Hello,
I think I have summarized the discussion for clarity:
1. rep nop can used with all x86 boxes, unless a valid example can be found where it doesn't work. Athlon works with the rep nop.
2. There's a bug in get_model_name(),
cpuid(0x80000001, &dummy, &dummy, &dummy, &(c->x86_capability));
that overwrites the capability state. It will be fixed in 2.2.18 by Dave Jones. Should we also look at Peter Anvin's fix for the problem that Linus mentioned? What are the other features of the Pentium IV should be included in the kernel pending the capability state fix?
3. 2.4.x may support processor speeds up to 100GHz, as well as Pentium IV. Linus will have a Pentium IV available soon, but can someone test the kernel with a Pentium IV sooner?
Regards,
Frank
Followup to: <[email protected]>
By author: Frank Davis <[email protected]>
In newsgroup: linux.dev.kernel
>
> 2. There's a bug in get_model_name(),
> cpuid(0x80000001, &dummy, &dummy, &dummy, &(c->x86_capability));
>
> that overwrites the capability state. It will be fixed in 2.2.18 by
> Dave Jones. Should we also look at Peter Anvin's fix for the problem
> that Linus mentioned? What are the other features of the Pentium IV
> should be included in the kernel pending the capability state fix?
>
I'm working on a revamp of the i386 CPUID detection code, which will
fix this problem as a side effect.
-hpa
--
<[email protected]> at work, <[email protected]> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt
On Wed, Nov 08, 2000 at 12:36:21PM -0800, H. Peter Anvin wrote:
> Followup to: <[email protected]>
> By author: Frank Davis <[email protected]>
> In newsgroup: linux.dev.kernel
> >
> > 2. There's a bug in get_model_name(),
> > cpuid(0x80000001, &dummy, &dummy, &dummy, &(c->x86_capability));
> >
> > that overwrites the capability state. It will be fixed in 2.2.18 by
> > Dave Jones. Should we also look at Peter Anvin's fix for the problem
> > that Linus mentioned? What are the other features of the Pentium IV
> > should be included in the kernel pending the capability state fix?
> >
>
> I'm working on a revamp of the i386 CPUID detection code, which will
> fix this problem as a side effect.
Is this revamp only for processors that actually support the
CPUID-instruction, or will you fix the CPU-detection for non-CPUID
processors too?! There are quite a few processors that can be detected
properly but aren't (for instance, IBM 486slc/slc2/slc3)
/David Weinehall
_ _
// David Weinehall <[email protected]> /> Northern lights wander \\
// Project MCA Linux hacker // Dance across the winter sky //
\> http://www.acc.umu.se/~tao/ </ Full colour fire </
> Is this revamp only for processors that actually support the
> CPUID-instruction, or will you fix the CPU-detection for non-CPUID
> processors too?! There are quite a few processors that can be detected
> properly but aren't (for instance, IBM 486slc/slc2/slc3)
Linus refused code to ident the ones that didnt matter because the code was
(neccessarily) obscure, weird and didn't change anything but the string in
procfs.
Followup to: <[email protected]>
By author: Alan Cox <[email protected]>
In newsgroup: linux.dev.kernel
>
> > Is this revamp only for processors that actually support the
> > CPUID-instruction, or will you fix the CPU-detection for non-CPUID
> > processors too?! There are quite a few processors that can be detected
> > properly but aren't (for instance, IBM 486slc/slc2/slc3)
>
> Linus refused code to ident the ones that didnt matter because the code was
> (neccessarily) obscure, weird and didn't change anything but the string in
> procfs.
>
It should be a lot cleaner to do that kind of stuff -- if you feel
it's worth bothering to -- in my changed version. It's taking a bit
longer than I'd like, because I keep uncovering, ahem, "issues" with
the current code that can be easily fixed once one adds a real
framework for these things. I hopefully will have something RSN.
-hpa
--
<[email protected]> at work, <[email protected]> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt
On Wed, Nov 08, 2000 at 03:14:22PM -0500, Frank Davis wrote:
> 3. 2.4.x may support processor speeds up to 100GHz, as well as Pentium IV. Linus will have a Pentium IV available soon, but can someone test the kernel with a Pentium IV sooner?
I've gotta Pentium IV 800MHz working on 2.4.0-test10. Just started testing it. It actually seems pretty stable.
--
Prasanna P Subash
[email protected]