Re: [BUG 2.6.20-rc3-mm1] raid1 mount blocks for ever

[Jens Axboe <jens.axboe@oracle.com>]

On Sat, Jan 06 2007, Torsten Kaiser wrote:
> On Saturday 06 January 2007 11:02, Jens Axboe wrote:
> > On Sat, Jan 06 2007, Torsten Kaiser wrote:
> > > On Saturday 06 January 2007 04:59, Andrew Morton wrote:
> > > > http://userweb.kernel.org/~akpm/2.6.20-rc3-mm1x.bz2 is basically
> > > > 2.6.20-rc3-mm1, minus git-block.patch.  Can you and Torsten please
> > > > test that, see if the hangs go away?
> > >
> > > Works for me too.
> >
> > Torsten, can you test XFS with barriers disabled? I have a feeling that
> > is where the problem lies.
> 
> With your patch from Bug 7775: Same hang detected by NMI watchdog

Does this fix it? Only apply this one, not the patch from 7775.

diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index ec40e44..bae57e0 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -1542,7 +1542,7 @@ static inline void queue_sync_plugs(request_queue_t *q)
 	 * If the current process is plugged and has barriers submitted,
 	 * we will livelock if we don't unplug first.
 	 */
-	blk_unplug_current();
+	blk_replug_current_nested();
 
 	synchronize_qrcu(&q->qrcu);
 }

-- 
Jens Axboe