[Spice-devel] [PATCH spice v0.12.6] Flush display commands before validating the surface.

[Frediano Ziglio]

> 
> On Tue, Dec 08, 2015 at 09:11:08AM -0600, Jeremy White wrote:
> > This fixes a display glitch in xspice which is caused when
> > a surface create is queued, but then a direct call to update
> > the area is issued.  Unless we flush the queue, the surface
> > does not exist, and we fail.
> 
> This also matches what handle_dev_update_async() is doing (flush, call
> VALIDATE_SURFACE_RET()).
> 
> Acked-by: Christophe Fergeau <cfergeau at redhat.com>
> 
> I'll push this once a 0.12 branch exists.
> 
> Christophe
> 

I would ack too, the patch is not harmful

However: Is this patch fixing a bug on spice-server or is just a workaround
of a bug of xspice?

I personally think the second. If you issue an asynchronous command you should
not try to issue another synchronous command depending on the first one but
you first should wait for first command finish and then issue the second.
Unless these commands are all defined to go into a single queue to they are
all serialized together.

Frediano