I just fixed a bug in fsck_ffs that would cause it, when finding a
problem and clearing an inode, to clear the next inode as well.
I introduced the bug by accidentally reversing an if-else in
the UFS2 merge.

If the wiped inode happened to allocated, this would cause symptoms
like 'bad filedescriptor' errors, when the file was next accessed.
This may have gone unnoticed for a while, since fsck would have
marked the filesystems clean.

Please update sbin/fsck_ffs, with fsck.h revision 1.34. If you
never had an fsck run that cleared inodes since the UFS commit,
you won't have any problems. Otherwise, you might want to force
a filesystem check with the new fsck_ffs to make sure there are
no erroneously cleared inodes out there.

- Frank

-- 
Frank van der Linden                                   fvdl@wasabisystems.com
-----------------------http://www.wasabisystems.com/--------------------------
NetBSD development                                     Embedded, Storage, other