[Simh] DZ11 vs DZV/DZQ11

[Mark Pizzolato]

On Monday, April 9, 2018 at 12:13 PM, Robert Armstrong wrote:
> >…. single PDP11 simulator can simulate different PDP11 models …
>
>  Yes, and the UNIBUS and QBUS peripherals don’t always map 
> exactly in functionality.  Actually I’m surprised this situation 
> hasn’t come up before with other devices.

The only other device which I can think of that has a different 
number of ports/units on the Qbus version vs Unibus is the VH
and indeed this bus change problem will also exist there for the 
PDP11 simulator.

> >Feel free not to use the extra ports that are provided free of charge.  ☺
>
>  You’re right – for the most part it doesn’t matter.   But, … if 
> you’re using the DZV/DZQ simulation with a TELNET port (which 
> I imagine is what most people do) then I don’t think you can 
> stop simh from accepting a connection for lines 4..7 if the first 
> four lines are already busy.  

True, and I see that as the problem/bug you are reporting and 
which needs fixing.

> And if you want to simulate a QBUS machine with 8 lines and 
> two DVZ/DZQ cards, then what the OS actually sees is all eight 
> connections coming in on the first card and none on the second 
> card.  I haven’t tried it yet, but I don’t know what the various -11 
> OSes will do when they get activity on lines that can’t physically 
> exist.  Probably they use the same driver for both the DZ and 
> DZV so they won’t care, but I don’t know that.

Well, since the drivers are common, and the driver knows that 
it is on a Qbus machine, it will only respond to and/or use 4 lines
per controller.  Lines 4-7, 12-15, etc. won't be recognized by the 
OS.  Incoming telnet connections won't behave reasonably since 
the OS won't see them properly.

> >  What should happen when an odd number of DZV11’s are 
> > configured and the CPU type is changed …
>
> Well, ideally there should be an error or warning message, but 
> I imagine that’s difficult to implement.

Difficult or not, when someone changes the CPU type all devices 
get to see that this has happened and some devices are 
automatically disabled if a variant they support isn't available on 
the new bus type.

I'm going to have the number of available lines round up to the
appropriate number that fits.  So, if lines is 4 and a switch is 
made from Qbus to Unibus, then the DZ will be enhanced to 
have 8 lines.

>  For me, I’d be happy if simh was changed to accept 
> “SET LINES=…” in multiples of 4 rather than multiples of 8.  
> It’s true that this would possibly allow you to create invalid 
> DZ configurations, but in my mind the possibility of deliberately 
> defining an invalid configuration is less annoying than the 
> impossibility of defining a valid configuration.

Do you see a down side to the rounding up so the number of 
lines fits?

- Mark