Re: 4.1-rc6 radi5 OOPS

[Neil Brown <neilb@suse.de>]

On Thu, 11 Jun 2015 17:02:57 +1000
Neil Brown <neilb@suse.de> wrote:

> On Thu, 11 Jun 2015 16:48:47 +1000
> Neil Brown <neilb@suse.de> wrote:
> 
> 
> > Would you be able to test with the following patch?  There is a chance it might
> > confirm whether two sync threads are running at the same time.
> 
> 
> I tried the patch, and was fortunate:
> 
> [  208.626468] Register thread ffff88003cf53d00 md0_resync
> [  208.629801] Started thread ffff88003cf53d00 md0_resync
> [  208.629801] Finished thread ffff88003cf53d00
> [  208.639800] Reap thread ffff88003cf53d00
> [  208.639800] Finished thread ffff88003cf53d00 md0_resync
> [  208.643133] Register thread ffff88003cf53700 md0_resync
> [  208.643133] Reap thread           (null)
> [  208.646466] Started thread ffff88003cf53700 md0_resync
> [  208.659799] Reap thread ffff88003cf53700
> -HANG-
> 
> That "(null)" is a problem - that will clear MD_RECOVERY_RUNNING, but there
> is a thread there.
> Will keep looking.
> 
> NeilBrown

Could you try this?  It is just a hunch, but I'm about to stop for the evening
so I won't find anything more until tomorrow.

Thanks,
NeilBrown

diff --git a/drivers/md/md.c b/drivers/md/md.c
index d4f31e1..b966b5b 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -4217,13 +4217,13 @@ action_store(struct mddev *mddev, const char *page, size_t len)
 		clear_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
 
 	if (cmd_match(page, "idle") || cmd_match(page, "frozen")) {
-		flush_workqueue(md_misc_wq);
-		if (mddev->sync_thread) {
-			set_bit(MD_RECOVERY_INTR, &mddev->recovery);
-			if (mddev_lock(mddev) == 0) {
+		if (mddev_lock(mddev) == 0) {
+			flush_workqueue(md_misc_wq);
+			if (mddev->sync_thread) {
+				set_bit(MD_RECOVERY_INTR, &mddev->recovery);
 				md_reap_sync_thread(mddev);
-				mddev_unlock(mddev);
 			}
+			mddev_unlock(mddev);
 		}
 	} else if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery) ||
 		   test_bit(MD_RECOVERY_NEEDED, &mddev->recovery))