sk->sk_socket will be assigned in case of marginal crossed SYN, but also
in other cases, e.g.

 - With TCP Fast Open, if the connection got accept()'ed before
   receiving the 3rd ACK ;

 - With MPTCP, when accepting additional subflows to an existing MPTCP
   connection.

In these cases, the switch to TCP_ESTABLISHED is done when receiving the
3rd ACK, without the SYN flag then.

To properly restrict the wake-up to crossed SYN cases, it is then
required to also limit the check to packets containing the SYN-ACK
flags.

While at it, also update the attached comment: sk->sk_sleep has been
removed in 2010, and replaced by sk->sk_wq in commit 43815482370c ("net:
sock_def_readable() and friends RCU conversion").

Fixes: 168a8f58059a ("tcp: TCP Fast Open Server - main code path")
Suggested-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
Notes:
 - The above 'Fixes' tag should correspond to the commit introducing the
   possibility to have sk->sk_socket being set there in other cases than
   the crossed SYN one. But I might have missed other cases. Maybe
   1da177e4c3f4 ("Linux-2.6.12-rc2") might be safer? On the other hand,
   I don't think this wake-up was causing any visible issue, apart from
   not being needed.
---
 net/ipv4/tcp_input.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index bfe1bc69dc3e..5cebb389bf71 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6797,9 +6797,9 @@ tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb)
 
 		/* Note, that this wakeup is only for marginal crossed SYN case.
 		 * Passively open sockets are not waked up, because
-		 * sk->sk_sleep == NULL and sk->sk_socket == NULL.
+		 * sk->sk_wq == NULL and sk->sk_socket == NULL.
 		 */
-		if (sk->sk_socket)
+		if (sk->sk_socket && th->syn)
 			sk_wake_async(sk, SOCK_WAKE_IO, POLL_OUT);
 
 		tp->snd_una = TCP_SKB_CB(skb)->ack_seq;

-- 
2.45.2

On Thu, Jul 18, 2024 at 12:34 PM Matthieu Baerts (NGI0)
<matttbe@kernel.org> wrote:
>
> sk->sk_socket will be assigned in case of marginal crossed SYN, but also
> in other cases, e.g.
>
>  - With TCP Fast Open, if the connection got accept()'ed before
>    receiving the 3rd ACK ;
>
>  - With MPTCP, when accepting additional subflows to an existing MPTCP
>    connection.
>
> In these cases, the switch to TCP_ESTABLISHED is done when receiving the
> 3rd ACK, without the SYN flag then.
>
> To properly restrict the wake-up to crossed SYN cases, it is then
> required to also limit the check to packets containing the SYN-ACK
> flags.
>
> While at it, also update the attached comment: sk->sk_sleep has been
> removed in 2010, and replaced by sk->sk_wq in commit 43815482370c ("net:
> sock_def_readable() and friends RCU conversion").
>
> Fixes: 168a8f58059a ("tcp: TCP Fast Open Server - main code path")
> Suggested-by: Kuniyuki Iwashima <kuniyu@amazon.com>
> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
> ---
> Notes:
>  - The above 'Fixes' tag should correspond to the commit introducing the
>    possibility to have sk->sk_socket being set there in other cases than
>    the crossed SYN one. But I might have missed other cases. Maybe
>    1da177e4c3f4 ("Linux-2.6.12-rc2") might be safer? On the other hand,
>    I don't think this wake-up was causing any visible issue, apart from
>    not being needed.

This seems a net-next candidate to me ?

Hi Eric,

Thank you for the review!

On 23/07/2024 16:32, Eric Dumazet wrote:
> On Thu, Jul 18, 2024 at 12:34 PM Matthieu Baerts (NGI0)
> <matttbe@kernel.org> wrote:
>>
>> sk->sk_socket will be assigned in case of marginal crossed SYN, but also
>> in other cases, e.g.
>>
>>  - With TCP Fast Open, if the connection got accept()'ed before
>>    receiving the 3rd ACK ;
>>
>>  - With MPTCP, when accepting additional subflows to an existing MPTCP
>>    connection.
>>
>> In these cases, the switch to TCP_ESTABLISHED is done when receiving the
>> 3rd ACK, without the SYN flag then.
>>
>> To properly restrict the wake-up to crossed SYN cases, it is then
>> required to also limit the check to packets containing the SYN-ACK
>> flags.
>>
>> While at it, also update the attached comment: sk->sk_sleep has been
>> removed in 2010, and replaced by sk->sk_wq in commit 43815482370c ("net:
>> sock_def_readable() and friends RCU conversion").
>>
>> Fixes: 168a8f58059a ("tcp: TCP Fast Open Server - main code path")
>> Suggested-by: Kuniyuki Iwashima <kuniyu@amazon.com>
>> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
>> ---
>> Notes:
>>  - The above 'Fixes' tag should correspond to the commit introducing the
>>    possibility to have sk->sk_socket being set there in other cases than
>>    the crossed SYN one. But I might have missed other cases. Maybe
>>    1da177e4c3f4 ("Linux-2.6.12-rc2") might be safer? On the other hand,
>>    I don't think this wake-up was causing any visible issue, apart from
>>    not being needed.
> 
> This seems a net-next candidate to me ?

Fine by me!

I modified this line mainly because Kuniyuki mentioned that it was the
same check as the new one, modified in patch 1/2. I didn't find any
visible issue with the wakeup, so I guess it can go to net-next.

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.