Bug 28360

Comment 1 ru 2001-06-23 09:26:13 UTC

Responsible Changed
From-To: freebsd-bugs->ru

Will look into this next week.

Comment 2 ru 2001-06-23 12:44:06 UTC

On Fri, Jun 22, 2001 at 10:38:56PM -0700, David W. Chapman Jr. wrote:
> 
> I want internet traffic to come from source ip 66.64.12.249 when
> using internet from router.
> 
> >How-To-Repeat:
> route add default 66.64.6.1 -ifa 66.64.12.249
> >Fix:
> temp workaround:
> 
> route add default 66.64.6.1
> route change default -ifa 66.64.12.249
> 
This isn't a bug in the route(8) command.  Kernel routing code
(in rtsock.c) checks for -ifa address only when processing the
RTM_CHANGE command.  Look here (from the route(8) manpage):

: In a change or add command where the destination and gateway
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
: are not sufficient to specify the route (as in the ISO case
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
: where several interfaces may have the same address), the -ifp
: or -ifa modifiers may be used to determine the interface or
: interface address.

But I agree that having such a functionality in `route add'
would be useful too.  The following patch duplicates some code
from RTM_CHANGE.  Please test.  (The `ifp' part should probably
be duplicated too.)

Index: rtsock.c
===================================================================
RCS file: /home/ncvs/src/sys/net/rtsock.c,v
retrieving revision 1.44.2.2
diff -u -p -r1.44.2.2 rtsock.c
--- rtsock.c	2000/08/03 00:09:34	1.44.2.2
+++ rtsock.c	2001/06/23 11:35:38
@@ -282,7 +282,7 @@ route_output(m, so)
 	struct rt_addrinfo info;
 	int len, error = 0;
 	struct ifnet *ifp = 0;
-	struct ifaddr *ifa = 0;
+	struct ifaddr *ifa = 0, *oifa;
 
 #define senderr(e) { error = e; goto flush;}
 	if (m == 0 || ((m->m_len < sizeof(long)) &&
@@ -332,6 +332,18 @@ route_output(m, so)
 		error = rtrequest(RTM_ADD, dst, gate, netmask,
 					rtm->rtm_flags, &saved_nrt);
 		if (error == 0 && saved_nrt) {
+			if (ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) {
+				oifa = saved_nrt->rt_ifa;
+				if (oifa != ifa) {
+					if (oifa && oifa->ifa_rtrequest)
+						oifa->ifa_rtrequest(RTM_DELETE,
+						    saved_nrt, gate);
+					IFAFREE(saved_nrt->rt_ifa);
+					saved_nrt->rt_ifa = ifa;
+					ifa->ifa_refcnt++;
+					saved_nrt->rt_ifp = ifa->ifa_ifp;
+				}
+			}
 			rt_setmetrics(rtm->rtm_inits,
 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
 			saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
@@ -424,7 +436,7 @@ route_output(m, so)
 							rt_key(rt), gate))))
 				ifp = ifa->ifa_ifp;
 			if (ifa) {
-				register struct ifaddr *oifa = rt->rt_ifa;
+				oifa = rt->rt_ifa;
 				if (oifa != ifa) {
 				    if (oifa && oifa->ifa_rtrequest)
 					oifa->ifa_rtrequest(RTM_DELETE,



Cheers,
-- 
Ruslan Ermilov		Oracle Developer/DBA,
ru@sunbay.com		Sunbay Software AG,
ru@FreeBSD.org		FreeBSD committer,
+380.652.512.251	Simferopol, Ukraine

http://www.FreeBSD.org	The Power To Serve
http://www.oracle.com	Enabling The Information Age

I must have misplaced my reply.  This patch works great!

I forget, what is the reason this hasn't been in -current yet?

Comment 5 ru 2001-09-24 10:36:47 UTC

On Sun, Sep 23, 2001 at 01:00:33PM -0500, David W. Chapman Jr. wrote:
> I forget, what is the reason this hasn't been in -current yet?
> 
NetBSD has a more generic solution to this problem, which also
allows the -ifp to be specified with RTM_ADD.  This is very
useful in the IPv6 case.

This requires some substantial changes to sys/net/rtsock.c, and
I haven't yet found the time to do it.

Sorry for the delay, but please be patient.


Thanks,
-- 
Ruslan Ermilov		Oracle Developer/DBA,
ru@sunbay.com		Sunbay Software AG,
ru@FreeBSD.org		FreeBSD committer,
+380.652.512.251	Simferopol, Ukraine

http://www.FreeBSD.org	The Power To Serve
http://www.oracle.com	Enabling The Information Age

Comment 6 ru 2001-10-17 18:49:39 UTC

State Changed
From-To: open->closed

Fixed in 5.0-CURRENT, sys/net/rtsock.c,v 1.61. 
MFC after 1 month.