[OpenMP] Fixed an issue that taskwait doesn't work on detachable task

[tianshilei1992]

Summary

D77609 mistakenly changed the bebavior of task waiting on detachable task that a detachable task is not waited, based on https://lists.llvm.org/pipermail/openmp-dev/2021-February/003836.html. This patch fixed it. Thank Raúl for the report.

[jdoerfert]

Can we use the email reproducer as test?

[jdoerfert]

Also, include a reference to the email in the commit message https://lists.llvm.org/pipermail/openmp-dev/2021-February/003836.html

[tianshilei1992]

Added the reproducer

[jdoerfert]

LG

[tianshilei1992]

Actually I'm pretty curious about how Windows tests passed...It is using pthread...

[tianshilei1992]

Make the test target independent

openmp/runtime/src/kmp_tasking.cpp

if (!KMP_TASKING_ENABLED(task_team)) {
       kmp_int32 tid = encountering_thread->th.th_info.ds.ds_tid;
       kmp_thread_data_t *thread_data = &task_team->tt.tt_threads_data[tid];
       // No lock needed since only owner can allocate
       if (thread_data->td.td_deque == NULL) {
         __kmp_alloc_task_deque(encountering_thread, thread_data);
       }
     }
 
-    if (flags->proxy == TASK_PROXY &&
+    if ((flags->proxy == TASK_PROXY || flags->detachable == TASK_DETACHABLE) &&
         task_team->tt.tt_found_proxy_tasks == FALSE)
       TCW_4(task_team->tt.tt_found_proxy_tasks, TRUE);
     if (flags->hidden_helper &&
         task_team->tt.tt_hidden_helper_task_encountered == FALSE)
       TCW_4(task_team->tt.tt_hidden_helper_task_encountered, TRUE);
   }
 
   // Calculate shared structure offset including padding after kmp_task_t struct