[src/trunk]: src/sys/net/npf Mmh, put back the RFC6946 check (about dummy fra...

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

details:   https://anonhg.NetBSD.org/src/rev/0e69e7d92a81
branches:  trunk
changeset: 321367:0e69e7d92a81
user:      maxv <maxv%NetBSD.org@localhost>
date:      Tue Mar 13 16:23:40 2018 +0000

description:
Mmh, put back the RFC6946 check (about dummy fragments), otherwise NPF
is not happy in npf_reassembly, because NPC_IPFRAG is again returned after
the packet was reassembled.

I'm wondering whether it would not be better to just remove the fragment
header in frag6_input directly.

diffstat:

 sys/net/npf/npf_inet.c |  11 +++++++++--
 1 files changed, 9 insertions(+), 2 deletions(-)

diffs (32 lines):

diff -r c5c73ddbdfb5 -r 0e69e7d92a81 sys/net/npf/npf_inet.c
--- a/sys/net/npf/npf_inet.c    Tue Mar 13 14:54:13 2018 +0000
+++ b/sys/net/npf/npf_inet.c    Tue Mar 13 16:23:40 2018 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: npf_inet.c,v 1.40 2018/03/13 09:04:02 maxv Exp $       */
+/*     $NetBSD: npf_inet.c,v 1.41 2018/03/13 16:23:40 maxv Exp $       */
 
 /*-
  * Copyright (c) 2009-2014 The NetBSD Foundation, Inc.
@@ -40,7 +40,7 @@
 
 #ifdef _KERNEL
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: npf_inet.c,v 1.40 2018/03/13 09:04:02 maxv Exp $");
+__KERNEL_RCSID(0, "$NetBSD: npf_inet.c,v 1.41 2018/03/13 16:23:40 maxv Exp $");
 
 #include <sys/param.h>
 #include <sys/types.h>
@@ -392,6 +392,13 @@
                                if (ip6f == NULL)
                                        return NPC_FMTERR;
 
+                               /* RFC6946: Skip dummy fragments. */
+                               if (!ntohs(ip6f->ip6f_offlg & IP6F_OFF_MASK) &&
+                                   !(ip6f->ip6f_offlg & IP6F_MORE_FRAG)) {
+                                       hlen = sizeof(struct ip6_frag);
+                                       break;
+                               }
+
                                hlen = 0;
                                flags |= NPC_IPFRAG;