diff options
author | Shilei Tian <i@tianshilei.me> | 2023-06-14 11:45:49 -0400 |
---|---|---|
committer | Shilei Tian <i@tianshilei.me> | 2023-06-14 11:46:12 -0400 |
commit | 85592d3d4d402b99df32bcc711bca56a8a593c97 (patch) | |
tree | 3700eba12361be6971ceea9ba7b9fce8ffa20262 /openmp/runtime/src | |
parent | 27fac4a72ae54a471471a69c0ad999585ccbb026 (diff) | |
download | llvm-85592d3d4d402b99df32bcc711bca56a8a593c97.zip llvm-85592d3d4d402b99df32bcc711bca56a8a593c97.tar.gz llvm-85592d3d4d402b99df32bcc711bca56a8a593c97.tar.bz2 |
[OpenMP] Fix the issue where `num_threads` still takes effect incorrectly
This patch fixes the issue that, if we have a compile-time serialized parallel
region (such as `if (0)`) with `num_threads`, followed by a regular parallel
region, the regular parallel region will pick up the value set in the serialized
parallel region incorrectly. The reason is, in the front end, if we can prove a
parallel region has to serialized, instead of emitting `__kmpc_fork_call`, the
front end directly emits `__kmpc_serialized_parallel`, body, and `__kmpc_end_serialized_parallel`.
However, this "optimization" doesn't consider the case where `num_threads` is
used such that `__kmpc_push_num_threads` is still emitted. Since we don't reset
the value in `__kmpc_serialized_parallel`, it will affect the next parallel region
followed by it.
Fix #63197.
Reviewed By: tlwilmar
Differential Revision: https://reviews.llvm.org/D152883
Diffstat (limited to 'openmp/runtime/src')
-rw-r--r-- | openmp/runtime/src/kmp_runtime.cpp | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/openmp/runtime/src/kmp_runtime.cpp b/openmp/runtime/src/kmp_runtime.cpp index 54d2805..03956ac 100644 --- a/openmp/runtime/src/kmp_runtime.cpp +++ b/openmp/runtime/src/kmp_runtime.cpp @@ -1153,6 +1153,9 @@ void __kmp_serialized_parallel(ident_t *loc, kmp_int32 global_tid) { // Reset for next parallel region this_thr->th.th_set_proc_bind = proc_bind_default; + // Reset num_threads for next parallel region + this_thr->th.th_set_nproc = 0; + #if OMPT_SUPPORT ompt_data_t ompt_parallel_data = ompt_data_none; void *codeptr = OMPT_LOAD_RETURN_ADDRESS(global_tid); |