Jetpack 6.0: Clusters of null values on the captured camera image

Hi All,

I’m trying to migrate our V4L2 camera sensor driver to the Jetpack 6.0 production release SDK, which has the Jetson Linux 36.3 driver package. The hardware setup is a Jetson Orin Nano Developer Kit, and the custom camera device tree overlay is applied over the flash.sh.

The camera sensor is connected to the CAM0 connector of the Orin Nano Devkit. It was probed successfully.

Frame capture command
v4l2-ctl --device /dev/video0 --set-fmt-video=width=2560,height=512,pixelformat=RGGB --stream-mmap --stream-to=frame.bin --stream-count=1

Using Jetpack 6.0, the frames are captured, and the saved bin file has clusters of null value occurrences in multiple places.

There are no CSI or CRC error events on the debug traces.

log1_JP6_frame_capture_debug_traces.txt (11.0 KB)

tegra234-p3767-camera-p3768-nv-cam.txt (17.7 KB)

Previously, we were using Jetpack 5.1.2, and the captured image didn’t have any clusters of null values. The issue seems to be reproducible only on the Jetpack 6 production release SDK (L4T_36_3).

Can you provide your suggestions to debug this issue.

Thanks

Hello @sivasubramaniyan.padmanab,

Just thinking out loud…

Could it be a timming issue ?

How does the image look when you debayer it ?

Regards,
Andrew
support@proventusnova.com

It’s know issue and should be fix in next release.
Please try skip frames (as many as possible) to check for current release.

Can you please update the targeted date for next release.

Could be end of year but I can’t promise.

Thanks

@ShaneCCC I think we are also hitting the same problem. Could you please point us to the Bug report / description so we can better understand what is happening. Does this affect any mipi camera under JP6 ?!

It could be kernel regression relative to the memory control. Suppose doesn’t matter with the MIPI.

@ShaneCCC it is really hard to figure out what you want to tell us with those really short sentences. Can you please provide us a link to the issue report?

It’s kernel regression cause the issue we are debugging to get the root cause.

Thanks

Related: [Camera][VI][JP-6.0/r36.3] Fixes for raw image corruption (line artifacts) through VI-mode

3 Likes