Re: [PATCH mptcp-net] mptcp: clear scheduled subflows on retransmit

[Geliang Tang <geliang@kernel.org>]

Hi Paolo,

On Thu, 2025-11-20 at 23:16 +0100, Paolo Abeni wrote:
> When __mptcp_retrans() kicks-in, it schedules one or more subflow for
> retransmission, but such subflows could be actually left alone if
> there
> is no more data to retransmit and/or in case of concurrent fallback.
> 
> Scheduled subflows could be processed much later in time, i.e. when
> new
> data will be transmitted, leading to bad subflow selection.
> 
> Explicitly clear all scheduled subflows before leaving the
> retransmission
> function.
> 
> Fixes: ee2708aedad0 ("mptcp: use get_retrans wrapper")
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> ---
>  net/mptcp/protocol.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 6f4220c8b5bb..3b327e9ccb78 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -2763,7 +2763,7 @@ static void __mptcp_retrans(struct sock *sk)
>  		}
>  
>  		if (!mptcp_send_head(sk))
> -			return;
> +			goto clear_scheduled;
>  
>  		goto reset_timer;
>  	}
> @@ -2794,7 +2794,7 @@ static void __mptcp_retrans(struct sock *sk)
>  			if (__mptcp_check_fallback(msk)) {
>  				spin_unlock_bh(&msk->fallback_lock);
>  				release_sock(ssk);
> -				return;
> +				goto clear_scheduled;
>  			}
>  
>  			while (info.sent < info.limit) {
> @@ -2826,6 +2826,15 @@ static void __mptcp_retrans(struct sock *sk)
>  
>  	if (!mptcp_rtx_timer_pending(sk))
>  		mptcp_reset_rtx_timer(sk);
> +
> +clear_scheduled:
> +	/* If no rtx data was available or in case of fallback,
> there
> +	 * could be left-over scheduled subflows; clear them all
> +	 * or later xmit could use bad ones
> +	 */
> +	mptcp_for_each_subflow(msk, subflow)
> +		if (READ_ONCE(subflow->scheduled))
> +			mptcp_subflow_set_scheduled(subflow, false);
>  }

If this __mptcp_retrans() needs to clear scheduled subflows, then
shouldn't similar this clear_scheduled operations also be added at the
end of both the __mptcp_push_pending() and
__mptcp_subflow_push_pending()?

Thanks,
-Geliang

>  
>  /* schedule the timeout timer for the relevant event: either close
> timeout