Subject: Re: 3100 woes [was Re: libdl?]
To: Ted Lemon <firstname.lastname@example.org>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
Date: 04/25/1997 11:35:14
Bill Studenmund writes:
>Wasn't there some mention of a bug in that one of the scsi chips (on an
>alpha?) was getting inited to the wrong clock frequency? Could there be
>some problem like that floating around?
There *were*, but it's a different model of decstation, and a
different SCSI chip.
I duplicated a 53c94 clock-speed initialization bug from the Alpha
port into the 5000/1xx and 5000/2x models. That bug is fixed. I
don't think it has anything to do with the DEC-specific SCSI/DSSI asic
(SII) in the 3100/2100 (aka pmax).
If the 5000s are anything to go by, then using internal disks with the
NetBSD/pmax drivers is probably a mistake. Even if a system worked
under Ultrix. The Ultrix drivers can recover from poor termination
problems that cause the NetBSD/pmax drivers to just give up.
The 4.4bsd/pmax scsi drivers -- especially the sii driver -- have a
long history of unpleasant bugs. The publically-available
documentation on the proprietary SII chip is good, but not *quite*
detailed enough to write a really good driver. I think it's safer for
the driver to not try and recover from error conditions rather than
(for example) silently corrupting data.
I do agree that hanging the entire machine is another story altogether:).