Skip to content

wait-group: Only decrement counter to zero with mutex held #38

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

Merged
avdgrinten merged 1 commit into from
Feb 7, 2026
Merged

wait-group: Only decrement counter to zero with mutex held #38

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
25 changes: 13 additions & 12 deletions include/async/wait-group.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,20 +46,19 @@ struct wait_group {
>
> items;

while (1) {
auto v = ctr_.load(std::memory_order_acquire);
if (ctr_.compare_exchange_strong(v, v - 1, std::memory_order_acq_rel, std::memory_order_acquire)) {
assert(v > 0);
if (v != 1) {
return;
} else {
break;
}
}
// We can decrement outside of the lock as long as we do not reach zero.
auto v = ctr_.load(std::memory_order_relaxed);
while (v > 1) {
if (ctr_.compare_exchange_strong(v, v - 1, std::memory_order_acq_rel, std::memory_order_relaxed))
return;
}
assert(v > 0);

{
frg::unique_lock lock(mutex_);
// Only wake waiters if we reach zero.
if (ctr_.fetch_sub(1, std::memory_order_acq_rel) > 1)
return;
items.splice(items.end(), queue_);
}

Expand Down Expand Up @@ -89,7 +88,8 @@ struct wait_group {
{
frg::unique_lock lock(wg_->mutex_);

if(wg_->ctr_.load(std::memory_order_acquire) > 0) {
// Relaxed since non-zero -> zero transitions cannot happen while the mutex is held.
if(wg_->ctr_.load(std::memory_order_relaxed) > 0) {
if(!cobs_.try_set(ct_)) {
cancelled = true;
}else{
Expand All @@ -108,7 +108,8 @@ struct wait_group {
{
frg::unique_lock lock(wg_->mutex_);

if(wg_->ctr_.load(std::memory_order_acquire) > 0) {
// Relaxed since non-zero -> zero transitions cannot happen while the mutex is held.
if(wg_->ctr_.load(std::memory_order_relaxed) > 0) {
cancelled = true;
auto it = wg_->queue_.iterator_to(this);
wg_->queue_.erase(it);
Expand Down
Loading