Re: IBM 390
On Thu, 25 Mar 1999, Sid Sorenson wrote:
> Woha Phillip, I am impressed. I am trying to get used to this mailing
> list, these are some of the responses that I tried to post earlier. Is
> this showing up on the list????
Aie! I seem to be losing emails! DRAT DRAT DRAT! Stupid ISP. Don't worry
folks; I *should* be on a MUCH more reliable address with more time to
read/reply within 4 to 6 weeks. My company is moving to a totally
different building, and I've got total control of the data center. Which
includes the email servers. ;)
> I fully realize that there is no way linux can get the same throughput
> from a 390 that OS/390 gets. I have been around both mainframes and Unix
> machines for some time and I am intimately aware of the power and
> flexability of a purely Mainframe OS in a production environment.
> Buttttt, I talked the boss into letting me play with this thing till it
> breaks. This is an F12 and we are replacing it with an F55 in Q4. I was
> just trying to see if anyone wants to tinker as well.
Actually, this is *totally* untrue. Providing we can get the necessary
documentation, I don't see ANY reason why Linux can't equal or even
SURPASS OS/390. It just takes time, information, and hard work.
> We have to wait a while, the box may become available as early as
> August, but it will definately be availabe late November. Mantenance
> runs out December 30 1999, and my boss says I can play with it and drop
> a line into it from the Internet until it breaks.
> Philip R Jaenke has expressed an intrest, but if I remember right, the
> Power PC processors were used in the 370 line. I can check on that
> though.
I haven't had a chance to look, so if anyone out there could please
confirm or refute this, please let me know. If it's true, anyone who can
get me access to S/370 hardware, PLEASE get in touch ASAP. };)
> Any idea what it would take to get the kernel to suport it? I figured
> the approach would be to get gas working for that archetecture and then
> get gcc up. The next thing I thought whuld be needed was a tftp loader
> so the box could be IPL'd That would probably have to be written to tape
> and loaded from there. From there, the kernel should at least compile.
> tftp should allow us to load a kernel and see how it dies.
>
> I was thinking of dropping a vpn into a PC running 2.1 on the same
> network and making that and the 390 available on the net. The pipe would
> be small (Probably a 56K modem) but it would be accessible for people to
> hack on.
The best approach IMNSHO, is to get the COMPILER first. Once you have the
compiler, cross-compiling is near trivial. That's a *BIG* key in
developing for a currently unsupported architecture. Then you can play a
little safer.
I don't recommend a tftp bootload or *any* sort of 'not-normal' way of
booting the machine, even during development, until *AFTER* you have a
working bootstrap. You can quote me on this. It's my opinion that you
stand a chance of skewing results and possibly just plain screwing up if
you concentrate on alternate methods before what should be the primary
method. Key word there being 'opinion.' I've only worked with RS/6000's.
And the bootloader was relatively easy. It may be harder with an S/390. I
honestly can't say for sure.
As to availability, it's mostly a moot point as far as I'm concerned, due
to my ISP being a bunch of MORONS. I'm lucky if I can stay connected for
>15 minutes anymore. And, no, it's not my phone line. It's remote
disconnects according to my logs and testing. And they're on PRIs. So you
figure it out. ;P Should be on a decent connect soon enough, though, or
so I hope. We'll see. Only time will tell.
+----------------------------------------------------------------+
| Phillip R. Jaenke | "Not all wisdom comes from without; |
| Head of Unix Systems | much wisdom can only come from |
| Unicent Telecom | within. Only you can teach yourself |
| prj@nls.net (RISCy@IRC) | some of the most important lessons |
+-------------------------+ of life." --Takes-Many-Roads, |
Project Head | Silent Strider Theurge |
The Linux RS/6000 Project +--------------------------------------+
-PESTER ME!- PJaenke@unicent.com / 800.319.9250 x4268 -PESTER ME!-
Reply to: