Re: [PATCH 10/16] x86/amd_nb: Move SMN access code to a new amd_smn driver

[Yazen Ghannam <yazen.ghannam@amd.com>]

On Mon, Nov 04, 2024 at 03:29:58PM +0100, Borislav Petkov wrote:
> On Wed, Oct 23, 2024 at 05:21:44PM +0000, Yazen Ghannam wrote:
> > diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> > index ba5252d8e21c..a03ffa5b6bb1 100644
> > --- a/arch/x86/Kconfig
> > +++ b/arch/x86/Kconfig
> > @@ -3128,6 +3128,9 @@ config AMD_NODE
> >  	def_bool y
> >  	depends on CPU_SUP_AMD && PCI
> >  
> > +config AMD_SMN
> > +	def_bool y
> > +	depends on AMD_NODE
> 
> Why is this a separate compilation unit and not part of amd_node.c? Especially
> if it depends on it.
> 
> I don't see the real need for having smaller compilation units. Both the node
> and the smn stuff will end up being built-in in 99% of the configs. So why are
> we making separate Kconfig items and yadda yadda?
> 

AMD_NB (legacy systems) and AMD_SMN (Zen-based systems) would both
depend on AMD_NODE. The thinking is that a user could build a minimal
config for either legacy or Zen-based systems.

> Just do a single amd_node and that's it. We can always split later, if really
> needed.
>

Okay, will do.

Thanks,
Yazen