Re: [PATCH] mm: mempolicy: fix automatic numa balancing for shmem

[Yosry Ahmed]

On Thu, Jul 23, 2026 at 11:45 AM Gregory Price <gourry@xxxxxxxxxx> wrote:
>
> On Thu, Jul 23, 2026 at 01:48:27PM -0400, Gregory Price wrote:
> > On Thu, Jul 23, 2026 at 01:07:04PM -0400, Johannes Weiner wrote:
> >
> > So the default global mempolicy is effectively:
> >
> >     (MPOL_PREFERRED | (MPOL_F_MOF | MPOL_F_MORON)
> >
> > which is just MPOL_F_NUMA_BALANCING without MPOL_F_NUMA_BALANCING.
> >
> > The user can't actually see that - because the MOF/MORON flags get
> > stripped out when it's queried (because they're internal only).
> >
> > Maybe we should just set MPOL_F_NUMA_BALANCING in the default/preferred
> > policies and document that the global preferred policy default-enables
> > MPOL_F_NUMA_BALANCING - and that if a user (KVM) doesn't want to be
> > affected by numa balancing it needs to set a policy to disable it.
> >
>
> Walking this all back:
>
>   - MPOL_DEFAULT is the default policy
>   - numa_balancing=on means enable NUMA_BALANCING on the defaul policy
>   - shmem was broken and did not respect this
>
> Johannes' fix is correct, and there are no additional changes required.
>
> Users enabling numa balancing while running KVM with shared regions are
> not getting what they're asking for presently - that is a bug.
>
> They should get the performance hit they're asking for (and then either
> turn off numa balancing or set an explicit mempolicy for those tasks /
> memfd's).

Does it make sense to add a pr_warn() or pr_info() to help users
figure this out when they see a regression?