quick fix for lingering snapdir unmount problems

Unfortunately, even after e79b680, I still, much more rarely,
tripped asserts when playing with many ctldir mounts at once.

Since this appears to happen if we dispatched twice too fast, just
ignore it. We don't actually need to do anything if someone already
started doing it for us.

Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Rich Ercolani <[email protected]>
Closes #14462

Author: rincebrain

module/os/linux/zfs/zfs_ctldir.c

@@ -392,7 +392,20 @@ zfsctl_snapshot_unmount_delay_impl(zfs_snapentry_t *se, int delay)
 
 	zfsctl_snapshot_hold(se);
 	rw_enter(&se->se_taskqid_lock, RW_WRITER);
-	ASSERT3S(se->se_taskqid, ==, TASKQID_INVALID);
+	/*
+	 * If this condition happens, we managed to:
+	 * - dispatch once
+	 * - want to dispatch _again_ before it returned
+	 *
+	 * So let's just return - if that task fails at unmounting,
+	 * we'll eventually dispatch again, and if it succeeds,
+	 * no problem.
+	 */
+	if (se->se_taskqid != TASKQID_INVALID) {
+		rw_exit(&se->se_taskqid_lock);
+		zfsctl_snapshot_rele(se);
+		return;
+	}
 	se->se_taskqid = taskq_dispatch_delay(system_delay_taskq,
 	    snapentry_expire, se, TQ_SLEEP, ddi_get_lbolt() + delay * HZ);
 	rw_exit(&se->se_taskqid_lock);