Skip to content

Bugfix：Avoid deadlock issues caused by logging in a loop #1265

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Bugfix：Avoid deadlock issues caused by logging in a loop #1265

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bfdb006
Avoid deadlock issues caused by logging in a loop.
wangfakang Apr 25, 2024
ad475ae
remove some duplicats logical thx @kiskra-nvidia.
wangfakang Apr 27, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 15 additions & 11 deletions src/misc/param.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,17 +65,21 @@ void ncclLoadParam(char const* env, int64_t deftVal, int64_t uninitialized, int6
if (__atomic_load_n(cache, __ATOMIC_RELAXED) == uninitialized) {
const char* str = ncclGetEnv(env);
int64_t value = deftVal;
if (str && strlen(str) > 0) {
errno = 0;
value = strtoll(str, nullptr, 0);
if (errno) {
value = deftVal;
INFO(NCCL_ALL,"Invalid value %s for %s, using default %lld.", str, env, (long long)deftVal);
} else {
INFO(NCCL_ENV,"%s set by environment to %lld.", env, (long long)value);
}
if (!str || strlen(str) <= 0) {
__atomic_store_n(cache, value, __ATOMIC_RELAXED);
pthread_mutex_unlock(&mutex);
return;
}
errno = 0;
value = strtoll(str, nullptr, 0);
// To prevent deadlock issues caused by logging in a loop,
// so cache the value before the log operation.
__atomic_store_n(cache, errno ? deftVal : value, __ATOMIC_RELAXED);
if (errno) {
INFO(NCCL_ALL,"Invalid value %s for %s, using default %lld.", str, env, (long long)deftVal);
} else {
INFO(NCCL_ENV,"%s set by environment to %lld.", env, (long long)value);
}
__atomic_store_n(cache, value, __ATOMIC_RELAXED);
}
pthread_mutex_unlock(&mutex);
}
Expand All @@ -84,4 +88,4 @@ const char *ncclGetEnv(const char *name) {
static pthread_once_t once = PTHREAD_ONCE_INIT;
pthread_once(&once, initEnv);
return getenv(name);
}
}