Re: [PATCH 3/3] EDAC: carve out AMD MCE decoding logic

From: Ingo Molnar
Date: Fri Oct 02 2009 - 14:47:49 EST



* Borislav Petkov <borislav.petkov@xxxxxxx> wrote:

> On Fri, Oct 02, 2009 at 03:39:54PM +0200, Ingo Molnar wrote:
> > I suspect this is fine currently because no two EDAC modules should be
> > active at the same time. A followup cleanup patch would be nice
> > nevertheless that uses a notifier chain here with proper
> > register/unregister locking.
>
> Please do elaborate for I can't seem to follow.
>
> AFAICT, we are going to have max one MCE decoder at any one time so no
> need to notify it.
>
> The x86_mce_decode_callback assignment probably needs locking just to
> be on the safe side. We can do a
>
> mce_set_decode_callback(void (*f)(struct mce*m))
>
> helper which falls back to default_decode_mce() whenever its being
> passed a NULL for *f using proper locking...
>
> I'm pretty sure I'm missing something though...

No, the locking was all that i meant. Using atomic_notifier would solve
that. Make the default decoder low-prio, that way there's no need to do
the callback save/restore sequence either.

Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/