[Openmp-commits] [openmp] r317339 - Updating implementation of OMPT as specified in OpenMP 5.0 Preview 2 (TR6)

[Jonas Hahnfeld via Openmp-commits]

Am 2017-11-03 13:09, schrieb Joachim Protze via Openmp-commits:
> Author: jprotze
> Date: Fri Nov  3 10:09:00 2017
> New Revision: 317339
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=317339&view=rev
> Log:
> Updating implementation of OMPT as specified in OpenMP 5.0 Preview 2 
> (TR6)
> 
> The TR6 document is expected to be publically released around November 
> 15.
> This patch does not implement OMPT for libomptarget.
> 
> Patch by Simon Convent and Joachim Protze
> 
> Differential Revision: https://reviews.llvm.org/D39182
> 
> Removed:
>     openmp/trunk/runtime/test/lock/omp_init_lock.c
> Modified:
>     openmp/trunk/CREDITS.txt
>     openmp/trunk/runtime/.clang-format
>     openmp/trunk/runtime/CMakeLists.txt
>     openmp/trunk/runtime/src/exports_so.txt
>     openmp/trunk/runtime/src/include/50/ompt.h.var
>     openmp/trunk/runtime/src/kmp_csupport.cpp
>     openmp/trunk/runtime/src/kmp_dispatch.cpp
>     openmp/trunk/runtime/src/kmp_gsupport.cpp
>     openmp/trunk/runtime/src/kmp_i18n.cpp
>     openmp/trunk/runtime/src/kmp_itt.inl
>     openmp/trunk/runtime/src/kmp_lock.cpp
>     openmp/trunk/runtime/src/kmp_runtime.cpp
>     openmp/trunk/runtime/src/kmp_tasking.cpp
>     openmp/trunk/runtime/src/kmp_threadprivate.cpp
>     openmp/trunk/runtime/src/ompt-event-specific.h
>     openmp/trunk/runtime/src/ompt-general.cpp
>     openmp/trunk/runtime/src/ompt-internal.h
>     openmp/trunk/runtime/src/ompt-specific.cpp
>     openmp/trunk/runtime/test/ompt/callback.h
>     openmp/trunk/runtime/test/ompt/cancel/cancel_parallel.c
>     openmp/trunk/runtime/test/ompt/cancel/cancel_worksharing.c
>     openmp/trunk/runtime/test/ompt/loadtool/tool_available.c
>     openmp/trunk/runtime/test/ompt/ompt-signal.h
>     openmp/trunk/runtime/test/ompt/tasks/dependences.c
> 

The two OMPT commits revert some recent patches (libomptarget in 
r317339, omp_init_lock and __kmp_fatal_message in this one). I'd propose 
to revert both of them and recommit them properly so that we don't need 
to track down each lost change manually.

Jonas