On Mon, Sep 02, 2024 at 03:21:12PM +0800, alexs@xxxxxxxxxx wrote:
> And keep the memcg_data member, since as Yosry pointed out:
> "When the pages are freed, put_page() -> folio_put() -> __folio_put()
> will call mem_cgroup_uncharge(). The latter will call folio_memcg()
> (which reads folio->memcg_data) to figure out if uncharging needs to
> be done.
... why does that mean we need to keep memcg_data in this definition?
If it's unused, it will stay unused.
> There are also other similar code paths that will check
> folio->memcg_data. It is currently expected to be present for all
> folios. So until we have custom code paths per-folio type for
> allocation/freeing/etc, we need to keep folio->memcg_data present and
> properly initialized."
...
> +struct zpdesc {
> + unsigned long flags;
> + struct list_head lru;
> + struct movable_operations *mops;
Hm. I'm not sure this is a wise definition. While yes, we know it
will only point to the mops, we always set the bottom two bits to '10'.
So I think it's safer to make this 'unsigned long mops' and force people
to _not_ use it directly.
