PDP-11/23 Debugging

Noel Chiappa jnc at mercury.lcs.mit.edu
Tue Nov 18 17:24:35 CST 2014

    > From: Ben Sinclair

    > I don't believe I have any discontinuity, but how would I tell?

Well, usually, it's caused by leaving a card or grant jumper out. I don't
think I've ever seen anything else cause it, although I suppose an
electronics failure on a card (failure to pass the grant) could also cause
it. But neither of those sounds like it's happening here (since you have only
one card that can interrupt...)

    > I think with my backplane (H9273) I'm supposed to put everything down
    > the left side, with the BVD11 at the bottom.

Right. The 9273 is a so-called Q/CD backplane, which means the QBUS (which
only takes one dual slot) runs down the left side (when looking at it from
the side the boards plug into); all the RHS slots are wired for C-D

    > Currently I have starting from the top left and going down the left
    > side: M8186, M8044, M8043, then the BVD at the very bottom.

That sounds right. And the chances of the 8044 not passing the grant are
zilch, since it has very substantial grant jumpers. So unless your backplane
is damaged (again, very unlikely), or the grant drive on the CPU card is bad
(possible, I suppose), the grant should be getting to the DLV11-J (I'm
assuming that test uses the console DL11 to test interrupts - can't think what
else it could possibly be using :-).

So, if the CPU checks OK, maybe the DLV11-J has a problem with interrupts? Is
there a DL11 diag you can run on it? (If not, I can whip up a simple test to
check it out.)

    > I have another memory board, but I know it's bad, so I can't try that!

Probably the one you have passes the memory diag, though, right?

If it's not an issue with the DL11 interrupting, it sounds like we may have
to dig into that diag to find out what it thinks it's doing when it blows up.
If I can get my tool-chain working (currently downloading the abs-loader
isn't working right somehow), I'll see if I can figure out how to load it,
and DDT along with it, and poke into it.

Locating the fiche would be another way to go...


More information about the cctech mailing list