Physical disk size 65536 blocks for an 8G drive

[Jim Hickstein <jxh@mirapoint.com>]

Sorry to post what amounts to a support question on -dev, but it did not
get an answer on -user.  I'm afraid it's too arcane, so I appeal to you.

I have a PBG3 Wallstreet and recently bought a new 8GB drive to put in
it, an IBM DYLA-28100.  MacOS 8.6, update the drivers, so far, so good. 
I managed to get pdisk to make partitions (after umounting /mnt/cdrom
from /dev/hda to silence a warning in writing the partition table). 
MacOS runs fine, with volumes in hda10 (5GB hfs+) and hda11 (600MB hfs),
_above_ Linux.  But Linux barfs mounting / on /dev/hda7 (and even worse
for /usr on /hda9), saying that the superblock reports a size of
so-and-so (~131000 for /) but the _physical size of the disk_ is 65536 blocks.

Evidently something is reporting a block count of 0xffff, and it's
getting incremented to tell me the "length".  Yet, the drive probes
correctly (apparently) and reports a size of 7145MB or so.

fsck is the one telling me this, for /dev/hda7.  For /dev/hda9, it
really complains, saying that inode 3 is corrupt, etc.  Part of hda7 is
below 0xffff blocks, but all of hda9 is above it.

Any ideas?  This is 2.2.10 off the Q3 distribution.

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/