Re: arch_set_info_guest() producing inconsistent state on x86?

[Keir Fraser <keir.xen@gmail.com>]

On 29/03/2011 13:01, "Keir Fraser" <keir.xen@gmail.com> wrote:

> On 29/03/2011 12:45, "Jan Beulich" <JBeulich@novell.com> wrote:
> 
>>> It's probably used by debuggers running in dom0? Also see
>>> modify_returncode() in libxc/xc_resume.c -- so it's used on suspend resume
>>> in the failure case.
>>> 
>>> I doubt anything other than GPRs are ever modified after first
>>> initialisation.
>> 
>> So should we then perhaps make the function check the bits
>> it doesn't really update match what is in place already?
> 
> I suppose it would be nice. I can't say I care much one way or the other.

By which I mean: if you want to make the change, and do it in a way that is
clean and clear (maybe you can improve the function's readability while
there, since it is a bit of a mess) then I'll apply it happily. I don't want
to make the function even worse and more unmaintainable than it already is,
however.

 -- Keir

>  -- Keir
> 
>