mostly silent failure in uvcvideo patch process #70
Comments
|
Hi, I strongly recommend using a Windows machine to update your firmware. 1.0.72.06 is the latest and 1.0.71.xx has known issues. Installing the DCM on a Windows 10 machine with a camera plugged in will update the firmware. If you don't have Win10 handy, I can dig up the utility for Win 8.1 (seems like they recently removed it!). |
|
@prarobo so I accidentally replicated this issue today installing a new machine and saw this exact behavior (RGB only, bad magic numbers for all streams) and it was because I forgot to run the 4.4 uvcvideo patching script:
|
|
@ddiakopoulos Thanks for your response. I updated the firmware to 1.0.72.06. I also reapplied the uvcvideo patch. I am still seeing the same error. Bad magic number and no depth stream. Any thoughts? I am able to see the individual streams (color, IR and depth through mplayer) but cpp-multicam shows only color stream. |
|
@prarobo what about cpp-capture with a single camera? |
|
Only color image shows, no depth image when running cpp-capture. This is the terminal output. Capturing DEPTH at 480 x 360, fov = 55.1 x 42.8, distortion = NONE |
|
This is the exact behavior of a system that hasn't been correctly patched. I'm not sure the script is successfully being applied on your system. Can you run the following commands and post your dmesg output?
Also after streaming, could you most the last few lines of dmesg? Unknown frame formats are also a telltale sign. |
|
This what I am seeing from dmesg (see below). I am seeing an unknown video format. I am applying the patch the though. Is there a way to check if the patch is getting applied? [ 368.323530] usb 4-3.3: new SuperSpeed USB device number 5 using xhci_hcd |
|
You need to monitor what is failing in the patch script. The patch is not being applied correctly. When the script is finished, it should copy a new version of uvcvideo.ko to your system directory and then you manually re-load it with the |
|
Hello, I have the same problem, executing cpp-capture I only see rgb image: I have applied several times the patch uvcvideo-4.4.sh v4.4-wily and I haven't resolved the issue. The dmesg info: The patch results: 100%[===============================================>] 231 --.-K/s en 0s 2016-03-19 17:12:14 (38,7 MB/s) - “SOURCES.2” guardado [231/231] Shallow cloning Linux source repository... (~100mb) You are in 'detached HEAD' state. You can look around, make experimental If you want to create a new branch to retain commits you create, you may git checkout -b new_branch_name Checking out files: 100% (52221/52221), done. 2016-03-19 17:14:39 (1,73 MB/s) - “index.html” guardado [9003] --2016-03-19 17:14:39-- http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.4-wily/linux-headers-4.4.0-040400_4.4.0-040400.201601101930_all.deb 100%[===============================================>] 9.758.244 1,40MB/s en 6,9s 2016-03-19 17:14:46 (1,35 MB/s) - “linux-headers-4.4.0-040400_4.4.0-040400.201601101930_all.deb” guardado [9758244/9758244] --2016-03-19 17:14:46-- http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.4-wily/linux-headers-4.4.0-040400-generic_4.4.0-040400.201601101930_amd64.deb 100%[===============================================>] 741.242 981KB/s en 0,7s 2016-03-19 17:14:47 (981 KB/s) - “linux-headers-4.4.0-040400-generic_4.4.0-040400.201601101930_amd64.deb” guardado [741242/741242] (Leyendo la base de datos ... 331704 ficheros o directorios instalados actualmente.) 100%[===============================================>] 18.903.697 1,40MB/s en 13s 2016-03-19 17:15:04 (1,37 MB/s) - “0001-base-packaging.patch” guardado [18903697/18903697] --2016-03-19 17:15:04-- http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.4-wily/0002-debian-changelog.patch 100%[===============================================>] 426.590 740KB/s en 0,6s 2016-03-19 17:15:05 (740 KB/s) - “0002-debian-changelog.patch” guardado [426590/426590] --2016-03-19 17:15:05-- http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.4-wily/0003-configs-based-on-Ubuntu-4.4.0-0.10.patch 100%[===============================================>] 6.536 --.-K/s en 0,002s 2016-03-19 17:15:05 (2,72 MB/s) - “0003-configs-based-on-Ubuntu-4.4.0-0.10.patch” guardado [6536/6536] patching file debian.master/NOTES Thanks |
|
@javcursor One of the first lines in the dmesg is this, showing that the driver hasn't been patched.
Thanks for attaching the output of the patching log. I believe this is identical to the error documented here: IntelRealSense/realsense-ros#12 (comment) This is the line where the script is failing: https://github.com/IntelRealSense/librealsense/blob/master/scripts/patch-uvcvideo-4.4.sh#L50. You need manually execute the rest of the steps in a new shell to complete the patch. Unfortunately, you will need to unwind some of the variables like KBASE, but you could run the script once again and add some print statements to to check if you have trouble. |
|
Problem solved, many thanks for your time. |
|
I solved my problem by using the uvcvideo.ko file generated from a different identical system. I just copied and pasted this into the appropriate location. Do you see any caveats in this approach @ddiakopoulos ? Will I run into problems if I do this for several robots? |
|
@prarobo should be fine so long as the |
|
I had the same problem. Fixed it by deleting lines 50 & 51 in the patch_uvcvideo-4.4.sh script and used the following instead: |
ddiakopoulos
changed the title
|
Hi, I was having some problems with the camera but thanks to this issue and to this other issue I was able to fix it. However, I wonder why the changes that fix the problem in the script have been undone. Thanks! |
|
When you remove the lines 50 & 51 then the two files are not copied i.e. missing. Thomas |
|
Hi, I'm running into a similar problem with the SR300: modprobe reveals repeated warnings involving uvcvideo, and if I try and start up a window with a stream (eg. using cpp-tutorial-2-streams) neither colour, depth or IR come up at all (not even a black window*). I'm using the Video4Linux backend, to minimize potential issues. I tried both the 4.4 kernel script and the mainline script patch, on clean installs, but had the same problem on both. Since it seems related, I thought I'd post here as it could well be a patch error, given the uvcvideo warnings. However none of the fixes listed here seem to work - let me know if it's more appropriate to start a new issue. Changing the patch script as suggested above simply throws up more errors. I ran the SR300 live unit tests script and it reported no problems ("all tests passed), however a further modprobe of uvc after running the script showed multiple "uvcvideo: Non-zero status" and "uvcvideo: Failed to query UVC control" errors (modprobe output copied below, sorry for the length).
I should note that I have previously set up this same SR300 using identical hardware and an identical installation process, and it worked fine (still does!). Completely baffled at this point. [ 213.505694] uvcvideo: Found UVC 1.10 device Intel(R) RealSense(TM) Camera SR300 (8086:0aa5) |
|
For whatever reason, the patch isn't being applied correctly on your system (unknown video formats are a telltale sign). Follow the script closely, and don't forget to |
|
update: tried on a completely clean Ubuntu 14.04 install on a different machine - no disturbing messages from uvcvideo in dmesg, and the camera seems to be working. No doubt I did something careless during the patch application on the previous system, I will roll back (again!) and see if I can get it working. |
|
Further update: re-installing Ubuntu and running the installation procedure exactly as on the first and third machine (which both work fine) still produces no useable image. After applying the patch, the modprobe reveals nothing unusual (except the module verification failure, but this seems to happen whether or not the patch succeeds as I get the same message on other machines) [ 351.929478] uvcvideo: module verification failed: signature and/or required key missing - tainting kernel After plugging in the realsense and running cpp-tutorials-2-streams, I get nothing at all in the glfw window and running modprobe produces the following dmesg output: [ 662.018483] uvcvideo: Unknown video format 435a4e49-b602-480f-978c-e4e88ae89b89 So at this point it looks like there's something more fundamental going on, and the problem is not with the patch itself. I am getting a weirdly persistent grub screen on boot, which might mean there's another kernel installed, although I am using the same installation USB as on the working machines, and when I reinstalled ubuntu I erased the disk. I will check the partitions and try again I suppose. |
|
when running all patches, etc... I got the same issue in msg: |
|
I too have the error Perhaps the most interesting thing here is that I had been using the camera fine the previous day, then after a reboot of the system both of these problems arose. There may have been an apt-get upgrade before the reboot - is that expected to break the driver? |
|
Rerunning the patch script did not fix the issue described above, but rerunning it, telling it 'yes' when it asks a question about the patch already being applied that means "do you want to unpatch instead of patch" (I forget how it's phrased, but not very helpfully!), then rerunning it a second time, allows the script to run to completion, which fixed my issue. |
|
mine, it's still not working! |
|
I also had the issue of unknown video format after running the patching script and loading the uvcvideo module. It turns out that the script was not running successfully for me even though no errors were shown. This was due to the 'realsense-camera-format patch' having a relative address (line 47) and was not being found. Re-running the script from the 'librealsense' dir rather than 'librealsense/scripts' fixed this issue for me. |
|
I did everything from the begining in another folder and I always get this when trying to run any example: rs_error was raised when calling rs_create_context(api_version:11201): |
Camera Model: R200
Firmware Version: 1.0.71.06
Operating System & Version: Linux 14.04.04 x86_64
Kernel Version (Linux Only): 4.4
I am only able to visualize the rbg camera stream. I don't see depth and IR streams. When I run cpp-multicam, this is the terminal output
xxx@xxx:~/librealsense/bin$ ./cpp-multicam
Starting Intel RealSense R200... done.
rs.warn: Subdevice 1 bad magic number 0x0
rs.warn: Subdevice 1 bad magic number 0x0
...
[this warning keeps repeating]
I am able to visualize the IR and depth streams using mplayer. So my realsense device is working correctly. Something should be going wrong with the driver.
Is it a syncing issue that makes the driver drop frames?
The text was updated successfully, but these errors were encountered: