Re: [PATCH 3/6] macvtap: zerocopy: validate vector length beforepinning user pages

[Eric Dumazet]

On Mon, 2012-04-16 at 14:08 +0800, Jason Wang wrote:
> Currently we do not validate the vector length before calling
> get_user_pages_fast(), host stack would be easily overflowed by
> malicious guest driver who give us a descriptor with length greater
> than MAX_SKB_FRAGS. Solve this problem by checking the free entries
> before trying to pin user pages.
> 
> Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx>
> ---
>  drivers/net/macvtap.c |    2 ++
>  1 files changed, 2 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/net/macvtap.c b/drivers/net/macvtap.c
> index 7cb2684..d197a78 100644
> --- a/drivers/net/macvtap.c
> +++ b/drivers/net/macvtap.c
> @@ -529,6 +529,8 @@ static int zerocopy_sg_from_iovec(struct sk_buff *skb, const struct iovec *from,
>  		}
>  		base = (unsigned long)from->iov_base + offset;
>  		size = ((base & ~PAGE_MASK) + len + ~PAGE_MASK) >> PAGE_SHIFT;
> +		if (i + size >= MAX_SKB_FRAGS)
> +			return -EFAULT;
>  		num_pages = get_user_pages_fast(base, size, 0, &page[i]);
>  		if ((num_pages != size) ||
>  		    (num_pages > MAX_SKB_FRAGS - skb_shinfo(skb)->nr_frags))
> 

Hi Jason

Why is -EFAULT the right error code ?

Also, why not removing the "(num_pages > MAX_SKB_FRAGS -
skb_shinfo(skb)->nr_frags)" test done few lines after ?

Also, if get_user_pages_fast() returns truncated result (size - 1 for
example), we apparently dont free the references on pages.

The comment applies only on pages that were added in skb frags.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/