Re: Re: MVS under VM\ESA

kerravon86 wrote:
> 
> I looked up SIE in the index of the ESA POPS,
> but didn't see anything.
> 
> http://publib.boulder.ibm.com/cgi-bin/bookmgr/BOOKS/DZ9AR006/INDEX?DT=19990630131355

The Start Interpretive Execution (SIE) instruction isn't described in 
the Principle Of Operations. However, it is essential to the operations 
of VM/XA SF and SP, VM/ESA and z/VM. All XA and above systems implement 
it. Without this instruction, none of those virtualisation hypervisors work.

> 
> I guess it's to do with single instruction
> execution. What's that got to do with 2K
> vs 4K pages that Dave mentioned?
> 
> And surely pretending to protect 4K pages
> when you're only protecting 2K is going to
> have an adverse effect on something?

Your message also indicated how VM/ESA could handle S/370 DAT or S/370 
I/O.. That's what I was answering to.

Now...

When a S/370 program running on a S/370 with 4K protection keys issues a 
SSK or ISK instruction, it will be greeted with a program interrupt 
*UNLESS* CR0 Bit 7 is set to 1 in which case, SSK and ISK operate in the 
same manner as as SSKE/ISKE.

Since all programs executing as a S/370 SIE guest (i.e. under PR/SM, 
VM/XA SF, VM/XA SP, VM/ESA (ESA edition, not the 370 edition) or z/VM 
(up to 4.3)) execute under the protection keys provided by the SIE host 
which is *ALWAYS* a 4K protection key host (because there is no 2K keys 
in XA+), then all S/370 guests executing under those environments ALWAYS 
have a 4K protection key and therefore, SSK/ISK won't work unless CR0 
bit 7 is set to 1.

> 
>> my (not talking to me) friend.. SIE
> 
> Pardon? Who isn't talking to who?
> 

Your message dated July 9th 2010 on this list seemed pretty clear to me 
about this (a quite rude message along of the line of "messages are 
being utterly ignored", "pissing in the wind" and such language).

--Ivan