Re: [PATCH] memcg: keep propagating stats updates to ancestors

[Mingyu He]

On Mon, Oct 12, 2026 at 7:08 PM Shakeel Butt <shakeel.butt@xxxxxxxxx> wrote:

> Please share your custom in-kernel consumer. Let's not make changes in the
> kernel for things which are not in upstreame kernel.

My custom kernel only has a few changes from upstream.
I just call mem_cgroup_flush_stats() and then call memcg_page_state() in
a kworker to judge whether to do a page cache shrinking.

> >   Regarding the BPF interface, the implementation in the tree I found is:
> >
> >   __bpf_kfunc void bpf_mem_cgroup_flush_stats(struct mem_cgroup *memcg)
> >   {
> >           mem_cgroup_flush_stats(memcg);
>
> Let's make this force flush.

I am not sure what you mean. Do you mean that only change this bpf function to
force flush, and keep memcg_rstat_updated unchanged?
If you mean that, I think this is a good approach.
But I think we also had better to add some comments to memcg_rstat_updated()
to inform callers about this behavior ( maybe has a large data flush latency).

Thanks,
Runli