GEOM topology still+more output/USB Crashes

Hi, (More info)

Despite some digging in the source, I didn't yet fix my geom topology
bug / slice issue yet.  Since I'm low on time to mess around and want to
move on to crypto stuff... please assist if you have time.

I'm writing an architecture manual and other projects at the moment
so it would be nice to have this workstation/server back as other
work depends on it.  My laptop doing double duty is grinding to a
halt/swapping excessively.

I'd love to dabble with kernel internals but there comes a point
when I have to consider what is just basic needs for a production machine
vs. dev workstation.

I did appreciate the chance to review GEOM sources and the GEOM
tutorial but that's about all I've done so far.  I assume debugging
will take half a day unless someone knows exactly where to point me.

 -

I got it to boot using bootloader from floppy but didn't ever manage
to get it back to bootable state from ad0.

I did change the mbr to be at offset 0 instead of 63 which drastically
reduced the radical/loopy slice entries.

It looks like the difference between offset of 0 and offset of 63
causing something wrong with the hashing/MD5 function avoiding duplicate
specification.

This wasn't occurring previously even though the disk layout on this
machine may be dated/dumb on my part as admin.

See:
 -  http://afields.ca/bsd/messed-geom.ps		(dot geom tree)
 -  http://afields.ca/bsd/messed-geom.xml		(sysctl -b kern.geom.confxml )
 -  http://afields.ca/bsd/afields-5.4-p6-dmesg.txt	(dmesg)


FreeBSD 5.4-RELEASE-p6 #0: Tue Aug 16 10:34:25 EDT 2005
..
WARNING: Expected rawoffset 0, found 63
WARNING: Expected rawoffset 0, found 63
...

Also with this kernel: USB crashes both on disconnect of flash
device w/ volumes mounted (on umount/at shutdown) -- often I have
a loose cable and when usb keyboard is removed/KVM cycled.

Thanks,
-- 
Allan Fields