skbuff: fix ftrace handling in skb_unshare

If the skb is not dropped afterwards we should run consume_skb instead kfree_skb. Inside of function skb_unshare we do always a kfree_skb, doesn't depend if skb_copy failed or was successful.

This patch switch this behaviour like skb_share_check, if allocation of sk_buff failed we use kfree_skb otherwise consume_skb.

---
 include/linux/skbuff.h | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

index abde271..d150734 100644
@@ -1116,7 +1116,12 @@ static inline struct sk_buff *skb_unshare(struct sk_buff *skb,
 	might_sleep_if(pri & __GFP_WAIT);
 	if (skb_cloned(skb)) {
 		struct sk_buff *nskb = skb_copy(skb, pri);
-		kfree_skb(skb);	/* Free our shared copy */
+
+		/* Free our shared copy */
+		if (likely(nskb))
+			consume_skb(skb);
+		else
+			kfree_skb(skb);
 		skb = nskb;
 	}
 	return skb;

2.1.2

From: Alexander Aring <[email protected]>
Date: Fri, 10 Oct 2014 23:10:47 +0200

Applied, thanks.