like them to be as simple as possible. Every xattr inode would have a
flag set that says I'm an xattr inode!, and it shouldn't be all that
hard on fsck.
Yeah, that's what ocfs2 does.
Maybe it'll be easier because btrfs can just toss inodes (leafs with
inodes, I suppose) whose checksums are invalid. fsck.{ext*,ocfs2}
gets to play the fun game of trying to figure out which of a set of
conflicting fields is actually the one that isn't corrupt. It's not
great.
(rule of thumb: every time fsck has to ask an interactive question,
instead of being able to tell from disk structures, there's a design
failure.)
So maybe it doesn't matter. Still, it seems sloppy.
Maybe I could roll a patch some day to demonstrate what I'm talking
about instead of just whining in email. That'd be something!
- z
_______________________________________________
Btrfs-devel mailing list
[email protected]
http://oss.oracle.com/mailman/listinfo/btrfs-devel
