Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 beta 3/9/83; site sdcrdcf.UUCP
Path: utzoo!linus!security!genrad!decvax!harpo!seismo!hao!hplabs!sdcrdcf!mike
From: mike@sdcrdcf.UUCP (Michael Williams)
Newsgroups: net.bugs.2bsd
Subject: Seen this before ?
Message-ID: <739@sdcrdcf.UUCP>
Date: Wed, 14-Dec-83 13:44:26 EST
Article-I.D.: sdcrdcf.739
Posted: Wed Dec 14 13:44:26 1983
Date-Received: Sat, 17-Dec-83 01:07:39 EST
Reply-To: mike@sdcrdcf.UUCP (Michael Williams)
Organization: System Development Corporation, Santa Monica
Lines: 37
Keywords: disk driver V7

I am running V7 kernel with Berkeley utilities, until my 2.9
tape arrives.  I have two disk controllers, the 2nd is an SI
9900, with two 9766 300 Mbyte drives, keyed 1 and 2.  (The first
is the usual with two RP04's, switched to 0 & 1).  The SI drives
are emulating RP04's, which I believe means that they only read/write
from even tracks (thus 820 physical, 410 logical).  
Problems:
	1) When I run icheck (yes, remember that old goody ?)
	   on the raw device (e.g. rzp2a (I'll explain the zp
	   in a minute) it gives me "bad mode: 162"
				    "bad mode: 163"
	   etc.  It always starts at block 162, and doesn't
	   complain about every eighth block (ie 161, 169, 177
	   etc are good).  (Remember 8 inodes per block ?)
	   Icheck doesn't complain on the block device (zp2a).
	2) When I run mkfs on this drive, the system "hangs",
	   regardless of the partition I am trying to create.
	   (I had been playing around with zp_sizes, trying to
	   get three or four filesystems of the same size).
	   Throwing the reset switch on the controller at this
	   point makes the read/write flash for a second.  The
	   system is still hung.  Throwing the switch again
	   frees the system and causes mkfs to go away.

The zp is a cludge, I used the slots in [bc]devsw from the rp
driver and filled them with "zp".  Zp.c is a copy of hp.c
with a different {HP,ZP}ADDR for the second controller, and at
times different sizes tables. (oh yes, all the variables are changed
from hpcsr1 to something like zp1CntrlStatReg, and they are unique).
I hope to figure this out today, but I thought I'd post a note to see
if you'd seen it before.

Mike Williams

{allegra,cbosgd,hplabs,ihnp4,trw-unix}!sdcrdcf!mike
(805) 987-6811 x4528