svn commit: r291991 - stable/9/sys/netinet
Andrey V. Elsukov
ae at FreeBSD.org
Tue Dec 8 07:47:02 UTC 2015
Author: ae
Date: Tue Dec 8 07:47:01 2015
New Revision: 291991
URL: https://svnweb.freebsd.org/changeset/base/291991
Log:
MFC r291579:
In the same way fix the problem described in r291578 for IGMPv3.
In case when router has a lot of multicast groups, the reply can take
several packets due to MTU limitation.
Also we have a limit IGMP_MAX_RESPONSE_BURST == 4, that limits the number
of packets we send in one shot. Then we recalculate the timer value and
schedule the remaining packets for sending.
The problem is that when we call igmp_v3_dispatch_general_query() to send
remaining packets, we queue new reply in the same mbuf queue. And when
number of packets is bigger than IGMP_MAX_RESPONSE_BURST, we get endless
reply of IGMPv3 reports.
To fix this, add the check for remaining packets in the queue.
Modified:
stable/9/sys/netinet/igmp.c
Directory Properties:
stable/9/sys/ (props changed)
Modified: stable/9/sys/netinet/igmp.c
==============================================================================
--- stable/9/sys/netinet/igmp.c Tue Dec 8 07:43:12 2015 (r291990)
+++ stable/9/sys/netinet/igmp.c Tue Dec 8 07:47:01 2015 (r291991)
@@ -3326,6 +3326,15 @@ igmp_v3_dispatch_general_query(struct ig
KASSERT(igi->igi_version == IGMP_VERSION_3,
("%s: called when version %d", __func__, igi->igi_version));
+ /*
+ * Check that there are some packets queued. If so, send them first.
+ * For large number of groups the reply to general query can take
+ * many packets, we should finish sending them before starting of
+ * queuing the new reply.
+ */
+ if (igi->igi_gq.ifq_head != NULL)
+ goto send;
+
ifp = igi->igi_ifp;
IF_ADDR_RLOCK(ifp);
@@ -3361,6 +3370,7 @@ igmp_v3_dispatch_general_query(struct ig
}
IF_ADDR_RUNLOCK(ifp);
+send:
loop = (igi->igi_flags & IGIF_LOOPBACK) ? 1 : 0;
igmp_dispatch_queue(&igi->igi_gq, IGMP_MAX_RESPONSE_BURST, loop);
More information about the svn-src-stable-9
mailing list