[PATCH net 0/2] geneve fixes

[PATCH net 0/2] geneve fixes

[Tariq Toukan]

Hi,

This small patchset by Gal provides bug fixes to the geneve tunnels flows.

Patch 1 fixes an incorrect value returned by the inner network header
offset helper.
Patch 2 fixes an issue inside the mlx5e tunneling flow. It 'happened' to
be harmless so far, before applying patch 1.

Series generated against:
commit d30d0e49da71 ("Merge tag 'net-6.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net")

Regards,
Tariq

Gal Pressman (2):
  geneve: Fix incorrect inner network header offset when
    innerprotoinherit is set
  net/mlx5e: Fix features validation check for tunneled UDP (non-VXLAN)
    packets

 drivers/net/ethernet/mellanox/mlx5/core/en_main.c |  3 +--
 drivers/net/geneve.c                              | 10 ++++++----
 include/net/ip_tunnels.h                          |  5 +++--
 3 files changed, 10 insertions(+), 8 deletions(-)

-- 
2.31.1

[PATCH net 1/2] geneve: Fix incorrect inner network header offset when innerprotoinherit is set

[Tariq Toukan]

From: Gal Pressman <gal@nvidia.com>

When innerprotoinherit is set, the tunneled packets do not have an inner
Ethernet header.
Change 'maclen' to not always assume the header length is ETH_HLEN, as
there might not be a MAC header.

This resolves issues with drivers (e.g. mlx5, in
mlx5e_tx_tunnel_accel()) who rely on the skb inner network header offset
to be correct, and use it for TX offloads.

Fixes: d8a6213d70ac ("geneve: fix header validation in geneve[6]_xmit_skb")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/geneve.c     | 10 ++++++----
 include/net/ip_tunnels.h |  5 +++--
 2 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index 51495cb4b9be..838e85ddec67 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -815,6 +815,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 			   struct geneve_dev *geneve,
 			   const struct ip_tunnel_info *info)
 {
+	bool inner_proto_inherit = geneve->cfg.inner_proto_inherit;
 	bool xnet = !net_eq(geneve->net, dev_net(geneve->dev));
 	struct geneve_sock *gs4 = rcu_dereference(geneve->sock4);
 	const struct ip_tunnel_key *key = &info->key;
@@ -826,7 +827,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be16 sport;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb))
+	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
 		return -EINVAL;
 
 	if (!gs4)
@@ -908,7 +909,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	}
 
 	err = geneve_build_skb(&rt->dst, skb, info, xnet, sizeof(struct iphdr),
-			       geneve->cfg.inner_proto_inherit);
+			       inner_proto_inherit);
 	if (unlikely(err))
 		return err;
 
@@ -925,6 +926,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 			    struct geneve_dev *geneve,
 			    const struct ip_tunnel_info *info)
 {
+	bool inner_proto_inherit = geneve->cfg.inner_proto_inherit;
 	bool xnet = !net_eq(geneve->net, dev_net(geneve->dev));
 	struct geneve_sock *gs6 = rcu_dereference(geneve->sock6);
 	const struct ip_tunnel_key *key = &info->key;
@@ -935,7 +937,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be16 sport;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb))
+	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
 		return -EINVAL;
 
 	if (!gs6)
@@ -997,7 +999,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 		ttl = ttl ? : ip6_dst_hoplimit(dst);
 	}
 	err = geneve_build_skb(dst, skb, info, xnet, sizeof(struct ipv6hdr),
-			       geneve->cfg.inner_proto_inherit);
+			       inner_proto_inherit);
 	if (unlikely(err))
 		return err;
 
diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 9a6a08ec7713..1db2417b8ff5 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -461,9 +461,10 @@ static inline bool pskb_inet_may_pull(struct sk_buff *skb)
 
 /* Variant of pskb_inet_may_pull().
  */
-static inline bool skb_vlan_inet_prepare(struct sk_buff *skb)
+static inline bool skb_vlan_inet_prepare(struct sk_buff *skb,
+					 bool inner_proto_inherit)
 {
-	int nhlen = 0, maclen = ETH_HLEN;
+	int nhlen = 0, maclen = inner_proto_inherit ? 0 : ETH_HLEN;
 	__be16 type = skb->protocol;
 
 	/* Essentially this is skb_protocol(skb, true)
-- 
2.31.1

[PATCH net 2/2] net/mlx5e: Fix features validation check for tunneled UDP (non-VXLAN) packets

[Tariq Toukan]

From: Gal Pressman <gal@nvidia.com>

Move the vxlan_features_check() call to after we verified the packet is
a tunneled VXLAN packet.

Without this, tunneled UDP non-VXLAN packets (for ex. GENENVE) might
wrongly not get offloaded.
In some cases, it worked by chance as GENEVE header is the same size as
VXLAN, but it is obviously incorrect.

Fixes: e3cfc7e6b7bd ("net/mlx5e: TX, Add geneve tunnel stateless offload support")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index c53c99dde558..a605eae56685 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -4875,7 +4875,7 @@ static netdev_features_t mlx5e_tunnel_features_check(struct mlx5e_priv *priv,
 
 		/* Verify if UDP port is being offloaded by HW */
 		if (mlx5_vxlan_lookup_port(priv->mdev->vxlan, port))
-			return features;
+			return vxlan_features_check(skb, features);
 
 #if IS_ENABLED(CONFIG_GENEVE)
 		/* Support Geneve offload for default UDP port */
@@ -4901,7 +4901,6 @@ netdev_features_t mlx5e_features_check(struct sk_buff *skb,
 	struct mlx5e_priv *priv = netdev_priv(netdev);
 
 	features = vlan_features_check(skb, features);
-	features = vxlan_features_check(skb, features);
 
 	/* Validate if the tunneled packet is being offloaded by HW */
 	if (skb->encapsulation &&
-- 
2.31.1

Re: [PATCH net 1/2] geneve: Fix incorrect inner network header offset when innerprotoinherit is set

[Wojciech Drewek]

On 06.06.2024 22:32, Tariq Toukan wrote:
> From: Gal Pressman <gal@nvidia.com>
> 
> When innerprotoinherit is set, the tunneled packets do not have an inner
> Ethernet header.
> Change 'maclen' to not always assume the header length is ETH_HLEN, as
> there might not be a MAC header.
> 
> This resolves issues with drivers (e.g. mlx5, in
> mlx5e_tx_tunnel_accel()) who rely on the skb inner network header offset
> to be correct, and use it for TX offloads.
> 
> Fixes: d8a6213d70ac ("geneve: fix header validation in geneve[6]_xmit_skb")
> Signed-off-by: Gal Pressman <gal@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> ---

Thanks!
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>

>  drivers/net/geneve.c     | 10 ++++++----
>  include/net/ip_tunnels.h |  5 +++--
>  2 files changed, 9 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
> index 51495cb4b9be..838e85ddec67 100644
> --- a/drivers/net/geneve.c
> +++ b/drivers/net/geneve.c
> @@ -815,6 +815,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  			   struct geneve_dev *geneve,
>  			   const struct ip_tunnel_info *info)
>  {
> +	bool inner_proto_inherit = geneve->cfg.inner_proto_inherit;
>  	bool xnet = !net_eq(geneve->net, dev_net(geneve->dev));
>  	struct geneve_sock *gs4 = rcu_dereference(geneve->sock4);
>  	const struct ip_tunnel_key *key = &info->key;
> @@ -826,7 +827,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  	__be16 sport;
>  	int err;
>  
> -	if (!skb_vlan_inet_prepare(skb))
> +	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
>  		return -EINVAL;
>  
>  	if (!gs4)
> @@ -908,7 +909,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  	}
>  
>  	err = geneve_build_skb(&rt->dst, skb, info, xnet, sizeof(struct iphdr),
> -			       geneve->cfg.inner_proto_inherit);
> +			       inner_proto_inherit);
>  	if (unlikely(err))
>  		return err;
>  
> @@ -925,6 +926,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  			    struct geneve_dev *geneve,
>  			    const struct ip_tunnel_info *info)
>  {
> +	bool inner_proto_inherit = geneve->cfg.inner_proto_inherit;
>  	bool xnet = !net_eq(geneve->net, dev_net(geneve->dev));
>  	struct geneve_sock *gs6 = rcu_dereference(geneve->sock6);
>  	const struct ip_tunnel_key *key = &info->key;
> @@ -935,7 +937,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  	__be16 sport;
>  	int err;
>  
> -	if (!skb_vlan_inet_prepare(skb))
> +	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
>  		return -EINVAL;
>  
>  	if (!gs6)
> @@ -997,7 +999,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
>  		ttl = ttl ? : ip6_dst_hoplimit(dst);
>  	}
>  	err = geneve_build_skb(dst, skb, info, xnet, sizeof(struct ipv6hdr),
> -			       geneve->cfg.inner_proto_inherit);
> +			       inner_proto_inherit);
>  	if (unlikely(err))
>  		return err;
>  
> diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
> index 9a6a08ec7713..1db2417b8ff5 100644
> --- a/include/net/ip_tunnels.h
> +++ b/include/net/ip_tunnels.h
> @@ -461,9 +461,10 @@ static inline bool pskb_inet_may_pull(struct sk_buff *skb)
>  
>  /* Variant of pskb_inet_may_pull().
>   */
> -static inline bool skb_vlan_inet_prepare(struct sk_buff *skb)
> +static inline bool skb_vlan_inet_prepare(struct sk_buff *skb,
> +					 bool inner_proto_inherit)
>  {
> -	int nhlen = 0, maclen = ETH_HLEN;
> +	int nhlen = 0, maclen = inner_proto_inherit ? 0 : ETH_HLEN;
>  	__be16 type = skb->protocol;
>  
>  	/* Essentially this is skb_protocol(skb, true)

Re: [PATCH net 2/2] net/mlx5e: Fix features validation check for tunneled UDP (non-VXLAN) packets

[Wojciech Drewek]

On 06.06.2024 22:32, Tariq Toukan wrote:
> From: Gal Pressman <gal@nvidia.com>
> 
> Move the vxlan_features_check() call to after we verified the packet is
> a tunneled VXLAN packet.
> 
> Without this, tunneled UDP non-VXLAN packets (for ex. GENENVE) might
> wrongly not get offloaded.
> In some cases, it worked by chance as GENEVE header is the same size as
> VXLAN, but it is obviously incorrect.
> 
> Fixes: e3cfc7e6b7bd ("net/mlx5e: TX, Add geneve tunnel stateless offload support")
> Signed-off-by: Gal Pressman <gal@nvidia.com>
> Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> ---

Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>

>  drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> index c53c99dde558..a605eae56685 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> @@ -4875,7 +4875,7 @@ static netdev_features_t mlx5e_tunnel_features_check(struct mlx5e_priv *priv,
>  
>  		/* Verify if UDP port is being offloaded by HW */
>  		if (mlx5_vxlan_lookup_port(priv->mdev->vxlan, port))
> -			return features;
> +			return vxlan_features_check(skb, features);
>  
>  #if IS_ENABLED(CONFIG_GENEVE)
>  		/* Support Geneve offload for default UDP port */
> @@ -4901,7 +4901,6 @@ netdev_features_t mlx5e_features_check(struct sk_buff *skb,
>  	struct mlx5e_priv *priv = netdev_priv(netdev);
>  
>  	features = vlan_features_check(skb, features);
> -	features = vxlan_features_check(skb, features);
>  
>  	/* Validate if the tunneled packet is being offloaded by HW */
>  	if (skb->encapsulation &&

Re: [PATCH net 0/2] geneve fixes

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net.git (main)
by David S. Miller <davem@davemloft.net>:

On Thu, 6 Jun 2024 23:32:47 +0300 you wrote:
> Hi,
> 
> This small patchset by Gal provides bug fixes to the geneve tunnels flows.
> 
> Patch 1 fixes an incorrect value returned by the inner network header
> offset helper.
> Patch 2 fixes an issue inside the mlx5e tunneling flow. It 'happened' to
> be harmless so far, before applying patch 1.
> 
> [...]

Here is the summary with links:
  - [net,1/2] geneve: Fix incorrect inner network header offset when innerprotoinherit is set
    https://git.kernel.org/netdev/net/c/c6ae073f5903
  - [net,2/2] net/mlx5e: Fix features validation check for tunneled UDP (non-VXLAN) packets
    https://git.kernel.org/netdev/net/c/791b4089e326

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html