eglglessink on Raspberry PI

Julian Scheel julian at jusst.de
Thu Jan 31 05:56:37 PST 2013


Hi Sebastian,

Am 31.01.2013 um 09:16 schrieb Sebastian Dröge <sebastian.droege at collabora.co.uk>:

> On Di, 2013-01-22 at 13:31 +0100, Julian Scheel wrote:
>> Hi,
>> 
>> I have done some work on adding a Raspberry PI support to the
>> eglglessink. A very rough preliminary patch is attached. Especially the
>> configure.ac is evil, but I haven't had the time to properly sort out
>> the build issues I faced yet.
>> Actually the gl context is created and all gl calls seem successfull,
>> but the output is only a single color plane. The color seems to be what
>> is in the top left corner of the input picture.
>> I wonder about what possibly could cause this? One thing I could imagine
>> would be the VertexAttribPointer settings. I don't really understand the
>> code there:
>> 
>>  glVertexAttribPointer (eglglessink->eglglesctx.position_loc[0],
>> 	3, GL_FLOAT, GL_FALSE, sizeof (coord5),
>> 	(gpointer) (0 * sizeof (coord5)));
>> 
>> What does the (gpointer) (0 * sizeof (coord5)) as vertex array pointer
>> mean? For texloc the code is quite similiar.
> 
> It's the offset in the vertex array.

Thanks, in the meantime I was able to understand the code. Actually it seems we hit a bug in the broadcom GLES implementation. I have opened a bug in the raspberry firmware project: https://github.com/raspberrypi/firmware/issues/142
So it should be under investigation by broadcom now.

> Could you open a bugreport for this?

Yes I will open a bug and attach my code there.

Regards,
Julian


More information about the gstreamer-devel mailing list