Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220713150143.147537-1-mlombard@redhat.com>
In-Reply-To: <20220713150143.147537-1-mlombard@redhat.com>
From:   Alexander Duyck <alexander.duyck@gmail.com>
Date:   Wed, 13 Jul 2022 08:12:49 -0700
Message-ID: <CAKgT0Uet==0v3bFV8KcnXLxB9BoC8qyMRkeH-X5sfWE7Bm7ikg@mail.gmail.com>
Subject: Re: [PATCH V2] mm: prevent page_frag_alloc() from corrupting the memory
To:     Maurizio Lombardi <mlombard@redhat.com>
Cc:     Jakub Kicinski <kuba@kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        linux-mm <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Netdev <netdev@vger.kernel.org>, Chen Lin <chen45464546@163.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Jul 13, 2022 at 8:01 AM Maurizio Lombardi <mlombard@redhat.com> 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;
> In this case, 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.
>
> V2: do not free the cache page because this could make memory pressure
> even worse, just return NULL.
>
> Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>
> ---
>  mm/page_alloc.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e008a3df0485..b1407254a826 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5617,6 +5617,8 @@ void *page_frag_alloc_align(struct page_frag_cache *nc,
>                 /* 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))
> +                       return NULL;
>         }
>
>         nc->pagecnt_bias--;

This works for me. If I am not mistaken it should be only adding one
conditional jump that isn't taken to the fast path based on a
calculation we were already doing.

Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>