Tegra194-vi5 15c10000.vi: corr_err: discarding frame then reboot

xavier dmesg report discarding frame then reboot, attached file is the log and the trace file.

jetson_release: JetPack 4.4, L4T 32.4.3

minicom.txt (118.1 KB)
trace_1702.txt (11.5 MB)

We connect cameras through FPD-Link-III in custom carrier board,and some port report discarding frame, maybe that is my board’s problem about hardware design, but why it cause reboot, that is my question.

please help to have a check, thanks.

hello user98628,

may I know who’s the sensor vendor. thanks

SG2-AR0231C-0202-FPDLink from SENSING
datasheet

hello user98628,

discarding frame it’s actually a warning messages, it’ll drop the current failed capture request and queue the frames again. however, there’s a potential failure, no reply from camera processor, which means the signaling never come to the VI engine. if you kept low-level driver stuck for waiting frames, it eventually trigger watch-dog timeout to force reboot the system.
you may dig into the driver side to figure out whether this board output the signaling to CSI channel, then please review DT settings to examine the hardware configurations.
BTW,
please also check these Supported Cameras by Jetson Camera Partners.
thanks

how VI engine driver will do when stuck for waiting frames? watchdog reboot, maybe that’s not a good method, which part of VI engine driver should i check to avoid the reboot? thanks.

hello user98628,

you should ensure the streaming sending to CSI brick without failure, please contact with Jetson Camera Partners.

besides,
if you look into low-level driver, i.e. kernel/nvidia/drivers/media/platform/tegra/camera/vi/vi5_fops.c
it’s VI engine sending a capture request, and waiting for buffer complete to return a capture frame.
for example,

static void vi5_capture_enqueue()
{
               ...
               err = vi_capture_request(chan->tegra_vi_channel[vi_port], &request[vi_port]);