I have read the thread.
A few points :
Can an application spoof 'top' into reporting a CPU usage that is not really
there ? If it is just 'reporting' the NOT CPU usage, how does that make its
way to 'top' as actual CPU usage.
A bit of perseverance and, like the inetd, you can get it to the much
healthier state of priority = -1 and CPU usage of 0%. So what have I just
done. Stopped the CPU idle loop or stopped the APM daemon from hogging the
CPU ? Given the responsiveness of the system I don't believe that the idle
loop has been affected, and given APM all works then nothing has happened to
it either !
According to the kernel mailing lists the name kapm-idled was just put there
to stop people complaining that kapmd was using too much CPU ! Surely a
better fix would have been to look into why kapmd was using / reporting such
a high CPU usage.
If it is just 'reporting' that the system is idle why bother ! Surely the big
black bar on 'gtop' is far more comforting than 40 - 80% CPU usage !!!
I realise this is out of Mandrake's remit, but it is certainly confusing. Do
Red Hat have a newer kernel with a better fix ? Please observe - Red Hat's
Rawhide 2.4 kernel DOES NOT display this perverse behaviour (their kapm-idled
has a PRI of -1 and CPU of 0), so why does Mandrakes ?
Owen
On Friday 15 December 2000 11:48 pm, you wrote:
> > If it "doen't do anything" then why is it using 40 - 80% CPU resource ?
>
> You really should read the message thread he pointed you to (URL included
> below). It IS NOT using 40 - 80% CPU resource. It IS reporting that the
> CPU is IDLE for 40 - 80% of the time. Worry if it drops down to 0%. That
> is the reason for the "-idle" part of the name. It is reporting idle time
> instead of usage time.
>
> > > http://marc.theaimsgroup.com/?l=linux-kernel&m=97063175201750&w=2