leaking memory and semaphores

[jonas.kirilla]

In hrev21877, doing a

while (true); do clear; sysinfo | tail -n 5; done;

shows the semaphore and memory count rising continously.

ProcessController shows System Resources & Caches rising whereas the kernel_team's memory is stable (but fluctuating some).

[axeld]

As I've written in the commit message to hrev21862, the kernel is currently leaking address spaces. Unfortunately, fixing it the obvious way did not work out as expected (due to the team tear down code). Will have to investigate this some day :-)

[jonas.kirilla]

Thanks for the explanation, Axel! It's good to know that you're on top of things. :)

On each fork two more semaphores seem to get used: one 'address space' and one 'translation map'. Forking eventually deplete resources, resulting in a panic saying the "heap overgrew itself".

[axeld]

Fixed in hrev22130.