? ompt_task_final : 0x0) |                         \
+      (info->td_flags.merged_if0 ? ompt_task_mergeable : 0x0)
 
 typedef struct {
   ompt_frame_t frame;
-  void *function;
-  ompt_task_id_t task_id;
+  ompt_data_t task_data;
+  kmp_taskdata_t *scheduling_parent;
 #if OMP_40_ENABLED
   int ndeps;
   ompt_task_dependence_t *deps;
 #endif /* OMP_40_ENABLED */
 } ompt_task_info_t;
 
 typedef struct {
-  ompt_parallel_id_t parallel_id;
-  void *microtask;
+  ompt_data_t parallel_data;
+  void *master_return_address;
 } ompt_team_info_t;
 
 typedef struct ompt_lw_taskteam_s {
   ompt_team_info_t ompt_team_info;
   ompt_task_info_t ompt_task_info;
+  int heap;
   struct ompt_lw_taskteam_s *parent;
 } ompt_lw_taskteam_t;
 
-typedef struct ompt_parallel_info_s {
-  ompt_task_id_t parent_task_id; /* id of parent task            */
-  ompt_parallel_id_t parallel_id; /* id of parallel region        */
-  ompt_frame_t *parent_task_frame; /* frame data of parent task    */
-  void *parallel_function; /* pointer to outlined function */
-} ompt_parallel_info_t;
-
 typedef struct {
-  ompt_state_t state;
+  ompt_data_t thread_data;
+  ompt_data_t task_data; /* stored here from implicit barrier-begin until
+                            implicit-task-end */
+  void *return_address; /* stored here on entry of runtime */
+  omp_state_t state;
   ompt_wait_id_t wait_id;
+  int ompt_task_yielded;
   void *idle_frame;
 } ompt_thread_info_t;
 
-extern ompt_callbacks_t ompt_callbacks;
+extern ompt_callbacks_internal_t ompt_callbacks;
 
-#if OMP_40_ENABLED && OMPT_SUPPORT && OMPT_TRACE
+#if OMP_40_ENABLED && OMPT_SUPPORT && OMPT_OPTIONAL
 #if USE_FAST_MEMORY
 #define KMP_OMPT_DEPS_ALLOC __kmp_fast_allocate
 #define KMP_OMPT_DEPS_FREE __kmp_fast_free
 #else
 #define KMP_OMPT_DEPS_ALLOC __kmp_thread_malloc
 #define KMP_OMPT_DEPS_FREE __kmp_thread_free
 #endif
-#endif /* OMP_40_ENABLED && OMPT_SUPPORT && OMPT_TRACE */
+#endif /* OMP_40_ENABLED && OMPT_SUPPORT && OMPT_OPTIONAL */
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
 void ompt_pre_init(void);
 void ompt_post_init(void);
 void ompt_fini(void);
 
-extern int ompt_enabled;
+#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level)
+#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level)
+
+int __kmp_control_tool(uint64_t command, uint64_t modifier, void *arg);
+
+extern ompt_callbacks_active_t ompt_enabled;
+
+#if KMP_OS_WINDOWS
+#define UNLIKELY(x) (x)
+#define OMPT_NOINLINE __declspec(noinline)
+#else
+#define UNLIKELY(x) __builtin_expect(!!(x), 0)
+#define OMPT_NOINLINE __attribute__((noinline))
+#endif
 
 #ifdef __cplusplus
 };
 #endif
 
 #endif

runtime/src/ompt-specific.h

 #ifndef OMPT_SPECIFIC_H
 #define OMPT_SPECIFIC_H
 
 #include "kmp.h"
 
 /*****************************************************************************
  * types
  ****************************************************************************/
 
 typedef kmp_info_t ompt_thread_t;
 
 /*****************************************************************************
  * forward declarations
  ****************************************************************************/
 
-void __ompt_team_assign_id(kmp_team_t *team, ompt_parallel_id_t ompt_pid);
+void __ompt_team_assign_id(kmp_team_t *team, ompt_data_t ompt_pid);
 void __ompt_thread_assign_wait_id(void *variable);
 
 void __ompt_lw_taskteam_init(ompt_lw_taskteam_t *lwt, ompt_thread_t *thr,
-                             int gtid, void *microtask,
-                             ompt_parallel_id_t ompt_pid);
+                             int gtid, ompt_data_t *ompt_pid, void *codeptr);
 
-void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, ompt_thread_t *thr);
+void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, ompt_thread_t *thr,
+                             int on_heap);
 
-ompt_lw_taskteam_t *__ompt_lw_taskteam_unlink(ompt_thread_t *thr);
-
-ompt_parallel_id_t __ompt_parallel_id_new(int gtid);
-ompt_task_id_t __ompt_task_id_new(int gtid);
+void __ompt_lw_taskteam_unlink(ompt_thread_t *thr);
 
 ompt_team_info_t *__ompt_get_teaminfo(int depth, int *size);
 
-ompt_task_info_t *__ompt_get_taskinfo(int depth);
-
-void __ompt_thread_begin(ompt_thread_type_t thread_type, int gtid);
+ompt_task_info_t *__ompt_get_task_info_object(int depth);
 
-void __ompt_thread_end(ompt_thread_type_t thread_type, int gtid);
+int __ompt_get_parallel_info_internal(int ancestor_level,
+                                      ompt_data_t **parallel_data,
+                                      int *team_size);
 
-int __ompt_get_parallel_team_size_internal(int ancestor_level);
+int __ompt_get_task_info_internal(int ancestor_level, int *type,
+                                  ompt_data_t **task_data,
+                                  ompt_frame_t **task_frame,
+                                  ompt_data_t **parallel_data, int *thread_num);
 
-ompt_task_id_t __ompt_get_task_id_internal(int depth);
+ompt_data_t *__ompt_get_thread_data_internal();
 
-ompt_frame_t *__ompt_get_task_frame_internal(int depth);
+static uint64_t __ompt_get_get_unique_id_internal();
 
 /*****************************************************************************
  * macros
  ****************************************************************************/
 
+#define OMPT_CUR_TASK_INFO(thr) (&(thr->th.th_current_task->ompt_task_info))
+#define OMPT_CUR_TASK_DATA(thr)                                                \
+  (&(thr->th.th_current_task->ompt_task_info.task_data))
+#define OMPT_CUR_TEAM_INFO(thr) (&(thr->th.th_team->t.ompt_team_info))
+#define OMPT_CUR_TEAM_DATA(thr)                                                \
+  (&(thr->th.th_team->t.ompt_team_info.parallel_data))
+
 #define OMPT_HAVE_WEAK_ATTRIBUTE KMP_HAVE_WEAK_ATTRIBUTE
 #define OMPT_HAVE_PSAPI KMP_HAVE_PSAPI
 #define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle)
 
+inline void *__ompt_load_return_address(int gtid) {
+  kmp_info_t *thr = __kmp_threads[gtid];
+  void *return_address = thr->th.ompt_thread_info.return_address;
+  thr->th.ompt_thread_info.return_address = NULL;
+  return return_address;
+}
+
+#define OMPT_STORE_RETURN_ADDRESS(gtid)                                        \
+  if (ompt_enabled.enabled && gtid >= 0 && __kmp_threads[gtid] &&              \
+      !__kmp_threads[gtid]->th.ompt_thread_info.return_address)                \
+  __kmp_threads[gtid]->th.ompt_thread_info.return_address =                    \
+      __builtin_return_address(0)
+#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid)
+
 //******************************************************************************
 // inline functions
 //******************************************************************************
 
 inline ompt_thread_t *ompt_get_thread_gtid(int gtid) {
   return (gtid >= 0) ? __kmp_thread_from_gtid(gtid) : NULL;
 }
 
 inline ompt_thread_t *ompt_get_thread() {
   int gtid = __kmp_get_gtid();
   return ompt_get_thread_gtid(gtid);
 }
 
-inline void ompt_set_thread_state(ompt_thread_t *thread, ompt_state_t state) {
+inline void ompt_set_thread_state(ompt_thread_t *thread, omp_state_t state) {
   thread->th.ompt_thread_info.state = state;
 }
 
 inline const char *ompt_get_runtime_version() {
   return &__kmp_version_lib_ver[KMP_VERSION_MAGIC_LEN];
 }
 
 #endif

runtime/src/ompt-specific.cpp

 //******************************************************************************
 // include files
 //******************************************************************************
 
 #include "kmp.h"
 #include "ompt-internal.h"
 #include "ompt-specific.h"
 
+#if KMP_OS_UNIX
+#include <dlfcn.h>
+#endif
+
+#if KMP_OS_WINDOWS
+#define THREAD_LOCAL __declspec(thread)
+#else
+#define THREAD_LOCAL __thread
+#endif
+
 //******************************************************************************
 // macros
 //******************************************************************************
 
-#define GTID_TO_OMPT_THREAD_ID(id) ((ompt_thread_id_t)(id >= 0) ? id + 1 : 0)
-
-#define LWT_FROM_TEAM(team) (team)->t.ompt_serialized_team_info;
+#define LWT_FROM_TEAM(team) (team)->t.ompt_serialized_team_info
 
 #define OMPT_THREAD_ID_BITS 16
 
-// 2013 08 24 - John Mellor-Crummey
-//   ideally, a thread should assign its own ids based on thread private data.
-//   however, the way the intel runtime reinitializes thread data structures
-//   when it creates teams makes it difficult to maintain persistent thread
-//   data. using a shared variable instead is simple. I leave it to intel to
-//   sort out how to implement a higher performance version in their runtime.
-
-// when using fetch_and_add to generate the IDs, there isn't any reason to waste
-// bits for thread id.
-#if 0
-#define NEXT_ID(id_ptr, tid)                                                   \
-  ((KMP_TEST_THEN_INC64(id_ptr) << OMPT_THREAD_ID_BITS) | (tid))
-#else
-#define NEXT_ID(id_ptr, tid) (KMP_TEST_THEN_INC64((volatile kmp_int64 *)id_ptr))
-#endif
-
 //******************************************************************************
 // private operations
 //******************************************************************************
 
 //----------------------------------------------------------
 // traverse the team and task hierarchy
-// note: __ompt_get_teaminfo and __ompt_get_taskinfo
+// note: __ompt_get_teaminfo and __ompt_get_task_info_object
 //       traverse the hierarchy similarly and need to be
 //       kept consistent
 //----------------------------------------------------------
 
 ompt_team_info_t *__ompt_get_teaminfo(int depth, int *size) {
   kmp_info_t *thr = ompt_get_thread();
 
   if (thr) {
     kmp_team *team = thr->th.th_team;
     if (team == NULL)
       return NULL;
 
-    ompt_lw_taskteam_t *lwt = LWT_FROM_TEAM(team);
+    ompt_lw_taskteam_t *next_lwt = LWT_FROM_TEAM(team), *lwt = NULL;
 
     while (depth > 0) {
       // next lightweight team (if any)
       if (lwt)
         lwt = lwt->parent;
 
       // next heavyweight team (if any) after
       // lightweight teams are exhausted
       if (!lwt && team) {
+        if (next_lwt) {
+          lwt = next_lwt;
+          next_lwt = NULL;
+        } else {
           team = team->t.t_parent;
           if (team) {
-          lwt = LWT_FROM_TEAM(team);
+            next_lwt = LWT_FROM_TEAM(team);
+          }
         }
       }
 
       depth--;
     }
 
     if (lwt) {
       // lightweight teams have one task
if (lwt) {
       // return team info for heavyweight team
       return &team->t.ompt_team_info;
     }
   }
 
   return NULL;
 }
 
-ompt_task_info_t *__ompt_get_taskinfo(int depth) {
+ompt_task_info_t *__ompt_get_task_info_object(int depth) {
   ompt_task_info_t *info = NULL;
   kmp_info_t *thr = ompt_get_thread();
 
   if (thr) {
     kmp_taskdata_t *taskdata = thr->th.th_current_task;
-    ompt_lw_taskteam_t *lwt = LWT_FROM_TEAM(taskdata->td_team);
+    ompt_lw_taskteam_t *lwt = NULL,
+                       *next_lwt = LWT_FROM_TEAM(taskdata->td_team);
 
     while (depth > 0) {
       // next lightweight team (if any)
       if (lwt)
         lwt = lwt->parent;
 
       // next heavyweight team (if any) after
       // lightweight teams are exhausted
       if (!lwt && taskdata) {
+        if (next_lwt) {
+          lwt = next_lwt;
+          next_lwt = NULL;
+        } else {
           taskdata = taskdata->td_parent;
           if (taskdata) {
-          lwt = LWT_FROM_TEAM(taskdata->td_team);
+            next_lwt = LWT_FROM_TEAM(taskdata->td_team);
+          }
         }
       }
       depth--;
     }
 
     if (lwt) {
       info = &lwt->ompt_task_info;
     } else if (taskdata) {
       info = &taskdata->ompt_task_info;
     }
   }
 
   return info;
 }
 
-//******************************************************************************
-// interface operations
-//******************************************************************************
+ompt_task_info_t *__ompt_get_scheduling_taskinfo(int depth) {
+  ompt_task_info_t *info = NULL;
+  kmp_info_t *thr = ompt_get_thread();
 
-//----------------------------------------------------------
-// thread support
-//----------------------------------------------------------
+  if (thr) {
+    kmp_taskdata_t *taskdata = thr->th.th_current_task;
+
+    ompt_lw_taskteam_t *lwt = NULL,
+                       *next_lwt = LWT_FROM_TEAM(taskdata->td_team);
+
+    while (depth > 0) {
+      // next lightweight team (if any)
+      if (lwt)
+        lwt = lwt->parent;
 
-ompt_parallel_id_t __ompt_thread_id_new() {
-  static uint64_t ompt_thread_id = 1;
-  return NEXT_ID(&ompt_thread_id, 0);
+      // next heavyweight team (if any) after
+      // lightweight teams are exhausted
+      if (!lwt && taskdata) {
+        // first try scheduling parent (for explicit task scheduling)
+        if (taskdata->ompt_task_info.scheduling_parent) {
+          taskdata = taskdata->ompt_task_info.scheduling_parent;
+        } else if (next_lwt) {
+          lwt = next_lwt;
+          next_lwt = NULL;
+        } else {
+          // then go for implicit tasks
+          taskdata = taskdata->td_parent;
+          if (taskdata) {
+            next_lwt = LWT_FROM_TEAM(taskdata->td_team);
+          }
+        }
+      }
+      depth--;
     }
 
-void __ompt_thread_begin(ompt_thread_type_t thread_type, int gtid) {
-  ompt_callbacks.ompt_callback(ompt_event_thread_begin)(
-      thread_type, GTID_TO_OMPT_THREAD_ID(gtid));
+    if (lwt) {
+      info = &lwt->ompt_task_info;
+    } else if (taskdata) {
+      info = &taskdata->ompt_task_info;
+    }
   }
 
-void __ompt_thread_end(ompt_thread_type_t thread_type, int gtid) {
-  ompt_callbacks.ompt_callback(ompt_event_thread_end)(
-      thread_type, GTID_TO_OMPT_THREAD_ID(gtid));
+  return info;
 }
 
-ompt_thread_id_t __ompt_get_thread_id_internal() {
-  // FIXME: until we have a better way of assigning ids, use __kmp_get_gtid
-  // since the return value might be negative, we need to test that before
-  // assigning it to an ompt_thread_id_t, which is unsigned.
-  int id = __kmp_get_gtid();
-  assert(id >= 0);
+//******************************************************************************
+// interface operations
+//******************************************************************************
 
-  return GTID_TO_OMPT_THREAD_ID(id);
+//----------------------------------------------------------
+// thread support
+//----------------------------------------------------------
+
+ompt_data_t *__ompt_get_thread_data_internal() {
+  if (__kmp_get_gtid() >= 0) {
+    kmp_info_t *thread = ompt_get_thread();
+    if (thread == NULL)
+      return NULL;
+    return &(thread->th.ompt_thread_info.thread_data);
+  }
+  return NULL;
 }
 
 //----------------------------------------------------------
 // state support
 //----------------------------------------------------------
 
 void __ompt_thread_assign_wait_id(void *variable) {
-  int gtid = __kmp_gtid_get_specific();
-  kmp_info_t *ti = ompt_get_thread_gtid(gtid);
+  kmp_info_t *ti = ompt_get_thread();
 
   ti->th.ompt_thread_info.wait_id = (ompt_wait_id_t)variable;
 }
 
-ompt_state_t __ompt_get_state_internal(ompt_wait_id_t *ompt_wait_id) {
+omp_state_t __ompt_get_state_internal(ompt_wait_id_t *ompt_wait_id) {
   kmp_info_t *ti = ompt_get_thread();
 
   if (ti) {
     if (ompt_wait_id)
       *ompt_wait_id = ti->th.ompt_thread_info.wait_id;
     return ti->th.ompt_thread_info.state;
   }
-  return ompt_state_undefined;
-}
-
-//----------------------------------------------------------
-// idle frame support
-//----------------------------------------------------------
-
-void *__ompt_get_idle_frame_internal(void) {
-  kmp_info_t *ti = ompt_get_thread();
-  return ti ? ti->th.ompt_thread_info.idle_frame : NULL;
+  return omp_state_undefined;
 }
 
 //----------------------------------------------------------
 // parallel region support
 //----------------------------------------------------------
 
-ompt_parallel_id_t __ompt_parallel_id_new(int gtid) {
-  static uint64_t ompt_parallel_id = 1;
-  return gtid >= 0 ? NEXT_ID(&ompt_parallel_id, gtid) : 0;
-}
-
-void *__ompt_get_parallel_function_internal(int depth) {
-  ompt_team_info_t *info = __ompt_get_teaminfo(depth, NULL);
-  void *function = info ? info->microtask : NULL;
-  return function;
-}
-
-ompt_parallel_id_t __ompt_get_parallel_id_internal(int depth) {
-  ompt_team_info_t *info = __ompt_get_teaminfo(depth, NULL);
-  ompt_parallel_id_t id = info ? info->parallel_id : 0;
-  return id;
-}
-
-int __ompt_get_parallel_team_size_internal(int depth) {
-  // initialize the return value with the error value.
-  // if there is a team at the specified depth, the default
-  // value will be overwritten the size of that team.
-  int size = -1;
-  (void)__ompt_get_teaminfo(depth, &size);
-  return size;
+int __ompt_get_parallel_info_internal(int ancestor_level,
+                                      ompt_data_t **parallel_data,
+                                      int *team_size) {
+  ompt_team_info_t *info;
+  if (team_size) {
+    info = __ompt_get_teaminfo(ancestor_level, team_size);
+  } else {
+    info = __ompt_get_teaminfo(ancestor_level, NULL);
+  }
+  if (parallel_data) {
+    *parallel_data = info ? &(info->parallel_data) : NULL;
+  }
+  return info ? 2 : 0;
 }
 
 //----------------------------------------------------------
 // lightweight task team support
 //----------------------------------------------------------
 
 void __ompt_lw_taskteam_init(ompt_lw_taskteam_t *lwt, kmp_info_t *thr, int gtid,
-                             void *microtask, ompt_parallel_id_t ompt_pid) {
-  lwt->ompt_team_info.parallel_id = ompt_pid;
-  lwt->ompt_team_info.microtask = microtask;
-  lwt->ompt_task_info.task_id = 0;
+                             ompt_data_t *ompt_pid, void *codeptr) {
+  // initialize parallel_data with input, return address to parallel_data on
+  // exit
+  lwt->ompt_team_info.parallel_data = *ompt_pid;
+  lwt->ompt_team_info.master_return_address = codeptr;
+  lwt->ompt_task_info.task_data.value = 0;
   lwt->ompt_task_info.frame.reenter_runtime_frame = NULL;
   lwt->ompt_task_info.frame.exit_runtime_frame = NULL;
-  lwt->ompt_task_info.function = NULL;
+  lwt->ompt_task_info.scheduling_parent = NULL;
+  lwt->ompt_task_info.deps = NULL;
+  lwt->ompt_task_info.ndeps = 0;
+  lwt->heap = 0;
   lwt->parent = 0;
 }
 
-void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, kmp_info_t *thr) {
-  ompt_lw_taskteam_t *my_parent = thr->th.th_team->t.ompt_serialized_team_info;
-  lwt->parent = my_parent;
-  thr->th.th_team->t.ompt_serialized_team_info = lwt;
+void __ompt_lw_taskteam_link(ompt_lw_taskteam_t *lwt, kmp_info_t *thr,
+                             int on_heap) {
+  ompt_lw_taskteam_t *link_lwt = lwt;
+  if (thr->th.th_team->t.t_serialized >
+      1) { // we already have a team, so link the new team and swap values
+    if (on_heap) { // the lw_taskteam cannot stay on stack, allocate it on heap
+      link_lwt =
+          (ompt_lw_taskteam_t *)__kmp_allocate(sizeof(ompt_lw_taskteam_t));
+    }
+    link_lwt->heap = on_heap;
+
+    // would be swap in the (on_stack) case.
+    ompt_team_info_t tmp_team = lwt->ompt_team_info;
+    link_lwt->ompt_team_info = *OMPT_CUR_TEAM_INFO(thr);
+    *OMPT_CUR_TEAM_INFO(thr) = tmp_team;
+
+    ompt_task_info_t tmp_task = lwt->ompt_task_info;
+    link_lwt->ompt_task_info = *OMPT_CUR_TASK_INFO(thr);
+    *OMPT_CUR_TASK_INFO(thr) = tmp_task;
+
+    // link the taskteam into the list of taskteams:
+    ompt_lw_taskteam_t *my_parent =
+        thr->th.th_team->t.ompt_serialized_team_info;
+    link_lwt->parent = my_parent;
+    thr->th.th_team->t.ompt_serialized_team_info = link_lwt;
+  } else {
+    // this is the first serialized team, so we just store the values in the
+    // team and drop the taskteam-object
+    *OMPT_CUR_TEAM_INFO(thr) = lwt->ompt_team_info;
+    *OMPT_CUR_TASK_INFO(thr) = lwt->ompt_task_info;
+  }
 }
 
-ompt_lw_taskteam_t *__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;
-  if (lwtask)
+  if (lwtask) {
     thr->th.th_team->t.ompt_serialized_team_info = lwtask->parent;
-  return lwtask;
+
+    ompt_team_info_t tmp_team = lwtask->ompt_team_info;
+    lwtask->ompt_team_info = *OMPT_CUR_TEAM_INFO(thr);
+    *OMPT_CUR_TEAM_INFO(thr) = tmp_team;
+
+    ompt_task_info_t tmp_task = lwtask->ompt_task_info;
+    lwtask->ompt_task_info = *OMPT_CUR_TASK_INFO(thr);
+    *OMPT_CUR_TASK_INFO(thr) = tmp_task;
+
+    if (lwtask->heap) {
+      __kmp_free(lwtask);
+      lwtask = NULL;
+    }
+  }
+  //    return lwtask;
 }
 
 //----------------------------------------------------------
 // task support
 //----------------------------------------------------------
 
-ompt_task_id_t __ompt_task_id_new(int gtid) {
-  static uint64_t ompt_task_id = 1;
-  return NEXT_ID(&ompt_task_id, gtid);
-}
+int __ompt_get_task_info_internal(int ancestor_level, int *type,
+                                  ompt_data_t **task_data,
+                                  ompt_frame_t **task_frame,
+                                  ompt_data_t **parallel_data,
+                                  int *thread_num) {
+  if (ancestor_level < 0)
+    return 0;
 
-ompt_task_id_t __ompt_get_task_id_internal(int depth) {
-  ompt_task_info_t *info = __ompt_get_taskinfo(depth);
-  ompt_task_id_t task_id = info ? info->task_id : 0;
-  return task_id;
-}
+  // copied from __ompt_get_scheduling_taskinfo
+  ompt_task_info_t *info = NULL;
+  ompt_team_info_t *team_info = NULL;
+  kmp_info_t *thr = ompt_get_thread();
 
-void *__ompt_get_task_function_internal(int depth) {
-  ompt_task_info_t *info = __ompt_get_taskinfo(depth);
-  void *function = info ? info->function : NULL;
-  return function;
+  if (thr) {
+    kmp_taskdata_t *taskdata = thr->th.th_current_task;
+    if (taskdata == NULL)
+      return 0;
+    kmp_team *team = thr->th.th_team;
+    if (team == NULL)
+      return 0;
+    ompt_lw_taskteam_t *lwt = NULL,
+                       *next_lwt = LWT_FROM_TEAM(taskdata->td_team);
+
+    while (ancestor_level > 0) {
+      // next lightweight team (if any)
+      if (lwt)
+        lwt = lwt->parent;
+
+      // next heavyweight team (if any) after
+      // lightweight teams are exhausted
+      if (!lwt && taskdata) {
+        // first try scheduling parent (for explicit task scheduling)
+        if (taskdata->ompt_task_info.scheduling_parent) {
+          taskdata = taskdata->ompt_task_info.scheduling_parent;
+        } else if (next_lwt) {
+          lwt = next_lwt;
+          next_lwt = NULL;
+        } else {
+          // then go for implicit tasks
+          taskdata = taskdata->td_parent;
+          if (team == NULL)
+            return 0;
+          team = team->t.t_parent;
+          if (taskdata) {
+            next_lwt = LWT_FROM_TEAM(taskdata->td_team);
+          }
+        }
+      }
+      ancestor_level--;
     }
 
-ompt_frame_t *__ompt_get_task_frame_internal(int depth) {
-  ompt_task_info_t *info = __ompt_get_taskinfo(depth);
-  ompt_frame_t *frame = info ? frame = &info->frame : NULL;
-  return frame;
+    if (lwt) {
+      info = &lwt->ompt_task_info;
+      team_info = &lwt->ompt_team_info;
+      if (type) {
+        *type = ompt_task_implicit;
+      }
+    } else if (taskdata) {
+      info = &taskdata->ompt_task_info;
+      team_info = &team->t.ompt_team_info;
+      if (type) {
+        if (taskdata->td_parent) {
+          *type = (taskdata->td_flags.tasktype ? ompt_task_explicit
+                                               : ompt_task_implicit) |
+                  TASK_TYPE_DETAILS_FORMAT(taskdata);
+        } else {
+          *type = ompt_task_initial;
+        }
+      }
+    }
+    if (task_data) {
+      *task_data = info ? &info->task_data : NULL;
+    }
+    if (task_frame) {
+      // OpenMP spec asks for the scheduling task to be returned.
+      *task_frame = info ? &info->frame : NULL;
+    }
+    if (parallel_data) {
+      *parallel_data = team_info ? &(team_info->parallel_data) : NULL;
+    }
+    return info ? 2 : 0;
+  }
+  return 0;
 }
 
 //----------------------------------------------------------
 // team support
 //----------------------------------------------------------
 
-void __ompt_team_assign_id(kmp_team_t *team, ompt_parallel_id_t ompt_pid) {
-  team->t.ompt_team_info.parallel_id = ompt_pid;
+void __ompt_team_assign_id(kmp_team_t *team, ompt_data_t ompt_pid) {
+  team->t.ompt_team_info.parallel_data = ompt_pid;
+}
+
+//----------------------------------------------------------
+// misc
+//----------------------------------------------------------
+
+static uint64_t __ompt_get_unique_id_internal() {
+  static uint64_t thread = 1;
+  static THREAD_LOCAL uint64_t ID = 0;
+  if (ID == 0) {
+    uint64_t new_thread = KMP_TEST_THEN_INC64((kmp_int64 *)&thread);
+    ID = new_thread << (sizeof(uint64_t) * 8 - OMPT_THREAD_ID_BITS);
+  }
+  return ++ID;
 }

runtime/src/z_Linux_util.cpp

 int __kmp_invoke_microtask(microtask_t pkfn, int gtid, int tid, int argc,
                            void *p_argv[]
 #if OMPT_SUPPORT
                            ,
                            void **exit_frame_ptr
 #endif
                            ) {
 #if OMPT_SUPPORT
-  *exit_frame_ptr = __builtin_frame_address(0);
+  *exit_frame_ptr = OMPT_GET_FRAME_ADDRESS(0);
 #endif
 
   switch (argc) {
   default:
     fprintf(stderr, "Too many args to microtask: %d!\n", argc);
     fflush(stderr);
     exit(-1);
   case 0:

runtime/test/CMakeLists.txt

if (${var})
     set(${var} True)
   else()
     set(${var} False)
   endif()
 endmacro()
 
 pythonize_bool(LIBOMP_USE_HWLOC)
 pythonize_bool(LIBOMP_OMPT_SUPPORT)
-pythonize_bool(LIBOMP_OMPT_BLAME)
-pythonize_bool(LIBOMP_OMPT_TRACE)
+pythonize_bool(LIBOMP_OMPT_OPTIONAL)
 pythonize_bool(LIBOMP_HAVE_LIBM)
 pythonize_bool(LIBOMP_HAVE_LIBATOMIC)
 
 set(LIBOMP_TEST_CFLAGS "" CACHE STRING
   "Extra compiler flags to send to the test compiler")
 
 if(${LIBOMP_STANDALONE_BUILD})
   # Make sure we can use the console pool for recent cmake and ninja > 1.5

runtime/test/lit.cfg

if config.has_ompt and config.test_filecheck == "":
     lit_config.note("Not testing OMPT because FileCheck was not found")
     config.has_ompt = False
 
 if config.has_ompt:
     config.available_features.add("ompt")
     # for callback.h
     config.test_cflags += " -I " + config.test_source_root + "/ompt"
 
+if 'Linux' in config.operating_system:
+    config.available_features.add("linux")
+
 # to run with icc INTEL_LICENSE_FILE must be set
 if 'INTEL_LICENSE_FILE' in os.environ:
     config.environment['INTEL_LICENSE_FILE'] = os.environ['INTEL_LICENSE_FILE']
 
-# substitutions
-if config.has_ompt:
-    config.substitutions.append(("FileCheck", config.test_filecheck))
-    config.substitutions.append(("%sort-threads", "sort --numeric-sort --stable"))
 
+# substitutions
 config.substitutions.append(("%libomp-compile-and-run", \
     "%libomp-compile && %libomp-run"))
 config.substitutions.append(("%libomp-cxx-compile-and-run", \
     "%libomp-cxx-compile && %libomp-run"))
 config.substitutions.append(("%libomp-cxx-compile", \
     "%clangXX %cflags -std=c++11 %s -o %t" + libs))
 config.substitutions.append(("%libomp-compile", \
     "%clang %cflags %s -o %t" + libs))
+config.substitutions.append(("%libomp-tool", \
+    "%clang %cflags -shared -fPIC -o %T/tool.so" + libs))
 config.substitutions.append(("%libomp-run", "%t"))
 config.substitutions.append(("%clangXX", config.test_cxx_compiler))
 config.substitutions.append(("%clang", config.test_compiler))
 config.substitutions.append(("%openmp_flag", config.test_openmp_flag))
 config.substitutions.append(("%cflags", config.test_cflags))
 
+if config.has_ompt:
+    config.substitutions.append(("FileCheck", config.test_filecheck))
+    config.substitutions.append(("%sort-threads", "sort --numeric-sort --stable"))

runtime/test/lit.site.cfg.in

 @AUTO_GEN_COMMENT@
 
 config.test_compiler = "@LIBOMP_TEST_COMPILER@"
 config.test_cxx_compiler = "@LIBOMP_TEST_CXX_COMPILER@"
 config.test_filecheck = "@FILECHECK_EXECUTABLE@"
 config.test_openmp_flag = "@LIBOMP_TEST_OPENMP_FLAG@"
 config.test_extra_cflags = "@LIBOMP_TEST_CFLAGS@"
 config.libomp_obj_root = "@CMAKE_CURRENT_BINARY_DIR@"
 config.library_dir = "@LIBOMP_LIBRARY_DIR@"
 config.omp_header_directory = "@LIBOMP_BINARY_DIR@/src"
 config.operating_system = "@CMAKE_SYSTEM_NAME@"
 config.hwloc_library_dir = "@LIBOMP_HWLOC_LIBRARY_DIR@"
 config.using_hwloc = @LIBOMP_USE_HWLOC@
-config.has_ompt = @LIBOMP_OMPT_SUPPORT@ and @LIBOMP_OMPT_BLAME@ and @LIBOMP_OMPT_TRACE@
+config.has_ompt = @LIBOMP_OMPT_SUPPORT@ and @LIBOMP_OMPT_OPTIONAL@
 config.has_libm = @LIBOMP_HAVE_LIBM@
 config.has_libatomic = @LIBOMP_HAVE_LIBATOMIC@
 
 # Let the main config do the real work.
 lit_config.load_config(config, "@LIBOMP_BASE_DIR@/test/lit.cfg")

runtime/test/ompt/callback.h

+#define _BSD_SOURCE
 #include <stdio.h>
 #include <inttypes.h>
+#include <omp.h>
 #include <ompt.h>
+#include "ompt-signal.h"
 
-static ompt_get_task_id_t ompt_get_task_id;
-static ompt_get_task_frame_t ompt_get_task_frame;
-static ompt_get_thread_id_t ompt_get_thread_id;
-static ompt_get_parallel_id_t ompt_get_parallel_id;
+static const char* ompt_thread_type_t_values[] = {
+  NULL,
+  "ompt_thread_initial",
+  "ompt_thread_worker",
+  "ompt_thread_other"
+};
+
+static const char* ompt_task_status_t_values[] = {
+  NULL,
+  "ompt_task_complete",
+  "ompt_task_yield",
+  "ompt_task_cancel",
+  "ompt_task_others"
+};
+static const char* ompt_cancel_flag_t_values[] = {
+  "ompt_cancel_parallel",
+  "ompt_cancel_sections",
+  "ompt_cancel_do",
+  "ompt_cancel_taskgroup",
+  "ompt_cancel_activated",
+  "ompt_cancel_detected",
+  "ompt_cancel_discarded_task"
+};
+
+static ompt_set_callback_t ompt_set_callback;
+static ompt_get_task_info_t ompt_get_task_info;
+static ompt_get_thread_data_t ompt_get_thread_data;
+static ompt_get_parallel_info_t ompt_get_parallel_info;
+static ompt_get_unique_id_t ompt_get_unique_id;
+static ompt_get_num_places_t ompt_get_num_places;
+static ompt_get_place_proc_ids_t ompt_get_place_proc_ids;
+static ompt_get_place_num_t ompt_get_place_num;
+static ompt_get_partition_place_nums_t ompt_get_partition_place_nums;
+static ompt_get_proc_id_t ompt_get_proc_id;
+static ompt_enumerate_states_t ompt_enumerate_states;
+static ompt_enumerate_mutex_impls_t ompt_enumerate_mutex_impls;
 
 static void print_ids(int level)
 {
-  ompt_frame_t* frame = ompt_get_task_frame(level);
-  printf("%" PRIu64 ": level %d: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", exit_frame=%p, reenter_frame=%p\n", ompt_get_thread_id(), level, ompt_get_parallel_id(level), ompt_get_task_id(level), frame->exit_runtime_frame, frame->reenter_runtime_frame);
+  ompt_frame_t* frame ;
+  ompt_data_t* parallel_data;
+  ompt_data_t* task_data;
+  int exists_task = ompt_get_task_info(level, NULL, &task_data, &frame, &parallel_data, NULL);
+  if (frame)
+  {
+    printf("%" PRIu64 ": task level %d: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", exit_frame=%p, reenter_frame=%p\n", ompt_get_thread_data()->value, level, exists_task ? parallel_data->value : 0, exists_task ? task_data->value : 0, frame->exit_runtime_frame, frame->reenter_runtime_frame);
+  }
+  else
+    printf("%" PRIu64 ": task level %d: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", frame=%p\n", ompt_get_thread_data()->value, level, exists_task ? parallel_data->value : 0, exists_task ? task_data->value : 0, frame);
 }
 
 #define print_frame(level)\
 do {\
-  printf("%" PRIu64 ": __builtin_frame_address(%d)=%p\n", ompt_get_thread_id(), level, __builtin_frame_address(level));\
+  printf("%" PRIu64 ": __builtin_frame_address(%d)=%p\n", ompt_get_thread_data()->value, level, __builtin_frame_address(level));\
 } while(0)
 
+#define print_current_address(id)\
+{}              /* Empty block between "#pragma omp ..." and __asm__ statement as a workaround for icc bug */ \
+__asm__("nop"); /* provide an instruction as jump target (compiler would insert an instruction if label is target of a jmp ) */ \
+ompt_label_##id:\
+    printf("%" PRIu64 ": current_address=%p or %p\n", ompt_get_thread_data()->value, (char*)(&& ompt_label_##id)-1, (char*)(&& ompt_label_##id)-4) 
+    /* "&& label" returns the address of the label (GNU extension); works with gcc, clang, icc */
+    /* for void-type runtime function, the label is after the nop (-1), for functions with return value, there is a mov instruction before the label (-4) */
+
+#define print_fuzzy_address(id)\
+{}              /* Empty block between "#pragma omp ..." and __asm__ statement as a workaround for icc bug */ \
+__asm__("nop"); /* provide an instruction as jump target (compiler would insert an instruction if label is target of a jmp ) */ \
+ompt_label_##id:\
+    printf("%" PRIu64 ": fuzzy_address=0x%lx or 0x%lx\n", ompt_get_thread_data()->value, ((uint64_t)(char*)(&& ompt_label_##id))/256-1, ((uint64_t)(char*)(&& ompt_label_##id))/256) 
+    /* "&& label" returns the address of the label (GNU extension); works with gcc, clang, icc */
+    /* for void-type runtime function, the label is after the nop (-1), for functions with return value, there is a mov instruction before the label (-4) */
+
+static void format_task_type(int type, char* buffer)
+{
+  char* progress = buffer;
+  if(type & ompt_task_initial) progress += sprintf(progress, "ompt_task_initial");
+  if(type & ompt_task_implicit) progress += sprintf(progress, "ompt_task_implicit");
+  if(type & ompt_task_explicit) progress += sprintf(progress, "ompt_task_explicit");
+  if(type & ompt_task_target) progress += sprintf(progress, "ompt_task_target");
+  if(type & ompt_task_undeferred) progress += sprintf(progress, "|ompt_task_undeferred");
+  if(type & ompt_task_untied) progress += sprintf(progress, "|ompt_task_untied");
+  if(type & ompt_task_final) progress += sprintf(progress, "|ompt_task_final");
+  if(type & ompt_task_mergeable) progress += sprintf(progress, "|ompt_task_mergeable");
+  if(type & ompt_task_merged) progress += sprintf(progress, "|ompt_task_merged");
+}
+
+static void
+on_ompt_callback_mutex_acquire(
+  ompt_mutex_kind_t kind,
+  unsigned int hint,
+  unsigned int impl,
+  ompt_wait_id_t wait_id,
+  const void *codeptr_ra)
+{
+  switch(kind)
+  {
+    case ompt_mutex_lock:
+      printf("%" PRIu64 ": ompt_event_wait_lock: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    case ompt_mutex_nest_lock:
+      printf("%" PRIu64 ": ompt_event_wait_nest_lock: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    case ompt_mutex_critical:
+      printf("%" PRIu64 ": ompt_event_wait_critical: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    case ompt_mutex_atomic:
+      printf("%" PRIu64 ": ompt_event_wait_atomic: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    case ompt_mutex_ordered:
+      printf("%" PRIu64 ": ompt_event_wait_ordered: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    default:
+      break;
+  }
+}
+
+static void
+on_ompt_callback_mutex_acquired(
+  ompt_mutex_kind_t kind,
+  ompt_wait_id_t wait_id,
+  const void *codeptr_ra)
+{
+  switch(kind)
+  {
+    case ompt_mutex_lock:
+      printf("%" PRIu64 ": ompt_event_acquired_lock: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_nest_lock:
+      printf("%" PRIu64 ": ompt_event_acquired_nest_lock_first: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_critical:
+      printf("%" PRIu64 ": ompt_event_acquired_critical: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_atomic:
+      printf("%" PRIu64 ": ompt_event_acquired_atomic: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_ordered:
+      printf("%" PRIu64 ": ompt_event_acquired_ordered: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    default:
+      break;
+  }
+}
+
+static void
+on_ompt_callback_mutex_released(
+  ompt_mutex_kind_t kind,
+  ompt_wait_id_t wait_id,
+  const void *codeptr_ra)
+{
+  switch(kind)
+  {
+    case ompt_mutex_lock:
+      printf("%" PRIu64 ": ompt_event_release_lock: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_nest_lock:
+      printf("%" PRIu64 ": ompt_event_release_nest_lock_last: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_critical:
+      printf("%" PRIu64 ": ompt_event_release_critical: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_atomic:
+      printf("%" PRIu64 ": ompt_event_release_atomic: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_ordered:
+      printf("%" PRIu64 ": ompt_event_release_ordered: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    default:
+      break;
+  }
+}
+
+static void
+on_ompt_callback_nest_lock(
+    ompt_scope_endpoint_t endpoint,
+    ompt_wait_id_t wait_id,
+    const void *codeptr_ra)
+{
+  switch(endpoint)
+  {
+    case ompt_scope_begin:
+      printf("%" PRIu64 ": ompt_event_acquired_nest_lock_next: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_scope_end:
+      printf("%" PRIu64 ": ompt_event_release_nest_lock_prev: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+  }
+}
 
 static void
-on_ompt_event_barrier_begin(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id)
+on_ompt_callback_sync_region(
+  ompt_sync_region_kind_t kind,
+  ompt_scope_endpoint_t endpoint,
+  ompt_data_t *parallel_data,
+  ompt_data_t *task_data,
+  const void *codeptr_ra)
+{
+  switch(endpoint)
   {
-  printf("%" PRIu64 ": ompt_event_barrier_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 "\n", ompt_get_thread_id(), parallel_id, task_id);
+    case ompt_scope_begin:
+      switch(kind)
+      {
+        case ompt_sync_region_barrier:
+          printf("%" PRIu64 ": ompt_event_barrier_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
           print_ids(0);
+          break;
+        case ompt_sync_region_taskwait:
+          printf("%" PRIu64 ": ompt_event_taskwait_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskgroup:
+          printf("%" PRIu64 ": ompt_event_taskgroup_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+          break;
+      }
+      break;
+    case ompt_scope_end:
+      switch(kind)
+      {
+        case ompt_sync_region_barrier:
+          printf("%" PRIu64 ": ompt_event_barrier_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskwait:
+          printf("%" PRIu64 ": ompt_event_taskwait_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskgroup:
+          printf("%" PRIu64 ": ompt_event_taskgroup_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+      }
+      break;
+  }
+}
+
+static void
+on_ompt_callback_sync_region_wait(
+  ompt_sync_region_kind_t kind,
+  ompt_scope_endpoint_t endpoint,
+  ompt_data_t *parallel_data,
+  ompt_data_t *task_data,
+  const void *codeptr_ra)
+{
+  switch(endpoint)
+  {
+    case ompt_scope_begin:
+      switch(kind)
+      {
+        case ompt_sync_region_barrier:
+          printf("%" PRIu64 ": ompt_event_wait_barrier_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskwait:
+          printf("%" PRIu64 ": ompt_event_wait_taskwait_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskgroup:
+          printf("%" PRIu64 ": ompt_event_wait_taskgroup_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+          break;
+      }
+      break;
+    case ompt_scope_end:
+      switch(kind)
+      {
+        case ompt_sync_region_barrier:
+          printf("%" PRIu64 ": ompt_event_wait_barrier_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskwait:
+          printf("%" PRIu64 ": ompt_event_wait_taskwait_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+        case ompt_sync_region_taskgroup:
+          printf("%" PRIu64 ": ompt_event_wait_taskgroup_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, codeptr_ra);
+          break;
+      }
+      break;
+  }
 }
 
 static void
-on_ompt_event_barrier_end(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id)
+on_ompt_callback_flush(
+    ompt_data_t *thread_data,
+    const void *codeptr_ra)
 {
-  printf("%" PRIu64 ": ompt_event_barrier_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 "\n", ompt_get_thread_id(), parallel_id, task_id);
+  printf("%" PRIu64 ": ompt_event_flush: codeptr_ra=%p\n", thread_data->value, codeptr_ra);
 }
 
 static void
-on_ompt_event_implicit_task_begin(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id)
+on_ompt_callback_cancel(
+    ompt_data_t *task_data,
+    int flags,
+    const void *codeptr_ra)
+{
+  const char* first_flag_value;
+  const char* second_flag_value;
+  if(flags & ompt_cancel_parallel)
+    first_flag_value = ompt_cancel_flag_t_values[0];
+  else if(flags & ompt_cancel_sections)
+    first_flag_value = ompt_cancel_flag_t_values[1];
+  else if(flags & ompt_cancel_do)
+    first_flag_value = ompt_cancel_flag_t_values[2];
+  else if(flags & ompt_cancel_taskgroup)
+    first_flag_value = ompt_cancel_flag_t_values[3];
+
+  if(flags & ompt_cancel_activated)
+    second_flag_value = ompt_cancel_flag_t_values[4];
+  else if(flags & ompt_cancel_detected)
+    second_flag_value = ompt_cancel_flag_t_values[5];
+  else if(flags & ompt_cancel_discarded_task)
+    second_flag_value = ompt_cancel_flag_t_values[6];
+    
+  printf("%" PRIu64 ": ompt_event_cancel: task_data=%" PRIu64 ", flags=%s|%s=%" PRIu32 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, task_data->value, first_flag_value, second_flag_value, flags,  codeptr_ra);
+}
+
+static void
+on_ompt_callback_idle(
+  ompt_scope_endpoint_t endpoint)
+{
+  switch(endpoint)
   {
-  printf("%" PRIu64 ": ompt_event_implicit_task_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 "\n", ompt_get_thread_id(), parallel_id, task_id);
+    case ompt_scope_begin:
+      printf("%" PRIu64 ": ompt_event_idle_begin:\n", ompt_get_thread_data()->value);
+      break;
+    case ompt_scope_end:
+      printf("%" PRIu64 ": ompt_event_idle_end:\n", ompt_get_thread_data()->value);
+      break;
+  }
 }
 
 static void
-on_ompt_event_implicit_task_end(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id)
+on_ompt_callback_implicit_task(
+    ompt_scope_endpoint_t endpoint,
+    ompt_data_t *parallel_data,
+    ompt_data_t *task_data,
+    unsigned int team_size,
+    unsigned int thread_num)
+{
+  switch(endpoint)
+  {
+    case ompt_scope_begin:
+      if(task_data->ptr)
+        printf("%s\n", "0: task_data initially not null");
+      task_data->value = ompt_get_unique_id();
+      printf("%" PRIu64 ": ompt_event_implicit_task_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", team_size=%" PRIu32 ", thread_num=%" PRIu32 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, team_size, thread_num);
+      break;
+    case ompt_scope_end:
+      printf("%" PRIu64 ": ompt_event_implicit_task_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", team_size=%" PRIu32 ", thread_num=%" PRIu32 "\n", ompt_get_thread_data()->value, (parallel_data)?parallel_data->value:0, task_data->value, team_size, thread_num);
+      break;
+  }
+}
+
+static void
+on_ompt_callback_lock_init(
+  ompt_mutex_kind_t kind,
+  unsigned int hint,
+  unsigned int impl,
+  ompt_wait_id_t wait_id,
+  const void *codeptr_ra)
+{
+  switch(kind)
+  {
+    case ompt_mutex_lock:
+      printf("%" PRIu64 ": ompt_event_init_lock: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    case ompt_mutex_nest_lock:
+      printf("%" PRIu64 ": ompt_event_init_nest_lock: wait_id=%" PRIu64 ", hint=%" PRIu32 ", impl=%" PRIu32 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, hint, impl, codeptr_ra);
+      break;
+    default:
+      break;
+  }
+}
+
+static void
+on_ompt_callback_lock_destroy(
+  ompt_mutex_kind_t kind,
+  ompt_wait_id_t wait_id,
+  const void *codeptr_ra)
+{
+  switch(kind)
+  {
+    case ompt_mutex_lock:
+      printf("%" PRIu64 ": ompt_event_destroy_lock: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    case ompt_mutex_nest_lock:
+      printf("%" PRIu64 ": ompt_event_destroy_nest_lock: wait_id=%" PRIu64 ", codeptr_ra=%p \n", ompt_get_thread_data()->value, wait_id, codeptr_ra);
+      break;
+    default:
+      break;
+  }
+}
+
+static void
+on_ompt_callback_work(
+  ompt_work_type_t wstype,
+  ompt_scope_endpoint_t endpoint,
+  ompt_data_t *parallel_data,
+  ompt_data_t *task_data,
+  uint64_t count,
+  const void *codeptr_ra)
+{
+  switch(endpoint)
+  {
+    case ompt_scope_begin:
+      switch(wstype)
+      {
+        case ompt_work_loop:
+          printf("%" PRIu64 ": ompt_event_loop_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_sections:
+          printf("%" PRIu64 ": ompt_event_sections_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_single_executor:
+          printf("%" PRIu64 ": ompt_event_single_in_block_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_single_other:
+          printf("%" PRIu64 ": ompt_event_single_others_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_workshare:
+          //impl
+          break;
+        case ompt_work_distribute:
+          printf("%" PRIu64 ": ompt_event_distribute_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_taskloop:
+          //impl
+          printf("%" PRIu64 ": ompt_event_taskloop_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+      }
+      break;
+    case ompt_scope_end:
+      switch(wstype)
+      {
+        case ompt_work_loop:
+          printf("%" PRIu64 ": ompt_event_loop_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_sections:
+          printf("%" PRIu64 ": ompt_event_sections_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_single_executor:
+          printf("%" PRIu64 ": ompt_event_single_in_block_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_single_other:
+          printf("%" PRIu64 ": ompt_event_single_others_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_workshare:
+          //impl
+          break;
+        case ompt_work_distribute:
+          printf("%" PRIu64 ": ompt_event_distribute_end: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+        case ompt_work_taskloop:
+          //impl
+          printf("%" PRIu64 ": ompt_event_taskloop_end: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", codeptr_ra=%p, count=%" PRIu64 "\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra, count);
+          break;
+      }
+      break;
+  }
+}
+
+static void
+on_ompt_callback_master(
+  ompt_scope_endpoint_t endpoint,
+  ompt_data_t *parallel_data,
+  ompt_data_t *task_data,
+  const void *codeptr_ra)
+{
+  switch(endpoint)
+  {
+    case ompt_scope_begin:
+      printf("%" PRIu64 ": ompt_event_master_begin: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+      break;
+    case ompt_scope_end:
+      printf("%" PRIu64 ": ompt_event_master_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, codeptr_ra);
+      break;
+  }
+}
+
+static void
+on_ompt_callback_parallel_begin(
+  ompt_data_t *parent_task_data,
+  const ompt_frame_t *parent_task_frame,
+  ompt_data_t* parallel_data,
+  uint32_t requested_team_size,
+  ompt_invoker_t invoker,
+  const void *codeptr_ra)
 {
-  printf("%" PRIu64 ": ompt_event_implicit_task_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 "\n", ompt_get_thread_id(), parallel_id, task_id);
+  if(parallel_data->ptr)
+    printf("%s\n", "0: parallel_data initially not null");
+  parallel_data->value = ompt_get_unique_id();
+  printf("%" PRIu64 ": ompt_event_parallel_begin: parent_task_id=%" PRIu64 ", parent_task_frame.exit=%p, parent_task_frame.reenter=%p, parallel_id=%" PRIu64 ", requested_team_size=%" PRIu32 ", codeptr_ra=%p, invoker=%d\n", ompt_get_thread_data()->value, parent_task_data->value, parent_task_frame->exit_runtime_frame, parent_task_frame->reenter_runtime_frame, parallel_data->value, requested_team_size, codeptr_ra, invoker);
 }
 
 static void
-on_ompt_event_loop_begin(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t parent_task_id,
-  void *workshare_function)
+on_ompt_callback_parallel_end(
+  ompt_data_t *parallel_data,
+  ompt_data_t *task_data,
+  ompt_invoker_t invoker,
+  const void *codeptr_ra)
 {
-  printf("%" PRIu64 ": ompt_event_loop_begin: parallel_id=%" PRIu64 ", parent_task_id=%" PRIu64 ", workshare_function=%p\n", ompt_get_thread_id(), parallel_id, parent_task_id, workshare_function);
+  printf("%" PRIu64 ": ompt_event_parallel_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", invoker=%d, codeptr_ra=%p\n", ompt_get_thread_data()->value, parallel_data->value, task_data->value, invoker, codeptr_ra);
+}
+
+static void
+on_ompt_callback_task_create(
+    ompt_data_t *parent_task_data,     /* id of parent task            */
+    const ompt_frame_t *parent_frame,  /* frame data for parent task   */
+    ompt_data_t* new_task_data,        /* id of created task           */
+    int type,
+    int has_dependences,
+    const void *codeptr_ra)               /* pointer to outlined function */
+{
+  if(new_task_data->ptr)
+    printf("%s\n", "0: new_task_data initially not null");
+  new_task_data->value = ompt_get_unique_id();
+  char buffer[2048];
+
+  format_task_type(type, buffer);
+
+  //there is no paralllel_begin callback for implicit parallel region
+  //thus it is initialized in initial task
+  if(type & ompt_task_initial)
+  {
+    ompt_data_t *parallel_data;
+    ompt_get_parallel_info(0, &parallel_data, NULL);
+    if(parallel_data->ptr)
+      printf("%s\n", "0: parallel_data initially not null");
+    parallel_data->value = ompt_get_unique_id();
+  }
+
+  printf("%" PRIu64 ": ompt_event_task_create: parent_task_id=%" PRIu64 ", parent_task_frame.exit=%p, parent_task_frame.reenter=%p, new_task_id=%" PRIu64 ", codeptr_ra=%p, task_type=%s=%d, has_dependences=%s\n", ompt_get_thread_data()->value, parent_task_data ? parent_task_data->value : 0, parent_frame ? parent_frame->exit_runtime_frame : NULL, parent_frame ? parent_frame->reenter_runtime_frame : NULL, new_task_data->value, codeptr_ra, buffer, type, has_dependences ? "yes" : "no");
 }
 
 static void
-on_ompt_event_loop_end(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id)
+on_ompt_callback_task_schedule(
+    ompt_data_t *first_task_data,
+    ompt_task_status_t prior_task_status,
+    ompt_data_t *second_task_data)
 {
-  printf("%" PRIu64 ": ompt_event_loop_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 "\n", ompt_get_thread_id(), parallel_id, task_id);
+  printf("%" PRIu64 ": ompt_event_task_schedule: first_task_id=%" PRIu64 ", second_task_id=%" PRIu64 ", prior_task_status=%s=%d\n", ompt_get_thread_data()->value, first_task_data->value, second_task_data->value, ompt_task_status_t_values[prior_task_status], prior_task_status);
+  if(prior_task_status == ompt_task_complete)
+  {
+    printf("%" PRIu64 ": ompt_event_task_end: task_id=%" PRIu64 "\n", ompt_get_thread_data()->value, first_task_data->value);
+  }
 }
 
 static void
-on_ompt_event_parallel_begin(
-  ompt_task_id_t parent_task_id,
-  ompt_frame_t *parent_task_frame,
-  ompt_parallel_id_t parallel_id,
-  uint32_t requested_team_size,
-  void *parallel_function,
-  ompt_invoker_t invoker)
+on_ompt_callback_task_dependences(
+  ompt_data_t *task_data,
+  const ompt_task_dependence_t *deps,
+  int ndeps)
 {
-  printf("%" PRIu64 ": ompt_event_parallel_begin: parent_task_id=%" PRIu64 ", parent_task_frame.exit=%p, parent_task_frame.reenter=%p, parallel_id=%" PRIu64 ", requested_team_size=%" PRIu32 ", parallel_function=%p, invoker=%d\n", ompt_get_thread_id(), parent_task_id, parent_task_frame->exit_runtime_frame, parent_task_frame->reenter_runtime_frame, parallel_id, requested_team_size, parallel_function, invoker);
+  printf("%" PRIu64 ": ompt_event_task_dependences: task_id=%" PRIu64 ", deps=%p, ndeps=%d\n", ompt_get_thread_data()->value, task_data->value, (void *)deps, ndeps);
 }
 
 static void
-on_ompt_event_parallel_end(
-  ompt_parallel_id_t parallel_id,
-  ompt_task_id_t task_id,
-  ompt_invoker_t invoker)
+on_ompt_callback_task_dependence(
+  ompt_data_t *first_task_data,
+  ompt_data_t *second_task_data)
 {
-  printf("%" PRIu64 ": ompt_event_parallel_end: parallel_id=%" PRIu64 ", task_id=%" PRIu64 ", invoker=%d\n", ompt_get_thread_id(), parallel_id, task_id, invoker);
+  printf("%" PRIu64 ": ompt_event_task_dependence_pair: first_task_id=%" PRIu64 ", second_task_id=%" PRIu64 "\n", ompt_get_thread_data()->value, first_task_data->value, second_task_data->value);
 }
 
+static void
+on_ompt_callback_thread_begin(
+  ompt_thread_type_t thread_type,
+  ompt_data_t *thread_data)
+{
+  if(thread_data->ptr)
+    printf("%s\n", "0: thread_data initially not null");
+  thread_data->value = ompt_get_unique_id();
+  printf("%" PRIu64 ": ompt_event_thread_begin: thread_type=%s=%d, thread_id=%" PRIu64 "\n", ompt_get_thread_data()->value, ompt_thread_type_t_values[thread_type], thread_type, thread_data->value);
+}
 
-void ompt_initialize(
-  ompt_function_lookup_t lookup,
-  const char *runtime_version,
-  unsigned int ompt_version)
+static void
+on_ompt_callback_thread_end(
+  ompt_data_t *thread_data)
 {
-  ompt_set_callback_t ompt_set_callback = (ompt_set_callback_t) lookup("ompt_set_callback");
-  ompt_get_task_id = (ompt_get_task_id_t) lookup("ompt_get_task_id");
-  ompt_get_task_frame = (ompt_get_task_frame_t) lookup("ompt_get_task_frame");
-  ompt_get_thread_id = (ompt_get_thread_id_t) lookup("ompt_get_thread_id");
-  ompt_get_parallel_id = (ompt_get_parallel_id_t) lookup("ompt_get_parallel_id");
-
-  ompt_set_callback(ompt_event_barrier_begin, (ompt_callback_t) &on_ompt_event_barrier_begin);
-  ompt_set_callback(ompt_event_barrier_end, (ompt_callback_t) &on_ompt_event_barrier_end);
-  ompt_set_callback(ompt_event_implicit_task_begin, (ompt_callback_t) &on_ompt_event_implicit_task_begin);
-  ompt_set_callback(ompt_event_implicit_task_end, (ompt_callback_t) &on_ompt_event_implicit_task_end);
-  ompt_set_callback(ompt_event_loop_begin, (ompt_callback_t) &on_ompt_event_loop_begin);
-  ompt_set_callback(ompt_event_loop_end, (ompt_callback_t) &on_ompt_event_loop_end);
-  ompt_set_callback(ompt_event_parallel_begin, (ompt_callback_t) &on_ompt_event_parallel_begin);
-  ompt_set_callback(ompt_event_parallel_end, (ompt_callback_t) &on_ompt_event_parallel_end);
-  printf("0: NULL_POINTER=%p\n", NULL);
+  printf("%" PRIu64 ": ompt_event_thread_end: thread_id=%" PRIu64 "\n", ompt_get_thread_data()->value, thread_data->value);
 }
 
-ompt_initialize_t ompt_tool()
+static int
+on_ompt_callback_control_tool(
+  uint64_t command,
+  uint64_t modifier,
+  void *arg,
+  const void *codeptr_ra)
+{
+  ompt_frame_t* omptTaskFrame;
+  ompt_get_task_info(0, NULL, (ompt_data_t**) NULL, &omptTaskFrame, NULL, NULL);
+  printf("%" PRIu64 ": ompt_event_control_tool: command=%" PRIu64 ", modifier=%" PRIu64 ", arg=%p, codeptr_ra=%p, current_task_frame.exit=%p, current_task_frame.reenter=%p \n", ompt_get_thread_data()->value, command, modifier, arg, codeptr_ra, omptTaskFrame->exit_runtime_frame, omptTaskFrame->reenter_runtime_frame);
+  return 0; //success
+}
+
+#define register_callback_t(name, type)                       \
+do{                                                           \
+  type f_##name = &on_##name;                                 \
+  if (ompt_set_callback(name, (ompt_callback_t)f_##name) ==   \
+      ompt_set_never)                                         \
+    printf("0: Could not register callback '" #name "'\n");   \
+}while(0)
+
+#define register_callback(name) register_callback_t(name, name##_t)
+
+int ompt_initialize(
+  ompt_function_lookup_t lookup,
+  ompt_fns_t* fns)
+{
+  ompt_set_callback = (ompt_set_callback_t) lookup("ompt_set_callback");
+  ompt_get_task_info = (ompt_get_task_info_t) lookup("ompt_get_task_info");
+  ompt_get_thread_data = (ompt_get_thread_data_t) lookup("ompt_get_thread_data");
+  ompt_get_parallel_info = (ompt_get_parallel_info_t) lookup("ompt_get_parallel_info");
+  ompt_get_unique_id = (ompt_get_unique_id_t) lookup("ompt_get_unique_id");
+
+  ompt_get_num_places = (ompt_get_num_places_t) lookup("ompt_get_num_places");
+  ompt_get_place_proc_ids = (ompt_get_place_proc_ids_t) lookup("ompt_get_place_proc_ids");
+  ompt_get_place_num = (ompt_get_place_num_t) lookup("ompt_get_place_num");
+  ompt_get_partition_place_nums = (ompt_get_partition_place_nums_t) lookup("ompt_get_partition_place_nums");
+  ompt_get_proc_id = (ompt_get_proc_id_t) lookup("ompt_get_proc_id");
+  ompt_enumerate_states = (ompt_enumerate_states_t) lookup("ompt_enumerate_states");
+  ompt_enumerate_mutex_impls = (ompt_enumerate_mutex_impls_t) lookup("ompt_enumerate_mutex_impls");
+
+  register_callback(ompt_callback_mutex_acquire);
+  register_callback_t(ompt_callback_mutex_acquired, ompt_callback_mutex_t);
+  register_callback_t(ompt_callback_mutex_released, ompt_callback_mutex_t);
+  register_callback(ompt_callback_nest_lock);
+  register_callback(ompt_callback_sync_region);
+  register_callback_t(ompt_callback_sync_region_wait, ompt_callback_sync_region_t);
+  register_callback(ompt_callback_control_tool);
+  register_callback(ompt_callback_flush);
+  register_callback(ompt_callback_cancel);
+  register_callback(ompt_callback_idle);
+  register_callback(ompt_callback_implicit_task);
+  register_callback_t(ompt_callback_lock_init, ompt_callback_mutex_acquire_t);
+  register_callback_t(ompt_callback_lock_destroy, ompt_callback_mutex_t);
+  register_callback(ompt_callback_work);
+  register_callback(ompt_callback_master);
+  register_callback(ompt_callback_parallel_begin);
+  register_callback(ompt_callback_parallel_end);
+  register_callback(ompt_callback_task_create);
+  register_callback(ompt_callback_task_schedule);
+  register_callback(ompt_callback_task_dependences);
+  register_callback(ompt_callback_task_dependence);
+  register_callback(ompt_callback_thread_begin);
+  register_callback(ompt_callback_thread_end);
+  printf("0: NULL_POINTER=%p\n", (void*)NULL);
+  return 1; //success
+}
+
+void ompt_finalize(ompt_fns_t* fns)
+{
+  printf("0: ompt_event_runtime_shutdown\n");
+}
+
+ompt_fns_t* ompt_start_tool(
+  unsigned int omp_version,
+  const char *runtime_version)
 {
-  return &ompt_initialize;
+  static ompt_fns_t ompt_fns = {&ompt_initialize,&ompt_finalize};
+  return &ompt_fns;
 }

runtime/test/ompt/cancel/cancel_parallel.c

+// RUN: %libomp-compile && env OMP_CANCELLATION=true %libomp-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// Current GOMP interface implementation does not support cancellation
+// XFAIL: gcc
+
+#include "callback.h"
+#include "omp.h"
+
+int main()
+{
+  #pragma omp parallel num_threads(2)
+  {
+    if(omp_get_thread_num() == 0)
+    {
+      printf("%" PRIu64 ": fuzzy_address=0x%lx or 0x%lx\n", ompt_get_thread_data()->value, ((uint64_t)(char*)(&& ompt_label_1))/256-1, ((uint64_t)(char*)(&& ompt_label_1))/256);
+      #pragma omp cancel parallel
+      print_fuzzy_address(1); //does not actually print the address but provides a label
+    }
+    else
+    {
+      usleep(100);
+      printf("%" PRIu64 ": fuzzy_address=0x%lx or 0x%lx\n", ompt_get_thread_data()->value, ((uint64_t)(char*)(&& ompt_label_2))/256-1, ((uint64_t)(char*)(&& ompt_label_2))/256);
+      #pragma omp cancellation point parallel
+      print_fuzzy_address(2); //does not actually print the address but provides a label
+    }
+  }
+
+
+  // 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_cancel'
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[NULL]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[NULL]], task_type=ompt_task_initial=1, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_parallel|ompt_cancel_activated=17, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: fuzzy_address={{.*}}[[OTHER_RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_parallel|ompt_cancel_detected=33, codeptr_ra=[[OTHER_RETURN_ADDRESS]]{{[0-f][0-f]}}
+
+  return 0;
+}

runtime/test/ompt/cancel/cancel_taskgroup.c

+// RUN:  %libomp-compile && env OMP_CANCELLATION=true %libomp-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// Current GOMP interface implementation does not support cancellation
+// XFAIL: gcc
+
+#include "callback.h"
+#include <unistd.h>  
+#include <stdio.h>
+
+int main()
+{
+  int condition=0;
+  #pragma omp parallel num_threads(2)
+  {}
+
+  print_frame(0);
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+      #pragma omp taskgroup
+      {
+        #pragma omp task shared(condition)
+        {
+          printf("start execute task 1\n");
+          OMPT_SIGNAL(condition);
+          OMPT_WAIT(condition,2);
+          #pragma omp cancellation point taskgroup
+          printf("end execute task 1\n");
+        }
+        #pragma omp task shared(condition)
+        {
+          printf("start execute task 2\n");
+          OMPT_SIGNAL(condition);
+          OMPT_WAIT(condition,2);
+          #pragma omp cancellation point taskgroup
+          printf("end execute task 2\n");
+        }
+      #pragma omp task shared(condition)
+        {
+          printf("start execute task 3\n");
+          OMPT_SIGNAL(condition);
+          OMPT_WAIT(condition,2);
+          #pragma omp cancellation point taskgroup
+          printf("end execute task 3\n");
+        }
+      #pragma omp task if(0) shared(condition)
+        {
+          printf("start execute task 4\n");
+          OMPT_WAIT(condition,1);
+          #pragma omp cancel taskgroup
+          printf("end execute task 4\n");
+        }
+        OMPT_SIGNAL(condition);
+      }
+    }
+    #pragma omp barrier
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_master'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_cancel'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_master_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]*}}
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID]], parent_task_frame.exit={{0x[0-f]*}}, parent_task_frame.reenter={{0x[0-f]*}}, new_task_id=[[FIRST_TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]*}}, task_type=ompt_task_explicit=4, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID]], parent_task_frame.exit={{0x[0-f]*}}, parent_task_frame.reenter={{0x[0-f]*}}, new_task_id=[[SECOND_TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]*}}, task_type=ompt_task_explicit=4, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID]], parent_task_frame.exit={{0x[0-f]*}}, parent_task_frame.reenter={{0x[0-f]*}}, new_task_id=[[THIRD_TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]*}}, task_type=ompt_task_explicit=4, has_dependences=no
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID]], parent_task_frame.exit={{0x[0-f]*}}, parent_task_frame.reenter={{0x[0-f]*}}, new_task_id=[[CANCEL_TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]*}}, task_type=ompt_task_explicit|ompt_task_undeferred=134217732, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[PARENT_TASK_ID]], second_task_id=[[CANCEL_TASK_ID]], prior_task_status=ompt_task_others=4
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_cancel: task_data=[[CANCEL_TASK_ID]], flags=ompt_cancel_taskgroup|ompt_cancel_activated=24, codeptr_ra={{0x[0-f]*}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[CANCEL_TASK_ID]], second_task_id=[[PARENT_TASK_ID]], prior_task_status=ompt_task_cancel=3
+
+  // CHECK-DAG: {{^}}{{[0-9]+}}: ompt_event_cancel: task_data={{[0-9]+}}, flags=ompt_cancel_taskgroup|ompt_cancel_discarded_task=72, codeptr_ra=[[NULL]]
+  // CHECK-DAG: {{^}}{{[0-9]+}}: ompt_event_cancel: task_data={{[0-9]+}}, flags=ompt_cancel_taskgroup|ompt_cancel_discarded_task=72, codeptr_ra=[[NULL]]
+  
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_cancel: task_data={{[0-9]+}}, flags=ompt_cancel_taskgroup|ompt_cancel_detected=40, codeptr_ra={{0x[0-f]*}}
+
+  return 0;
+}

runtime/test/ompt/cancel/cancel_worksharing.c

+// RUN: %libomp-compile && env OMP_CANCELLATION=true %libomp-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// Current GOMP interface implementation does not support cancellation
+// XFAIL: gcc
+
+
+#include "callback.h"
+#include <unistd.h>
+
+int main()
+{
+  int condition=0;
+  #pragma omp parallel num_threads(2)
+  {
+    int x = 0;
+    int i;
+    #pragma omp for
+    for(i = 0; i < 2; i++)
+    {
+      if(i == 0)
+      {
+        x++;
+        OMPT_SIGNAL(condition);
+        #pragma omp cancel for
+      }
+      else
+      {
+        x++;
+        OMPT_WAIT(condition,1);
+        usleep(10000);
+        #pragma omp cancellation point for
+      }
+    }
+  }
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp sections
+    {
+      #pragma omp section
+      {
+        OMPT_SIGNAL(condition);
+        #pragma omp cancel sections
+      }
+      #pragma omp section
+      {
+        OMPT_WAIT(condition,2);
+        usleep(10000);
+        #pragma omp cancellation point sections
+      }
+    }
+  }
+
+
+  // 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_cancel'
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_task_create: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[NULL]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[NULL]], task_type=ompt_task_initial=1, has_dependences=no
+  
+  // cancel for and sections
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_do|ompt_cancel_activated=20, codeptr_ra={{0x[0-f]*}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_sections|ompt_cancel_activated=18, codeptr_ra={{0x[0-f]*}}
+  // CHECK: {{^}}[[OTHER_THREAD_ID:[0-9]+]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_do|ompt_cancel_detected=36, codeptr_ra={{0x[0-f]*}}
+  // CHECK: {{^}}[[OTHER_THREAD_ID:[0-9]+]]: ompt_event_cancel: task_data=[[TASK_ID:[0-9]+]], flags=ompt_cancel_sections|ompt_cancel_detected=34, codeptr_ra={{0x[0-f]*}}
+
+  return 0;
+}

runtime/test/ompt/loadtool/tool_available.c

+// RUN: %libomp-compile -DCODE && %libomp-compile -DTOOL -o%T/tool.so -shared -fPIC && env OMP_TOOL_LIBRARIES=%T/tool.so %libomp-run | FileCheck %s
+// REQUIRES: ompt
+
+/*
+ *  This file contains code for an OMPT shared library tool to be 
+ *  loaded and the code for the OpenMP executable. 
+ *  -DTOOL enables the code for the tool during compilation
+ *  -DCODE enables the code for the executable during compilation
+ *  The RUN line compiles the two binaries and then tries to load
+ *  the tool using the OMP_TOOL_LIBRARIES environmental variable.
+ */
+
+#ifdef CODE
+#include "omp.h"
+
+int main()
+{
+  #pragma omp parallel num_threads(2)
+  {
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 
+  
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}0: ompt_event_runtime_shutdown
+
+  return 0;
+}
+
+#endif /* CODE */
+
+#ifdef TOOL
+
+#include <stdio.h>
+#include <ompt.h>
+
+int ompt_initialize(
+  ompt_function_lookup_t lookup,
+  ompt_fns_t* fns)
+{
+  printf("0: NULL_POINTER=%p\n", (void*)NULL);
+  return 1; //success
+}
+
+void ompt_finalize(ompt_fns_t* fns)
+{
+  printf("%d: ompt_event_runtime_shutdown\n", omp_get_thread_num());
+}
+
+ompt_fns_t* ompt_start_tool(
+  unsigned int omp_version,
+  const char *runtime_version)
+{
+  static ompt_fns_t ompt_fns = {&ompt_initialize,&ompt_finalize};
+  return &ompt_fns;
+}
+#endif /* TOOL */

runtime/test/ompt/misc/api_calls.c

+// RUN: %libomp-compile && env OMP_PLACES=cores %libomp-run | FileCheck %s
+// REQUIRES: ompt, linux
+#include "callback.h"
+#include <omp.h>
+#define __USE_GNU
+#include <sched.h>
+#undef __USE_GNU
+
+void print_list(char* function_name, int list[])
+{
+  printf("%" PRIu64 ": %s(0)=(%d", ompt_get_thread_data()->value, function_name, list[0]);
+  int i;
+  for(i = 1; i < omp_get_place_num_procs(0); i++)
+  {
+    printf(",%d", list[i]);
+  }
+  printf(")\n");
+}
+
+int main()
+{
+  #pragma omp parallel num_threads(1)
+  {
+    printf("%" PRIu64 ": omp_get_num_places()=%d\n", ompt_get_thread_data()->value, omp_get_num_places());
+    printf("%" PRIu64 ": ompt_get_num_places()=%d\n", ompt_get_thread_data()->value, ompt_get_num_places());
+
+    int omp_ids[omp_get_place_num_procs(0)];
+    omp_get_place_proc_ids(0, omp_ids);
+    print_list("omp_get_place_proc_ids" ,omp_ids);
+    int ompt_ids[omp_get_place_num_procs(0)];
+    ompt_get_place_proc_ids(0, omp_get_place_num_procs(0), ompt_ids);
+    print_list("ompt_get_place_proc_ids", ompt_ids);
+
+    printf("%" PRIu64 ": omp_get_place_num()=%d\n", ompt_get_thread_data()->value, omp_get_place_num());
+    printf("%" PRIu64 ": ompt_get_place_num()=%d\n", ompt_get_thread_data()->value, ompt_get_place_num());
+
+    int omp_nums[omp_get_partition_num_places()];
+    omp_get_partition_place_nums(omp_nums);
+    print_list("omp_get_partition_place_nums" ,omp_nums);
+    int ompt_nums[omp_get_partition_num_places()];
+    ompt_get_partition_place_nums(omp_get_partition_num_places(), ompt_nums);
+    print_list("ompt_get_partition_place_nums", ompt_nums);
+
+    printf("%" PRIu64 ": sched_getcpu()=%d\n", ompt_get_thread_data()->value, sched_getcpu());
+    printf("%" PRIu64 ": ompt_get_proc_id()=%d\n", ompt_get_thread_data()->value, ompt_get_proc_id());
+  }
+
+  // Check if libomp supports the callbacks for this test.
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: omp_get_num_places()=[[NUM_PLACES:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_get_num_places()=[[NUM_PLACES]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: omp_get_place_proc_ids(0)=([[PROC_IDS:[0-9\,]+]])
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_get_place_proc_ids(0)=([[PROC_IDS]])
+
+  // CHECK: {{^}}[[MASTER_ID]]: omp_get_place_num()=[[PLACE_NUM:[-]?[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_get_place_num()=[[PLACE_NUM]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: sched_getcpu()=[[CPU_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_get_proc_id()=[[CPU_ID]]
+
+
+  return 0;
+}

runtime/test/ompt/misc/control_tool.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  #pragma omp parallel num_threads(1)
+  {
+    print_frame(1);
+    print_frame(0);
+    omp_control_tool(omp_control_tool_flush, 1, NULL);
+    print_current_address(0);
+  }
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_control_tool'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: __builtin_frame_address(1)=[[EXIT_FRAME:0x[0-f]*]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER_FRAME:0x[0-f]*]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_control_tool: command=3, modifier=1, arg=[[NULL]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]*]], current_task_frame.exit=[[EXIT_FRAME]], current_task_frame.reenter=[[REENTER_FRAME]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/misc/control_tool_no_ompt_support.c

+// RUN: %libomp-compile-and-run
+#include <omp.h>
+
+int main()
+{
+  #pragma omp parallel num_threads(1)
+  {
+    omp_control_tool(omp_control_tool_flush, 1, NULL);
+  }
+
+  return 0;
+}

runtime/test/ompt/misc/idle.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+  #pragma omp parallel num_threads(3)
+  {
+    #pragma omp atomic
+    x++;
+  }
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp atomic
+    x++;
+  }
+
+
+  printf("x=%d\n", x);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_idle'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_idle_begin:
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_idle_end:
+
+  return 0;
+}

runtime/test/ompt/ompt-signal.h

+// These functions are used to provide a signal-wait mechanism to enforce expected scheduling for the test cases.
+// Conditional variable (s) needs to be shared! Initialize to 0
+#include <unistd.h>
+
+#define OMPT_SIGNAL(s) ompt_signal(&s)
+//inline 
+void ompt_signal(int* s) 
+{                
+  #pragma omp atomic
+  (*s)++;
+}
+                
+#define OMPT_WAIT(s,v) ompt_wait(&s,v)
+// wait for s >= v
+//inline 
+void ompt_wait(int *s, int v)
+{
+  int wait=0;
+  do{
+    usleep(10);
+    #pragma omp atomic read
+    wait = (*s);
+  }while(wait<v);
+}

runtime/test/ompt/parallel/dynamic_enough_threads.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+
+int main()
+{
+  omp_set_dynamic(1);
+
+  #pragma omp parallel num_threads(4)
+  {
+    print_ids(0);
+    print_ids(1);
+  }
+  print_fuzzy_address(1);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+
+  //team-size of 1-4 is expected
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]], team_size={{[1-4]}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/parallel/dynamic_not_enough_threads.c

+// RUN: %libomp-compile && env OMP_THREAD_LIMIT=2 %libomp-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+
+int main()
+{
+  omp_set_dynamic(1);
+
+  #pragma omp parallel num_threads(4)
+  {
+    print_ids(0);
+    print_ids(1);
+  }
+  print_fuzzy_address(1);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+
+  //team-size of 1-4 is expected
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]], team_size={{[1-4]}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/parallel/max_active_levels_serialized.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=THREADS %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  omp_set_nested(1);
+  omp_set_max_active_levels(1);
+
+  #pragma omp parallel num_threads(2)
+  {
+    print_ids(0);
+    print_ids(1);
+    #pragma omp parallel num_threads(2)
+    {
+      print_ids(0);
+      print_ids(1);
+      print_ids(2);
+    }
+    print_fuzzy_address(1);
+  }
+  print_fuzzy_address(2);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+  // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+  
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=[[NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  return 0;
+}

runtime/test/ompt/parallel/nested.c

 // RUN: %libomp-compile-and-run | FileCheck %s
 // RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=THREADS %s
 // REQUIRES: ompt
 #include "callback.h"
 #include <omp.h>
+#include <unistd.h>
 
 int main()
 {
+  int condition=0;
   omp_set_nested(1);
   print_frame(0);
 
   #pragma omp parallel num_threads(4)
   {
     print_frame(1);
     print_ids(0);
     print_ids(1);
     print_frame(0);
+
+    //get all implicit task events before starting nested:
+    #pragma omp barrier
+    
     #pragma omp parallel num_threads(4)
     {
       print_frame(1);
       print_ids(0);
       print_ids(1);
       print_ids(2);
       print_frame(0);
+      OMPT_SIGNAL(condition);
+      OMPT_WAIT(condition,16);
       #pragma omp barrier
+      print_fuzzy_address(1);
       print_ids(0);
     }
+    print_fuzzy_address(2);
     print_ids(0);
   }
+  print_fuzzy_address(3);
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
 
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
 
 
   // THREADS: {{^}}0: NULL_POINTER=[[NULL:.*$]]
   // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: __builtin_frame_address(0)=[[MAIN_REENTER:0x[0-f]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // nested parallel masters
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // THREADS: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
   // THREADS: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION:0x[0-f]+]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
   // THREADS: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[NESTED_EXIT:0x[0-f]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
   // THREADS: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[NESTED_REENTER:0x[0-f]+]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
   // explicit barrier
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NESTED_REENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], codeptr_ra=[[BARRIER_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NESTED_REENTER]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[BARRIER_RETURN_ADDRESS]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_EXIT]], reenter_frame=[[NULL]]
   // implicit barrier
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]], codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]], codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
   // implicit barrier
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // nested parallel worker threads
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   return 0;
 }

runtime/test/ompt/parallel/nested_lwt.c

 // RUN: %libomp-compile-and-run | FileCheck %s
 // RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=THREADS %s
 // REQUIRES: ompt
 #include "callback.h"
 #include <omp.h>
+#include <unistd.h>
 
 int main()
 {
   omp_set_nested(1);
+  int condition;
 
   #pragma omp parallel num_threads(4)
   {
     print_ids(0);
     print_ids(1);
+    //get all implicit task events before starting nested:
+    #pragma omp barrier
     #pragma omp parallel num_threads(1)
     {
       print_ids(0);
       print_ids(1);
       print_ids(2);
+      //get all implicit task events before starting nested:
+      #pragma omp barrier
       #pragma omp parallel num_threads(4)
       {
         print_ids(0);
         print_ids(1);
         print_ids(2);
         print_ids(3);
+      OMPT_SIGNAL(condition);
+      OMPT_WAIT(condition,16);
       }
+      print_fuzzy_address(1);
     }
+    print_fuzzy_address(2);
   }
+  print_fuzzy_address(3);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
 
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
 
 
   // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
-  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // nested parallel masters
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
 
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=[[NESTED_PARALLEL_FUNCTION:0x[0-f]+]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
 
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_NESTED_PARALLEL_FUNCTION:0x[0-f]+]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit=[[NESTED_NESTED_TASK_FRAME_EXIT]], parent_task_frame.reenter=[[NESTED_NESTED_TASK_FRAME_ENTER:0x[0-f]+]], parallel_id=[[NESTED_NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_NESTED_TASK_FRAME_EXIT:0x[0-f]+]], reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame=[[NESTED_NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NESTED_TASK_FRAME_EXIT]], reenter_frame=[[NESTED_TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // nested parallel worker threads
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
   // can't reliably tell which parallel region is the parent...
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
-  // THREADS: {{^}}[[THREAD_ID]]: level 3: parallel_id=0, task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}
+  // THREADS: {{^}}[[THREAD_ID]]: task level 3: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[TASK_FRAME_ENTER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
 
   return 0;
 }

runtime/test/ompt/parallel/nested_serialized.c

#pragma omp parallel num_threads(4)
     print_ids(0);
     print_ids(1);
     #pragma omp parallel num_threads(4)
     {
       print_ids(0);
       print_ids(1);
       print_ids(2);
     }
+    print_fuzzy_address(1);
   }
+  print_fuzzy_address(2);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
 
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
 
 
   // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
-  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION:0x[0-f]+]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=[[NESTED_PARALLEL_FUNCTION]], invoker=[[PARALLEL_INVOKER]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[NESTED_RETURN_ADDRESS]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 2: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[NESTED_IMPLICIT_TASK_ID]]
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   return 0;
 }

runtime/test/ompt/parallel/no_thread_num_clause.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=THREADS %s
+// REQUIRES: ompt
+#include "callback.h"
+
+int main()
+{
+  omp_set_num_threads(4);
+  #pragma omp parallel
+  {
+    print_ids(0);
+    print_ids(1);
+  }
+  print_fuzzy_address(1);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+
+
+  // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_initial=1, thread_id=[[MASTER_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=0, parent_task_frame.exit=(nil), parent_task_frame.reenter=(nil), new_task_id=281474976710658, codeptr_ra=(nil), task_type=ompt_task_initial=1, has_dependences=no
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker={{[0-9]+}}
+
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  return 0;
+}

runtime/test/ompt/parallel/normal.c

 // RUN: %libomp-compile-and-run | FileCheck %s
 // RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=THREADS %s
 // REQUIRES: ompt
 #include "callback.h"
 
 int main()
 {
   #pragma omp parallel num_threads(4)
   {
     print_ids(0);
     print_ids(1);
   }
+  print_fuzzy_address(1);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+
 
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+  
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
   // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
 
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
 
 
   // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
-  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker={{.*}}
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_initial=1, thread_id=[[MASTER_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker={{.*}}
 
   // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-
-  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
-  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
-  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
   // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
   // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   return 0;
 }

runtime/test/ompt/parallel/not_enough_threads.c

+// RUN: %libomp-compile && env OMP_THREAD_LIMIT=2 %libomp-run | FileCheck %s
+// RUN: %libomp-compile && env OMP_THREAD_LIMIT=2 %libomp-run | %sort-threads | FileCheck --check-prefix=THREADS %s
+// REQUIRES: ompt
+#include "callback.h"
+
+int main()
+{
+  #pragma omp parallel num_threads(4)
+  {
+    print_ids(0);
+    print_ids(1);
+  }
+  print_fuzzy_address(1);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // Note that we cannot ensure that the worker threads have already called barrier_end and implicit_task_end before parallel_end!
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK-DAG: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK-DAG: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+
+
+  // THREADS: 0: NULL_POINTER=[[NULL:.*$]]
+  // THREADS: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_initial=1, thread_id=[[MASTER_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker={{[0-9]+}}
+
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // THREADS: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_thread_begin: thread_type=ompt_thread_worker=2, thread_id=[[THREAD_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]]
+  // THREADS-NOT: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // THREADS: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  return 0;
+}

runtime/test/ompt/parallel/parallel_if0.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+
+int main()
+{
+//  print_frame(0);
+  #pragma omp parallel if(0)
+  {
+//    print_frame(1);
+    print_ids(0);
+    print_ids(1);
+//    print_frame(0);
+    #pragma omp parallel if(0)
+    {
+//      print_frame(1);
+      print_ids(0);
+      print_ids(1);
+      print_ids(2);
+//      print_frame(0);
+      #pragma omp task
+      {
+//        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+        print_ids(3);
+      }
+    }
+    print_fuzzy_address(1);
+  }
+  print_fuzzy_address(2);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_event_implicit_task_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_event_implicit_task_end'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[NESTED_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame={{0x[0-f]+}}
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id=[[EXPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[NESTED_IMPLICIT_TASK_ID]], second_task_id=[[EXPLICIT_TASK_ID]], prior_task_status=ompt_task_others=4
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[EXPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[EXPLICIT_TASK_ID]], second_task_id=[[NESTED_IMPLICIT_TASK_ID]], prior_task_status=ompt_task_complete=1
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_end: task_id=[[EXPLICIT_TASK_ID]]
+
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=0, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[NESTED_RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=0, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/parallel/serialized.c

 // RUN: %libomp-compile-and-run | FileCheck %s
 // REQUIRES: ompt
 #include "callback.h"
 
 int main()
 {
+//  print_frame(0);
   #pragma omp parallel num_threads(1)
   {
+//    print_frame(1);
     print_ids(0);
     print_ids(1);
+//    print_frame(0);
+    #pragma omp parallel num_threads(1)
+    {
+//      print_frame(1);
+      print_ids(0);
+      print_ids(1);
+      print_ids(2);
+//      print_frame(0);
+      #pragma omp task
+      {
+//        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+        print_ids(3);
+      }
     }
+    print_fuzzy_address(1);
+  }
+  print_fuzzy_address(2);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_event_implicit_task_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_event_implicit_task_end'
 
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:.+]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK-NOT: 0: task_data initially not null
+  // CHECK-NOT: 0: thread_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[OUTER_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
 
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[MASTER_ID]]: level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[MASTER_ID]]: level 1: parallel_id=0, task_id=[[PARENT_TASK_ID]]
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[NESTED_PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=[[INNER_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame={{0x[0-f]+}}
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[NESTED_IMPLICIT_TASK_ID]], parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id=[[EXPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[NESTED_IMPLICIT_TASK_ID]], second_task_id=[[EXPLICIT_TASK_ID]], prior_task_status=ompt_task_others=4
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[EXPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[NESTED_IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame={{0x[0-f]+}}, reenter_frame={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[EXPLICIT_TASK_ID]], second_task_id=[[NESTED_IMPLICIT_TASK_ID]], prior_task_status=ompt_task_complete=1
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_end: task_id=[[EXPLICIT_TASK_ID]]
+
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=0, task_id=[[NESTED_IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[NESTED_PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], invoker=[[PARALLEL_INVOKER]], codeptr_ra=[[INNER_RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[INNER_RETURN_ADDRESS]]
 
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=0, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], invoker=[[PARALLEL_INVOKER]], codeptr_ra=[[OUTER_RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[OUTER_RETURN_ADDRESS]]
 
   return 0;
 }

runtime/test/ompt/synchronization/barrier/explicit.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp atomic
+    x++;
+
+    #pragma omp barrier
+    print_current_address();
+
+    #pragma omp atomic
+    x++;
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread explicit barrier 
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  // master thread implicit barrier at parallel end
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+
+
+  // worker thread explicit barrier 
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[THREAD_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  // worker thread implicit barrier at parallel end
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/barrier/for_loop.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int y[] = {0,1,2,3};
+
+  #pragma omp parallel num_threads(2)
+  {
+    //implicit barrier at end of for loop
+    int i;
+    #pragma omp for
+    for (i = 0; i < 4; i++)
+    {
+      y[i]++;
+    }
+    print_current_address();
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread implicit barrier at loop end 
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+  // master thread implicit barrier at parallel end
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+
+  // worker thread explicit barrier 
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+  // worker thread implicit barrier after parallel
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/barrier/for_simd.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int y[] = {0,1,2,3};
+
+  int i;
+  #pragma omp for simd
+  for (i = 0; i < 4; i++)
+  {
+    y[i]++;
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread implicit barrier at simd loop end 
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+  return 0;
+}

runtime/test/ompt/synchronization/barrier/parallel_region.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+
+  //implicit barrier at end of a parallel region
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp atomic
+    x++;
+  }
+  print_fuzzy_address();
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread implicit barrier at parallel end
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+
+  // worker thread implicit barrier at parallel end
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/barrier/sections.c

+// RUN: %libomp-compile-and-run | %sort-threads  | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+
+  #pragma omp parallel num_threads(2)
+  {
+    //implicit barrier after sections with nowait but with lastprivates
+    //implicit barrier at end of sections
+    #pragma omp sections
+    {
+      #pragma omp section 
+      {
+        #pragma omp atomic
+        x++;
+      }
+      
+      #pragma omp section 
+      {
+        #pragma omp atomic
+        x++;
+      }
+    }
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread implicit barrier at sections end
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+  // master thread implicit barrier at parallel end
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+
+  // worker thread implicit barrier at sections end
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+  // worker thread implicit barrier at parallel end
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/barrier/single.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+
+  #pragma omp parallel num_threads(2)
+  {
+    //implicit barrier at end of single
+    #pragma omp single
+    {
+      x++;
+    }
+    print_fuzzy_address();
+    //critical section to avoid merge of two barriers into one
+    #pragma omp critical
+    {
+      x++;
+    }
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // master thread implicit barrier at single end
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // master thread implicit barrier at parallel end
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+
+
+  // worker thread implicit barrier at single end
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // worker thread implicit barrier at parallel end
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_wait_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[NULL]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/critical.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  #pragma omp critical
+  {
+    print_current_address(1);
+    print_ids(0);
+  }
+  print_current_address(2);
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_wait_critical: wait_id=[[WAIT_ID:[0-9]+]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_critical: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_critical: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/flush.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// GCC generates code that does not call the runtime for the flush construct
+// XFAIL: gcc
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  #pragma omp parallel num_threads(2)
+  {
+    int tid = omp_get_thread_num();
+    
+    #pragma omp flush
+    print_current_address(1);
+  }
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_flush'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_flush: codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: current_address=[[RETURN_ADDRESS]]
+  //
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_flush: codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: current_address=[[RETURN_ADDRESS]]
+
+
+
+  return 0;
+}

runtime/test/ompt/synchronization/lock.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  //need to use an OpenMP construct so that OMPT will be initalized
+  #pragma omp parallel num_threads(1)
+    print_ids(0);
+
+  omp_lock_t lock;
+  printf("%" PRIu64 ": &lock: %lli\n", ompt_get_thread_data()->value, (long long) &lock);
+  omp_init_lock(&lock);
+  print_current_address(1);
+  omp_set_lock(&lock);
+  print_current_address(2);
+  omp_unset_lock(&lock);
+  print_current_address(3);
+  omp_destroy_lock(&lock);
+  print_current_address(4);
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: &lock: [[WAIT_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_init_lock: wait_id=[[WAIT_ID]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_lock: wait_id=[[WAIT_ID]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_destroy_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+ 
+  return 0;
+}

runtime/test/ompt/synchronization/master.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+// GCC generates code that does not call the runtime for the master construct
+// XFAIL: gcc
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+      print_fuzzy_address(1);
+      x++;
+    }
+    print_current_address(2);
+  }
+
+  printf("%" PRIu64 ": x=%d\n", ompt_get_thread_data()->value, x);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_master'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_master_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_master_end: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra=[[RETURN_ADDRESS_END:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: current_address=[[RETURN_ADDRESS_END]]
+
+
+  return 0;
+}

runtime/test/ompt/synchronization/nest_lock.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  //need to use an OpenMP construct so that OMPT will be initalized
+  #pragma omp parallel num_threads(1)
+    print_ids(0);
+
+  omp_nest_lock_t nest_lock;
+  printf("%" PRIu64 ": &nest_lock: %lli\n", ompt_get_thread_data()->value, (long long) &nest_lock);
+  omp_init_nest_lock(&nest_lock);
+  print_current_address(1);
+  omp_set_nest_lock(&nest_lock);
+  print_current_address(2);
+  omp_set_nest_lock(&nest_lock);
+  print_current_address(3);
+  omp_unset_nest_lock(&nest_lock);
+  print_current_address(4);
+  omp_unset_nest_lock(&nest_lock);
+  print_current_address(5);
+  omp_destroy_nest_lock(&nest_lock);
+  print_current_address(6);
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_init_nest_lock: wait_id=[[WAIT_ID:[0-9]+]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_first: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_next: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_prev: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_last: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_destroy_nest_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/ordered.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  #pragma omp ordered
+  {
+    print_current_address(1);
+    print_ids(0);
+  }
+  print_current_address(2);
+  
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_wait_ordered: wait_id=[[WAIT_ID:[0-9]+]], hint={{[0-9]+}}, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_ordered: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_ordered: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  
+  return 0;
+}

runtime/test/ompt/synchronization/taskgroup.c

+// RUN:  %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <unistd.h>  
+#include <stdio.h>
+
+int main()
+{
+  int condition=0;
+  int x=0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+      #pragma omp taskgroup
+      {
+        print_current_address(1);
+        #pragma omp task
+        {
+          #pragma omp atomic
+          x++;
+        }
+      }
+      print_current_address(2);
+    }
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_master'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_cancel'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_thread_begin'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_taskgroup_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_taskgroup_begin: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_taskgroup_end: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_taskgroup_end: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/taskwait.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+      #pragma omp task
+      {
+        x++;
+      }
+      #pragma omp taskwait
+      print_current_address(1);
+    }
+  }
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_sync_region_wait'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_taskwait_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_taskwait_begin: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_taskwait_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: ompt_event_taskwait_end: parallel_id={{[0-9]+}}, task_id={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS]]
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/test_lock.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  omp_lock_t lock;
+  omp_init_lock(&lock);
+  print_current_address(1);
+
+  omp_test_lock(&lock);
+  print_current_address(2);
+  omp_unset_lock(&lock);
+  print_current_address(3);
+
+  omp_set_lock(&lock);
+  print_current_address(4);
+  omp_test_lock(&lock);
+  print_current_address(5);
+  omp_unset_lock(&lock);
+  print_current_address(6);
+
+  omp_destroy_lock(&lock);
+  print_current_address(7);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_init_lock: wait_id=[[WAIT_ID:[0-9]+]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_destroy_lock: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]  
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/synchronization/test_nest_lock.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  omp_nest_lock_t nest_lock;
+  omp_init_nest_lock(&nest_lock);
+
+  omp_test_nest_lock(&nest_lock);
+  omp_unset_nest_lock(&nest_lock);
+
+  omp_set_nest_lock(&nest_lock);
+  omp_test_nest_lock(&nest_lock);
+  omp_unset_nest_lock(&nest_lock);
+  omp_unset_nest_lock(&nest_lock);
+
+  omp_destroy_nest_lock(&nest_lock);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_init_nest_lock: wait_id=[[WAIT_ID:[0-9]+]], hint=0, impl={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_first: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_last: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_first: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_next: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_prev: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}  
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_last: wait_id=[[WAIT_ID]], codeptr_ra={{0x[0-f]+}}  
+
+  return 0;
+}

runtime/test/ompt/synchronization/test_nest_lock_parallel.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  omp_nest_lock_t nest_lock;
+  omp_init_nest_lock(&nest_lock);
+
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+      omp_set_nest_lock(&nest_lock);
+      print_current_address(1);
+    }
+    #pragma omp barrier
+    omp_test_nest_lock(&nest_lock); //should fail for non-master
+    print_current_address(2);
+    #pragma omp barrier
+    #pragma omp master
+    {
+      omp_unset_nest_lock(&nest_lock);
+      print_current_address(3);
+      omp_unset_nest_lock(&nest_lock);
+      print_current_address(4);
+    }
+  }
+
+  omp_destroy_nest_lock(&nest_lock);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_nest_lock'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID:[0-9]+]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_first: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]] 
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_acquired_nest_lock_next: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS]] 
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_prev: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_release_nest_lock_last: wait_id=[[WAIT_ID]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK-NEXT: {{^}}[[MASTER_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_destroy_nest_lock: wait_id=[[WAIT_ID]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_wait_nest_lock: wait_id=[[WAIT_ID]], hint=0, impl={{[0-9]+}}, codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]] 
+  // CHECK-NOT: {{^}}[[THREAD_ID]]: ompt_event_acquired_nest_lock_next: wait_id=[[WAIT_ID]]
+  // CHECK-NEXT: {{^}}[[THREAD_ID]]: current_address={{.*}}[[RETURN_ADDRESS]]
+
+  return 0;
+}

runtime/test/ompt/tasks/dependences.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <omp.h>   
+#include <math.h>
+#include <unistd.h>
+
+int main()
+{
+  int x = 0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {  
+      #pragma omp task depend(out:x)
+      {
+        x++;
+        usleep(100);
+      }
+      print_fuzzy_address(1);
+    
+      #pragma omp task depend(in:x)
+      {
+        x = -1;
+      }
+    }
+  }
+
+  x++;
+
+
+  // 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_task_dependences'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_dependence'
+  
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+
+  // CHECK: {{^}}{{[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter=[[NULL]], new_task_id=[[FIRST_TASK:[0-f]+]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, task_type=ompt_task_explicit=4, has_dependences=yes
+  // CHECK: {{^}}{{[0-9]+}}: ompt_event_task_dependences: task_id=[[FIRST_TASK]], deps={{0x[0-f]+}}, ndeps=1
+  // CHECK: {{^}}{{[0-9]+}}: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+
+  // CHECK: {{^}}{{[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter=[[NULL]], new_task_id=[[SECOND_TASK:[0-f]+]], codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit=4, has_dependences=yes
+  // CHECK: {{^}}{{[0-9]+}}: ompt_event_task_dependences: task_id=[[SECOND_TASK]], deps={{0x[0-f]+}}, ndeps=1
+  // CHECK: {{^}}{{[0-9]+}}: ompt_event_task_dependence_pair: first_task_id=[[FIRST_TASK]], second_task_id=[[SECOND_TASK]]
+
+
+  return 0;
+}

runtime/test/ompt/tasks/explicit_task.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h> 
+
+int main()
+{
+  int condition=0;
+  omp_set_nested(0);
+  print_frame(0);
+  #pragma omp parallel num_threads(2)
+  {
+    print_frame(1);
+    print_ids(0);
+    print_ids(1);
+    print_frame(0);
+    #pragma omp master
+    {
+      print_ids(0);
+      #pragma omp task shared(condition)
+      {
+        OMPT_SIGNAL(condition);
+        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+      }
+      print_fuzzy_address(1);
+      OMPT_WAIT(condition,1);
+      print_ids(0);
+    }
+    #pragma omp barrier
+    print_ids(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_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: __builtin_frame_address(0)=[[MAIN_REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+  // nested parallel masters
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // <- ompt_event_task_create would be expected here
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // explicit barrier after master
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // implicit barrier parallel
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // this is expected to come earlier and at MASTER:
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[IMPLICIT_TASK_ID]], second_task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[TASK_EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], exit_frame=[[TASK_EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[TASK_ID]], second_task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_end: task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+
+
+  return 0;
+}

runtime/test/ompt/tasks/serialized.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>   
+#include <math.h>
+
+int main()
+{
+  omp_set_nested(0);
+  print_frame(0);
+  #pragma omp parallel num_threads(2)
+  {
+    print_frame(1);
+    print_ids(0);
+    print_ids(1);
+    print_frame(0);
+    #pragma omp master
+    {
+      print_ids(0);
+      int t = (int)sin(0.1);
+      #pragma omp task if(t)
+      {
+        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+      }
+      print_fuzzy_address(1);
+      print_ids(0);
+    }
+    print_ids(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_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+  
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[NULL]], new_task_id={{[0-9]+}}, codeptr_ra=[[NULL]], task_type=ompt_task_initial=1, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[MAIN_REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+
+  // nested parallel masters
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // <- ompt_event_task_schedule ([[IMPLICIT_TASK_ID]], [[TASK_ID]]) would be expected here
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[IMPLICIT_TASK_ID]], second_task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[TASK_EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], exit_frame=[[TASK_EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // <- ompt_event_task_schedule ([[TASK_ID]], [[IMPLICIT_TASK_ID]]) would be expected here
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[TASK_ID]], second_task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_end: task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reen
+
+  // implicit barrier parallel
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+
+
+  return 0;
+}

runtime/test/ompt/tasks/task_in_joinbarrier.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h> 
+
+int main()
+{
+  int condition=0;
+  omp_set_nested(0);
+  print_frame(0);
+  #pragma omp parallel num_threads(2)
+  {
+    print_frame(1);
+    print_ids(0);
+    print_ids(1);
+    print_frame(0);
+    #pragma omp master
+    {
+      print_ids(0);
+      #pragma omp task shared(condition)
+      {
+        OMPT_SIGNAL(condition);
+        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+      }
+      OMPT_WAIT(condition,1);
+      print_ids(0);
+    }
+    print_ids(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_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: __builtin_frame_address(0)=[[MAIN_REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+  // nested parallel masters
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // <- ompt_event_task_create would be expected here
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[TASK_FUNCTION:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // implicit barrier parallel
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // implicit barrier parallel
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[IMPLICIT_TASK_ID]], second_task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[TASK_EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], exit_frame=[[TASK_EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[TASK_ID]], second_task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_end: task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+
+
+  return 0;
+}

runtime/test/ompt/tasks/task_types.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h>
+#include <math.h>
+
+
+void print_task_type(int id)
+{
+  #pragma omp critical
+  {
+    int task_type;
+    char buffer[2048];
+    ompt_get_task_info(0, &task_type, NULL, NULL, NULL, NULL);
+    format_task_type(task_type, buffer);
+    printf("%" PRIu64 ": id=%d task_type=%s=%d\n", ompt_get_thread_data()->value, id, buffer, task_type);
+  }
+};
+
+int main()
+{
+  //initial task
+  print_task_type(0);
+
+  int x;
+  //implicit task
+  #pragma omp parallel num_threads(1)
+  {
+    print_task_type(1);
+    x++;
+  }
+
+  #pragma omp parallel num_threads(2)
+  #pragma omp master
+  {
+    //explicit task
+    #pragma omp task
+    {
+      print_task_type(2);
+      x++;
+    }
+
+    //explicit task with undeferred
+    #pragma omp task if(0)
+    {
+      print_task_type(3);
+      x++;
+    }
+
+    //explicit task with untied
+    #pragma omp task untied
+    {
+      print_task_type(4);
+      x++;
+    }
+
+    //explicit task with final
+    #pragma omp task final(1)
+    {
+      print_task_type(5);
+      x++;
+      //nested explicit task with final and undeferred
+      #pragma omp task
+      {
+        print_task_type(6);
+        x++;
+      }
+    }
+
+    //Mergeable task test deactivated for now
+    //explicit task with mergeable
+    /*
+    #pragma omp task mergeable if((int)sin(0))
+    {
+      print_task_type(7);
+      x++;
+    }
+    */
+
+    //TODO: merged task
+  }
+
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_task_create: parent_task_id=0, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[NULL]], new_task_id={{[0-9]+}}, codeptr_ra=[[NULL]], task_type=ompt_task_initial=1, has_dependences=no
+  // CHECK-NOT: 0: parallel_data initially not null
+  // CHECK: {{^}}[[MASTER_ID]]: id=0 task_type=ompt_task_initial=1
+  // CHECK: {{^}}[[MASTER_ID]]: id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730
+
+  // CHECK-DAG: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit=4, has_dependences=no
+  // CHECK-DAG: {{^[0-9]+}}: id=2 task_type=ompt_task_explicit=4
+
+  // CHECK-DAG: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred=134217732, has_dependences=no
+  // CHECK-DAG: {{^[0-9]+}}: id=3 task_type=ompt_task_explicit|ompt_task_undeferred=134217732
+
+  // CHECK-DAG: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_untied=268435460, has_dependences=no
+  // CHECK-DAG: {{^[0-9]+}}: id=4 task_type=ompt_task_explicit|ompt_task_untied=268435460
+
+  // CHECK-DAG: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_final=536870916, has_dependences=no
+  // CHECK-DAG: {{^[0-9]+}}: id=5 task_type=ompt_task_explicit|ompt_task_final=536870916
+
+  // CHECK-DAG: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644, has_dependences=no
+  // CHECK-DAG: {{^[0-9]+}}: id=6 task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644
+
+  return 0;
+}

runtime/test/ompt/tasks/task_types_serialized.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+
+#include "callback.h"
+#include <omp.h>
+
+void print_task_type(int id)
+{
+  #pragma omp critical
+  {
+    int task_type;
+    char buffer[2048];
+    ompt_get_task_info(0, &task_type, NULL, NULL, NULL, NULL);
+    format_task_type(task_type, buffer);
+    printf("%" PRIu64 ": id=%d task_type=%s=%d\n", ompt_get_thread_data()->value, id, buffer, task_type);
+  }
+};
+
+int main()
+{
+  //initial task
+  print_task_type(0);
+
+  int x;
+  //implicit task
+  #pragma omp parallel num_threads(1)
+  {
+    print_task_type(1);
+    x++;
+  }
+
+  #pragma omp parallel num_threads(1)
+  #pragma omp master
+  {
+    //explicit task
+    #pragma omp task
+    {
+      print_task_type(2);
+      x++;
+    }
+
+    //explicit task with undeferred
+    #pragma omp task if(0)
+    {
+      print_task_type(3);
+      x++;
+    }
+
+    //explicit task with untied
+    #pragma omp task untied
+    {
+      print_task_type(4);
+      x++;
+    }
+
+    //explicit task with final
+    #pragma omp task final(1)
+    {
+      print_task_type(5);
+      x++;
+      //nested explicit task with final and undeferred
+      #pragma omp task
+      {
+        print_task_type(6);
+        x++;
+      }
+    }
+
+/*
+    //TODO:not working
+    //explicit task with mergeable
+    #pragma omp task mergeable
+    {
+      print_task_type(7);
+      x++;
+    }
+*/
+
+    //TODO: merged task
+  }
+
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_task_create'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_task_create: parent_task_id=0, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[NULL]], new_task_id={{[0-9]+}}, codeptr_ra=[[NULL]], task_type=ompt_task_initial=1, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: id=0 task_type=ompt_task_initial=1
+  // CHECK: {{^}}[[MASTER_ID]]: id=1 task_type=ompt_task_implicit|ompt_task_undeferred=134217730
+
+  // CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred=134217732, has_dependences=no
+  // CHECK: {{^[0-9]+}}: id=2 task_type=ompt_task_explicit|ompt_task_undeferred=134217732
+
+  // CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred=134217732, has_dependences=no
+  // CHECK: {{^[0-9]+}}: id=3 task_type=ompt_task_explicit|ompt_task_undeferred=134217732
+
+  // CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_untied=402653188, has_dependences=no
+  // CHECK: {{^[0-9]+}}: id=4 task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_untied=402653188
+
+  // CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644, has_dependences=no
+  // CHECK: {{^[0-9]+}}: id=5 task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644
+
+  // CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644, has_dependences=no
+  // CHECK: {{^[0-9]+}}: id=6 task_type=ompt_task_explicit|ompt_task_undeferred|ompt_task_final=671088644
+
+  // ___CHECK: {{^[0-9]+}}: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id={{[0-9]+}}, codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit|ompt_task_undeferred=134217732, has_dependences=no
+  // ___CHECK: {{^[0-9]+}}: id=7 task_type=ompt_task_explicit|ompt_task_undeferred=134217732
+
+  return 0;
+}

runtime/test/ompt/tasks/taskyield.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// Current GOMP interface implements taskyield as stub
+// XFAIL: gcc
+
+#include "callback.h"
+#include <omp.h>   
+#include <unistd.h>
+
+int main()
+{
+  int condition=0, x=0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp master
+    {
+        #pragma omp task shared(condition)
+        {
+          OMPT_SIGNAL(condition);
+          OMPT_WAIT(condition,2);
+        }
+        OMPT_WAIT(condition,1);
+        #pragma omp task shared(x)
+        {
+          x++;
+        }
+        printf("%" PRIu64 ": before yield\n", ompt_get_thread_data()->value);
+        #pragma omp taskyield
+        printf("%" PRIu64 ": after yield\n", ompt_get_thread_data()->value);
+        OMPT_SIGNAL(condition);
+    }
+  }
+
+
+  // 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_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID:[0-9]+]], team_size={{[0-9]+}}, thread_num={{[0-9]+}}
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id=[[WORKER_TASK:[0-9]+]], codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit=4, has_dependences=no
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id={{[0-9]+}}, parent_task_frame.exit={{0x[0-f]+}}, parent_task_frame.reenter={{0x[0-f]+}}, new_task_id=[[MAIN_TASK:[0-9]+]], codeptr_ra={{0x[0-f]+}}, task_type=ompt_task_explicit=4, has_dependences=no
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[IMPLICIT_TASK_ID]], second_task_id=[[MAIN_TASK]], prior_task_status=ompt_task_yield=2
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_schedule: first_task_id=[[MAIN_TASK]], second_task_id=[[IMPLICIT_TASK_ID]], prior_task_status=ompt_task_complete=1
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_task_schedule: first_task_id={{[0-9]+}}, second_task_id=[[WORKER_TASK]], prior_task_status=ompt_task_others=4
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[WORKER_TASK]], second_task_id={{[0-9]+}}, prior_task_status=ompt_task_complete=1
+
+
+
+
+
+  return 0;
+}

runtime/test/ompt/tasks/untied_task.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+#include "callback.h"
+#include <omp.h> 
+
+int main()
+{
+  int condition=0;
+  omp_set_nested(0);
+  print_frame(0);
+  #pragma omp parallel num_threads(2)
+  {
+    print_frame(1);
+    print_ids(0);
+    print_ids(1);
+    print_frame(0);
+    #pragma omp master
+    {
+      print_ids(0);
+      #pragma omp task untied shared(condition)
+      {
+        OMPT_SIGNAL(condition);
+        print_frame(1);
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+        #pragma omp task if(0)
+        {
+          print_ids(0);
+          print_ids(1);
+          print_ids(2);
+        }
+        print_ids(0);
+        print_ids(1);
+        print_ids(2);
+      }
+      OMPT_WAIT(condition,1);
+      print_ids(0);
+    }
+    #pragma omp barrier
+    print_ids(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_task_schedule'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquire'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_acquired'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_mutex_released'
+
+
+  // CHECK: {{^}}0: NULL_POINTER=[[NULL:.*$]]
+
+  // make sure initial data pointers are null
+  // CHECK-NOT: 0: new_task_data initially not null
+  
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: __builtin_frame_address(0)=[[MAIN_REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter=[[MAIN_REENTER]], parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=2, codeptr_ra=0x{{[0-f]+}}, invoker=[[PARALLEL_INVOKER:[0-9]+]]
+  // nested parallel masters
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID:[0-9]+]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // <- ompt_event_task_create would be expected here
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_task_create: parent_task_id=[[IMPLICIT_TASK_ID]], parent_task_frame.exit=[[EXIT]], parent_task_frame.reenter=[[REENTER]], new_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[TASK_FUNCTION:0x[0-f]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // explicit barrier after master
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // implicit barrier parallel
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(0)=[[REENTER:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // this is expected to come earlier and at MASTER:
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[IMPLICIT_TASK_ID]], second_task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: __builtin_frame_address(1)=[[TASK_EXIT:0x[0-f]+]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], exit_frame=[[TASK_EXIT]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 1: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[EXIT]], reenter_frame=[[REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 2: parallel_id=[[IMPLICIT_PARALLEL_ID]], task_id=[[PARENT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[MAIN_REENTER]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_schedule: first_task_id=[[TASK_ID]], second_task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_task_end: task_id=[[TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: task level 0: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]], exit_frame=[[NULL]], reenter_frame=[[NULL]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_barrier_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
+
+
+
+  return 0;
+}

runtime/test/ompt/worksharing/for/auto_split.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %S/base_split.h
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=CHECK-LOOP %S/base_split.h
+// REQUIRES: ompt
+// GCC doesn't call runtime for auto = static schedule
+// XFAIL: gcc
+
+#define SCHEDULE auto
+#include "base_split.h"

runtime/test/ompt/worksharing/for/base.h

 #include "callback.h"
 #include <omp.h>
 
 int main()
 {
   unsigned int i;
 
   #pragma omp parallel for num_threads(4) schedule(SCHEDULE)
   for (i = 0; i < 4; i++) {
   }
 
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_work'
+
+
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id={{[0-9]+}}, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, parallel_function=0x{{[0-f]+}}, invoker={{.*}}
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id={{[0-9]+}}, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=0x{{[0-f]+}}, invoker={{[0-9]+}}
 
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], workshare_function=0x{{[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=0x{{[0-f]+}}
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], workshare_function=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=0x{{[0-f]+}}
   // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], workshare_function=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=0x{{[0-f]+}}
   // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], workshare_function=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=0x{{[0-f]+}}
   // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_implicit_task_end: parallel_id={{[0-9]+}}, task_id=[[IMPLICIT_TASK_ID]]
 
   return 0;
 }

runtime/test/ompt/worksharing/for/base_serialized.h

 #include "callback.h"
 #include <omp.h>
 
 int main()
 {
   unsigned int i;
 
   #pragma omp parallel for num_threads(1) schedule(SCHEDULE)
   for (i = 0; i < 1; i++) {
   }
   
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_work'
+
+
   // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
-  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=1, parallel_function=0x{{[0-f]+}}, invoker={{.+}}
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id=[[PARENT_TASK_ID:[0-9]+]], parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=1, codeptr_ra=0x{{[0-f]+}}, invoker={{[0-9]+}}
 
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], workshare_function=0x{{[0-f]+}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], task_id=[[IMPLICIT_TASK_ID:[0-9]+]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[IMPLICIT_TASK_ID]], codeptr_ra=0x{{[0-f]+}}
   // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
-  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id=[[PARALLEL_ID]], task_id=[[IMPLICIT_TASK_ID]]
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_implicit_task_end: parallel_id={{[PARALLEL_ID,0]}}, task_id=[[IMPLICIT_TASK_ID]]
 
   return 0;
 }

runtime/test/ompt/worksharing/for/base_split.h

+#include "callback.h"
+#include <omp.h>
+
+/* With the combined parallel-for construct (base.h), the return-addresses are hard to compare.
+   With the separate parallel and for-nowait construct, the addresses become more predictable,
+   but the begin of the for-loop still generates additional code, so the offset of loop-begin 
+   to the label is >4 Byte.
+*/
+
+int main()
+{
+  unsigned int i;
+
+  #pragma omp parallel num_threads(4) 
+  {
+    print_current_address(0);
+    #pragma omp for schedule(SCHEDULE) nowait
+    for (i = 0; i < 4; i++) {
+      print_fuzzy_address(1);
+    }
+    print_fuzzy_address(2);
+  }
+  print_fuzzy_address(3);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_begin'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_parallel_end'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_implicit_task'
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_work'
+
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id={{[0-9]+}}, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra=[[PARALLEL_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}, invoker={{[0-9]+}}
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id={{[0-9]+}}, codeptr_ra=[[LOOP_BEGIN_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[LOOP_END_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[LOOP_END_RETURN_ADDRESS]]
+
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_parallel_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, invoker={{[0-9]+}}, codeptr_ra=[[PARALLEL_RETURN_ADDRESS]]
+  // CHECK: {{^}}[[MASTER_ID]]: fuzzy_address={{.*}}[[PARALLEL_RETURN_ADDRESS]]
+  
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[LOOP_END_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[LOOP_END_RETURN_ADDRESS]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[LOOP_END_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[LOOP_END_RETURN_ADDRESS]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id={{[0-9]+}}, codeptr_ra=0x{{[0-f]+}}
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_loop_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[LOOP_END_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK: {{^}}[[THREAD_ID]]: fuzzy_address={{.*}}[[LOOP_END_RETURN_ADDRESS]]
+
+
+  // CHECK-LOOP: 0: NULL_POINTER=[[NULL:.*$]]
+  // CHECK-LOOP: 0: ompt_event_runtime_shutdown
+  // CHECK-LOOP: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_parallel_begin: parent_task_id={{[0-9]+}}, parent_task_frame.exit=[[NULL]], parent_task_frame.reenter={{0x[0-f]+}}, parallel_id=[[PARALLEL_ID:[0-9]+]], requested_team_size=4, codeptr_ra={{0x[0-f]+}}, invoker={{[0-9]+}}
+  // CHECK-LOOP: {{^}}[[MASTER_ID]]: ompt_event_loop_begin: parallel_id=[[PARALLEL_ID]], parent_task_id={{[0-9]+}}, codeptr_ra=[[LOOP_BEGIN_RETURN_ADDRESS:0x[0-f]+]]{{[0-f][0-f]}}
+  // CHECK-LOOP: {{^}}{{[0-9]+}}: fuzzy_address={{.*}}[[LOOP_BEGIN_RETURN_ADDRESS]]
+  // CHECK-LOOP: {{^}}{{[0-9]+}}: fuzzy_address={{.*}}[[LOOP_BEGIN_RETURN_ADDRESS]]
+  // CHECK-LOOP: {{^}}{{[0-9]+}}: fuzzy_address={{.*}}[[LOOP_BEGIN_RETURN_ADDRESS]]
+  // CHECK-LOOP: {{^}}{{[0-9]+}}: fuzzy_address={{.*}}[[LOOP_BEGIN_RETURN_ADDRESS]]
+
+
+  return 0;
+}

runtime/test/ompt/worksharing/for/dynamic_split.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %S/base_split.h
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=CHECK-LOOP %S/base_split.h
+// REQUIRES: ompt
+
+#define SCHEDULE dynamic
+#include "base_split.h"

runtime/test/ompt/worksharing/for/guided_split.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %S/base_split.h
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=CHECK-LOOP %S/base_split.h
+// REQUIRES: ompt
+
+#define SCHEDULE guided
+#include "base_split.h"

runtime/test/ompt/worksharing/for/runtime_split.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %S/base_split.h
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=CHECK-LOOP %S/base_split.h
+// REQUIRES: ompt
+
+#define SCHEDULE runtime
+#include "base_split.h"

runtime/test/ompt/worksharing/for/static_split.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %S/base_split.h
+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck --check-prefix=CHECK-LOOP %S/base_split.h
+// REQUIRES: ompt
+// GCC doesn't call runtime for static schedule
+// XFAIL: gcc
+
+#define SCHEDULE static
+#include "base_split.h"

runtime/test/ompt/worksharing/sections.c

+// RUN: %libomp-compile-and-run | FileCheck %s
+// REQUIRES: ompt
+// GCC generates code that does not distinguish between sections and loops
+// XFAIL: gcc
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  #pragma omp parallel sections num_threads(2)
+  {
+    #pragma omp section
+    {
+      printf("%lu: section 1\n", ompt_get_thread_data()->value);
+    }
+    #pragma omp section
+    {
+      printf("%lu: section 2\n", ompt_get_thread_data()->value);
+    }
+  }
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_work'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[MASTER_ID:[0-9]+]]: ompt_event_sections_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], parent_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[SECT_BEGIN:0x[0-f]+]], count=2
+  // CHECK: {{^}}[[MASTER_ID]]: ompt_event_sections_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[SECT_END:0x[0-f]+]]
+
+  // CHECK: {{^}}[[THREAD_ID:[0-9]+]]: ompt_event_sections_begin: parallel_id=[[PARALLEL_ID]], parent_task_id=[[TASK_ID:[0-9]+]], codeptr_ra=[[SECT_BEGIN]], count=2
+  // CHECK: {{^}}[[THREAD_ID]]: ompt_event_sections_end: parallel_id=[[PARALLEL_ID]], task_id={{[0-9]+}}, codeptr_ra=[[SECT_END]]
+
+
+  return 0;
+}

runtime/test/ompt/worksharing/single.c

+// RUN: %libomp-compile-and-run | %sort-threads | FileCheck %s
+// REQUIRES: ompt
+// GCC generates code that does not call the runtime for the single construct
+// XFAIL: gcc
+
+#include "callback.h"
+#include <omp.h>
+
+int main()
+{
+  int x = 0;
+  #pragma omp parallel num_threads(2)
+  {
+    #pragma omp single
+    {
+      x++;
+    }
+  }
+
+  printf("x=%d\n", x);
+
+  // Check if libomp supports the callbacks for this test.
+  // CHECK-NOT: {{^}}0: Could not register callback 'ompt_callback_work'
+
+  // CHECK: 0: NULL_POINTER=[[NULL:.*$]]
+
+  // CHECK: {{^}}[[THREAD_ID_1:[0-9]+]]: ompt_event_single_in_block_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], parent_task_id=[[TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]+}}, count=1
+  // CHECK: {{^}}[[THREAD_ID_1]]: ompt_event_single_in_block_end: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra={{0x[0-f]+}}, count=1
+
+  // CHECK: {{^}}[[THREAD_ID_2:[0-9]+]]: ompt_event_single_others_begin: parallel_id=[[PARALLEL_ID:[0-9]+]], task_id=[[TASK_ID:[0-9]+]], codeptr_ra={{0x[0-f]+}}, count=1
+  // CHECK: {{^}}[[THREAD_ID_2]]: ompt_event_single_others_end: parallel_id=[[PARALLEL_ID]], task_id=[[TASK_ID]], codeptr_ra={{0x[0-f]+}}, count=1
+
+
+
+  return 0;
+}