netlink: Added VRF support

Added VRF support to the iproute routines.
Updated iproute* calls in the ipoe, radius, and dhcpv6 code.

Signed-off-by: Andrii Melnychenko <a.melnychenko@vyos.io>

6 files changed, 137 insertions, 28 deletions

diff --git a/accel-pppd/ctrl/ipoe/ipoe.c b/accel-pppd/ctrl/ipoe/ipoe.c
index 6ce35a69..bc55aa69 100644
--- a/accel-pppd/ctrl/ipoe/ipoe.c
+++ b/accel-pppd/ctrl/ipoe/ipoe.c
@@ -1022,9 +1022,9 @@ static void __ipoe_session_activate(struct ipoe_session *ses)
 			in_addr_t gw;
 			iproute_get(ses->router, &gw, NULL);
 			if (gw)
-				iproute_add(0, ses->siaddr, ses->yiaddr, gw, conf_proto, 32);
+				iproute_add(0, ses->siaddr, ses->yiaddr, gw, conf_proto, 32, NULL);
 			else
-				iproute_add(0, ses->siaddr, ses->router, gw, conf_proto, 32);
+				iproute_add(0, ses->siaddr, ses->router, gw, conf_proto, 32, NULL);
 		}*/
 
 		if (serv->opt_mode == MODE_L3)
@@ -1068,9 +1068,9 @@ static void __ipoe_session_activate(struct ipoe_session *ses)
 
 	if (ses->ifindex == -1 && !serv->opt_ifcfg) {
 		if (!serv->opt_ip_unnumbered)
-			iproute_add(serv->ifindex, ses->router, ses->yiaddr, 0, conf_proto, ses->mask, 0);
+			iproute_add(serv->ifindex, ses->router, ses->yiaddr, 0, conf_proto, ses->mask, 0, NULL);
 		else
-			iproute_add(serv->ifindex, serv->opt_src ?: ses->router, ses->yiaddr, 0, conf_proto, 32, 0);
+			iproute_add(serv->ifindex, serv->opt_src ?: ses->router, ses->yiaddr, 0, conf_proto, 32, 0, NULL);
 	}
 
 	if (ses->l4_redirect)
@@ -1171,7 +1171,7 @@ static void ipoe_session_started(struct ap_session *s)
 
 	if (ses->ses.ipv4->peer_addr != ses->yiaddr)
 		//ipaddr_add_peer(ses->ses.ifindex, ses->router, ses->yiaddr); // breaks quagga
-		iproute_add(ses->ses.ifindex, ses->router, ses->yiaddr, 0, conf_proto, 32, 0);
+		iproute_add(ses->ses.ifindex, ses->router, ses->yiaddr, 0, conf_proto, 32, 0, NULL);
 
 	if (ses->ifindex != -1 && ses->xid) {
 		ses->dhcpv4 = dhcpv4_create(ses->ctrl.ctx, ses->ses.ifname, "");
@@ -1255,9 +1255,9 @@ static void ipoe_session_finished(struct ap_session *s)
 	} else if (ses->started) {
 		if (!serv->opt_ifcfg) {
 			if (!serv->opt_ip_unnumbered)
-				iproute_del(serv->ifindex, ses->router, ses->yiaddr, 0, conf_proto, ses->mask, 0);
+				iproute_del(serv->ifindex, ses->router, ses->yiaddr, 0, conf_proto, ses->mask, 0, NULL);
 			else
-				iproute_del(serv->ifindex, serv->opt_src ?: ses->router, ses->yiaddr, 0, conf_proto, 32, 0);
+				iproute_del(serv->ifindex, serv->opt_src ?: ses->router, ses->yiaddr, 0, conf_proto, 32, 0, NULL);
 		}
 	}
 
diff --git a/accel-pppd/ifcfg.c b/accel-pppd/ifcfg.c
index 3750060b..4e4a381a 100644
--- a/accel-pppd/ifcfg.c
+++ b/accel-pppd/ifcfg.c
@@ -234,7 +234,7 @@ void __export ap_session_ifdown(struct ap_session *ses)
 			if (!a->installed)
 				continue;
 			if (a->prefix_len > 64)
-				ip6route_del(ses->ifindex, &a->addr, a->prefix_len, NULL, 0, 0);
+				ip6route_del(ses->ifindex, &a->addr, a->prefix_len, NULL, 0, 0, ses->vrf_name);
 			else {
 				struct in6_addr addr;
 				memcpy(addr.s6_addr, &a->addr, 8);
diff --git a/accel-pppd/ipv6/dhcpv6.c b/accel-pppd/ipv6/dhcpv6.c
index 85863e0c..92a6ed6c 100644
--- a/accel-pppd/ipv6/dhcpv6.c
+++ b/accel-pppd/ipv6/dhcpv6.c
@@ -162,7 +162,7 @@ static void ev_ses_finished(struct ap_session *ses)
 		if (pd->dp_active) {
 			struct ipv6db_addr_t *p;
 			list_for_each_entry(p, &ses->ipv6_dp->prefix_list, entry)
-				ip6route_del(0, &p->addr, p->prefix_len, NULL, 0, 0);
+				ip6route_del(0, &p->addr, p->prefix_len, NULL, 0, 0, NULL);
 		}
 
 		ipdb_put_ipv6_prefix(ses, ses->ipv6_dp);
@@ -184,7 +184,7 @@ static void insert_dp_routes(struct ap_session *ses, struct dhcpv6_pd *pd, struc
 		addr = NULL;
 
 	list_for_each_entry(p, &ses->ipv6_dp->prefix_list, entry) {
-		if (ip6route_add(ses->ifindex, &p->addr, p->prefix_len, addr, 0, 0)) {
+		if (ip6route_add(ses->ifindex, &p->addr, p->prefix_len, addr, 0, 0, NULL)) {
 			err = errno;
 			inet_ntop(AF_INET6, &p->addr, str1, sizeof(str1));
 			if (addr)
diff --git a/accel-pppd/libnetlink/iputils.c b/accel-pppd/libnetlink/iputils.c
index 4bf7fd01..34181e2a 100644
--- a/accel-pppd/libnetlink/iputils.c
+++ b/accel-pppd/libnetlink/iputils.c
@@ -11,10 +11,13 @@
 #include <errno.h>
 #include <time.h>
 #include <sys/uio.h>
-//#include <linux/if_link.h>
+#ifdef HAVE_VRF
+#include <linux/if_link.h>
+#endif
 //#include <linux/if_addr.h>
 //#include <linux/rtnetlink.h>
 #include <linux/fib_rules.h>
+#include <net/if.h>
 
 #include "log.h"
 
@@ -458,7 +461,81 @@ int __export ipaddr_del_peer(int ifindex, in_addr_t addr, in_addr_t peer)
 	return r;
 }
 
-int __export iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio)
+#ifdef HAVE_VRF
+static int ipvrf_get_table(uint32_t *tb_id, const char *vrf_name)
+{
+	struct iplink_req {
+		struct nlmsghdr n;
+		struct ifinfomsg i;
+		char buf[4096];
+	} req;
+	struct rtnl_handle *rth = net->rtnl_get();
+	struct rtattr *tb[IFLA_MAX+1];
+	struct rtattr *li[IFLA_INFO_MAX+1];
+	struct rtattr *vrf_attr[IFLA_VRF_MAX + 1];
+	struct ifinfomsg *ifi;
+	int len;
+	int r = -1;
+	*tb_id = RT_TABLE_MAIN;
+
+	log_ppp_info2("utils: getting route table for %s\n", vrf_name);
+
+	if (!vrf_name)
+		return 0;
+
+	memset(&req, 0, sizeof(req) - 4096);
+
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_GETLINK;
+	req.i.ifi_family = AF_UNSPEC;
+
+	addattr_l(&req.n, 4096, IFLA_IFNAME, vrf_name, strnlen(vrf_name, IFNAMSIZ));
+
+	r = rtnl_talk(rth, &req.n, 0, 0, &req.n, NULL, NULL, 0);
+	if (r < 0) {
+		log_ppp_error("BUG: route table \"%s\" not found\n", vrf_name);
+		return r;
+	}
+
+	ifi = NLMSG_DATA(&req.n);
+
+	len = req.n.nlmsg_len;
+
+	len -= NLMSG_LENGTH(sizeof(*ifi));
+	if (len < 0) {
+		goto error;
+	}
+
+	parse_rtattr(tb, IFLA_MAX, IFLA_RTA(ifi), len);
+
+	if (!tb[IFLA_LINKINFO])
+		goto error;
+
+	parse_rtattr_nested(li, IFLA_INFO_MAX, tb[IFLA_LINKINFO]);
+
+	if (!li[IFLA_INFO_KIND] || !li[IFLA_INFO_DATA])
+		goto error;
+
+	if (strncmp(RTA_DATA(li[IFLA_INFO_KIND]), "vrf", 4)) {
+		log_ppp_error("BUG: link \"%s\" is not VRF\n", vrf_name);
+		goto error;
+	}
+
+	parse_rtattr_nested(vrf_attr, IFLA_VRF_MAX, li[IFLA_INFO_DATA]);
+	if (vrf_attr[IFLA_VRF_TABLE])
+		*tb_id = *(uint32_t *)RTA_DATA(vrf_attr[IFLA_VRF_TABLE]);
+
+	if (*tb_id == 0)
+		log_ppp_error("BUG: VRF %s is missing table id\n", vrf_name);
+
+	return 0;
+error:
+	return -1;
+}
+#endif
+
+int __export iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio, const char *vrf_name)
 {
 	struct ipaddr_req {
 		struct nlmsghdr n;
@@ -473,16 +550,24 @@ int __export iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw
 
 	memset(&req, 0, sizeof(req) - 4096);
 
+	uint32_t rt_table = RT_TABLE_MAIN;
+#ifdef HAVE_VRF
+	if (ipvrf_get_table(&rt_table, vrf_name) < 0)
+		return -1;
+#endif
+
 	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
 	req.n.nlmsg_flags = NLM_F_REQUEST | NLM_F_CREATE;
 	req.n.nlmsg_type = RTM_NEWROUTE;
 	req.i.rtm_family = AF_INET;
-	req.i.rtm_table = RT_TABLE_MAIN;
+	req.i.rtm_table = rt_table;
 	req.i.rtm_scope = gw ? RT_SCOPE_UNIVERSE : RT_SCOPE_LINK;
 	req.i.rtm_protocol = proto;
 	req.i.rtm_type = RTN_UNICAST;
 	req.i.rtm_dst_len = mask;
 
+	if (rt_table != RT_TABLE_MAIN)
+		addattr32(&req.n, sizeof(req), RTA_TABLE, rt_table);
 	if (ifindex)
 		addattr32(&req.n, sizeof(req), RTA_OIF, ifindex);
 	if (src)
@@ -501,7 +586,7 @@ int __export iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw
 	return r;
 }
 
-int __export iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio)
+int __export iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio, const char *vrf_name)
 {
 	struct ipaddr_req {
 		struct nlmsghdr n;
@@ -516,16 +601,24 @@ int __export iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw
 
 	memset(&req, 0, sizeof(req) - 4096);
 
+	uint32_t rt_table = RT_TABLE_MAIN;
+#ifdef HAVE_VRF
+	if (ipvrf_get_table(&rt_table, vrf_name) < 0)
+		return -1;
+#endif
+
 	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
 	req.n.nlmsg_flags = NLM_F_REQUEST;
 	req.n.nlmsg_type = RTM_DELROUTE;
 	req.i.rtm_family = AF_INET;
-	req.i.rtm_table = RT_TABLE_MAIN;
+	req.i.rtm_table = rt_table;
 	req.i.rtm_scope = gw ? RT_SCOPE_UNIVERSE : RT_SCOPE_LINK;
 	req.i.rtm_protocol = proto;
 	req.i.rtm_type = RTN_UNICAST;
 	req.i.rtm_dst_len = mask;
 
+	if (rt_table != RT_TABLE_MAIN)
+		addattr32(&req.n, sizeof(req), RTA_TABLE, rt_table);
 	if (ifindex)
 		addattr32(&req.n, sizeof(req), RTA_OIF, ifindex);
 	if (src)
@@ -544,7 +637,7 @@ int __export iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw
 	return r;
 }
 
-int __export ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio)
+int __export ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio, const char *vrf_name)
 {
 	struct ipaddr_req {
 		struct nlmsghdr n;
@@ -559,17 +652,25 @@ int __export ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len,
 
 	memset(&req, 0, sizeof(req) - 4096);
 
+	uint32_t rt_table = RT_TABLE_MAIN;
+#ifdef HAVE_VRF
+	if (ipvrf_get_table(&rt_table, vrf_name) < 0)
+		return -1;
+#endif
+
 	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
 	req.n.nlmsg_flags = NLM_F_REQUEST | NLM_F_CREATE;
 	req.n.nlmsg_type = RTM_NEWROUTE;
 	req.i.rtm_family = AF_INET6;
-	req.i.rtm_table = RT_TABLE_MAIN;
+	req.i.rtm_table = rt_table;
 	req.i.rtm_scope = RT_SCOPE_UNIVERSE;
 	req.i.rtm_protocol = proto;
 	req.i.rtm_type = RTN_UNICAST;
 	req.i.rtm_dst_len = pref_len;
 
 	addattr_l(&req.n, sizeof(req), RTA_DST, dst, sizeof(*dst));
+	if (rt_table != RT_TABLE_MAIN)
+		addattr32(&req.n, sizeof(req), RTA_TABLE, rt_table);
 	if (ifindex)
 		addattr32(&req.n, sizeof(req), RTA_OIF, ifindex);
 	if (gw)
@@ -585,7 +686,7 @@ int __export ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len,
 	return r;
 }
 
-int __export ip6route_del(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio)
+int __export ip6route_del(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio, const char *vrf_name)
 {
 	struct ipaddr_req {
 		struct nlmsghdr n;
@@ -600,17 +701,25 @@ int __export ip6route_del(int ifindex, const struct in6_addr *dst, int pref_len,
 
 	memset(&req, 0, sizeof(req) - 4096);
 
+	uint32_t rt_table = RT_TABLE_MAIN;
+#ifdef HAVE_VRF
+	if (ipvrf_get_table(&rt_table, vrf_name) < 0)
+		return -1;
+#endif
+
 	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
 	req.n.nlmsg_flags = NLM_F_REQUEST;
 	req.n.nlmsg_type = RTM_DELROUTE;
 	req.i.rtm_family = AF_INET6;
-	req.i.rtm_table = RT_TABLE_MAIN;
+	req.i.rtm_table = rt_table;
 	req.i.rtm_scope = RT_SCOPE_UNIVERSE;
 	req.i.rtm_protocol = proto;
 	req.i.rtm_type = RTN_UNICAST;
 	req.i.rtm_dst_len = pref_len;
 
 	addattr_l(&req.n, sizeof(req), RTA_DST, dst, sizeof(*dst));
+	if (rt_table != RT_TABLE_MAIN)
+		addattr32(&req.n, sizeof(req), RTA_TABLE, rt_table);
 	if (ifindex)
 		addattr32(&req.n, sizeof(req), RTA_OIF, ifindex);
 	if (gw)
diff --git a/accel-pppd/libnetlink/iputils.h b/accel-pppd/libnetlink/iputils.h
index 9292cea2..86500882 100644
--- a/accel-pppd/libnetlink/iputils.h
+++ b/accel-pppd/libnetlink/iputils.h
@@ -4,6 +4,7 @@
 #include <linux/if_link.h>
 #include <netinet/in.h>
 #include <stdint.h>
+#include <config.h>
 
 typedef int (*iplink_list_func)(int index, int flags, const char *name, int iflink, int vid, void *arg);
 
@@ -20,12 +21,12 @@ int ipaddr_add_peer(int ifindex, in_addr_t addr, in_addr_t peer_addr);
 int ipaddr_del(int ifindex, in_addr_t addr, int mask);
 int ipaddr_del_peer(int ifindex, in_addr_t addr, in_addr_t peer);
 
-int iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio);
-int iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio);
+int iproute_add(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio, const char *vrf_name);
+int iproute_del(int ifindex, in_addr_t src, in_addr_t dst, in_addr_t gw, int proto, int mask, uint32_t prio, const char *vrf_name);
 in_addr_t iproute_get(in_addr_t dst, in_addr_t *gw);
 
-int ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio);
-int ip6route_del(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio);
+int ip6route_add(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio, const char *vrf_name);
+int ip6route_del(int ifindex, const struct in6_addr *dst, int pref_len, const struct in6_addr *gw, int proto, uint32_t prio, const char *vrf_name);
 int ip6addr_add(int ifindex, struct in6_addr *addr, int prefix_len);
 int ip6addr_add_peer(int ifindex, struct in6_addr *addr, struct in6_addr *peer_addr);
 int ip6addr_del(int ifindex, struct in6_addr *addr, int prefix_len);
diff --git a/accel-pppd/radius/radius.c b/accel-pppd/radius/radius.c
index 8dbbe1cd..143c237c 100644
--- a/accel-pppd/radius/radius.c
+++ b/accel-pppd/radius/radius.c
@@ -801,8 +801,7 @@ static void ses_started(struct ap_session *ses)
 		bool gw_spec = !IN6_IS_ADDR_UNSPECIFIED(&fr6->gw);
 		char nbuf[INET6_ADDRSTRLEN];
 		char gwbuf[INET6_ADDRSTRLEN];
-
-		if (ip6route_add(gw_spec ? 0 : rpd->ses->ifindex, &fr6->prefix, fr6->plen, gw_spec ? &fr6->gw : NULL, 3, fr6->prio)) {
+		if (ip6route_add(gw_spec ? 0 : rpd->ses->ifindex, &fr6->prefix, fr6->plen, gw_spec ? &fr6->gw : NULL, 3, fr6->prio, rpd->ses->vrf_name)) {
 			log_ppp_warn("radius: failed to add route %s/%hhu %s %u\n",
 				     u_ip6str(&fr6->prefix, nbuf), fr6->plen,
 				     u_ip6str(&fr6->gw, gwbuf), fr6->prio);
@@ -810,7 +809,7 @@ static void ses_started(struct ap_session *ses)
 	}
 
 	for (fr = rpd->fr; fr; fr = fr->next) {
-		if (iproute_add(fr->gw ? 0 : rpd->ses->ifindex, 0, fr->dst, fr->gw, 3, fr->mask, fr->prio)) {
+		if (iproute_add(fr->gw ? 0 : rpd->ses->ifindex, 0, fr->dst, fr->gw, 3, fr->mask, fr->prio, rpd->ses->vrf_name)) {
 			char dst[17], gw[17];
 			u_inet_ntoa(fr->dst, dst);
 			u_inet_ntoa(fr->gw, gw);
@@ -849,12 +848,12 @@ static void ses_finishing(struct ap_session *ses)
 		 * when the interface is removed.
 		 */
 		if (!IN6_IS_ADDR_UNSPECIFIED(&fr6->gw))
-			ip6route_del(0, &fr6->prefix, fr6->plen, &fr6->gw, 3, fr6->prio);
+			ip6route_del(0, &fr6->prefix, fr6->plen, &fr6->gw, 3, fr6->prio, rpd->ses->vrf_name);
 	}
 
 	for (fr = rpd->fr; fr; fr = fr->next) {
 		if (fr->gw)
-			iproute_del(0, 0, fr->dst, fr->gw, 3, fr->mask, fr->prio);
+			iproute_del(0, 0, fr->dst, fr->gw, 3, fr->mask, fr->prio, rpd->ses->vrf_name);
 	}
 
 	if (rpd->acct_started || rpd->acct_req)