Re: [PATCH 3/5] x86: Kconfig: Remove CONFIG_NODES_SPAN_OTHER_NODES

From: Thomas Gleixner
Date: Wed Jul 10 2019 - 01:59:14 EST


Hoan,

On Wed, 10 Jul 2019, Hoan Tran OS wrote:
> On 6/25/19 3:45 PM, Thomas Gleixner wrote:
> > On Tue, 25 Jun 2019, Hoan Tran OS wrote:
> >> @@ -1567,15 +1567,6 @@ config X86_64_ACPI_NUMA
> >> ---help---
> >> Enable ACPI SRAT based node topology detection.
> >>
> >> -# Some NUMA nodes have memory ranges that span
> >> -# other nodes. Even though a pfn is valid and
> >> -# between a node's start and end pfns, it may not
> >> -# reside on that node. See memmap_init_zone()
> >> -# for details.
> >> -config NODES_SPAN_OTHER_NODES
> >> - def_bool y
> >> - depends on X86_64_ACPI_NUMA
> >
> > the changelog does not mention that this lifts the dependency on
> > X86_64_ACPI_NUMA and therefore enables that functionality for anything
> > which has NUMA enabled including 32bit.
> >
>
> I think this config is used for a NUMA layout which NUMA nodes addresses
> are spanned to other nodes. I think 32bit NUMA also have the same issue
> with that layout. Please correct me if I'm wrong.

I'm not saying you're wrong, but it's your duty to provide the analysis why
this is correct for everything which has NUMA enabled.

> > The core mm change gives no helpful information either. You just copied the
> > above comment text from some random Kconfig.
>
> Yes, as it's a correct comment and is used at multiple places.

Well it maybe correct in terms of explaining what this is about, it still
does not explain why this is needed by default on everything which has NUMA
enabled.

Thanks,

tglx