Skip to contentSkip to conversation

net: Fix oops from tcp_collapse() when using splice()

PATCH In review Steven J. Magnani 2 participants 3 messages Mar 30, 2010

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: Fix oops from tcp_collapse() when using splice()

tcp_read_sock() can have a eat skbs without immediately advancing copied_seq. This can cause a panic in tcp_collapse() if it is called as a result of the recv_actor dropping the socket lock.

A userspace program that splices data from a socket to either another socket or to a file can trigger this bug.

---

net/ipv4/tcp.c +1 -0
diff -uprN a/net/ipv4/tcp.c b/net/ipv4/tcp.c
@@ -1335,6 +1335,7 @@ int tcp_read_sock(struct sock *sk, read_
 		sk_eat_skb(sk, skb, 0);
 		if (!desc->count)
 			break;
+		tp->copied_seq = seq;
 	}
 	tp->copied_seq = seq;
 
maintainer
#1 lore raw

From: "Steven J. Magnani" <[email protected]>
Date: Tue, 30 Mar 2010 15:45:13 -0500

Quoted 8 lines

tcp_read_sock() can have a eat skbs without immediately advancing copied_seq. This can cause a panic in tcp_collapse() if it is called as a result of the recv_actor dropping the socket lock.

A userspace program that splices data from a socket to either another socket or to a file can trigger this bug.

Thanks for fixing this I'll look at your patch more closely right now.

maintainer
#2 lore raw

From: David Miller <[email protected]>
Date: Tue, 30 Mar 2010 13:47:41 -0700 (PDT)

Quoted 7 lines

From: "Steven J. Magnani" <[email protected]>
Date: Tue, 30 Mar 2010 15:45:13 -0500

Quoted 4 lines

tcp_read_sock() can have a eat skbs without immediately advancing copied_seq. This can cause a panic in tcp_collapse() if it is called as a result of the recv_actor dropping the socket lock.

Quoted 7 lines
Quoted 4 lines

A userspace program that splices data from a socket to either another socket or to a file can trigger this bug.

Thanks for fixing this I'll look at your patch more closely right now.

Patch applied, thanks Steven!