Pascal

I must do this on many robot to get a chance of crash:( A stress tester is a good thing todo

The problem occur only when there is also PIGPIO initialization with the v4l2-ctl, it use DMA trick to make precise PWM. But I can't get 100% sure But that might...

I make a working stress tester, first time I can reproduce the bug !!!

first try without debug=1 sorry The last line is the problem. ``` (root|~) vcdbg log msg 2988287.376: venc: venc_rc: frame 244 aborted bits 43424 > -57415 (HRD), QP=51 2988322.911: venc:...

I must manual stress for about 30 minutes to get theses information : Only the last line appear specific to the bug ``` (root|~) vcdbg log msg 1952515.424: venc: venc_rc:...

I just get a crash without entire configuration applying (after 1 day uptime), Only v4l2-ctl with brighness and/or bitrate

Thanks for you feedback 6by9. In the meantime I will set up an automatic restart on the robots with statistics on crashes, this will serve as a test once the...

Now we use a WatchDog to detect the "venc: venc_exit: cnt != out_cnt" here is frequency occurrence of this problem : ``` 14/09 08:36:07:922 | | | 176.179.178.138 | raspA...

Another rare videocore crash. (where v4l2-ctl --all freeze also) ``` Sep 29 18:58:52 raspberrypi kernel: [196478.158463] ------------[ cut here ]------------ Sep 29 18:58:52 raspberrypi kernel: [196478.158488] WARNING: CPU: 3 PID:...

Another one (a PI 3 A+ this time) only few hours after clean reboot : All robots run the current stable Kernel (apt update/upgrade without rpi-update) The current stable Kernel...