[01:08] *** fixxxermet has left [09:44] <hverkuil> mchehab: the media-commits mailinglist is missing some commits (e.g. a5d42b8c3b3ddccd88dc1c70957177d31a6699fb, "vivid: Fix iteration in driver removal path"). [09:45] <hverkuil> Do you need to run some script to generate the mails to that ml? [09:54] <mchehab> hverkuil: no, it is a commit hook at git [09:55] <mchehab> (or an update hook - need to double check) [09:55] <mchehab> anyway, the server runs automatically when patches are merged [09:58] <mchehab> the script is based on some hook from an earlier version of git... [09:59] <mchehab> I had to modify it to avoiding mailbombing people when I'm pulling back from upstream [10:00] <mchehab> (basically, I added a logic there that avoids sending patches if more than a certain number of changes got merged on a single git push) [10:00] <mchehab> the number is configurable via a git var [10:01] <mchehab> There are actually two configs there: [10:01] <mchehab> maxsize = 100000 [10:01] <mchehab> maxmsgs = 100 [10:01] <mchehab> maxmsgs is the max number of patches [10:01] <mchehab> maxsize is the max size of a message [10:02] <mchehab> if the size is too big, it discards the diff [10:02] <mchehab> but sends the email [10:04] <hverkuil> Hmm, commits f4f24d1fd0803e8b5de5da373276f5046bef7463 to 2788052a8ac9326723c8526f9c5953b8cfff8ab5 are missing from media-commits. [10:05] <hverkuil> BTW, I plan on creating the first draft of the media workshop today & review the vb2 split patches. [10:06] <hverkuil> My laptop died two weeks ago while in Ireland, so I haven't been able to do the vb2 review earlier :-( [10:07] <mchehab> :( [10:07] <mchehab> weird that those patches got missed [10:07] <mchehab> just 7 patches... not enough to hit the 100 patches threshold [10:08] <hverkuil> I wouldn't have noticed if it wasn't for the vivid patches. [10:10] <mchehab> with regards to the workshop, the only thing related to MC from my side is to discuss about dynamic support (basically, removing entities/interfaces)... [10:10] <mchehab> we may need to support dynamic number of PADs on some future, but I'm not working on this yet [10:11] <mchehab> yet, I'm not sure if we should discuss MC dynamic support... as it seems that pinchartl and sailus_ didn't finish reviewing what we have there already [10:12] <mchehab> so, not much sense on getting further while we don't finish what's there [10:13] <mchehab> so, please put this theme to the end [10:13] <hverkuil> will do. [10:13] <hverkuil> When I have posted the draft agenda, can you make a link available somewhere so people who register for the workshop know what will be discussed? [10:14] <mchehab> yeah, I can publish the agenda at the news section of linuxtv.org [10:15] <hverkuil> Actually, how to do this in the future would be a good topic for the workshop as well. We should look at how other subsystems organize such workshops. [10:16] <mchehab> yeah, we can add a topic to discuss improvements to the next workshops [10:16] <mchehab> I know that ALSA uses Google drive for people/themes... [10:17] <mchehab> like we're doing this time [10:17] <mchehab> it would be great to know if this is working well or not [10:17] <mchehab> about that, Junghak was unable to access it, because of some firewall restrictions where he sits [10:18] <mchehab> (I had to add his proposal there, and add the names from people on his office that will be joining us) [10:18] <mchehab> not sure if someone else had the same problems [10:20] <mchehab> I need to go. Today is a Holiday here... I won't stay available most of the day [10:22] <hverkuil> enjoy your day! [10:34] <mchehab> thanks! [10:35] <mchehab> ah, one last thing: please add a topic from my side about documentation... [10:35] <mchehab> I'd like to discuss some things about the work I've done to improve the kABI documentation there [10:36] <hverkuil> ok [10:37] <mchehab> thanks! [10:37] * mchehab is gone [15:56] *** benjiG has left [16:08] <sailus_> mchehab: There's another patch that's needed on top of your set to make drivers such as smiapp function. [16:09] <sailus_> I'm still trying to figure out which patch causes the issue to begin with. I guess it's related to turning the links a linked list instead of an array. [16:23] <mchehab> sailus_: what's happening? [16:26] <sailus_> Well, a NULL pointer exception on my N900. [16:26] <sailus_> I have to go for a moment. [16:31] <mchehab> sailus_: where? did you have the OOPS? [16:31] <mchehab> javier__ tested the patches with smiapp/omap3 and fixed the issues he found [16:32] <mchehab> the issues were basically because the links were created before the entity register [16:33] <mchehab> maybe he lost some special case that the N900 code hits [16:39] <javier__> sailus_: what media-ctl pipeline are you using and how are you testing? I started testing omap3 isp video capture by faking having a smiapp camera sensor in the DT [16:40] <javier__> that's how I found the pad links creation before media entity registration and the bugs in the graph traversal logic in the first versions [16:42] <javier__> mchehab, sailus_: but for the real video capture, I used a board that has a tvp5150 video decoder connected to the OMAP3 ISP [16:42] <sailus_> I'm back now; Agnes is sick and her tablet on which she watches cartoons run out of battery. [16:42] <sailus_> mchehab, javier__: The crash happens with the extra patches. [16:43] <sailus_> The underlying problem is that the smiapp driver does register additional entities in its registered() callback. [16:43] <sailus_> Other sub-device drivers I'm aware of don't need that. [16:44] <sailus_> I think the real issue was that a link was created between a registered and an unregistered entity. [16:44] <sailus_> So the patch I have registers the media entity before calling the registered() callback. [16:44] <mchehab> sailus_, hverkuil: just mailbomb the 83 patch series to the Workshop ML. [16:44] <sailus_> I'm not that certain it's a good solution but nevertheless it resolves a problem. [16:45] <sailus_> mchehab: It's not prohibited to send large number of patches. :-) [16:45] <mchehab> sailus_: I don't want anyone to put a prize on my head for mailbombing 83 patches that are already at the ML [16:45] <mchehab> :) [16:46] <mchehab> nor people leaving the ML [16:46] <mchehab> we had already people leaving the ML due to large mailbombs [16:46] <javier__> sailus_: I see, that explains why I didn't see it since I hacked smiapp_probe() to be almost a no-op (besides registering the entities [16:46] <sailus_> Most MUAs can mark entire threads read by a single key combination. [16:46] <mchehab> in the past [16:47] <sailus_> Then there are people using MUAs that do not support threads... [16:47] <javier__> sailus_: later when I got access to the board with the tvp5150 video decoder, I could test that video capture actually worked [16:47] <mchehab> sailus_: likely yes... [16:47] <sailus_> The workshop list is fine. [16:48] <mchehab> but there were cases of people that got their mailbox full due to such mailbombs [16:48] <sailus_> X-) [16:48] <sailus_> I think patches should no longer be a problem. [16:48] <sailus_> Not nowadays. [16:48] <mchehab> maybe [16:48] <sailus_> At least I get multi-megabyte Power point presentations and such monsters weekly. [16:49] <sailus_> Not that my preference would be that they were so big, but anyway... ;) [16:49] <mchehab> well, my old employer account had a very low limit at the mailbox [16:49] <mchehab> that was 2 years ago [16:49] <sailus_> I had my mail server reject messages larger than 10 MiB or such. [16:50] <sailus_> I had to increase that to 20 MiB recently for practical reasons. [16:50] <mchehab> the presentation I sent to Tizen Dev Conference were higher than their limit [16:50] <mchehab> I had to use a high compression at the PDF [16:50] <mchehab> reducing a lot the resolution for it to arrive there [16:51] <sailus_> mchehab: Thanks for resending the set! [16:51] <mchehab> anytime [16:53] <mchehab> so far, only 21 patches arrived here from the ML [16:54] <mchehab> sailus_: did the 83 patches arrive to you? [16:54] <mchehab> 84 emails, actually [16:55] <sailus_> I'm getting them bit by bit. [16:55] <sailus_> I guess they'll eventually arrive. [16:57] <javier__> most of them already arrived here [17:14] <sailus_> mchehab: Sent. [17:17] <mchehab> sailus_: thanks! looks ok to me [17:57] <mchehab> sailus_: could you please send us the .dot produced by mc_nextgen_test? [17:58] <mchehab> http://git.linuxtv.org/cgit.cgi/mchehab/experimental-v4l-utils.git/tree/contrib/test/mc_nextgen_test.c?h=mc-next-gen&id=91a479e9f998cba4754e182b7baefcca731b59f5 [17:58] <mchehab> on the N900 [22:16] <x41> Is there a way to have multiple programs capture from one webcam? [23:16] <pinchartl> hverkuil: have you ever noticed that adding device_caps broke userspace ? [23:16] <pinchartl> I've only noticed it now :-) [23:17] <pinchartl> the OMAP3 ISP driver was setting the capabilities field to per-device-node caps [23:17] <pinchartl> after adding device_caps the capabilities field became the overall device caps [23:17] <pinchartl> so it moved from being either V4L2_CAP_VIDEO_CAPTURE or V4L2_CAP_VIDEO_OUTPUT to being V4L2_CAP_VIDEO_CAPTURE | V4L2_CAP_VIDEO_OUTPUT [23:18] <pinchartl> and here's my userspace [23:18] <pinchartl> if (cap.capabilities & V4L2_CAP_VIDEO_CAPTURE) [23:18] <pinchartl> dev->type = V4L2_BUF_TYPE_VIDEO_CAPTURE; [23:18] <pinchartl> else if (cap.capabilities & V4L2_CAP_VIDEO_OUTPUT) [23:18] <pinchartl> dev->type = V4L2_BUF_TYPE_VIDEO_OUTPUT; [23:46] <pinchartl> second userspace breakage with the OMAP3 ISP driver, my fault this time [23:46] <pinchartl> when adding crop support for the CCDC [23:46] <pinchartl> my userspace code does something stupid and breaks [23:51] <javier__> pinchartl: what user-space tool are you using ooi? I could capture frames using your yavta tool just fine [23:51] <javier__> at least I didn't notice any issues [23:53] <pinchartl> http://git.ideasonboard.org/omap3-isp-live.git [23:53] <javier__> pinchartl: cool, I didn't know about this one. Thanks a lot for the pointer [23:53] <pinchartl> you're welcome [23:54] <pinchartl> it needs love :-) [23:54] <javier__> :) [23:55] <pinchartl> the application uses omapfb and omap_vout [23:55] <pinchartl> it should be ported to omapdrm [23:57] <pinchartl> mchehab: how long do you plan to stay in Seoul ? [23:58] <mchehab> I should be returning on Oct, 30, I guess [23:58] <mchehab> why? [23:59] <pinchartl> I wonder if I should plan a day to work on MC on Thursday [23:59] <pinchartl> or rather whether you plan to lock me up in the Samsung HQ for that ;-)