Re: [RFC PATCH v2 0/3] BPF-based per-cgroup NUMA balancing mode selection

[Yafang Shao]

On Tue, Oct 13, 2026 at 1:51 AM Tejun Heo <tj@xxxxxxxxxx> wrote:
>
> Hello,
>
> On Mon, Oct 12, 2026 at 08:04:35PM +0800, Yafang Shao wrote:
> > This series adds a bpf_numab_ops struct_ops which attaches to a
> > cgroup. The kernel consults the policies of the matching cgroup and of
> > every ancestor and combines what they return.
>
> The op doesn't compute anything. It returns a fixed mode per cgroup and
> that's all the series uses it for. BPF is unnecessary here. This is a
> prctl() feature, not a struct_ops one.

Part of the consults are keyed on struct folio rather than struct
task_struct: the folio surface answers from the folio's memcg, the
owner's request, not the touching task's. I don't see how to key
these callsites to a per-task setting.  Please correct me if I'm
wrong.

prctl() is also a stable uAPI rather than a kernel-internal
interface: a new option is frozen the day it merges, which is a
poor fit for a feature whose semantics are still moving.  The
struct_ops/kfunc surface is explicitly exempt from the uAPI
stability rules, which is what lets us experiment with the kernel
feature rapidly.

-- 
Regards
Yafang