Re: [PATCH] NMI trap revised

Riccardo Facchetti (
Tue, 13 May 1997 18:50:24 +0200 (MET DST)

On Fri, 9 May 1997, Alan Cox wrote:

> > Hmmm ... is there a way to force an NMI (hardware, not just int 0x02) to
> > simulate a memory error, just to see if the relevant bits are set ?
> On a pentium board with an APIC bus (ie all the dual cpu ones) you can
> send other processors an NMI at will

Oh ... I could buy another P133 and install it on my Tyan Tomcat IIID, but
this seems to be useful only for those who have this HW, so we can not
write some kind of autodetection.
I ask myself why NMI exist (and while we are at it APM too), if the specs
followed are not unified and there is no way to autodetect the behaviour
of the chipset. Bah.