Re: [Xen-devel] [Patch 2/6] Xen/MCE: remove mcg_ctl and other adjustment for future vMCE

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 06.08.12 at 15:51, "Liu, Jinsong" <jinsong.liu@xxxxxxxxx> wrote:
>>>> 
>>>> Is there a particular reason to make this access fault here, when
>>>> it didn't before? I.e. was there anything wrong with the previous
>>>> approach of returning zero on reads and ignoring writes when
>>>> !MCG_CTL_P? 
>>>> 
>>> 
>>> Semantically this code is better than previous approach, since
>>> !MCG_CTL_P means unimplemented MCG_CTL so access it would generate
>>> GP#. 
>> 
>> Agreed. But nevertheless I'd like to be a little more conservative
>> here. After all, "knowing" that this won't break Windows or Linux
>> isn't covering all possible HVM guests (and the quotes are there
>> to indicate that (a) unless you have access to Windows sources,
>> you can't really know, you may at best have empirical data
>> suggesting so, and (b) makes you/us dependent on all older
>> Windows/Linux versions you didn't try out/look at behave
>> correctly here too).
> 
> OK, fine to me to use previous approach, updated as attached.

Thanks, committed. Now what's your take on pulling the previous
patch #5 ahead (at once addressing the compatibility issue I
pointed out), as that's really the meat of the save/restore forward
compatibility? The other four patches could then be postponed
until after 4.2 went out afaict.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel