< curious>
I just noticed that our 8-way X440 is showing (in /proc/interrupts)
about 100 NMIs / second, on each CPU. Would some kind soul please tell
me if this is OK? A brief explanation of what this interrupt is being
used for would be appreciated.
We're running the latest RH Advanced server kernel.
</curious>
Sorry if this is a stupid question.
Thanks,
John
On Fri, Sep 19, 2003 at 07:31:03PM -0700, John Wendel wrote:
> I just noticed that our 8-way X440 is showing (in /proc/interrupts)
> about 100 NMIs / second, on each CPU. Would some kind soul please tell
> me if this is OK? A brief explanation of what this interrupt is being
> used for would be appreciated.
> We're running the latest RH Advanced server kernel.
> </curious>
> Sorry if this is a stupid question.
A common use for intentional NMI's is NMI-based profiling, i.e. oprofile.
I can't say for sure this is where your NMI's are coming from without
seeing more about your kernel.
-- wli
On Fri, 19 Sep 2003 19:34:15 -0700, William Lee Irwin III wrote:
>On Fri, Sep 19, 2003 at 07:31:03PM -0700, John Wendel wrote:
>> I just noticed that our 8-way X440 is showing (in /proc/interrupts)
>> about 100 NMIs / second, on each CPU. Would some kind soul please tell
>> me if this is OK? A brief explanation of what this interrupt is being
>> used for would be appreciated.
>> We're running the latest RH Advanced server kernel.
>> </curious>
>> Sorry if this is a stupid question.
>
>A common use for intentional NMI's is NMI-based profiling, i.e. oprofile.
>I can't say for sure this is where your NMI's are coming from without
>seeing more about your kernel.
This case is more likely to be the NMI watchdog.
Having it enabled is normal and good, but you can disable it
by booting without the nmi_watchdog= parameter.