Re: usb webcam: infinite reset/resume bug

Ricardo Ribalda <[email protected]>
Newsgroups org.kernel.vger.linux-media
Message-ID <CANiDSCvt4f=NYCV3Y0eD=mKNkKknKv1WUz0gXcWBcrM6sW4Htw@mail.gmail.com>
Hi Flynn

Just to understand what is going on:

OBS with pipewire+v4l2 plugins -> fails
OBS with v4l2 plugin -> works
OBS with pipewire plugin -> works

?

On Sun, 9 Aug 2026 at 03:50, Flynn <[email protected]> wrote:
>
> Hi, sorry for the late reply.
>
> I could have sworn I removed it, but when I went to check the camera situation today I realized I had a Pipewire beta camera input in my OBS in addition to the V4L2 one. The Pipewire one seems to work even when the V4L2 one is fucked, and removing the Pipewire one seems to fix the V4L2 one. I really don't know how I missed this for so long. So after that discovery everything seems to work fine but I'll update you if it breaks again. Thanks for your patience
>
> Regards,
> Flynn
>
> On Fri, Jul 31, 2026 at 1:43 AM Ricardo Ribalda <[email protected]> wrote:
>>
>> Hi Flynn
>>
>>
>> On Sat, 25 Jul 2026 at 00:54, Flynn <[email protected]> wrote:
>> >
>> > Just thought I should note I was just on a Discord call using this same camera and it worked fine with no issues.
>>
>> Looks like the camera is marking frames as erroneous after you change
>> the control:
>>
>> [674830.694385] uvcvideo 1-4.3.2:1.0: Control 1/4 value change len 9
>> [674830.710385] uvcvideo 1-4.3.2:1.0: Control 2/10 value change len 7
>> [674830.725403] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>> [674830.725408] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>> [674830.725410] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>> [674830.725411] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>> [674830.725412] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>> [674830.725413] uvcvideo 1-4.3.2:1.0: Marking buffer as bad (error bit set)
>>
>>
>> So we have to debug two different things:
>>
>> - Is the error bit confusing OBS?
>>
>> You can check if that is the case by
>> rmmod uvcvideo
>> modprobe uvcvideo trace=0xffffffff nodrop=0
>> CRASH YOUR CAMERA
>> dmesg > dmesg.log
>>
>> If that works, then we need ping the obs studio guys.
>>
>>
>> - Is your camera unable to recover? (is it broken?)
>>
>> If the previous test do not work. We should probably ask Logitech what
>> is going on.
>>
>>
>> But lets go step by step and check the nodrop test.
>> Regards!
>> >
>> > On Fri, Jul 24, 2026 at 2:52 PM Flynn <[email protected]> wrote:
>> >>
>> >> Hi Ricardo, I only managed to get any output from the second test (the first one printed nothing at all to dmesg). I've attached it as dmesg.txt.
>> >>
>> >> You'll note there are two different devices doing stuff in the log. One is my laptop's inbuilt webcam and the other is the USB webcam I'm primarily troubleshooting (which I bought so I had a camera to move around). After realizing they were both using the same driver, I did another test that went like this:
>> >> - opened OBS to see USB camera input currently broken
>> >> - opened camera input properties, upon which it fixed itself
>> >> - switched to my inbuilt webcam (causing it to freeze again on a frame of the USB webcam)
>> >> - switched back to the usb webcam (it did not recover from the freeze)
>> >> I've attached that log as dmesg3.txt, though I sure don't recognize any of the specific events in it. Pretty sure the USB webcam is on 1-4 or whatever and the inbuilt is 1-13. Thanks for the help
>> >>
>> >> Regards,
>> >> Flynn
>> >>
>> >> On Fri, Jul 24, 2026 at 12:17 AM Ricardo Ribalda <[email protected]> wrote:
>> >>>
>> >>> Hi Flynn
>> >>>
>> >>> Is the error reproducible? If so, could you share the output of dmesg
>> >>> after the camera freezes?
>> >>>
>> >>> Also, try this:
>> >>>
>> >>> dmesg -c
>> >>> rmmod uvcvideo
>> >>> modprobe uvcvideo trace=0xffffffff
>> >>> CRASH YOUR CAMERA
>> >>> dmesg > dmesg.log
>> >>>
>> >>> And then share the dmesg, hopefully it has enough information for us
>> >>> to figure out what is going on
>> >>>
>> >>>
>> >>> On Fri, 24 Jul 2026 at 02:18, Flynn <[email protected]> wrote:
>> >>> >
>> >>> > Hello mailing list, please go easy on me as I've never reported a bug this way before.
>> >>> >
>> >>> > So I have a Logitech USB webcam using uvcvideo and v4l2, and I'm trying to use it with OBS Studio (the Flatpak if it matters). However, upon changing the device's video settings (possibly to an impossible/unsupported combination? i did try to increase the exposure AND set the framerate to max?), the video stream froze fully and as far as I can tell didn't recover until I manually replugged it. (Frankly I don't have the patience today to check what makes it start working again, lol)
>> >>> >
>> >>> > Expected behavior is that the device should either not fail in the first place, or should recover the first time it tries to.
>> >>> >
>> >>> > Kernel: Linux pop-os 7.0.9-76070009-generic #202605191504~1780431819~24.04~94774ef SMP PREEMPT_DYNAMIC Wed J x86_64 x86_64 x86_64 GNU/Linux
>> >>> > Distro: Pop OS 24.04 LTS
>> >>> >
>> >>> > UVCVideo quirks: 4294967295
>> >>> > modinfo uvcvideo | head -20 attached as uvcvideo.txt
>> >>> > lsusb -v -d 046d:0825 attached as webcam_info.txt
>> >>> > v4l2-ctl --device=/dev/video3 --all attached as capabilities.txt
>> >>> > v4l2-ctl --device=/dev/video3 --list-formats-ext attached as formats.txt
>> >>> > the full OBS run log is attached as obs_log.txt
>> >>> >
>> >>> > I'm ready and willing to provide further info, do more diagnostics or run test patches if it would help. Thank you
>> >>> >
>> >>> > Regards,
>> >>> > Flynn
>> >>> >
>> >>> >
>> >>>
>> >>>
>> >>> --
>> >>> Ricardo Ribalda
>>
>>
>>
>> --
>> Ricardo Ribalda



-- 
Ricardo Ribalda
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.