Re: Stops Pentiums dead (and how does Cyrix perform?)

linux kernel account (linker@nightshade.z.ml.org)
Sun, 9 Nov 1997 00:54:36 -0500 (EST)


Make you could go a step further, when it traps, it could analyze the
offending code and compair it to a comprehensive table of 'bad opcodes'
and decided to fail on not..

On Sat, 8 Nov 1997, Richard B. Johnson wrote:

> On Sat, 8 Nov 1997, linux kernel account wrote:
>
> >
> > The guy running x86.org doesnt want anything to do with this bug... I've
> > already emaild him.. and it's "for personal reasons".. So please dont
> > flood him..
>
> [SNIPPED]
>
> I'm talking to x86.org. They are advocates and are providing technical
> assistance. Don't bug them. They know about the problem.
>
> We may be able to set Pentiums to trap on the lock prefix when in
> user-mode. User mode code doesn't (shouldn't) use this instruction.
> This should fix the problem. I have already grepped through the latest
> gcc c-library and none of that code seems to require the lock prefix.
>
>
> Cheers,
> Dick Johnson
>
> Richard B. Johnson
> Project Engineer
> Analogic Corporation
> Penguin : Linux version 2.1.60 on an i586 machine (66.15 BogoMips).
> Warning : It's hard to remain at the trailing edge of technology.
>