On Thu, May 5, 2022 at 4:03 PM Adam Thornton <athornton@gmail.com> wrote:


On May 5, 2022, at 2:32 PM, Tom Lyon via TUHS <tuhs@minnie.tuhs.org> wrote:

I was (re?)introduced to Chuck Haley recently and discovered he had a copy of a Bell Labs memo from himself, London, Maranzaro, and Ritchie.  They suggest that the path pursued to get UNIX running in/under TSS/370 was the hard way to go.


Oh my.

I mean, I know that OS (and descendants--for those of you without this particular trauma, we mostly mean MVS (now z/OS), but of course there's OS/MFT, OS/MVT, and then OS/MVS)  was always trying to kill VM, and we went through similar crap with Linux/390 (and zLinux), which contains code to let it run directly on the iron, even though even production shops are not going to *do* that (OpenSolaris-for-z required VM underneath it; hell, it required a couple new DIAGs we requested).

That actually makes it easy to run zLinux on Hercules, so I'm not complaining, but...it's not how any shop running it as anything other than a curiosity would do it.  I mean, OK, I guess you could have a teeny little LPAR but let's face it the LPAR is basically VM.

Possibly metaphorically but this LPAR facility is provided by PR/SM starting with later 3090s and ES/9000.  There is good coverage in IBM Systems Journal or IBM R&D journal (I can check my archives for which one if anyone needs a reference). It’s a firmware hypervisor like pHyp or Sun’s ldom.  VM is a strange yet delightful strain of operating system where full virtualization was front and center for both time sharing and isolation.  There’s really nothing quite like it although lately things like Qubes and MirageOS may share some of the delights.