On Mon, 11 Jul 2022 21:53:04 +0100 Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
> On Fri, Jun 17, 2022 at 04:42:47PM +0100, Matthew Wilcox (Oracle) wrote:
> > @@ -2478,23 +2477,21 @@ static void shrink_active_list(unsigned long nr_to_scan,
> >
> > while (!list_empty(&l_hold)) {
> > struct folio *folio;
> > - struct page *page;
> >
> > cond_resched();
> > folio = lru_to_folio(&l_hold);
> > list_del(&folio->lru);
> > - page = &folio->page;
> >
> > - if (unlikely(!page_evictable(page))) {
> > - putback_lru_page(page);
> > + if (unlikely(!folio_evictable(folio))) {
> > + folio_putback_lru(folio);
> > continue;
> > }
> >
> > if (unlikely(buffer_heads_over_limit)) {
> > - if (page_has_private(page) && trylock_page(page)) {
> > - if (page_has_private(page))
> > - try_to_release_page(page, 0);
> > - unlock_page(page);
> > + if (folio_get_private(folio) && folio_trylock(folio)) {
> > + if (folio_get_private(folio))
> > + filemap_release_folio(folio, 0);
> > + folio_unlock(folio);
> > }
> > }
> >
>
> Hi Andrew. Hugh points out that the above is not an equivalent
> transformation for pages which are in the swapcache. Can you apply
> this fix, or would you like a full patch?
>
> ...
The original is in mm-stable and rebasing that is bad (sigh) so I'll
add this as a standalone patch. So yes, please send along the real
thing.
