Re: [PATCH net 2/4] xen-netback: Fix releasing frag_list skbs in error path

[Wei Liu <wei.liu2@citrix.com>]

On Thu, Jul 17, 2014 at 08:09:50PM +0100, Zoltan Kiss wrote:
> When the grant operations failed, the skb is freed up eventually, and it tries
> to release the frags, if there is any. For the main skb nr_frags is set to 0 to
> avoid this, but on the frag_list it iterates through the frags array, and tries
> to call put_page on the page pointer which contains garbage at that time.
> 
> Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
> Reported-by: Armin Zentai <armin.zentai@ezit.hu>
> Cc: netdev@vger.kernel.org
> Cc: linux-kernel@vger.kernel.org
> Cc: xen-devel@lists.xenproject.org
> ---
> diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
> index 604ff71..e9ffb05 100644
> --- a/drivers/net/xen-netback/netback.c
> +++ b/drivers/net/xen-netback/netback.c
> @@ -1522,6 +1522,11 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
>  		/* Check the remap error code. */
>  		if (unlikely(xenvif_tx_check_gop(queue, skb, &gop_map, &gop_copy))) {

It's worth adding a comment here saying that all those pages have been
freed in xenvif_tx_check_gop so that we don't leak any page even if we
manually set nr_frags to 0.

Wei.

>  			skb_shinfo(skb)->nr_frags = 0;
> +			if (skb_has_frag_list(skb)) {
> +				struct sk_buff *nskb =
> +						skb_shinfo(skb)->frag_list;
> +				skb_shinfo(nskb)->nr_frags = 0;
> +			}
>  			kfree_skb(skb);
>  			continue;
>  		}