Re: [PATCH net-next v3 2/6] mptcp: keep rcv_mwnd_seq in sync with subflow rcv_wnd

[Matthieu Baerts <matttbe@kernel.org>]

Hi Simon,

On 11/03/2026 23:08, Simon Baatz wrote:
> On Wed, Mar 11, 2026 at 07:27:34PM +0100, Matthieu Baerts wrote:
>> On 09/03/2026 09:02, Simon Baatz via B4 Relay wrote:
>>> From: Simon Baatz <gmbnomis@gmail.com>
>>>
>>> MPTCP shares a receive window across subflows and applies it at the
>>> subflow level by adjusting each subflow's rcv_wnd when needed.  With
>>> the new TCP tracking of the maximum advertised window sequence,
>>> rcv_mwnd_seq must stay consistent with these subflow-level rcv_wnd
>>> adjustments.
>>
>> Thank you for these modifications!
>>
>>> Signed-off-by: Simon Baatz <gmbnomis@gmail.com>
>>> ---
>>>  net/mptcp/options.c | 6 ++++--
>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/net/mptcp/options.c b/net/mptcp/options.c
>>> index 43df4293f58bfbd8a8df6bf24b9f15e0f9e238f6..8a1c5698983cff3082d68290626dd8f1e044527f 100644
>>> --- a/net/mptcp/options.c
>>> +++ b/net/mptcp/options.c
>>
>> (...)
>>
>>> @@ -1338,8 +1339,9 @@ static void mptcp_set_rwin(struct tcp_sock *tp, struct tcphdr *th)
>>>  		 */
>>>  		rcv_wnd_new = rcv_wnd_old;
>>>  		win = rcv_wnd_old - ack_seq;
>>> -		tp->rcv_wnd = min_t(u64, win, U32_MAX);
>>> -		new_win = tp->rcv_wnd;
>>> +		new_win = min_t(u64, win, U32_MAX);
>>> +		tp->rcv_wnd = new_win;
>>
>> Out of curiosity, why did you change the two lines above?
>> (even if it makes sense, the diff is a bit confusing, and the commit
>> message doesn't mention this :) )
> 
> I wanted to keep tcp_update_max_rcv_wnd_seq() calls close to the
> respective update sites (same pattern everywhere).

Thanks, I now understand the reason.

> In the original form
> 
> tp->rcv_wnd = min_t(u64, win, U32_MAX);
> tcp_update_max_rcv_wnd_seq(tp);
> new_win = tp->rcv_wnd;
> 
> the ordering suggests that tcp_update_max_rcv_wnd_seq() might modify
> tp->rcv_wnd.

Note that if tp->rcv_mwnd_seq always needs to be modified when
tp->rcv_wnd and/or tp->rcv_wup are modified, maybe a single helper could
be called to modify all of them, so it might be less likely to forget
about modifying tp->rcv_mwnd_seq as well in the future.

But probably it might be unlikely to have new places where tp->rcv_wnd
and/or tp->rcv_wup need to be modified like here with MPTCP. So probably
fine like that.

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