"net: Add support for hardware-offloaded encapsulation" introduced
the encapsulation field of struct skb, which when set provides hints
that GSO should handle an skb that encapsulates a packet.
This patch adds an encapsulation_features field which provides
a hint to dev dev_hard_start_xmit() that harware-offload encapsulation
features should be used. Previously this hint was provided by the
encapsulation field.
The other uses of the encapsulation field are left unchanged.
The two in-tree locations that set the encapsulation have been updated to
also set encapsulation_field.
The motivation for this is to provide support segmentation of GSO MPLS skbs.
This may be necessary when a non-MPLS GSO skb is turned into an MPLS GSO
skb by Open vSwtich and its MPLS push action.
In this case it harware-offload encapsulation features should be used,
actually to be more exact software segmentation should be selected, but
other hints provided by the encapsulation field are not applicable.
Cc: Joseph Gasparakis <joseph.gaspara...@intel.com>
Cc: Peter P Waskiewicz Jr <peter.p.waskiewicz...@intel.com>
Cc: Alexander Duyck <alexander.h.du...@intel.com>
Signed-off-by: Simon Horman <ho...@verge.net.au>
---
include/linux/skbuff.h | 9 ++++++++-
net/core/dev.c | 2 +-
net/core/skbuff.c | 1 +
net/ipv4/gre.c | 1 +
net/ipv4/ipip.c | 1 +
5 files changed, 12 insertions(+), 2 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index e27d1c7..42e61c1 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -492,7 +492,14 @@ struct sk_buff {
* headers if needed
*/
__u8 encapsulation:1;
- /* 7/9 bit hole (depending on ndisc_nodetype presence) */
+ /* Encapsulation protocol and NIC drivers should use
+ * this flag to indicate to each other if the skb contains
+ * encapsulated packet and GSO should use encapsulation features
+ * instead of standard features for the netdev. This is typically
+ * a subset of cases where skb->encapsulation is set.
+ */
+ __u8 encapsulation_features:1;
+ /* 6/8 bit hole (depending on ndisc_nodetype presence) */
kmemcheck_bitfield_end(flags2);
#ifdef CONFIG_NET_DMA
diff --git a/net/core/dev.c b/net/core/dev.c
index 3655ff9..ca81ac8 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2494,7 +2494,7 @@ int dev_hard_start_xmit(struct sk_buff *skb, struct
net_device *dev,
* hardware encapsulation features instead of standard
* features for the netdev
*/
- if (skb->encapsulation)
+ if (skb->encapsulation_features)
features &= dev->hw_enc_features;
if (netif_needs_gso(skb, features)) {
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index ba64614..7504371 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -680,6 +680,7 @@ static void __copy_skb_header(struct sk_buff *new, const
struct sk_buff *old)
new->l4_rxhash = old->l4_rxhash;
new->no_fcs = old->no_fcs;
new->encapsulation = old->encapsulation;
+ new->encapsulation_features = old->encapsulation_features;
#ifdef CONFIG_XFRM
new->sp = secpath_get(old->sp);
#endif
diff --git a/net/ipv4/gre.c b/net/ipv4/gre.c
index 0ae998b..8420f29 100644
--- a/net/ipv4/gre.c
+++ b/net/ipv4/gre.c
@@ -157,6 +157,7 @@ static struct sk_buff *gre_gso_segment(struct sk_buff *skb,
}
skb->encapsulation = 0;
+ skb->encapsulation_features = 0;
if (unlikely(!pskb_may_pull(skb, ghl)))
goto out;
diff --git a/net/ipv4/ipip.c b/net/ipv4/ipip.c
index 77bfcce..a6db3c0 100644
--- a/net/ipv4/ipip.c
+++ b/net/ipv4/ipip.c
@@ -220,6 +220,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb,
struct net_device *dev)
if (likely(!skb->encapsulation)) {
skb_reset_inner_headers(skb);
skb->encapsulation = 1;
+ skb->encapsulation_features = 1;
}
ip_tunnel_xmit(skb, dev, tiph);
--
1.7.10.4
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
