On Tue, 2016-09-06 at 18:20 +0100, Graham Cobb wrote: > they know the UUID of the subvolume?
Unfortunately, btrfs seems to be pretty problematic when anyone knows your UUIDs... Look for my thread "attacking btrfs filesystems via UUID collisions?" in the list archives. From accidental corruptions to intentional attacks, a lot seemed to be quite possible... and I do not even talk about general attacks that any such systems that may do auto-assembly/auto-rebuilds (e.g. of multi- device containers/fs) or similar likely suffer from (I've also discussed this in the thread back then at some of the later mails). Don't think anything has happened in that area yet (or ever?)... at least I wouldn't have noticed that these security showstoppers would have got fixed. Cheers, Chris.
smime.p7s
Description: S/MIME cryptographic signature
