Anyway, I fatfingered my remounting of the swap partition on /dev/hdb, and
entered /dev/hdb3 rather than /dev/hdb4. /dev/hdb3 is my /usr partition, and
was already mounted. 'swapon' complained, of course, saying "hbd3: invalid
argument," and aborted. I mounted /dev/hdb4 as swap, and was promptly hit
with this:
Ext2-fs panic (device 03:43):read_inode_bitmap: cannot read inode bitmap.
block_group=108, inode_bitmap=884744.
No entry in the log, nothing. I also had similar problems with kernel panics
when I changed the partitioning on my drives, altering some ext2 partitions,
and rebooting to reload the partition table before running mke2fs, and
forgot to remove the partitions from the /etc/fstab file. In that situation,
it claims to try to read beyond the end of the device. I had to get a
bootdisk, mount the root fs, and hack up /etc/fstab manually.
I understand the second error ("attempt to read beyond end of device"), but
I'm curious as to why this is an offence punishable by panic. The first
error, though, is something that should not happen.
Normally, I'd go in, try to "fix" the problem, and then send out a patch.
I'm not doing so because this machine is incredibly slow (5-7 hours for a
kernel compile), and I'm not sure if this is an intended behavior.