There have been many reports of the timer overflow that makes the camera grabbing crash after 49.7 days with errors like
Apr 1 07:30:49 rc-visard-ng-1422324319939 kernel: [4295001.509376] [RCE] BUG: common/rtcpu-queue.c:22 [safe_queue_recv] "ERROR 0x2000000a: safe_queue_recv failed"
Apr 1 07:30:49 rc-visard-ng-1422324319939 kernel: [4295001.545218] tegra186-cam-rtcpu bc00000.rtcpu: Alert: Camera RTCPU gone bad! restoring it immediately!!
We are still encountering this severe but on our Orin Nano and tried many versions, none of which fixed the issue.
Affected L4T releases with rce-fw sha1:
- L4T 35.5.0 with rce-fw sha1=55ecd57df677fd722e795b49ccb283f065b222a1
- L4T 35.5.0 with rce-fw sha1=761d9d48563b4131a5af3b34f2c9f19a713a222d (rce-fw from fused-orin-nano-35-5-0-camera-rtcpu-gone-bad )
- L4T 35.6.4 with rce-fw sha1=738ae3ed66400a5562093b958d218bf596c1f623
- L4T 36.4.3 with rce-fw sha1=e2238c99959d2df9350d393f04e1f44e5bef98cb
It is not even listed under known issues in the release notes!!
Is there a rce-fw that fixes this for L4T 35.X? Or when will it be available?
Is it really fixed with 36.5.0 as https://forums.developer.nvidia.com/t/jetson-agx-orin-camera-failure-safe-queue-recv-0x2000000a-camera-rtcpu-gone-bad-no-recovery-until-reboot-connect-tech-carrier/358992/15 suggests?
I can start another long-term test with 36.5.0 now, but it will obviously take another 49 days until I can be sure!
We really need a fix for this!