bpf: Allow bpf_tcp_sock() helper in XDP

[PATCH bpf-next] bpf: Allow bpf_tcp_sock() helper in XDP

Posted by Arthur Fabre 4 days, 15 hours ago

We can already lookup sockets in XDP via the bpf_sk(c)_lookup_(udp|tcp)
helpers. This can be used to determine if a received packet is for an
established socket or not, for example in a load balancer.

But the TCP TIME-WAIT state needs special handling for this: packets
received for a TIME-WAIT socket can either belong to the current
incarnation of the connection, or a new incarnation. The only way to
determine this is to compare the sequence numbers.

Add the bpf_tcp_sock() helper in XDP to allow the socket sequence
numbers to be read.

---
Signed-off-by: Arthur Fabre <afabre@cloudflare.com>
---
 net/core/filter.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/core/filter.c b/net/core/filter.c
index f04982d79d72e92f1b46ccd87633391e38c3da81..5e3a1951130369f070c3a373ea649bd10db412b6 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -8506,6 +8506,8 @@ xdp_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_sk_release_proto;
 	case BPF_FUNC_skc_lookup_tcp:
 		return &bpf_xdp_skc_lookup_tcp_proto;
+	case BPF_FUNC_tcp_sock:
+		return &bpf_tcp_sock_proto;
 	case BPF_FUNC_tcp_check_syncookie:
 		return &bpf_tcp_check_syncookie_proto;
 	case BPF_FUNC_tcp_gen_syncookie:

---
base-commit: d95d76aa772bf94df353b015b1cb38303d4a415d
change-id: 20260204-afabre-bpf_tcp_sock-117d03c6f0e3

Best regards,
-- 
Arthur Fabre <afabre@cloudflare.com>

Re: [PATCH bpf-next] bpf: Allow bpf_tcp_sock() helper in XDP

Posted by bot+bpf-ci@kernel.org 4 days, 15 hours ago

>     But the TCP TIME-WAIT state needs special handling for this: packets
>     received for a TIME-WAIT socket can either belong to the current
>     incarnation of the connection, or a new incarnation. The only way to
>     determine this is to compare the sequence numbers.
>
>     Add the bpf_tcp_sock() helper in XDP to allow the socket sequence
>     numbers to be read.

Can this approach work for TIME-WAIT sockets? Looking at bpf_tcp_sock() in
net/core/filter.c:

    if (sk_fullsock(sk) && sk->sk_protocol == IPPROTO_TCP)
        return (unsigned long)sk;
    return (unsigned long)NULL;

And sk_fullsock() in include/net/sock.h:

    static inline bool sk_fullsock(const struct sock *sk)
    {
        return (1 << sk->sk_state) & ~(TCPF_TIME_WAIT | TCPF_NEW_SYN_RECV);
    }

TIME-WAIT sockets are explicitly excluded by sk_fullsock(), so bpf_tcp_sock()
returns NULL for them. The sequence numbers stored in TIME-WAIT sockets
(skc_tw_rcv_nxt and skc_tw_snd_nxt in sock_common) would not be accessible
through this helper.

The change itself is useful for accessing tcp_sock fields from established
connections in XDP, but the commit message implies this solves the TIME-WAIT
sequence number problem described, which doesn't appear to be the case.


---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/21671977729

AI-authorship-score: low
AI-authorship-explanation: The commit message style is consistent with human-written kernel patches, with natural technical language and standard formatting; the technical error is more likely human oversight than AI hallucination.
issues-found: 1
issue-severity-score: low
issue-severity-explanation: The issue is a misleading commit message that claims the change enables TIME-WAIT socket sequence number reading, but the bpf_tcp_sock() helper explicitly returns NULL for TIME-WAIT sockets.

Re: [PATCH bpf-next] bpf: Allow bpf_tcp_sock() helper in XDP

Posted by Martin KaFai Lau 4 days, 5 hours ago


On 2/4/26 5:04 AM, bot+bpf-ci@kernel.org wrote:
>>      But the TCP TIME-WAIT state needs special handling for this: packets
>>      received for a TIME-WAIT socket can either belong to the current
>>      incarnation of the connection, or a new incarnation. The only way to
>>      determine this is to compare the sequence numbers.
>>
>>      Add the bpf_tcp_sock() helper in XDP to allow the socket sequence
>>      numbers to be read.
> 
> Can this approach work for TIME-WAIT sockets? Looking at bpf_tcp_sock() in
> net/core/filter.c:
> 
>      if (sk_fullsock(sk) && sk->sk_protocol == IPPROTO_TCP)
>          return (unsigned long)sk;
>      return (unsigned long)NULL;
> 
> And sk_fullsock() in include/net/sock.h:
> 
>      static inline bool sk_fullsock(const struct sock *sk)
>      {
>          return (1 << sk->sk_state) & ~(TCPF_TIME_WAIT | TCPF_NEW_SYN_RECV);
>      }
> 
> TIME-WAIT sockets are explicitly excluded by sk_fullsock(), so bpf_tcp_sock()
> returns NULL for them. The sequence numbers stored in TIME-WAIT sockets
> (skc_tw_rcv_nxt and skc_tw_snd_nxt in sock_common) would not be accessible
> through this helper.
> 
> The change itself is useful for accessing tcp_sock fields from established
> connections in XDP, but the commit message implies this solves the TIME-WAIT
> sequence number problem described, which doesn't appear to be the case.

or use "tp = bpf_core_cast(sk, struct tcp_sock /* or tcp_timewait_sock 
*/);". The bpf prog can also read all members of the kernel struct 
instead of a subset defined in bpf_tcp_sock.

pw-bot: cr

> 
> 
> ---
> AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
> See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
> 
> CI run summary: https://github.com/kernel-patches/bpf/actions/runs/21671977729
> 
> AI-authorship-score: low
> AI-authorship-explanation: The commit message style is consistent with human-written kernel patches, with natural technical language and standard formatting; the technical error is more likely human oversight than AI hallucination.
> issues-found: 1
> issue-severity-score: low
> issue-severity-explanation: The issue is a misleading commit message that claims the change enables TIME-WAIT socket sequence number reading, but the bpf_tcp_sock() helper explicitly returns NULL for TIME-WAIT sockets.

Re: [PATCH bpf-next] bpf: Allow bpf_tcp_sock() helper in XDP

Posted by Arthur Fabre 3 days, 12 hours ago

On Wed Feb 4, 2026 at 11:27 PM CET, Martin KaFai Lau wrote:
>
>
> On 2/4/26 5:04 AM, bot+bpf-ci@kernel.org wrote:
> >>      But the TCP TIME-WAIT state needs special handling for this: packets
> >>      received for a TIME-WAIT socket can either belong to the current
> >>      incarnation of the connection, or a new incarnation. The only way to
> >>      determine this is to compare the sequence numbers.
> >>
> >>      Add the bpf_tcp_sock() helper in XDP to allow the socket sequence
> >>      numbers to be read.
> > 
> > Can this approach work for TIME-WAIT sockets? Looking at bpf_tcp_sock() in
> > net/core/filter.c:
> > 
> >      if (sk_fullsock(sk) && sk->sk_protocol == IPPROTO_TCP)
> >          return (unsigned long)sk;
> >      return (unsigned long)NULL;
> > 
> > And sk_fullsock() in include/net/sock.h:
> > 
> >      static inline bool sk_fullsock(const struct sock *sk)
> >      {
> >          return (1 << sk->sk_state) & ~(TCPF_TIME_WAIT | TCPF_NEW_SYN_RECV);
> >      }
> > 
> > TIME-WAIT sockets are explicitly excluded by sk_fullsock(), so bpf_tcp_sock()
> > returns NULL for them. The sequence numbers stored in TIME-WAIT sockets
> > (skc_tw_rcv_nxt and skc_tw_snd_nxt in sock_common) would not be accessible
> > through this helper.
> > 
> > The change itself is useful for accessing tcp_sock fields from established
> > connections in XDP, but the commit message implies this solves the TIME-WAIT
> > sequence number problem described, which doesn't appear to be the case.
>
> or use "tp = bpf_core_cast(sk, struct tcp_sock /* or tcp_timewait_sock 
> */);". The bpf prog can also read all members of the kernel struct 
> instead of a subset defined in bpf_tcp_sock.

Brilliant, that works! Thanks Martin.

>
> pw-bot: cr
>
> > 
> > 
> > ---
> > AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
> > See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
> > 
> > CI run summary: https://github.com/kernel-patches/bpf/actions/runs/21671977729
> > 
> > AI-authorship-score: low
> > AI-authorship-explanation: The commit message style is consistent with human-written kernel patches, with natural technical language and standard formatting; the technical error is more likely human oversight than AI hallucination.
> > issues-found: 1
> > issue-severity-score: low
> > issue-severity-explanation: The issue is a misleading commit message that claims the change enables TIME-WAIT socket sequence number reading, but the bpf_tcp_sock() helper explicitly returns NULL for TIME-WAIT sockets.