We are implementing a hardware trigger mode for a camera on a jetson Xavier NX.
The basic process is:
We start the application;
Connect camera using Argus (enableCanInifiniteTimeout=1);
Use the i2c controller to put the camera into a still state (no frame from it until hardware trigger);
We trigger camera using GPIO, read image and process it.
The appliacation works for a while, the enableCanInifiniteTimeout=1 does a great job on an application level on waiting.
The problem is that if we do not trigger the camera for ~2-5 minutes, then the whole Jetson reboots. We think it is a driver or kernel level timeout rebooting the system if there is no response from the camera for a while, but it would be good to keep going.
Is there any way for disabling the low level timeout, making the device wait for a signal from the camera hardware indefinately?
that’s correct, it should be low level timeout to trigger watchdog.
could you please setup a terminal and gathering the logs for reference, for example, $ dmesg --follow
thanks
I’ve started the app at [61.365317] secs, device rebooted at ~[240.0] secs, and there is nothing there.
I’m attaching the full system log from that time. syslog.txt (22.4 KB)
There are some NvCaptureStatusErrorDecode messages, but that is there eve is the camera is working fine. Apart from that I can’t see any message corresponding to the reboot.
here’s HW throttling (OC alarm) happened. it may be the thermal issue then hardware shutdown the chip.
do you have thermal solution? please also setup tegrastats utility to monitor the process usage.