[Libdlo] [ANNOUNCE] udlfb 0.2.2 and xf86-driver-displaylink 0.2

[Roberto De Ioris]

> Roberto De Ioris wrote:
>> I think that the "displaylink is a card" approach, is not a good choice
>> in the modern time.
>>
>> I will try to follow the "displaylink is an output" way, hacking the
>> intel
>> driver. I will post some code as soon as possible.
>>
>>
>
> not entirely sure I understand the concept of using displaylink as an
> output. Does this mean that if the GPU on the card does accelerated 3D
> then this could be done then the rendered output sent out to the
> displaylink? Compiz on the displaylink would be kind of amazing.
> I think the displaylink as a card approach would still be useful
> especially for embedded applications where the controlling computer
> might not have anything resembling a GPU on it.
>
> Or did I miss the point?
>
>
> Alan.
>

Yes, you have got the point.
The trick is using a bigger framebuffer in the card device (i am hacking
on the intel driver) and attach a Damage Block Handler that send commands
to the displaylink when something in the framebuffer zone of interest is
changed.
In this way you can use the 3d feutures of the gfx card and use the
displaylink as a simple (slow and complex) monitor.

Obviously the xf86-driver-displaylink approach must continue to exist for
devices like
the sheevaplug or more generally embedded system without graphics hardware.

--
Roberto