dump/dd/kernel bugs in 4.0.1 ??

[Eric Ho]

Has anyone experienced any kernel related bug(s) in 4.0.1 that may affect
dump & dd ?  It seems that they always got into a disk-wait state whenever
I do a large dump onto a scsi tape or whenever I need to do a remote
extract from a dataless client (during a remote install).  This is very
annoying.

This happens on my Sun-4/280S-24 under 4.0.1.  The whole net is running
4.0.1 and it seems that this doesn't happen under 4.0 (so much for the
upgrade then).  I didn't do anything fancy though -- I just extracted the
4.0.1 upgrade tape while I'm still at 4.0, then select all the options,
and then rebuild the kernel (discarding unneeded options from my kernel
config files of course).

I don't have any fancy hardware on my 4/280S either -- just a plain old
4/280S-8 plus an extra 16 meg of Clearpoint memory, a 1/4-inch scsi
cartridage tape drive from Sun, a 380meg Fuji Eagle and a 892meg Hitachi
hooking off a Xylogics 451 (all three from Sun).

The only other SunOS patch I did applied is the ufs patch that is supposed
to cure the 'itrunc' barfs that I get from time to time whenever the
filesystem is busy.

I didn't get this sort of thing before when I was using 4.0 (generic
kernel) though.

Can someone enlighten me as to what have I done wrong or may have missed ?

Eric Ho  
Princeton University
eho at confidence.princeton.edu