Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: Notesfiles $Revision: 1.6.2.17 $; site uiucdcs.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxj!ihnp4!inuxc!pur-ee!uiucdcs!irwin
From: irwin@uiucdcs.UUCP
Newsgroups: net.unix-wizards
Subject: Re: 780 system hanging
Message-ID: <13700073@uiucdcs.UUCP>
Date: Tue, 23-Oct-84 17:12:00 EDT
Article-I.D.: uiucdcs.13700073
Posted: Tue Oct 23 17:12:00 1984
Date-Received: Thu, 25-Oct-84 03:18:00 EDT
References: <13205@sri-arpa.UUCP>
Lines: 26
Nf-ID: #R:sri-arpa:-1320500:uiucdcs:13700073:000:1275
Nf-From: uiucdcs!irwin    Oct 23 16:12:00 1984

The MS780C memory is the 4 meg (256k X 16 boards). You do not say how
much you have, and if the two faulty boards are in the last two slots.
The memory has to be continuous, so they need to be in the last two
slots (not counting terminator) if you disable them.

The probable cause of the machine hanging is that the controller is
having compound errors. The memory is error correcting type, but if
the controller is correcting an error and gets a second error at the
same time it gets confused and hangs the buss. This happened on our
780. We were having single bit errors (correctable) so ignored them.
When the errors got to confusing the controller, we had to get the bad
boards repaired.

If you have the 780 on DEC maintenance and if the diagnostic floppies
are on site, you can boot the console floppy, enter micro diag #2 and
type T on the console, it will test memory and point out the bad boards.

When the machine hangs, if you do the ^p and H to get the >>> prompt,
you can do a E 20002000/N:2 which is the examine of the memory controller
registers at TR1.

If you have the docs on the controller and decipher the printout you get
on the console, you may determine what the controller is doing.

If the above is not enough help, mail me at pur-ee!uiucdcs!irwin.