Use of CONFIG_M686

Use of CONFIG_M686

[J.A. Magallon]

Hi all...

Grepping through the sources or the kernel in search of CONFIG_M686
occurences, there are some places where it looks like that flag is
used as 'Anything bigger than a Pentium'. Now kernel has configs
for PIII, P4, probably PII.

It is the f00f bug handling. Files:

arch/i386/kernel/traps.c:

#ifndef CONFIG_M686 <=================== which also passes if PII, P4...
void __init trap_init_f00f_bug(void)
...

arch/i386/kernel/setup.c:

static void __init init_intel(struct cpuinfo_x86 *c)
{
#ifndef CONFIG_M686 <=================== again
    static int f00f_workaround_enabled = 0;
...


So thats why I asked if we could use a CONFIG_MPENTIUMPRO, and make
CONFIG_M686 a generic flag that is also defined for anything bigger
than a Pentium (that looks like the current usage).

So:
Pentium -> M586
PPro    -> MPENTIUMPRO M686
PII     -> MPENTIUMII  M686
PIII    -> MPENTIUMIII M686
P4      -> MPENTIUM4   M686


-- 
J.A. Magallon                           #  Let the source be with you...        
mailto:jamagallon@able.es
Mandrake Linux release 8.3 (Cooker) for i586
Linux werewolf 2.4.19-pre8-jam4 #2 SMP dom may 26 11:20:42 CEST 2002 i686

Re: Use of CONFIG_M686

[J.A. Magallon]

On 2002.05.28 J.A. Magallon wrote:
>Hi all...
>
>arch/i386/kernel/traps.c:
>
>#ifndef CONFIG_M686 <=================== which also passes if PII, P4...
>void __init trap_init_f00f_bug(void)
>...

Would it be enough with

#if defined(CONFIG_M586) || defined(CONFIG_M586TSC) || defined(CONFIG_M586MMX)

386-486 do not have the bug, do ?


-- 
J.A. Magallon                           #  Let the source be with you...        
mailto:jamagallon@able.es
Mandrake Linux release 8.3 (Cooker) for i586
Linux werewolf 2.4.19-pre8-jam4 #2 SMP dom may 26 11:20:42 CEST 2002 i686

Re: Use of CONFIG_M686

[Brian Gerst]

J.A. Magallon wrote:
> Hi all...
> 
> Grepping through the sources or the kernel in search of CONFIG_M686
> occurences, there are some places where it looks like that flag is
> used as 'Anything bigger than a Pentium'. Now kernel has configs
> for PIII, P4, probably PII.
> 
> It is the f00f bug handling. Files:
> 
> arch/i386/kernel/traps.c:
> 
> #ifndef CONFIG_M686 <=================== which also passes if PII, P4...
> void __init trap_init_f00f_bug(void)
> ...
> 
> arch/i386/kernel/setup.c:
> 
> static void __init init_intel(struct cpuinfo_x86 *c)
> {
> #ifndef CONFIG_M686 <=================== again
>     static int f00f_workaround_enabled = 0;
> ...
> 
> 
> So thats why I asked if we could use a CONFIG_MPENTIUMPRO, and make
> CONFIG_M686 a generic flag that is also defined for anything bigger
> than a Pentium (that looks like the current usage).
> 
> So:
> Pentium -> M586
> PPro    -> MPENTIUMPRO M686
> PII     -> MPENTIUMII  M686
> PIII    -> MPENTIUMIII M686
> P4      -> MPENTIUM4   M686
> 
> 

I fixed that in 2.5 by introducing CONFIG_X86_F00F_BUG.

http://marc.theaimsgroup.com/?l=linux-kernel&m=101416800017102&w=4

-- 

						Brian Gerst

Re: Use of CONFIG_M686

[Alan Cox]

On Mon, 2002-05-27 at 23:29, J.A. Magallon wrote:
> 
> On 2002.05.28 J.A. Magallon wrote:
> >Hi all...
> >
> >arch/i386/kernel/traps.c:
> >
> >#ifndef CONFIG_M686 <=================== which also passes if PII, P4...
> >void __init trap_init_f00f_bug(void)
> >...
> 
> Would it be enough with
> 
> #if defined(CONFIG_M586) || defined(CONFIG_M586TSC) || defined(CONFIG_M586MMX)
> 
> 386-486 do not have the bug, do ?

You misunderstand the intent. A 386 or 486 kernel will run on a Pentium
and could therefore hit the error. A PPro kernel would die earlier
anyway. Of course its long been PPRO|Athlon|... and the ifdef wanted
updating. I'd ifdef it on CONFIG_X86_FOOF_BUG and put the FOOF thing
into arch/i386/Config.in nicely with the other stuff

Re: Use of CONFIG_M686

[Dave Jones]

On Tue, May 28, 2002 at 01:05:46AM +0100, Alan Cox wrote:

 > You misunderstand the intent. A 386 or 486 kernel will run on a Pentium
 > and could therefore hit the error. A PPro kernel would die earlier
 > anyway. Of course its long been PPRO|Athlon|... and the ifdef wanted
 > updating. I'd ifdef it on CONFIG_X86_FOOF_BUG and put the FOOF thing
 > into arch/i386/Config.in nicely with the other stuff

Agreed. This is what's done in 2.5 btw.

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs