#linux-media 2025-02-11,Tue

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
***danitool has quit IRC (Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos) [00:41]
...... (idle for 26mn)
Ampera has quit IRC (Remote host closed the connection)
Ampera has joined #linux-media
[01:07]
............... (idle for 1h14mn)
smlavine has quit IRC (Remote host closed the connection)
jluthra has quit IRC (Read error: Connection reset by peer)
ezequielg has quit IRC (Read error: Connection reset by peer)
jluthra has joined #linux-media
ezequielg has joined #linux-media
[02:22]
.... (idle for 16mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[02:40]
Mo_ has joined #linux-media [02:54]
Mo has quit IRC (Ping timeout: 480 seconds) [03:00]
................. (idle for 1h21mn)
thansen has quit IRC (Quit: The Lounge - https://thelounge.github.io)
thansen has joined #linux-media
[04:21]
......... (idle for 40mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[05:04]
Mo_ has quit IRC (Remote host closed the connection) [05:13]
............. (idle for 1h0mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[06:13]
Mo has joined #linux-media [06:26]
.... (idle for 18mn)
wenscould someone merge the remaining media dt-binding patches in https://patchwork.linuxtv.org/project/linux-media/list/?series=13680 ? [06:44]
....... (idle for 34mn)
***mvaittin has joined #linux-media
tmerciai has joined #linux-media
tmerciai has quit IRC (Remote host closed the connection)
tmerciai has joined #linux-media
[07:18]
jmassot has joined #linux-media [07:33]
....... (idle for 34mn)
ao2 has joined #linux-media [08:07]
GBenji has joined #linux-media
djrscally has joined #linux-media
[08:12]
......... (idle for 44mn)
nas_chung has joined #linux-media [08:58]
linkmauve has left
linkmauve has joined #linux-media
[09:05]
..... (idle for 24mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
paulk-bis has quit IRC ()
paulk has joined #linux-media
[09:29]
mvaittin has quit IRC (Ping timeout: 480 seconds)
frieder has joined #linux-media
[09:41]
....... (idle for 31mn)
paulk-bis has joined #linux-media
paulk has quit IRC (Ping timeout: 480 seconds)
[10:14]
tleb has quit IRC (Remote host closed the connection)
mvaittin has joined #linux-media
[10:21]
paulk-bis has quit IRC ()
paulk has joined #linux-media
[10:35]
......... (idle for 42mn)
tonyk has joined #linux-media
tonyk has left The Lounge - https://thelounge.chat
frieder has quit IRC (Ping timeout: 480 seconds)
[11:17]
frieder has joined #linux-media [11:29]
...... (idle for 29mn)
tleb has joined #linux-media [11:58]
....... (idle for 32mn)
dcz_ has joined #linux-media [12:30]
dcz_what does it mean for a video driver to be "video-node-centric"? I can't find a way to relate mbus formats to fourcc codes in that case, does it mean there will be no configurable link to the video device?
can I assume anything about the topology of such a media device or the existence of subdevices?
are there any other such drivers apart from uvc?
uvc seems to have a pretty rigid topology, is it the same as being video-node-centric?
sorry for too many questions, I'm getting confused about the generality of the interace as I try to implement a userspace driver
[12:33]
***mvaittin has quit IRC (Ping timeout: 480 seconds) [12:40]
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[12:48]
frieder has quit IRC (Ping timeout: 480 seconds)
BrianG61UK has joined #linux-media
dcz_ has quit IRC (Read error: No route to host)
dcz_ has joined #linux-media
frieder has joined #linux-media
[13:02]
jmondidcz_: a "video-node centric" driver means that all the confgiguration and control of the camera pipeline is performed on a single video device node (/dev/videoX) and the driver internally propagate the settings to other entities in the system. video-node centric drivers are sometimes (maybe unfairly) called "legacy" as they offer a simplified view of the system where none of the
subdevices part of the pipeline are exposed to userspace. It is not a suitable design for modern camera systems where multiple entities concur in realizing the capture pipeline (generally at least an image sensor, a csi-2 reciver, some form of ISP and dma engines) while it is valid (afaict) for USB cameras as all the image processing is performed on the camera module rather than
being realized by a pipeline of processing blocks on the SoC
[13:19]
dcz_thanks, that's what I was missing. On the first read, it looked like it was just a single property that was affected [13:22]
jmondiI presume modern uvc cameras do anyway expose a media topology to allow enumerating the entities in the system, but they're anyway controlled by /dev/video0 only [13:23]
dcz_what's the point of that? [13:23]
jmonditopology enumeration you mean ? [13:23]
dcz_I have 3 entities but I haven't found a way to reference them directly
yes
[13:23]
jmondiI'm abolutely not a uvc expert, but I presume modern webcam have additional features exposed as additional video devices/subdevices
you can have additional data streams like a point cloud to represent depth maps in examples, and I presume in UVC they get represented as additional video devices
[13:24]
dcz_I have one device with just metadata here [13:26]
jmondibut again, I know nothing about UVC so take this as my own understanding only [13:26]
dcz_thanks
I have found some UVC ioctls for controls, but there doesn't seem to be a connection to specific entities (but I don't understand how to use them anyway)
different question: does all this mean that a MC-centric driver will have a subdevice node for every entity, or I cannot make that assumption?
[13:26]
jmondidcz_: it really depends on the platform. Components in the system which needs to be configured with specific settings to realize a function should be exposed to userspace
but there's no requirement to have one subdev for each component of the pipeline if the component is pass-through or has unmodifiable settings I think
(or simply the driver does not support those functions)
(however in this case, my suggestion would be to expose the full topology from the very beginning so that userspace doesn't need to be completely rewritten when a new function/block is added to the driver)
dcz_: and for controls (if you're talking about v4l2-controls) for video-centric drivers the controls of the internal entities are collected and exposed through the control handler of the video device
[13:28]
dcz_jmondi: I meant the special UVC controls that can be mapped to v4l2 controls https://www.kernel.org/doc/html/next/userspace-api/media/drivers/uvcvideo.html#introduction
I tried to run someone's software to take a look at this but didn't get anywhere
[13:33]
jmondisorry no idea, seems like a uvc specific feature [13:34]
dcz_thanks for the explanations [13:34]
jmondiyw [13:35]
dcz_it's awkward for the model I'm using to not be able to query some pads on the path, so knowing that I can't make assumptions here is useful [13:35]
pinchartldcz_: the UVC video node exposes an ioctl to get/set controls on XUs. the media graph is there to allow enumerating the XUs
and what do you mean by "implement a userspace driver" ?
[13:40]
dcz_by analogy to mesa, there's the kernel driver and the user driver to every complex device
in libcamera they are called pipelines IIRC
pinchartl: oh, now I vaguely recall, getting from the entity to the XU involved going through the USB /sys hierarchy
[13:42]
jmondimaybe there's something I don't know about uvc, but does it require "a driver" since most of the image transform happen "magically" inside the camera itself ?
* "a -userspace- driver"
(yes, we have a pipeline for uvc in libcamera, but offers nothing more than what an application could do by poking at /dev/video0 directly)
[13:49]
dcz_I think of a driver as something that abstracts a complex interface rather than offers more possibilities [13:55]
jmondithat's my point.. what's complex about UVC ? :)
(from a userspace perspective I mean)
[13:56]
dcz_nothing. it's having a unified interface across different things of the same type (cameras) that warrants a driver
otherwise, you can bang MMIO registers and never need a driver for anything
[13:57]
jmondieheh true that
ah well, if you're aiming at complex devices, having uvc as a bonus is perfectly fine
[13:58]
........ (idle for 37mn)
***mvaittin has joined #linux-media [14:35]
...... (idle for 25mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[15:00]
Mo has quit IRC (Ping timeout: 480 seconds) [15:07]
mvaittin has quit IRC (Ping timeout: 480 seconds) [15:12]
.............. (idle for 1h5mn)
frieder has quit IRC (Ping timeout: 480 seconds) [16:17]
PedroAmarante has joined #linux-media
prabhakalad has quit IRC (Remote host closed the connection)
[16:24]
frieder has joined #linux-media [16:29]
prabhakalad has joined #linux-media [16:35]
frieder has quit IRC (Remote host closed the connection) [16:48]
.......... (idle for 45mn)
GBenji has left [17:33]
.......... (idle for 46mn)
ao2 has quit IRC (Quit: Leaving) [18:19]
PedroAmarante_ has joined #linux-media [18:28]
PedroAmarante has quit IRC (Ping timeout: 480 seconds)
danitool has joined #linux-media
[18:36]
.... (idle for 19mn)
tmerciai has quit IRC (Ping timeout: 480 seconds) [18:55]
...... (idle for 27mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[19:22]
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[19:30]
..... (idle for 24mn)
dcz_ has quit IRC (Quit: Konversation terminated!)
BrianG61UK has quit IRC (Read error: Connection reset by peer)
[19:56]
..... (idle for 20mn)
BrianG61UK has joined #linux-media [20:18]
jmassot has quit IRC (Remote host closed the connection) [20:27]
..... (idle for 20mn)
Mo has joined #linux-media [20:47]
............ (idle for 59mn)
PedroAmarante_Hey, anyone still talks here? [21:46]
djrscallyPlenty [21:52]
***BrianG61UK has quit IRC (Remote host closed the connection) [21:55]
Mo has quit IRC (Remote host closed the connection) [22:00]
..... (idle for 24mn)
ten157237743246305066182150355 has quit IRC (Remote host closed the connection)
ten157237743246305066182150355 has joined #linux-media
[22:24]
......... (idle for 41mn)
PedroAmarante_ has quit IRC (Ping timeout: 480 seconds) [23:07]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)