Re: [PATCH RFC mptcp] net: tcp: prevent transition to FIN_WAIT2 when mptcp has unacked data_fin

[Paolo Abeni <pabeni@redhat.com>]

Hello,

sharing here private discussion to keep all on the same page...

On Mon, 2022-08-22 at 14:32 +0200, Florian Westphal wrote:
> Paolo Abeni <pabeni@redhat.com> wrote:
> > > Full test case stashed here for the time being:
> > > https://github.com/fw-strlen/packetdrill/commit/60a3f57ea309f910643e06dbf123236741c7f8d9
> > > 
> > > Its possible to either change tw_sk to have enough mptcp context to be
> > > able to send a packet back, or we can supress/delay the tw_sock
> > > transition.  This is what classic TCP is doing when it receives another
> > > FIN while in FIN1 state.
> > > 
> > > This change makes the test case work (another dss ack is sent), but
> > > there may be other corner cases where we need to delay the sk ->
> > > tw_sk transition.
> > > 
> > > If the general idea looks ok, perhaps its better to replace
> > > 
> > > tcp_time_wait(sk, skb, ..
> > > 
> > > with a mptcp aware helper, e.g.  tcp_time_wait_check(sk, skb, ..
> > > and place the option parsing for mptcp-subflows there?
> > 
> > _if_ (big if) I read correctly, this patch "always" (most common
> > shutdown sequence should match the condition checked here, right?)
> 
> Really?  I had to fudge with the test case a long time to get the needed
> transition, I'm not even sure the test case is legit.
> 
> https://github.com/fw-strlen/packetdrill/commit/60a3f57ea309f910643e06dbf123236741c7f8d9
> 
> is that even correct?!

I mean/fear that a simple shutdown sequence (local close(), followed by
remote data fin) could end-up matching the test implemented by this
patch (even if the data fin retransmission will really happen on a much
more rare condition). 

That is, possibly the test is a little more relaxed then actual packet
sequence necessary to trigger the issue (but I don't know how to make
it more strict). 

I'll try to double check the above.

Thanks!

Paolo