Re: [PATCH v4] target/s390x: Implement the MVPG condition-code-option bit

[Cornelia Huck]

On Thu, 4 Mar 2021 09:10:44 +0100
David Hildenbrand <david@redhat.com> wrote:

> On 03.03.21 22:36, Richard Henderson wrote:
> > On 3/3/21 1:22 PM, David Hildenbrand wrote:  
> >>  
> >>> Am 03.03.2021 um 22:19 schrieb Richard Henderson 
> >>> <richard.henderson@linaro.org>:
> >>>
> >>> On 3/3/21 1:11 PM, David Hildenbrand wrote:  
> >>>> MMIO on s390x? :)  
> >>>
> >>> hw/s390x/s390-pci-bus.c, memory_region_init_io*().
> >>>  
> >>
> >> ... part of system address space where a CPU could stumble over it?  
> > 
> > Impossible to tell within 3 layers of object wrappers.  :-(
> > I suppose I have no idea how "pci" was hacked onto s390x.  
> 
> You've used the right words to describe "pci" (!) on s390x.
> 
> IIRC, there is no MMIO: configuration space accesses etc. are performed 
> using special access instructions - which will "emulate" the MMIO access 
> performed on other archs via simple read/write instructions.

Yes, it's all specialized instructions. (QEMU emulates one set, there's
a newer one supported by the Linux kernel that is a bit more mmio-like
in semantics, but I don't know the details.)

> 
> Ordinary instructions (e.g., mvpg) that operate on the system address 
> space should never stumble over MMIO regions - because that concept does 
> not exist on s390x.
> 

That's my understanding as well.