| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| *** | ChanServ sets mode: +v mchehab | [03:29] |
| ....................................................................................... (idle for 7h14mn) | ||
| hverkuil | sailus: is the imx v12 series now OK? | [10:43] |
| ...... (idle for 29mn) | ||
| sailus | hverkuil: Good question.
I think so. I'll send my ack. But... is there a plan to get the driver out of the staging tree? That'd be nice to see happen. | [11:12] |
| hverkuil | OK, then I'll go over it one more time and if I don't see anything weird I'll make a PR.
Yeah, we need to see what is still blocking it from being mainlined. | [11:13] |
| sailus | I guess it hasn't changed over the time, uAPI semantics at least.
There are dependencies between links and controls as well ass media bus codes and pixel formats. These should be put to the TODO file unless they're there already. | [11:14] |
| hverkuil | Can you take a look at drivers/staging/media/imx/TODO?
I think the first two points are not blocking (can always be addressed later as it is internal to the kernel). | [11:18] |
| sailus | The format thingy doesn't seem to be there.
Indeed. | [11:19] |
| hverkuil | I will have to think (again) about point 3 since it is really not quite how it should work but perhaps it can be accepted.
I'm not sure about the last point, you might be able to say more about that. It looks like an imx driver limitation to me, but not blocking mainlining. I.e. something that may have to be addressed later. | [11:20] |
| sailus | The control handling is a hack, there's no way around that.
I'd like to see it go. Instead we needed a helper library in the user space. | [11:21] |
| hverkuil | libcamera, basically. | [11:22] |
| sailus | Yes, libcamera is much more than that though. | [11:22] |
| hverkuil | I want to sit down and think some more about this and see if there isn't something that can be done to simplify this. I do understand why it is done like that in imx, since it is a pain in the posterior to have to go through subdevs, but the current implementation simply won't work in general.
I had some ideas floating in my brain, but I don't know if any of them are feasible. need to sit down for that. | [11:26] |
| sailus | hverkuil: Do you have something user space related in mind?
I think we needed better libraries for the user space. Not a single big one, but a handful of basic building blocks that can be used to create the functionality needed without making use case specific changes to the kernel. | [11:39] |
| hverkuil | pinchartl: regarding sailus' comments above: wasn't that planned as part of libcamera? I think that was discussed at some point at least.
sailus: one thing that can help when fiddling around with this on the command line is if v4l2-ctl had an option that could show all the subdevs in a video pipeline ending at a video device. Basically traversing up the graph and showing all the subdevs. It's a bit complex right now to determine that. (a lot complex, actually) | [11:42] |
| sailus | Yeah, it'd be good to use common code for parsing the graph, for instance.
That'd make the implementation quite a bit easier. | [11:47] |
| hverkuil | I still think that the data returned by g_topology can be much more useful. I needed to determine which device node was associated with an entity, and instead of just getting all the interface links with G_TOPOLOGY and just searching for interfaces connected to that entity you need to get the pads, links and interfaces :-(
Anyway, that's another discussion. | [11:53] |
| pinchartl | hverkuil: what's the qeustion ?
what is planned as part of libcamera ? | [11:58] |
| hverkuil | Is there a low-level library planned that provides basic helper functions to e.g. traverse media controller graphs? | [11:59] |
| pinchartl | we have some helpers related to that in libcamera, but that's internal, and in C++
a mediactl library would make sense | [12:00] |
| ....... (idle for 32mn) | ||
| *** | sailus has quit IRC (Quit: leaving) | [12:32] |
| ................................. (idle for 2h40mn) | ||
| mchehab has quit IRC (Ping timeout: 246 seconds)
ChanServ sets mode: +v mchehab | [15:12] | |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |