Moved lock calls out of OMPT code

The removal of the idle callback was done in D48362 and is removed from this Patch.

Add comment explaining the use of alloca()

Use alloca() to force availability of frame pointer

Add a testcase that calls the runtime entry points when OMPT is not enabled.

Add check to ompt_get_state()

In D47709#1125178, @protze.joachim wrote:

taskloop is actually a tasking construct and explicitly no worksharing construct. So, please move the test in the task directory.

Add support for OMPT return addresses.
Also test them.

Move ompt_enabled.enabled checks out of internal functions

Use a stride of 3 in testcase loop

Use tc for the iteration count

Address Jonas' comments:

Use kmpc_omp_task() instead of changing kmp_omp_task().

Remove formatting change unrelated to this revision.

Fix uninitialized step count

Implement suggestions

Reformat

Reformat using git-clang-format
Applying clang-format to callback.h does not improve readability

Fix small copy-and-paste mistake

Update Formatting

Fix indentation

Add some comments to checks and reformat as suggested

Check task_ids in task_types.c testcase

Check ompt_get_partition_place_nums().
Fix small bug in implementation of ompt_get_partition_place_nums(): return correct length.

Exclude formatting of ompt_try_start_tool

Implemented suggestions

Merge D42470 into this revision

Implemented suggestions

In D41904#972169, @Hahnfeld wrote:

Hmm, can we also somehow test that the runtime continues searching for a tool in that case? For example having OMP_TOOL_LIBRARIES with a tool that has no ompt_start_tool, one that returns NULL and a final one that does activate OMPT?

Use omp_control_tool for an additional check

Added testcase

Added reference to reported issue in comments

Updated testcase

Implemented suggestion

Added Testcase

Updated revision

Removed unnecessary definition.

Implemented suggestion

Implemented suggestion

Added testcase

Implemented suggestions.

Include another return address handling fix

Fixed Diff

Fix formatting

Implemented all suggestions.

The use of UNSUPPORTED seems to be necessary. I updated the Diff.

In D40384#939025, @Hahnfeld wrote:

In D40384#938762, @omalyshe wrote:

In D40384#938757, @protze.joachim wrote:

@omalyshe Which version of icc introduced the ident information for workshare constructs?

ICC with the ident information has not been released yet

In that case cancel/cancel_worksharing.c should fail for icc-17, icc-18 too?