[PATCH net] tunnels: Don't require remote endpoint or ID during creation.

[PATCH net] tunnels: Don't require remote endpoint or ID during creation.

[Jesse Gross]

Before lightweight tunnels existed, it really didn't make sense to
create a tunnel that was not fully specified, such as without a
destination IP address - the resulting packets would go nowhere.
However, with lightweight tunnels, the opposite is true - it doesn't
make sense to require this information when it will be provided later
on by the route. This loosens the requirements for this information.

An alternative would be to allow the relaxed version only when
COLLECT_METADATA is enabled. However, since there are several
variations on this theme (such as NBMA tunnels in GRE), just dropping
the restrictions seems the most consistent across tunnels and with
the existing configuration.

CC: John Linville <linville@tuxdriver.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
 drivers/net/geneve.c | 12 ++++++------
 drivers/net/vxlan.c  |  7 +++----
 2 files changed, 9 insertions(+), 10 deletions(-)

diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index 8f5c02e..cde29f8 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -870,14 +870,14 @@ static int geneve_newlink(struct net *net, struct net_device *dev,
 	__be16 dst_port = htons(GENEVE_UDP_PORT);
 	__u8 ttl = 0, tos = 0;
 	bool metadata = false;
-	__be32 rem_addr;
-	__u32 vni;
+	__be32 rem_addr = 0;
+	__u32 vni = 0;
 
-	if (!data[IFLA_GENEVE_ID] || !data[IFLA_GENEVE_REMOTE])
-		return -EINVAL;
+	if (data[IFLA_GENEVE_ID])
+		vni = nla_get_u32(data[IFLA_GENEVE_ID]);
 
-	vni = nla_get_u32(data[IFLA_GENEVE_ID]);
-	rem_addr = nla_get_in_addr(data[IFLA_GENEVE_REMOTE]);
+	if (data[IFLA_GENEVE_REMOTE])
+		rem_addr = nla_get_in_addr(data[IFLA_GENEVE_REMOTE]);
 
 	if (data[IFLA_GENEVE_TTL])
 		ttl = nla_get_u8(data[IFLA_GENEVE_TTL]);
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index bbac1d3..afdc65f 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -2745,11 +2745,10 @@ static int vxlan_newlink(struct net *src_net, struct net_device *dev,
 	struct vxlan_config conf;
 	int err;
 
-	if (!data[IFLA_VXLAN_ID])
-		return -EINVAL;
-
 	memset(&conf, 0, sizeof(conf));
-	conf.vni = nla_get_u32(data[IFLA_VXLAN_ID]);
+
+	if (data[IFLA_VXLAN_ID])
+		conf.vni = nla_get_u32(data[IFLA_VXLAN_ID]);
 
 	if (data[IFLA_VXLAN_GROUP]) {
 		conf.remote_ip.sin.sin_addr.s_addr = nla_get_in_addr(data[IFLA_VXLAN_GROUP]);
-- 
2.1.4

Re: [PATCH net] tunnels: Don't require remote endpoint or ID during creation.

[Thomas Graf]

On 10/16/15 at 04:36pm, Jesse Gross wrote:
> Before lightweight tunnels existed, it really didn't make sense to
> create a tunnel that was not fully specified, such as without a
> destination IP address - the resulting packets would go nowhere.
> However, with lightweight tunnels, the opposite is true - it doesn't
> make sense to require this information when it will be provided later
> on by the route. This loosens the requirements for this information.
> 
> An alternative would be to allow the relaxed version only when
> COLLECT_METADATA is enabled. However, since there are several
> variations on this theme (such as NBMA tunnels in GRE), just dropping
> the restrictions seems the most consistent across tunnels and with
> the existing configuration.
> 
> CC: John Linville <linville@tuxdriver.com>
> Signed-off-by: Jesse Gross <jesse@nicira.com>

I left it like it was for VXLAN with the thought that you had
specifically enable the metadata driven TX but I'm perfectly
fine with this as well.

Signed-off-by: Thomas Graf <tgraf@suug.ch>

Re: [PATCH net] tunnels: Don't require remote endpoint or ID during creation.

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Fri, 16 Oct 2015 16:36:00 -0700

> Before lightweight tunnels existed, it really didn't make sense to
> create a tunnel that was not fully specified, such as without a
> destination IP address - the resulting packets would go nowhere.
> However, with lightweight tunnels, the opposite is true - it doesn't
> make sense to require this information when it will be provided later
> on by the route. This loosens the requirements for this information.
> 
> An alternative would be to allow the relaxed version only when
> COLLECT_METADATA is enabled. However, since there are several
> variations on this theme (such as NBMA tunnels in GRE), just dropping
> the restrictions seems the most consistent across tunnels and with
> the existing configuration.
> 
> CC: John Linville <linville@tuxdriver.com>
> Signed-off-by: Jesse Gross <jesse@nicira.com>

Applied, thanks Jesse.