tcp: fix 0 divide in __tcp_select_window()
authorEric Dumazet <edumazet@google.com>
Wed, 1 Feb 2017 16:33:53 +0000 (08:33 -0800)
committerBen Hutchings <ben@decadent.org.uk>
Thu, 16 Mar 2017 02:18:57 +0000 (02:18 +0000)
[ Upstream commit 06425c308b92eaf60767bc71d359f4cbc7a561f8 ]

syszkaller fuzzer was able to trigger a divide by zero, when
TCP window scaling is not enabled.

SO_RCVBUF can be used not only to increase sk_rcvbuf, also
to decrease it below current receive buffers utilization.

If mss is negative or 0, just return a zero TCP window.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
net/ipv4/tcp_output.c

index 8327d11..1c96e9b 100644 (file)
@@ -1923,9 +1923,11 @@ u32 __tcp_select_window(struct sock *sk)
        int full_space = min_t(int, tp->window_clamp, tcp_full_space(sk));
        int window;
 
-       if (mss > full_space)
+       if (unlikely(mss > full_space)) {
                mss = full_space;
-
+               if (mss <= 0)
+                       return 0;
+       }
        if (free_space < (full_space >> 1)) {
                icsk->icsk_ack.quick = 0;