Skip to contentSkip to conversation

udp gso fixes

PATCH In review Willem de Bruijn 2 participants 4 messages May 22, 2018 – May 23, 2018

Subsystems (5)
lore
Share & tools
lore

Willem de Bruijn

author maintainer
#0 lore raw

v2 cover udp gso fixes

From: Willem de Bruijn <[email protected]>

A few small fixes:
- disallow segmentation with XFRM
- do not leak gso packets into the ingress path

Changes
v1 -> v2
- fix build failure in team.c
- drop scatter-gather fix:
this is now fixed by commit 113f99c33585 ("net: test tailroom before appending to linear skb"). After this patch gso skbs are built non-linear regardless of NETIF_F_SG and skb_segment builds linear segs.

Willem de Bruijn (4):
udp: exclude gso from xfrm paths
gso: limit udp gso to egress-only virtual devices

 drivers/net/bonding/bond_main.c | 5 +++--
 drivers/net/team/team.c         | 5 +++--
 include/linux/netdev_features.h | 1 -
 net/ipv4/udp.c                  | 3 ++-
 net/ipv6/udp.c                  | 3 ++-
 5 files changed, 10 insertions(+), 7 deletions(-)
Signature · 1 line
2.17.0.441.gb46fe60e1d-goog
author maintainer
#1 lore raw

PATCH v2 1/2 udp: exclude gso from xfrm paths

From: Willem de Bruijn <[email protected]>

UDP GSO delays final datagram construction to the GSO layer. This conflicts with protocol transformations.

Fixes bec1f6f69736 ("udp: generate gso with UDP_SEGMENT")
CC Michal Kubecek <[email protected]>
---
 net/ipv4/udp.c | 3 ++-
 net/ipv6/udp.c | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)
net/ipv4/udp.c +2 -1
index ff4d4ba67735..d71f1f3e1155 100644
@@ -788,7 +788,8 @@ static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4,
 			return -EINVAL;
 		if (sk->sk_no_check_tx)
 			return -EINVAL;
-		if (skb->ip_summed != CHECKSUM_PARTIAL || is_udplite)
+		if (skb->ip_summed != CHECKSUM_PARTIAL || is_udplite ||
+		    dst_xfrm(skb_dst(skb)))
 			return -EIO;
 
 		skb_shinfo(skb)->gso_size = cork->gso_size;
net/ipv6/udp.c +2 -1
index 2839c1bd1e58..426c9d2b418d 100644
@@ -1053,7 +1053,8 @@ static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6,
 			return -EINVAL;
 		if (udp_sk(sk)->no_check6_tx)
 			return -EINVAL;
-		if (skb->ip_summed != CHECKSUM_PARTIAL || is_udplite)
+		if (skb->ip_summed != CHECKSUM_PARTIAL || is_udplite ||
+		    dst_xfrm(skb_dst(skb)))
 			return -EIO;
 
 		skb_shinfo(skb)->gso_size = cork->gso_size;
Signature · 1 line
2.17.0.441.gb46fe60e1d-goog
author maintainer
#2 lore raw

PATCH v2 2/2 gso: limit udp gso to egress-only virtual devices

In reply to Willem de Bruijn

From: Willem de Bruijn <[email protected]>

Until the udp receive stack supports large packets (UDP GRO), GSO packets must not loop from the egress to the ingress path.

Revert the change that added NETIF_F_GSO_UDP_L4 to various virtual devices through NETIF_F_GSO_ENCAP_ALL as this included devices that may loop packets, such as veth and macvlan.

Instead add it to specific devices that forward to another device's egress path, bonding and team.

Fixes 83aa025f535f ("udp: add gso support to virtual devices")
CC Alexander Duyck <[email protected]>
---
 drivers/net/bonding/bond_main.c | 5 +++--
 drivers/net/team/team.c         | 5 +++--
 include/linux/netdev_features.h | 1 -
 3 files changed, 6 insertions(+), 5 deletions(-)
drivers/net/bonding/bond_main.c +3 -2
index 06efdf6a762b..fea17b92b1ae 100644
@@ -1107,7 +1107,8 @@ static void bond_compute_features(struct bonding *bond)
 
 done:
 	bond_dev->vlan_features = vlan_features;
-	bond_dev->hw_enc_features = enc_features | NETIF_F_GSO_ENCAP_ALL;
+	bond_dev->hw_enc_features = enc_features | NETIF_F_GSO_ENCAP_ALL |
+				    NETIF_F_GSO_UDP_L4;
 	bond_dev->gso_max_segs = gso_max_segs;
 	netif_set_gso_max_size(bond_dev, gso_max_size);
 
@@ -4268,7 +4269,7 @@ void bond_setup(struct net_device *bond_dev)
 				NETIF_F_HW_VLAN_CTAG_RX |
 				NETIF_F_HW_VLAN_CTAG_FILTER;
 
-	bond_dev->hw_features |= NETIF_F_GSO_ENCAP_ALL;
+	bond_dev->hw_features |= NETIF_F_GSO_ENCAP_ALL | NETIF_F_GSO_UDP_L4;
 	bond_dev->features |= bond_dev->hw_features;
 }
 
drivers/net/team/team.c +3 -2
index 9dbd390ace34..d6ff881165d0 100644
@@ -1026,7 +1026,8 @@ static void __team_compute_features(struct team *team)
 	}
 
 	team->dev->vlan_features = vlan_features;
-	team->dev->hw_enc_features = enc_features | NETIF_F_GSO_ENCAP_ALL;
+	team->dev->hw_enc_features = enc_features | NETIF_F_GSO_ENCAP_ALL |
+				     NETIF_F_GSO_UDP_L4;
 	team->dev->hard_header_len = max_hard_header_len;
 
 	team->dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
@@ -2117,7 +2118,7 @@ static void team_setup(struct net_device *dev)
 			   NETIF_F_HW_VLAN_CTAG_RX |
 			   NETIF_F_HW_VLAN_CTAG_FILTER;
 
-	dev->hw_features |= NETIF_F_GSO_ENCAP_ALL;
+	dev->hw_features |= NETIF_F_GSO_ENCAP_ALL | NETIF_F_GSO_UDP_L4;
 	dev->features |= dev->hw_features;
 }
 
include/linux/netdev_features.h +0 -1
index c87c3a3453c1..623bb8ced060 100644
@@ -220,7 +220,6 @@ enum {
 				 NETIF_F_GSO_GRE_CSUM |			\
 				 NETIF_F_GSO_IPXIP4 |			\
 				 NETIF_F_GSO_IPXIP6 |			\
-				 NETIF_F_GSO_UDP_L4 |			\
 				 NETIF_F_GSO_UDP_TUNNEL |		\
 				 NETIF_F_GSO_UDP_TUNNEL_CSUM)
 
Signature · 1 line
2.17.0.441.gb46fe60e1d-goog
maintainer
#3 lore raw
In reply to Willem de Bruijn

From: Willem de Bruijn <[email protected]>
Date: Tue, 22 May 2018 11:34:38 -0400

Quoted 6 lines

From: Willem de Bruijn <[email protected]>

A few small fixes:
- disallow segmentation with XFRM
- do not leak gso packets into the ingress path

Quoted 8 lines

Changes
v1 -> v2
- fix build failure in team.c
- drop scatter-gather fix:
this is now fixed by commit 113f99c33585 ("net: test tailroom before appending to linear skb"). After this patch gso skbs are built non-linear regardless of NETIF_F_SG and skb_segment builds linear segs.

Series applied, thanks.