Prev: [PATCH] init - dont loose initcall code returned v2
Next: It's time to hack on dentry_stat?
From: Andi Kleen on 7 May 2008 10:20
stephane eranian wrote:
> Hello,
>
>
> Here is a first perfmon patch. It provides the following:
>
> - minimal perfmon source tree layout with a topdir perfmon for
> generic code and
> the x86 arch specific subdir in arch/x86.

How much larger would a patch be that does at least something minimally
useful? (like context switching the performance registers)

-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: stephane eranian on 7 May 2008 10:30
Andi,

It would be much bigger because it would pull in a bunch of generic
and model-specific
data structures and code.

To context switch, you need to know what the registers are, thus you
need the PMU
mapping table, which is implemented by a kernel module, so it needs
its registration
framework....


On Wed, May 7, 2008 at 4:17 PM, Andi Kleen <andi(a)firstfloor.org> wrote:
> stephane eranian wrote:
> > Hello,
> >
> >
> > Here is a first perfmon patch. It provides the following:
> >
> > - minimal perfmon source tree layout with a topdir perfmon for
> > generic code and
> > the x86 arch specific subdir in arch/x86.
>
> How much larger would a patch be that does at least something minimally
> useful? (like context switching the performance registers)
>
> -Andi
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Andi Kleen on 7 May 2008 10:40
stephane eranian wrote:
> Andi,
>
> It would be much bigger because it would pull in a bunch of generic
> and model-specific
> data structures and code.
>
> To context switch, you need to know what the registers are, thus you
> need the PMU
> mapping table, which is implemented by a kernel module, so it needs
> its registration
> framework....

Can you strip it down to be simpler?

Minimally generic x86 perfctr switching code just needs an array with
MSR numbers, doesn't it? And there are only a few variants.

-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: [PATCH] init - dont loose initcall code returned v2
Next: It's time to hack on dentry_stat?