Re: [PATCH] virtio-balloon spec: provide a version of the "silent deflate" feature that works

[Paolo Bonzini <pbonzini@redhat.com>]

Il 10/09/2012 08:03, Michael S. Tsirkin ha scritto:
> On Mon, Sep 10, 2012 at 07:50:13AM +0200, Paolo Bonzini wrote:
>> Il 09/09/2012 00:22, Michael S. Tsirkin ha scritto:
>>>> Almost.  One is "the guest, if really needed, can tell the host of
>>>> pages".  If not negotiated, and the host does not support it, the host
>>>> must break the guest (e.g. fail to offer any virtqueues).
>>>
>>> There is no way in spec to break the guest.
>>> You can not fail to offer virtqueues.
>>
>> You can always return 0 for the first queue.
> 
> I don't think guest drivers recover gracefully from this.
> Do they?

No, that's the point ("break the guest" is really "break the driver").

>>> Besides, there is no guarantee that virtqueue setup
>>> happens after feature negotiation.
>>
>> It is the only way that makes sense though (unless the guest would write
>> 0 for its features).  Should we change that?
> 
> I do think it would be nice to add a generic way for device to
> notify guest about an internal failure.
> This can only happen after DRIVER_OK status is written though,
> and since existing drivers do not expect such failure, it might
> be too late.

Agreed.

>>>> The other is "the guest, though, would prefer not to do so".  It is
>>>> different because the guest can proceed in a fallback mode even if the
>>>> host doesn't offer it.
>>>
>>> I think I get what your proposed SILENT means what I do not get
>>> is the motivation. It looks like a premature optimization to me.
>>
>> The motivation is to let the driver choose between two behaviors: the
>> current one where ballooning is only done on request, and a more
>> aggressive one.
> 
> Yes but why is being silent any good? Optimization?
> Any data to show that it will help some workload?

Idle guests can move cache pages to the balloon.  You can overcommit
more aggressively, because the host can madvise away a lot more memory.

>>> OK so TELL says *when* to notify host, SILENT if set allows guest
>>> to skip leak notifications? In this case TELL should just be ignored
>>> when SILENT is set.
>>
>> Yeah, that was my first idea.  However, there are existing drivers that
>> ignore SILENT, so that would not be 100% exact.
> 
> Not sure I follow the logic.
> They don't ack SILENT so that would be 100% exact.

Hmm, then I'm not sure I follow yours.  We agreed that delaying
notifications or skipping them is really the same thing, right?

I think we're just stuck in a linguistic problem, with "must not" being
wrong and "does not have to" being too verbose.  Calling it
VIRTIO_BALLOON_F_SILENT_DEFLATE was a workaround for this, but perhaps
it adds more confusion.

Paolo