PERFORCE change 129560 for review

Rui Paulo rpaulo at FreeBSD.org
Mon Nov 26 06:09:30 PST 2007


http://perforce.freebsd.org/chv.cgi?CH=129560

Change 129560 by rpaulo at rpaulo_zoo on 2007/11/26 14:08:46

	* Correct a sequence number check.
	* Ignore TH_ECE and TH_CWR in the header prediction check.
	
	Tested in the netperf cluster. The tcpdump was analysed with
	wireshark and no packets were lost when ECN was enabled.

Affected files ...

.. //depot/projects/tcpecn/netinet/tcp_input.c#3 edit

Differences ...

==== //depot/projects/tcpecn/netinet/tcp_input.c#3 (text+ko) ====

@@ -968,7 +968,7 @@
 		 * Ignore if we are already trying to recover.
 		 */
 		if ((thflags & TH_ECE) &&
-		    SEQ_GEQ(tp->snd_una, tp->snd_recover))
+		    SEQ_LEQ(th->th_ack, tp->snd_recover))
 			tcp_congestion_exp(tp);
 	}
 
@@ -1040,8 +1040,7 @@
 	 */
 	if (tp->t_state == TCPS_ESTABLISHED &&
 	    th->th_seq == tp->rcv_nxt &&
-	    (thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK|TH_ECE|TH_CWR))
-	        == TH_ACK &&
+	    (thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
 	    tp->snd_nxt == tp->snd_max &&
 	    tiwin && tiwin == tp->snd_wnd && 
 	    ((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) &&


More information about the p4-projects mailing list