Hi QNX experts,
Any tips on how to debug a complete system lockup? I have a repeatable test scenario which causes my telnet and console sessions to totally freeze-up.
Kernel is not crashing since there is no register dump on serial or VGA consoles, and Caps/Numlock keys are still active. Ping requests are ignored.
Qconn is down so IDE doesn’t show any useful information. If I run hogs at high priority on the console it also hangs. I’m running QNX 6.5.0.
DT
How about using “APS” partitioning scheduler… Try putting devc-ser and tinit in a single partition with say 10% of CPU. Then, if the rest of the system locks up due to a run-away CPU, you should still be able to login.
Of course if memory is exhausted, this won’t help. Then you may want to try “memory partitioning” which is also supported.
This is a little more abstract approach. They scenario you describe sounds a lot like a runaway process at a high priority. You might want to think about what processes you have that could do this. You might try starting the system with everything you wrote put at a lower process then things you didn’t write, like telnet. You might also try starting a shell at a high priority and seeing if you can run anything when the system hangs.
If everything you are running is generic QNX, you might want to consider some kind of hardware problem or failure.