[lwip-devel] [bug #27445] PCB hangs in Fast Retransmit due to unchanging cwnd

[Kieran Mansley]

Follow-up Comment #3, bug #27445 (project lwip):

OK, combining the fix for this and bug #27329 was not as trivial as I hoped. 
I've ended up with that bit of code reading as follows:


      if (pcb->snd_wl2 + pcb->snd_wnd == right_wnd_edge){
        ++pcb->dupacks;

        if (pcb->dupacks >= 3) {
          if (!(pcb->flags & TF_INFR) && pcb->unacked != NULL) {
            /* This is fast retransmit. Retransmit the first unacked segment.
*/
            LWIP_DEBUGF(TCP_FR_DEBUG, ("tcp_receive: dupacks %"U16_F"
(%"U32_F"), fast retransmit %"U32_F"n",
                                       (u16_t)pcb->dupacks, pcb->lastack,
                                       ntohl(pcb->unacked->tcphdr->seqno)));
            tcp_rexmit(pcb);
            /* Set ssthresh to max (FlightSize / 2, 2*SMSS) */
            /*pcb->ssthresh = LWIP_MAX((pcb->snd_max -
              pcb->lastack) / 2,
              2 * pcb->mss);*/
            /* Set ssthresh to half of the minimum of the current cwnd and
the advertised window */
            if (pcb->cwnd > pcb->snd_wnd)
              pcb->ssthresh = pcb->snd_wnd / 2;
            else
              pcb->ssthresh = pcb->cwnd / 2;
            
            /* The minimum value for ssthresh should be 2 MSS */
            if (pcb->ssthresh < 2*pcb->mss) {
              LWIP_DEBUGF(TCP_FR_DEBUG, ("tcp_receive: The minimum value for
ssthresh %"U16_F" should be min 2 mss %"U16_F"...n", pcb->ssthresh,
2*pcb->mss));
              pcb->ssthresh = 2*pcb->mss;
            }
            
            pcb->cwnd = pcb->ssthresh + 3 * pcb->mss;
            pcb->flags |= TF_INFR;
          } else {
            /* Inflate the congestion window, but not if it means that
               the value overflows. */
            if ((u16_t)(pcb->cwnd + pcb->mss) > pcb->cwnd) {
              pcb->cwnd += pcb->mss;
            }
          }
        }

        if (pcb->unacked == NULL && pcb->unsent == NULL)
          pcb->dupacks = 0;
      }


Doest that look OK?

    _______________________________________________________

Reply to this item at:

  <http://savannah.nongnu.org/bugs/?27445>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.nongnu.org/