On Wed, Jul 10, 2019 at 2:32 PM Henry Burns <henryburns@xxxxxxxxxx> wrote:
>
> z3fold_page_migrate() will never succeed because it attempts to acquire a
> lock that has already been taken by migrate.c in __unmap_and_move().
>
> __unmap_and_move() migrate.c
> trylock_page(oldpage)
> move_to_new_page(oldpage_newpage)
> a_ops->migrate_page(oldpage, newpage)
> z3fold_page_migrate(oldpage, newpage)
> trylock_page(oldpage)
>
>
> Signed-off-by: Henry Burns <henryburns@xxxxxxxxxx>
Reviewed-by: Shakeel Butt <shakeelb@xxxxxxxxxx>
Please add the Fixes tag as well.
> ---
> mm/z3fold.c | 6 ------
> 1 file changed, 6 deletions(-)
>
> diff --git a/mm/z3fold.c b/mm/z3fold.c
> index 985732c8b025..9fe9330ab8ae 100644
> --- a/mm/z3fold.c
> +++ b/mm/z3fold.c
> @@ -1335,16 +1335,11 @@ static int z3fold_page_migrate(struct address_space *mapping, struct page *newpa
> zhdr = page_address(page);
> pool = zhdr_to_pool(zhdr);
>
> - if (!trylock_page(page))
> - return -EAGAIN;
> -
> if (!z3fold_page_trylock(zhdr)) {
> - unlock_page(page);
> return -EAGAIN;
> }
> if (zhdr->mapped_count != 0) {
> z3fold_page_unlock(zhdr);
> - unlock_page(page);
> return -EBUSY;
> }
> new_zhdr = page_address(newpage);
> @@ -1376,7 +1371,6 @@ static int z3fold_page_migrate(struct address_space *mapping, struct page *newpa
> queue_work_on(new_zhdr->cpu, pool->compact_wq, &new_zhdr->work);
>
> page_mapcount_reset(page);
> - unlock_page(page);
> put_page(page);
> return 0;
> }
> --
> 2.22.0.410.gd8fdbe21b5-goog
>
