diff mbox series

[bpf,v4] bpf: Support dual-stack sockets in bpf_tcp_check_syncookie

Message ID 20220329133001.2283509-1-maximmi@nvidia.com
State New
Headers show
Series [bpf,v4] bpf: Support dual-stack sockets in bpf_tcp_check_syncookie | expand

Commit Message

Maxim Mikityanskiy March 29, 2022, 1:30 p.m. UTC
bpf_tcp_gen_syncookie looks at the IP version in the IP header and
validates the address family of the socket. It supports IPv4 packets in
AF_INET6 dual-stack sockets.

On the other hand, bpf_tcp_check_syncookie looks only at the address
family of the socket, ignoring the real IP version in headers, and
validates only the packet size. This implementation has some drawbacks:

1. Packets are not validated properly, allowing a BPF program to trick
   bpf_tcp_check_syncookie into handling an IPv6 packet on an IPv4
   socket.

2. Dual-stack sockets fail the checks on IPv4 packets. IPv4 clients end
   up receiving a SYNACK with the cookie, but the following ACK gets
   dropped.

This patch fixes these issues by changing the checks in
bpf_tcp_check_syncookie to match the ones in bpf_tcp_gen_syncookie. IP
version from the header is taken into account, and it is validated
properly with address family.

Fixes: 399040847084 ("bpf: add helper to check for a valid SYN cookie")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
---
 net/core/filter.c                             | 17 +++-
 .../bpf/test_tcp_check_syncookie_user.c       | 78 ++++++++++++++-----
 2 files changed, 72 insertions(+), 23 deletions(-)

v2 changes: moved from bpf-next to bpf.

v3 changes: added a selftest.

v4 changes: none, CCed Jakub and Arthur from Cloudflare.

Comments

Maxim Mikityanskiy March 31, 2022, 1:58 p.m. UTC | #1
Jakub, Arthur, could you review and ACK the patch?

> -----Original Message-----
> From: Maxim Mikityanskiy <maximmi@nvidia.com>
> 
> bpf_tcp_gen_syncookie looks at the IP version in the IP header and
> validates the address family of the socket. It supports IPv4 packets in
> AF_INET6 dual-stack sockets.
> 
> On the other hand, bpf_tcp_check_syncookie looks only at the address
> family of the socket, ignoring the real IP version in headers, and
> validates only the packet size. This implementation has some drawbacks:
> 
> 1. Packets are not validated properly, allowing a BPF program to trick
>    bpf_tcp_check_syncookie into handling an IPv6 packet on an IPv4
>    socket.
> 
> 2. Dual-stack sockets fail the checks on IPv4 packets. IPv4 clients end
>    up receiving a SYNACK with the cookie, but the following ACK gets
>    dropped.
> 
> This patch fixes these issues by changing the checks in
> bpf_tcp_check_syncookie to match the ones in bpf_tcp_gen_syncookie. IP
> version from the header is taken into account, and it is validated
> properly with address family.
> 
> Fixes: 399040847084 ("bpf: add helper to check for a valid SYN cookie")
> Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
> Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
> ---
>  net/core/filter.c                             | 17 +++-
>  .../bpf/test_tcp_check_syncookie_user.c       | 78 ++++++++++++++-----
>  2 files changed, 72 insertions(+), 23 deletions(-)
> 
> v2 changes: moved from bpf-next to bpf.
> 
> v3 changes: added a selftest.
> 
> v4 changes: none, CCed Jakub and Arthur from Cloudflare.
> 
> diff --git a/net/core/filter.c b/net/core/filter.c
> index a7044e98765e..64470a727ef7 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -7016,24 +7016,33 @@ BPF_CALL_5(bpf_tcp_check_syncookie, struct sock *,
> sk, void *, iph, u32, iph_len
>  	if (!th->ack || th->rst || th->syn)
>  		return -ENOENT;
> 
> +	if (unlikely(iph_len < sizeof(struct iphdr)))
> +		return -EINVAL;
> +
>  	if (tcp_synq_no_recent_overflow(sk))
>  		return -ENOENT;
> 
>  	cookie = ntohl(th->ack_seq) - 1;
> 
> -	switch (sk->sk_family) {
> -	case AF_INET:
> -		if (unlikely(iph_len < sizeof(struct iphdr)))
> +	/* Both struct iphdr and struct ipv6hdr have the version field at the
> +	 * same offset so we can cast to the shorter header (struct iphdr).
> +	 */
> +	switch (((struct iphdr *)iph)->version) {
> +	case 4:
> +		if (sk->sk_family == AF_INET6 && ipv6_only_sock(sk))
>  			return -EINVAL;
> 
>  		ret = __cookie_v4_check((struct iphdr *)iph, th, cookie);
>  		break;
> 
>  #if IS_BUILTIN(CONFIG_IPV6)
> -	case AF_INET6:
> +	case 6:
>  		if (unlikely(iph_len < sizeof(struct ipv6hdr)))
>  			return -EINVAL;
> 
> +		if (sk->sk_family != AF_INET6)
> +			return -EINVAL;
> +
>  		ret = __cookie_v6_check((struct ipv6hdr *)iph, th, cookie);
>  		break;
>  #endif /* CONFIG_IPV6 */
> diff --git a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> index b9e991d43155..e7775d3bbe08 100644
> --- a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> +++ b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> @@ -18,8 +18,9 @@
>  #include "bpf_rlimit.h"
>  #include "cgroup_helpers.h"
> 
> -static int start_server(const struct sockaddr *addr, socklen_t len)
> +static int start_server(const struct sockaddr *addr, socklen_t len, bool
> dual)
>  {
> +	int mode = !dual;
>  	int fd;
> 
>  	fd = socket(addr->sa_family, SOCK_STREAM, 0);
> @@ -28,6 +29,14 @@ static int start_server(const struct sockaddr *addr,
> socklen_t len)
>  		goto out;
>  	}
> 
> +	if (addr->sa_family == AF_INET6) {
> +		if (setsockopt(fd, IPPROTO_IPV6, IPV6_V6ONLY, (char *)&mode,
> +			       sizeof(mode)) == -1) {
> +			log_err("Failed to set the dual-stack mode");
> +			goto close_out;
> +		}
> +	}
> +
>  	if (bind(fd, addr, len) == -1) {
>  		log_err("Failed to bind server socket");
>  		goto close_out;
> @@ -47,24 +56,17 @@ static int start_server(const struct sockaddr *addr,
> socklen_t len)
>  	return fd;
>  }
> 
> -static int connect_to_server(int server_fd)
> +static int connect_to_server(const struct sockaddr *addr, socklen_t len)
>  {
> -	struct sockaddr_storage addr;
> -	socklen_t len = sizeof(addr);
>  	int fd = -1;
> 
> -	if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
> -		log_err("Failed to get server addr");
> -		goto out;
> -	}
> -
> -	fd = socket(addr.ss_family, SOCK_STREAM, 0);
> +	fd = socket(addr->sa_family, SOCK_STREAM, 0);
>  	if (fd == -1) {
>  		log_err("Failed to create client socket");
>  		goto out;
>  	}
> 
> -	if (connect(fd, (const struct sockaddr *)&addr, len) == -1) {
> +	if (connect(fd, (const struct sockaddr *)addr, len) == -1) {
>  		log_err("Fail to connect to server");
>  		goto close_out;
>  	}
> @@ -116,7 +118,8 @@ static int get_map_fd_by_prog_id(int prog_id, bool *xdp)
>  	return map_fd;
>  }
> 
> -static int run_test(int server_fd, int results_fd, bool xdp)
> +static int run_test(int server_fd, int results_fd, bool xdp,
> +		    const struct sockaddr *addr, socklen_t len)
>  {
>  	int client = -1, srv_client = -1;
>  	int ret = 0;
> @@ -142,7 +145,7 @@ static int run_test(int server_fd, int results_fd, bool
> xdp)
>  		goto err;
>  	}
> 
> -	client = connect_to_server(server_fd);
> +	client = connect_to_server(addr, len);
>  	if (client == -1)
>  		goto err;
> 
> @@ -199,12 +202,30 @@ static int run_test(int server_fd, int results_fd,
> bool xdp)
>  	return ret;
>  }
> 
> +static bool get_port(int server_fd, in_port_t *port)
> +{
> +	struct sockaddr_in addr;
> +	socklen_t len = sizeof(addr);
> +
> +	if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
> +		log_err("Failed to get server addr");
> +		return false;
> +	}
> +
> +	/* sin_port and sin6_port are located at the same offset. */
> +	*port = addr.sin_port;
> +	return true;
> +}
> +
>  int main(int argc, char **argv)
>  {
>  	struct sockaddr_in addr4;
>  	struct sockaddr_in6 addr6;
> +	struct sockaddr_in addr4dual;
> +	struct sockaddr_in6 addr6dual;
>  	int server = -1;
>  	int server_v6 = -1;
> +	int server_dual = -1;
>  	int results = -1;
>  	int err = 0;
>  	bool xdp;
> @@ -224,25 +245,43 @@ int main(int argc, char **argv)
>  	addr4.sin_family = AF_INET;
>  	addr4.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
>  	addr4.sin_port = 0;
> +	memcpy(&addr4dual, &addr4, sizeof(addr4dual));
> 
>  	memset(&addr6, 0, sizeof(addr6));
>  	addr6.sin6_family = AF_INET6;
>  	addr6.sin6_addr = in6addr_loopback;
>  	addr6.sin6_port = 0;
> 
> -	server = start_server((const struct sockaddr *)&addr4, sizeof(addr4));
> -	if (server == -1)
> +	memset(&addr6dual, 0, sizeof(addr6dual));
> +	addr6dual.sin6_family = AF_INET6;
> +	addr6dual.sin6_addr = in6addr_any;
> +	addr6dual.sin6_port = 0;
> +
> +	server = start_server((const struct sockaddr *)&addr4, sizeof(addr4),
> +			      false);
> +	if (server == -1 || !get_port(server, &addr4.sin_port))
>  		goto err;
> 
>  	server_v6 = start_server((const struct sockaddr *)&addr6,
> -				 sizeof(addr6));
> -	if (server_v6 == -1)
> +				 sizeof(addr6), false);
> +	if (server_v6 == -1 || !get_port(server_v6, &addr6.sin6_port))
> +		goto err;
> +
> +	server_dual = start_server((const struct sockaddr *)&addr6dual,
> +				   sizeof(addr6dual), true);
> +	if (server_dual == -1 || !get_port(server_dual, &addr4dual.sin_port))
> +		goto err;
> +
> +	if (run_test(server, results, xdp,
> +		     (const struct sockaddr *)&addr4, sizeof(addr4)))
>  		goto err;
> 
> -	if (run_test(server, results, xdp))
> +	if (run_test(server_v6, results, xdp,
> +		     (const struct sockaddr *)&addr6, sizeof(addr6)))
>  		goto err;
> 
> -	if (run_test(server_v6, results, xdp))
> +	if (run_test(server_dual, results, xdp,
> +		     (const struct sockaddr *)&addr4dual, sizeof(addr4dual)))
>  		goto err;
> 
>  	printf("ok\n");
> @@ -252,6 +291,7 @@ int main(int argc, char **argv)
>  out:
>  	close(server);
>  	close(server_v6);
> +	close(server_dual);
>  	close(results);
>  	return err;
>  }
> --
> 2.30.2
Arthur Fabre April 3, 2022, 4:28 p.m. UTC | #2
On Tue, Mar 29, 2022 at 1:30 PM Maxim Mikityanskiy <maximmi@nvidia.com> wrote:
>
> bpf_tcp_gen_syncookie looks at the IP version in the IP header and
> validates the address family of the socket. It supports IPv4 packets in
> AF_INET6 dual-stack sockets.
>
> On the other hand, bpf_tcp_check_syncookie looks only at the address
> family of the socket, ignoring the real IP version in headers, and
> validates only the packet size. This implementation has some drawbacks:
>
> 1. Packets are not validated properly, allowing a BPF program to trick
>    bpf_tcp_check_syncookie into handling an IPv6 packet on an IPv4
>    socket.
>
> 2. Dual-stack sockets fail the checks on IPv4 packets. IPv4 clients end
>    up receiving a SYNACK with the cookie, but the following ACK gets
>    dropped.
>
> This patch fixes these issues by changing the checks in
> bpf_tcp_check_syncookie to match the ones in bpf_tcp_gen_syncookie. IP
> version from the header is taken into account, and it is validated
> properly with address family.
>
> Fixes: 399040847084 ("bpf: add helper to check for a valid SYN cookie")
> Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
> Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
> ---
>  net/core/filter.c                             | 17 +++-
>  .../bpf/test_tcp_check_syncookie_user.c       | 78 ++++++++++++++-----
>  2 files changed, 72 insertions(+), 23 deletions(-)
>
> v2 changes: moved from bpf-next to bpf.
>
> v3 changes: added a selftest.
>
> v4 changes: none, CCed Jakub and Arthur from Cloudflare.
>
> diff --git a/net/core/filter.c b/net/core/filter.c
> index a7044e98765e..64470a727ef7 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -7016,24 +7016,33 @@ BPF_CALL_5(bpf_tcp_check_syncookie, struct sock *, sk, void *, iph, u32, iph_len
>         if (!th->ack || th->rst || th->syn)
>                 return -ENOENT;
>
> +       if (unlikely(iph_len < sizeof(struct iphdr)))
> +               return -EINVAL;
> +
>         if (tcp_synq_no_recent_overflow(sk))
>                 return -ENOENT;
>
>         cookie = ntohl(th->ack_seq) - 1;
>
> -       switch (sk->sk_family) {
> -       case AF_INET:
> -               if (unlikely(iph_len < sizeof(struct iphdr)))
> +       /* Both struct iphdr and struct ipv6hdr have the version field at the
> +        * same offset so we can cast to the shorter header (struct iphdr).
> +        */
> +       switch (((struct iphdr *)iph)->version) {
> +       case 4:
> +               if (sk->sk_family == AF_INET6 && ipv6_only_sock(sk))
>                         return -EINVAL;
>
>                 ret = __cookie_v4_check((struct iphdr *)iph, th, cookie);
>                 break;
>
>  #if IS_BUILTIN(CONFIG_IPV6)
> -       case AF_INET6:
> +       case 6:
>                 if (unlikely(iph_len < sizeof(struct ipv6hdr)))
>                         return -EINVAL;
>
> +               if (sk->sk_family != AF_INET6)
> +                       return -EINVAL;
> +
>                 ret = __cookie_v6_check((struct ipv6hdr *)iph, th, cookie);
>                 break;
>  #endif /* CONFIG_IPV6 */
> diff --git a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> index b9e991d43155..e7775d3bbe08 100644
> --- a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> +++ b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
> @@ -18,8 +18,9 @@
>  #include "bpf_rlimit.h"
>  #include "cgroup_helpers.h"
>
> -static int start_server(const struct sockaddr *addr, socklen_t len)
> +static int start_server(const struct sockaddr *addr, socklen_t len, bool dual)
>  {
> +       int mode = !dual;
>         int fd;
>
>         fd = socket(addr->sa_family, SOCK_STREAM, 0);
> @@ -28,6 +29,14 @@ static int start_server(const struct sockaddr *addr, socklen_t len)
>                 goto out;
>         }
>
> +       if (addr->sa_family == AF_INET6) {
> +               if (setsockopt(fd, IPPROTO_IPV6, IPV6_V6ONLY, (char *)&mode,
> +                              sizeof(mode)) == -1) {
> +                       log_err("Failed to set the dual-stack mode");
> +                       goto close_out;
> +               }
> +       }
> +
>         if (bind(fd, addr, len) == -1) {
>                 log_err("Failed to bind server socket");
>                 goto close_out;
> @@ -47,24 +56,17 @@ static int start_server(const struct sockaddr *addr, socklen_t len)
>         return fd;
>  }
>
> -static int connect_to_server(int server_fd)
> +static int connect_to_server(const struct sockaddr *addr, socklen_t len)
>  {
> -       struct sockaddr_storage addr;
> -       socklen_t len = sizeof(addr);
>         int fd = -1;
>
> -       if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
> -               log_err("Failed to get server addr");
> -               goto out;
> -       }
> -
> -       fd = socket(addr.ss_family, SOCK_STREAM, 0);
> +       fd = socket(addr->sa_family, SOCK_STREAM, 0);
>         if (fd == -1) {
>                 log_err("Failed to create client socket");
>                 goto out;
>         }
>
> -       if (connect(fd, (const struct sockaddr *)&addr, len) == -1) {
> +       if (connect(fd, (const struct sockaddr *)addr, len) == -1) {
>                 log_err("Fail to connect to server");
>                 goto close_out;
>         }
> @@ -116,7 +118,8 @@ static int get_map_fd_by_prog_id(int prog_id, bool *xdp)
>         return map_fd;
>  }
>
> -static int run_test(int server_fd, int results_fd, bool xdp)
> +static int run_test(int server_fd, int results_fd, bool xdp,
> +                   const struct sockaddr *addr, socklen_t len)
>  {
>         int client = -1, srv_client = -1;
>         int ret = 0;
> @@ -142,7 +145,7 @@ static int run_test(int server_fd, int results_fd, bool xdp)
>                 goto err;
>         }
>
> -       client = connect_to_server(server_fd);
> +       client = connect_to_server(addr, len);
>         if (client == -1)
>                 goto err;
>
> @@ -199,12 +202,30 @@ static int run_test(int server_fd, int results_fd, bool xdp)
>         return ret;
>  }
>
> +static bool get_port(int server_fd, in_port_t *port)
> +{
> +       struct sockaddr_in addr;
> +       socklen_t len = sizeof(addr);
> +
> +       if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
> +               log_err("Failed to get server addr");
> +               return false;
> +       }
> +
> +       /* sin_port and sin6_port are located at the same offset. */
> +       *port = addr.sin_port;
> +       return true;
> +}
> +
>  int main(int argc, char **argv)
>  {
>         struct sockaddr_in addr4;
>         struct sockaddr_in6 addr6;
> +       struct sockaddr_in addr4dual;
> +       struct sockaddr_in6 addr6dual;
>         int server = -1;
>         int server_v6 = -1;
> +       int server_dual = -1;
>         int results = -1;
>         int err = 0;
>         bool xdp;
> @@ -224,25 +245,43 @@ int main(int argc, char **argv)
>         addr4.sin_family = AF_INET;
>         addr4.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
>         addr4.sin_port = 0;
> +       memcpy(&addr4dual, &addr4, sizeof(addr4dual));
>
>         memset(&addr6, 0, sizeof(addr6));
>         addr6.sin6_family = AF_INET6;
>         addr6.sin6_addr = in6addr_loopback;
>         addr6.sin6_port = 0;
>
> -       server = start_server((const struct sockaddr *)&addr4, sizeof(addr4));
> -       if (server == -1)
> +       memset(&addr6dual, 0, sizeof(addr6dual));
> +       addr6dual.sin6_family = AF_INET6;
> +       addr6dual.sin6_addr = in6addr_any;
> +       addr6dual.sin6_port = 0;
> +
> +       server = start_server((const struct sockaddr *)&addr4, sizeof(addr4),
> +                             false);
> +       if (server == -1 || !get_port(server, &addr4.sin_port))
>                 goto err;
>
>         server_v6 = start_server((const struct sockaddr *)&addr6,
> -                                sizeof(addr6));
> -       if (server_v6 == -1)
> +                                sizeof(addr6), false);
> +       if (server_v6 == -1 || !get_port(server_v6, &addr6.sin6_port))
> +               goto err;
> +
> +       server_dual = start_server((const struct sockaddr *)&addr6dual,
> +                                  sizeof(addr6dual), true);
> +       if (server_dual == -1 || !get_port(server_dual, &addr4dual.sin_port))
> +               goto err;
> +
> +       if (run_test(server, results, xdp,
> +                    (const struct sockaddr *)&addr4, sizeof(addr4)))
>                 goto err;
>
> -       if (run_test(server, results, xdp))
> +       if (run_test(server_v6, results, xdp,
> +                    (const struct sockaddr *)&addr6, sizeof(addr6)))
>                 goto err;
>
> -       if (run_test(server_v6, results, xdp))
> +       if (run_test(server_dual, results, xdp,
> +                    (const struct sockaddr *)&addr4dual, sizeof(addr4dual)))
>                 goto err;
>
>         printf("ok\n");
> @@ -252,6 +291,7 @@ int main(int argc, char **argv)
>  out:
>         close(server);
>         close(server_v6);
> +       close(server_dual);
>         close(results);
>         return err;
>  }
> --
> 2.30.2

Thanks for the fix and the test! Looks sane to me, and it passes the
tests we have internally (but we only test IPv4 :/).

Acked-by: Arthur Fabre <afabre@cloudflare.com>
Daniel Borkmann April 5, 2022, 10:18 p.m. UTC | #3
Hey Maxim,

On 3/29/22 3:30 PM, Maxim Mikityanskiy wrote:
> bpf_tcp_gen_syncookie looks at the IP version in the IP header and
> validates the address family of the socket. It supports IPv4 packets in
> AF_INET6 dual-stack sockets.
> 
> On the other hand, bpf_tcp_check_syncookie looks only at the address
> family of the socket, ignoring the real IP version in headers, and
> validates only the packet size. This implementation has some drawbacks:
> 
> 1. Packets are not validated properly, allowing a BPF program to trick
>     bpf_tcp_check_syncookie into handling an IPv6 packet on an IPv4
>     socket.
> 
> 2. Dual-stack sockets fail the checks on IPv4 packets. IPv4 clients end
>     up receiving a SYNACK with the cookie, but the following ACK gets
>     dropped.
> 
> This patch fixes these issues by changing the checks in
> bpf_tcp_check_syncookie to match the ones in bpf_tcp_gen_syncookie. IP
> version from the header is taken into account, and it is validated
> properly with address family.
> 
> Fixes: 399040847084 ("bpf: add helper to check for a valid SYN cookie")
> Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
> Reviewed-by: Tariq Toukan <tariqt@nvidia.com>

Could you do one final spin (while retaining Arthur's Ack) by splitting
off the selftest into a separate commit? Otherwise stable folks likely
won't be able to pick the fix up. Otherwise lgtm.

Thanks,
Daniel
diff mbox series

Patch

diff --git a/net/core/filter.c b/net/core/filter.c
index a7044e98765e..64470a727ef7 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -7016,24 +7016,33 @@  BPF_CALL_5(bpf_tcp_check_syncookie, struct sock *, sk, void *, iph, u32, iph_len
 	if (!th->ack || th->rst || th->syn)
 		return -ENOENT;
 
+	if (unlikely(iph_len < sizeof(struct iphdr)))
+		return -EINVAL;
+
 	if (tcp_synq_no_recent_overflow(sk))
 		return -ENOENT;
 
 	cookie = ntohl(th->ack_seq) - 1;
 
-	switch (sk->sk_family) {
-	case AF_INET:
-		if (unlikely(iph_len < sizeof(struct iphdr)))
+	/* Both struct iphdr and struct ipv6hdr have the version field at the
+	 * same offset so we can cast to the shorter header (struct iphdr).
+	 */
+	switch (((struct iphdr *)iph)->version) {
+	case 4:
+		if (sk->sk_family == AF_INET6 && ipv6_only_sock(sk))
 			return -EINVAL;
 
 		ret = __cookie_v4_check((struct iphdr *)iph, th, cookie);
 		break;
 
 #if IS_BUILTIN(CONFIG_IPV6)
-	case AF_INET6:
+	case 6:
 		if (unlikely(iph_len < sizeof(struct ipv6hdr)))
 			return -EINVAL;
 
+		if (sk->sk_family != AF_INET6)
+			return -EINVAL;
+
 		ret = __cookie_v6_check((struct ipv6hdr *)iph, th, cookie);
 		break;
 #endif /* CONFIG_IPV6 */
diff --git a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
index b9e991d43155..e7775d3bbe08 100644
--- a/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
+++ b/tools/testing/selftests/bpf/test_tcp_check_syncookie_user.c
@@ -18,8 +18,9 @@ 
 #include "bpf_rlimit.h"
 #include "cgroup_helpers.h"
 
-static int start_server(const struct sockaddr *addr, socklen_t len)
+static int start_server(const struct sockaddr *addr, socklen_t len, bool dual)
 {
+	int mode = !dual;
 	int fd;
 
 	fd = socket(addr->sa_family, SOCK_STREAM, 0);
@@ -28,6 +29,14 @@  static int start_server(const struct sockaddr *addr, socklen_t len)
 		goto out;
 	}
 
+	if (addr->sa_family == AF_INET6) {
+		if (setsockopt(fd, IPPROTO_IPV6, IPV6_V6ONLY, (char *)&mode,
+			       sizeof(mode)) == -1) {
+			log_err("Failed to set the dual-stack mode");
+			goto close_out;
+		}
+	}
+
 	if (bind(fd, addr, len) == -1) {
 		log_err("Failed to bind server socket");
 		goto close_out;
@@ -47,24 +56,17 @@  static int start_server(const struct sockaddr *addr, socklen_t len)
 	return fd;
 }
 
-static int connect_to_server(int server_fd)
+static int connect_to_server(const struct sockaddr *addr, socklen_t len)
 {
-	struct sockaddr_storage addr;
-	socklen_t len = sizeof(addr);
 	int fd = -1;
 
-	if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
-		log_err("Failed to get server addr");
-		goto out;
-	}
-
-	fd = socket(addr.ss_family, SOCK_STREAM, 0);
+	fd = socket(addr->sa_family, SOCK_STREAM, 0);
 	if (fd == -1) {
 		log_err("Failed to create client socket");
 		goto out;
 	}
 
-	if (connect(fd, (const struct sockaddr *)&addr, len) == -1) {
+	if (connect(fd, (const struct sockaddr *)addr, len) == -1) {
 		log_err("Fail to connect to server");
 		goto close_out;
 	}
@@ -116,7 +118,8 @@  static int get_map_fd_by_prog_id(int prog_id, bool *xdp)
 	return map_fd;
 }
 
-static int run_test(int server_fd, int results_fd, bool xdp)
+static int run_test(int server_fd, int results_fd, bool xdp,
+		    const struct sockaddr *addr, socklen_t len)
 {
 	int client = -1, srv_client = -1;
 	int ret = 0;
@@ -142,7 +145,7 @@  static int run_test(int server_fd, int results_fd, bool xdp)
 		goto err;
 	}
 
-	client = connect_to_server(server_fd);
+	client = connect_to_server(addr, len);
 	if (client == -1)
 		goto err;
 
@@ -199,12 +202,30 @@  static int run_test(int server_fd, int results_fd, bool xdp)
 	return ret;
 }
 
+static bool get_port(int server_fd, in_port_t *port)
+{
+	struct sockaddr_in addr;
+	socklen_t len = sizeof(addr);
+
+	if (getsockname(server_fd, (struct sockaddr *)&addr, &len)) {
+		log_err("Failed to get server addr");
+		return false;
+	}
+
+	/* sin_port and sin6_port are located at the same offset. */
+	*port = addr.sin_port;
+	return true;
+}
+
 int main(int argc, char **argv)
 {
 	struct sockaddr_in addr4;
 	struct sockaddr_in6 addr6;
+	struct sockaddr_in addr4dual;
+	struct sockaddr_in6 addr6dual;
 	int server = -1;
 	int server_v6 = -1;
+	int server_dual = -1;
 	int results = -1;
 	int err = 0;
 	bool xdp;
@@ -224,25 +245,43 @@  int main(int argc, char **argv)
 	addr4.sin_family = AF_INET;
 	addr4.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
 	addr4.sin_port = 0;
+	memcpy(&addr4dual, &addr4, sizeof(addr4dual));
 
 	memset(&addr6, 0, sizeof(addr6));
 	addr6.sin6_family = AF_INET6;
 	addr6.sin6_addr = in6addr_loopback;
 	addr6.sin6_port = 0;
 
-	server = start_server((const struct sockaddr *)&addr4, sizeof(addr4));
-	if (server == -1)
+	memset(&addr6dual, 0, sizeof(addr6dual));
+	addr6dual.sin6_family = AF_INET6;
+	addr6dual.sin6_addr = in6addr_any;
+	addr6dual.sin6_port = 0;
+
+	server = start_server((const struct sockaddr *)&addr4, sizeof(addr4),
+			      false);
+	if (server == -1 || !get_port(server, &addr4.sin_port))
 		goto err;
 
 	server_v6 = start_server((const struct sockaddr *)&addr6,
-				 sizeof(addr6));
-	if (server_v6 == -1)
+				 sizeof(addr6), false);
+	if (server_v6 == -1 || !get_port(server_v6, &addr6.sin6_port))
+		goto err;
+
+	server_dual = start_server((const struct sockaddr *)&addr6dual,
+				   sizeof(addr6dual), true);
+	if (server_dual == -1 || !get_port(server_dual, &addr4dual.sin_port))
+		goto err;
+
+	if (run_test(server, results, xdp,
+		     (const struct sockaddr *)&addr4, sizeof(addr4)))
 		goto err;
 
-	if (run_test(server, results, xdp))
+	if (run_test(server_v6, results, xdp,
+		     (const struct sockaddr *)&addr6, sizeof(addr6)))
 		goto err;
 
-	if (run_test(server_v6, results, xdp))
+	if (run_test(server_dual, results, xdp,
+		     (const struct sockaddr *)&addr4dual, sizeof(addr4dual)))
 		goto err;
 
 	printf("ok\n");
@@ -252,6 +291,7 @@  int main(int argc, char **argv)
 out:
 	close(server);
 	close(server_v6);
+	close(server_dual);
 	close(results);
 	return err;
 }