Skip to contentSkip to conversation

vrf: check accept_source_route on the original netdevice

PATCH Superseded Stephen Suryaputra 3 participants 5 messages Mar 27, 2019 – Apr 1, 2019

Subsystems (3)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 vrf: check accept_source_route on the original netdevice

Configuration check to accept source route IP options should be made on the incoming netdevice when the skb->dev is an l3mdev master. The route lookup for the source route next hop also needs the incoming netdev.

---
 net/ipv4/ip_input.c   | 3 +++
 net/ipv4/ip_options.c | 6 +++++-
 2 files changed, 8 insertions(+), 1 deletion(-)
net/ipv4/ip_input.c +3 -0
index ecce2dc78f17..754176222ef6 100644
@@ -263,6 +263,9 @@ static inline bool ip_rcv_options(struct sk_buff *skb)
 	const struct iphdr *iph;
 	struct net_device *dev = skb->dev;
 
+	if (netif_is_l3_master(dev))
+		dev = __dev_get_by_index(dev_net(dev), IPCB(skb)->iif);
+
 	/* It looks as overkill, because not all
 	   IP options require packet mangling.
 	   But it is the easiest for now, especially taking
net/ipv4/ip_options.c +5 -1
index 32a35043c9f5..99f37591d7c1 100644
@@ -620,6 +620,7 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 	struct iphdr *iph = ip_hdr(skb);
 	unsigned char *optptr = skb_network_header(skb) + opt->srr;
 	struct rtable *rt = skb_rtable(skb);
+	struct net_device *dev = skb->dev;
 	struct rtable *rt2;
 	unsigned long orefdst;
 	int err;
@@ -638,6 +639,9 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 	if (rt->rt_type != RTN_LOCAL)
 		return -EINVAL;
 
+	if (netif_is_l3_master(dev))
+		dev = __dev_get_by_index(dev_net(dev), IPCB(skb)->iif);
+
 	for (srrptr = optptr[2], srrspace = optptr[1]; srrptr <= srrspace; srrptr += 4) {
 		if (srrptr + 3 > srrspace) {
 			icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((opt->srr+2)<<24));
@@ -647,7 +651,7 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 
 		orefdst = skb->_skb_refdst;
 		skb_dst_set(skb, NULL);
-		err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, skb->dev);
+		err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, dev);
 		rt2 = skb_rtable(skb);
 		if (err || (rt2->rt_type != RTN_UNICAST && rt2->rt_type != RTN_LOCAL)) {
 			skb_dst_drop(skb);
Signature · 1 line
2.17.1
maintainer
#1 lore raw

From: Stephen Suryaputra <[email protected]>
Date: Wed, 27 Mar 2019 15:58:48 -0400

Quoted 5 lines

Configuration check to accept source route IP options should be made on the incoming netdevice when the skb->dev is an l3mdev master. The route lookup for the source route next hop also needs the incoming netdev.

David, please review this.

Thanks.

#2 lore raw
Quoted 7 lines · David Miller

From: Stephen Suryaputra <[email protected]>
Date: Wed, 27 Mar 2019 15:58:48 -0400

Quoted 4 lines

Configuration check to accept source route IP options should be made on the incoming netdevice when the skb->dev is an l3mdev master. The route lookup for the source route next hop also needs the incoming netdev.

Quoted 6 lines
Quoted 1 line

David, please review this.

Thanks.

now that my heart rate is back to normal following the Auburn game ...

The change can be much simpler by passing the device down the stack from where it is saved before the skb->dev switch.

#3 lore raw
In reply to Stephen Suryaputra
Quoted 15 lines · Stephen Suryaputra

Configuration check to accept source route IP options should be made on the incoming netdevice when the skb->dev is an l3mdev master. The route lookup for the source route next hop also needs the incoming netdev.

---
 net/ipv4/ip_input.c   | 3 +++
 net/ipv4/ip_options.c | 6 +++++-
 2 files changed, 8 insertions(+), 1 deletion(-)
net/ipv4/ip_input.c +0 -0
index ecce2dc78f17..754176222ef6 100644

net/ipv4/ip_input.c static inline bool ip_rcv_options(struct sk_buff *skb) · line 267

Quoted 5 lines
net/ipv4/ip_input.c +2 -0
@@ -263,6 +263,9 @@ static inline bool ip_rcv_options(struct sk_buff *skb)
 	const struct iphdr *iph;
 	struct net_device *dev = skb->dev;
 
+	if (netif_is_l3_master(dev))
+		dev = __dev_get_by_index(dev_net(dev), IPCB(skb)->iif);

ip_rcv_options is called from ip_rcv_finish_core which already has this dev. Passing it as an arg saves the lookup. I believe you can always use dev as the input argument vs skb->dev.

Quoted 17 lines
net/ipv4/ip_input.c +1 -0
+
 	/* It looks as overkill, because not all
 	   IP options require packet mangling.
 	   But it is the easiest for now, especially taking
net/ipv4/ip_options.c +1 -0
index 32a35043c9f5..99f37591d7c1 100644
@@ -620,6 +620,7 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 	struct iphdr *iph = ip_hdr(skb);
 	unsigned char *optptr = skb_network_header(skb) + opt->srr;
 	struct rtable *rt = skb_rtable(skb);
+	struct net_device *dev = skb->dev;
 	struct rtable *rt2;
 	unsigned long orefdst;
 	int err;

net/ipv4/ip_options.c int ip_options_rcv_srr(struct sk_buff *skb) · line 643

Quoted 5 lines
net/ipv4/ip_options.c +2 -0
@@ -638,6 +639,9 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 	if (rt->rt_type != RTN_LOCAL)
 		return -EINVAL;
 
+	if (netif_is_l3_master(dev))
+		dev = __dev_get_by_index(dev_net(dev), IPCB(skb)->iif);

same here - just pass net_device from ip_rcv_options

Quoted 14 lines
net/ipv4/ip_options.c +2 -1
+
 	for (srrptr = optptr[2], srrspace = optptr[1]; srrptr <= srrspace; srrptr += 4) {
 		if (srrptr + 3 > srrspace) {
 			icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((opt->srr+2)<<24));
@@ -647,7 +651,7 @@ int ip_options_rcv_srr(struct sk_buff *skb)
 
 		orefdst = skb->_skb_refdst;
 		skb_dst_set(skb, NULL);
-		err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, skb->dev);
+		err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, dev);
 		rt2 = skb_rtable(skb);
 		if (err || (rt2->rt_type != RTN_UNICAST && rt2->rt_type != RTN_LOCAL)) {
 			skb_dst_drop(skb);