Re: [RFC PATCH 1/3] mm/numa: add exclusive node pool and numa=standby boot parameter

[Gregory Price <gourry@gourry.net>]

On Thu, Jul 02, 2026 at 02:09:06PM +0300, Mike Rapoport wrote:
> On Tue, Jun 23, 2026 at 12:36:19PM -0400, Gregory Price wrote:
> > On Sun, Jun 14, 2026 at 12:08:31PM +0300, Mike Rapoport wrote:
> > > On Thu, Jun 11, 2026 at 10:04:01AM -0400, Gregory Price wrote:
> > > > On Thu, Jun 11, 2026 at 12:00:17PM +0300, Mike Rapoport wrote:
> > >  
> > > > So really i think you're pointing out that futex_init() here probably
> > > > shouldn't be using num_possible_nodes?
> > > 
> > > I'd rather say that num_possible_nodes() with and without CXL (or other
> > > differentiated memory) has different semantics.
> > > Maybe we need to add a new primitive for possible differentiated nodes and
> > > keep num_possible_nodes() to mean "number of possible nodes with normal
> > > memory".
> > >  
> > 
> > We'd have to define "normal" here a little more discretely.
> > 
> > Normal = N_MEMORY at __init?
> > Normal = N_MEMORY in the future?
> 
> Normal = not differentiated, no matter at __init or in the future.
> I.e. memory that kernel will use with existing allocation primitives.
> 
> > We also use the possible_nodes() mask to allocate per-node pgdat, so
> > the futex example is largely just another "hey look at this thing,
> > I wonder what other stuff is out there".
> 
> Right, futex is only one example. My point is that with multiplication of
> possible and not populated nodes the possible_nodes() mask may not reflect
> adequately the limits its callers look for.
> 
> And that maybe it's time to audit possible_nodes() callers.
>  

Totally reasonable, I have been thinking the same thing.

~Gregory