jnc at mercury.lcs.mit.edu
Tue Jun 7 08:03:29 CDT 2016
>> Unexpected trap to location 114
> Well, 0114 is the 'memory system error' vector - i.e. parity, or
> un-recoverable ECC error.
> I'm too lazy to read the 11/94 and J-11 manuals to see what the 11/94
> has in the way of registers that record memory issues
Finally found a moment to take a gander at that:
The KDJ11-E memory seems to be parity only, not ECC.
There's a Parity CSR at 17772100, which will tell us if a parity error was
detected in the main (on-board) memory. There's also a Memory System Error
Register at 17777744, but that's only there for backward compatability, and
I'm not sure it will tell us much.
It might also be good to look at the system CSR at 17777526, which will tell
us if the firmware has set the memory size to how much memory is actually on
There does not appear, alas, to be a register that indicates _where_ the
parity error happened. However, if it's happening regularly (and still, as
in 'problem disappeared in the process of diagnosis') we should be able
to work out where it is.
Although I'm still somewhat astonished that a persistent parity error wasn't
picked up by the self-test on power-on.
> From: Rod Smallwood
> If we look at what an 11/94 is then its a single board CPU ... in a
> three slot back plane. There after its UNIBUS So is there any thing we
> can put in the spare QBUS slot without doing any damage?
In theory, you should be able to plug most QBUS boards into those slots
(although if there are QBUS cards that barf in a Q/CD slot, one couldn't use
them - I can't recall if any such exist, though), but... no device board is
going to work properly since the two QBUS bus grant lines (BDMG and BIAK) will
be wired past those slots, directly from the CPU to the KTJ11.
On the 11/84 backplane, there are a couple of jumpers that _can_ send the bus
grants through the 'QBUS' slots on the 11/84 backplane (so _in theory_ one
should be able to plus QBUS devices into those slots, and have them function
correctly, when those jumpers are set appropriately, but we have yet to
confirm that), but I don't know if the /94 backplane has anything
similar. Alas, there are no 11/94 prints online (that I know of) to confirm
> I found not one but two RXV21 RX02 controllers. So I'll give them a try
> in the 11/83 and see what happens. I'll soon know if the RX02 is good
> or not.
Excellent! The more known good, working components we have to play with, the
More information about the cctech