Re: [PATCH RESENT v7 mptcp-next 4/4] mptcp: leverage the backlog for RX packet processing

[Paolo Abeni <pabeni@redhat.com>]

On 11/1/25 1:04 AM, Mat Martineau wrote:
> On Mon, 27 Oct 2025, Paolo Abeni wrote:
>> @@ -3573,6 +3586,8 @@ static void mptcp_release_cb(struct sock *sk)
>>
>> 		cond_resched();
>> 		spin_lock_bh(&sk->sk_lock.slock);
>> +		if (spool_bl)
>> +			mptcp_backlog_spooled(sk, moved, &skbs);
> 
> Hi Paolo -
> 
> Given the discussion in v5, to address the "wild producer" scenario this 
> loop can keep track of total_moved and exit the loop when that gets too 
> large.
> 
> The question is what the total limit would be - available rcvbuf when 
> entering the loop, or some multiple of that?

I spent a lot of time around that choice. Limiting the loop here is
quite tricky. Stopping the loop after sk_rcvbuf bytes causes mptcp-level
OoO and adds additional complexity to the code (as the receiver will
need to check the backlog even when the receive queue is not empty.

Note that the code proposed in this series is as robust as the current
(i.e. release_cb() can already be bothered by a wild producer in exactly
the same way).

I propose to decouple the wild producer problem solution from these patches.

/P