PDP-11/23 Debugging

Ben Sinclair bensinc at gmail.com
Wed Nov 19 13:36:20 CST 2014


Thanks! I'll try that out tonight and see what happens. My loader just
takes address/value pairs. It basically just automates me typing them at
the ODT console.

I don't have a memory test yet, but Glen just pointed out a test that I
should be able to run. I should have noticed that earlier since I've been
using that disk image forever.

On Wed, Nov 19, 2014 at 12:39 PM, Noel Chiappa <jnc at mercury.lcs.mit.edu>
wrote:

>     > From: Ben Sinclair
>
>     > Thanks a lot for your help Noel!
>
> Sure, glad to help. I'm at just about this stage with my 11/23's, so it's
> no
> biggie.
>
>     > I did write a Ruby program to enter programs in from files via ODT,
> so
>     > I can probably run your test that way, assuming it's a set of
> addresses
>     > and values.
>
> I can do that. Here you go (in Unix assembler, which is somewhat different
> from the DEC one in some syntax details, notably '/' for comment, not ';'):
>
>         br4 = 200
>
>         stack = 2000
>
>         conregs = 177560
>
>         rcsr = 0
>         rbuf = 2
>         tcsr = 4
>         tbuf = 6
>
>         rint = 100
>
>         . = 60^.
>         conr; br4                       / 1020 200
>
>         . = 1000^.
>
> test:   mov     $stack, sp              / 12706 2000
>         mov     $conregs, r1            / 12701 177560
> /       mov     $rint, rcsr(r1)         / 12761 100 0
>         mov     $rint, (r1)             / 12711 100
> 1:      wait                            / 1
>         br      1b                      / 776
>
> conr:   movb    rbuf(r1), tbuf(r1)      / 116161 2 6
>         rti                             / 2
>
> (The numbers on the right are hand-assembled octal, but I did put it
> through the assembler to make sure I'd gotten it right. The commented
> out instruction is clearer, but adds one word to the required type-in.)
>
> Here it is in address/location form:
>
> 60      1020
> 62      200
>
> 1000    12706
> 1002    2000
> 1004    12701
> 1006    177560
> 1010    12711
> 1012    100
> 1014    1
> 1016    776
> 1020    116161
> 1022    2
> 1024    6
> 1026    2
>
> Start with "1000G" to ODT. It should echo everything you type, using
> interrupts from the console DL11 to do so.
>
> I did try this on my machine to debug it, and fixed one minor thinko (I had
> used "rts pc" instead of "rti"). After that, worked like it says on the
> tin.
>
> My machine is (at the moment) basically identical config to yours: an
> 11/23,
> a memory card (NatSemi), and a DLV11-J. (No terminator/ROM, it's in a Sigma
> box which has termination built in.)
>
>
>     > That PDF is the one I was looking at
>
> OK.
>
>     > though it looks like there are more diagnostics available for the
> 23+,
>     > so that might not be completely equivalent to the test available for
>     > the 23.
>
> Yes, that was my sense too.
>
>
>     > I'm hoping a memory test shows that I have some problems there, which
>     > is potentially the easiest thing to fix.
>
> Do you have a memory test, yet? I am planning on whipping up one of those
> too, for my own use, although it will be more elaborate than that console
> interrupt test.
>
> If you can use octal dumps to load, I can produce that form pretty easily:
> let me know what syntax your loader takes.
>
>     > Someone told me a while back that the M8044 memory tends to be fail,
> so
>     > maybe I should look for something else.
>
> Interesting; I have a whole stack, and the first one I tried seemed not
> to work.
>
>         Noel
>



-- 
Ben Sinclair
ben at bensinclair.com


More information about the cctalk mailing list