[vfio-users] Framebuffer intercept: is it possible?

[Alex Williamson]

On Tue, 2015-09-15 at 13:20 +0200, Leonhard Preis wrote:
> Hi!
> 
> Since the card outputs the signal directly to the monitor, never passing  
> through the domain of the host that is very unlikely. Only thing I could  
> imagine is that you could find a way to access the cards memory and grab  
> something there but that’s not the same thing, right?

Yes, this is how you'd need to do it and it is theoretically possible.
A privileged user in the host could mmap device memory and read from it
(writing is most likely a very bad idea).  I don't know enough about the
internal operation of a graphics card, but I assume the framebuffer
would be accessible or could be pieced together if the user had
sufficient knowledge of the programming model of the card.

> Am 15.09.2015, 10:01 Uhr, schrieb francesco dicarlo  
> <evilsephiroth at gmail.com>:
> 
> > Reallly interested in this. It would be a good way to analyse also  
> > performance.
> >
> > 2015-09-15 9:09 GMT+02:00 Blank Field <ihatethisfield at gmail.com>:
> >>
> >> Is it theoretically possible to capture the video output of the VM's  
> >> p-t'd GPU using some host facilities?
> >> It would be extra awesome to be able to capture exactly what you see on  
> >> the screen, like any stuttering, corruption, directX shaders and  
> >> accelerated video playback.
> >>
> >> _______________________________________________
> >> vfio-users mailing list
> >> vfio-users at redhat.com
> >> https://www.redhat.com/mailman/listinfo/vfio-users
> >>
> >
> 
> 
> 
> _______________________________________________
> vfio-users mailing list
> vfio-users at redhat.com
> https://www.redhat.com/mailman/listinfo/vfio-users