Re: [PATCH] x86/mce: Streamline MCE subsystem's naming
From: Ingo Molnar
Date: Thu Dec 06 2018 - 11:27:10 EST
* Borislav Petkov <bp@xxxxxxxxx> wrote:
> On Wed, Dec 05, 2018 at 07:01:58PM +0100, Ingo Molnar wrote:
> > Oh - I thought we'd have arch/x86/mce/ or so?
> >
> > There's machine check events that are not tied to any particular CPU,
> > correct? So this would be the right conceptual level - and it would also
> > remove the somewhat redundant 'kernel' part.
>
> Well, all the MCE events reported are some way or the other tied to the
> CPU and they're reported in the CPU's MCA banks so I think we want
>
> arch/x86/cpu/mce/
Well, *everything* the kernel does is in some way connected to a CPU,
because we always execute the kernel on a CPU, still we have things like
discrete PMUs and abstractions for other pieces of hardware that are not
per CPU.
So the real question is, is there a signifcant class of MCE events that
are not tied to the reporting channel which is per CPU (-ish ...) MCA
banks?
Thanks,
Ingo