[09:23] <hverkuil> mchehab: ping
[09:26] <mchehab> hverkuil: pong
[09:27] <hverkuil> I received a report that em28xx is no longer loading in 5.7. It was tracked to commit a3fbc2e6bb05a3b1ea341cd29dea09b4a033727b (media: mc-entity.c: use WARN_ON, validate link pads).
[09:28] <hverkuil> That commit validates that a link is always between a source and a sink pad.
[09:29] <hverkuil> The underlying cause is that dvb_create_media_graph is trying to create a link between two sink pads: Xceive XC3028:0 -> Micronas DRXD DVB-T:0
[09:29] <hverkuil> pad 0 of the xceive tuner is a sink, not a source.
[09:29] <hverkuil> I'm not sure where in the code this is done. Can you give some pointers?
[09:30] <mchehab> bbiab
[09:35] <hverkuil> I think I see the bug:
[09:35] <hverkuil>                 pad_source = media_get_pad_index(tuner, true,
[09:35] <hverkuil>                                                  PAD_SIGNAL_ANALOG);
[09:35] <hverkuil> true should be false
[09:36] <hverkuil> it's getting the sink pad, not the source pad.
[09:36] <hverkuil> (in dvb_create_media_graph())
[09:37] <hverkuil> Yup, that's it.
[09:39] <hverkuil> I'll post a patch
[09:40] <hverkuil> Ah, this is fixed already in our master.
[09:41] <hverkuil> commit 9f984cacf4f4d53fd8a3f44d7f13528b81c1f6a8
[09:41] <hverkuil> This needs to be added to stable
[09:41] <hverkuil> Should have had a Fixes: tag
[09:42] <hverkuil> Hmm, it is likely that this will be picked up as soon as this is merged into Linus' tree.
[09:44] <syoung> sorry this is my bad. It should have gone into 5.7 as a fix.
[09:45] <syoung> Once it's merged into Linus' tree, I'll ask for it be merged into stable
[09:50] <hverkuil> syoung: thanks!
[09:51] <hverkuil> at least now I know what's wrong. The bug report came to me since if you bisect it it ends up with my commit.
[14:37] <b-rad> i cc'd you on the patch hverkuil, i guess you didn't see it. If there's a better email to use in the future just let me know
[14:42] <hverkuil> b-rad: I tend to ignore dvb related patches :-)
[14:43] <hverkuil> Either that, or I just forgot (equally likely)
[14:43] <ezequielg> hverkuil: Pleas note that I've reviewed Cedrus patch as you requested, but it seems it's not merged yet.
[14:43] <b-rad> yeh i know you do, i was hoping the media controller bit might at least catch your attention for a looksie :)
[14:44] <hverkuil> ezequielg: "[PATCH v3 2/2] media: cedrus: Implement runtime PM"?
[14:46] <ezequielg> [PATCH v3 1/2] media: cedrus: Program output format during each run
[14:46] <ezequielg> and 2/2 as well, yes.
[14:49] <hverkuil> It looks like those fell through the cracks. Only 1/2 is for stable, is that correct?
[14:52] <ezequielg> yes.
[14:52] <hverkuil> OK. I'll make a PR for 5.8 this week.
[14:57] <ezequielg> thanks a lot.
[15:02] <hverkuil> if not this week, then early next week. But I hope I'll have time tomorrow.
[15:07] <ezequielg> there's no hurries on my side.
[15:07] <ezequielg> and it's going to -stable, so...
[19:16] <ezequielg> Kwiboo: in case it wasn't clear from my review, the rkvdec scaling fix looks good, i just would like to do the naming fix for clarity.
[19:49] *** benjiG has left 
[23:43] <styl1te> I'm trying to use the v4l2loopback module to create a fake webcam device to share my screen, as described here: https://superuser.com/questions/411897/using-desktop-as-fake-webcam-on-linux  However, cheese won't read the device, and guvcview says 'V4L2_CORE: V4L2_CAP_TIMEPERFRAME not supported'