Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!tut.cis.ohio-state.edu!bloom-beacon!think!husc6!genrad!decvax!rsp
From: rsp@decvax.dec.com (Ricky Palmer - (603)881-0370 - ZK3-3/T74)
Newsgroups: comp.sys.dec
Subject: Re: DS 3100 serial ports and other
Message-ID: <5170@decvax.dec.com>
Date: 19 Jun 89 20:57:23 GMT
References: <2390@blake.acs.washington.edu>
Sender: news@decvax.dec.com
Lines: 22

-------
I have seen problems with Wren V's on the DS3100.
The problem is that the Wren does a seek, disconnects,
and never reselects the initiator (in this case the 3100).
Various people I have talked with are looking into getting
a microcode fix from CDC. At one point I was told that
microcode 5466 worked correctly with "fast machines"
but this does not appear to be accurate. The problem appears
quite readily when using the Wren in synchronous mode with
the cache enable bit turned on (maximum performance mode).
I was able to cause it to occur by attempting to execute
a large binary off the Wren drive such as "dxpsview" (the
Postscript previewer). The binary has a very large text and
data space and leads to some nice seek requests at the drive.
The next thing you know the "dxpsview" process is hung in
a pagein state waiting for the disk and the disk never returns.


---   Ricky Palmer	 Ultrix Advanced Development
---   Digital Equipment Corporation
---   Nashua, New Hampshire
---   ... One of the fathers of PMAX ...