Re: [Qemu-devel] Re: [PATCH 0/8] kvm: Fixes, cleanups and live migration

[Gleb Natapov]

On Sun, May 03, 2009 at 08:57:58AM +0300, Avi Kivity wrote:
> Gleb Natapov wrote:
>> On Sat, May 02, 2009 at 10:12:57PM +0300, Avi Kivity wrote:
>>   
>>> Gleb Natapov wrote:
>>>     
>>>>> I think the right thing to do with this is introduce a kvm-cpu 
>>>>> savevm  that stores this information since it isn't relevant to 
>>>>> TCG.  I think  it's arguable whether you want instruction length 
>>>>> there (can you get it  reliably on SVM?).
>>>>>
>>>>>             
>>>> We can't get it on SVM without instruction decoding, but it is not required
>>>> on SVM. It is absolutely essential for soft interrupt/exception injection
>>>> on VMX and has to be a part of migratable state.
>>>>         
>>> We need it in some neutral form so cross-vendor migration can work.
>>>
>>>     
>> VMX->SVM No problem.
>> SVM->VMX bad luck :)  We will have to decode instruction ourself.   
>
> I don't think it's necessary.  We can record the software interrupt at  
> the end of the instruction that generated it, and give it higher  
> priority than a pending external interrupt.  On vmx, decrement RIP and  
> set entry instruction length = 1 before injection.
And get wrong error value when exception happens during soft interrupt
delivery? I don't like all those tricks. They work only if everything
happens like you expected and breaks completely when it is not.

>                                                     On svm, use EVENTINJ  
> and forget about the instruction length.
>
On SVM we do not re-inject soft int/exception at all, but re-execute the
offending instruction.

--
                        Gleb.