> scsibus0 at esp0
> esp0 targ 1 lun 0: <QUANTUM, P105SS 910-10-94, 8.9> SCSI1 0/direct fixed
> sd0 at scsibus0: 100MB, 1019 cyl, 6 head, 33 sec, 512 bytes/sec
> error... error... error... error... error... error... error... error... error...
>  error... error... error... error... error... error... error... error... error..
> . error... error... error... error... error... error... error... error... error.
> .. error... error... error... error... error... error... error... error... error
> ... error... error... <I hit L1-a here>

I had a problem like this about a week ago.  I kicked a "possible
patch" that fixed about half of the problem to charles & pk then,
and i believe that charles committed it (or at least, a variant of it)
this morning.

for me, it fixed half the problem.  However, the other half was
DMA-related, and i don't think it would show up on the sparc.  (this
was on an alpha.)

just fyi: i ran into the problem with an IBM lightning disk.

> Interestingly enough, the third disk *does* work with the old SCSI 
> system in NetBSD 1.0.   Maybe we can borrow some ideas from the old
> driver to get the new one up to speed?   I currently don't know enough 
> about SCSI to debug this properly.

it's fairly well known that the current sparc driver leaves a _lot_ to
be desired.  last i heard, charles was in the process of rewriting it,
but was in "doc wait."  however, either the docs are on their way to
him by now, or he already has them...  8-)  I have no idea when it
will be fit for public consumption, however.


chris