Understood, thanks for your explanation :) 在 2022/7/27 19:40, William Kucharski 写道:
This isn't a NULL pointer dereference; returning &(0->page) is completely legal as was discussed regarding this exact code back in April: https://lore.kernel.org/lkml/YmfgqKcMmstgfz+0@xxxxxxxxxxxxxxxxxxxx/On Jul 27, 2022, at 12:46 AM, Yongqiang Liu <liuyongqiang13@xxxxxxxxxx> wrote: When __filemap_get_folio() failed and returned NULL, we would get a NULL pointer dereference in pagecache_get_page. Fixes: 3f0c6a07fee6 ("mm/filemap: Add filemap_get_folio") Signed-off-by: Yongqiang Liu <liuyongqiang13@xxxxxxxxxx> Cc: <stable@xxxxxxxxxxxxxxx> # 5.16 --- mm/folio-compat.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/mm/folio-compat.c b/mm/folio-compat.c index 20bc15b57d93..7b21393480e0 100644 --- a/mm/folio-compat.c +++ b/mm/folio-compat.c @@ -124,7 +124,9 @@ struct page *pagecache_get_page(struct address_space *mapping, pgoff_t index, struct folio *folio; folio = __filemap_get_folio(mapping, index, fgp_flags, gfp); - if ((fgp_flags & FGP_HEAD) || !folio || xa_is_value(folio)) + if (!folio) + return NULL; + if ((fgp_flags & FGP_HEAD) || xa_is_value(folio)) return &folio->page; return folio_file_page(folio, index); } -- 2.25.1.
