I would like to use my VAX station currently running OpenVMS 7.3 as an
hypervisor to install and run inside it a little instance of NetBSD/vax...

Is there any ( free ) virtualization software for OpenVMS like the kvm
or qemu for Linux, or does the VAX processor lack capabilities of this kind ?

Best regards

On Saturday, April 1, 2023 at 12:18:27 PM UTC-7, I Kea wrote:
> I would like to use my VAX station currently running OpenVMS 7.3 as an
> hypervisor to install and run inside it a little instance of NetBSD/vax....
> Is there any ( free ) virtualization software for OpenVMS like the kvm
> or qemu for Linux, or does the VAX processor lack capabilities of this kind ?
There are rules on virtualizable architectures, but I am not sure about VAX..
The important rule is that you have to be able to not tell if you are in user
or supervisor mode. Virtual machines run in user mode, where the
hypervisor emulates any privileged instructions. There are some where
the mode bits are visible, and so technically can't be truly virtualizable.

In any case, I don't know that anyone has written one for VAX.

On 02/04/2023 00:15, gah4 wrote:
> On Saturday, April 1, 2023 at 12:18:27 PM UTC-7, I Kea wrote:
>> I would like to use my VAX station currently running OpenVMS 7.3 as an
>> hypervisor to install and run inside it a little instance of NetBSD/vax...
>
>> Is there any ( free ) virtualization software for OpenVMS like the kvm
>> or qemu for Linux, or does the VAX processor lack capabilities of this kind ?
>
> There are rules on virtualizable architectures, but I am not sure about VAX.
> The important rule is that you have to be able to not tell if you are in user
> or supervisor mode. Virtual machines run in user mode, where the
> hypervisor emulates any privileged instructions. There are some where
> the mode bits are visible, and so technically can't be truly virtualizable.
>
> In any case, I don't know that anyone has written one for VAX.
>
>

Is there nota VAX version of NetBSD? Could you install that on an
alternate disk, then boot that?

--
Chris

On 01/04/2023 20:18, I Kea wrote:
>
> I would like to use my VAX station currently running OpenVMS 7.3 as an
> hypervisor to install and run inside it a little instance of NetBSD/vax...
>
> Is there any ( free ) virtualization software for OpenVMS like the kvm
> or qemu for Linux, or does the VAX processor lack capabilities of this kind ?

Well QEMU and KVM are totally different animals.

KVM is a hypervisor and only allows you to run "like for like" code. It
relies on extensions in modern Intel and AMD CPUs that allow a
"hypervisor" to run native code in a "sandbox" and the Hypervisor only
has to deal with a limited number of exceptions and so the Virtual
Machines can deliver near-native performance.

QEMU is a simulator or emulator, and simulates every instruction in
software. This means performance is no where near native performance. I
am not familiar with QEMU but I suspect it may need feature that VMS 7.x
does not provide.

If you are desperate to try this then I suggest the SIMH simulators may
be a better option, but performance is likely to be a dog...

>
> Best regards

Dave

In article <23caf9c0-a773-4476-b821-493ed2989960n@googlegroups.com>,
gah4@u.washington.edu (gah4) wrote:

> There are rules on virtualizable architectures, but I am not sure
> about VAX. . . .
> In any case, I don't know that anyone has written one for VAX.

It appears that DEC developed one called VMM in the 1980s, but it was
abandoned before being released.

<https://en.wikipedia.org/wiki/VAX#Operating_systems>

John

On 4/1/2023 3:18 PM, I Kea wrote:
>
> I would like to use my VAX station currently running OpenVMS 7.3 as an
> hypervisor to install and run inside it a little instance of NetBSD/vax...

Doesn't SIMH still build on VMS?

>
> Is there any ( free ) virtualization software for OpenVMS like the kvm
> or qemu for Linux, or does the VAX processor lack capabilities of this kind ?
>
> Best regards

bill

it would have been really nice to have a "virtual box" for OpenVMS, it would
have allowed not to give up running its basic operating system, OpenVMS, and
at the same time to be able to run more recent Linux versions such as
NetBSD/vax 9.3 in the background. A VAX with only 32 MB of memory, maybe
even less, could easily take charge of one or more virtual machines inside it,
as far as CPU consumption is concerned I noticed that my VAX with OpenVMS
just loaded consumes 0% of cpu (according to MONITOR utility).

A test that I would like to do in the future, if I have time and just for
education and test purpose, is the reverse, and that is to try to install
OpenVMS inside a VAX where NetBSD/vax is already installed, but this was not the
original topic.

Naturally it is also possible to have a VAX with 2 internal disks, on one you
can install OpenVMS and on the other NetBSD/vax, from the prompt >>> one decides
which of the two RZ disks to boot. Once NetBSD/vax is started it may be possible
to browse (mount) the contents of the other disk where OpenVMS is installed
using the ODS-2 utilities, I don't know if they have been ported for NetBSD/vax
as well. And I don't know if the same thing can be done in reverse, ie being able
to read the contents of the other disk where NetBSD/vax is installed from the
other environment where instead we have just started with OpenVMS.

In any case, the fact that DEC had already thought about virtual machine concepts
in the 1980s ( VMM ) is further proof that they were already far ahead of
everyone at the time.

On 2023-04-03, I Kea <keai73942@gmail.com> wrote:
>
> In any case, the fact that DEC had already thought about virtual machine concepts
> in the 1980s ( VMM ) is further proof that they were already far ahead of
> everyone at the time.

IBM beat them, and everyone else, to this big time.

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

On 4/3/2023 1:04 AM, I Kea wrote:
>
>
> it would have been really nice to have a "virtual box" for OpenVMS, it would
> have allowed not to give up running its basic operating system, OpenVMS, and
> at the same time to be able to run more recent Linux versions such as
> NetBSD/vax 9.3 in the background.

NetBSD is not Linux and has no connection to it at all.

bill

On Sun, 02 Apr 2023 22:04:39 -0700, I Kea wrote:

> In any case, the fact that DEC had already thought about virtual machine
> concepts in the 1980s ( VMM ) is further proof that they were already
> far ahead of everyone at the time.

I think the IBM paper is dated around 1965.

On 03/04/2023 19:13, Bob Eager wrote:
> On Sun, 02 Apr 2023 22:04:39 -0700, I Kea wrote:
>
>> In any case, the fact that DEC had already thought about virtual machine
>> concepts in the 1980s ( VMM ) is further proof that they were already
>> far ahead of everyone at the time.
>

Much much older.... e.g.

https://en.wikipedia.org/wiki/Popek_and_Goldberg_virtualization_requirements

> I think the IBM paper is dated around 1965.

I think "virtual machine" concepts go back to the start of computing
history. In hist paper "On Computable Numbers" Turing talks about
"Turing Machines" and "A Universal Turing Machine" which can be any of
the other Turing Machines.

The first program to run on an Electronic Computer was on ENIAC in 1948.
To do this ENIAC was programmed using the plug boards to emulate an
EDVAC type machine. As there wasn't enough RAM the program had to be
stored in switch banks, so ROM not RAM...

... and as a "virtual machines" is en effect just emulating the same
machine as you are running on, on may argue VMs are as old as computing...

... I think the first "Hypervisor"s were CP40 and CP67 which ran on the
IBM 360/40 and 360/67 respectively. These date from around 1970.

In 1970 IBM announced the 370 and in 1972 VM/370 which comprised a
hypervisor (VM aka CP) and CMS a simple single user operating system.
Each user has their own virtual machine into which they load CMS.

So some IBM users have been computing virtually since 1970, so rather
than being ahead of the game, DEC was a long way behind.

Dave

On Monday, April 3, 2023 at 11:13:26 AM UTC-7, Bob Eager wrote:
(snip)
> I think the IBM paper is dated around 1965.

While designing the virtual addressing system for the 360/67, IBM had a specially
modified 360/40 to practice on. That ran CP-40, and then led to CP-67 on
the 360/67. That was in 1967, and led to CP/CMS.