I'm pretty sure this is not softupdates related. It could either be bad RAM
or a bad disk block, there is no way entryoffsetinblock could be 21209,
the block size is only 8192. And you need over 2000 files to fill the
directory to i_offset == 37593, assuming an average file name length of
10 bytes.
I've got a crash in ufs_lookup, and I'm trying to assess responsibility. I
use SoftUpdates on all drives and _MAY_ have some bad RAM. This crash was
during a make -j4 -DNOCLEAN world, and maybe it may be due to SoftUpdates not
completely having finished the dir, but I'm thinking it could've been
