On Tue, Aug 27, 2013 at 11:39:37AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@xxxxxxxxxx>
> CRC enabled filesystems fail log recovery with 100% reliability on
> xfstests xfs/085 with the following failure:
Unfortunately I have not been able to hit this one... not sure why.
> XFS (vdb): Mounting Filesystem
> XFS (vdb): Starting recovery (logdev: internal)
> XFS (vdb): Corruption detected. Unmount and run xfs_repair
> XFS (vdb): bad inode magic/vsn daddr 144 #0 (magic=0)
> XFS: Assertion failed: 0, file: fs/xfs/xfs_inode_buf.c, line: 95
> The problem is that the inode buffer has not been recovered before
> the readahead on the inode buffer is issued. The checkpoint being
> recovered actually allocates the inode chunk we are doing readahead
> from, so what comes from disk during readahead is essentially
> random and the verifier barfs on it.
> This inode buffer readahead problem affects non-crc filesystems,
> too, but xfstests does not trigger it at all on such
> Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
I've been mulling this one over for a bit, and I'm not quite sure this
My feeling is that in light of commit 9222a9cf, if we do take part of a
buffer back in time, the write verifier should fail. I think for a v2
inode the read and write verifiers should both be disabled for the
duration of recovery. For v3 inodes, I suspect the current situation
where we do use write verifiers is broken in the same way, at least
until we pull in 'xfs: prevent transient corrupt states during log
recovery', which, as you say, won't fix the problem for the v2 inode.
I'll pull this in and send a patch to that effect.
Reviewed-by: Ben Myers <bpm@xxxxxxx>