[src/netbsd-1-5]: src/sys/netinet6 Pull up revision 1.63 (requested by itojun):

[he <he%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/1c70c2081b1d
branches:  netbsd-1-5
changeset: 491084:1c70c2081b1d
user:      he <he%NetBSD.org@localhost>
date:      Wed Apr 04 17:14:31 2001 +0000

description:
Pull up revision 1.63 (requested by itojun):
  Make sure rcvif is sane on call to icmp6_reflect().
  Fixes panic in certain configurations / instances.

diffstat:

 sys/netinet6/icmp6.c |  11 ++++++++++-
 1 files changed, 10 insertions(+), 1 deletions(-)

diffs (25 lines):

diff -r 1dbfa0e26f1f -r 1c70c2081b1d sys/netinet6/icmp6.c
--- a/sys/netinet6/icmp6.c      Wed Apr 04 17:05:14 2001 +0000
+++ b/sys/netinet6/icmp6.c      Wed Apr 04 17:14:31 2001 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: icmp6.c,v 1.33.2.8 2001/03/11 21:11:58 he Exp $        */
+/*     $NetBSD: icmp6.c,v 1.33.2.9 2001/04/04 17:14:31 he Exp $        */
 /*     $KAME: icmp6.c,v 1.146 2000/10/01 12:37:20 itojun Exp $ */
 
 /*
@@ -269,6 +269,15 @@
        icmp6->icmp6_code = code;
        icmp6->icmp6_pptr = htonl((u_int32_t)param);
 
+       /*
+        * icmp6_reflect() is designed to be in the input path.
+        * icmp6_error() can be called from both input and outut path,
+        * and if we are in output path rcvif could contain bogus value.
+        * clear m->m_pkthdr.rcvif for safety, we should have enough scope
+        * information in ip header (nip6).
+        */
+       m->m_pkthdr.rcvif = NULL;
+
        icmp6stat.icp6s_outhist[type]++;
        icmp6_reflect(m, sizeof(struct ip6_hdr)); /*header order: IPv6 - ICMPv6*/