netdev - [PATCH net] tcp: make sure skb is not shared before using skb

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <1423831632.4942.25.camel@edumazet-glaptop2.roam.corp.google.com>
Date:	Fri, 13 Feb 2015 04:47:12 -0800
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	David Miller <davem@...emloft.net>
Cc:	netdev <netdev@...r.kernel.org>, Yuchung Cheng <ycheng@...gle.com>
Subject: [PATCH net] tcp: make sure skb is not shared before using skb_get()

From: Eric Dumazet <edumazet@...gle.com>

IPv6 can keep a copy of SYN message using skb_get() in
tcp_v6_conn_request() so that caller wont free the skb when calling
kfree_skb() later.
    
Therefore TCP fast open has to clone the skb it is queuing in
child->sk_receive_queue, as all skbs consumed from receive_queue are
freed using __kfree_skb() (ie assuming skb->users == 1)
    
Signed-off-by: Eric Dumazet <edumazet@...gle.com>
Signed-off-by: Yuchung Cheng <ycheng@...gle.com>
Fixes: 5b7ed0892f2af ("tcp: move fastopen functions to tcp_fastopen.c")
---
 net/ipv4/tcp_fastopen.c |   32 ++++++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 8 deletions(-)

diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c
index 53db2c309572..ea82fd492c1b 100644
--- a/net/ipv4/tcp_fastopen.c
+++ b/net/ipv4/tcp_fastopen.c
@@ -134,6 +134,7 @@ static bool tcp_fastopen_create_child(struct sock *sk,
 	struct tcp_sock *tp;
 	struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue;
 	struct sock *child;
+	u32 end_seq;
 
 	req->num_retrans = 0;
 	req->num_timeout = 0;
@@ -185,20 +186,35 @@ static bool tcp_fastopen_create_child(struct sock *sk,
 
 	/* Queue the data carried in the SYN packet. We need to first
 	 * bump skb's refcnt because the caller will attempt to free it.
+	 * Note that IPv6 might also have used skb_get() trick
+	 * in tcp_v6_conn_request() to keep this SYN around (treq->pktopts)
+	 * So we need to eventually get a clone of the packet,
+	 * before inserting it in sk_receive_queue.
 	 *
 	 * XXX (TFO) - we honor a zero-payload TFO request for now,
 	 * (any reason not to?) but no need to queue the skb since
 	 * there is no data. How about SYN+FIN?
 	 */
-	if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq + 1) {
-		skb = skb_get(skb);
-		skb_dst_drop(skb);
-		__skb_pull(skb, tcp_hdr(skb)->doff * 4);
-		skb_set_owner_r(skb, child);
-		__skb_queue_tail(&child->sk_receive_queue, skb);
-		tp->syn_data_acked = 1;
+	end_seq = TCP_SKB_CB(skb)->end_seq;
+	if (end_seq != TCP_SKB_CB(skb)->seq + 1) {
+		struct sk_buff *skb2;
+
+		if (unlikely(skb_shared(skb)))
+			skb2 = skb_clone(skb, GFP_ATOMIC);
+		else
+			skb2 = skb_get(skb);
+
+		if (likely(skb2)) {
+			skb_dst_drop(skb2);
+			__skb_pull(skb2, tcp_hdrlen(skb));
+			skb_set_owner_r(skb2, child);
+			__skb_queue_tail(&child->sk_receive_queue, skb2);
+			tp->syn_data_acked = 1;
+		} else {
+			end_seq = TCP_SKB_CB(skb)->seq + 1;
+		}
 	}
-	tcp_rsk(req)->rcv_nxt = tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+	tcp_rsk(req)->rcv_nxt = tp->rcv_nxt = end_seq;
 	sk->sk_data_ready(sk);
 	bh_unlock_sock(child);
 	sock_put(child);


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html