Re: [PATCH] NMI trap revised (was Re: NMI errors in 2.0.30??)

Riccardo Facchetti (
Fri, 9 May 1997 22:21:41 +0200 (MET DST)

On Fri, 9 May 1997 wrote:

> On Fri, 9 May 1997 17:01:56 +0200, you wrote:
> > After studying lots of specs and testing it on several motherboards, it
> > seems that those bits are implemented almost everywhere and hard-wired to zeroes
> > on the remaining MB's. We probably should honor them and report the reason
> > to the user if we find any.
> >
> Well, I have very little faith in motherboard manufacturers and BIOS writers, so
> I wouldn't believe anything I didn't set myself.

This is not a point. The BIOS setup is done by you with your hands. You
can decide to enable/disable the NMIs and enable/disable Parity or ECC
checks. Since this configuration is alredy done by the BIOS, which tests
the machine before booting it, you should be confident that when the "Now
booting the kernel" message is displayed, the machine is reasonably set
up. No kernel initialization is needed, and forcing some initialization
not done at BIOS stage can be very dangerous (See my PATCH3 e-mail
about NMI tests with non-NMI memory).

> > > This is really an issue about where hardware initialization should occur. The
> > > real-mode setup.S seems to be the intended place to do this sort of thing, but
> > > very few people like writing assembler code so it tends to clutter up the kernel
> > > proper. I have a few ideas about this. Maybe something will happen in a few
> > > weeks.
> >
> > We probably don't have any hardware to be initialized as we must not touch
> > the enables (see the previous discussion).
> >
> > Martin
> Again, this depends on how much trust you have in the hardware setup.

You must trust your hardware setup. It boots or not. It works or not.
Depending on the BIOS setups.