Power cycling during hardware detection on bootup causes kudzu to lose its stored config settings and redetect all hardware as 'newly installed' on the next boot. This happens on both ext2 and ext3 filesystems. Whether this behavior is correct, I'm not sure, but I'm inclined to not think so.
The resulting filesystem check *probably* removed /etc/sysconfig/hwconf, in which case there's not much that can be done.
This first happened on the latest ext3fs; I later reproduced it on ext2. Do you suppose a journal replay would result in the same recovered fs state as an e2fsck run? I'd tend to think they're not *quite* the same, since journaling can repair things that e2fsck cannot. In any case, I guess redetecting everything is better than trying to recover possibly stale configuration data.