Re: Load average and CPU hogs in realtime scheduling

Olaf Titz (olaf@bigred.inka.de)
Sun, 16 Nov 1997 15:48:11 +0100

> because during the time the SCHED_FIFO process sucks all the CPU the
> scheduler will never select one of the other processes waiting to run.
> So when MAME quits all those processes get run, finish their jobs and

I thought of this too, but how do they _become_ waiting to run in the
first place? I've started with all them in S state and no other
process can unblock them...

...seems I've not really thought about interrupts. Okay, you're right.

(But even this is interesting: although the skyrocketing load may be
correct, the system shows no signs of overload, no frantic disk
activity, etc. Linux has really got better in the 2.0.31 series.)

olaf