2016-11-09 Multimedia group chat report

We had a multimedia group meeting on 2016-11-09. Here's a summary of the 
discussions. Please correct any mistake you would notice.

Attendees:

- Kieran
- Laurent
- Morimoto-san
- Niklas

Magnus and Ulrich were absent.


Topic 1. Status check for the multimedia tasks
----------------------------------------------

* Kieran

Since last meeting:

- VSP1 writeback prototype submitted

Frames can be successfully grabbed from the V4L2 video node. Awaiting reviews 
on DRI/media lists.

- Partition Algorithm improvements posted

Still need to see if/how display list objects can be reused.

For the next two weeks:

- Start on Q4/2 tasks
and/or
- Try to reuse display lists for VSP partitions

Issues and Blockers: None


* Laurent

Since last meeting:

- Attended Kernel Summit and Linux Plumbers Conference

Posted a few reports on the periperi mailing list on topics of interest for 
the upstreaming team. No more report planned, but please feel free to ask 
questions.

For the next two weeks:

- Follow up on the LVDS mode selection patches.
- Post a proposal for the rotation and histogram APIs.
- Post a few misc DU and VSP patches forward-ported from the BSP.

Issues and Blockers:


* Morimoto-san

Since last meeting:

- Posted HDMI+sound patch.
- Posted ALSA SoC OF-graph base patch-set.
- Posted ALSA SoC framework cleanup patch-set.

All the above are under review.

- Worked on BSP team sound issue
- M3 board export paper work for new guys

For the next two weeks:

- Follow up on the posted patches

Issues and Blockers:

- American people seem to have selected Mr. Trump.


* Niklas

Since last meeting:

- Posted new rewritten VIN Gen3 driver (uses media controller 
framework) with M3-W support and a refreshed CSI2 driver.

- Got a VIN->DU test-case working on Gen2 and Gen3

A VIN tests repository has been created at https://git.ragnatech.se/vin-tests/. Jacopo is also working on a buffer sharing test application that will 
support VIN, VSP and DU, we'll likely merge both.

HDMI loopback should be used for more automated testing, but it's too early 
for Gen3. We'll revisit the topic later.

For the next two weeks:

- Address the review comments on the VIN and CSI2 driver.

Issues and Blockers: None


Topic 2. Additional tasks for Q4/2 2016
---------------------------------------

SoWs haven't been finalized yet, but should hopefully be by the end of the 
week. At this time it looks like we'll have the following tasks.

- VSP1 race conditions fixes, VSP suspend/resume (Gen2 only, Gen3 requires 
more work due to FCP dependency) and Gen3 HDMI output for Kieran

- Gen3 HDMI output and V4L2 cache management API (upstream discussions only, 
actual implementation likely in 2017/Q1) for Laurent

- Nothing for Niklas as all his time will be consumed by other groups)

- IPMMU multimedia development (including DU horizontal lines issue) for 
Ulrich. Help from the hardware team might be needed, we don't know whether the 
issue is caused by software or hardware.

In parallel Jacopo is working as a trial task on a test application to measure 
performances issues, and performance improvements when patches will be ready.

Fences support isn't planned for Q4 at this time, but will be built on top of 
V4L2 fences support that Collabora is working on, likely in 2017/Q1.


Topic 3. Next meeting
---------------------

The next meeting will be held on 2016-11-23 at 08:00 GMT / 09:00 CET / 10:00 
EET / 17:00 JST.

h1. Multimedia-chat-meeting-2016-11-09

<neg> morning  [16:59]
<pinchartl> hello
<horms> hi pinchartl, i will be lurking this morning if you don't mind  [17:00]
<pinchartl> no worries  [17:01]
<morimoto> hi
<kbingham> Morning all.
<pinchartl> Magnus told me 15 minutes ago he was going back home so I assume
	    he won't join  [17:02]
<pinchartl> let's wait a few minutes for Ulrich
<pinchartl> neg: you thus have a few minutes to send your report by e-mail
								        [17:03]
<neg> :-)  [17:05]
<pinchartl> let's get started  [17:06]
<pinchartl> topics for today
<pinchartl> - Status check for the multimedia tasks
<pinchartl> - Additional tasks for Q4/2 2016
<pinchartl> anything else ?  [17:07]
<neg> not that I can think of
<pinchartl> kbingham: you can start  [17:08]
<pinchartl> kbingham: it was both a permission and a request :-)  [17:09]
<kbingham> A) - I've posted a series for review to implement a prototype
	   writeback extension to VSP1. Frames can successfully be grabbed
	   using V4L2 interfaces from the established video node.
<kbingham> hehe
<kbingham> I've also posted a couple of changes to the partition-algorithm to
	   move some of the calculations to stream on - rather than on every
	   frame - and provide better restrictions to match the hw
	   restrictions for the partition-algorithm  [17:10]
<kbingham> I've also been playing with ways to try to re-use DL objects - but
	   that's still a WIP set ...  [17:11]
<kbingham> B) - whilst lacking in budget - I will either start on Q4/2 tasks -
	   or carry on playing with DL's  - or continue working to convert my
	   garage office :)  [17:12]
<kbingham> C) No particular blockers currently.
<pinchartl> I think starting with Q4/2 tasks makes sense, let's discuss that
	    together after this meeting
<kbingham> Ack :)  [17:13]
<kbingham> That's me done then ! :-)
<pinchartl> thank you
<pinchartl> my turn
<pinchartl> I've attended the kernel summit and the Linux Plumbers Conference
								        [17:14]
<pinchartl> I've posted a few reports on the periperi mailing list on topics
	    of interest for us
<horms> pinchartl: thanks for your report on s2ram
<pinchartl> I don't plan to post any additional report, but if you have
	    specific questions, please ask
<pinchartl> horms: you're welcome
*** neg_ (~neg@unaffiliated/neg) has joined channel #periperi  [17:15]
*** neg (~neg@unaffiliated/neg) has quit: Quit: Reconnecting
*** neg_ (~neg@unaffiliated/neg) is now known as neg
<pinchartl> for the next two weeks I plan to follow up on LVDS mode selection
	    (I've received a few replies)  [17:16]
<pinchartl> post a proposal to the V4L2 mailing list to try and move forward
	    on VSP rotation and histogram API
<pinchartl> and post a few misc patches ported from the BSP for the VSP and DU
	    drivers
<pinchartl> no issue or blocker  [17:17]
<pinchartl> morimoto: you're next
<morimoto> I'm very glad to start, my lord !
<morimoto> A) What have I done since last time  [17:18]
<morimoto>   1) I posted HDMI+sound patch to ML.
<morimoto>   2) I posted ALSA SoC OF-graph base patch-set.
<morimoto>   3) I posted ALSA SoC framework cleanup patch-set. 
<morimoto> these are all under review now
<morimoto> 
<morimoto>   a) I worked for BSP team sound issue
<morimoto>   b) M3 board export paper work for new guys
<morimoto> B) What I plan to do till next time
<morimoto>  - continue to post patches
<morimoto> C) Problems I have currently
<morimoto>  - American people seems selected Mr.Trump.
<morimoto> --end--
* horms falls of chair
<pinchartl> regarded the OF-graph implementation, I think you've replied to
	    John Stulz about his ADV7511 HDMI sound patch series, right ?
								        [17:19]
<pinchartl> s/regarded/regarding/
<morimoto> Yes, I think so (I forgot his name, though)
<pinchartl> I briefly talked to him at LPC  [17:20]
<pinchartl> you're the expert on OF graph for sound :-)
<pinchartl> but please let me know if I should try to spend more time on those
	    patches
<morimoto> for DT review ?  [17:21]
<pinchartl> yes
<morimoto> If he can follow my OF-graph base HDMI sound DT, he should wait
								        [17:22]
<morimoto> because my patch-set is not yet accepted.
<morimoto> there is no responce form OF maintainer (= Rob) 
<morimoto> s/form/from/
<pinchartl> do you think he should wait, or use non OF-graph DT bindings ?
								        [17:23]
<morimoto> I think HDMI should use OF-graph DT, and this is your opinion,
	   right ?
<pinchartl> yes  [17:24]
<pinchartl> thanks for the confirmation
<pinchartl> neg: your turn  [17:25]
<neg> A)
<neg>  - Posted new rewritten VIN Gen3 driver (uses the media controller
      framework) with M3-W support and a refreshed CSI2 driver
<neg>  - Got a VIN->DU test-case working on Gen2 and Gne3 \o/
<morimoto> pinchartl: your welcome
<neg> B) Address the review comments on the VIN and CSI2 driver
<neg> C) None
<pinchartl> nice work for VIN -> DU  [17:26]
<pinchartl> it would make sense to add that to a test suite
<pinchartl> I wonder which one though
<pinchartl> maybe a new one ?
<pinchartl> Jacopo is working on a test application for buffer sharing that
	    will also support the VSP
<pinchartl> so merging the two would make sense  [17:27]
<neg> yes I have started on a vin-tests repo at
      https://git.ragnatech.se/vin-tests/
<pinchartl> what do you think ?
<neg> but it's quiet inmature and could use a lot more work
<neg> yes I spoken a bit with Jacopo and I hope it could be a good fit for
      incorperating the VIN->DU test case  [17:28]
<pinchartl> VSP tests were easier as they're self-contained
<pinchartl> but VIN unit tests would definitely be useful
<pinchartl> let's keep this in mind for now, it's a bit too early to decide
	    what to do there  [17:29]
<pinchartl> especially for Gen3
<neg> yes, maybe down the road a HDMI loopback could be used for some
      automated testing
<neg> it would be usefull so yes lets keep that in mind  [17:30]
<pinchartl> that's what I was thinking about
<pinchartl> next topic, additional tasks for Q4/2  [17:31]
<pinchartl> SoWs have still not been finalized, but we're getting close
	    (hopefully by the end of this week)  [17:32]
<pinchartl> in addition to what has already been proposed, Renesas requested
	    Kieran to help with the HDMI output upstreaming for Gen3
								        [17:33]
<morimoto> for next additional ?
<pinchartl> for second half of Q4
<morimoto> OK
<kbingham> pinchartl: OK.  [17:34]
<pinchartl> does anyone have anything they would like to discuss ?
<kbingham> pinchartl: Is there an existing branch/patchset somewhere?
<pinchartl> kbingham: yes, let's sync up after the meeting  [17:35]
<kbingham> Ok.
<morimoto> pinchartl: can I confirm (?)
<morimoto> this is not related to additional, but will be additional
<pinchartl> I'm not sure to understand the difference :-)  [17:36]
<morimoto> do you have plan for "cache less" task ?
<pinchartl> yes, I'll start discussions on V4L2 API extensions for that
								        [17:37]
<pinchartl> in parallel Jacopo is working on a test application to measure
	    performances issues
<pinchartl> and performance improvements, when patches will be ready
<pinchartl> that's all for Q4/2  [17:38]
<pinchartl> note that only API discussions are planned for Q4/2, the actual
	    implementation will likely be in 2017/Q1
<morimoto> OK, nice to know. and how about "Fence" ?
<pinchartl> fences support isn't planned yet, it will likely be for 2017/Q1
<morimoto> OK  [17:39]
<pinchartl> Collabora will likely post patches to add fence support in V4L2 in
	    the next few months
<pinchartl> so the plan is to do something else until they have posted those
	    patches, and add fence support to the VSP driver on top of them
<pinchartl> to avoid duplicating the work
<morimoto> OK, nice plan  [17:40]
<morimoto> How about "horizontal bug" ?
<pinchartl> the IPMMU issue ?
<morimoto> Maybe
<pinchartl> Ulrich will look at that, again for Q4/2
<morimoto> I wonder does he need Renesas HW guy's help ?  [17:41]
<morimoto> is it HW issue ? or SW issue ?
<pinchartl> that's a good question
<pinchartl> I don't know at this point
<morimoto> OK, I will talk with BSP team about it.  [17:42]
<pinchartl> thanks
<morimoto> and last question. do you have plan about suspend/resume bug fix ?
<morimoto> vsp1_pm_suspend()
<morimoto> Oops, it need normal suspend/resume base. maybe next year ?  [17:43]
<horms> pinchartl: i have two questions if there is time at the end  [17:44]
<pinchartl> Kieran will look at that for Q4/2 on Gen2. for Gen3 we neeed
	    additional suspend/resume core work, that will be for 2017/Q1
<kbingham> pinchartl: You know I don't have any G2 hardware right :D  [17:45]
<pinchartl> kbingham: good point :-) it can be emulated on Gen3 though, with a
	    small hack to make sure the FCP won't be suspended at the wrong
	    time  [17:46]
<pinchartl> the point isn't so much to work on Gen2 but to fix the issues
	    internal to the VSP driver
<kbingham> Ok!
<pinchartl> the suspend/resume issues caused by the dependency on FCP are out
	    of scope at this time  [17:47]
<pinchartl> horms: go ahead  [17:48]
<horms> 1. How does Jacopo fit into things? Is he working with us these days?
<pinchartl> he's working on a trial task
<horms> ok, thanks.  [17:49]
<pinchartl> his task is to write a buffer sharing application
<pinchartl> and we'll use it to measure performance issues and improvements
	    related to cache management
<horms> 2. I am wondering what the status of your upporting from BSP work
	is. Mainly because as you know its reporting season and I'd like to be
	sure I have any pending updates in my report.
<pinchartl> I have to send a report about that by 11/M  [17:50]
<horms> ok, I got a patch from him last week. I wasn't aware he was on our
	team. Its good to know.
<pinchartl> I'll submit the simple patches
<pinchartl> and will list areas that need more work
<pinchartl> based on the spreadsheet that you provided
<horms> superb, thanks
<horms> fwiw kaneko san will start analysing the 3.4.0.rc1 BSP  [17:51]
<pinchartl> by the way, the spreadsheets hosted on google docs and attached in
	    the osdr wiki are different
<pinchartl> that's a bit confusing
<horms> which iirc was released last week
<horms> ok
<horms> i'll look into that
<horms> probably best to delete the copy in the wiki as the google one keeps
	changing
<pinchartl> sounds good to me  [17:52]
<morimoto> According to Renesas guys, you can buy ULCB board by AMAZON (Now
	   Gen2, Gen3 soon). I didn't check yet
<horms> thanks, that is also good to know  [17:53]
<morimoto> pinchartl: +1 question. Is HDMI out upstreaming for long-term ?
								        [17:54]
<morimoto> I mean it takes long-term ?
<pinchartl> the idea is to post a patch series for upstreaming before the end
	    of the year
<pinchartl> it might not contain all features
<pinchartl> but it should be an upstream candidate, with no known local hacks
								        [17:55]
<morimoto> what is the difference between Ulrich's prototype ?
<horms> pinchartl: which link are you referring to on the wiki?
<pinchartl> the prototype contains hacks
<pinchartl> I'll clean that up to make it all upstreamable
<morimoto> Ahh, OK. 
<pinchartl> horms:
	    https://osdr.renesas.com/projects/linux-kernel-development/wiki/Miniperi-2016-10
								        [17:56]
<pinchartl> there's a bsp_patch_list_20160930.xlsx file attached there
<horms> ok, i see
<horms> that is the list generated by the BSP team  [17:57]
<pinchartl> ok  [17:58]
<pinchartl> I've used that as a starting point  [17:59]
<pinchartl> next topic: next meeting
<pinchartl> I proposed two weeks from now
<pinchartl> at the same time
<pinchartl> 2016-11-23 at 08:00 GMT / 09:00 CET / 10:00  [18:00]
<pinchartl> EET / 17:00 JST
<neg> that works for me
<kbingham> should be fine here.  [18:01]
<pinchartl> approved :-)
<pinchartl> I propose adjourning this meeting
<pinchartl> does anyone second ?
<neg> second  [18:02]
<pinchartl> thank you everybody for attending
<kbingham> Have a good day all!
<neg> thanks all, bye
<pinchartl> :q  [18:04]
<pinchartl> oops :-)
<morimoto> bye