Add rib_lookup_info() to provide API for retrieving individual route

  entries data in unified format.

There are control plane functions that require information other than
  just next-hop data (e.g. individual rtentry fields like flags or
  prefix/mask). Given that the goal is to avoid rte reference/refcounting,
  re-use rt_addrinfo structure to store most rte fields. If caller wants
  to retrieve key/mask or gateway (which are sockaddrs and are allocated
  separately), it needs to provide sufficient-sized sockaddrs structures
  w/ ther pointers saved in passed rt_addrinfo.

Convert:
  * lltable new records checks (in_lltable_rtcheck(),
    nd6_is_new_addr_neighbor().
  * rtsock pre-add/change route check.
  * IPv6 NS ND-proxy check (RADIX_MPATH code was eliminated because
     1) we don't support RTF_ANNOUNCE ND-proxy for networks and there should
       not be multiple host routes for such hosts 2) if we have multiple
       routes we should inspect them (which is not done). 3) the entire idea
       of abusing KRT as storage for ND proxy seems odd. Userland programs
       should be used for that purpose).

3 files changed, 165 insertions, 11 deletions

diff --git a/sys/net/route.c b/sys/net/route.c
index 86f99ee..11a5b8d 100644
--- a/sys/net/route.c
+++ b/sys/net/route.c
@@ -147,6 +147,8 @@ static void rt_notifydelete(struct rtentry *rt, struct rt_addrinfo *info);
 static struct radix_node *rt_mpath_unlink(struct radix_node_head *rnh,
     struct rt_addrinfo *info, struct rtentry *rto, int *perror);
 #endif
+static int rt_exportinfo(struct rtentry *rt, struct rt_addrinfo *info,
+    int flags);
 
 struct if_mtuinfo
 {
@@ -832,6 +834,147 @@ rtrequest_fib(int req,
 
 
 /*
+ * Copy most of @rt data into @info.
+ *
+ * If @flags contains NHR_COPY, copies dst,netmask and gw to the
+ * pointers specified by @info structure. Assume such pointers
+ * are zeroed sockaddr-like structures with sa_len field initialized
+ * to reflect size of the provided buffer. if no NHR_COPY is specified,
+ * point dst,netmask and gw @info fields to appropriate @rt values.
+ *
+ * if @flags contains NHR_REF, do refcouting on rt_ifp.
+ *
+ * Returns 0 on success.
+ */
+int
+rt_exportinfo(struct rtentry *rt, struct rt_addrinfo *info, int flags)
+{
+	struct rt_metrics *rmx;
+	struct sockaddr *src, *dst;
+	int sa_len;
+
+	if (flags & NHR_COPY) {
+		/* Copy destination if dst is non-zero */
+		src = rt_key(rt);
+		dst = info->rti_info[RTAX_DST];
+		sa_len = src->sa_len;
+		if (src != NULL && dst != NULL) {
+			if (src->sa_len > dst->sa_len)
+				return (ENOMEM);
+			memcpy(dst, src, src->sa_len);
+			info->rti_addrs |= RTA_DST;
+		}
+
+		/* Copy mask if set && dst is non-zero */
+		src = rt_mask(rt);
+		dst = info->rti_info[RTAX_NETMASK];
+		if (src != NULL && dst != NULL) {
+
+			/*
+			 * Radix stores different value in sa_len,
+			 * assume rt_mask() to have the same length
+			 * as rt_key()
+			 */
+			if (sa_len > dst->sa_len)
+				return (ENOMEM);
+			memcpy(dst, src, src->sa_len);
+			info->rti_addrs |= RTA_NETMASK;
+		}
+
+		/* Copy gateway is set && dst is non-zero */
+		src = rt->rt_gateway;
+		dst = info->rti_info[RTAX_GATEWAY];
+		if ((rt->rt_flags & RTF_GATEWAY) && src != NULL && dst != NULL){
+			if (src->sa_len > dst->sa_len)
+				return (ENOMEM);
+			memcpy(dst, src, src->sa_len);
+			info->rti_addrs |= RTA_GATEWAY;
+		}
+	} else {
+		info->rti_info[RTAX_DST] = rt_key(rt);
+		info->rti_addrs |= RTA_DST;
+		if (rt_mask(rt) != NULL) {
+			info->rti_info[RTAX_NETMASK] = rt_mask(rt);
+			info->rti_addrs |= RTA_NETMASK;
+		}
+		if (rt->rt_flags & RTF_GATEWAY) {
+			info->rti_info[RTAX_GATEWAY] = rt->rt_gateway;
+			info->rti_addrs |= RTA_GATEWAY;
+		}
+	}
+
+	rmx = info->rti_rmx;
+	if (rmx != NULL) {
+		info->rti_mflags |= RTV_MTU;
+		rmx->rmx_mtu = rt->rt_mtu;
+	}
+
+	info->rti_flags = rt->rt_flags;
+	info->rti_ifp = rt->rt_ifp;
+	info->rti_ifa = rt->rt_ifa;
+
+	if (flags & NHR_REF) {
+		/* Do 'traditional' refcouting */
+		if_ref(info->rti_ifp);
+	}
+
+	return (0);
+}
+
+/*
+ * Lookups up route entry for @dst in RIB database for fib @fibnum.
+ * Exports entry data to @info using rt_exportinfo().
+ *
+ * if @flags contains NHR_REF, refcouting is performed on rt_ifp.
+ *   All references can be released later by calling rib_free_info()
+ *
+ * Returns 0 on success.
+ * Returns ENOENT for lookup failure, ENOMEM for export failure.
+ */
+int
+rib_lookup_info(uint32_t fibnum, const struct sockaddr *dst, uint32_t flags,
+    uint32_t flowid, struct rt_addrinfo *info)
+{
+	struct radix_node_head *rh;
+	struct radix_node *rn;
+	struct rtentry *rt;
+	int error;
+
+	KASSERT((fibnum < rt_numfibs), ("rib_lookup_rte: bad fibnum"));
+	rh = rt_tables_get_rnh(fibnum, dst->sa_family);
+	if (rh == NULL)
+		return (ENOENT);
+
+	RADIX_NODE_HEAD_RLOCK(rh);
+	rn = rh->rnh_matchaddr(__DECONST(void *, dst), rh);
+	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
+		rt = RNTORT(rn);
+		/* Ensure route & ifp is UP */
+		if (RT_LINK_IS_UP(rt->rt_ifp)) {
+			flags = (flags & NHR_REF) | NHR_COPY;
+			error = rt_exportinfo(rt, info, flags);
+			RADIX_NODE_HEAD_RUNLOCK(rh);
+
+			return (error);
+		}
+	}
+	RADIX_NODE_HEAD_RUNLOCK(rh);
+
+	return (ENOENT);
+}
+
+/*
+ * Releases all references acquired by rib_lookup_info() when
+ * called with NHR_REF flags.
+ */
+void
+rib_free_info(struct rt_addrinfo *info)
+{
+
+	if_rele(info->rti_ifp);
+}
+
+/*
  * Iterates over all existing fibs in system calling
  *  @setwa_f function prior to traversing each fib.
  *  Calls @wa_f function for each element in current fib.
diff --git a/sys/net/route.h b/sys/net/route.h
index 847722c..7c69e1c 100644
--- a/sys/net/route.h
+++ b/sys/net/route.h
@@ -197,6 +197,9 @@ struct rtentry {
 #define	NHR_IFAIF		0x01	/* Return ifa_ifp interface */
 #define	NHR_REF			0x02	/* For future use */
 
+/* Control plane route request flags */
+#define	NHR_COPY		0x100	/* Copy rte data */
+
 /* rte<>nhop translation */
 static inline uint16_t
 fib_rte_to_nh_flags(int rt_flags)
@@ -460,6 +463,9 @@ void	 rtredirect_fib(struct sockaddr *, struct sockaddr *,
 int	 rtrequest_fib(int, struct sockaddr *,
 	    struct sockaddr *, struct sockaddr *, int, struct rtentry **, u_int);
 int	 rtrequest1_fib(int, struct rt_addrinfo *, struct rtentry **, u_int);
+int	rib_lookup_info(uint32_t, const struct sockaddr *, uint32_t, uint32_t,
+	    struct rt_addrinfo *);
+void	rib_free_info(struct rt_addrinfo *info);
 
 #include <sys/eventhandler.h>
 typedef void (*rtevent_redirect_fn)(void *, struct rtentry *, struct rtentry *, struct sockaddr *);
diff --git a/sys/net/rtsock.c b/sys/net/rtsock.c
index 6fcbbc6..a5ca9fa 100644
--- a/sys/net/rtsock.c
+++ b/sys/net/rtsock.c
@@ -614,11 +614,16 @@ route_output(struct mbuf *m, struct socket *so, ...)
 	 */
 	if (info.rti_info[RTAX_GATEWAY] != NULL &&
 	    info.rti_info[RTAX_GATEWAY]->sa_family != AF_LINK) {
-		struct route gw_ro;
+		struct rt_addrinfo ginfo;
+		struct sockaddr *gdst;
+
+		bzero(&ginfo, sizeof(ginfo));
+		bzero(&ss, sizeof(ss));
+		ss.ss_len = sizeof(ss);
+
+		ginfo.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&ss;
+		gdst = info.rti_info[RTAX_GATEWAY];
 
-		bzero(&gw_ro, sizeof(gw_ro));
-		gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY];
-		rtalloc_ign_fib(&gw_ro, 0, fibnum);
 		/* 
 		 * A host route through the loopback interface is 
 		 * installed for each interface adddress. In pre 8.0
@@ -629,14 +634,14 @@ route_output(struct mbuf *m, struct socket *so, ...)
 		 * 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;
-			info.rti_flags |= RTF_GWFLAG_COMPAT;
+		if (rib_lookup_info(fibnum, gdst, NHR_REF, 0, &ginfo) == 0) {
+			if (ss.ss_family == AF_LINK &&
+			    ginfo.rti_ifp->if_flags & IFF_LOOPBACK) {
+				info.rti_flags &= ~RTF_GATEWAY;
+				info.rti_flags |= RTF_GWFLAG_COMPAT;
+			}
+			rib_free_info(&ginfo);
 		}
-		if (gw_ro.ro_rt != NULL)
-			RTFREE(gw_ro.ro_rt);
 	}
 
 	switch (rtm->rtm_type) {