Re: [Qemu-devel] [RFC 5/7] qxl-render: call ppm_save on callback

[Alon Levy]

On Wed, Feb 22, 2012 at 01:55:45PM -0200, Luiz Capitulino wrote:
> On Wed, 22 Feb 2012 15:29:33 +0100
> Alon Levy <address@hidden> wrote:
> 
> > On Wed, Feb 22, 2012 at 03:22:11PM +0100, Gerd Hoffmann wrote:
> > >   Hi,
> > > 
> > > > Honestly, for this particular case, I'm not 100% sure that having an id 
> > > > is
> > > > _required_, as I don't expect a client to submit multiple screendump 
> > > > calls
> > > > in parallel and we don't "officially" support multiple QMP clients 
> > > > either.
> > > > Also, having the screendump filename in the event will serve as a form 
> > > > of
> > > > identifier too.
> > > 
> > > That is exactly my thinking, echo the filename written in the event.
> > > 
> > > > Btw, are you planning to add the event to the already existing 
> > > > screendump
> > > > command? Adding a new command that doesn't use the monitor async API and
> > > > is truly asynchronous wouldn't better?
> > > 
> > > Good question.  I'd tend to just let the existing command send trigger
> > > an event.  But libvirt needs some way to figure whenever it should wait
> > > for an event ...
> > 
> > Right, that's the second reason I think a new command is needed.
> > Additionally a new command can be implemented only by qxl and not by
> > anything else (although I guess that would be a NACK?)
> 
> Is there anything specific to qlx in the command? If there's, then you
> should also consider making this a QOM device property. The downside is
> that QOM commands are not going to be stable for 1.1.

qxl is the only one that needs the async stuff since it needs to ask
spice-server to do the rendering, which is done in another thread
outside of qemu control. The rest of the screendump implementations
afaict don't need any such complexity.

The command itself would not be specific to qxl.

I thought I just need a QAPI command, and docs/ say QMP doesn't use that
yet, so what's the connection to QMP?