* [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
@ 2023-09-20 10:20 Jakub Sitnicki
2023-09-20 18:19 ` Kui-Feng Lee
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Jakub Sitnicki @ 2023-09-20 10:20 UTC (permalink / raw)
To: bpf
Cc: netdev, kernel-team, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, John Fastabend, Cong Wang
With a SOCKMAP/SOCKHASH map and an sk_msg program user can steer messages
sent from one TCP socket (s1) to actually egress from another TCP
socket (s2):
tcp_bpf_sendmsg(s1) // = sk_prot->sendmsg
tcp_bpf_send_verdict(s1) // __SK_REDIRECT case
tcp_bpf_sendmsg_redir(s2)
tcp_bpf_push_locked(s2)
tcp_bpf_push(s2)
tcp_rate_check_app_limited(s2) // expects tcp_sock
tcp_sendmsg_locked(s2) // ditto
There is a hard-coded assumption in the call-chain, that the egress
socket (s2) is a TCP socket.
However in commit 122e6c79efe1 ("sock_map: Update sock type checks for
UDP") we have enabled redirects to non-TCP sockets. This was done for the
sake of BPF sk_skb programs. There was no indention to support sk_msg
send-to-egress use case.
As a result, attempts to send-to-egress through a non-TCP socket lead to a
crash due to invalid downcast from sock to tcp_sock:
BUG: kernel NULL pointer dereference, address: 000000000000002f
...
Call Trace:
<TASK>
? show_regs+0x60/0x70
? __die+0x1f/0x70
? page_fault_oops+0x80/0x160
? do_user_addr_fault+0x2d7/0x800
? rcu_is_watching+0x11/0x50
? exc_page_fault+0x70/0x1c0
? asm_exc_page_fault+0x27/0x30
? tcp_tso_segs+0x14/0xa0
tcp_write_xmit+0x67/0xce0
__tcp_push_pending_frames+0x32/0xf0
tcp_push+0x107/0x140
tcp_sendmsg_locked+0x99f/0xbb0
tcp_bpf_push+0x19d/0x3a0
tcp_bpf_sendmsg_redir+0x55/0xd0
tcp_bpf_send_verdict+0x407/0x550
tcp_bpf_sendmsg+0x1a1/0x390
inet_sendmsg+0x6a/0x70
sock_sendmsg+0x9d/0xc0
? sockfd_lookup_light+0x12/0x80
__sys_sendto+0x10e/0x160
? syscall_enter_from_user_mode+0x20/0x60
? __this_cpu_preempt_check+0x13/0x20
? lockdep_hardirqs_on+0x82/0x110
__x64_sys_sendto+0x1f/0x30
do_syscall_64+0x38/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd
Reject selecting a non-TCP sockets as redirect target from a BPF sk_msg
program to prevent the crash. When attempted, user will receive an EACCES
error from send/sendto/sendmsg() syscall.
Fixes: 122e6c79efe1 ("sock_map: Update sock type checks for UDP")
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
---
FYI, I'm working on revamping the sockmap_listen selftest, which exercises
some of redirect combinations, to cover the whole combination matrix so
that we can catch these kinds of problems early on.
net/core/sock_map.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/net/core/sock_map.c b/net/core/sock_map.c
index cb11750b1df5..4292c2ed1828 100644
--- a/net/core/sock_map.c
+++ b/net/core/sock_map.c
@@ -668,6 +668,8 @@ BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
sk = __sock_map_lookup_elem(map, key);
if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
return SK_DROP;
+ if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
+ return SK_DROP;
msg->flags = flags;
msg->sk_redir = sk;
@@ -1267,6 +1269,8 @@ BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
sk = __sock_hash_lookup_elem(map, key);
if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
return SK_DROP;
+ if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
+ return SK_DROP;
msg->flags = flags;
msg->sk_redir = sk;
--
2.41.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
2023-09-20 10:20 [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets Jakub Sitnicki
@ 2023-09-20 18:19 ` Kui-Feng Lee
2023-09-20 20:59 ` Jakub Sitnicki
2023-09-25 18:27 ` John Fastabend
2023-09-29 15:20 ` patchwork-bot+netdevbpf
2 siblings, 1 reply; 6+ messages in thread
From: Kui-Feng Lee @ 2023-09-20 18:19 UTC (permalink / raw)
To: Jakub Sitnicki, bpf
Cc: netdev, kernel-team, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, John Fastabend, Cong Wang
On 9/20/23 03:20, Jakub Sitnicki wrote:
>
> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
> index cb11750b1df5..4292c2ed1828 100644
> --- a/net/core/sock_map.c
> +++ b/net/core/sock_map.c
> @@ -668,6 +668,8 @@ BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
> sk = __sock_map_lookup_elem(map, key);
> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
> return SK_DROP;
> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
> + return SK_DROP;
>
> msg->flags = flags;
> msg->sk_redir = sk;
> @@ -1267,6 +1269,8 @@ BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
> sk = __sock_hash_lookup_elem(map, key);
> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
> return SK_DROP;
> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
> + return SK_DROP;
>
> msg->flags = flags;
> msg->sk_redir = sk;
Just be curious! Can it happen to other socket types?
I mean to redirect a msg from a sk of any type to one of another type.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
2023-09-20 18:19 ` Kui-Feng Lee
@ 2023-09-20 20:59 ` Jakub Sitnicki
2023-09-20 21:11 ` Kui-Feng Lee
0 siblings, 1 reply; 6+ messages in thread
From: Jakub Sitnicki @ 2023-09-20 20:59 UTC (permalink / raw)
To: Kui-Feng Lee
Cc: bpf, netdev, kernel-team, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, John Fastabend, Cong Wang
On Wed, Sep 20, 2023 at 11:19 AM -07, Kui-Feng Lee wrote:
> On 9/20/23 03:20, Jakub Sitnicki wrote:
>> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
>> index cb11750b1df5..4292c2ed1828 100644
>> --- a/net/core/sock_map.c
>> +++ b/net/core/sock_map.c
>> @@ -668,6 +668,8 @@ BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
>> sk = __sock_map_lookup_elem(map, key);
>> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
>> return SK_DROP;
>> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
>> + return SK_DROP;
>> msg->flags = flags;
>> msg->sk_redir = sk;
>> @@ -1267,6 +1269,8 @@ BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
>> sk = __sock_hash_lookup_elem(map, key);
>> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
>> return SK_DROP;
>> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
>> + return SK_DROP;
>> msg->flags = flags;
>> msg->sk_redir = sk;
>
> Just be curious! Can it happen to other socket types?
> I mean to redirect a msg from a sk of any type to one of another type.
Today sk_msg redirects are implemented only for tcp4 and tcp6.
Here's a full matrix of what redirects are supported [1].
[1] https://gist.github.com/jsitnicki/578fdd614d181bed2b02922b17972b4e
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
2023-09-20 20:59 ` Jakub Sitnicki
@ 2023-09-20 21:11 ` Kui-Feng Lee
0 siblings, 0 replies; 6+ messages in thread
From: Kui-Feng Lee @ 2023-09-20 21:11 UTC (permalink / raw)
To: Jakub Sitnicki
Cc: bpf, netdev, kernel-team, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, John Fastabend, Cong Wang
On 9/20/23 13:59, Jakub Sitnicki wrote:
> On Wed, Sep 20, 2023 at 11:19 AM -07, Kui-Feng Lee wrote:
>> On 9/20/23 03:20, Jakub Sitnicki wrote:
>>> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
>>> index cb11750b1df5..4292c2ed1828 100644
>>> --- a/net/core/sock_map.c
>>> +++ b/net/core/sock_map.c
>>> @@ -668,6 +668,8 @@ BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
>>> sk = __sock_map_lookup_elem(map, key);
>>> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
>>> return SK_DROP;
>>> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
>>> + return SK_DROP;
>>> msg->flags = flags;
>>> msg->sk_redir = sk;
>>> @@ -1267,6 +1269,8 @@ BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
>>> sk = __sock_hash_lookup_elem(map, key);
>>> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
>>> return SK_DROP;
>>> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
>>> + return SK_DROP;
>>> msg->flags = flags;
>>> msg->sk_redir = sk;
>>
>> Just be curious! Can it happen to other socket types?
>> I mean to redirect a msg from a sk of any type to one of another type.
>
> Today sk_msg redirects are implemented only for tcp4 and tcp6.
>
> Here's a full matrix of what redirects are supported [1].
>
> [1] https://gist.github.com/jsitnicki/578fdd614d181bed2b02922b17972b4e
Thanks!
^ permalink raw reply [flat|nested] 6+ messages in thread
* RE: [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
2023-09-20 10:20 [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets Jakub Sitnicki
2023-09-20 18:19 ` Kui-Feng Lee
@ 2023-09-25 18:27 ` John Fastabend
2023-09-29 15:20 ` patchwork-bot+netdevbpf
2 siblings, 0 replies; 6+ messages in thread
From: John Fastabend @ 2023-09-25 18:27 UTC (permalink / raw)
To: Jakub Sitnicki, bpf
Cc: netdev, kernel-team, Alexei Starovoitov, Daniel Borkmann,
Andrii Nakryiko, John Fastabend, Cong Wang
Jakub Sitnicki wrote:
> With a SOCKMAP/SOCKHASH map and an sk_msg program user can steer messages
> sent from one TCP socket (s1) to actually egress from another TCP
> socket (s2):
>
> tcp_bpf_sendmsg(s1) // = sk_prot->sendmsg
> tcp_bpf_send_verdict(s1) // __SK_REDIRECT case
> tcp_bpf_sendmsg_redir(s2)
> tcp_bpf_push_locked(s2)
> tcp_bpf_push(s2)
> tcp_rate_check_app_limited(s2) // expects tcp_sock
> tcp_sendmsg_locked(s2) // ditto
>
> There is a hard-coded assumption in the call-chain, that the egress
> socket (s2) is a TCP socket.
>
> However in commit 122e6c79efe1 ("sock_map: Update sock type checks for
> UDP") we have enabled redirects to non-TCP sockets. This was done for the
> sake of BPF sk_skb programs. There was no indention to support sk_msg
> send-to-egress use case.
>
> As a result, attempts to send-to-egress through a non-TCP socket lead to a
> crash due to invalid downcast from sock to tcp_sock:
>
> BUG: kernel NULL pointer dereference, address: 000000000000002f
> ...
> Call Trace:
> <TASK>
> ? show_regs+0x60/0x70
> ? __die+0x1f/0x70
> ? page_fault_oops+0x80/0x160
> ? do_user_addr_fault+0x2d7/0x800
> ? rcu_is_watching+0x11/0x50
> ? exc_page_fault+0x70/0x1c0
> ? asm_exc_page_fault+0x27/0x30
> ? tcp_tso_segs+0x14/0xa0
> tcp_write_xmit+0x67/0xce0
> __tcp_push_pending_frames+0x32/0xf0
> tcp_push+0x107/0x140
> tcp_sendmsg_locked+0x99f/0xbb0
> tcp_bpf_push+0x19d/0x3a0
> tcp_bpf_sendmsg_redir+0x55/0xd0
> tcp_bpf_send_verdict+0x407/0x550
> tcp_bpf_sendmsg+0x1a1/0x390
> inet_sendmsg+0x6a/0x70
> sock_sendmsg+0x9d/0xc0
> ? sockfd_lookup_light+0x12/0x80
> __sys_sendto+0x10e/0x160
> ? syscall_enter_from_user_mode+0x20/0x60
> ? __this_cpu_preempt_check+0x13/0x20
> ? lockdep_hardirqs_on+0x82/0x110
> __x64_sys_sendto+0x1f/0x30
> do_syscall_64+0x38/0x90
> entry_SYSCALL_64_after_hwframe+0x63/0xcd
>
> Reject selecting a non-TCP sockets as redirect target from a BPF sk_msg
> program to prevent the crash. When attempted, user will receive an EACCES
> error from send/sendto/sendmsg() syscall.
>
> Fixes: 122e6c79efe1 ("sock_map: Update sock type checks for UDP")
> Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
> ---
> FYI, I'm working on revamping the sockmap_listen selftest, which exercises
> some of redirect combinations, to cover the whole combination matrix so
> that we can catch these kinds of problems early on.
Yes this would be appreciated.
>
> net/core/sock_map.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
> index cb11750b1df5..4292c2ed1828 100644
> --- a/net/core/sock_map.c
> +++ b/net/core/sock_map.c
> @@ -668,6 +668,8 @@ BPF_CALL_4(bpf_msg_redirect_map, struct sk_msg *, msg,
> sk = __sock_map_lookup_elem(map, key);
> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
> return SK_DROP;
> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
> + return SK_DROP;
>
> msg->flags = flags;
> msg->sk_redir = sk;
> @@ -1267,6 +1269,8 @@ BPF_CALL_4(bpf_msg_redirect_hash, struct sk_msg *, msg,
> sk = __sock_hash_lookup_elem(map, key);
> if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
> return SK_DROP;
> + if (!(flags & BPF_F_INGRESS) && !sk_is_tcp(sk))
> + return SK_DROP;
As a stop gap I think this is fine. If anyone wants to add support though
I do think as a use case it would make sense to redirect TCP into an
AF_UNIX socket and vice versa.
Acked-by: John Fastabend <john.fastabend@gmail.com>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
2023-09-20 10:20 [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets Jakub Sitnicki
2023-09-20 18:19 ` Kui-Feng Lee
2023-09-25 18:27 ` John Fastabend
@ 2023-09-29 15:20 ` patchwork-bot+netdevbpf
2 siblings, 0 replies; 6+ messages in thread
From: patchwork-bot+netdevbpf @ 2023-09-29 15:20 UTC (permalink / raw)
To: Jakub Sitnicki
Cc: bpf, netdev, kernel-team, ast, daniel, andrii, john.fastabend, cong.wang
Hello:
This patch was applied to bpf/bpf.git (master)
by Daniel Borkmann <daniel@iogearbox.net>:
On Wed, 20 Sep 2023 12:20:55 +0200 you wrote:
> With a SOCKMAP/SOCKHASH map and an sk_msg program user can steer messages
> sent from one TCP socket (s1) to actually egress from another TCP
> socket (s2):
>
> tcp_bpf_sendmsg(s1) // = sk_prot->sendmsg
> tcp_bpf_send_verdict(s1) // __SK_REDIRECT case
> tcp_bpf_sendmsg_redir(s2)
> tcp_bpf_push_locked(s2)
> tcp_bpf_push(s2)
> tcp_rate_check_app_limited(s2) // expects tcp_sock
> tcp_sendmsg_locked(s2) // ditto
>
> [...]
Here is the summary with links:
- [bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets
https://git.kernel.org/bpf/bpf/c/b80e31baa436
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2023-09-29 15:20 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-09-20 10:20 [PATCH bpf] bpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets Jakub Sitnicki
2023-09-20 18:19 ` Kui-Feng Lee
2023-09-20 20:59 ` Jakub Sitnicki
2023-09-20 21:11 ` Kui-Feng Lee
2023-09-25 18:27 ` John Fastabend
2023-09-29 15:20 ` patchwork-bot+netdevbpf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).