On Sun, 30 Aug 2009, Qing Li wrote:
In ip_output(), the flow-table module must not try to cache L2/L3
information for interface of IFF_POINTOPOINT or IFF_LOOPBACK type.
Since the L2 information (rt_lle) is invalid for these interface
types, accidental caching attempt will trigger panic when the invalid
rt_lle reference is accessed.
When installing a new route, or when updating an existing route, the
user supplied gateway address may be an interface address (this is
particularly true for point-to-point interface related modules such
as ppp, if_tun, if_gif). Currently the routing command handler always
set the RTF_GATEWAY flag if the gateway address is given as part of the
command paramters. Therefore the gateway address must be verified against
interface addresses or else the route would be treated as an indirect
route, thus making that route unusable.
With this change, can I mark the following two TODO items as done:
* RTM_CHANGE in net/rtsock.c can incorrectly set RTF_GATEWAY (QingLi) (in
progress)
* flowtable mishandles gateway (G) routes on POINT2POINT interfaces
(BrianSomers) (in progress)
Also, do we believe your last few commits fix the three issues raised by
Sato-san in his IPv6 reports:
* IPv6 regression on 8.x (QingLi) (in progress)
Thanks!
Robert N M Watson
Computer Laboratory
University of Cambridge
Reviewed by: kmacy, julian, rwatson
Approved by: re
Modified:
stable/8/sys/ (props changed)
stable/8/sys/amd64/include/xen/ (props changed)
stable/8/sys/cddl/contrib/opensolaris/ (props changed)
stable/8/sys/contrib/dev/acpica/ (props changed)
stable/8/sys/contrib/pf/ (props changed)
stable/8/sys/dev/xen/xenpci/ (props changed)
stable/8/sys/net/flowtable.c
stable/8/sys/net/rtsock.c
Modified: stable/8/sys/net/flowtable.c
==
--- stable/8/sys/net/flowtable.cSun Aug 30 22:39:49 2009
(r196672)
+++ stable/8/sys/net/flowtable.cSun Aug 30 22:42:32 2009
(r196673)
@@ -692,6 +692,12 @@ uncached:
struct rtentry *rt = ro->ro_rt;
struct ifnet *ifp = rt->rt_ifp;
+ if (ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) {
+ RTFREE(rt);
+ ro->ro_rt = NULL;
+ return (ENOENT);
+ }
+
if (rt->rt_flags & RTF_GATEWAY)
l3addr = rt->rt_gateway;
else
Modified: stable/8/sys/net/rtsock.c
==
--- stable/8/sys/net/rtsock.c Sun Aug 30 22:39:49 2009(r196672)
+++ stable/8/sys/net/rtsock.c Sun Aug 30 22:42:32 2009(r196673)
@@ -513,6 +513,39 @@ route_output(struct mbuf *m, struct sock
senderr(error);
}
+ /*
+* The given gateway address may be an interface address.
+* For example, issuing a "route change" command on a route
+* entry that was created from a tunnel, and the gateway
+* address given is the local end point. In this case the
+* RTF_GATEWAY flag must be cleared or the destination will
+* not be reachable even though there is no error message.
+*/
+ if (info.rti_info[RTAX_GATEWAY] != NULL &&
+ info.rti_info[RTAX_GATEWAY]->sa_family != AF_LINK) {
+ struct route gw_ro;
+
+ bzero(&gw_ro, sizeof(gw_ro));
+ gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY];
+ rtalloc_ign(&gw_ro, 0);
+ /*
+* A host route through the loopback interface is
+* installed for each interface adddress. In pre 8.0
+* releases the interface address of a PPP link type
+* is not reachable locally. This behavior is fixed as
+* part of the new L2/L3 redesign and rewrite work. The
+* signature of this interface address route is the
+* AF_LINK sa_family type of the rt_gateway, and the
+* rt_ifp has the IFF_LOOPBACK flag set.
+*/
+ if (gw_ro.ro_rt != NULL &&
+ gw_ro.ro_rt->rt_gateway->sa_family == AF_LINK &&
+ gw_ro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)
+ info.rti_flags &= ~RTF_GATEWAY;
+ if (gw_ro.ro_rt != NULL)
+ RTFREE(gw_ro.ro_rt);
+ }
+
switch (rtm->rtm_type) {
struct rtentry *saved_nrt;
@@ -714,7 +747,7 @@ route_output(struct mbuf *m, struct sock
RT_UNLOCK(rt);
senderr(error);
}
- rt->rt_flags |= RTF_GATEWAY;
+ rt->rt_flags |= (RTF_GATEWAY & info.rti_flags);
}
