[02:57] *** c0d1n61at3 has left [03:03] *** c0d1n61at3 has left [07:26] <paulk-leonov> hverkuil: looks like I won't make it in time for the hevc series btw, sorry about that [07:26] <hverkuil> No problem. [08:04] <hverkuil> mripard: can you review https://patchwork.linuxtv.org/patch/58388/ ? [08:05] <hverkuil> snawrocki: can you review https://patchwork.linuxtv.org/patch/58138/ ? [08:40] <snawrocki> hverkuil: done, thanks for testing it [08:51] <hverkuil> snawrocki: no problem. BTW, testing on the odroid is still a huge pain due to the fact that after using tftp in u-boot, ethernet is now dead when in the kernel is loaded. [08:52] <hverkuil> The patch series to fix that seems to have been abandoned. [08:59] <snawrocki> hverkuil: On our test farm, as a workaround, we do board reset in u-boot after uploading the kernel through tftp, mszyprow can provide you with the details on how to do this in u-boot [10:16] <dafna2> <pinchartl "the pipeline needs to start with"> So, before pluggin vimc, the configfs needs to traverse the graph and make sure that any path in the graph starts with a sensor? [10:16] <dafna2> <dafna2 "So, before pluggin vimc, the con"> this will be quit a lot af added code, other option is in the function vimc_streamer_pipeline_init, to make sure that the last entitiy added to the pipe is a sensor, and fail if it is not [10:21] <neg> hverkuil: About rcar-vin FIELD_ALTERNATE patch series, would it be possible to have your feedback on Laurent's comment on '[PATCH v2 4/6] rcar-vin: Do not reset the crop and compose rectangles in s_fmt' ? [10:25] <hverkuil> neg: done [10:26] <neg> hverkuil: Thanks! [11:43] <dafna2> <dafna2 "this will be quit a lot af added"> pinchartl: koike I think this should be checked anyway since it could be that the topology have a sensor at the start of the pipe but some of the links on the way are not enabled [12:05] <pinchartl> dafna2: the quoting in your irc client is a bit annoying :-S [12:06] <pinchartl> if it's possible for applications to disable links and create an invalid pipeline, then the right point to check pipeline validity is at stream on time [12:06] <pinchartl> and invalid pipelines should return -EPIPE [12:18] <dafna2> @pin [12:20] <dafna2> pinchartl: sorry, I'm using Riot, :\ Ill try to figure out how to remove it. so you mean not from the vimc_streamer_pipeline_init function? [12:21] <pinchartl> dafna2: I'm not sure what that function does :-) [12:21] <pinchartl> I haven't checked the internals of vimc for some time [12:22] <dafna2> it is called from the capture's vimc_cap_start_streaming [12:23] <dafna2> https://git.linuxtv.org/media_tree.git/tree/drivers/media/platform/vimc/vimc-streamer.c#n80 [12:23] <dafna2> at line 109, inside the !entity condition we should check that the last entity is a sensor [12:25] <dafna2> indeed when I run "media-ctl -d1 -l "5:1->21:0[0]" -v" to disable the only enabled link non-mutable link and then try to stream I get a crash [12:26] <pinchartl> that looks like a good place [12:37] <kbingham> to link a csi2 device to it's sensor subdevice, are there any media-controller helpers to create the pads, and link ? or do they just have to be created manually. [12:38] <kbingham> I.e., like v4l2_m2m_register_media_controller, but for a device which has a single subdev to be linked ? [12:38] <kbingham> Perhaps it's not something that can be generalised enough though ... [12:44] *** c0d1n61at3 has left [13:10] *** c0d1n61at3 has left "Leaving" [13:26] *** c0d1n61a13 has left [13:30] <dafna2> @pi [13:31] <dafna2> pinchartl: Checking if the last entity has a sink pad is make sense? so that assuming it excpect a non NULL frame [13:44] <hverkuil> dafna2: did you see my irc message I send you directly? [13:45] <hverkuil> send -> sent [13:52] *** jichang has left [13:57] *** c0d1n61at3 has left [14:03] <hverkuil> c0d1n61at3: Jiunn, do you see messages here? [14:05] <c0d1n61at3> hverkuil: i see you [14:05] <hverkuil> c0d1n61at3: I replied privately to your 'one, two, three...' message, but I guess you didn't get it? [15:33] *** john3voltas[m] has left [20:10] *** c0d1n61at3 has left [20:55] *** c0d1n61at3 has left