On Thu, Mar 29, 2012 at 03:12:03PM -0400, Christoph Hellwig wrote: > The actual fix looks good: > > > Reviewed-by: Christoph Hellwig <hch@xxxxxx> > > But shou;dn't we move the setting of b_error to zero into common buffer > code instead of adding it to the callers? We generally do - in xfs_buf_iodone_callbacks() - which covers pretty much all cached buffer IO. However, for uncached buffers, it is only zeroed in xfs_buf_alloc(). For single use buffers, this is fine, but for multiple use buffers like the log uses during recovery for log IO, it doesn't get reset anywhere because the buffer doesn't go anywhere near common buffer code between IOs. Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
