Is there any way to check the reason for a segmentation fault?

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
Jetson Nano
• DeepStream Version
5.0.1
• JetPack Version (valid for Jetson only)
4.4
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

I’m having a Python script with deals with 3 USB cams simultaneously and does inference on it. The script is running on a pre-canned downloaded image hosting JetPack 4.4. It is crashing with segmentation fault on the latest JP 4.5 image available for download.

I have no clue, what to inspect to get the reason for this crash. Do you?

Hi,
We have python samples in

deepstream-test1-usbcam may be close to your usecase. Is it possible to share a patch to the sample so that we can replicate the issue?

Hi @DaneLLL,

Thanks for the follow up. Since my code is a continuation of the nice work of @kassian (published here Python USB Camera Deepstream 5.0 Sample with Multiple Cameras?) I have no problems to share it too. It is basically a merge of your one USB camera sample and the multi RTSP input sample with some changes to enable capturing of image/jpeg.

  • Install a fresh system from the latest development SD image which you provide for download (jp45) on a Jetson Nano 4G development kit device
  • I started (as I’m used to do that) with sudo update && sudo upgrade -y
  • Install DeepStream SDK following your own guides (Note: Don’t change to r32.4 main, leave it as r32.5 main - this is what I did)
  • Install your Python sample apps as linked above.

This code works perfect on an earlier SD image (jp44), but crashes immediately with segmentation fault once more than one camera is used simultaneously.

I didn’t double check my results with JP45 again, since I returned to JP44. I hope you can confirm.

On the way to install the python samples I had to sudo chown 777 -R /opt in order to gain myself RW access to /opt.

Please find my config.txt and the inference.py python3 script here:

Extract both files into a new subdirectory at the same level as all the python sample apps reside, so e.g. to

deepstream_python_apps/tree/master/apps/three-cams

Then run

python3 inference.py /dev/video0 /dev/video1 /dev/video2

You might need to adapt the video URIs to your situation. Please choose the device(s), which support image/jpeg capture or you need to change the source bin configuration. I was not able to make three USB cams working while capturing YUV due to the known USB bus congestion problems on a Jetson Nano. Cameras are of this brand, three devices. https://www.amazon.de/gp/product/B07GSS4LT5/ref=ppx_yo_dt_b_asin_title_o04_s00?ie=UTF8&psc=1

Let me know, if something is unclear and if I can help. I don’t want to be fixed to an outdated version, for sure.

Regards

This morning I double checked the issue. You may probably disregard. As of now it works.

But w/o sudo update && sudo upgrade the installation of the deepstream requirement libraries fails:

neil@jetson:~$ sudo apt install \
> libssl1.0.0 \
> libgstreamer1.0-0 \
> gstreamer1.0-tools \
> gstreamer1.0-plugins-good \
> gstreamer1.0-plugins-bad \
> gstreamer1.0-plugins-ugly \
> gstreamer1.0-libav \
> libgstrtspserver-1.0-0 \
> libjansson4=2.11-1
[sudo] password for neil: 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
E: Unable to locate package libgstrtspserver-1.0-0
E: Couldn't find any package by glob 'libgstrtspserver-1.0-0'
E: Couldn't find any package by regex 'libgstrtspserver-1.0-0'
neil@jetson:~$

This only works after the update. Anyway, I don’t have the observed crash anymore as I had with my first attempt.

Sorry for the hassle.

28 fps on each cam. Works