| Message ID | 20220421005026.686A45EC01F2@us226.sjc.aristanetworks.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 5b0b9e4c2c895227c8852488b3f09839233bba54 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [v2,net] tcp: md5: incorrect tcp_header_len for incoming connections | expand |
On Wed, Apr 20, 2022 at 5:50 PM Francesco Ruggeri <fruggeri@arista.com> wrote: > > In tcp_create_openreq_child we adjust tcp_header_len for md5 using the > remote address in newsk. But that address is still 0 in newsk at this > point, and it is only set later by the callers (tcp_v[46]_syn_recv_sock). > Use the address from the request socket instead. > > > > Fixes: cfb6eeb4c860 ("[TCP]: MD5 Signature Option (RFC2385) support.") > Signed-off-by: Francesco Ruggeri <fruggeri@arista.com> Thanks for fixing this Francesco. Reviewed-by: Eric Dumazet <edumazet@google.com>
Hello: This patch was applied to netdev/net.git (master) by Jakub Kicinski <kuba@kernel.org>: On Wed, 20 Apr 2022 17:50:26 -0700 you wrote: > In tcp_create_openreq_child we adjust tcp_header_len for md5 using the > remote address in newsk. But that address is still 0 in newsk at this > point, and it is only set later by the callers (tcp_v[46]_syn_recv_sock). > Use the address from the request socket instead. > > v2: Added "Fixes:" line. > > [...] Here is the summary with links: - [v2,net] tcp: md5: incorrect tcp_header_len for incoming connections https://git.kernel.org/netdev/net/c/5b0b9e4c2c89 You are awesome, thank you!
On Wed, Apr 20, 2022 at 5:50 PM Francesco Ruggeri <fruggeri@arista.com> wrote: > > In tcp_create_openreq_child we adjust tcp_header_len for md5 using the > remote address in newsk. But that address is still 0 in newsk at this > point, and it is only set later by the callers (tcp_v[46]_syn_recv_sock). > Use the address from the request socket instead. > > v2: Added "Fixes:" line. > > Fixes: cfb6eeb4c860 ("[TCP]: MD5 Signature Option (RFC2385) support.") > Signed-off-by: Francesco Ruggeri <fruggeri@arista.com> > --- > net/ipv4/tcp_minisocks.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c > index 6366df7aaf2a..6854bb1fb32b 100644 > --- a/net/ipv4/tcp_minisocks.c > +++ b/net/ipv4/tcp_minisocks.c > @@ -531,7 +531,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, > newtp->tsoffset = treq->ts_off; > #ifdef CONFIG_TCP_MD5SIG > newtp->md5sig_info = NULL; /*XXX*/ > - if (newtp->af_specific->md5_lookup(sk, newsk)) > + if (treq->af_specific->req_md5_lookup(sk, req_to_sk(req))) Wait a minute. Are you sure treq->af_specific is initialized at this point ? I should have tested this one liner patch really :/ I think that for syncookies, treq->af_specific is not initialized, because we do not go through tcp_conn_request() helper, but instead use cookie_tcp_reqsk_alloc() Before your patch treq->af_specific was only used during SYNACK generation, which does not happen in syncookie more while receiving the third packet. I will test something like this patch. We could move the init after cookie_tcp_reqsk_alloc() has been called, but I prefer using the same construct than tcp_conn_request() diff --git a/include/net/tcp.h b/include/net/tcp.h index be712fb9ddd71b2b320356677f3aa1c6759e2698..9987b3fba9f202632916cc439af9d17f1e68bcd3 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -480,6 +480,7 @@ int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th, u32 cookie); struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb); struct request_sock *cookie_tcp_reqsk_alloc(const struct request_sock_ops *ops, + const struct tcp_request_sock_ops *af_ops, struct sock *sk, struct sk_buff *skb); #ifdef CONFIG_SYN_COOKIES diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c index 2cb3b852d14861231ac47f0b3e4daeb57682ffd2..f191bfa996d71f11ef786a0a3bcb8f737622d37a 100644 --- a/net/ipv4/syncookies.c +++ b/net/ipv4/syncookies.c @@ -281,6 +281,7 @@ bool cookie_ecn_ok(const struct tcp_options_received *tcp_opt, EXPORT_SYMBOL(cookie_ecn_ok); struct request_sock *cookie_tcp_reqsk_alloc(const struct request_sock_ops *ops, + const struct tcp_request_sock_ops *af_ops, struct sock *sk, struct sk_buff *skb) { @@ -297,6 +298,8 @@ struct request_sock *cookie_tcp_reqsk_alloc(const struct request_sock_ops *ops, return NULL; treq = tcp_rsk(req); + /* treq->af_specific might be used to perform TCP_MD5 lookup */ + treq->af_specific = af_ops; treq->syn_tos = TCP_SKB_CB(skb)->ip_dsfield; #if IS_ENABLED(CONFIG_MPTCP) treq->is_mptcp = sk_is_mptcp(sk); @@ -364,7 +367,8 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb) goto out; ret = NULL; - req = cookie_tcp_reqsk_alloc(&tcp_request_sock_ops, sk, skb); + req = cookie_tcp_reqsk_alloc(&tcp_request_sock_ops, + &tcp_request_sock_ipv4_ops, sk, skb); if (!req) goto out; diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c index d1b61d00368e1f58725e9997f74a0b144901277e..9cc123f000fbcfbeff7728bfee5339d6dd6470f9 100644 --- a/net/ipv6/syncookies.c +++ b/net/ipv6/syncookies.c @@ -170,7 +170,8 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb) goto out; ret = NULL; - req = cookie_tcp_reqsk_alloc(&tcp6_request_sock_ops, sk, skb); + req = cookie_tcp_reqsk_alloc(&tcp6_request_sock_ops, + &tcp_request_sock_ipv6_ops, sk, skb); if (!req) goto out;
On Sun, Apr 24, 2022 at 12:37 PM Eric Dumazet <edumazet@google.com> wrote: > > Wait a minute. > > Are you sure treq->af_specific is initialized at this point ? > > I should have tested this one liner patch really :/ > > I think that for syncookies, treq->af_specific is not initialized, > because we do not go through > tcp_conn_request() helper, but instead use cookie_tcp_reqsk_alloc() > > Before your patch treq->af_specific was only used during SYNACK > generation, which does not happen in syncookie more while receiving > the third packet. > > I will test something like this patch. We could move the init after > cookie_tcp_reqsk_alloc() has been called, but I prefer using the same > construct than tcp_conn_request() > Thanks for fixing this Eric. I had not considered syncookies. Francesco
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 6366df7aaf2a..6854bb1fb32b 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -531,7 +531,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->tsoffset = treq->ts_off; #ifdef CONFIG_TCP_MD5SIG newtp->md5sig_info = NULL; /*XXX*/ - if (newtp->af_specific->md5_lookup(sk, newsk)) + if (treq->af_specific->req_md5_lookup(sk, req_to_sk(req))) newtp->tcp_header_len += TCPOLEN_MD5SIG_ALIGNED; #endif if (skb->len >= TCP_MSS_DEFAULT + newtp->tcp_header_len)
In tcp_create_openreq_child we adjust tcp_header_len for md5 using the remote address in newsk. But that address is still 0 in newsk at this point, and it is only set later by the callers (tcp_v[46]_syn_recv_sock). Use the address from the request socket instead. v2: Added "Fixes:" line. Fixes: cfb6eeb4c860 ("[TCP]: MD5 Signature Option (RFC2385) support.") Signed-off-by: Francesco Ruggeri <fruggeri@arista.com> --- net/ipv4/tcp_minisocks.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)