Re: [Xen-devel] Re: [PATCH 06/16] xen: disable PAT

["Jan Beulich" <jbeulich@xxxxxxxxxx>]

>>> Keir Fraser <keir.fraser@xxxxxxxxxxxxx> 10.05.09 19:34 >>>
>On 10/05/2009 14:29, "Alan Cox" <alan@xxxxxxxxxxxxxxxxxxx> wrote:
>
>>> have matching PAT configuration. No elfnote would mean use Xen's existing
>>> PAT setup (or if that's very dangerous then disable PAT altogether, perhaps
>>> dependent on CPU model/stepping?).
>> 
>> Hiding it on errata hit processors if the guest cannot support PAT
>> safely on such processors sounds a good policy and its one being a
>> hypervisor you can do neatly.
>> 
>> There are quite a few different CPUs with PAT errata. I've no idea why
>> there are so many errata about that specific bit.
>
>Okay, this sounds like a good way to then. I can work out details with
>Jeremy from here, and this will then have no impact on Linux PAT logic. If
>we advertise PAT to Linux via CPUID, that will mean PAT is set up just as
>Linux requires.

Depending on what exact plans you have here, I would question whether
having an upper layer dictate mechanisms used by a lower layer. In
particular, just like with the same mistake made with kexec,
- Linux's use of PAT may change (specifically, the value it writes to the PAT
  MSR may not be statically determinable at some point, making the ELF note
  approach you indicated as a possible solution unusable)
- the interface may not be suitable for non-Linux
Hence I'd appreciate if working out the details here could be done in public.

Thanks, Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel