Running ardupilot 3.5.0 rc 7 on Erle brain 3 turns out to be bit dangerous: the ardupilot would occasionally crush and restart, even without the copter being armed.
This happened with other dev- or 3.5 version too. Attached is a log for this matter. I am running on erle brain 3 with only PWM output and receiver attached.
Two logs before and after the ardupilot crash.
00000057.BIN (120 KB)
00000056.BIN (256 KB)
Similar behavior is observed at rc8:
00000072.BIN (864 KB)
00000073.BIN (656 KB)
Attaching also the Parameter Settings:
ParamListUpdated2.param (12.9 KB)
The quadcopter has PX4flow and LidarLitev3 connected, pwm output and rc receiver connected.
It would crash at random time without even armed.
Can you run it under gdb and check the backtrace?
Does it crash without the lidar and px4flow?
yes it crushes even when its just powered by micro usb and nothing else.
Not entirely sure how gdb works in this case, any instructions?
@JD-ETH before going too deep, please confirm that the new code is stable with a newly flashed OS.
That is the test asked you to do a few days ago and you confirmed it was stable right ?
I did a clean build with erle debian package just now. But it has still happened again. It happened with only pwm and radio receiver attached.
Trying different ways to see whats happening, so far no sucess:
sudo ./arducopter -A udp:"10.0.0.2:6000"
Raspberry Pi 2/3 with BCM2709!
MS5611 found on bus 0 address 0x00
PulsedLightI2C: bad version 0x00 0x00
MPU: temp reset 6627 0
Interrupted: Segmentation fault
with rc8. Was not yet able to reproduce the same error in gdb yet…
with your hardware configuration I’m not able to reproduce your issue. Could you please, put here more info/videos…
The video is not that entertaining to watch
Hardware setup: 3S battery powers the 2.25 A 3DR Power module,
Connected output: PWM and Receiver
Setup: New Erle Image 09-12-16 and apm 3.5.0 rc7
[ESC beeps when powered but no pwm signal received]
After indefinite time, leads to crash and ESC beeps for several seconds again.
Doesn’t happen with the older release 3.4-rc2!
Back then I didn’t look so carefully at the light I guess. I have this very interesting behavior, just starting arducopter it crashes consistantly, while in gdb in seems to run just fine. (same binary)
Yes it crashes in all conditions, but doesn’t in debug mode. Check my videos in the reply.
First time seg fault in debug mode:
[Thread debugging using libthread_db enabled]
Using host libthread_db library “/lib/arm-linux-gnueabihf/libthread_db.so.1”.
Raspberry Pi 2/3 with BCM2709!
[New Thread 0x76ca3450 (LWP 1391)]
[New Thread 0x769ff450 (LWP 1392)]
[New Thread 0x768ff450 (LWP 1393)]
[New Thread 0x765ff450 (LWP 1394)]
[New Thread 0x764ff450 (LWP 1395)]
[New Thread 0x75b8e450 (LWP 1396)]
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x768ff450 (LWP 1393)]
0x000b6408 in Linux::RCInput_RPI::_timer_tick() ()
Yay! At least we now know it happens in debug!
Can your provide a backtrace?
thread apply bt all (or whatever that syntax is…)
Well I didn’t know what to do back then so I don’t have one.
But this seems to be known failure:
And the issue:
Thanks to report by @ppoirier
The issue you are linking is fixed some time ago. What’s the exact version you are running? I don’t have a erlebrain 3. But I don’t see much difference from the others that would cause it on erlebrain3 and not on other rpi-based boards.
I’ll be back from vacations next week and can try to reproduce on some RPI boards
i can help you in something ?
But I don’t see much difference from the others that would cause it on erlebrain3 and not on other rpi-based boards.
The best guy to help with that is maybe @LanderU
I will create a new OS for our boards, and test it.
If you can debug the firmware, please do it. Run with
gdb and when the firmware crashed, type
You think that we can just let the drone running without props in a desk and wait?
yes, please. And run it with
gdb, thank you!!!