Jetson AGX 32Gb - Gstreamer interruptions

I’m using an AGX with Jetpack 5.1.1. Our system uses 4 CSI cameras (Connect Tech Rogue Carrier Board with JCB005 Camera Adapter Board), and we notice that after a while, the streams stop capturing frames, but GStreamer isn’t crashing.

When a camera failure occurs, we notice when the fourth occurrence of the error below happens, a stream stops:

[ 1716.389079] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194404
[ 1985.877492] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194404
[ 2022.375924] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194404
[ 2064.474117] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194404

Furthermore, if we let Gstreamer continue, another camera failure will eventually follow with similar logging behavior. However, the err_data value is different:

[ 2698.946814] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194405
[ 2800.842440] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194405
[ 2960.135685] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194405
[ 3044.532002] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 1, flags: 0, err_data 4194405

I want to point out that the camera failure does not follow the device order. Among video0, video1, video2, and video3, the first failure consistently occurs in video2, followed by video3.

  • Is there a reason why the failure order remains the same across different AGX systems we have tested (we have tried 4 identical systems)?
  • Why does the error code vary depending on the camera? As mentioned, video2 has an err_data value of 4194404, while video3 has an err_data value of 4194405.
  • Why does the error occur four times before the system completely fails? Is it because we are using four cameras?
  • We also notice that when the error occurs, a CPU core becomes maxed out (we’re using all 8 cores). If 1 camera has failed, one core is maxed out. If 2 cameras fail, two cores are maxed out.

Thanks in advance for any assistance provided.

Hi,
For Jetpack 5.1.1(r35.3.1), there are several bug fixes in

Jetson/L4T/r35.3.x patches - eLinux.org

Please check the list and apply the related fixes for a try. Also suggest upgrade to latest Jetpack 5.1.4.

Our BSP is only available up to Jetpack 5.1.2. Are the patches that you linked for 5.1.1 incorporated in the 5.1.2 release? We upgraded our system to 5.1.2 and there were no differences.

We found another post that appears similar to our issues (Continuously streamming even when there are frame errors)

The post I linked above has a reply with a patch. Can you confirm if the patch in the response is also applied to Jetpack 5.1.2?

Hi,
For Jetpack 5.1.2, please check

Jetson/L4T/r35.4.x patches - eLinux.org

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.