> On Wed, Apr 04, 2001 at 01:32:09PM -0500, Steve Lord wrote:
> > Mount by label is all user space, there is no kernel code involved,
> Ah. There was talk on lkml about allowing the kernel to find the root
> fs by UUID or label, so I assumed all of this was done in the kernel.
> > beyond reading data out the block device interface to look for the
> > magic numbers. You need a version of the mount command with this
> > support in it for the code to function.
> > I am using mount-2.10m-6 but I have to admit I have not tried a mount
> > by label recently.
> 2.10r-4 from Wolverine here. Looking at the source, it appears XFS
> label support is already implemented. I'm investigating.
It is possible our labeling code got broken, have you tried the uuid
version of the mount? You can also use xfs_db directly to dump this
xfs_db -r /dev/xxxx