After some more experiments, it seems to be related to length of kernel execution. For kernels that takes about < .2 secs it runs fine. For kernels that takes > .5 sec it hangs the system.
Is there VPU recover type of functionality in the linux driver, that might be causing issues?
Well, 1.3 should be right around the corner so try with that....
I've only barely gotten my kernels up to around .4 seconds but I'm not having any issues like that.... Save for an accidental infinite loop will crash the program, and occasionally will cause my system to hang.
However I would think if this was some sort of VPU recovery mechanism, that it would do something more (IE reboot/shutdown the comp or the module).
At any rate, when I get back from work maybe I'll check out your program and see if I can reproduce the error with my flavor of stuff...
- Can you kill the X-server or the process instead of rebooting? I'll admit it looks like a case of picking your poison.
- Is it also related to how much drawing (to this or other windows) the GPU is doing at the same time?
- if you have a test case, send us an email at firstname.lastname@example.org