Hi all, This patch series creates a couple of new ioctls to query and set the filesystem uuid of mounted filesystems. This is apparently desirous for some cloud providers who wish to be able to deploy pre-baked rootfs images to a machine, boot it, and have the client machine be able to change the label and uuid to reflect that it's now derivative of the original image. For ext4 this is pretty easy to do as all the pieces are already in place. For XFS this is a little more difficult because we need to get our house in order w.r.t. dependencies between the log uuid and filesystem superblock uuid, which means that this is really new ext4 functionality that I'd like to share with the other filesystems. I'm particularly curious to hear what people think about the FORCE_INCOMPAT flag. There are some circumstances (namely when the entire fs metadata is keyed to a certain uuid) where we can only change the uuid by turning on an incompat feature flag. The currently running kernel should be able to handle that just fine, but older kernels won't be able to mount the fs after that. We (XFS) normally don't do things like that, which is why I require positive affirmation from userspace that doing so is ok. This is an extraordinary way to destroy everything. Enjoy! Comments and questions are, as always, welcome. --D
