Skip to contentSkip to conversation

ip: on queued skb use skb_header_pointer instead of pskb_may_pull

PATCH In review Willem de Bruijn 2 participants 2 messages Jan 7, 2019 – Jan 10, 2019

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

PATCH 1/1 ip: on queued skb use skb_header_pointer instead of pskb_may_pull

From: Willem de Bruijn <[email protected]>

Commit 2efd4fca703a ("ip: in cmsg IP(V6)_ORIGDSTADDR call pskb_may_pull") avoided a read beyond the end of the skb linear segment by calling pskb_may_pull.

That function can trigger a BUG_ON in pskb_expand_head if the skb is shared, which it is when when peeking. It can also return ENOMEM.

Avoid both by switching to safer skb_header_pointer.

Fixes 2efd4fca703a ("ip: in cmsg IP(V6)_ORIGDSTADDR call pskb_may_pull")
---
 net/ipv4/ip_sockglue.c | 12 +++++-------
 net/ipv6/datagram.c    | 10 ++++------
 2 files changed, 9 insertions(+), 13 deletions(-)
net/ipv4/ip_sockglue.c +5 -7
index fffcc130900e..82f341e84fae 100644
@@ -148,19 +148,17 @@ static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
 
 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb)
 {
+	__be16 _ports[2], *ports;
 	struct sockaddr_in sin;
-	__be16 *ports;
-	int end;
-
-	end = skb_transport_offset(skb) + 4;
-	if (end > 0 && !pskb_may_pull(skb, end))
-		return;
 
 	/* All current transport protocols have the port numbers in the
 	 * first four bytes of the transport header and this function is
 	 * written with this assumption in mind.
 	 */
-	ports = (__be16 *)skb_transport_header(skb);
+	ports = skb_header_pointer(skb, skb_transport_offset(skb),
+				   sizeof(_ports), &_ports);
+	if (!ports)
+		return;
 
 	sin.sin_family = AF_INET;
 	sin.sin_addr.s_addr = ip_hdr(skb)->daddr;
net/ipv6/datagram.c +4 -6
index bde08aa549f3..c2262a7e2088 100644
@@ -700,17 +700,15 @@ void ip6_datagram_recv_specific_ctl(struct sock *sk, struct msghdr *msg,
 	}
 	if (np->rxopt.bits.rxorigdstaddr) {
 		struct sockaddr_in6 sin6;
-		__be16 *ports;
-		int end;
+		__be16 _ports[2], *ports;
 
-		end = skb_transport_offset(skb) + 4;
-		if (end <= 0 || pskb_may_pull(skb, end)) {
+		ports = skb_header_pointer(skb, skb_transport_offset(skb),
+					   sizeof(_ports), &_ports);
+		if (ports) {
 			/* All current transport protocols have the port numbers in the
 			 * first four bytes of the transport header and this function is
 			 * written with this assumption in mind.
 			 */
-			ports = (__be16 *)skb_transport_header(skb);
-
 			sin6.sin6_family = AF_INET6;
 			sin6.sin6_addr = ipv6_hdr(skb)->daddr;
 			sin6.sin6_port = ports[1];
Signature · 1 line
2.20.1.97.g81188d93c3-goog
maintainer
#1 lore raw

From: Willem de Bruijn <[email protected]>
Date: Mon, 7 Jan 2019 16:47:33 -0500

Quoted 9 lines

From: Willem de Bruijn <[email protected]>

Commit 2efd4fca703a ("ip: in cmsg IP(V6)_ORIGDSTADDR call pskb_may_pull") avoided a read beyond the end of the skb linear segment by calling pskb_may_pull.

That function can trigger a BUG_ON in pskb_expand_head if the skb is shared, which it is when when peeking. It can also return ENOMEM.

Quoted 6 lines

Avoid both by switching to safer skb_header_pointer.

Fixes 2efd4fca703a ("ip: in cmsg IP(V6)_ORIGDSTADDR call pskb_may_pull")

Applied and queued up for -stable.