| Message ID | 20230427092159.44998-1-atenart@kernel.org (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | dc6456e938e938d64ffb6383a286b2ac9790a37f |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] net: ipv6: fix skb hash for some RST packets | expand |
On Thu, Apr 27, 2023 at 11:22 AM Antoine Tenart <atenart@kernel.org> wrote: > > The skb hash comes from sk->sk_txhash when using TCP, except for some > IPv6 RST packets. This is because in tcp_v6_send_reset when not in > TIME_WAIT the hash is taken from sk->sk_hash, while it should come from > sk->sk_txhash as those two hashes are not computed the same way. > > Packetdrill script to test the above, > > 0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3 > +0 fcntl(3, F_SETFL, O_RDWR|O_NONBLOCK) = 0 > +0 connect(3, ..., ...) = -1 EINPROGRESS (Operation now in progress) > > +0 > (flowlabel 0x1) S 0:0(0) <...> > > // Wrong ack seq, trigger a rst. > +0 < S. 0:0(0) ack 0 win 4000 > > // Check the flowlabel matches prior one from SYN. > +0 > (flowlabel 0x1) R 0:0(0) <...> > > Fixes: 9258b8b1be2e ("ipv6: tcp: send consistent autoflowlabel in RST packets") > Signed-off-by: Antoine Tenart <atenart@kernel.org> > --- Good catch, thanks ! Reviewed-by: Eric Dumazet <edumazet@google.com>
Hello: This patch was applied to netdev/net.git (main) by David S. Miller <davem@davemloft.net>: On Thu, 27 Apr 2023 11:21:59 +0200 you wrote: > The skb hash comes from sk->sk_txhash when using TCP, except for some > IPv6 RST packets. This is because in tcp_v6_send_reset when not in > TIME_WAIT the hash is taken from sk->sk_hash, while it should come from > sk->sk_txhash as those two hashes are not computed the same way. > > Packetdrill script to test the above, > > [...] Here is the summary with links: - [net] net: ipv6: fix skb hash for some RST packets https://git.kernel.org/netdev/net/c/dc6456e938e9 You are awesome, thank you!
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index 1bf93b61aa06..e2898912c90c 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -1064,7 +1064,7 @@ static void tcp_v6_send_reset(const struct sock *sk, struct sk_buff *skb) if (np->repflow) label = ip6_flowlabel(ipv6h); priority = sk->sk_priority; - txhash = sk->sk_hash; + txhash = sk->sk_txhash; } if (sk->sk_state == TCP_TIME_WAIT) { label = cpu_to_be32(inet_twsk(sk)->tw_flowlabel);
The skb hash comes from sk->sk_txhash when using TCP, except for some IPv6 RST packets. This is because in tcp_v6_send_reset when not in TIME_WAIT the hash is taken from sk->sk_hash, while it should come from sk->sk_txhash as those two hashes are not computed the same way. Packetdrill script to test the above, 0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3 +0 fcntl(3, F_SETFL, O_RDWR|O_NONBLOCK) = 0 +0 connect(3, ..., ...) = -1 EINPROGRESS (Operation now in progress) +0 > (flowlabel 0x1) S 0:0(0) <...> // Wrong ack seq, trigger a rst. +0 < S. 0:0(0) ack 0 win 4000 // Check the flowlabel matches prior one from SYN. +0 > (flowlabel 0x1) R 0:0(0) <...> Fixes: 9258b8b1be2e ("ipv6: tcp: send consistent autoflowlabel in RST packets") Signed-off-by: Antoine Tenart <atenart@kernel.org> --- net/ipv6/tcp_ipv6.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)