path: root/wiki/Chat_log/20180419-mm-chatlog

Multimedia-chat-meeting-2018-04-19

10:24 < pinchartl> anyway, let's get started
10:24 < pinchartl> Topic 1. Status Check for the Multimedia Tasks
10:24 < pinchartl> * Jacopo
10:24 < pinchartl> Since last meeting:
10:24 < pinchartl> - V3M Eagle display enablement
10:24 < pinchartl> Patches have been posted as "[PATCH v2 0/4] V3M-Eagle HDMI output enablement".
10:24 < pinchartl> - LVDS decoder support
10:24 < pinchartl> Patches have been posted as "[PATCH v9 0/2] drm: Add Thine THC63LVD1024 LVDS decoder bridge". Not yet but almost ready features includes
10:24 < pinchartl>   - Add static image format support to DRM bridges
10:24 < pinchartl>   - Make du_lvds use DRM bridge image format
10:24 < pinchartl> - Patch review
10:24 < pinchartl> Until next meeting:
10:24 < pinchartl> - VIN enablement on D3 Draak
10:24 < pinchartl> - soc_camera removal
10:24 < pinchartl> Issues and Blockers: None
10:24 < pinchartl> jmondi: any comment ?
10:25 < jmondi> pinchartl: not really
10:25 < jmondi> apart from DRM people being a bit unresponsive
10:25 < pinchartl> please ping them
10:25 < jmondi> and that I'm holding the DRM bridge format series beacause of that
10:25 < jmondi> I setn v9 yesterday iirc
10:25 < jmondi> I'll wait a few days maybe
10:26 < pinchartl> ok
10:26 < pinchartl> do you think it's ready to be merged ?
10:27 < jmondi> I guess so, I have collected 3 or 4 reviewed-bys
10:27 < jmondi> rob's one on bindings
10:27 < jmondi> yours and Andrzej's one on driver
10:27 < pinchartl> you can ping Archit
10:27 < pinchartl> architt on IRC
10:27 < jmondi> it should be good I guess
10:27 < jmondi> I will
10:28 < pinchartl> thank you
10:28 < pinchartl> * Laurent
10:28 < pinchartl> Since last meeting:
10:28 < pinchartl> - Patch review
10:28 < pinchartl> - Additional tasks preparation
10:28 < pinchartl> - LVDS OF overlay fixes
10:28 < pinchartl> Until next meeting:
10:28 < pinchartl> - Get the GMSL patches posted to public mailing lists
10:28 < pinchartl> - Upstream pending VSP patches
10:28 < pinchartl> - Trip to Japan (29/04 to 17/05)
10:28 < pinchartl> Issues and blockers: None
10:28 < pinchartl> any question ?
10:29 < dammsan> will you post GMSL before going to Japan? =)
10:29 < pinchartl> I want to
10:29 < morimoto> you came to Japan from 29/04 ?
10:29 < pinchartl> yes, and I will be in Tokyo from 14/04 to 17/04
10:30 < geertu>  /05?
10:30 < pinchartl> sorry, 14/05 to 17/05
10:30 < morimoto> for 1 month !?
10:30 < jmondi> 2019?
10:30  * jmondi hides
10:31 < pinchartl> landing in Haneda on 14/05 at 14:55 and leaving from Narita on 17/05 at 09:50
10:31 < morimoto> And came again in Jun ?
10:31 < pinchartl> two weeks and a half :-)
10:31 < pinchartl> yes, then coming back in June
10:31 < morimoto> sound nice
10:32 < pinchartl> I hope it will be :-)
10:32 < pinchartl> it will be my first Golden Week in Japan
10:33 < pinchartl> * Magnus:
10:33 < pinchartl> Since last meeting: None
10:33 < pinchartl> Until next meeting: None
10:33 < pinchartl> Issues and blockers: None
10:33 < morimoto> :)
10:33 < pinchartl> dammsan: any comment ? :-)
10:34 < pinchartl> * Morimoto-san
10:34 < pinchartl> Since last meeting:
10:34 < pinchartl> - Wash up random request from BSP team
10:34 < pinchartl> As far as understood, almost all BSP team requests are already solved. The only exception is "[PATCH] media: rcar-vin: Fix image alignment for setting pre clipping".
10:34 < pinchartl> Until next meeting: None
10:34 < pinchartl> Issues and Blockers: None
10:34 < pinchartl> morimoto: any comment ?
10:34 < dammsan> no comment from me
10:34 < morimoto> no comment
10:34 < pinchartl> * Niklas
10:34 < pinchartl> Since last meeting:
10:34 < pinchartl> - [PATCH v14 00/33] rcar-vin: Add Gen3 with media controller
10:34 < pinchartl> Hans has sent a pull-request for VIN Gen3 patches \o/
10:34 < pinchartl> - Found issue on v4.17-rc1 related to V4L2 and ADV748x, see blockers
10:34 < pinchartl> - Addressed all review comments on the CSI-2 driver, will repost after syncing with Laurent
10:34 < pinchartl> - Implemented Morimoto-sans request of extending media bus format support for VIN and CSI-2
10:34 < pinchartl> - [PATCH] media: entity: fix spelling for media_entity_get_fwnode_pad()
10:34 < pinchartl> Until next meeting:
10:34 < pinchartl> - VIN and CSI-2 enhancements and upport
10:34 < pinchartl> Once VIN Gen3 and/or CSI-2 driver is in media-tree start posting up-port patches and patches held back for VIN due to dependency on Gen3 patches.
10:34 < pinchartl> - Start M3-N VIN integration
10:34 < pinchartl> Issues and blockers:
10:34 < pinchartl> - CMA problems on v4.17-rc1
10:34 < pinchartl> v4.17-rc1 brings in bad8c6c0b1144694 ("mm/cma: manage the memory of the CMA area by using the ZONE_MOVABLE"). This issue brings a issue we are already aware of into a critical problem.
10:34 < pinchartl> The patch in question plays with CMA, The technical analysis is currently missing, but the result however is very noticeable. If the format configured in the media graph don't match reality more data then is allocated in the capture buffer can be written to the buffer. This was known before but this change appears to place the capture closer to other data which now easily crashes the whole system.
10:34 < pinchartl> To make matters worse there seems to be no way to set the video standard (PAL/NTSC/..) on a subdevice that is part of a media graph. This currently makes CVBS capture dangerous to to use with anything but a NTSC source as that is the adv748x drivers default standard. Niklas has developed a patch for the kernel subdevice implementation that allows for QUERYSTD, G_STD and S_STD to be controlled as for a 
10:34 < pinchartl> video device, similar work have already been done for EDID and other similar IOCTLS.
10:34 < pinchartl> Before posting the patch Niklas wanted to discuss the solution with the multimedia team. As extending the IOCTLS that are allowed for a subdev might be useful the core of the problem is still the same. If the user selects NTSC and feed it a PAL source things will crash which is not really good. One idea is to have the get_fmt/set_fmt of the AFE in the adv748x detect the current standard and return 
10:34 < pinchartl> format based on that, but we tried that in the past and it was rejected.
10:35 < dammsan> pinchartl: i want to ask you about ipmmu support at some point
10:35 < pinchartl> dammsan: sure, let's discuss that after the status update
10:35 < pinchartl> neg: any comment on done/todo ?
10:36 < pinchartl> have we lost Niklas ?
10:37 < pinchartl> seems so
10:37 < dammsan> pinchartl: just fyi there is a measels outbreak in okinawa now
10:37 < pinchartl> I'll reply to his issue by e-mail then
10:37 < pinchartl> dammsan: lovely...
10:38 < dammsan> you might want to get a vaccine if you are not covered already
10:38 < pinchartl> I think I got vaccinated when I was a kid
10:38 < dammsan> our generation of japanese are not fully vaccinated unfortunately
10:38 < dammsan> so it seems to be a bit of a mess
10:39 < dammsan> so i would enjoy seeing you _before_ your okinawa trip
10:39 < geertu> pinchartl: As a Belgian, you should have gotten either the disease, or the vaccine.
10:40 < jmondi> I got vaccinated when I was a kid, but they gave me another shot before I left for thailand
10:40 < pinchartl> dammsan: I'm afraid that would be difficult :-/
10:41 < dammsan> might be good to be careful
10:41 < jmondi> pinchartl: you may want to ask, they told me the new raccomandation is to have 4 shots not just 3 as it used to
10:41 < dammsan> pinchartl: yeah
10:41 < pinchartl> as I'll land on the 30th in the morning in Narita and leave from Haneda on the same day in the afternoon
10:41 < jmondi> ofc you'll be autistic after that according to novax.info and other reliable sources
10:42 < dammsan> pinchartl: maybe we can meet at haneda?
10:42 < geertu> jmondi: Only 2 shots for my kids, according to their vaccination card.
10:42 < dammsan> anyway lets discuss later
10:42 < pinchartl> dammsan: :-)
10:42 < pinchartl> so neg is absent
10:42 < pinchartl> let's move on
10:43 < pinchartl> Topic 2. BSP Team Requests
10:43 < pinchartl> - kms-test-brxalloc.py failures
10:43 < pinchartl> The test team noticed that kms-test-brxalloc.py failed. The problem has been reported to Laurent who only came back from holidays yesterday. It will be investigated this week.
10:44 < pinchartl> morimoto: is that ok ?
10:44 < morimoto> Not yet OK. Is this kernel .config settings ?
10:44 < morimoto> Ahh.. sorry.
10:45 < morimoto> OK, please report to me
10:45 < pinchartl> sure :-)
10:45 < pinchartl> that's all I have for questions from the BSP team, anything else ?
10:45 < morimoto> Other questions are already solved. thank you for your help M/M member
10:46 < pinchartl> you're welcome
10:47 < pinchartl> Topic 3. Discussions
10:47 < pinchartl> dammsan: you wanted to discuss IPMMU ?
10:48 < dammsan> right, sorry for the delay
10:48 < dammsan> i noticed a topic branch in latest renesas-drivers
10:49 < pinchartl> yes ?
10:50 < dammsan> drm-du-iommu-v1-20171115
10:50 < dammsan> i was wondering if there is any special target date for this one?
10:50 < dammsan> for upstream merge i mean
10:51 < pinchartl> let me have a look
10:51 < dammsan> thanks
10:51 < dammsan> just wanted to ping you about it if you had forgotten =)
10:51 < pinchartl> no there isn't. Daniel Vetter has pushed back on my proposal, I have to either convince him, or come up with a different implementation
10:52 < pinchartl> it's part of the backlog I need to handle
10:52 < dammsan> ok no stress
10:52 < pinchartl> I've decided to focus on review first in order to avoid blocking other people's backlogs :-)
10:52 < dammsan> slow and steady
10:52 < dammsan> sounds good
10:53 < dammsan> thanks
10:53 < pinchartl> you're welcome
10:53 < dammsan> and i wonder about vin for r-car gen3
10:53 < dammsan> niklas is carrying a pretty heavy burden it seems to me
10:54 < pinchartl> we should ask Niklas, but as far as I know it will be merged in v4.18-rc1
10:54 < dammsan> that would be good
10:55 < pinchartl> any other question ?
10:55 < dammsan> it almost looks like the ipmmu will be free of errata before vin gets support upstream =)
10:55 < pinchartl> :-D
10:55 < dammsan> (obviously a joke)
10:56 < dammsan> apart from that nothing special from my side
10:57 < pinchartl> if there's no other question
10:57 < pinchartl> Topic 4. Additional Tasks for 2018 Q2
10:57 < pinchartl> dammsan: have all additional tasks for Q2/1 been submitted ?
10:58 < dammsan> yes and no
10:58 < pinchartl> I like the yes part of that answer
10:58 < dammsan> paper work for jacopo has one final step that will happen tomorrow
10:58 < dammsan> and i have not seen any signed document from your side either =)
10:59 < dammsan> but apart from that we're good
10:59 < pinchartl> but no blocker ?
10:59 < dammsan> no blocker
10:59 < pinchartl> I was on holidays, hence the lack of signature :-)
10:59 < pinchartl> for Q2/2
10:59 < dammsan> i was waiting for papers myself too
10:59 < pinchartl> we have the following candidates
10:59 < pinchartl> - Jacopo: VIN scaler (UDS)
10:59 < pinchartl> - Kieran: DMA virtualization
10:59 < pinchartl> - Laurent: Display virtualization performance improvements
10:59 < pinchartl> - Niklas: VIN and CSI-2 power management
10:59 < pinchartl> - Ulrich: DU LVDS dual-link mode, DU LVDS PLL and DU DPLL fixes
10:59 < pinchartl> we need to estimate the effort
11:00 < dammsan> from my side
11:00 < pinchartl> but it's already known that VIN scaler would take more than 5 days, which is the budget available for Q2/2 for Jacopo
11:00 < dammsan> i think the virtualization bits are important
11:00 < pinchartl> that's why I've included them :-)
11:00 < dammsan> but the rest smells like best effort base task activity in my mind
11:00 < pinchartl> what's the different between base and additional from that point of view ?
11:01 < pinchartl> I mean, any work could be done under either umbrella
11:01 < dammsan> i don't want to order more VIN stuff when things are not in upstream yet
11:01 < dammsan> to give you time     to focus on getting that sorted
11:01 < dammsan> only paper work is teh difference
11:02 < pinchartl> that at least matches my understanding :-)
11:02 < pinchartl> regarding VIN
11:02 < pinchartl> a pull request has been submitted to Mauro
11:02 < pinchartl> so Gen3 support will be in v4.18-rc1
11:03 < dammsan> sounds good
11:03 < dammsan> i guess batch 1 additional adds on top of that perhaps
11:03 < pinchartl> https://www.spinics.net/lists/linux-media/msg132192.html
11:03 < pinchartl> it was sent on Monday
11:03 < pinchartl> I expect it to be merged in the very near future
11:04 < pinchartl> so I don't think it would block VIN development for Q2/2
11:05 < dammsan> well, i think we've been carrying VIN for quite a few kernel versions
11:05 < pinchartl> yes, and now it's going upstream :-)
11:05 < dammsan> i'm currently in the "i believe it when i see it" mode
11:06 < pinchartl> that's why I pasted
11:06 < pinchartl> https://www.spinics.net/lists/linux-media/msg132192.html
11:06 < dammsan> excellent - looking forward to that!
11:06 < pinchartl> :-)
11:07 < dammsan> that's good
11:07 < dammsan> so at this point i'm inclined to move forward with the virtualization tasks for you and kieran
11:07 < dammsan> i am hoping that ulrich can do some early memory autodetection hacking
11:08 < dammsan> but i'm yet to hear back from him
11:08 < pinchartl> early memory autodetection ?
11:08 < dammsan> that would take care of all members with larger additional task pools
11:08 < dammsan> yeah some packages for R-Car Gen3 come with more memory mounted
11:08 < dammsan> BSP team is busy
11:09 < dammsan> shooting themselves in the foot by duplicating DTS
11:09 < dammsan> lets see how much time he needs for the memory stuff
11:09 < pinchartl> detection in U-Boot with automatic .dtb updates ?
11:09 < dammsan> yeah something like that
11:10 < dammsan> maybe board/package specific ATF
11:10 < dammsan> and read-only memory controller configuration
11:10 < dammsan> and some magic in u-boot to populate the memory nodes during runtime
11:10 < pinchartl> ok, if it's in U-Boot it sounds good to me
11:10 < dammsan> first a prototype for ulrich
11:10 < dammsan> and if it works out well then we will involve marek and perhaps otehrs
11:12 < geertu> dammsan: U-Boot already does the /memory population magic
11:12 < pinchartl> ok, so I'll put the DU tasks on hold
11:12 < dammsan> geertu: that's good - i guess we just need to learn how to detect
11:13 < geertu> dammsan: Indeed
11:13 < dammsan> geertu: and perhaps strip away some nodes from upstream DTS at some point later on if needed
11:13 < pinchartl> last topic
11:13 < pinchartl> Topic 5. Next Meeting
11:13 < pinchartl> The next meeting will be held three weeks from now, on Wednesday 2018-05-09 at 
11:13 < pinchartl> 09:00 BST / 10:00 CEST / 11:00 EEST / 17:00 JST as a combined meeting with the 
11:13 < pinchartl> Core and I/O groups.
11:13 < pinchartl> Please note that this it the time of the multimedia portion of the meeting, 
11:13 < pinchartl> the I/O and core portions will start an hour earlier:
11:13 < pinchartl> I/O:    08:00 BST / 09:00 CEST / 10:00 EEST / 16:00 JST
11:13 < pinchartl> Core:   08:30 BST / 09:30 CEST / 10:30 EEST / 16:30 JST
11:13 < pinchartl> MM:     09:00 BST / 10:00 CEST / 11:00 EEST / 17:00 JST
11:13 < pinchartl> I will very likely not be able to join the meeting, in which case I will appoint a deputy.
11:13 < geertu> dammsan: Nah, Xen has already learned to ignore duplicate memory blocks ;-)
11:14 < dammsan> sounds good
11:14 < dammsan> geertu: yum
11:15 < pinchartl> that's all I have for today
11:15 < pinchartl> any other question ?
11:15 < dammsan> don't eat too much goya shampure in okinawa =)
11:16 < pinchartl> what is that ?
11:17 < dammsan> guya is the bitter cucumber thing
11:17 < shimoda> maybe Goya Chample?
11:17 < dammsan> shampure is like pad thai
11:17 < shimoda> ゴーヤチャンプル
11:17 < pinchartl> ah that :-)
11:18 < dammsan> anyway, enjoy your time off!
11:18 < pinchartl> I will
11:18 < pinchartl> thank you all for attending the meeting today