]> asedeno.scripts.mit.edu Git - linux.git/commitdiff
net: drop skb on failure in ip_check_defrag()
authorCong Wang <xiyou.wangcong@gmail.com>
Thu, 1 Nov 2018 19:02:37 +0000 (12:02 -0700)
committerDavid S. Miller <davem@davemloft.net>
Thu, 1 Nov 2018 20:55:30 +0000 (13:55 -0700)
Most callers of pskb_trim_rcsum() simply drop the skb when
it fails, however, ip_check_defrag() still continues to pass
the skb up to stack. This is suspicious.

In ip_check_defrag(), after we learn the skb is an IP fragment,
passing the skb to callers makes no sense, because callers expect
fragments are defrag'ed on success. So, dropping the skb when we
can't defrag it is reasonable.

Note, prior to commit 88078d98d1bb, this is not a big problem as
checksum will be fixed up anyway. After it, the checksum is not
correct on failure.

Found this during code review.

Fixes: 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/ip_fragment.c

index 9b0158fa431f2245c0fa7e21d62e3ac01296dc20..d6ee343fdb8647ea96240d017b72aef2f6790299 100644 (file)
@@ -722,10 +722,14 @@ struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
        if (ip_is_fragment(&iph)) {
                skb = skb_share_check(skb, GFP_ATOMIC);
                if (skb) {
-                       if (!pskb_may_pull(skb, netoff + iph.ihl * 4))
-                               return skb;
-                       if (pskb_trim_rcsum(skb, netoff + len))
-                               return skb;
+                       if (!pskb_may_pull(skb, netoff + iph.ihl * 4)) {
+                               kfree_skb(skb);
+                               return NULL;
+                       }
+                       if (pskb_trim_rcsum(skb, netoff + len)) {
+                               kfree_skb(skb);
+                               return NULL;
+                       }
                        memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));
                        if (ip_defrag(net, skb, user))
                                return NULL;