Skip to content

Conversation

@hieblmi
Copy link
Collaborator

@hieblmi hieblmi commented Jun 13, 2025

Fixes a deadlock in the deposit recovery that arises if deposits expired while the client was offline.
Since recoverDeposits blocks on fsm.SendEvent(ctx, OnRecover, nil) the main run loop never gets a chance to read from m.finalizedDepositChan.

@hieblmi hieblmi marked this pull request as draft June 13, 2025 11:01
@hieblmi hieblmi force-pushed the fix-deadlock branch 3 times, most recently from eb832fd to 7bd246e Compare June 13, 2025 11:59
@hieblmi hieblmi self-assigned this Jun 13, 2025
@hieblmi hieblmi marked this pull request as ready for review June 13, 2025 12:01
@hieblmi hieblmi requested review from bhandras, sputn1ck and starius June 13, 2025 12:01
@hieblmi hieblmi force-pushed the fix-deadlock branch 3 times, most recently from de3c60e to 8285bc7 Compare June 13, 2025 13:42
@hieblmi hieblmi mentioned this pull request Jun 16, 2025
if err != nil {
log.Errorf("Error sending OnStart event: %v", err)
}
go func(fsm *FSM) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wonder what the tradeoff is between having this in a go func or more asynchronous fsm actions? from what I can tell in effect it is the same?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But with most recent projects I tend to force tha actions to be asynchronous as this means we can go events only vs listening for channels everywhere where we need to interact with a blocking action

Copy link
Collaborator

@starius starius left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! 🌴
One proposal to prevent this issue from silently recurring.

log.Errorf("Error sending OnStart event: %v",
err)
}
}(fsm)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I propose to add a unit test that would fail with the dead-lock before this change and to make sure this goroutine doesn't leak in that test.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Given that is is a pattern we use frequently I only left a unit-test todo here.

@hieblmi hieblmi merged commit f7769d2 into lightninglabs:master Jun 23, 2025
4 checks passed
@hieblmi hieblmi deleted the fix-deadlock branch June 23, 2025 08:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants