[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Ksoftirq running wild .. What is it?

Hi Nicolas,

> >   a process called ksoftirqd_CPU0 is eating up my cpu ... I recently
> > returned from apm suspend and now this process eats 95% of my cpu. Oddly
> > enough I am not able to kill it with kill -9 pid.

> It's not odd at all. See below. As to the high CPU usage, it would
> appear that either it really has a lot of work to do, or it's a sampling
> issue.
I doubt it has much to do, as it happened just once after wakeup. It *usually* doesn't eat up my cpu.
Having said that I need to admit that I still don't know what soft irqs
are and what a sampling issue is.

To save myself from another hint to search the fine web: I found that,
but I am not much smarter after browsing through it.

Anyway. It seems odd *to me* as it doesn't happen often and it doesn't
seem to calm down again.

> At any rate, I would suggest not to kill -9 everything that is unknown
> on sight. It might be important.
Oh really? Lol, ok I will take that into consideration in the future.
To be serious again, what should I do in such a case? Reboot the machine
would be the only other option I could think of and that reminds me too
much of the other os.
I've seen that a process id of "4" hints to an important system process
and top also showed that the cpu was eaten up by a system process, so I
understood that the process might be important, but I issued sync as a
precaution and then tried to kill the process. Wouldn't that be enough
precaution for most cases, I mean better than rebooting? 

Until kill -9 doesn't seem to take effect I believed in kill -9 as the
last resort which will eventually kill a process. This doesn't seem to
the case. 

> >   Any idea what this process does?
> STFW. http://lwn.net/2001/0726/a/ugly.php3
> In short: That is, as the name suggests, a kernel daemon handling soft
> IRQs which hannot be handled at the spot.

But something seems to flood my systems with "soft irqs".


Reply to: