path: root/wiki/Chat_log/20161222-mm-chatlog

Multimedia-chat-meeting-2016-12-22

09:05 < pinchartl> topics for today
09:06 < pinchartl> - status check as usual
09:06 < pinchartl> anything else for anyone ?
09:07 < pinchartl> I'll take that as a no
09:07 < pinchartl> kbingham: as you're here you can start :-)
09:07  * kbingham should get a name change to zebra 
09:08 < pinchartl> be careful what nickname you wish for, they tend to stick around
09:09  * wsa_ used 'sort -R' to generate the speaking order in IO meetings ;)
09:09 < kbingham> So, in the last two weeks, A), I've submitted vsp1 suspend resume patches, and they have been reviewed, and I've done some ftrace measurements on the performance impacts of the changes to where partitions are calculated, (out of interest, and because they are useful to know the numbers/process)
09:10 < kbingham> Externally (but still sort of related), i've spent the majority of this week on my garage office conversion with the Father In Law
09:11 < kbingham> B) - I still intend to follow up on the review comments from the suspend resume patches, and of course we have Christmas holidays here.
09:11  * morimoto start chat in Renesas
09:12 < kbingham> - That will probably be the most I'll achieve in the next two weeks - so the 2016/q1 work will likely start in 2016/q1 :D
09:12 < pinchartl> sounds fair :-)
09:13 < kbingham> C) - I may have sprained my wrist, doing DIY work in the garage.
09:13  * kbingham EOF
09:13 < pinchartl> thank you
09:14 < pinchartl> seems it's time to take a bit of time off the computer to let your wrist rest then
09:14  * kbingham ponders the meaning of 'rest'
09:15 < pinchartl> don't ask me
09:15 < pinchartl> I'm next in alphabetical order
09:15 < pinchartl> over the past two weeks
09:16 < pinchartl> I've started V4L2 cache API analysis and discussions with the V4L2 community
09:16 < pinchartl> I revived a patch series that has been posted a year ago and submitted a new version
09:16 < pinchartl> which has been acked by Hans Verkuil already
09:16 < pinchartl> bus still requires a few changes
09:17 < pinchartl> the code covers V4L2 native buffers, but not buffers imported by V4L2 from dmabuf
09:17 < pinchartl> so I'll still need to address that part
09:17 < pinchartl> and it will require broadening the discussion to include the dmabuf and DRM developers
09:18 < pinchartl> I've also posted a new version of the HDMI output on Gen3 patches
09:19 < pinchartl> the dw-hdmi code has been significantly reworked compared to the previous version and now handles different versions of the IP core in a much cleaner way
09:19 < pinchartl> I've already received a few comments from review, a third version will likely be needed
09:20 < pinchartl> and finally I've spent time fighting with the V4L2 maintainer as usual
09:20 < kbingham> I'm sure synopsis are happy that the underlying hardware abstractions are now less obfuscated :D
09:20 < pinchartl> kbingham: they actually acked a few of the patches, so I think they're happy, yes
09:21 < pinchartl> we have issues in the media controller core that need to be addressed. they've been there from the start, but have been made worse by the maintainer
09:21 < pinchartl> this will require a face-to-face meeting, and one has been proposed at ELC time
09:22 < pinchartl> however, I've decided to organize another one sooner, at mid-January in Oslo
09:22 < pinchartl> Sakari Ailus, Hans Verkuil and myself will attend
09:22  * morimoto back from chat
09:23 < pinchartl> my hope is to come to an agreement with Hans first, without involving other V4L2 maintainers
09:24 < pinchartl> during the next two week I'll continue reviewing, reworking and resubmitting patches that haven't been accepted yet
09:24 < morimoto> Is this issue related to HDMI output ? or just media controller ?
09:25 < pinchartl> media controller only
09:25 < pinchartl> it's not related to any of my additional tasks in particular
09:25 < pinchartl> and it's taking more time from my base contract that I would like :-/
09:26 < pinchartl> which brings me to the issues and blockers
09:26 < pinchartl> no blocker as such, but an issue
09:26 < morimoto> OK, I understand.
09:26 < pinchartl> I have too much work for my base contract
09:26 < morimoto> :)
09:26 < pinchartl> with all the maintainership and involvement in the V4L2 and DRM/KMS core
09:27 < pinchartl> they're not covered by any additional task, and they take a large amount of time
09:28 < pinchartl> given the current 50/50 split there isn't really room to handle the problem
09:28 < pinchartl> but I hope to be able to offload some of the work to Kieran if we can get budget
09:29 < pinchartl> kbingham: don't worry, I'll still handle the fights, you would mostly need to handle driver maintainership :-)
09:31 < kbingham> pinchartl: will I need more popcorn in that case :D
09:31 < pinchartl> :-)
09:31 < pinchartl> that's all for me
09:31 < pinchartl> morimoto: you're next
09:31 < morimoto> OK
09:31 < morimoto> I asked about ALSA SoC + HDMI + DT things to pinchartl, but couldn't get answer in time
09:32 < morimoto> so I worked other things this week
09:32 < morimoto> I posted few patches to ALSA ML for cleanup etc
09:32 < morimoto> I would like to solve HDMI sound things, but it seems difficult
09:32 < morimoto> Blocker is it.
09:33 < morimoto> pinchartl said that V4L2 side is thinking about .query method ?
09:33 < morimoto> but, I still wondering about it.
09:33 < pinchartl> it's a problem that spans multiple subsystems, so it should be implemented at the media controller level in my opinion
09:34 < pinchartl> but we're definitely not there yet and it will require a lot of work
09:34 < morimoto> does your "we" mean PeriMulti ? or V4L2 ?
09:35 < morimoto> or Linux ?
09:35 < pinchartl> Linux :-)
09:35 < morimoto> hehe, OK :)
09:35 < pinchartl> I know how I'd like to see this being implemented
09:35 < pinchartl> and it would require using the media controller framework in DRM/KMS and ALSA
09:36 < pinchartl> which means that we need to solve the MC core issues first
09:36 < morimoto> MC core ?
09:36 < pinchartl> otherwise it will not be possible to convince other subsystems to use it
09:36 < pinchartl> media controller core
09:36 < morimoto> Ahh OK
09:37 < morimoto> From ALSA side point of view, it need to care about non-HDMI and HDMI. So my question is that
09:37 < morimoto> Does your "MC" is related to both non-HDMI and HDMI ?
09:37 < morimoto> or only HDMI ?
09:37 < pinchartl> both
09:37 < morimoto> Hmmm
09:37 < morimoto> This means, existing ALSA itself need to be update ?
09:38 < morimoto> without HDMI support
09:38 < pinchartl> the problem is that we need to know about port types to parse and walk the DT OF graph
09:38 < pinchartl> that information isn't available in DT
09:38 < pinchartl> it is only known to drivers
09:38 < morimoto> Yes.
09:38 < morimoto> But, in existing ALSA, all are sound, no question.
09:38 < pinchartl> so we can't implement generic OF graph parsing and walking code without help from drivers
09:39 < pinchartl> we need a new API for drivers to provide the information
09:39 < morimoto> OK, I see. but I wonder, this means, I can't rework for HDMI sound at this point ?
09:40 < pinchartl> when an OF graph is limited to one of the ALSA, DRM and V4L2 subsystems only, we can implement a subsystem-specific solution
09:40 < pinchartl> but in the general case, we need a solution that is compatible will all three subsystems
09:40 < pinchartl> so we need a common object that can implement this API
09:40 < pinchartl> and at the moment we have none
09:41 < morimoto> My ? is related to here
09:41 < pinchartl> the best candidate in my opinion is the media_entity object, that would need to be used by all three subsystems
09:41 < pinchartl> HDMI sound doesn't involve V4L2 but it involves the two other subsystems
09:41 < morimoto> three = ALSA / DRM / V4L2 ?
09:41 < pinchartl> yes
09:42 < morimoto> If this OF graph is related to only multimedia case, right ?
09:42 < pinchartl> it should be yes
09:42 < morimoto> If non multimedia want to use OF, it will not be "media_entity"
09:42 < morimoto> Hmm...
09:43 < pinchartl> correct, but non-multimedia devices don't involve ALSA :-)
09:43 < pinchartl> I'm open to other proposals, but in any case I believe this will be lots of work
09:43 < morimoto> But can use OF-graph ?
09:44 < morimoto> Is this related to know "which port is what type" ?
09:44 < morimoto> = OF-graph
09:44 < pinchartl> it's related to OF graph parsing
09:44 < pinchartl> not only about port type
09:45 < pinchartl> a DT node has ports that describe its external connections
09:45 < morimoto> OK, if so, naming should use "media" ?
09:45 < morimoto> should not
09:45 < pinchartl> but doesn't include information that describe the internals of the IP core
09:45 < pinchartl> that's the information missing at the moment
09:46 < pinchartl> we could use a different namespace, implement the API at a different level
09:46 < pinchartl> I'm not sure where though
09:46 < morimoto> I understand basic ideas
09:47 < morimoto> I wonder where do you (or other guys ?) discuss this topics ?
09:47 < morimoto> I mean which ML ?
09:47 < morimoto> ALSA side guys need to join it
09:47 < pinchartl> MC discussions usually happen on the linux-media mailing list
09:48 < pinchartl> if we implement the API somewhere else, I don't know
09:48 < morimoto> This is "OF-graph" releated method, right ?
09:48 < morimoto> we can get information from this new API
09:48 < morimoto> not only type, but many information
09:49 < pinchartl> yes, we need more than just the type
09:49 < morimoto> But this OF-graph use is not only V4L2, ALSA.
09:49 < pinchartl> if you have a DT node with 4 ports
09:49 < pinchartl> two inputs and two outputs
09:49 < morimoto> Other device can use it ?
09:50 < morimoto> I don't know who, but not only media I think ?
09:50 < pinchartl> it could be that internally input 1 is connected to output 1 and input 2 to output 2, creating two separate unrelated pipelines
09:50 < pinchartl> when parsing the graph we'd need to know about that
09:50 < pinchartl> possibly, but at the moment only multimedia devices use OF graph
09:51 < morimoto> OK
09:51 < morimoto> So, this means, my HDMI sound can't upstream at this point, right ?
09:52 < morimoto> It will be step 3 or later
09:52 < morimoto> not step 1
09:53 < pinchartl> yes, I think we're blocked at the moment
09:53 < morimoto> OK. can you update multimedia/todo ?
09:53 < morimoto> It is not my fault :P
09:53 < morimoto> It is = delaying
09:54 < pinchartl> obviously another approach would be to use a different style of DT bindings that would be easier to handle. lots of options are possible. we're at a brainstorming stage, someone needs to drive the effort with all the upstream subsystems. it's lots of work :-/
09:54 < pinchartl> sure, it's not your fault :-)
09:55 < morimoto> Hehe :) thanks
09:55 < pinchartl> I'll remove the target version from the todo list
09:55 < morimoto> So, I will start other tasks.
09:55 < morimoto> Thanks
09:55 < pinchartl> well
09:55 < pinchartl> if you move to other tasks
09:55 < pinchartl> who will solve this problem ? :-)
09:56 < morimoto> Oops, OK
09:56 < morimoto> I am 1 of un-lucky guys :)
09:56 < pinchartl> welcome to my world :-)
09:56 < morimoto> Hehe :)
09:57 < morimoto> I think I want to talk to you in F2F in some day
09:57 < morimoto> Maybe ELC
09:57 < morimoto> ?
09:57 < morimoto> Do you go there ?
09:57 < pinchartl> I will be there
09:57 < morimoto> OK
09:57 < pinchartl> I've submitted a talk proposal, I hope it will be accepted
09:58 < morimoto> OK
09:58 < morimoto> that's it from me
09:58 < pinchartl> it's quite a lot already :-)
09:58 < pinchartl> thank you
09:58 < pinchartl> (now I have to summarize this in the report...)
09:59 < pinchartl> I assume that your plans for the next two weeks is holidays ? :-)
09:59 < pinchartl> (29th to 9th if I remember correctly)
10:00 < morimoto> Me ? yes
10:01 < pinchartl> neg: you're next
10:01 < neg> a) Started to address CSI2 review comments, more work needed.
10:01 < neg> b) Continue addressing CSI2 comments, but don't expect to resend before next meeting due to holidays and vacation.
10:02 < neg> c) None.
10:02 < neg> EOT
10:02 < pinchartl> thank you
10:03 < pinchartl> next topic: next meeting
10:03 < pinchartl> if we keep the current biweekly schedule that would be January the 4th
10:04 < pinchartl> however, Niklas and Morimoto-san will still be on holidays
10:04 < pinchartl> and Ulrich and Kieran will just be back
10:04 < kbingham> pinchartl: Bump it by a week?
10:04 < pinchartl> so I don't expect there will be much to discuss
10:04 < pinchartl> kbingham: that's my proposal, yes
10:04 < kbingham> pinchartl: then seconded :D
10:04 < pinchartl> about about January the 11th at 08:00 GMT / 09:00 
10:04 < pinchartl> CET / 10:00 EET / 17:00 JST 
10:05 < neg> works for me
10:05 < kbingham> day after my birthday ... I'll bring cake
10:05 < pinchartl> :-)
10:05 < pinchartl> how old will you be ?
10:05 < morimoto> it is OK for me
10:06 < kbingham> 0x21
10:06 < kbingham> :D
10:06 < pinchartl> :-)
10:07  * morimoto I booked Kieran's birthday on my calender
10:07 < kbingham> hehe
10:07 < pinchartl> that's all I had for today
10:08 < pinchartl> I propose adjourning the meeting
10:08 < pinchartl> does anyone second ?
10:08 < kbingham> I
10:08 < pinchartl> meeting adjourned, thank you all for attending
10:08 < morimoto> 2nd !
10:09 < morimoto> Thank you