| To: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: 3.2.9 and locking problem |
| From: | Dave Chinner <david@xxxxxxxxxxxxx> |
| Date: | Tue, 13 Mar 2012 11:00:52 +1100 |
| Cc: | Arkadiusz Mi??kiewicz <arekm@xxxxxxxx>, xfs@xxxxxxxxxxx |
| In-reply-to: | <20120312161425.GA23815@xxxxxxxxxxxxx> |
| References: | <201203092028.47177.arekm@xxxxxxxx> <20120312005325.GX5091@dastard> <201203121443.59217.arekm@xxxxxxxx> <20120312161425.GA23815@xxxxxxxxxxxxx> |
| User-agent: | Mutt/1.5.21 (2010-09-15) |
On Mon, Mar 12, 2012 at 12:14:25PM -0400, Christoph Hellwig wrote: > In the 3.2 version xfs_sync_flags does a double unlock of the ilock - > change the last argument of xfs_trans_ijoin from XFS_ILOCK_EXCL to 0 > will fix this. The 3.0 version doesn't have this bug. > > And btw, the additions to the on-disk format are gross - this stuff > broke before and will break again when we add new features. Any idea why they aren't sending stuff like this upstream to us so they can be implemented correctly, robustly and in a future-proof manner? vserver users are going to be unhappy when their filesystems get broken because they are using out-of-tree, incompatible on-disk formats.... FWIW, are the vserver foilks distributing modified versions of xfsprogs to support these changes? Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH 11/19] xfs: Convert to new freezing code, Dave Chinner |
|---|---|
| Next by Date: | Re: 1B files, slow file creation, only AG0 used, Dave Chinner |
| Previous by Thread: | Re: 3.2.9 and locking problem, Christoph Hellwig |
| Next by Thread: | 1B files, slow file creation, only AG0 used, Michael Spiegle |
| Indexes: | [Date] [Thread] [Top] [All Lists] |