[Spice-devel] [Qemu-devel] [RFC PATCH spice v3 1/3] QXL interface: add a function to identify monitors in the guest

[Frediano Ziglio]

> On Thu, Nov 08, 2018 at 11:05:10AM +0100, Lukáš Hrázký wrote:
> > Hello,
> > 
> > On Thu, 2018-11-08 at 07:49 +0100, Gerd Hoffmann wrote:
> > >   Hi,
> > > 
> > > > + * The device_display_id_{start,count} denotes the sequence of device
> > > > display
> > > > + * IDs that map to the zero-based sequence of monitor IDs provided by
> > > > monitors
> > > > + * config on this interface. For example with device_display_id_start
> > > > = 2 and
> > > > + * device_display_id_count = 3 you get the following mapping:
> > > > + * monitor_id  ->  device_display_id
> > > > + *          0  ->  2
> > > > + *          1  ->  3
> > > > + *          2  ->  4
> > > > + *
> > > > + * Note this example is unsupported in practice. The only supported
> > > > cases are
> > > > + * either a single device display ID (count = 1) or multiple device
> > > > display IDs
> > > > + * in a sequence starting from 0.
> > > 
> > > This is confusing.  The usage of this api in the qemu counterpart looks
> > > sane though.
> > 
> > Not sure what you find confusing in particular... The example? Using an
> > example and then saying it's not supported?
> 
> Yes, that for example.  Also wondering why device_display_id doesn't
> start at zero.
> 

You introduced this ID so you should remember.
It starts from 0, just this is not the first registration, the other displays
were registered to other QXL instances.
This was introduced for virtio-gpu to be able to register the different
outputs to different QXL instances so to have

first output/display:
  spice_qxl_set_device_info(instance1, path1, 0, 1);
second:
  spice_qxl_set_device_info(instance2, path1, 1, 1);
third:
  spice_qxl_set_device_info(instance3, path1, 2, 1);

So as you can see device_display_id start from 0 but does not mean
is always 0 for every call.

> cheers,
>   Gerd
> 
> 
> 

Frediano