A potential error due to the unreleased lock

[ycaibb]

Dear developers: Thank you for your checking. It seems there is a lock resource leak on the lock paused. I think the lock should be released at the end of do_pause?

nbdkit/filters/pause/pause.c

Lines 146 to 163 in 9c19c0b

	static void
	do_pause (void)
	{
	if (is_paused) return;
	/* Grabbing the paused lock is enough to stop request processing. */
	pthread_mutex_lock (&paused);
	is_paused = true;
	/* However we must also wait until all outstanding requests have
	* been completed before we send the acknowledgement.
	*/
	nbdkit_debug ("pause: pausing, waiting for requests to complete");
	ACQUIRE_LOCK_FOR_CURRENT_SCOPE (&count_lock);
	while (count_requests > 0)
	pthread_cond_wait (&count_cond, &count_lock);
	nbdkit_debug ("pause: paused");
	}

[rwmjones]

Which lock do you mean?

[ycaibb]

This one

nbdkit/filters/pause/pause.c

Line 152 in 9c19c0b

pthread_mutex_lock (&paused);

[rwmjones]

It is unlocked in the next function.

[ycaibb]

But here is a break?

nbdkit/filters/pause/pause.c

Lines 198 to 202 in 9c19c0b

	case 'p':
	do_pause ();
	break;
	case 'r':
	do_resume ();

nbdkit/filters/pause/pause.c

Lines 166 to 174 in 9c19c0b

	do_resume (void)
	{
	if (!is_paused) return;
	/* Release the worker threads. */
	pthread_mutex_unlock (&paused);
	is_paused = false;
	nbdkit_debug ("pause: resumed");
	}

[rwmjones]

https://libguestfs.org/nbdkit-pause-filter.1.html#Control-socket
Would suggest reading how the pause filter works. Unless there is a specific problem you are seeing I will close this issue. BTW gitlab is the site used for nbdkit: https://gitlab.com/nbdkit/nbdkit

[ycaibb]

This case is detected by our static code analyzer, which might be a false positive. Thank you very much for your time.