(My sentmail shows I sent this reply last Tuesday, but since
the archives don't have it, I guess that explains why I got
no response. Here it is again)

Thank you Alan for the information, but it is still not totally clear to me
what is happening.

It vaguely occurred to me that perhaps somehow SIE was doing it all
now (as I wrote "bypass the simulation") at least when no CPUID had
been set by the user, so I *did* do a CP SET CPUID to a different serial
number, and it still did NOT pass control to HCPBIS.

So, how can I control the value if I do not get control passed to me?
SIE just saved the CPUID (including the user's SET CPUID serial) in
user storage, and I got no chance to alter it. (For reasons I won't go
into now, the part I have been modifying is the model number).

Is there perhaps some way to "degrade" my CP back to software
simulation of that particular instruction?

Thanks,
Shimon

> On Tuesday, 11/11/2008 at 11:49 EST, Shimon Lebowitz <[EMAIL PROTECTED]>
> wrote:
>
> > This exit is working perfectly in a z/VM 4.4 system, but I am now
> > having a problem in 5.3.
>
> > In the past, STIDP was simulated by HCPBIS. Is it
> > possible that now it is somehow managing to bypass
> > that simulation? It still knows to put in any CPUID
> > I set, so something seems to be intervening in the
> > normal hardware output. I am at a loss...  Maybe I am just
> > looking in the wrong place.
>
> STIDP is assisted by SIE (now).  If you want to control the value, you
> need to do an implicit SET CPUID at some other point in time, such as when
> the CPU is created.
>
> It's interesting that it's working perfectly on a z/VM 4.4 system since
> according to the notations in the code, it was added in z/VM 4.3!
>
> Alan Altmark
> z/VM Development
> IBM Endicott
>



--
************************************************************************
Shimon Lebowitz                mailto:[EMAIL PROTECTED]
VM System Programmer           .
Israel Police National HQ.    
Jerusalem, Israel              phone: +972 2 542-9877  fax: 542-9308
************************************************************************

Reply via email to