ipv6: restore correct ECN handling on TCP xmit
authorSteinar H. Gunderson <sgunderson@bigfoot.com>
Fri, 6 May 2011 23:44:46 +0000 (23:44 +0000)
committerDavid S. Miller <davem@davemloft.net>
Thu, 12 May 2011 22:52:14 +0000 (18:52 -0400)
Since commit e9df2e8fd8fbc9 (Use appropriate sock tclass setting for
routing lookup) we lost ability to properly add ECN codemarks to ipv6
TCP frames.

It seems like TCP_ECN_send() calls INET_ECN_xmit(), which only sets the
ECN bit in the IPv4 ToS field (inet_sk(sk)->tos), but after the patch,
what's checked is inet6_sk(sk)->tclass, which is a completely different
field.

Close bug https://bugzilla.kernel.org/show_bug.cgi?id=34322

[Eric Dumazet] : added the INET_ECN_dontxmit() fix and replace macros
by inline functions for clarity.

Signed-off-by: Steinar H. Gunderson <sgunderson@bigfoot.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/net/inet_ecn.h

index 88bdd010d65d3791eb634cae72709e01fc2c2090..2fa8d1341a0afa291e1fd5aa74bf0f7c4c91806f 100644 (file)
@@ -38,9 +38,19 @@ static inline __u8 INET_ECN_encapsulate(__u8 outer, __u8 inner)
        return outer;
 }
 
-#define        INET_ECN_xmit(sk) do { inet_sk(sk)->tos |= INET_ECN_ECT_0; } while (0)
-#define        INET_ECN_dontxmit(sk) \
-       do { inet_sk(sk)->tos &= ~INET_ECN_MASK; } while (0)
+static inline void INET_ECN_xmit(struct sock *sk)
+{
+       inet_sk(sk)->tos |= INET_ECN_ECT_0;
+       if (inet6_sk(sk) != NULL)
+               inet6_sk(sk)->tclass |= INET_ECN_ECT_0;
+}
+
+static inline void INET_ECN_dontxmit(struct sock *sk)
+{
+       inet_sk(sk)->tos &= ~INET_ECN_MASK;
+       if (inet6_sk(sk) != NULL)
+               inet6_sk(sk)->tclass &= ~INET_ECN_MASK;
+}
 
 #define IP6_ECN_flow_init(label) do {          \
       (label) &= ~htonl(INET_ECN_MASK << 20);  \