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

tenthumbs@cybernex.net
Fri, 09 May 1997 16:59:46 GMT


On Fri, 9 May 1997 18:48:52 +0200, you wrote:

> I think of something else: I currently work on a generic chipset support
> module which will have several options for turning on various features. I simply
> can add an option forcing setting of the memory parity check logic.
>

That works well until you change something that has side effects. Consider the
Cyrix 5x86/6x86 chips. On the 5x86, you can set some bits and enable/disable the
branch target buffer. This, of course, dramatically affects BogoMips. So, if you
do this, you have to fiddle with a running kernel.

Most of the time, you can alter a running kernel, but not always. Both methods
are probably necessary.

John