diff options
| author | john stultz <johnstul@us.ibm.com> | 2010-06-04 15:32:19 -0700 |
|---|---|---|
| committer | Thomas Gleixner <tglx@linutronix.de> | 2010-06-08 20:02:30 +0200 |
| commit | bdeadaf2f7daf6c098e83506d4d99142ca3912f8 (patch) | |
| tree | 9f7f69d68375190321eb340573fc996c8fa5eac1 | |
| parent | 697684652c217b241a07d9e261c7a20e9b072d43 (diff) | |
| download | lwn-bdeadaf2f7daf6c098e83506d4d99142ca3912f8.tar.gz lwn-bdeadaf2f7daf6c098e83506d4d99142ca3912f8.zip | |
nfs: Avoid igrab deadlock.
On Fri, 2010-05-21 at 11:04 -0400, Jan Kansky wrote:
> It's crashing when I try to login as a user that has a nfs4 mounted home
> directory. See attached.
>
> I saw the discussion here:
>
> http://kerneltrap.org/mailarchive/linux-kernel/2010/5/7/4566886/thread
>
> That patch you mentioned there is already included in 2.6.33.4, and yet
> I still see:
>
> kernel BUG at kernel/rtmutex.c:808!
So the issue is nfs4_get_open_state() is calling igrab() while holding
the inode->i_lock. With the vfs scalability patches, igrab() takes that
lock as well, so we deadlock.
I believe the following resolves the issue, however I'm not sure if its
totally correct or not, since I'm not totally sure if the __iget()
rules.
With this patch, Jan Kansky (the original reporter) still had booting
issues, but was unable to provide details.
However, this patch has been tested by John Kacur and resolves the hang
for him, however we still see MNT_MOUNTED WARN_ONs that I'm still trying
to decide how to fix, but are otherwise harmless.
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Cc: "Luis Claudio R. Goncalves" <lclaudio@uudg.org>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Jan Kansky <kansky@ll.mit.edu>
Cc: John Kacur <jkacur@redhat.com>
Cc: Clark Williams <williams@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
| -rw-r--r-- | fs/nfs/nfs4state.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index c1e2733f4fa4..7ae43df98d04 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -501,7 +501,8 @@ nfs4_get_open_state(struct inode *inode, struct nfs4_state_owner *owner) state->owner = owner; atomic_inc(&owner->so_count); list_add(&state->inode_states, &nfsi->open_states); - state->inode = igrab(inode); + __iget(inode); + state->inode = inode; spin_unlock(&inode->i_lock); /* Note: The reclaim code dictates that we add stateless * and read-only stateids to the end of the list */ |