[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <0699747bc9bd7aaf7dc87efd33aa6b95de7d793e.1689677201.git.jbenc@redhat.com>
Date: Tue, 18 Jul 2023 12:50:13 +0200
From: Jiri Benc <jbenc@...hat.com>
To: netdev@...r.kernel.org
Cc: Jesse Brandeburg <jesse.brandeburg@...el.com>,
Tony Nguyen <anthony.l.nguyen@...el.com>
Subject: [PATCH net] vxlan: calculate correct header length for GPE
VXLAN-GPE does not add an extra inner Ethernet header. Take that into
account when calculating header length.
This causes problems in skb_tunnel_check_pmtu, where incorrect PMTU is
cached. If the VXLAN-GPE interface has MTU 1464 set (with the underlying
interface having the usual MTU of 1500), a TCP stream sent over the
tunnel is first segmented to 1514 byte frames only to be immediatelly
followed by a resend with 1500 bytes frames, before the other side even
has a chance to ack them.
Fixes: e1e5314de08ba ("vxlan: implement GPE")
Signed-off-by: Jiri Benc <jbenc@...hat.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 2 +-
drivers/net/vxlan/vxlan_core.c | 23 ++++++++-----------
include/net/vxlan.h | 13 +++++++----
3 files changed, 20 insertions(+), 18 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 1726297f2e0d..8eb9839a3ca6 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -8479,7 +8479,7 @@ static void ixgbe_atr(struct ixgbe_ring *ring,
struct ixgbe_adapter *adapter = q_vector->adapter;
if (unlikely(skb_tail_pointer(skb) < hdr.network +
- VXLAN_HEADROOM))
+ vxlan_headroom(0)))
return;
/* verify the port is recognized as VXLAN */
diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 78744549c1b3..42be8a26c171 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2516,7 +2516,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
}
ndst = &rt->dst;
- err = skb_tunnel_check_pmtu(skb, ndst, VXLAN_HEADROOM,
+ err = skb_tunnel_check_pmtu(skb, ndst, vxlan_headroom(flags & VXLAN_F_GPE),
netif_is_any_bridge_port(dev));
if (err < 0) {
goto tx_error;
@@ -2577,7 +2577,8 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
goto out_unlock;
}
- err = skb_tunnel_check_pmtu(skb, ndst, VXLAN6_HEADROOM,
+ err = skb_tunnel_check_pmtu(skb, ndst,
+ vxlan_headroom((flags & VXLAN_F_GPE) | VXLAN_F_IPV6),
netif_is_any_bridge_port(dev));
if (err < 0) {
goto tx_error;
@@ -2989,14 +2990,12 @@ static int vxlan_change_mtu(struct net_device *dev, int new_mtu)
struct vxlan_rdst *dst = &vxlan->default_dst;
struct net_device *lowerdev = __dev_get_by_index(vxlan->net,
dst->remote_ifindex);
- bool use_ipv6 = !!(vxlan->cfg.flags & VXLAN_F_IPV6);
/* This check is different than dev->max_mtu, because it looks at
* the lowerdev->mtu, rather than the static dev->max_mtu
*/
if (lowerdev) {
- int max_mtu = lowerdev->mtu -
- (use_ipv6 ? VXLAN6_HEADROOM : VXLAN_HEADROOM);
+ int max_mtu = lowerdev->mtu - vxlan_headroom(vxlan->cfg.flags);
if (new_mtu > max_mtu)
return -EINVAL;
}
@@ -3644,11 +3643,11 @@ static void vxlan_config_apply(struct net_device *dev,
struct vxlan_dev *vxlan = netdev_priv(dev);
struct vxlan_rdst *dst = &vxlan->default_dst;
unsigned short needed_headroom = ETH_HLEN;
- bool use_ipv6 = !!(conf->flags & VXLAN_F_IPV6);
int max_mtu = ETH_MAX_MTU;
+ u32 flags = conf->flags;
if (!changelink) {
- if (conf->flags & VXLAN_F_GPE)
+ if (flags & VXLAN_F_GPE)
vxlan_raw_setup(dev);
else
vxlan_ether_setup(dev);
@@ -3673,8 +3672,7 @@ static void vxlan_config_apply(struct net_device *dev,
dev->needed_tailroom = lowerdev->needed_tailroom;
- max_mtu = lowerdev->mtu - (use_ipv6 ? VXLAN6_HEADROOM :
- VXLAN_HEADROOM);
+ max_mtu = lowerdev->mtu - vxlan_headroom(flags);
if (max_mtu < ETH_MIN_MTU)
max_mtu = ETH_MIN_MTU;
@@ -3685,10 +3683,9 @@ static void vxlan_config_apply(struct net_device *dev,
if (dev->mtu > max_mtu)
dev->mtu = max_mtu;
- if (use_ipv6 || conf->flags & VXLAN_F_COLLECT_METADATA)
- needed_headroom += VXLAN6_HEADROOM;
- else
- needed_headroom += VXLAN_HEADROOM;
+ if (flags & VXLAN_F_COLLECT_METADATA)
+ flags |= VXLAN_F_IPV6;
+ needed_headroom += vxlan_headroom(flags);
dev->needed_headroom = needed_headroom;
memcpy(&vxlan->cfg, conf, sizeof(*conf));
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index 0be91ca78d3a..1648240c9668 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -386,10 +386,15 @@ static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
return features;
}
-/* IP header + UDP + VXLAN + Ethernet header */
-#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
-/* IPv6 header + UDP + VXLAN + Ethernet header */
-#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
+static inline int vxlan_headroom(u32 flags)
+{
+ /* VXLAN: IP4/6 header + UDP + VXLAN + Ethernet header */
+ /* VXLAN-GPE: IP4/6 header + UDP + VXLAN */
+ return (flags & VXLAN_F_IPV6 ? sizeof(struct ipv6hdr) :
+ sizeof(struct iphdr)) +
+ sizeof(struct udphdr) + sizeof(struct vxlanhdr) +
+ (flags & VXLAN_F_GPE ? 0 : ETH_HLEN);
+}
static inline struct vxlanhdr *vxlan_hdr(struct sk_buff *skb)
{
--
2.39.2
Powered by blists - more mailing lists