Hi all,
Here is a rough draft of the agenda. Note that I have postponed the ALSA/IIO parts to day 2 since Shuah arrives late on day 1.
Lars-Peter, if you think the IIO/ASoC presentations/discussions should be moved to day 1 since that is really relevant, then let us know.
Since this is a brainstorm meeting the agenda is very high level and totally unreliable :-)
The goal is to make the MC API work for new subsystems and if we are more likely to achieve that by rearranging the agenda, then we'll do that.
Day 0 (i.e. before the workshop starts!)
Everyone should read through the links provided by Mauro in the workshop announcement and the RFC for the property API (see the link below). We don't want to have to repeat everything again during the workshop.
Day 1:
Discuss the MC+DVB API.
In my view the core conflict is about how to represent interfaces and the control links between interfaces and entities. We need to decide on this first before we can go on to other MC topics,
I proposed a solution for this (http://www.spinics.net/lists/linux-media/msg91365.html) and we need to discuss if this works for modeling DVB and V4L devices.
If we agree that my proposal is a good starting point, then we need to discuss the following open issues that I didn't pin down in my proposal:
- how to represent the control links between interfaces and entities - property API (http://www.spinics.net/lists/linux-media/msg90160.html) - how should the 'dev' struct in struct media_entity_desc be used if we have multiple interfaces controlling this entity?
If my proposal is rejected, then we will have to come up with an alternative and the remainder of the agenda will be highly uncertain.
Day 2:
Start with the IIO and ASoC presentations.
Continue the Day 1 discussions if needed.
Day 2+3:
Discuss MC+ALSA+IIO.
Time Division Multiplexed pipelines: how to represent those in the MC?
Discuss my VIDIOC_SUBDEV_QUERYCAP proposal:
v2: https://www.mail-archive.com/linux-media@vger.kernel.org/msg88100.html v3: https://www.mail-archive.com/linux-media@vger.kernel.org/msg90155.html
The basic problem (related to the requirements of v4l2-compliance) is that today there is no unique ioctl that all subdev device nodes support (hence the addition of VIDIOC_SUBDEV_QUERYCAP) and that there is no link back from a V4L2, DVB, ALSA, etc. device to the media controller device that it is part of.
End of Day 3:
Create action items and who will work on each item.
Should we make a tech topic proposal for the Kernel Summit? (Need to send it in on this day as well)
Regards,
Hans