RealSense Viewer does not detect camera on Linux anymore - as it did before #4916
Comments
|
@neilyoung hello, |
|
@ev-mp Thanks for the follow up. I'm currently doing again a time-intensive thing on my Windows partition, but I'm able to provide the logs later. In the meantime it might be of interest, that I'm constantly getting this error on installation (tried several times): and the mentioned make.log reports things like this: Also the final check for the kernel patch does not show the expected string: At runtime the RealSense Viewer complains about missing |
|
@neilyoung thanks, we're are aware of the issue with DKMS with the recent releases on Ubuntu 4.15 branch - #4869, #4669. But those should not affect the presence of |
|
@ev-mp Thanks. The cable is the original cable. I also tried another one already to no avail. Rest of info will follow in an hour or so. The Windows process is at 50% :/ |
|
@ev-mp OK, here we have it. Right now it is again not working.
|
|
@neilyoung , did you manually disconnect/reset the camera at this point ?
If not - it would help to connect the camera via an external self-powered USB3 hub and see whether the disconnect events still occur. |
|
@ev-mp Good point. Let me test that. |
|
@ev-mp No, doesn't work either
Yes |
|
@neilyoung , I'm not sure I understand - this mean that the disconnect was made by you, not internal event, correct ? |
|
@ev-mp Right. I just plugged the device and unplugged it again. But of course while running the viewer it was plugged :) Did the I then opened a new console and run There was no visible output in the |
|
Here is the output of |
|
I tried to work around with a fresh live install of Ubuntu 18.04, but things are getting worse.
Giving up. Camera goes back |
|
@neilyoung, thanks for the informative logs and I took the time to review it -
Ubuntu 18: - to fix dependency issues with Ubuntu distro try to apply |
|
@ev-mp Thanks, I have returned the camera. Thanks for your efforts |
|
@neilyoung , thanks for the cooperation. |
|
@ev-mp Thank you. This is for sure not the end of the story :) |
|
has there been any update regarding this issue? We are experiencing the similar crash trace after stopping the RealSense 435i using an UP board with this setup:
Here is the output of dmesg -wT that we see: |
|
Have the problem solved? I have the same problem |
|
+1 |
|
Any update on this issue? @intel: Please take a look at this issue, it really renders these devices unreliable for serious applications. Our setup: Here is a portion of the output of dmesg -wT that we see: [Sat May 21 07:47:21 2022] zram5: detected capacity change from 0 to 2092478464 Here is the complete output of dmesg -wT that we see: |
|
Issue occurs on the lastest stable Debian release, but doesn't seem to actually affect the RealSense's functionality. Below is the
|
|
Hi @ciandonovan The RealSense SDK does not yet officially support kernel 6.1. At the time of writing this, the newest supported kernel version is 5.15. It is possible to use unsupported kernels with the SDK without applying a kernel patch but there may be unforeseen consequences in regards to stability. This issue can be resolved if the SDK is built from source code with CMake with the build flag -DFORCE_RSUSB_BACKEND=true included in the CMake build instruction. An SDK build based on the RSUSB Backend bypasses the kernel, so it is not dependent on Linux versions or kernel versions and does not require kernel patching. |
|
@MartyG-RealSense this issue occurs without any components of the RealSense SDK running. I imagine it's an issue somewhere between the upstream Linux kernel driver and the RealSense's firmware. Is there a better place than this to flag this issue then? Edit: Does the RealSense SDK also refer to the Linux kernel drivers, even those upstream? |
|
There have been occasional cases where a small difference in minor kernel version numbers of the same class has dictated whether the SDK works with that kernel or not (such as kernel 5.11.28 working but 5.11.30 not working). For non-RealSense Ubuntu questions, Ask Ubuntu is a popular support forum with a long history. If the SDK is installed from packages or built from source code with a patch applied to the kernel then the SDK will be interconnected with the kernel. Such a build is called 'native backend' or 'V4L2 backend'. With the package installation, a patch does not need to be manually applied because the patch is already bundled with the packages. If the SDK is built from source code with the RSUSB Backend method then the kernel is bypassed and so it does not affect the SDK. |
|
@MartyG-RealSense just it clarify, this issue occurs without the RealSense SDK, I've tested it with a bare kernel, nothing else RealSense related installed. However the RealSense firmware is presenting itself the kernel doesn't like, specifically the V4L2 component. But again, nothing to do with the RealSense SDK as it is not installed. Edit: Are you referring to certain kernels that the Intel RealSense carries patches for? We're using an unpatched kernel. |
|
When building librealsense from source code with the native / V4L2 backend method, the kernel patch is typically applied before the SDK is installed, and the patch modifies the kernel without the SDK being involved. I note your statement above that your kernel is unpatched though. Have you tried using the lsusb command in the Ubuntu terminal to detect the camera on the USB port? If it is detected then the lsusb log should contain a line such as |
|
@MartyG-RealSense the RealSense cameras do actually function perfectly correctly on our system, it's just that particular kernel trace that looked concerning to me. It only occurs once when the camera is plugged in however. We actually run the entire userspace component of the RealSense SDK in unprivileged user namespace containers, using a patch I wrote to librealsense that's awaiting a Pull Request into upstream. So definitely no kernel patches, modules, or other RealSense SDK components on the host system due to that sandboxing, but everything is still functional regardless. Anyway I'll flag this issue with the Debian kernel team, might be under their remit. |
|
This looks like the relevant patch addressing the camera formats in v4l2-ioctl.c for the RealSense, I'll see if it resolves the issue. |
|
Thanks very much for the further details. Please do update me when you have news. |
|
Any updates on the kernel patch or Bookworm support? Encountering a similar issue on my RPI Bookworm setup |
|
Hi @hexbabe RealSense does not provide support for specific Raspberry Pi OS versions. The SDK either works with a new Pi OS version or it doesn't. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Issue Description
While having RealSense Viewer running on Windows 10 on the same machine (Asus Vivobook S) and while having had it also running on Ubuntu 16.04 RealSense Viewer is no longer able to detect the camera.
lsusbis detecting the Intel device when plugged, but the app does not.Is there any means to approach the problem?
The text was updated successfully, but these errors were encountered: