Re: RE: [PATCH] Don't enable irq for machine check vmexit

[Keir Fraser <keir.fraser@eu.citrix.com>]

On 05/02/2010 14:36, "Jiang, Yunhong" <yunhong.jiang@intel.com> wrote:

>> How about the attached alternative, which avoids repeated reads of
>> VM_INTR_INFO? Also I'm not sure whether checking for
>> VM_EXIT_REASONS_FAILED_VMENTRY is useful, so I removed it. After all,
>> EXIT_REASON_MCE_DURING_VMENTRY should imply it anyway.
> 
> Thanks for your patch. Yes, it is much better to avoid the repeated read.
> 
> I'm not sure if it is ok if we don't check the VM_EXIT_REASONS_FAILED_VMENTRY.
> Checking the SDM and seems it is ok. In fact, I didn't find effective method
> to test this VMEntry MCE failed case, althgouh I can test MCE VMExit with
> EXIT_REASON_EXCEPTION_NMI case easily. (I will try to find a method to test
> this VMEntry failure case next week, maybe poison the VMCS range can trigger
> it, I'm not sure).

Well, we don't seem to know what bit 31 is for. Or, at least, we don't know
how it should affect our behaviour in the vmexit handler. So looking at it
does seem a bit pointless.

 -- Keir