Add sysctls to if_enc(4) to control whether the firewalls or

bpf will see inner and outer headers or just inner or outer
headers for incoming and outgoing IPsec packets.

This is useful in bpf to not have over long lines for debugging
or selcting packets based on the inner headers.
It also properly defines the behavior of what the firewalls see.

Last but not least it gives you if_enc(4) for IPv6 as well.

[ As some auxiliary state was not available in the later
  input path we save it in the tdbi. That way tcpdump can give a
  consistent view of either of (authentic,confidential) for both
  before and after states. ]

Discussed with:	thompsa (2007-04-25, basic idea of unifying paths)
Reviewed by:	thompsa, gnn

6 files changed, 146 insertions, 18 deletions

diff --git a/sys/net/if_enc.c b/sys/net/if_enc.c
index e8d279c..3ab9081 100644
--- a/sys/net/if_enc.c
+++ b/sys/net/if_enc.c
@@ -60,7 +60,9 @@
 #include <netinet6/ip6_var.h>
 #endif
 
+#include "opt_enc.h"
 #include <netipsec/ipsec.h>
+#include <netipsec/xform.h>
 
 #define ENCMTU		(1024+512)
 
@@ -90,6 +92,32 @@ static void	enc_clone_destroy(struct ifnet *);
 
 IFC_SIMPLE_DECLARE(enc, 1);
 
+/*
+ * Sysctls.
+ */
+
+/*
+ * Before and after are relative to when we are stripping the
+ * outer IP header.
+ */
+SYSCTL_NODE(_net, OID_AUTO, enc, CTLFLAG_RW, 0, "enc sysctl");
+
+SYSCTL_NODE(_net_enc, OID_AUTO, in, CTLFLAG_RW, 0, "enc input sysctl");
+static int ipsec_filter_mask_in = ENC_BEFORE;
+SYSCTL_XINT(_net_enc_in, OID_AUTO, ipsec_filter_mask, CTLFLAG_RW,
+	&ipsec_filter_mask_in, 0, "IPsec input firewall filter mask");
+static int ipsec_bpf_mask_in = ENC_BEFORE;
+SYSCTL_XINT(_net_enc_in, OID_AUTO, ipsec_bpf_mask, CTLFLAG_RW,
+	&ipsec_bpf_mask_in, 0, "IPsec input bpf mask");
+
+SYSCTL_NODE(_net_enc, OID_AUTO, out, CTLFLAG_RW, 0, "enc output sysctl");
+static int ipsec_filter_mask_out = ENC_BEFORE;
+SYSCTL_XINT(_net_enc_out, OID_AUTO, ipsec_filter_mask, CTLFLAG_RW,
+	&ipsec_filter_mask_out, 0, "IPsec output firewall filter mask");
+static int ipsec_bpf_mask_out = ENC_BEFORE|ENC_AFTER;
+SYSCTL_XINT(_net_enc_out, OID_AUTO, ipsec_bpf_mask, CTLFLAG_RW,
+	&ipsec_bpf_mask_out, 0, "IPsec output bpf mask");
+
 static void
 enc_clone_destroy(struct ifnet *ifp)
 {
@@ -194,16 +222,26 @@ enc_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
 }
 
 int
-ipsec_filter(struct mbuf **mp, int dir)
+ipsec_filter(struct mbuf **mp, int dir, int flags)
 {
 	int error, i;
 	struct ip *ip;
 
 	KASSERT(encif != NULL, ("%s: encif is null", __func__));
+	KASSERT(flags & (ENC_IN|ENC_OUT),
+		("%s: invalid flags: %04x", __func__, flags));
 
 	if ((encif->if_drv_flags & IFF_DRV_RUNNING) == 0)
 		return (0);
 
+	if (flags & ENC_IN) {
+		if ((flags & ipsec_filter_mask_in) == 0)
+			return (0);
+	} else {
+		if ((flags & ipsec_filter_mask_out) == 0)
+			return (0);
+	}
+
 	/* Skip pfil(9) if no filters are loaded */
 	if (!(PFIL_HOOKED(&inet_pfil_hook)
 #ifdef INET6
@@ -269,23 +307,48 @@ bad:
 }
 
 void
-ipsec_bpf(struct mbuf *m, struct secasvar *sav, int af)
+ipsec_bpf(struct mbuf *m, struct secasvar *sav, int af, int flags)
 {
-	int flags;
+	int mflags;
 	struct enchdr hdr;
 
 	KASSERT(encif != NULL, ("%s: encif is null", __func__));
-	KASSERT(sav != NULL, ("%s: sav is null", __func__));
+	KASSERT(flags & (ENC_IN|ENC_OUT),
+		("%s: invalid flags: %04x", __func__, flags));
 
 	if ((encif->if_drv_flags & IFF_DRV_RUNNING) == 0)
 		return;
 
+	if (flags & ENC_IN) {
+		if ((flags & ipsec_bpf_mask_in) == 0)
+			return;
+	} else {
+		if ((flags & ipsec_bpf_mask_out) == 0)
+			return;
+	}
+
 	if (bpf_peers_present(encif->if_bpf)) {
-		flags = 0;
-		if (sav->alg_enc != SADB_EALG_NONE)
-			flags |= M_CONF;
-		if (sav->alg_auth != SADB_AALG_NONE)
-			flags |= M_AUTH;
+		mflags = 0;
+		hdr.spi = 0;
+		if (!sav) {
+			struct m_tag *mtag;
+			mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
+			if (mtag != NULL) {
+				struct tdb_ident *tdbi;
+				tdbi = (struct tdb_ident *) (mtag + 1);
+				if (tdbi->alg_enc != SADB_EALG_NONE)
+					mflags |= M_CONF;
+				if (tdbi->alg_auth != SADB_AALG_NONE)
+					mflags |= M_AUTH;
+				hdr.spi = tdbi->spi;
+			}
+		} else {
+			if (sav->alg_enc != SADB_EALG_NONE)
+				mflags |= M_CONF;
+			if (sav->alg_auth != SADB_AALG_NONE)
+				mflags |= M_AUTH;
+			hdr.spi = sav->spi;
+		}
 
 		/*
 		 * We need to prepend the address family as a four byte
@@ -295,8 +358,8 @@ ipsec_bpf(struct mbuf *m, struct secasvar *sav, int af)
 		 * to it).
 		 */
 		hdr.af = af;
-		hdr.spi = sav->spi;
-		hdr.flags = flags;
+		/* hdr.spi already set above */
+		hdr.flags = mflags;
 
 		bpf_mtap2(encif->if_bpf, &hdr, sizeof(hdr), m);
 	}
diff --git a/sys/netipsec/ipsec.h b/sys/netipsec/ipsec.h
index 1464014..6efacc6 100644
--- a/sys/netipsec/ipsec.h
+++ b/sys/netipsec/ipsec.h
@@ -410,8 +410,15 @@ extern	void m_checkalignment(const char* where, struct mbuf *m0,
 extern	struct mbuf *m_makespace(struct mbuf *m0, int skip, int hlen, int *off);
 extern	caddr_t m_pad(struct mbuf *m, int n);
 extern	int m_striphdr(struct mbuf *m, int skip, int hlen);
-extern	int ipsec_filter(struct mbuf **, int);
-extern	void ipsec_bpf(struct mbuf *, struct secasvar *, int);
+
+#ifdef DEV_ENC
+#define	ENC_BEFORE	0x0001
+#define	ENC_AFTER	0x0002
+#define	ENC_IN		0x0100
+#define	ENC_OUT		0x0200
+extern	int ipsec_filter(struct mbuf **, int, int);
+extern	void ipsec_bpf(struct mbuf *, struct secasvar *, int, int);
+#endif
 #endif /* _KERNEL */
 
 #ifndef _KERNEL
diff --git a/sys/netipsec/ipsec_input.c b/sys/netipsec/ipsec_input.c
index cea8aff..63677ec 100644
--- a/sys/netipsec/ipsec_input.c
+++ b/sys/netipsec/ipsec_input.c
@@ -444,6 +444,9 @@ ipsec4_common_input_cb(struct mbuf *m, struct secasvar *sav,
 		bcopy(&saidx->dst, &tdbi->dst, saidx->dst.sa.sa_len);
 		tdbi->proto = sproto;
 		tdbi->spi = sav->spi;
+		/* Cache those two for enc(4) in xform_ipip. */
+		tdbi->alg_auth = sav->alg_auth;
+		tdbi->alg_enc = sav->alg_enc;
 
 		m_tag_prepend(m, mtag);
 	} else if (mt != NULL) {
@@ -458,10 +461,10 @@ ipsec4_common_input_cb(struct mbuf *m, struct secasvar *sav,
 	 * Pass the mbuf to enc0 for bpf and pfil. We will filter the IPIP
 	 * packet later after it has been decapsulated.
 	 */
-	ipsec_bpf(m, sav, AF_INET);
+	ipsec_bpf(m, sav, AF_INET, ENC_IN|ENC_BEFORE);
 
 	if (prot != IPPROTO_IPIP)
-		if ((error = ipsec_filter(&m, PFIL_IN)) != 0)
+		if ((error = ipsec_filter(&m, PFIL_IN, ENC_IN|ENC_BEFORE)) != 0)
 			return (error);
 #endif
 
@@ -703,6 +706,9 @@ ipsec6_common_input_cb(struct mbuf *m, struct secasvar *sav, int skip, int proto
 		bcopy(&saidx->dst, &tdbi->dst, sizeof(union sockaddr_union));
 		tdbi->proto = sproto;
 		tdbi->spi = sav->spi;
+		/* Cache those two for enc(4) in xform_ipip. */
+		tdbi->alg_auth = sav->alg_auth;
+		tdbi->alg_enc = sav->alg_enc;
 
 		m_tag_prepend(m, mtag);
 	} else {
@@ -713,6 +719,19 @@ ipsec6_common_input_cb(struct mbuf *m, struct secasvar *sav, int skip, int proto
 
 	key_sa_recordxfer(sav, m);
 
+#ifdef DEV_ENC
+	/*
+	 * Pass the mbuf to enc0 for bpf and pfil. We will filter the IPIP
+	 * packet later after it has been decapsulated.
+	 */
+	ipsec_bpf(m, sav, AF_INET6, ENC_IN|ENC_BEFORE);
+
+	/* XXX-BZ does not make sense. */
+	if (prot != IPPROTO_IPIP)
+		if ((error = ipsec_filter(&m, PFIL_IN, ENC_IN|ENC_BEFORE)) != 0)
+			return (error);
+#endif
+
 	/* Retrieve new protocol */
 	m_copydata(m, protoff, sizeof(u_int8_t), (caddr_t) &nxt8);
 
diff --git a/sys/netipsec/ipsec_output.c b/sys/netipsec/ipsec_output.c
index 27ad224..ae7c08b 100644
--- a/sys/netipsec/ipsec_output.c
+++ b/sys/netipsec/ipsec_output.c
@@ -362,8 +362,10 @@ ipsec4_process_packet(
 	sav = isr->sav;
 
 #ifdef DEV_ENC
+	/* pass the mbuf to enc0 for bpf processing */
+	ipsec_bpf(m, sav, AF_INET, ENC_OUT|ENC_BEFORE);
 	/* pass the mbuf to enc0 for packet filtering */
-	if ((error = ipsec_filter(&m, PFIL_OUT)) != 0)
+	if ((error = ipsec_filter(&m, PFIL_OUT, ENC_OUT|ENC_BEFORE)) != 0)
 		goto bad;
 #endif
 
@@ -466,7 +468,10 @@ ipsec4_process_packet(
 
 #ifdef DEV_ENC
 	/* pass the mbuf to enc0 for bpf processing */
-	ipsec_bpf(m, sav, AF_INET);
+	ipsec_bpf(m, sav, AF_INET, ENC_OUT|ENC_AFTER);
+	/* pass the mbuf to enc0 for packet filtering */
+	if ((error = ipsec_filter(&m, PFIL_OUT, ENC_OUT|ENC_AFTER)) != 0)
+		goto bad;
 #endif
 
 	/*
@@ -710,6 +715,14 @@ ipsec6_output_tunnel(struct ipsec_output_state *state, struct secpolicy *sp, int
 	if (isr == NULL)
 		goto bad;
 
+#ifdef DEV_ENC
+	/* pass the mbuf to enc0 for bpf processing */
+	ipsec_bpf(m, isr->sav, AF_INET6, ENC_OUT|ENC_BEFORE);
+	/* pass the mbuf to enc0 for packet filtering */
+	if ((error = ipsec_filter(&m, PFIL_OUT, ENC_OUT|ENC_BEFORE)) != 0)
+		goto bad;
+#endif
+
 	/*
 	 * There may be the case that SA status will be changed when
 	 * we are refering to one. So calling splsoftnet().
@@ -778,6 +791,15 @@ ipsec6_output_tunnel(struct ipsec_output_state *state, struct secpolicy *sp, int
 		goto bad;
 	}
 	ip6 = mtod(m, struct ip6_hdr *);
+
+#ifdef DEV_ENC
+	/* pass the mbuf to enc0 for bpf processing */
+	ipsec_bpf(m, isr->sav, AF_INET6, ENC_OUT|ENC_AFTER);
+	/* pass the mbuf to enc0 for packet filtering */
+	if ((error = ipsec_filter(&m, PFIL_OUT, ENC_OUT|ENC_AFTER)) != 0)
+		goto bad;
+#endif
+
 	error = (*isr->sav->tdb_xform->xf_output)(m, isr, NULL,
 		sizeof (struct ip6_hdr),
 		offsetof(struct ip6_hdr, ip6_nxt));
diff --git a/sys/netipsec/xform.h b/sys/netipsec/xform.h
index 58509c5..92f7866 100644
--- a/sys/netipsec/xform.h
+++ b/sys/netipsec/xform.h
@@ -57,6 +57,9 @@ struct tdb_ident {
 	u_int32_t spi;
 	union sockaddr_union dst;
 	u_int8_t proto;
+	/* Cache those two for enc(4) in xform_ipip. */
+	u_int8_t alg_auth;
+	u_int8_t alg_enc;
 };
 
 /*
diff --git a/sys/netipsec/xform_ipip.c b/sys/netipsec/xform_ipip.c
index 80fafd2..e0f54f5 100644
--- a/sys/netipsec/xform_ipip.c
+++ b/sys/netipsec/xform_ipip.c
@@ -348,8 +348,22 @@ _ipip_input(struct mbuf *m, int iphlen, struct ifnet *gifp)
 	ipipstat.ipips_ibytes += m->m_pkthdr.len - iphlen;
 
 #ifdef DEV_ENC
+	switch (v >> 4) {
+#ifdef INET
+	case 4:
+		ipsec_bpf(m, NULL, AF_INET, ENC_IN|ENC_AFTER);
+		break;
+#endif
+#ifdef INET6
+	case 6:
+		ipsec_bpf(m, NULL, AF_INET6, ENC_IN|ENC_AFTER);
+		break;
+#endif
+	default:
+		panic("%s: bogus ip version %u", __func__, v>>4);
+	}
 	/* pass the mbuf to enc0 for packet filtering */
-	if (ipsec_filter(&m, PFIL_IN) != 0)
+	if (ipsec_filter(&m, PFIL_IN, ENC_IN|ENC_AFTER) != 0)
 		return;
 #endif