uvcvideo: Failed to set UVC probe control after a few times using #618
Assignees
Comments
|
I should get this macroed: Please post a full How long is it between the initial use of the camera and when it first stops working? |
|
Was working for about 50minutes being up. Mpg-streamer was launched at least 5x, judging the jpg's it streamed. The jpg's are about 44kb, the frame rate is set to 2. About 20s long per mjg-stream session. The streaming itself is to a remote PC via openvpn. Below a dmesg after failure, one lsusb -v as user after failure and one as root after a reboot. dmesg after failure lsusb -v after failure as user |
|
Oops, wrong button :( |
|
Your lsusb details two quickcams attached - does the fault occur with both webcams attached or just one? Also, what capture method is used? Do you know if the MJPEG transport is used? |
|
Only one webcam, the other is a logitech wireless kb receiver (K400) |
|
Can you repeat the test after adding |
|
Survived an hour, also stressed it a bit more. Okay, looks I have a working setup with that kernel param. Of course it needs a few days more to be a bit more certain. |
|
Ok. Further to the format/bandwidth enquiry: can you either
and post a dmesg log. The UVC driver should spit out some more info about what bandwidth it selected. |
|
Camera still working btw, only got one line extra in dmesg yesterday night: [ 3518.656783] uvcvideo: Non-zero status (-5) in video completion handler. Did the rmod/modprobe thing, these lines were added to dmesg output: [39932.919609] usbcore: deregistering interface driver uvcvideo |
|
So MJPEG is being used as the stream with 192-byte frames. When the device starts to time out, do other USB devices stop working or is it just the webcam that fails? |
|
Wireless mouse and network still work, so just webcam I would say. |
|
Still running fine using the dwc_otg.fiq_fsm_enable=0 kernel param. I think we can conclude that not having dwc_otg.fiq_fsm_enable=0 is problematic for this use case. |
|
To narrow it down more, can you remove This should prevent the FIQ from performing isochronous transactions and will narrow it down further. |
|
So far it looks stable too. |
|
Ok, so the FIQ acceleration has the unintended consequence of making the camera disappear after a while. Weird, because all the webcams that I tested this with worked fine for extended periods. |
|
@koosvriezen has this issue been resolved? If yes, then please close this issue. |
|
Okay, I removed the mask option and rebooted, current stable raspian. I'll close the ticket if still working after a few weeks. |
|
No, I still need the Without this option I noticed that with the current kernel from stable |
|
Indeed the camera exposure settings is not related. I simply had to upgrade mjpg-streamer and use the |
|
Runs fine now for a few weeks, so closing this ticket. |
pfpacket
pushed a commit
to pfpacket/linux-rpi-rust
that referenced
this issue
Apr 7, 2023
rust: convert `platform` to use `driver`
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Using an alarm system that launches mjpg-streamer for a short period, after a while, maybe 5-10 times, I see these errors in dmesg
uvcvideo: Failed to set UVC probe control : -110 (exp. 26)
And next the device seem to be dead on any future mjpg-streamer launch
Here I think the relevant dmesg from the kernel log loading the driver
Linux video capture interface: v2.00
uvcvideo: Found UVC 1.00 device (046d:09a4)
input: UVC Camera (046d:09a4) as /devices/platform/bcm2708_usb/usb1/1-1/1-1.3/1-1.3:1.0/input/input2
usbcore: registered new interface driver uvcvideo
USB Video Class driver (1.1.1)
set volume quirk for QuickCam E3500
usbcore: registered new interface driver snd-usb-audio
I have an up-to-date raspian system, stable branch, updated with rpi-update
Linux raspberrypi 3.12.21+ #688 PREEMPT Wed Jun 4 20:47:24 BST 2014 armv6l GNU/Linux
Tried setting the uvcvideo quirks to 0x100:
media: Linux media interface: v0.10
Linux video capture interface: v2.00
bcm2708-i2s bcm2708-i2s.0: Failed to create debugfs directory
uvcvideo: Found UVC 1.00 device (046d:09a4)
uvcvideo: Forcing device quirks to 0x100 by module parameter for testing purpose.
uvcvideo: Please report required quirks to the linux-uvc-devel mailing list.
input: UVC Camera (046d:09a4) as /devices/platform/bcm2708_usb/usb1/1-1/1-1.3/1-1.3:1.0/input/input2
usbcore: registered new interface driver uvcvideo
USB Video Class driver (1.1.1)
set volume quirk for QuickCam E3500
usbcore: registered new interface driver snd-usb-audio
but that didn't help (maybe made it worse).
I happen to launch xbmc when this camera was stuck (but no mjpg-streamer was being launch or any other access to the camera being made), xbmc couldn't get its display up and the kernel log got a bunch of entries:
5:3:1: usb_set_interface failed (-110)
Given that this alarm system setup, with same camera used to work a while back when rpi was still on the 3.6 kernels (don't have any specifics date when exactly) and the camera works initially, I wonder whether this has to do with the usb driver rewrite.
The text was updated successfully, but these errors were encountered: