On Mon, Jul 11, 2022 at 12:52 AM Maurizio Lombardi <mlombard@xxxxxxxxxx> wrote:
>
> A number of drivers call page_frag_alloc() with a
> fragment's size > PAGE_SIZE.
> In low memory conditions, __page_frag_cache_refill() may fail the order 3
> cache allocation and fall back to order 0;
> If this happens, the cache will be smaller than the fragment, causing
> memory corruptions.
>
> Prevent this from happening by checking if the newly allocated cache
> is large enough for the fragment; if not, the allocation will fail
> and page_frag_alloc() will return NULL.
>
> Signed-off-by: Maurizio Lombardi <mlombard@xxxxxxxxxx>
> ---
> mm/page_alloc.c | 11 ++++++++---
> 1 file changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e008a3df0485..7fb000d7e90c 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5611,12 +5611,17 @@ void *page_frag_alloc_align(struct page_frag_cache *nc,
> /* if size can vary use size else just use PAGE_SIZE */
> size = nc->size;
> #endif
> - /* OK, page count is 0, we can safely set it */
> - set_page_count(page, PAGE_FRAG_CACHE_MAX_SIZE + 1);
> -
> /* reset page count bias and offset to start of new frag */
> nc->pagecnt_bias = PAGE_FRAG_CACHE_MAX_SIZE + 1;
> offset = size - fragsz;
> + if (unlikely(offset < 0)) {
> + free_the_page(page, compound_order(page));
> + nc->va = NULL;
> + return NULL;
> + }
> +
> + /* OK, page count is 0, we can safely set it */
> + set_page_count(page, PAGE_FRAG_CACHE_MAX_SIZE + 1);
> }
>
> nc->pagecnt_bias--;
Rather than forcing us to free the page it might be better to move the
lines getting the size and computing the offset to the top of the "if
(unlikely(offset < 0)) {" block. Then instead of freeing the page we
could just return NULL and don't have to change the value of any
fields in the page_frag_cache.
That way a driver performing bad requests can't force us to start
allocating and freeing pages like mad by repeatedly flushing the
cache.
