On Sun, Oct 09, 2011 at 07:52:40PM -0400, Christoph Hellwig wrote: > On Mon, Oct 10, 2011 at 10:11:46AM +1100, Dave Chinner wrote: > > From: Dave Chinner <dchinner@xxxxxxxxxx> > > > > Because repair uses direct IO, it cannot do IO smaller than a sector > > on the underlying device. When repairing a filesystem image, the > > filesystem hosting the image may have a sector size larger than the > > sector size of the image, and so single image sector reads and > > writes will fail. > > Does it? Given that the loop driver uses buffered I/O that would be > unexpected behaviour. I'll try to reproduce it as soon as my 4k > disk is available again. My 3TB, RAID6 based image storage filesystem has a 4k sector size, and I can't run xfs_repair on an image file that has smaller sector sizes than 4k without this mod. In that case, xfs_repair is doing direct IO on a file, not a device, and the file is on the 4k sector sized filesytem. I've been carrying this patch for a while. So it's definitely a real problem. Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
