From 55e3b2f45880faaf06f3c660ca9e8a6d9aa14bce Mon Sep 17 00:00:00 2001 From: Kuninori Morimoto Date: Mon, 9 Dec 2019 15:29:52 +0900 Subject: wiki: Porting wiki: Porting Chat Log Signed-off-by: Kuninori Morimoto --- wiki/Chat_log/20170412-mm-chatlog | 539 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 539 insertions(+) create mode 100644 wiki/Chat_log/20170412-mm-chatlog (limited to 'wiki/Chat_log/20170412-mm-chatlog') diff --git a/wiki/Chat_log/20170412-mm-chatlog b/wiki/Chat_log/20170412-mm-chatlog new file mode 100644 index 0000000..7cffe38 --- /dev/null +++ b/wiki/Chat_log/20170412-mm-chatlog @@ -0,0 +1,539 @@ +Multimedia-chat-meeting-2017-04-12 + + good morning [15:55] + morning + hi + hello [15:56] + let's wait a few minutes for Jacopo and Magnus to join [15:57] + Morning all + 'morning + now only Magnus is missing :-) [15:58] + while we wait [16:00] +*** dammsan (~dammsan@s214090.ppp.asahi-net.or.jp) has joined channel + #periperi [16:01] + morimoto: is it enugho infomrmation about 'VIN issues on H3 ES2.0 and M3 + ES1x/M3W' to talk about it today or should we postpon it? +* morimoto Renesas talk now + hi Magnus + neg: I think we should at least mention it + let's get started + topics for today + Topic 1. Status check for the multimedia tasks [16:02] + Topic 2. VIN issues on H3 ES2.0 and M3 ES1x/M3W + Topic 3. Upstream V4L2 development + anything else ? + ok, let's start with status updates [16:03] + maybe a check up on travel plans for OSS? + jmondi: good point + Stamp of approvel for Tokyo dates, the mail thread is dead and time to + buy tickets ;-) + Ack with those :) ^ + Kieran has to run away in the middle of the meeting, so he will + get started + kbingham: the stage is yours + Why thankyou maestro :-) +* morimoto back [16:04] + So I'm now back on RCar developments and I've commenced looking at + the ADV7482 work + I have had some difficulty capturing frames through VIN so far, + getting the links set up to capture. + neg: we have no W/A at this point, so, nothing to discuss today. it + is just information sharing [16:05] + (I have tried to capture using the existing code base, using HDMI + in ) - and a RPi to generate HDMI data + But the issue is getting the formats to propogate - checking the + formats shows 'unknown' in the FMT field .. and things aren't + propogating through media ctl. [16:06] + kbingham: do you have further things to try, or are you completely + blocked ? + if it's a format propagation issue I assume you can try to debug + it + I'll have more things to try - like getting some prints in to see + why the formats aren't there [16:07] + yes - It's just a debug issue really. + But will help me walk the code base anyway + ok + kbingham: did you try h3-compliance.sh from vin-tests? + neg: Yup ! Same issue + neg: And I set my RPi dev to 640x480 as you suggested - still no + avail - so I just need to dig and understand the issue + hum that is strange, guess I need to buy a RPI today :-) + If needed, I'll talk to (neg) later about that if possible. [16:08] + you can also ask me + not that I have tried it + but I can guide you through the format propagation process and + code + pinchartl: Thankyou :D Anyway - otherwise- Once that issue is + resolved, I have started looking at how to create the subdevs, as + per our disucussions + and once I have established a baseline - I should be comfortable + hacking away at that. [16:09] + ok, thank you + first guess is it's EDID related since the ADV7482 have no EDID support + I do all my testing using a Xbox as source so I assume it fallback to + something the ADV7482 likes + that is the ADV7482 driver have no EDID support + By default, the source is coming up as [fmt:unknown/1920x1080 + field:none] [16:10] + Anyway, we can go through that later :D [16:11] + kbingham: the ADV7482 prototype should autodetect the formats, I would + like to see the fill media graphyou get if possible, but maybe you and I + can hash this out after the meeting :-) + jmondi: you're next in resumed alpbabetical order :-) + I agree - we'll go through this after the meeting rather than in + the meeting :) + fine... [16:12] + so, I started looking at moving the CEU driver away from soc_camera + I'm still evaluating if it is better to rewrite it or patch the + existing + what's your opinion at this point ? [16:13] + I'm a bit more inclined to patch the existing, because at the moment + it seems to me the driver could be implemented without sub-devices, + just with a single video device + also, I saw there is a VEU driver, which seems simple enough to use + as "inspiration" [16:14] + even if it is a mem2mem device + that sounds good to me + you need a subdev for the camera sensor + but you don't need to expose it to userspace [16:15] + yeah, camera sensor apart... + because the CEU has a format conversion unit, that can be modeled as + sub-device visibile to userspace + the atmel/atmel-isc.c driver seems to be a good example + but I guess it can be left out for now [16:16] + you don't need to, you can implement format conversion internally + yes, also + pinchartl: thanks, I would have asked you suggestions on which driver + to look at + and that's all + I would have to clarify if OF support only is enough + because the remote mingor board we should start with, has no OF + support if I'm not wrong, right Magnus? [16:17] + pxa_camera should also be an option + pinchartl: I gave a look at pxa camera, mostly for OF parsing + correct, no OF support in Migor [16:18] + you need to use platform data :-/ + that's bad, it means I have to implement both :/ + you can start with platform data, and implement OF support later + [16:19] + we should also discuss about hw setup.. but it can be post-poned + after mingor has been made at least booting mainline + sounds good to me [16:20] + how about the next two weeks ? do you plan to continue working on + this ? + I'm still intrigued byt the idea of starting with an RZ device from + day 0 + pinchartl: yes, now that my other 2 tasks for IO/core are almost + done, I'll be fully on multimedia + ok [16:21] + I'll stop here, not to steal the stage to all the others :) + one more thing, could you please remember to send your status + report by e-mail before the meeting next time ? [16:22] + jmondi: yes there is no OF support for migor + pinchartl: uh, I'll do this right away + sorry about that + jmondi: no need to for today, I'll include it in the report + but it helps if I get them in advance + and, pinchartl, I'll ping you a bit more often in next days, hope you + won't mind ;) + no problem with that :-) [16:23] + (I'll be away from tomorrow to Tues. since we're speaking) + ok, done with me + next, me + I've mostly worked on code review and various discussions, + internally and upstream + as well as additional tasks negotiation [16:24] + from a code point of view, I've tried to get the VSP rotation and + histogram support merged upstream in v4.12 + it resulted in yet another fight with Mauro + up to a point where I can't take his abusive behaviour anymore + [16:25] + I took a few days to think this over + among the various options most of them seemed unreasonable to me + such as challenging him publicly for V4L2 maintainership for + instance + Hans Verkuil proposed to me to act as an interface between Mauro + and me [16:26] + to avoid all direct interactions + I don't really think this would work, but I might give it a try + in any case, I saw no other option that stepping out from upstream + V4L2 development, at least temporarily + I will reconsider that decision if the situation changes [16:27] + (or if I find a better idea) + in effect, this means I can continue reviewing code and driving + discussions internally + including architecture discussions + but I'll remain silent publicly for now + it also means I will likely need to refrain to author any V4L2 + patch [16:28] + I still need to discuss this with Hans today, after this meeting + sounds good to me + for the next two weeks, I plan to work on the multi-camera setup + and try to get it working + VSP fences support is effectively on hold, I can't work on that + [16:29] + you can also consider putting mauro on fire + I've considered that, but I don't think it will help + he will fight back, and it would just create a bit mess, most + likely without any concrete improvement + he looks like he may burn well + probably + dammsan: let's not resort to personal non-technical attacks + [16:30] + that's it for me for today + if any of you wants to discuss this further let's do so after the + status updates [16:31] + next, dammsan + quick question, 'reviewing code internally' means no public review of + patches? + no update from my side, currently installing migor in the remote + access rack + neg: correct. let's discuss the implications after the status + update [16:32] + dammsan: thank you + I assume you'll now try to get Migor to boot mainline ? + once it is installed yes [16:33] + thank you + thanks + next, Morimoto-san + A) What have I done since last time + continue OF-graph posting. I got response from Rob, but he mentions + small details which is related to ALSA SoC, not to OF-graph. + BSP team reported Sound issue. Almost all bugs are fixed. But some + noise issue was not yet solved. I'm thinking this is related to + board specific clk issue (L/R clock gets interference from Bit + clock) + 8ch camera datasheet export paper work + created periupport script + shared H3/M3 VIN issue to you guys ;P + B) What I plan to do till next time [16:34] + continue OF-graph posting + continue sound noise issue fix + 8ch camera board check by using demo program with BSP team + C) Problems I have currently + OF-graph stalling + --EOT-- + thank you + when do you think you will be able to verify the multi-camera + setup with the demo program ? [16:35] + According to BSP team, I and Magnus will check it on this Fri + huh, good to know + =) [16:36] + :-D + dammsan: this means you need to bring it to Renesas Office on Fri + :P + I'll wait until this weekend before working on it then + could you post a status update when you will be done on Friday ? + [16:37] + Of course I will check 2 board which will be ship to Laurent/Niklas + after May + and Magnus, will you be able to reinstall it in your remote access + setup after your done, to allow me to work on it during the + weekend ? + pinchartl: sure + thank you [16:39] + pinchartl: not sure about when i will get time to do that + I think all needed 8ch camera datasheet/information exist on Wiki + now. do we still missing something ? + dammsan: could you at least send me an e-mail when you will be + done ? and also notify me if you get delayed ? + it depends on how long time is needed in the renesas office [16:40] + morimoto: I still don't know what the small microcontroller inside + the cameras does exactly. do you have any information about that ? + it's connected to the I2C control bus, and controls the sensor + power down signal + once it is working i will reinstall it in the remote access rack + [16:41] + dammsan: you don't have to work around the clock obviously :-) but + for planning purposes, it would help if you could let me know at + the end of Friday if there will be delays + pinchartl: do you mean U402 ? and I think you are asking it to IMI + guys ? [16:42] + morimoto: yes, U402. I've asked the IMI person who sent me the + schematics, but haven't received any answer + i highly doubt things will start workin on friday + OK, I will push him + and we have meetings most of the afternoon + so i think you can assume you will not get it this weekend [16:43] + dammsan: ok. let me know if that changes then + will let you know once it is reinstalled + thanks + morimoto: I see your e-mail report contains a list of patches. do + I need to include them in the meeting report, or will they be + picked up by your scripts ? [16:44] + while waiting for that answer [16:45] + neg: your turn + a) + - Fixed up some small CSI-2 stuff, patches pending waiting further + work before I post them. + - Had discussions with Laurent, Kieran and Sakari about V4L2 + extensions needed to support ADV7482. + - Took a stab at at extending the of graph framework to be able to + iterate over all DT nodes in a graph connected by endpoints. But + ceased work after the talk above since the core issue can + hopefully be solved in a better way and then there would be no + users of this. [16:46] + - Tried to help out as much as possible with the 8-channel VIN + prototype Ulrich and Laurent are working on. + b) + - Respin CSI-2 and VIN patches with my new found knowledge of V4L2 + async framework. + - Keep supporting 8-channel VIN if needed. + - Address any review comments on VIN and CSI-2 patches. + c) + - No review or other activity on VIN patches, the series [PATCH + 00/16] rcar-vin: fix issues with format and capturing' IMHO ready + to be picked up upstream but needs reviews. + Also Friday 14/4 and Monday 17/4 are holidays in Sweden and I will be + out of town at least on Friday, so my response times are reduced between + those dates. + thanks for letting us know + pinchartl: I think bugfix patches of D) ? it will be pickuped + automatically by our script. you can drop it. thanks [16:47] + those two days are public holidays in Finland too + morimoto: thank you + regarding the rcar-vin series + I can review it internally [16:48] + again, let's discuss that after the status updates + uli___: your turn + so the csi purports to capture frames from the max9286, but it + doesn't really + i tried turning off integrity checks, but to no avail + i guess the data is not valid; maybe the virtual channel setup is + wrong? [16:49] + that's pretty much it. two suggestions for friday: + 1. dump the registers of the max92* and ov10635, if it works + 2. dump the microcontroller, if you have an isp programmer lying + around + it has 8k of rom, and most of it will be ov10635 register settings + should be easy to reverse-engineer... [16:50] + uli___: do you think that U402 is an I2C master ? + I thought it was a slave + the driver code suggests so + ouch + it says that linux doesn't support multi-master busses + and that it has to wait for the attiny to finish because of that + that would require opening the camera, I'm not sure if that's + feasible [16:51] + oh, i see... + but if the microcontroller indeed configures the sensor by itself, + we might have issues, yes + well, that's it from me [16:53] + thank you for the suggestions + that's it for the status updates + Topic 2. VIN issues on H3 ES2.0 and M3 ES1x/M3W [16:54] + Morimoto-san has reported VIN-related issues in the wiki + + (https://osdr.renesas.com/projects/linux-kernel-development/wiki/R-Car-Gen3#VIN-issue) + we don't have much information at this point though + so there isn't much to discuss + but I have a question [16:55] + Yes. but it is just information, and no W/A at this point. + morimoto: do those issues impact the multi-camera setup we're + trying to get working ? + morimoto: W/A == Wanted Action? + neg: Work Around + (I assume) + yes Work Around, sorry [16:56] + thanks :-) + pinchartl: According to BSP team, upstream is supporting 2 Lane ? + 1 lane, and 4 lane have no issue + but 2 lane has issue on M3 + BSP team is using H3 [16:57] + and our remote access is using M3 + dammsan: would it be possible to switch the remote access setup to + H3 ? + pinchartl: 1 question. are multimedia people tring to 8ch directly + ? [16:58] + I though 1ch camera as fist step [16:59] + morimoto: no, we're trying to capture from a single camera first + OK + So then it is using 1lane ? maybe M3 + s/mabe M3// + 4 lane [17:00] + pinchartl: i can attach the camera board to H3 instead [17:01] + but then our ground might become more stable + i mean more unstable since it is changing + I think this can wait until Friday + sorry I was wrong. CSI40's 1lane and 2lane are NG. CSI40 4lane is + OK. CSI20 is all OK + after the hardware has been verified in the renesas office i will + use the same one for the remote access + if you get it to work on H3 on Friday but can't on M3, then we'll + switch to H3 :-) [17:02] + of course if you have any special preference then we will follow + that + I think we're done with status updates [17:03] + sorry, with VIN :-) + Topic 3. OSS Japan trip + (FYI, my talk at OSS Japan has been approved) [17:04] + we need to start booking planes and hotels + morimoto: can we proceed with Renesas' blessing ? + blessing = trip cost ? [17:05] + or meeting ? + blessing as in getting authorisation to book the trip, and + invoicing it to Jinso [17:06] + Maybe it is OK unless you enjoyed party like Oil tycoon ;P + damn, my cunning plan has been understood :-) [17:07] + we'll keep the costs down as usual [17:08] + no champagne, we'll just drink nihonshu :-) + jmondi: kbingham: neg: you can proceed booking tickets + uli___: you could too, but you mentioned you won't be able to + join, right ? + no, i won't. sorry. [17:09] + Renesas will pay up to Economy class for flight. Of course you can + upgrade, but it is your cost ;P + I assume ppl will book the conference hotel even for code camp days? + "up to Economy class" - makes it sound ike there is something + lower? =) [17:10] + dammsan: United? + haha, if you prefer to be dragged out + haha :-) + what days we said again? 26th to the 2nd? [17:11] + jmondi: 29th to 3rd + (Monday to Saturday) + I suggest arriving on the 28th at the latest + (that's what I'll do, in the morning) + and leaving on the Sunday, or later if you want to take some time + to visit Tokyo [17:12] + please be careful about the jetlag though + flights from Europe often land early in the morning + if you have trouble sleeping in the plane, the next day is very + painful + you might want to arrive on the 27th to have more time to recover + [17:13] + looking at flights now + neg: I've booked the conference hotel for the first two days too + the rate allows changes + And we will have multimedia F2F meeting on 29th 30th ? + so I might move to a different location after we decide on the + meeting location + morimoto: correct. we still haven't decided where to host that + [17:14] + OK + I would appreciate if our valued Japanese and Japanese resident + team members could advise on appropriate meetings locations + OK, I and Magnus will try it. dammsan: please say YES [17:15] + YEESS + thank you :-) + How many member ? [17:16] + it should be 4 from Europe [17:17] + Jacopo, Kieran, Niklas and me + I assume Magnus and you will join too + I don't know who else would like to attend from Japan + we will have a multimedia team meeting, but I also want to use + this opportunity to work on the multi-camera setup with local + access to the hardware [17:18] + so I'd like part of those two days to be a code camp + next topic, upstream V4L2 development [17:19] + If I understnad the mail thread corretly Wolfram won't be there so no IO + meeting to conflict with, but can't see any feedback from Geert. Anyone + know if extra days will be needed for Core meeting? + neg: I don't know. let's ask geertu [17:20] + geertu: ^ :-) + I've already explained the situation briefly. I assume some of you + may have questions + (suggestions will be appreciated too, feel free to speak your mind + honestly) + pinchartl: is the discussion you had with Mauro on linux-media list? + [17:21] + no, it was an IRC discussion on #v4l + oh, that's why :) [17:22] + it's not that discussion as such that prompted me to reconsider my + involvement, but all the interactions I've had with Mauro over the + past 12 months (at least) + this was just the last straw + First off. I think it's problematic that you step away from V4L2, even + for a while. But I understand why you do it and support it. [17:23] + neg: thank you. for the record, when I said you're free to state + your mind, I could totally understand if you (or anyone else) + couldn't understand my decision, and thought it was a very bad + idea [17:24] + how can you really step away, involved as you are (saying this from + the point of view of someone outside the v4l2 community)? + jmondi: but stopping all my involvement + not sending any patch anymore + not participating in any public discussion (mailingl list, irc, + face to face meetings) [17:25] + not reviewing any code anymore + (at least publicly) + I mean, everything you guys have discussed here has been shared to + other v4l2 members (and that's great)... will you drop everything? + I plan to continue supporting the multimedia team and every member + who works on V4L2 through internal discussions and code reviews + [17:26] + but that might be the most I will be able to do + as mentioned earlier, I will discuss this with Hans Verkuil after + this meeting [17:27] + This creates voids since you is a big part when the media group interact + with the V4L2 community, I feel at least I would need to ask you from + time to time how to best proceed with things like that even if you won't + be a part of it + he proposed acting as middleman between Mauro and me + while I don't think it would work, I could still give it a try + neg: I'll answer all questions from the team and will help as much + as I can [17:28] + The thing closest to my heart is to get the VIN Gen3 patches upstream + and my interpretation of the situation have been that Hans have waited + for you to review the patches, and now that you can't public review/ack + those patches I feel worried on how to proceed to get those patches + upstream. + but you would need to be the one taking up the fight against Mauro + publicly + I won't be able to support you when it comes to defending your + work in public + v4l3? :p + jmondi: that's also an option I've considered, but I don't think + it's feasible. I might be wrong though + ah, another point, for the record [17:30] + Google will likely continue the work on the request API + (in particular Alexandre Courbot and Tomasz Figa from the chromeos + team) + I have a meeting scheduled tomorrow with them to explain how I + would like them to proceed + but their main use case is video codecs + they have no performance-related use case like we do [17:31] + and no android stupid HAL libraries to make happy + so I fear this means that whatever will be merged upstream will + not support our use cases + we might be left with no way to get decent performances from the + VSP with upstream APIs + that would be a big gap that the BSP would need to cover + I might be painting the situation blacker than it is really, but + it's a risk [17:32] + dammsan: morimoto: you're silent, any comment or question about + all this ? + I guess not [17:35] + if there's no other question or comment, the next topic is the + next meeting [17:36] + I propose two weeks from now, same time + Wednesday 2017-04-26 at + 08:00 GMT / 09:00 CEST / 10:00 EEST / 16:00 JST + works for me [17:37] + ok [17:38] + jmondi: does it work for you too ? + I think we've lost Morimoto-san and Magnus. I'll assume it works + for them too + pinchartl: maybe we can have a short chat after you talked to Hans about + who I can best procced with getting attention on the VIN patches? + it is fine =) + neg: sure + thanks + isn't there IO meeting on the 26th? [17:39] + no, it's the 18th sorry about that! + pinchartl: works for me then + ok, it's settled then [17:40] + if there's no other question, I propose adjourning this + meeting. you can always talk to me here or in private later today + or at any other time if there's something you'd like to discuss + does anyone second ? + second [17:41] + thank you all for attending, and have a nice day + second! + thanks all, have a nice day + can I send an email to periperi to discuss about OSS trip + organization? [17:43] + jmondi: sure [17:44] + jmondi: yes, we said at 19:00 [17:54] +* morimoto I'm back from Renesas talk [18:17] + pinchartl: I will share about Request API topic to BSP team. Thanks -- cgit v1.2.3