Re: Bug in computing data_len in tcp_sendmsg?

[Eric Dumazet <eric.dumazet@gmail.com>]

Le vendredi 02 décembre 2011 à 12:59 +0100, Eric Dumazet a écrit :

> Thanks for this detailed explanation !
> 
> And yes, you're probably right.
> 
> Are you willing to submit a patch to fix this ?
> 
> (If not, I can do it myself of course)
> 

[PATCH net-next] tcp: fix tcp_trim_head()

commit f07d960df3 (tcp: avoid frag allocation for small frames)
breaked assumption in tcp stack that skb is either linear (data_len ==
0), either fully fragged (data_len == len)

Thanks to Vijay for providing a very detailed explanation.

Reported-by: Vijay Subramanian <subramanian.vijay@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
 net/ipv4/tcp_output.c |   13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 58f69ac..4a0f54a 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1093,6 +1093,13 @@ static void __pskb_trim_head(struct sk_buff *skb, int len)
 {
 	int i, k, eat;
 
+	eat = min_t(int, len, skb_headlen(skb));
+	if (eat) {
+		__skb_pull(skb, eat);
+		len -= eat;
+		if (!len)
+			return;
+	}
 	eat = len;
 	k = 0;
 	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
@@ -1124,11 +1131,7 @@ int tcp_trim_head(struct sock *sk, struct sk_buff *skb, u32 len)
 	if (skb_cloned(skb) && pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
 		return -ENOMEM;
 
-	/* If len == headlen, we avoid __skb_pull to preserve alignment. */
-	if (unlikely(len < skb_headlen(skb)))
-		__skb_pull(skb, len);
-	else
-		__pskb_trim_head(skb, len - skb_headlen(skb));
+	__pskb_trim_head(skb, len);
 
 	TCP_SKB_CB(skb)->seq += len;
 	skb->ip_summed = CHECKSUM_PARTIAL;