Altp2m misbehavior

Altp2m misbehavior

[Lai, Paul C]

[-- Attachment #1.1: Type: text/plain, Size: 533 bytes --]

Hello list:

Been playing with the altp2m code and noticed that it's not behaving in Xen 4.7 tree.
Two changesets have been identified as where the be behavior changes.

The change is commit bd2239d9fa975a1ee5bcd27c218ae042cd0a57bc,
      x86/HVM: always intercept #AC and #DB
where an executable goes from working to crashing after a 10+ hypercalls.

The second change is in commit 484c14b7254e8d8936c05e3c28e332ea825c0155:
   x86/vmx: enable PML by default
This hangs the VM guest on the first few hypercalls.

-Paul

[-- Attachment #1.2: Type: text/html, Size: 2686 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

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

Re: Altp2m misbehavior

[Jan Beulich]

>>> On 02.05.16 at 18:28, <paul.c.lai@intel.com> wrote:
> Been playing with the altp2m code and noticed that it's not behaving in Xen 
> 4.7 tree.
> Two changesets have been identified as where the be behavior changes.
> 
> The change is commit bd2239d9fa975a1ee5bcd27c218ae042cd0a57bc,
>       x86/HVM: always intercept #AC and #DB
> where an executable goes from working to crashing after a 10+ hypercalls.
> 
> The second change is in commit 484c14b7254e8d8936c05e3c28e332ea825c0155:
>    x86/vmx: enable PML by default
> This hangs the VM guest on the first few hypercalls.

I'm sorry, Paul, but this is way too little information to be useful for
anything beyond making people aware there is a problem. I suppose
you have someone working on these already. If so, it would be nice
for them to communicate their findings on the list, unless they're
already close to having fixes available. If not, and especially if any
form of help is being seeked here, please provide details.

Jan


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