Marcelo Tosatti wrote:
> On Sat, Jan 19, 2008 at 04:37:43PM -0600, Anthony Liguori wrote:
>   
>> Avi Kivity wrote:
>>     
>>> Rusty Russell wrote:
>>>       
>>>> After discussions with Anthony Liguori, it seems that the virtio
>>>> balloon can be made even simpler.  Here's my attempt.
>>>>
>>>> Since the balloon requires Guest cooperation anyway, there seems
>>>> little reason to force it to tell the Host when it wants to reuse a
>>>> page.  It can simply fault it in.
>>>>
>>>>  
>>>>         
>>> Faulting is synchronous, while deflating is (or can be made) 
>>> asynchronous. If the host needs to do some work to get the memory, the 
>>> guest will be slowed down considerably.
>>>       
>> Good point.  Basically, we have two page hinting operations which 
>> roughly correspond to madvise(MADV_DONTNEED) and madvise(MADV_WILLNEED).
>>     
>
> Also, the simplified driver does not handle errors at all.
>
> I don't think that assuming madvise() can't fail is a good thing.
>   

I don't think the host should be communicating to the guest if madvise() 
fails.  The notification is a hint to the host.  The guest doesn't know 
what the host is doing in response to that hint.

Regards,

Anthony Liguori



-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-devel mailing list
kvm-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Reply via email to