Hi Naveen,

On Monday 17 October 2011 20:31:42 Naveen KRISHNAMURTHY wrote:

Hi Mauro,

Yes. From our (me and Alain) side we are quite interested in the actual guidelines for exposing the driver interfaces to user land through either V4L2 video device or sub-device or MC API.

I very much welcome your interest in the topic :-) Other chip vendors are interested as well, and the current lack of agreement doesn't help promiting MC/V4L2. The V4L community definitely needs to come up with a common position on this topic and to write guidelines for driver developers.

I'm fine with both extending the Sunday session and discussing this on Monday morning. I'm even willing to spend the entire night if that can help :-)

This is particularly in the context of STB / DTV platforms and complex SOC's. I think this is what you are proposing to allocate additional time as well. This would be welcomed change from our side. Monday morning is just fine.

I've included the feedback from the last round on our agenda for Oct 23 and 24.

I'm also adding a comment for each proposed theme, as the title sometimes may lead into confusion. Please review.

Sunday Oct 23 8:00 – 9:00 Breakfast 9:00 – 10:00 DVB API consistency: audio and video DVB API's - what to do? Mauro Chehab

The idea here is to discuss what should we do with the DVBv3 audio and video API, used only by one legacy driver on Kernel, and out of tree drivers. This API conflicts with V4L2 (and alsa?) API's. This theme may eventually be merged with the theme proposed by Naveen about using MC for DVB.

10:00 – 11:00 Videobuf2 – migration plans for legacy drivers Mauro Chehab

One hour may be too much for this theme. I've proposed this theme in order to discuss with videobuf2 maintainers/sponsors about how they're planning to add the missing features at VB2, and to start planning an strategy for migrating the existing drivers into it. There are some patches for saa7134 produced by VB2 maintainers, but they were incomplete, due to the missing features at VB2.

11:00 – 11:15 Break 11:15 – 12:15 V4L2 Spec: ambiguities fix Hans Verkuil

The idea here is to cover some aspects at the V4L2 spec that may lead into different implementations. IMO, this theme and the V4L/DVB on desktop x embedded are handling two aspects of the same subject.

12:15 – 13:15 Lunch Lunch 13:15 – 14:15 Buffer management: snapshot mode Guennadi

Guennadi proposed this subject in order to discuss how snapshot should work with a V4L2 device, when it is already streaming. There are some patches for buffer manager on my queue, but we may need more than that to properly support snapshots.

14:15 – 15:15 Multi FE support: one FE with multiple delivery systems like DRX-K frontend Mauro Chehab

How devices with one FE implementing two different DTV delivery systems should work? There are two ways of doing that, and they're implemented currently on a way that it is not very consistent.

15:15 – 15:30 Break 15:30 – 16:30 Multiple CI encoders and how to remove the current CI drivers from staging Mauro Chehab

Dmitri proposed this theme (and also I did), but it seems he can't come to the workshop. Basically, some CI drivers went into staging, because of the lack of a proper way to export those device capabilities via the DVB API. I'm not a CI expert. Is there anyone going to the workshop interested on leading this discussion?

16:30 – 17:00 Closing Notes Mauro Chehab

Sunday Oct 24 8:00 – 9:00 Breakfast 9:00 – 11:00 V4L2 / DVB on desktop vs. embedded systems Sakari Ailus

This is a hot theme at the ML. Basically, different types of devices require different ioctl support. The API spec doesn't say anything about what ioctl's are expected to be implemented on each device type. Worse than that, some things can now be implemented using a different API (MC/subdev API), making hard to review patches in a way that allows the new drivers to work with the existing applications.

11:00 – 11:15 Break 11:15 – 12:15 New DVB standards (ATSC-MH) Michael Krafky

Some new delivery systems emerged, and a proper support for them are needed at the DVB API.

12:15 – 13:15 Lunch 13:15 – 14:15 Media Controller & DVB Naveen Krishnamurthy

DVB SoC devices are as complex as V4L ones. A proper support for them will require extending the MC API to cover their needs. It would be good if Naveen/Valmat could give us a 15-20 min presentation explaining the needs from ST related to it.

14:15 – 15:15 Hacking day 15:15 – 15:30 Break 15:30 – 16:30 Hacking day 16:30 – 17:00 Closing Notes Mauro Chehab

Comments? Mauro

Hi,

On 10/21/2011 01:43 PM, Mauro Carvalho Chehab wrote:

Em 19-10-2011 11:56, Mauro Carvalho Chehab escreveu:

...

I've included the feedback from the last round on our agenda for Oct 23 and 24.

I'm also adding a comment for each proposed theme, as the title sometimes may lead into confusion. Please review.

Sunday Oct 23 8:00 – 9:00 Breakfast 9:00 – 10:00 DVB API consistency: audio and video DVB API's - what to do? Mauro Chehab

The idea here is to discuss what should we do with the DVBv3 audio and video API, used only by one legacy driver on Kernel, and out of tree drivers. This API conflicts with V4L2 (and alsa?) API's. This theme may eventually be merged with the theme proposed by Naveen about using MC for DVB.

10:00 – 11:00 Videobuf2 – migration plans for legacy drivers Mauro Chehab

One hour may be too much for this theme. I've proposed this theme in order to discuss with videobuf2 maintainers/sponsors about how they're planning to add the missing features at VB2, and to start planning an strategy for migrating the existing drivers into it. There are some patches for saa7134 produced by VB2 maintainers, but they were incomplete, due to the missing features at VB2.

11:00 – 11:15 Break 11:15 – 12:15 V4L2 Spec: ambiguities fix Hans Verkuil

The idea here is to cover some aspects at the V4L2 spec that may lead into different implementations. IMO, this theme and the V4L/DVB on desktop x embedded are handling two aspects of the same subject.

12:15 – 13:15 Lunch Lunch 13:15 – 14:15 Buffer management: snapshot mode Guennadi

Guennadi proposed this subject in order to discuss how snapshot should work with a V4L2 device, when it is already streaming. There are some patches for buffer manager on my queue, but we may need more than that to properly support snapshots.

14:15 – 15:15 Multi FE support: one FE with multiple delivery systems like DRX-K frontend Mauro Chehab

How devices with one FE implementing two different DTV delivery systems should work? There are two ways of doing that, and they're implemented currently on a way that it is not very consistent.

15:15 – 15:30 Break 15:30 – 16:30 Multiple CI encoders and how to remove the current CI drivers from staging Mauro Chehab

Dmitri proposed this theme (and also I did), but it seems he can't come to the workshop. Basically, some CI drivers went into staging, because of the lack of a proper way to export those device capabilities via the DVB API. I'm not a CI expert. Is there anyone going to the workshop interested on leading this discussion?

16:30 – 17:00 Closing Notes Mauro Chehab

Sunday Oct 24 8:00 – 9:00 Breakfast 9:00 – 11:00 V4L2 / DVB on desktop vs. embedded systems Sakari Ailus

This is a hot theme at the ML. Basically, different types of devices require different ioctl support. The API spec doesn't say anything about what ioctl's are expected to be implemented on each device type. Worse than that, some things can now be implemented using a different API (MC/subdev API), making hard to review patches in a way that allows the new drivers to work with the existing applications.

11:00 – 11:15 Break 11:15 – 12:15 New DVB standards (ATSC-MH) Michael Krafky

Some new delivery systems emerged, and a proper support for them are needed at the DVB API.

12:15 – 13:15 Lunch 13:15 – 14:15 Media Controller& DVB Naveen Krishnamurthy

DVB SoC devices are as complex as V4L ones. A proper support for them will require extending the MC API to cover their needs. It would be good if Naveen/Valmat could give us a 15-20 min presentation explaining the needs from ST related to it.

14:15 – 15:15 Hacking day 15:15 – 15:30 Break 15:30 – 16:30 Hacking day 16:30 – 17:00 Closing Notes Mauro Chehab

Comments? Mauro

No comments so far, so, I'll assume that everybody is ok with the current proposal ;)

I've been offline for a couple of days (PTO), fwiw I'm ok with the current proposal :)

Regards,

Hans