[Info-vax] Kittson question

[cmexec at gmail.com]

On Wednesday, January 7, 2015 2:54:47 AM UTC-5, David Froble wrote:

> The VAX instruction set is just that, the well defined HW instructions. 
>   Now, if someone is implementing a HW emulator, there would be no sense 
> in emulating just those instructions a MVII, for example, would support. 
>   It would be much better judgement to emulate the full set of VAX 
> instructions.  One emulator that can do everything a VAX can do.
> 
> Now, the HW emulator can have some data identifying some particular 
> model for licensing and other reasons.  Without those special reasons, a 
> VAX is a VAX is a VAX.  From a design perspective, there can be only one.

A VAX is more than just a CPU.  The kernel is the only software that normally
deals with issues such as different I/O busses and different privileged registers,
but they do exists and the kernel does deal with them.

A good example:  the VAX 11/780 privileged register set has a TOY (time of
year) register.  The MV II does not, it emulates the TOY register in software.
Yet, through hardware or emulation, user mode code can execute all user mode
instructions.  And the MV II specific kernel code can execute the privileged register
instructions for the privileged registers that it does have.

I had to explain this once to a customer who bought a new VAXStation that wouldn't
boot the previous years' version of VMS.  Last year's kernel didn't even know what I/O
busses were on a model it that didn't exist when it was written.

The new model was quite similar to an older model, and I suspect I could have
patched the older kernel to recognize it, but at that time I didn't even have current
source listings and wasn't going to try it without.