diff options
Diffstat (limited to 'wiki/Chat_log/20161222-mm-chatlog')
| -rw-r--r-- | wiki/Chat_log/20161222-mm-chatlog | 212 |
1 files changed, 212 insertions, 0 deletions
diff --git a/wiki/Chat_log/20161222-mm-chatlog b/wiki/Chat_log/20161222-mm-chatlog new file mode 100644 index 0000000..11b2ea6 --- /dev/null +++ b/wiki/Chat_log/20161222-mm-chatlog @@ -0,0 +1,212 @@ +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 |