Post by ***@vietfeir.com [hercules-390]Would anone know if Hercules would run these OSs and are there
publicly available distributions for them?
IMO Hercules would not run these OSes. Although most of the
architecture would be the same as pre-64-bit architecture of IBM design,
I'd expect enough differences to stop it working.
(Of course, were the technical specs ever made available, and if the
Herc developers chose to, they could probably add support for these
architectures to Herc in fairly short order.)
Beside which, I would be amazed if you could find any freely-available
version of such OSes. These vendors kept their stuff strictly OCO.
They got their start by copying IBM's freely available OSes and then did
their own thing with them. Sometimes they stole stuff they shouldn't
have, according to news reports I read decades ago, and folks got into
trouble.
Was the Hitachi OS called VOS3?
I think I heard that name but never worked on it.
Fujitsu's 370-like architecture had channel-DAT and used additional CCWs
which 370 did not have.
eg. TL (translate and lock) and TVL (translate virtual and lock).
The IOS EXCP driver (for one) used TVL so that the channel programs
could specify virtual addresses.
Regarding instructions, the Fj FLM (find left-most one) instruction is
much older and not the same as IBM's z instruction of the same name
(which has a mnemonic of FLOGR). Fj's superzap service aid used it
(FLM, not FLOGR). At least, that is the instruction it S0C1 abended on
when run on MVS/370 on a 158.
Another op-code I remember is x'A1' which was AI - Add Immediate.
Then there was SVG (IIRC) which did the STM 14,12,12(13) bizzo as well
as chaining the save areas somehow all in one instruction.
So, those operating systems were close enough for humans to recognise,
and for exchanging tape volumes and even disk volumes (when the device
type geometries matched), and sometimes even utility files (unloaded
PDSes, IEHDASDR disk dumps) but not close enough to reliably run on IBM
hardware.
Fj's 31 bit PSW had a different format. MSP's LOC=ABOVE GETMAIN flags
were different making it difficult to run code compiled on MVS. MSP
control blocks had different structures to the MVS counterparts in many
cases. For instance, in MVS, the new and old ASCB pointers in the PSA
are after the new and old TCB pointers, but on MSP they were before. I
had my F4 (later MSP) system monitor working so many of the control
block structure differences will be encoded in that, so if you want more
details I might be able to look it up.
We did not have Assembler H, so when I wanted to try out fullscreen ZAP
(TSO utility) from the CBT tape, I had to assemble it on the Fj system
because IFOX could not handle the code. PGM=JLBASM (IIRC) had no
problem assembling it.
Auxiliary storage shortage? No problem. The PAGEADD command had the
ability to create a new page data set. (Page data sets were non-VSAM in
the VTOC. To manually allocate them, use PGM=KDJBR14 (IEFBR14 was an
alias) with a DISP=NEW DD stmt, and then use PGM=KAXPGFM to format them.)
In the 1980s, Telecom (which is now called Telstra) were a large Fj
shop, but gave up on it due to its inability to run software products
built by ISVs for MVS/XA even though it had 31-bit addressing. BSM and
BASSM worked the same so that's what I used in my code to switch
A-modes. Geez, is it really 35-ish years ago?
I /believe/ many Japanese companies still run MSP or whatever it is
called these days, but I think that Fj does not distribute it much
outside Japan anymore, in contrast to the past where it was marketed
thoughout the Asia-Pacific (usually directly) and Europe (usually rebadged).
That's about as much as I (think I) know on the topic...
Cheers,
Greg