[Ocfs2-devel] [PATCH] ocfs2: Fix possible deadlock in ocfs2_global_read_dquot()

[Joel Becker <Joel.Becker@oracle.com>]

On Tue, Jun 02, 2009 at 02:23:59PM +0200, Jan Kara wrote:
> It is not possible to get a read lock and then try to get the same write lock
> in one thread as that can block on downconvert being requested by other node
> leading to deadlock. So first drop the quota lock for reading and only after
> that get it for writing.
> 
> Signed-off-by: Jan Kara <jack@suse.cz>
> ---
>  fs/ocfs2/quota_global.c |    4 +++-
>  1 files changed, 3 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/ocfs2/quota_global.c b/fs/ocfs2/quota_global.c
> index 1ed0f7c..edfa60c 100644
> --- a/fs/ocfs2/quota_global.c
> +++ b/fs/ocfs2/quota_global.c
> @@ -421,6 +421,7 @@ int ocfs2_global_read_dquot(struct dquot *dquot)
>  	OCFS2_DQUOT(dquot)->dq_originodes = dquot->dq_dqb.dqb_curinodes;
>  	if (!dquot->dq_off) {	/* No real quota entry? */
>  		/* Upgrade to exclusive lock for allocation */
> +		ocfs2_qinfo_unlock(info, 0);
>  		err = ocfs2_qinfo_lock(info, 1);
>  		if (err < 0)
>  			goto out_qlock;

	Can other nodes change the state of dq_off while you're
getting the exclusive?  I'm wondering if you have to retry the
qtree_read_dquot(), and I don't know enough about the code to say yes or
no.

Joel

-- 

	f/8 and be there.

Joel Becker
Principal Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127