pinn [Pi5] Lock-ups

Model: Pi5 PINN: v3.8.9 Occasionally and randomly, PINN will freeze, requiring a reboot. This seems to happen just after the wifi has established and the IP address has been set.

Dec 28 '23 14:12 procount

100% reproducible when moving the USB mouse

Jan 28 '24 19:01 Poopi

Yes I had suspected something similar myself.

Jan 28 '24 19:01 procount

Do you have a beta branch somewhere to look at?

Jan 30 '24 18:01 Poopi

Was thinking about source code rather that binaries that I already checked...

Feb 23 '24 11:02 Poopi

I need to get it into a fit state first, then I'll consider pushing it up. The odd thing is that apart from the removal of the alternate bootloading mechanism, the PINN source code is virtually the same as before. The 6.1 kernel is the same across all the models, but only the 64-bit version for the Pi5 exhibits this issue. I think it's more of a kernel issue, unless it's the kernel that's highlighting a latent problem in the application 🤷‍♂️

Feb 23 '24 11:02 procount

I do think it is the latter. I do have an old Xiaomi keyboard that generates spurious messages in kernel. When PINN hangs I can simply remove the dongle and the boot continues....

Feb 25 '24 20:02 Poopi

When it locks up, it seems to be stuck here:

(gdb) backtrace
#0  0xf6cc81c8 in pthread_cond_wait () from /home/pi5/buildroot/output/host/usr/arm-buildroot-linux-uclibcgnueabi/sysroot/lib/libpthread.so.0
#1  0x015f9750 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) bt
#0  0xf6cc81c8 in pthread_cond_wait () from /home/pi5/buildroot/output/host/usr/arm-buildroot-linux-uclibcgnueabi/sysroot/lib/libpthread.so.0
#1  0x015f9750 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

So, probably a thread or stack error.....?

May 24 '24 21:05 procount