net: ipmr: add support for passing full packet on wrong vif

This patch adds support for IGMPMSG_WRVIFWHOLE which is used to pass
full packet and real vif id when the incoming interface is wrong.
While the RP and FHR are setting up state we need to be sending the
registers encapsulated with all the data inside otherwise we lose it.
The RP then decapsulates it and forwards it to the interested parties.
Currently with WRONGVIF we can only be sending empty register packets
and will lose that data.
This behaviour can be enabled by using MRT_PIM with
val == IGMPMSG_WRVIFWHOLE. This doesn't prevent IGMPMSG_WRONGVIF from
happening, it happens in addition to it, also it is controlled by the same
throttling parameters as WRONGVIF (i.e. 1 packet per 3 seconds currently).
Both messages are generated to keep backwards compatibily and avoid
breaking someone who was enabling MRT_PIM with val == 4, since any
positive val is accepted and treated the same.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

3 files changed, 19 insertions, 5 deletions

diff --git a/include/linux/mroute_base.h b/include/linux/mroute_base.h
index fd436cd..6675b9f 100644
--- a/include/linux/mroute_base.h
+++ b/include/linux/mroute_base.h
@@ -254,6 +254,7 @@ struct mr_table {
 	atomic_t		cache_resolve_queue_len;
 	bool			mroute_do_assert;
 	bool			mroute_do_pim;
+	bool			mroute_do_wrvifwhole;
 	int			mroute_reg_vif_num;
 };
 
diff --git a/include/uapi/linux/mroute.h b/include/uapi/linux/mroute.h
index 10f9ff9..5d37a9c 100644
--- a/include/uapi/linux/mroute.h
+++ b/include/uapi/linux/mroute.h
@@ -120,6 +120,7 @@ enum {
 	IPMRA_TABLE_MROUTE_DO_ASSERT,
 	IPMRA_TABLE_MROUTE_DO_PIM,
 	IPMRA_TABLE_VIFS,
+	IPMRA_TABLE_MROUTE_DO_WRVIFWHOLE,
 	__IPMRA_TABLE_MAX
 };
 #define IPMRA_TABLE_MAX (__IPMRA_TABLE_MAX - 1)
@@ -173,5 +174,6 @@ enum {
 #define IGMPMSG_NOCACHE		1		/* Kern cache fill request to mrouted */
 #define IGMPMSG_WRONGVIF	2		/* For PIM assert processing (unused) */
 #define IGMPMSG_WHOLEPKT	3		/* For PIM Register processing */
+#define IGMPMSG_WRVIFWHOLE	4		/* For PIM Register and assert processing */
 
 #endif /* _UAPI__LINUX_MROUTE_H */
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 82f9141..5660adc 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -1052,7 +1052,7 @@ static int ipmr_cache_report(struct mr_table *mrt,
 	struct sk_buff *skb;
 	int ret;
 
-	if (assert == IGMPMSG_WHOLEPKT)
+	if (assert == IGMPMSG_WHOLEPKT || assert == IGMPMSG_WRVIFWHOLE)
 		skb = skb_realloc_headroom(pkt, sizeof(struct iphdr));
 	else
 		skb = alloc_skb(128, GFP_ATOMIC);
@@ -1060,7 +1060,7 @@ static int ipmr_cache_report(struct mr_table *mrt,
 	if (!skb)
 		return -ENOBUFS;
 
-	if (assert == IGMPMSG_WHOLEPKT) {
+	if (assert == IGMPMSG_WHOLEPKT || assert == IGMPMSG_WRVIFWHOLE) {
 		/* Ugly, but we have no choice with this interface.
 		 * Duplicate old header, fix ihl, length etc.
 		 * And all this only to mangle msg->im_msgtype and
@@ -1071,9 +1071,12 @@ static int ipmr_cache_report(struct mr_table *mrt,
 		skb_reset_transport_header(skb);
 		msg = (struct igmpmsg *)skb_network_header(skb);
 		memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr));
-		msg->im_msgtype = IGMPMSG_WHOLEPKT;
+		msg->im_msgtype = assert;
 		msg->im_mbz = 0;
-		msg->im_vif = mrt->mroute_reg_vif_num;
+		if (assert == IGMPMSG_WRVIFWHOLE)
+			msg->im_vif = vifi;
+		else
+			msg->im_vif = mrt->mroute_reg_vif_num;
 		ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2;
 		ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) +
 					     sizeof(struct iphdr));
@@ -1372,6 +1375,7 @@ int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval,
 	struct mr_table *mrt;
 	struct vifctl vif;
 	struct mfcctl mfc;
+	bool do_wrvifwhole;
 	u32 uval;
 
 	/* There's one exception to the lock - MRT_DONE which needs to unlock */
@@ -1502,10 +1506,12 @@ int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval,
 			break;
 		}
 
+		do_wrvifwhole = (val == IGMPMSG_WRVIFWHOLE);
 		val = !!val;
 		if (val != mrt->mroute_do_pim) {
 			mrt->mroute_do_pim = val;
 			mrt->mroute_do_assert = val;
+			mrt->mroute_do_wrvifwhole = do_wrvifwhole;
 		}
 		break;
 	case MRT_TABLE:
@@ -1983,6 +1989,9 @@ static void ip_mr_forward(struct net *net, struct mr_table *mrt,
 			       MFC_ASSERT_THRESH)) {
 			c->_c.mfc_un.res.last_assert = jiffies;
 			ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF);
+			if (mrt->mroute_do_wrvifwhole)
+				ipmr_cache_report(mrt, skb, true_vifi,
+						  IGMPMSG_WRVIFWHOLE);
 		}
 		goto dont_forward;
 	}
@@ -2659,7 +2668,9 @@ static bool ipmr_fill_table(struct mr_table *mrt, struct sk_buff *skb)
 			mrt->mroute_reg_vif_num) ||
 	    nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_ASSERT,
 		       mrt->mroute_do_assert) ||
-	    nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_PIM, mrt->mroute_do_pim))
+	    nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_PIM, mrt->mroute_do_pim) ||
+	    nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_WRVIFWHOLE,
+		       mrt->mroute_do_wrvifwhole))
 		return false;
 
 	return true;