history is hard

[Warner Losh]

On Sun, May 24, 2020 at 3:47 PM Peter Coghlan via cctalk <
cctalk at classiccmp.org> wrote:

> On Sun, 24 May 2020 at 15:18:34 -0600, Warner Losh wrote:
> >>
> >> >
> >> > The topic for my talk next week. Unix had virtualization in 74. The
> >> second
> >> > Unix port ran under OS/360's VM in 78.
> >> >
> >>
> >> What do you mean by "Unix had virtualization"?
> >>
> >
> > I mean that 4th edition UNIX ran under a hypervisor in MERT in 74 as a
> > process in that real-time executive.
> >
>
> Oh.  I thought maybe you meant Unix was able to do virtualization.
>
> What's special about being able to run under a hypervisor?  If the
> hypervisor does it's job right, whatever is running under it should
> not be aware that it is not running directly on hardware.
>

MERT was more a real-time executive than a hypervisor, so there was some
work needed to port UNIX to run as a process in MERT. The port was the unix
kernel, so that programs could have a UNIX API. It wasn't a pure
hypervisor, though, since a number of changes were required to Unix itself
to cope with running in what we'd likely call a paravirtualized environment.

>> Come to think of it, what do you mean by "OS/360's VM"?
> >>
> >
> > IBM's standard VM/360. Sorry for the confusion.
> >
>
> CP/67 or something like that maybe?  I don't think there was a VM/360
> either.
>

Sorry, it was VM/370, so the successor to CP/67 with virtual memory added.
https://akapugs.blog/2018/05/12/370unixpart2/

Warner