This is just a nit, but it seems to me that although the code
in tcp_reass() counts completely overlapping packets via the
statistics tcps_rcvduppack and tcps_rcvdupbyte, it does NOT
count partially overlapping packets as it should with
tcps_rcvpartduppack and tcps_rcvpartdupbyte.
I'm thinking that the following change ought to do the
right thing, but I'd like to know if anyone disagrees:
--- tcp_input.c Sat Dec 15 03:23:56 2001
+++ tcp_input.c+ Tue Mar 26 11:04:35 2002
@@ -223,6 +223,9 @@
* completes.
*/
goto present; /* ??? */
+ } else {
+ tcpstat.tcps_rcvpartduppack++;
+ tcpstat.tcps_rcvpartdupbyte += *tlenp;
}
m_adj(m, i);
*tlenp -= i;
--
G. Paul Ziemba [EMAIL PROTECTED]
FreeBSD unix:
12:36PM up 1 day, 2:44, 9 users, load averages: 0.04, 0.04, 0.01
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-net" in the body of the message
