Skip to contentSkip to conversation

vrf: sit mtu should not be updated when vrf netdev is the link

PATCH In review Stephen Suryaputra 3 participants 5 messages May 6, 2019 – May 7, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 vrf: sit mtu should not be updated when vrf netdev is the link

VRF netdev mtu isn't typically set and have an mtu of 65536. When the link of a tunnel is set, the tunnel mtu is changed from 1480 to the link mtu minus tunnel header. In the case of VRF netdev is the link, then the tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in this case.

---
 net/ipv6/sit.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv6/sit.c +1 -1
index b2109b74857d..971d60bf9640 100644
@@ -1084,7 +1084,7 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
 	if (!tdev && tunnel->parms.link)
 		tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
 
-	if (tdev) {
+	if (tdev && !netif_is_l3_master(tdev)) {
 		int t_hlen = tunnel->hlen + sizeof(struct iphdr);
 
 		dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);
Signature · 1 line
2.17.1
#1 lore raw
Quoted 17 lines · Stephen Suryaputra

VRF netdev mtu isn't typically set and have an mtu of 65536. When the link of a tunnel is set, the tunnel mtu is changed from 1480 to the link mtu minus tunnel header. In the case of VRF netdev is the link, then the tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in this case.

---
 net/ipv6/sit.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv6/sit.c +0 -0
index b2109b74857d..971d60bf9640 100644

net/ipv6/sit.c static void ipip6_tunnel_bind_dev(struct net_device *dev) · line 1090

Quoted 8 lines
net/ipv6/sit.c +1 -1
@@ -1084,7 +1084,7 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
 	if (!tdev && tunnel->parms.link)
 		tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
 
-	if (tdev) {
+	if (tdev && !netif_is_l3_master(tdev)) {
 		int t_hlen = tunnel->hlen + sizeof(struct iphdr);
 
 		dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);

can you explain how tdev is a VRF device? What's the config setup for this case?

#2 lore raw
Quoted 24 lines · David Ahern
Quoted 23 lines · Stephen Suryaputra

VRF netdev mtu isn't typically set and have an mtu of 65536. When the link of a tunnel is set, the tunnel mtu is changed from 1480 to the link mtu minus tunnel header. In the case of VRF netdev is the link, then the tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in this case.

---
 net/ipv6/sit.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv6/sit.c +1 -1
index b2109b74857d..971d60bf9640 100644
@@ -1084,7 +1084,7 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
 	if (!tdev && tunnel->parms.link)
 		tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
 
-	if (tdev) {
+	if (tdev && !netif_is_l3_master(tdev)) {
 		int t_hlen = tunnel->hlen + sizeof(struct iphdr);
 
Quoted 5 lines

net/ipv6/sit.c static void ipip6_tunnel_bind_dev(struct net_device *dev) · line 1090

Quoted 2 lines
net/ipv6/sit.c +0 -0
 		dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);

can you explain how tdev is a VRF device? What's the config setup for this case?

Hi David,

tdev is set to VRF device per your suggestion to my colleague back in 2017:

	https://www.spinics.net/lists/netdev/msg462706.html.
Specifically this on this follow up:
	https://www.spinics.net/lists/netdev/msg463287.html

His basic config before your suggestion is available in:

	https://www.spinics.net/lists/netdev/msg462770.html

He and I had a refresher discussion this am trying to figure out if tdev should be a slave device. This is true if the local addr is specified. In this case the addr has to bound to a slave device. Then the underlay VRF can be derived from it. But if only remote is specified, then there isn't a straightforward way to associate the remote with a VRF unless tdev is set to a VRF device.

Reviewed-by
#3 lore raw
Quoted 40 lines · Stephen Suryaputra
Quoted 24 lines · David Ahern
Quoted 23 lines · Stephen Suryaputra

VRF netdev mtu isn't typically set and have an mtu of 65536. When the link of a tunnel is set, the tunnel mtu is changed from 1480 to the link mtu minus tunnel header. In the case of VRF netdev is the link, then the tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in this case.

---
 net/ipv6/sit.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv6/sit.c +1 -1
index b2109b74857d..971d60bf9640 100644
@@ -1084,7 +1084,7 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
 	if (!tdev && tunnel->parms.link)
 		tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
 
-	if (tdev) {
+	if (tdev && !netif_is_l3_master(tdev)) {
 		int t_hlen = tunnel->hlen + sizeof(struct iphdr);
 
Quoted 5 lines

net/ipv6/sit.c static void ipip6_tunnel_bind_dev(struct net_device *dev) · line 1090

Quoted 2 lines
net/ipv6/sit.c +0 -0
 		dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);

can you explain how tdev is a VRF device? What's the config setup for this case?

Hi David,

tdev is set to VRF device per your suggestion to my colleague back in 2017:

	https://www.spinics.net/lists/netdev/msg462706.html.
Specifically this on this follow up:
	https://www.spinics.net/lists/netdev/msg463287.html

His basic config before your suggestion is available in:

Quoted 9 lines
	https://www.spinics.net/lists/netdev/msg462770.html

He and I had a refresher discussion this am trying to figure out if tdev should be a slave device. This is true if the local addr is specified. In this case the addr has to bound to a slave device. Then the underlay VRF can be derived from it. But if only remote is specified, then there isn't a straightforward way to associate the remote with a VRF unless tdev is set to a VRF device.

Right. Thanks for the reminder.

maintainer
#4 lore raw
In reply to Stephen Suryaputra

From: Stephen Suryaputra <[email protected]>
Date: Mon, 6 May 2019 15:00:01 -0400

Quoted 7 lines

VRF netdev mtu isn't typically set and have an mtu of 65536. When the link of a tunnel is set, the tunnel mtu is changed from 1480 to the link mtu minus tunnel header. In the case of VRF netdev is the link, then the tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in this case.

Applied and queued up for -stable.