At 13:53 Uhr -0500 29.10.2003, Robertdkeys@aol.com wrote:
>Are there issues in SS20 class machines in NetBSD 1.6.1
>or 1.6-current that would cause instability in the scsi bus,
>particularly as to reading of cdroms?  On most of my SS20
>(and even SS5) machines, I am having problems in 1.6.1 or
>current in reading the cdroms due to timeout errors and the
>bus resetting.  I don't see this in 1.5.2, for example, or slowlaris
>or OpenBSD.  It seems to suggest some possible scsi bus
>timing issues, but I can't quite put my finger on it exactly.
>All cdroms are standard sun things native to SS5/SS20 boxes.
>At one time I was thinking it was due to termination issues,
>but now, I am beginning to think not.
>
>If anyone knows of such problems, or solutions, kindly
>holler my way.

BTST. There was a time when NetBSD (1.5 then,
http://mail-index.netbsd.org/port-sparc/2000/09/21/0001.html) was plain
unusable  on my ss10 which would run Solaris 7 + 8 in various patchlevels
just fine. While the problem has become much better, it still happens
occasionally that the internal scsi bus gets into a state where it produces
an endless stream of error messages, and I have to reboot the box. I have
found that attaching an external device, properly terminated, helps here -
even if it is powered off.

I suspect the NetBSD esp(4) code sets up the scsi controller in a way that
makes it more sensitive to noise on the bus than with other OSes.

	hauke

--
/~\  The ASCII Ribbon Campaign
\ /    No HTML/RTF in email
 X     No Word docs in email
/ \  Respect for open standards