Re: [PATCH v2 10/15] mm, swap: wrap swap cache replacement with a helper

[Kairui Song <ryncsn@gmail.com>]

On Mon, Sep 8, 2025 at 11:52 AM Baolin Wang
<baolin.wang@linux.alibaba.com> wrote:
>
>
>
> On 2025/9/6 03:13, Kairui Song wrote:
> > From: Kairui Song <kasong@tencent.com>
> >
> > There are currently three swap cache users that are trying to replace an
> > existing folio with a new one: huge memory splitting, migration, and
> > shmem replacement. What they are doing is quite similar.
> >
> > Introduce a common helper for this. In later commits, they can be easily
> > switched to use the swap table by updating this helper.
> >
> > The newly added helper also makes the swap cache API better defined, and
> > debugging is easier.
> >
> > Signed-off-by: Kairui Song <kasong@tencent.com>
> > ---
> >   mm/huge_memory.c |  5 ++---
> >   mm/migrate.c     | 11 +++--------
> >   mm/shmem.c       | 10 ++--------
> >   mm/swap.h        |  3 +++
> >   mm/swap_state.c  | 32 ++++++++++++++++++++++++++++++++
> >   5 files changed, 42 insertions(+), 19 deletions(-)
> >
> > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > index 26cedfcd7418..a4d192c8d794 100644
> > --- a/mm/huge_memory.c
> > +++ b/mm/huge_memory.c
> > @@ -3798,9 +3798,8 @@ static int __folio_split(struct folio *folio, unsigned int new_order,
> >                        * NOTE: shmem in swap cache is not supported yet.
> >                        */
> >                       if (swap_cache) {
> > -                             __xa_store(&swap_cache->i_pages,
> > -                                        swap_cache_index(new_folio->swap),
> > -                                        new_folio, 0);
> > +                             __swap_cache_replace_folio(swap_cache, new_folio->swap,
> > +                                                        folio, new_folio);
> >                               continue;
> >                       }
>
> IIUC, it doesn't seem like a simple function replacement here. It
> appears that the original code has a bug: if the 'new_folio' is a large
> folio after split, we need to iterate over each swap entry of the large
> swapcache folio and then restore the new 'new_folio'.
>

That should be OK. We have a check in uniform_split_supported and
non_uniform_split_supported that swapcache folio can only be splitted
into order0. And it seems there is no support for splitting pure
swapcache folio now.

Maybe we can try to enable and make use of higher order split
after this series for swapcache. I just had a try to use some hackish
code to split random folios in the swap cache to larger order, it seems
fine after this series.