[GIT net-next] Open vSwitch

[GIT net-next] Open vSwitch

[Jesse Gross]

Open vSwitch changes for net-next/3.14. Highlights are:
 * Performance improvements in the mechanism to get packets to userspace
   using memory mapped netlink and skb zero copy where appropriate.
 * Per-cpu flow stats in situations where flows are likely to be shared
   across CPUs. Standard flow stats are used in other situations to save
   memory and allocation time.
 * A handful of code cleanups and rationalization.

The following changes since commit 6ce4eac1f600b34f2f7f58f9cd8f0503d79e42ae:

  Linux 3.13-rc1 (2013-11-22 11:30:55 -0800)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 443cd88c8a31379e95326428bbbd40af25c1d440:

  ovs: make functions local (2014-01-06 15:54:39 -0800)

----------------------------------------------------------------
Andy Zhou (1):
      openvswitch: Change ovs_flow_tbl_lookup_xx() APIs

Ben Pfaff (2):
      openvswitch: Correct comment.
      openvswitch: Shrink sw_flow_mask by 8 bytes (64-bit) or 4 bytes (32-bit).

Daniel Borkmann (1):
      net: ovs: use kfree_rcu instead of rcu_free_{sw_flow_mask_cb,acts_callback}

Jesse Gross (1):
      openvswitch: Silence RCU lockdep checks from flow lookup.

Pravin B Shelar (1):
      openvswitch: Per cpu flow stats.

Stephen Hemminger (1):
      ovs: make functions local

Thomas Graf (9):
      genl: Add genlmsg_new_unicast() for unicast message allocation
      netlink: Avoid netlink mmap alloc if msg size exceeds frame size
      openvswitch: Enable memory mapped Netlink i/o
      net: Export skb_zerocopy() to zerocopy from one skb to another
      openvswitch: Allow user space to announce ability to accept unaligned Netlink messages
      openvswitch: Drop user features if old user space attempted to create datapath
      openvswitch: Pass datapath into userspace queue functions
      openvswitch: Use skb_zerocopy() for upcall
      openvswitch: Compute checksum in skb_gso_segment() if needed

Wei Yongjun (1):
      openvswitch: remove duplicated include from flow_table.c

 include/linux/skbuff.h               |   3 +
 include/net/genetlink.h              |   4 +
 include/uapi/linux/openvswitch.h     |  14 ++-
 net/core/skbuff.c                    |  85 +++++++++++++
 net/netfilter/nfnetlink_queue_core.c |  59 +--------
 net/netlink/af_netlink.c             |   4 +
 net/netlink/genetlink.c              |  21 ++++
 net/openvswitch/datapath.c           | 231 +++++++++++++++++++----------------
 net/openvswitch/datapath.h           |   6 +-
 net/openvswitch/flow.c               |  96 +++++++++++++--
 net/openvswitch/flow.h               |  33 +++--
 net/openvswitch/flow_netlink.c       |  66 ++++++++--
 net/openvswitch/flow_netlink.h       |   1 +
 net/openvswitch/flow_table.c         |  60 ++++++---
 net/openvswitch/flow_table.h         |   6 +-
 net/openvswitch/vport.c              |   6 +-
 net/openvswitch/vport.h              |   1 -
 17 files changed, 483 insertions(+), 213 deletions(-)

[PATCH net-next 01/17] openvswitch: Correct comment.

[Jesse Gross]

From: Ben Pfaff <blp-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>

Signed-off-by: Ben Pfaff <blp-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/flow_table.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index e425427..f96ebd5 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -514,11 +514,7 @@ static struct sw_flow_mask *flow_mask_find(const struct flow_table *tbl,
 	return NULL;
 }
 
-/**
- * add a new mask into the mask list.
- * The caller needs to make sure that 'mask' is not the same
- * as any masks that are already on the list.
- */
+/* Add 'mask' into the mask list, if it is not already there. */
 static int flow_mask_insert(struct flow_table *tbl, struct sw_flow *flow,
 			    struct sw_flow_mask *new)
 {
-- 
1.8.3.2

[PATCH net-next 02/17] openvswitch: Shrink sw_flow_mask by 8 bytes (64-bit) or 4 bytes (32-bit).

[Jesse Gross]

From: Ben Pfaff <blp-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>

We won't normally have a ton of flow masks but using a size_t to store
values no bigger than sizeof(struct sw_flow_key) seems excessive.

This reduces sw_flow_key_range and sw_flow_mask by 4 bytes on 32-bit
systems.  On 64-bit systems it shrinks sw_flow_key_range by 12 bytes but
sw_flow_mask only by 8 bytes due to padding.

Compile tested only.

Signed-off-by: Ben Pfaff <blp-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Acked-by: Andy Zhou <azhou-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/flow.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 1510f51..176406d 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -122,8 +122,8 @@ struct sw_flow_key {
 } __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
 
 struct sw_flow_key_range {
-	size_t start;
-	size_t end;
+	unsigned short int start;
+	unsigned short int end;
 };
 
 struct sw_flow_mask {
-- 
1.8.3.2

Re: [PATCH net-next 02/17] openvswitch: Shrink sw_flow_mask by 8 bytes (64-bit) or 4 bytes (32-bit).

[Sergei Shtylyov]

On 07.01.2014 4:16, Jesse Gross wrote:

> From: Ben Pfaff <blp@nicira.com>

> We won't normally have a ton of flow masks but using a size_t to store
> values no bigger than sizeof(struct sw_flow_key) seems excessive.

> This reduces sw_flow_key_range and sw_flow_mask by 4 bytes on 32-bit
> systems.  On 64-bit systems it shrinks sw_flow_key_range by 12 bytes but
> sw_flow_mask only by 8 bytes due to padding.

> Compile tested only.

> Signed-off-by: Ben Pfaff <blp@nicira.com>
> Acked-by: Andy Zhou <azhou@nicira.com>
> Signed-off-by: Jesse Gross <jesse@nicira.com>
> ---
>   net/openvswitch/flow.h | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)

> diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
> index 1510f51..176406d 100644
> --- a/net/openvswitch/flow.h
> +++ b/net/openvswitch/flow.h
> @@ -122,8 +122,8 @@ struct sw_flow_key {
>   } __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
>
>   struct sw_flow_key_range {
> -	size_t start;
> -	size_t end;
> +	unsigned short int start;
> +	unsigned short int end;

    *short int* seems somewhat ambiguous, no?

WBR, Sergei

[PATCH net-next 03/17] openvswitch: Change ovs_flow_tbl_lookup_xx() APIs

[Jesse Gross]

From: Andy Zhou <azhou-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>

API changes only for code readability. No functional chnages.

This patch removes the underscored version. Added a new API
ovs_flow_tbl_lookup_stats() that returns the n_mask_hits.

Reported by: Ben Pfaff <blp-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Reviewed-by: Thomas Graf <tgraf-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Andy Zhou <azhou-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c   | 16 ++++------------
 net/openvswitch/flow_table.c | 10 +++++++++-
 net/openvswitch/flow_table.h |  4 +++-
 3 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 6f5e1dd..fcaed98 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -234,7 +234,7 @@ void ovs_dp_process_received_packet(struct vport *p, struct sk_buff *skb)
 	}
 
 	/* Look up flow. */
-	flow = ovs_flow_tbl_lookup(&dp->table, &key, &n_mask_hit);
+	flow = ovs_flow_tbl_lookup_stats(&dp->table, &key, &n_mask_hit);
 	if (unlikely(!flow)) {
 		struct dp_upcall_info upcall;
 
@@ -751,14 +751,6 @@ static struct sk_buff *ovs_flow_cmd_build_info(struct sw_flow *flow,
 	return skb;
 }
 
-static struct sw_flow *__ovs_flow_tbl_lookup(struct flow_table *tbl,
-					      const struct sw_flow_key *key)
-{
-	u32 __always_unused n_mask_hit;
-
-	return ovs_flow_tbl_lookup(tbl, key, &n_mask_hit);
-}
-
 static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 {
 	struct nlattr **a = info->attrs;
@@ -809,7 +801,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 		goto err_unlock_ovs;
 
 	/* Check if this is a duplicate flow */
-	flow = __ovs_flow_tbl_lookup(&dp->table, &key);
+	flow = ovs_flow_tbl_lookup(&dp->table, &key);
 	if (!flow) {
 		/* Bail out if we're not allowed to create a new flow. */
 		error = -ENOENT;
@@ -921,7 +913,7 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
 		goto unlock;
 	}
 
-	flow = __ovs_flow_tbl_lookup(&dp->table, &key);
+	flow = ovs_flow_tbl_lookup(&dp->table, &key);
 	if (!flow || !ovs_flow_cmp_unmasked_key(flow, &match)) {
 		err = -ENOENT;
 		goto unlock;
@@ -969,7 +961,7 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
 	if (err)
 		goto unlock;
 
-	flow = __ovs_flow_tbl_lookup(&dp->table, &key);
+	flow = ovs_flow_tbl_lookup(&dp->table, &key);
 	if (!flow || !ovs_flow_cmp_unmasked_key(flow, &match)) {
 		err = -ENOENT;
 		goto unlock;
diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index f96ebd5..261a54e 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -429,7 +429,7 @@ static struct sw_flow *masked_flow_lookup(struct table_instance *ti,
 	return NULL;
 }
 
-struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *tbl,
+struct sw_flow *ovs_flow_tbl_lookup_stats(struct flow_table *tbl,
 				    const struct sw_flow_key *key,
 				    u32 *n_mask_hit)
 {
@@ -447,6 +447,14 @@ struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *tbl,
 	return NULL;
 }
 
+struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *tbl,
+				    const struct sw_flow_key *key)
+{
+	u32 __always_unused n_mask_hit;
+
+	return ovs_flow_tbl_lookup_stats(tbl, key, &n_mask_hit);
+}
+
 int ovs_flow_tbl_num_masks(const struct flow_table *table)
 {
 	struct sw_flow_mask *mask;
diff --git a/net/openvswitch/flow_table.h b/net/openvswitch/flow_table.h
index fbe45d5..f54aa82 100644
--- a/net/openvswitch/flow_table.h
+++ b/net/openvswitch/flow_table.h
@@ -69,9 +69,11 @@ void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow);
 int  ovs_flow_tbl_num_masks(const struct flow_table *table);
 struct sw_flow *ovs_flow_tbl_dump_next(struct table_instance *table,
 				       u32 *bucket, u32 *idx);
-struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *,
+struct sw_flow *ovs_flow_tbl_lookup_stats(struct flow_table *,
 				    const struct sw_flow_key *,
 				    u32 *n_mask_hit);
+struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *,
+				    const struct sw_flow_key *);
 
 bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
 			       struct sw_flow_match *match);
-- 
1.8.3.2

[PATCH net-next 04/17] openvswitch: Silence RCU lockdep checks from flow lookup.

[Jesse Gross]

Flow lookup can happen either in packet processing context or userspace
context but it was annotated as requiring RCU read lock to be held. This
also allows OVS mutex to be held without causing warnings.

Reported-by: Justin Pettit <jpettit-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Reviewed-by: Thomas Graf <tgraf-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c   | 3 +--
 net/openvswitch/datapath.h   | 2 ++
 net/openvswitch/flow_table.c | 2 +-
 3 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index fcaed98..0727aaa 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -701,8 +701,7 @@ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp,
 	if (start) {
 		const struct sw_flow_actions *sf_acts;
 
-		sf_acts = rcu_dereference_check(flow->sf_acts,
-						lockdep_ovsl_is_held());
+		sf_acts = rcu_dereference_ovsl(flow->sf_acts);
 
 		err = ovs_nla_put_actions(sf_acts->actions,
 					  sf_acts->actions_len, skb);
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index 4067ea4..ba13be4 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -145,6 +145,8 @@ int lockdep_ovsl_is_held(void);
 #define ASSERT_OVSL()		WARN_ON(unlikely(!lockdep_ovsl_is_held()))
 #define ovsl_dereference(p)					\
 	rcu_dereference_protected(p, lockdep_ovsl_is_held())
+#define rcu_dereference_ovsl(p)					\
+	rcu_dereference_check(p, lockdep_ovsl_is_held())
 
 static inline struct net *ovs_dp_get_net(struct datapath *dp)
 {
diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index 261a54e..7b9cf2c 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -433,7 +433,7 @@ struct sw_flow *ovs_flow_tbl_lookup_stats(struct flow_table *tbl,
 				    const struct sw_flow_key *key,
 				    u32 *n_mask_hit)
 {
-	struct table_instance *ti = rcu_dereference(tbl->ti);
+	struct table_instance *ti = rcu_dereference_ovsl(tbl->ti);
 	struct sw_flow_mask *mask;
 	struct sw_flow *flow;
 
-- 
1.8.3.2

[PATCH net-next 05/17] genl: Add genlmsg_new_unicast() for unicast message allocation

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Allocates a new sk_buff large enough to cover the specified payload
plus required Netlink headers. Will check receiving socket for
memory mapped i/o capability and use it if enabled. Will fall back
to non-mapped skb if message size exceeds the frame size of the ring.

Signed-of-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Reviewed-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 include/net/genetlink.h |  4 ++++
 net/netlink/genetlink.c | 21 +++++++++++++++++++++
 2 files changed, 25 insertions(+)

diff --git a/include/net/genetlink.h b/include/net/genetlink.h
index 1b177ed..93695f0 100644
--- a/include/net/genetlink.h
+++ b/include/net/genetlink.h
@@ -73,6 +73,7 @@ struct genl_family {
  * @attrs: netlink attributes
  * @_net: network namespace
  * @user_ptr: user pointers
+ * @dst_sk: destination socket
  */
 struct genl_info {
 	u32			snd_seq;
@@ -85,6 +86,7 @@ struct genl_info {
 	struct net *		_net;
 #endif
 	void *			user_ptr[2];
+	struct sock *		dst_sk;
 };
 
 static inline struct net *genl_info_net(struct genl_info *info)
@@ -177,6 +179,8 @@ void genl_notify(struct genl_family *family,
 		 struct sk_buff *skb, struct net *net, u32 portid,
 		 u32 group, struct nlmsghdr *nlh, gfp_t flags);
 
+struct sk_buff *genlmsg_new_unicast(size_t payload, struct genl_info *info,
+				    gfp_t flags);
 void *genlmsg_put(struct sk_buff *skb, u32 portid, u32 seq,
 		  struct genl_family *family, int flags, u8 cmd);
 
diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c
index 4518a57..85bf42e 100644
--- a/net/netlink/genetlink.c
+++ b/net/netlink/genetlink.c
@@ -454,6 +454,26 @@ int genl_unregister_family(struct genl_family *family)
 EXPORT_SYMBOL(genl_unregister_family);
 
 /**
+ * genlmsg_new_unicast - Allocate generic netlink message for unicast
+ * @payload: size of the message payload
+ * @info: information on destination
+ * @flags: the type of memory to allocate
+ *
+ * Allocates a new sk_buff large enough to cover the specified payload
+ * plus required Netlink headers. Will check receiving socket for
+ * memory mapped i/o capability and use it if enabled. Will fall back
+ * to non-mapped skb if message size exceeds the frame size of the ring.
+ */
+struct sk_buff *genlmsg_new_unicast(size_t payload, struct genl_info *info,
+				    gfp_t flags)
+{
+	size_t len = nlmsg_total_size(genlmsg_total_size(payload));
+
+	return netlink_alloc_skb(info->dst_sk, len, info->snd_portid, flags);
+}
+EXPORT_SYMBOL_GPL(genlmsg_new_unicast);
+
+/**
  * genlmsg_put - Add generic netlink header to netlink message
  * @skb: socket buffer holding the message
  * @portid: netlink portid the message is addressed to
@@ -593,6 +613,7 @@ static int genl_family_rcv_msg(struct genl_family *family,
 	info.genlhdr = nlmsg_data(nlh);
 	info.userhdr = nlmsg_data(nlh) + GENL_HDRLEN;
 	info.attrs = attrbuf;
+	info.dst_sk = skb->sk;
 	genl_info_net_set(&info, net);
 	memset(&info.user_ptr, 0, sizeof(info.user_ptr));
 
-- 
1.8.3.2

[PATCH net-next 06/17] netlink: Avoid netlink mmap alloc if msg size exceeds frame size

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

An insufficent ring frame size configuration can lead to an
unnecessary skb allocation for every Netlink message. Check frame
size before taking the queue lock and allocating the skb and
re-check with lock to be safe.

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Reviewed-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/netlink/af_netlink.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index bca50b9..6433489 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -1769,6 +1769,9 @@ struct sk_buff *netlink_alloc_skb(struct sock *ssk, unsigned int size,
 	if (ring->pg_vec == NULL)
 		goto out_put;
 
+	if (ring->frame_size - NL_MMAP_HDRLEN < size)
+		goto out_put;
+
 	skb = alloc_skb_head(gfp_mask);
 	if (skb == NULL)
 		goto err1;
@@ -1778,6 +1781,7 @@ struct sk_buff *netlink_alloc_skb(struct sock *ssk, unsigned int size,
 	if (ring->pg_vec == NULL)
 		goto out_free;
 
+	/* check again under lock */
 	maxlen = ring->frame_size - NL_MMAP_HDRLEN;
 	if (maxlen < size)
 		goto out_free;
-- 
1.8.3.2

[PATCH net-next 07/17] openvswitch: Enable memory mapped Netlink i/o

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Use memory mapped Netlink i/o for all unicast openvswitch
communication if a ring has been set up.

Benchmark
  * pktgen -> ovs internal port
  * 5M pkts, 5M flows
  * 4 threads, 8 cores

Before:
Result: OK: 67418743(c67108212+d310530) usec, 5000000 (9000byte,0frags)
  74163pps 5339Mb/sec (5339736000bps) errors: 0
	+   2.98%     ovs-vswitchd  [k] copy_user_generic_string
	+   2.49%     ovs-vswitchd  [k] memcpy
	+   1.84%       kpktgend_2  [k] memcpy
	+   1.81%       kpktgend_1  [k] memcpy
	+   1.81%       kpktgend_3  [k] memcpy
	+   1.78%       kpktgend_0  [k] memcpy

After:
Result: OK: 24229690(c24127165+d102524) usec, 5000000 (9000byte,0frags)
  206358pps 14857Mb/sec (14857776000bps) errors: 0
	+   2.80%     ovs-vswitchd  [k] memcpy
	+   1.31%       kpktgend_2  [k] memcpy
	+   1.23%       kpktgend_0  [k] memcpy
	+   1.09%       kpktgend_1  [k] memcpy
	+   1.04%       kpktgend_3  [k] memcpy
	+   0.96%     ovs-vswitchd  [k] copy_user_generic_string

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Reviewed-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c | 56 ++++++++++++++++++++++++----------------------
 1 file changed, 29 insertions(+), 27 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 0727aaa..5da2534 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -402,6 +402,11 @@ static int queue_userspace_packet(struct net *net, int dp_ifindex,
 	struct sk_buff *nskb = NULL;
 	struct sk_buff *user_skb; /* to be queued to userspace */
 	struct nlattr *nla;
+	struct genl_info info = {
+		.dst_sk = net->genl_sock,
+		.snd_portid = upcall_info->portid,
+	};
+	size_t len;
 	int err;
 
 	if (vlan_tx_tag_present(skb)) {
@@ -422,7 +427,8 @@ static int queue_userspace_packet(struct net *net, int dp_ifindex,
 		goto out;
 	}
 
-	user_skb = genlmsg_new(upcall_msg_size(skb, upcall_info->userdata), GFP_ATOMIC);
+	len = upcall_msg_size(skb, upcall_info->userdata);
+	user_skb = genlmsg_new_unicast(len, &info, GFP_ATOMIC);
 	if (!user_skb) {
 		err = -ENOMEM;
 		goto out;
@@ -725,27 +731,30 @@ error:
 	return err;
 }
 
-static struct sk_buff *ovs_flow_cmd_alloc_info(struct sw_flow *flow)
+static struct sk_buff *ovs_flow_cmd_alloc_info(struct sw_flow *flow,
+					       struct genl_info *info)
 {
-	const struct sw_flow_actions *sf_acts;
+	size_t len;
 
-	sf_acts = ovsl_dereference(flow->sf_acts);
+	len = ovs_flow_cmd_msg_size(ovsl_dereference(flow->sf_acts));
 
-	return genlmsg_new(ovs_flow_cmd_msg_size(sf_acts), GFP_KERNEL);
+	return genlmsg_new_unicast(len, info, GFP_KERNEL);
 }
 
 static struct sk_buff *ovs_flow_cmd_build_info(struct sw_flow *flow,
 					       struct datapath *dp,
-					       u32 portid, u32 seq, u8 cmd)
+					       struct genl_info *info,
+					       u8 cmd)
 {
 	struct sk_buff *skb;
 	int retval;
 
-	skb = ovs_flow_cmd_alloc_info(flow);
+	skb = ovs_flow_cmd_alloc_info(flow, info);
 	if (!skb)
 		return ERR_PTR(-ENOMEM);
 
-	retval = ovs_flow_cmd_fill_info(flow, dp, skb, portid, seq, 0, cmd);
+	retval = ovs_flow_cmd_fill_info(flow, dp, skb, info->snd_portid,
+					info->snd_seq, 0, cmd);
 	BUG_ON(retval < 0);
 	return skb;
 }
@@ -826,8 +835,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 			goto err_flow_free;
 		}
 
-		reply = ovs_flow_cmd_build_info(flow, dp, info->snd_portid,
-						info->snd_seq, OVS_FLOW_CMD_NEW);
+		reply = ovs_flow_cmd_build_info(flow, dp, info, OVS_FLOW_CMD_NEW);
 	} else {
 		/* We found a matching flow. */
 		struct sw_flow_actions *old_acts;
@@ -855,8 +863,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 		rcu_assign_pointer(flow->sf_acts, acts);
 		ovs_nla_free_flow_actions(old_acts);
 
-		reply = ovs_flow_cmd_build_info(flow, dp, info->snd_portid,
-					       info->snd_seq, OVS_FLOW_CMD_NEW);
+		reply = ovs_flow_cmd_build_info(flow, dp, info, OVS_FLOW_CMD_NEW);
 
 		/* Clear stats. */
 		if (a[OVS_FLOW_ATTR_CLEAR]) {
@@ -918,8 +925,7 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
 		goto unlock;
 	}
 
-	reply = ovs_flow_cmd_build_info(flow, dp, info->snd_portid,
-					info->snd_seq, OVS_FLOW_CMD_NEW);
+	reply = ovs_flow_cmd_build_info(flow, dp, info, OVS_FLOW_CMD_NEW);
 	if (IS_ERR(reply)) {
 		err = PTR_ERR(reply);
 		goto unlock;
@@ -966,7 +972,7 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
 		goto unlock;
 	}
 
-	reply = ovs_flow_cmd_alloc_info(flow);
+	reply = ovs_flow_cmd_alloc_info(flow, info);
 	if (!reply) {
 		err = -ENOMEM;
 		goto unlock;
@@ -1118,17 +1124,17 @@ error:
 	return -EMSGSIZE;
 }
 
-static struct sk_buff *ovs_dp_cmd_build_info(struct datapath *dp, u32 portid,
-					     u32 seq, u8 cmd)
+static struct sk_buff *ovs_dp_cmd_build_info(struct datapath *dp,
+					     struct genl_info *info, u8 cmd)
 {
 	struct sk_buff *skb;
 	int retval;
 
-	skb = genlmsg_new(ovs_dp_cmd_msg_size(), GFP_KERNEL);
+	skb = genlmsg_new_unicast(ovs_dp_cmd_msg_size(), info, GFP_KERNEL);
 	if (!skb)
 		return ERR_PTR(-ENOMEM);
 
-	retval = ovs_dp_cmd_fill_info(dp, skb, portid, seq, 0, cmd);
+	retval = ovs_dp_cmd_fill_info(dp, skb, info->snd_portid, info->snd_seq, 0, cmd);
 	if (retval < 0) {
 		kfree_skb(skb);
 		return ERR_PTR(retval);
@@ -1223,8 +1229,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
 		goto err_destroy_ports_array;
 	}
 
-	reply = ovs_dp_cmd_build_info(dp, info->snd_portid,
-				      info->snd_seq, OVS_DP_CMD_NEW);
+	reply = ovs_dp_cmd_build_info(dp, info, OVS_DP_CMD_NEW);
 	err = PTR_ERR(reply);
 	if (IS_ERR(reply))
 		goto err_destroy_local_port;
@@ -1290,8 +1295,7 @@ static int ovs_dp_cmd_del(struct sk_buff *skb, struct genl_info *info)
 	if (IS_ERR(dp))
 		goto unlock;
 
-	reply = ovs_dp_cmd_build_info(dp, info->snd_portid,
-				      info->snd_seq, OVS_DP_CMD_DEL);
+	reply = ovs_dp_cmd_build_info(dp, info, OVS_DP_CMD_DEL);
 	err = PTR_ERR(reply);
 	if (IS_ERR(reply))
 		goto unlock;
@@ -1319,8 +1323,7 @@ static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info)
 	if (IS_ERR(dp))
 		goto unlock;
 
-	reply = ovs_dp_cmd_build_info(dp, info->snd_portid,
-				      info->snd_seq, OVS_DP_CMD_NEW);
+	reply = ovs_dp_cmd_build_info(dp, info, OVS_DP_CMD_NEW);
 	if (IS_ERR(reply)) {
 		err = PTR_ERR(reply);
 		genl_set_err(&dp_datapath_genl_family, sock_net(skb->sk), 0,
@@ -1351,8 +1354,7 @@ static int ovs_dp_cmd_get(struct sk_buff *skb, struct genl_info *info)
 		goto unlock;
 	}
 
-	reply = ovs_dp_cmd_build_info(dp, info->snd_portid,
-				      info->snd_seq, OVS_DP_CMD_NEW);
+	reply = ovs_dp_cmd_build_info(dp, info, OVS_DP_CMD_NEW);
 	if (IS_ERR(reply)) {
 		err = PTR_ERR(reply);
 		goto unlock;
-- 
1.8.3.2

[PATCH net-next 08/17] openvswitch: Per cpu flow stats.

[Jesse Gross]

From: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>

With mega flow implementation ovs flow can be shared between
multiple CPUs which makes stats updates highly contended
operation. This patch uses per-CPU stats in cases where a flow
is likely to be shared (if there is a wildcard in the 5-tuple
and therefore likely to be spread by RSS). In other situations,
it uses the current strategy, saving memory and allocation time.

Signed-off-by: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c     | 50 +++++++---------------
 net/openvswitch/flow.c         | 96 +++++++++++++++++++++++++++++++++++++++---
 net/openvswitch/flow.h         | 29 ++++++++++---
 net/openvswitch/flow_netlink.c | 56 ++++++++++++++++++++++--
 net/openvswitch/flow_netlink.h |  1 +
 net/openvswitch/flow_table.c   | 31 +++++++++++++-
 net/openvswitch/flow_table.h   |  2 +-
 7 files changed, 210 insertions(+), 55 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 5da2534..50d7782 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -251,9 +251,9 @@ void ovs_dp_process_received_packet(struct vport *p, struct sk_buff *skb)
 	OVS_CB(skb)->flow = flow;
 	OVS_CB(skb)->pkt_key = &key;
 
-	stats_counter = &stats->n_hit;
-	ovs_flow_used(OVS_CB(skb)->flow, skb);
+	ovs_flow_stats_update(OVS_CB(skb)->flow, skb);
 	ovs_execute_actions(dp, skb);
+	stats_counter = &stats->n_hit;
 
 out:
 	/* Update datapath statistics. */
@@ -459,14 +459,6 @@ out:
 	return err;
 }
 
-static void clear_stats(struct sw_flow *flow)
-{
-	flow->used = 0;
-	flow->tcp_flags = 0;
-	flow->packet_count = 0;
-	flow->byte_count = 0;
-}
-
 static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
 {
 	struct ovs_header *ovs_header = info->userhdr;
@@ -505,7 +497,7 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
 		packet->protocol = htons(ETH_P_802_2);
 
 	/* Build an sw_flow for sending this packet. */
-	flow = ovs_flow_alloc();
+	flow = ovs_flow_alloc(false);
 	err = PTR_ERR(flow);
 	if (IS_ERR(flow))
 		goto err_kfree_skb;
@@ -641,10 +633,10 @@ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp,
 	const int skb_orig_len = skb->len;
 	struct nlattr *start;
 	struct ovs_flow_stats stats;
+	__be16 tcp_flags;
+	unsigned long used;
 	struct ovs_header *ovs_header;
 	struct nlattr *nla;
-	unsigned long used;
-	u8 tcp_flags;
 	int err;
 
 	ovs_header = genlmsg_put(skb, portid, seq, &dp_flow_genl_family, flags, cmd);
@@ -673,24 +665,17 @@ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp,
 
 	nla_nest_end(skb, nla);
 
-	spin_lock_bh(&flow->lock);
-	used = flow->used;
-	stats.n_packets = flow->packet_count;
-	stats.n_bytes = flow->byte_count;
-	tcp_flags = (u8)ntohs(flow->tcp_flags);
-	spin_unlock_bh(&flow->lock);
-
+	ovs_flow_stats_get(flow, &stats, &used, &tcp_flags);
 	if (used &&
 	    nla_put_u64(skb, OVS_FLOW_ATTR_USED, ovs_flow_used_time(used)))
 		goto nla_put_failure;
 
 	if (stats.n_packets &&
-	    nla_put(skb, OVS_FLOW_ATTR_STATS,
-		    sizeof(struct ovs_flow_stats), &stats))
+	    nla_put(skb, OVS_FLOW_ATTR_STATS, sizeof(struct ovs_flow_stats), &stats))
 		goto nla_put_failure;
 
-	if (tcp_flags &&
-	    nla_put_u8(skb, OVS_FLOW_ATTR_TCP_FLAGS, tcp_flags))
+	if ((u8)ntohs(tcp_flags) &&
+	     nla_put_u8(skb, OVS_FLOW_ATTR_TCP_FLAGS, (u8)ntohs(tcp_flags)))
 		goto nla_put_failure;
 
 	/* If OVS_FLOW_ATTR_ACTIONS doesn't fit, skip dumping the actions if
@@ -770,6 +755,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 	struct datapath *dp;
 	struct sw_flow_actions *acts = NULL;
 	struct sw_flow_match match;
+	bool exact_5tuple;
 	int error;
 
 	/* Extract key. */
@@ -778,7 +764,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 		goto error;
 
 	ovs_match_init(&match, &key, &mask);
-	error = ovs_nla_get_match(&match,
+	error = ovs_nla_get_match(&match, &exact_5tuple,
 				  a[OVS_FLOW_ATTR_KEY], a[OVS_FLOW_ATTR_MASK]);
 	if (error)
 		goto error;
@@ -817,12 +803,11 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 			goto err_unlock_ovs;
 
 		/* Allocate flow. */
-		flow = ovs_flow_alloc();
+		flow = ovs_flow_alloc(!exact_5tuple);
 		if (IS_ERR(flow)) {
 			error = PTR_ERR(flow);
 			goto err_unlock_ovs;
 		}
-		clear_stats(flow);
 
 		flow->key = masked_key;
 		flow->unmasked_key = key;
@@ -866,11 +851,8 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
 		reply = ovs_flow_cmd_build_info(flow, dp, info, OVS_FLOW_CMD_NEW);
 
 		/* Clear stats. */
-		if (a[OVS_FLOW_ATTR_CLEAR]) {
-			spin_lock_bh(&flow->lock);
-			clear_stats(flow);
-			spin_unlock_bh(&flow->lock);
-		}
+		if (a[OVS_FLOW_ATTR_CLEAR])
+			ovs_flow_stats_clear(flow);
 	}
 	ovs_unlock();
 
@@ -908,7 +890,7 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
 	}
 
 	ovs_match_init(&match, &key, NULL);
-	err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL);
+	err = ovs_nla_get_match(&match, NULL, a[OVS_FLOW_ATTR_KEY], NULL);
 	if (err)
 		return err;
 
@@ -962,7 +944,7 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
 	}
 
 	ovs_match_init(&match, &key, NULL);
-	err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL);
+	err = ovs_nla_get_match(&match, NULL, a[OVS_FLOW_ATTR_KEY], NULL);
 	if (err)
 		goto unlock;
 
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index b409f52..16f4b46 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -35,6 +35,7 @@
 #include <linux/ip.h>
 #include <linux/ipv6.h>
 #include <linux/sctp.h>
+#include <linux/smp.h>
 #include <linux/tcp.h>
 #include <linux/udp.h>
 #include <linux/icmp.h>
@@ -60,10 +61,16 @@ u64 ovs_flow_used_time(unsigned long flow_jiffies)
 
 #define TCP_FLAGS_BE16(tp) (*(__be16 *)&tcp_flag_word(tp) & htons(0x0FFF))
 
-void ovs_flow_used(struct sw_flow *flow, struct sk_buff *skb)
+void ovs_flow_stats_update(struct sw_flow *flow, struct sk_buff *skb)
 {
+	struct flow_stats *stats;
 	__be16 tcp_flags = 0;
 
+	if (!flow->stats.is_percpu)
+		stats = flow->stats.stat;
+	else
+		stats = this_cpu_ptr(flow->stats.cpu_stats);
+
 	if ((flow->key.eth.type == htons(ETH_P_IP) ||
 	     flow->key.eth.type == htons(ETH_P_IPV6)) &&
 	    flow->key.ip.proto == IPPROTO_TCP &&
@@ -71,12 +78,87 @@ void ovs_flow_used(struct sw_flow *flow, struct sk_buff *skb)
 		tcp_flags = TCP_FLAGS_BE16(tcp_hdr(skb));
 	}
 
-	spin_lock(&flow->lock);
-	flow->used = jiffies;
-	flow->packet_count++;
-	flow->byte_count += skb->len;
-	flow->tcp_flags |= tcp_flags;
-	spin_unlock(&flow->lock);
+	spin_lock(&stats->lock);
+	stats->used = jiffies;
+	stats->packet_count++;
+	stats->byte_count += skb->len;
+	stats->tcp_flags |= tcp_flags;
+	spin_unlock(&stats->lock);
+}
+
+static void stats_read(struct flow_stats *stats,
+		       struct ovs_flow_stats *ovs_stats,
+		       unsigned long *used, __be16 *tcp_flags)
+{
+	spin_lock(&stats->lock);
+	if (time_after(stats->used, *used))
+		*used = stats->used;
+	*tcp_flags |= stats->tcp_flags;
+	ovs_stats->n_packets += stats->packet_count;
+	ovs_stats->n_bytes += stats->byte_count;
+	spin_unlock(&stats->lock);
+}
+
+void ovs_flow_stats_get(struct sw_flow *flow, struct ovs_flow_stats *ovs_stats,
+			unsigned long *used, __be16 *tcp_flags)
+{
+	int cpu, cur_cpu;
+
+	*used = 0;
+	*tcp_flags = 0;
+	memset(ovs_stats, 0, sizeof(*ovs_stats));
+
+	if (!flow->stats.is_percpu) {
+		stats_read(flow->stats.stat, ovs_stats, used, tcp_flags);
+	} else {
+		cur_cpu = get_cpu();
+		for_each_possible_cpu(cpu) {
+			struct flow_stats *stats;
+
+			if (cpu == cur_cpu)
+				local_bh_disable();
+
+			stats = per_cpu_ptr(flow->stats.cpu_stats, cpu);
+			stats_read(stats, ovs_stats, used, tcp_flags);
+
+			if (cpu == cur_cpu)
+				local_bh_enable();
+		}
+		put_cpu();
+	}
+}
+
+static void stats_reset(struct flow_stats *stats)
+{
+	spin_lock(&stats->lock);
+	stats->used = 0;
+	stats->packet_count = 0;
+	stats->byte_count = 0;
+	stats->tcp_flags = 0;
+	spin_unlock(&stats->lock);
+}
+
+void ovs_flow_stats_clear(struct sw_flow *flow)
+{
+	int cpu, cur_cpu;
+
+	if (!flow->stats.is_percpu) {
+		stats_reset(flow->stats.stat);
+	} else {
+		cur_cpu = get_cpu();
+
+		for_each_possible_cpu(cpu) {
+
+			if (cpu == cur_cpu)
+				local_bh_disable();
+
+			stats_reset(per_cpu_ptr(flow->stats.cpu_stats, cpu));
+
+			if (cpu == cur_cpu)
+				local_bh_enable();
+		}
+		put_cpu();
+	}
 }
 
 static int check_header(struct sk_buff *skb, int len)
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 176406d..2d770e2 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -19,6 +19,7 @@
 #ifndef FLOW_H
 #define FLOW_H 1
 
+#include <linux/cache.h>
 #include <linux/kernel.h>
 #include <linux/netlink.h>
 #include <linux/openvswitch.h>
@@ -146,6 +147,22 @@ struct sw_flow_actions {
 	struct nlattr actions[];
 };
 
+struct flow_stats {
+	u64 packet_count;		/* Number of packets matched. */
+	u64 byte_count;			/* Number of bytes matched. */
+	unsigned long used;		/* Last used time (in jiffies). */
+	spinlock_t lock;		/* Lock for atomic stats update. */
+	__be16 tcp_flags;		/* Union of seen TCP flags. */
+};
+
+struct sw_flow_stats {
+	bool is_percpu;
+	union {
+		struct flow_stats *stat;
+		struct flow_stats __percpu *cpu_stats;
+	};
+};
+
 struct sw_flow {
 	struct rcu_head rcu;
 	struct hlist_node hash_node[2];
@@ -155,12 +172,7 @@ struct sw_flow {
 	struct sw_flow_key unmasked_key;
 	struct sw_flow_mask *mask;
 	struct sw_flow_actions __rcu *sf_acts;
-
-	spinlock_t lock;	/* Lock for values below. */
-	unsigned long used;	/* Last used time (in jiffies). */
-	u64 packet_count;	/* Number of packets matched. */
-	u64 byte_count;		/* Number of bytes matched. */
-	__be16 tcp_flags;	/* Union of seen TCP flags. */
+	struct sw_flow_stats stats;
 };
 
 struct arp_eth_header {
@@ -177,7 +189,10 @@ struct arp_eth_header {
 	unsigned char       ar_tip[4];		/* target IP address        */
 } __packed;
 
-void ovs_flow_used(struct sw_flow *, struct sk_buff *);
+void ovs_flow_stats_update(struct sw_flow *flow, struct sk_buff *skb);
+void ovs_flow_stats_get(struct sw_flow *flow, struct ovs_flow_stats *stats,
+			unsigned long *used, __be16 *tcp_flags);
+void ovs_flow_stats_clear(struct sw_flow *flow);
 u64 ovs_flow_used_time(unsigned long flow_jiffies);
 
 int ovs_flow_extract(struct sk_buff *, u16 in_port, struct sw_flow_key *);
diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.c
index 2bc1bc1..3ccb92f 100644
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -266,6 +266,20 @@ static bool is_all_zero(const u8 *fp, size_t size)
 	return true;
 }
 
+static bool is_all_set(const u8 *fp, size_t size)
+{
+	int i;
+
+	if (!fp)
+		return false;
+
+	for (i = 0; i < size; i++)
+		if (fp[i] != 0xff)
+			return false;
+
+	return true;
+}
+
 static int __parse_flow_nlattrs(const struct nlattr *attr,
 				const struct nlattr *a[],
 				u64 *attrsp, bool nz)
@@ -487,8 +501,9 @@ static int metadata_from_nlattrs(struct sw_flow_match *match,  u64 *attrs,
 	return 0;
 }
 
-static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
-				const struct nlattr **a, bool is_mask)
+static int ovs_key_from_nlattrs(struct sw_flow_match *match,  bool *exact_5tuple,
+				u64 attrs, const struct nlattr **a,
+				bool is_mask)
 {
 	int err;
 	u64 orig_attrs = attrs;
@@ -545,6 +560,11 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
 		SW_FLOW_KEY_PUT(match, eth.type, htons(ETH_P_802_2), is_mask);
 	}
 
+	if (is_mask && exact_5tuple) {
+		if (match->mask->key.eth.type != htons(0xffff))
+			*exact_5tuple = false;
+	}
+
 	if (attrs & (1 << OVS_KEY_ATTR_IPV4)) {
 		const struct ovs_key_ipv4 *ipv4_key;
 
@@ -567,6 +587,13 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
 		SW_FLOW_KEY_PUT(match, ipv4.addr.dst,
 				ipv4_key->ipv4_dst, is_mask);
 		attrs &= ~(1 << OVS_KEY_ATTR_IPV4);
+
+		if (is_mask && exact_5tuple && *exact_5tuple) {
+			if (ipv4_key->ipv4_proto != 0xff ||
+			    ipv4_key->ipv4_src != htonl(0xffffffff) ||
+			    ipv4_key->ipv4_dst != htonl(0xffffffff))
+				*exact_5tuple = false;
+		}
 	}
 
 	if (attrs & (1 << OVS_KEY_ATTR_IPV6)) {
@@ -598,6 +625,13 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
 				is_mask);
 
 		attrs &= ~(1 << OVS_KEY_ATTR_IPV6);
+
+		if (is_mask && exact_5tuple && *exact_5tuple) {
+			if (ipv6_key->ipv6_proto != 0xff ||
+			    !is_all_set((u8 *)ipv6_key->ipv6_src, sizeof(match->key->ipv6.addr.src)) ||
+			    !is_all_set((u8 *)ipv6_key->ipv6_dst, sizeof(match->key->ipv6.addr.dst)))
+				*exact_5tuple = false;
+		}
 	}
 
 	if (attrs & (1 << OVS_KEY_ATTR_ARP)) {
@@ -640,6 +674,11 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
 					tcp_key->tcp_dst, is_mask);
 		}
 		attrs &= ~(1 << OVS_KEY_ATTR_TCP);
+
+		if (is_mask && exact_5tuple && *exact_5tuple &&
+		    (tcp_key->tcp_src != htons(0xffff) ||
+		     tcp_key->tcp_dst != htons(0xffff)))
+			*exact_5tuple = false;
 	}
 
 	if (attrs & (1 << OVS_KEY_ATTR_TCP_FLAGS)) {
@@ -671,6 +710,11 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match,  u64 attrs,
 					udp_key->udp_dst, is_mask);
 		}
 		attrs &= ~(1 << OVS_KEY_ATTR_UDP);
+
+		if (is_mask && exact_5tuple && *exact_5tuple &&
+		    (udp_key->udp_src != htons(0xffff) ||
+		     udp_key->udp_dst != htons(0xffff)))
+			*exact_5tuple = false;
 	}
 
 	if (attrs & (1 << OVS_KEY_ATTR_SCTP)) {
@@ -756,6 +800,7 @@ static void sw_flow_mask_set(struct sw_flow_mask *mask,
  * attribute specifies the mask field of the wildcarded flow.
  */
 int ovs_nla_get_match(struct sw_flow_match *match,
+		      bool *exact_5tuple,
 		      const struct nlattr *key,
 		      const struct nlattr *mask)
 {
@@ -803,10 +848,13 @@ int ovs_nla_get_match(struct sw_flow_match *match,
 		}
 	}
 
-	err = ovs_key_from_nlattrs(match, key_attrs, a, false);
+	err = ovs_key_from_nlattrs(match, NULL, key_attrs, a, false);
 	if (err)
 		return err;
 
+	if (exact_5tuple)
+		*exact_5tuple = true;
+
 	if (mask) {
 		err = parse_flow_mask_nlattrs(mask, a, &mask_attrs);
 		if (err)
@@ -844,7 +892,7 @@ int ovs_nla_get_match(struct sw_flow_match *match,
 			}
 		}
 
-		err = ovs_key_from_nlattrs(match, mask_attrs, a, true);
+		err = ovs_key_from_nlattrs(match, exact_5tuple, mask_attrs, a, true);
 		if (err)
 			return err;
 	} else {
diff --git a/net/openvswitch/flow_netlink.h b/net/openvswitch/flow_netlink.h
index 4401510..b31fbe2 100644
--- a/net/openvswitch/flow_netlink.h
+++ b/net/openvswitch/flow_netlink.h
@@ -45,6 +45,7 @@ int ovs_nla_put_flow(const struct sw_flow_key *,
 int ovs_nla_get_flow_metadata(struct sw_flow *flow,
 			      const struct nlattr *attr);
 int ovs_nla_get_match(struct sw_flow_match *match,
+		      bool *exact_5tuple,
 		      const struct nlattr *,
 		      const struct nlattr *);
 
diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index 7b9cf2c..299ea8b 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -72,19 +72,42 @@ void ovs_flow_mask_key(struct sw_flow_key *dst, const struct sw_flow_key *src,
 		*d++ = *s++ & *m++;
 }
 
-struct sw_flow *ovs_flow_alloc(void)
+struct sw_flow *ovs_flow_alloc(bool percpu_stats)
 {
 	struct sw_flow *flow;
+	int cpu;
 
 	flow = kmem_cache_alloc(flow_cache, GFP_KERNEL);
 	if (!flow)
 		return ERR_PTR(-ENOMEM);
 
-	spin_lock_init(&flow->lock);
 	flow->sf_acts = NULL;
 	flow->mask = NULL;
 
+	flow->stats.is_percpu = percpu_stats;
+
+	if (!percpu_stats) {
+		flow->stats.stat = kzalloc(sizeof(*flow->stats.stat), GFP_KERNEL);
+		if (!flow->stats.stat)
+			goto err;
+
+		spin_lock_init(&flow->stats.stat->lock);
+	} else {
+		flow->stats.cpu_stats = alloc_percpu(struct flow_stats);
+		if (!flow->stats.cpu_stats)
+			goto err;
+
+		for_each_possible_cpu(cpu) {
+			struct flow_stats *cpu_stats;
+
+			cpu_stats = per_cpu_ptr(flow->stats.cpu_stats, cpu);
+			spin_lock_init(&cpu_stats->lock);
+		}
+	}
 	return flow;
+err:
+	kfree(flow);
+	return ERR_PTR(-ENOMEM);
 }
 
 int ovs_flow_tbl_count(struct flow_table *table)
@@ -118,6 +141,10 @@ static struct flex_array *alloc_buckets(unsigned int n_buckets)
 static void flow_free(struct sw_flow *flow)
 {
 	kfree((struct sf_flow_acts __force *)flow->sf_acts);
+	if (flow->stats.is_percpu)
+		free_percpu(flow->stats.cpu_stats);
+	else
+		kfree(flow->stats.stat);
 	kmem_cache_free(flow_cache, flow);
 }
 
diff --git a/net/openvswitch/flow_table.h b/net/openvswitch/flow_table.h
index f54aa82..1996e34 100644
--- a/net/openvswitch/flow_table.h
+++ b/net/openvswitch/flow_table.h
@@ -55,7 +55,7 @@ struct flow_table {
 int ovs_flow_init(void);
 void ovs_flow_exit(void);
 
-struct sw_flow *ovs_flow_alloc(void);
+struct sw_flow *ovs_flow_alloc(bool percpu_stats);
 void ovs_flow_free(struct sw_flow *, bool deferred);
 
 int ovs_flow_tbl_init(struct flow_table *);
-- 
1.8.3.2

[PATCH net-next 09/17] net: ovs: use kfree_rcu instead of rcu_free_{sw_flow_mask_cb, acts_callback}

[Jesse Gross]

From: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>

As we're only doing a kfree() anyway in the RCU callback, we can
simply use kfree_rcu, which does the same job, and remove the
function rcu_free_sw_flow_mask_cb() and rcu_free_acts_callback().

Signed-off-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/flow_netlink.c | 10 +---------
 net/openvswitch/flow_table.c   |  9 +--------
 2 files changed, 2 insertions(+), 17 deletions(-)

diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.c
index 3ccb92f..4d000ac 100644
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -1176,19 +1176,11 @@ struct sw_flow_actions *ovs_nla_alloc_flow_actions(int size)
 	return sfa;
 }
 
-/* RCU callback used by ovs_nla_free_flow_actions. */
-static void rcu_free_acts_callback(struct rcu_head *rcu)
-{
-	struct sw_flow_actions *sf_acts = container_of(rcu,
-			struct sw_flow_actions, rcu);
-	kfree(sf_acts);
-}
-
 /* Schedules 'sf_acts' to be freed after the next RCU grace period.
  * The caller must hold rcu_read_lock for this to be sensible. */
 void ovs_nla_free_flow_actions(struct sw_flow_actions *sf_acts)
 {
-	call_rcu(&sf_acts->rcu, rcu_free_acts_callback);
+	kfree_rcu(sf_acts, rcu);
 }
 
 static struct nlattr *reserve_sfa_size(struct sw_flow_actions **sfa,
diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index 299ea8b..099a1a9 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -155,13 +155,6 @@ static void rcu_free_flow_callback(struct rcu_head *rcu)
 	flow_free(flow);
 }
 
-static void rcu_free_sw_flow_mask_cb(struct rcu_head *rcu)
-{
-	struct sw_flow_mask *mask = container_of(rcu, struct sw_flow_mask, rcu);
-
-	kfree(mask);
-}
-
 static void flow_mask_del_ref(struct sw_flow_mask *mask, bool deferred)
 {
 	if (!mask)
@@ -173,7 +166,7 @@ static void flow_mask_del_ref(struct sw_flow_mask *mask, bool deferred)
 	if (!mask->ref_count) {
 		list_del_rcu(&mask->list);
 		if (deferred)
-			call_rcu(&mask->rcu, rcu_free_sw_flow_mask_cb);
+			kfree_rcu(mask, rcu);
 		else
 			kfree(mask);
 	}
-- 
1.8.3.2

[PATCH net-next 10/17] openvswitch: remove duplicated include from flow_table.c

[Jesse Gross]

From: Wei Yongjun <yongjun_wei-zrsr2BFq86L20UzCJQGyNP8+0UxHXcjY@public.gmane.org>

Remove duplicated include.

Signed-off-by: Wei Yongjun <yongjun_wei-zrsr2BFq86L20UzCJQGyNP8+0UxHXcjY@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/flow_table.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index 099a1a9..51f6d7f 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -44,8 +44,6 @@
 #include <net/ipv6.h>
 #include <net/ndisc.h>
 
-#include "datapath.h"
-
 #define TBL_MIN_BUCKETS		1024
 #define REHASH_INTERVAL		(10 * 60 * HZ)
 
-- 
1.8.3.2

[PATCH net-next 11/17] net: Export skb_zerocopy() to zerocopy from one skb to another

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Make the skb zerocopy logic written for nfnetlink queue available for
use by other modules.

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Reviewed-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Acked-by: David S. Miller <davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 include/linux/skbuff.h               |  3 ++
 net/core/skbuff.c                    | 85 ++++++++++++++++++++++++++++++++++++
 net/netfilter/nfnetlink_queue_core.c | 59 ++-----------------------
 3 files changed, 92 insertions(+), 55 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index bec1cc7..7c48e2d 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -2345,6 +2345,9 @@ int skb_splice_bits(struct sk_buff *skb, unsigned int offset,
 		    struct pipe_inode_info *pipe, unsigned int len,
 		    unsigned int flags);
 void skb_copy_and_csum_dev(const struct sk_buff *skb, u8 *to);
+unsigned int skb_zerocopy_headlen(const struct sk_buff *from);
+void skb_zerocopy(struct sk_buff *to, const struct sk_buff *from,
+		  int len, int hlen);
 void skb_split(struct sk_buff *skb, struct sk_buff *skb1, const u32 len);
 int skb_shift(struct sk_buff *tgt, struct sk_buff *skb, int shiftlen);
 void skb_scrub_packet(struct sk_buff *skb, bool xnet);
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 2718fed..55859cb 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -2122,6 +2122,91 @@ __wsum skb_copy_and_csum_bits(const struct sk_buff *skb, int offset,
 }
 EXPORT_SYMBOL(skb_copy_and_csum_bits);
 
+ /**
+ *	skb_zerocopy_headlen - Calculate headroom needed for skb_zerocopy()
+ *	@from: source buffer
+ *
+ *	Calculates the amount of linear headroom needed in the 'to' skb passed
+ *	into skb_zerocopy().
+ */
+unsigned int
+skb_zerocopy_headlen(const struct sk_buff *from)
+{
+	unsigned int hlen = 0;
+
+	if (!from->head_frag ||
+	    skb_headlen(from) < L1_CACHE_BYTES ||
+	    skb_shinfo(from)->nr_frags >= MAX_SKB_FRAGS)
+		hlen = skb_headlen(from);
+
+	if (skb_has_frag_list(from))
+		hlen = from->len;
+
+	return hlen;
+}
+EXPORT_SYMBOL_GPL(skb_zerocopy_headlen);
+
+/**
+ *	skb_zerocopy - Zero copy skb to skb
+ *	@to: destination buffer
+ *	@source: source buffer
+ *	@len: number of bytes to copy from source buffer
+ *	@hlen: size of linear headroom in destination buffer
+ *
+ *	Copies up to `len` bytes from `from` to `to` by creating references
+ *	to the frags in the source buffer.
+ *
+ *	The `hlen` as calculated by skb_zerocopy_headlen() specifies the
+ *	headroom in the `to` buffer.
+ */
+void
+skb_zerocopy(struct sk_buff *to, const struct sk_buff *from, int len, int hlen)
+{
+	int i, j = 0;
+	int plen = 0; /* length of skb->head fragment */
+	struct page *page;
+	unsigned int offset;
+
+	BUG_ON(!from->head_frag && !hlen);
+
+	/* dont bother with small payloads */
+	if (len <= skb_tailroom(to)) {
+		skb_copy_bits(from, 0, skb_put(to, len), len);
+		return;
+	}
+
+	if (hlen) {
+		skb_copy_bits(from, 0, skb_put(to, hlen), hlen);
+		len -= hlen;
+	} else {
+		plen = min_t(int, skb_headlen(from), len);
+		if (plen) {
+			page = virt_to_head_page(from->head);
+			offset = from->data - (unsigned char *)page_address(page);
+			__skb_fill_page_desc(to, 0, page, offset, plen);
+			get_page(page);
+			j = 1;
+			len -= plen;
+		}
+	}
+
+	to->truesize += len + plen;
+	to->len += len + plen;
+	to->data_len += len + plen;
+
+	for (i = 0; i < skb_shinfo(from)->nr_frags; i++) {
+		if (!len)
+			break;
+		skb_shinfo(to)->frags[j] = skb_shinfo(from)->frags[i];
+		skb_shinfo(to)->frags[j].size = min_t(int, skb_shinfo(to)->frags[j].size, len);
+		len -= skb_shinfo(to)->frags[j].size;
+		skb_frag_ref(to, j);
+		j++;
+	}
+	skb_shinfo(to)->nr_frags = j;
+}
+EXPORT_SYMBOL_GPL(skb_zerocopy);
+
 void skb_copy_and_csum_dev(const struct sk_buff *skb, u8 *to)
 {
 	__wsum csum;
diff --git a/net/netfilter/nfnetlink_queue_core.c b/net/netfilter/nfnetlink_queue_core.c
index 21258cf..615ee12 100644
--- a/net/netfilter/nfnetlink_queue_core.c
+++ b/net/netfilter/nfnetlink_queue_core.c
@@ -235,51 +235,6 @@ nfqnl_flush(struct nfqnl_instance *queue, nfqnl_cmpfn cmpfn, unsigned long data)
 	spin_unlock_bh(&queue->lock);
 }
 
-static void
-nfqnl_zcopy(struct sk_buff *to, const struct sk_buff *from, int len, int hlen)
-{
-	int i, j = 0;
-	int plen = 0; /* length of skb->head fragment */
-	struct page *page;
-	unsigned int offset;
-
-	/* dont bother with small payloads */
-	if (len <= skb_tailroom(to)) {
-		skb_copy_bits(from, 0, skb_put(to, len), len);
-		return;
-	}
-
-	if (hlen) {
-		skb_copy_bits(from, 0, skb_put(to, hlen), hlen);
-		len -= hlen;
-	} else {
-		plen = min_t(int, skb_headlen(from), len);
-		if (plen) {
-			page = virt_to_head_page(from->head);
-			offset = from->data - (unsigned char *)page_address(page);
-			__skb_fill_page_desc(to, 0, page, offset, plen);
-			get_page(page);
-			j = 1;
-			len -= plen;
-		}
-	}
-
-	to->truesize += len + plen;
-	to->len += len + plen;
-	to->data_len += len + plen;
-
-	for (i = 0; i < skb_shinfo(from)->nr_frags; i++) {
-		if (!len)
-			break;
-		skb_shinfo(to)->frags[j] = skb_shinfo(from)->frags[i];
-		skb_shinfo(to)->frags[j].size = min_t(int, skb_shinfo(to)->frags[j].size, len);
-		len -= skb_shinfo(to)->frags[j].size;
-		skb_frag_ref(to, j);
-		j++;
-	}
-	skb_shinfo(to)->nr_frags = j;
-}
-
 static int
 nfqnl_put_packet_info(struct sk_buff *nlskb, struct sk_buff *packet,
 		      bool csum_verify)
@@ -304,7 +259,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 {
 	size_t size;
 	size_t data_len = 0, cap_len = 0;
-	int hlen = 0;
+	unsigned int hlen = 0;
 	struct sk_buff *skb;
 	struct nlattr *nla;
 	struct nfqnl_msg_packet_hdr *pmsg;
@@ -356,14 +311,8 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 		if (data_len > entskb->len)
 			data_len = entskb->len;
 
-		if (!entskb->head_frag ||
-		    skb_headlen(entskb) < L1_CACHE_BYTES ||
-		    skb_shinfo(entskb)->nr_frags >= MAX_SKB_FRAGS)
-			hlen = skb_headlen(entskb);
-
-		if (skb_has_frag_list(entskb))
-			hlen = entskb->len;
-		hlen = min_t(int, data_len, hlen);
+		hlen = skb_zerocopy_headlen(entskb);
+		hlen = min_t(unsigned int, hlen, data_len);
 		size += sizeof(struct nlattr) + hlen;
 		cap_len = entskb->len;
 		break;
@@ -504,7 +453,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 		nla->nla_type = NFQA_PAYLOAD;
 		nla->nla_len = nla_attr_size(data_len);
 
-		nfqnl_zcopy(skb, entskb, data_len, hlen);
+		skb_zerocopy(skb, entskb, data_len, hlen);
 	}
 
 	nlh->nlmsg_len = skb->len;
-- 
1.8.3.2

[PATCH net-next 12/17] openvswitch: Allow user space to announce ability to accept unaligned Netlink messages

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Reviewed-by: Daniel Borkmann <dborkman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 include/uapi/linux/openvswitch.h |  4 ++++
 net/openvswitch/datapath.c       | 14 ++++++++++++++
 net/openvswitch/datapath.h       |  2 ++
 3 files changed, 20 insertions(+)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index d120f9f..07ef2c3 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -75,6 +75,7 @@ enum ovs_datapath_attr {
 	OVS_DP_ATTR_UPCALL_PID,		/* Netlink PID to receive upcalls */
 	OVS_DP_ATTR_STATS,		/* struct ovs_dp_stats */
 	OVS_DP_ATTR_MEGAFLOW_STATS,	/* struct ovs_dp_megaflow_stats */
+	OVS_DP_ATTR_USER_FEATURES,	/* OVS_DP_F_*  */
 	__OVS_DP_ATTR_MAX
 };
 
@@ -106,6 +107,9 @@ struct ovs_vport_stats {
 	__u64   tx_dropped;		/* no space available in linux  */
 };
 
+/* Allow last Netlink attribute to be unaligned */
+#define OVS_DP_F_UNALIGNED	(1 << 0)
+
 /* Fixed logical ports. */
 #define OVSP_LOCAL      ((__u32)0)
 
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 50d7782..6a9b0cb 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -1040,6 +1040,7 @@ static const struct genl_ops dp_flow_genl_ops[] = {
 static const struct nla_policy datapath_policy[OVS_DP_ATTR_MAX + 1] = {
 	[OVS_DP_ATTR_NAME] = { .type = NLA_NUL_STRING, .len = IFNAMSIZ - 1 },
 	[OVS_DP_ATTR_UPCALL_PID] = { .type = NLA_U32 },
+	[OVS_DP_ATTR_USER_FEATURES] = { .type = NLA_U32 },
 };
 
 static struct genl_family dp_datapath_genl_family = {
@@ -1098,6 +1099,9 @@ static int ovs_dp_cmd_fill_info(struct datapath *dp, struct sk_buff *skb,
 			&dp_megaflow_stats))
 		goto nla_put_failure;
 
+	if (nla_put_u32(skb, OVS_DP_ATTR_USER_FEATURES, dp->user_features))
+		goto nla_put_failure;
+
 	return genlmsg_end(skb, ovs_header);
 
 nla_put_failure:
@@ -1144,6 +1148,12 @@ static struct datapath *lookup_datapath(struct net *net,
 	return dp ? dp : ERR_PTR(-ENODEV);
 }
 
+static void ovs_dp_change(struct datapath *dp, struct nlattr **a)
+{
+	if (a[OVS_DP_ATTR_USER_FEATURES])
+		dp->user_features = nla_get_u32(a[OVS_DP_ATTR_USER_FEATURES]);
+}
+
 static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
 {
 	struct nlattr **a = info->attrs;
@@ -1202,6 +1212,8 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
 	parms.port_no = OVSP_LOCAL;
 	parms.upcall_portid = nla_get_u32(a[OVS_DP_ATTR_UPCALL_PID]);
 
+	ovs_dp_change(dp, a);
+
 	vport = new_vport(&parms);
 	if (IS_ERR(vport)) {
 		err = PTR_ERR(vport);
@@ -1305,6 +1317,8 @@ static int ovs_dp_cmd_set(struct sk_buff *skb, struct genl_info *info)
 	if (IS_ERR(dp))
 		goto unlock;
 
+	ovs_dp_change(dp, info->attrs);
+
 	reply = ovs_dp_cmd_build_info(dp, info, OVS_DP_CMD_NEW);
 	if (IS_ERR(reply)) {
 		err = PTR_ERR(reply);
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index ba13be4..62619a4 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -88,6 +88,8 @@ struct datapath {
 	/* Network namespace ref. */
 	struct net *net;
 #endif
+
+	u32 user_features;
 };
 
 /**
-- 
1.8.3.2

[PATCH net-next 13/17] openvswitch: Drop user features if old user space attempted to create datapath

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Drop user features if an outdated user space instance that does not
understand the concept of user_features attempted to create a new
datapath.

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 include/uapi/linux/openvswitch.h | 10 +++++++++-
 net/openvswitch/datapath.c       | 21 +++++++++++++++++++++
 2 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 07ef2c3..970553c 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -40,7 +40,15 @@ struct ovs_header {
 
 #define OVS_DATAPATH_FAMILY  "ovs_datapath"
 #define OVS_DATAPATH_MCGROUP "ovs_datapath"
-#define OVS_DATAPATH_VERSION 0x1
+
+/* V2:
+ *   - API users are expected to provide OVS_DP_ATTR_USER_FEATURES
+ *     when creating the datapath.
+ */
+#define OVS_DATAPATH_VERSION 2
+
+/* First OVS datapath version to support features */
+#define OVS_DP_VER_FEATURES 2
 
 enum ovs_datapath_cmd {
 	OVS_DP_CMD_UNSPEC,
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 6a9b0cb..497b2fd 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -1148,6 +1148,18 @@ static struct datapath *lookup_datapath(struct net *net,
 	return dp ? dp : ERR_PTR(-ENODEV);
 }
 
+static void ovs_dp_reset_user_features(struct sk_buff *skb, struct genl_info *info)
+{
+	struct datapath *dp;
+
+	dp = lookup_datapath(sock_net(skb->sk), info->userhdr, info->attrs);
+	if (!dp)
+		return;
+
+	WARN(dp->user_features, "Dropping previously announced user features\n");
+	dp->user_features = 0;
+}
+
 static void ovs_dp_change(struct datapath *dp, struct nlattr **a)
 {
 	if (a[OVS_DP_ATTR_USER_FEATURES])
@@ -1220,6 +1232,15 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
 		if (err == -EBUSY)
 			err = -EEXIST;
 
+		if (err == -EEXIST) {
+			/* An outdated user space instance that does not understand
+			 * the concept of user_features has attempted to create a new
+			 * datapath and is likely to reuse it. Drop all user features.
+			 */
+			if (info->genlhdr->version < OVS_DP_VER_FEATURES)
+				ovs_dp_reset_user_features(skb, info);
+		}
+
 		goto err_destroy_ports_array;
 	}
 
-- 
1.8.3.2

[PATCH net-next 14/17] openvswitch: Pass datapath into userspace queue functions

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Allows removing the net and dp_ifindex argument and simplify the
code.

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c | 34 ++++++++++++++--------------------
 1 file changed, 14 insertions(+), 20 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 497b2fd..235acae 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -108,10 +108,9 @@ int lockdep_ovsl_is_held(void)
 #endif
 
 static struct vport *new_vport(const struct vport_parms *);
-static int queue_gso_packets(struct net *, int dp_ifindex, struct sk_buff *,
+static int queue_gso_packets(struct datapath *dp, struct sk_buff *,
 			     const struct dp_upcall_info *);
-static int queue_userspace_packet(struct net *, int dp_ifindex,
-				  struct sk_buff *,
+static int queue_userspace_packet(struct datapath *dp, struct sk_buff *,
 				  const struct dp_upcall_info *);
 
 /* Must be called with rcu_read_lock or ovs_mutex. */
@@ -277,7 +276,6 @@ int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
 		  const struct dp_upcall_info *upcall_info)
 {
 	struct dp_stats_percpu *stats;
-	int dp_ifindex;
 	int err;
 
 	if (upcall_info->portid == 0) {
@@ -285,16 +283,10 @@ int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
 		goto err;
 	}
 
-	dp_ifindex = get_dpifindex(dp);
-	if (!dp_ifindex) {
-		err = -ENODEV;
-		goto err;
-	}
-
 	if (!skb_is_gso(skb))
-		err = queue_userspace_packet(ovs_dp_get_net(dp), dp_ifindex, skb, upcall_info);
+		err = queue_userspace_packet(dp, skb, upcall_info);
 	else
-		err = queue_gso_packets(ovs_dp_get_net(dp), dp_ifindex, skb, upcall_info);
+		err = queue_gso_packets(dp, skb, upcall_info);
 	if (err)
 		goto err;
 
@@ -310,8 +302,7 @@ err:
 	return err;
 }
 
-static int queue_gso_packets(struct net *net, int dp_ifindex,
-			     struct sk_buff *skb,
+static int queue_gso_packets(struct datapath *dp, struct sk_buff *skb,
 			     const struct dp_upcall_info *upcall_info)
 {
 	unsigned short gso_type = skb_shinfo(skb)->gso_type;
@@ -327,7 +318,7 @@ static int queue_gso_packets(struct net *net, int dp_ifindex,
 	/* Queue all of the segments. */
 	skb = segs;
 	do {
-		err = queue_userspace_packet(net, dp_ifindex, skb, upcall_info);
+		err = queue_userspace_packet(dp, skb, upcall_info);
 		if (err)
 			break;
 
@@ -394,8 +385,7 @@ static size_t upcall_msg_size(const struct sk_buff *skb,
 	return size;
 }
 
-static int queue_userspace_packet(struct net *net, int dp_ifindex,
-				  struct sk_buff *skb,
+static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 				  const struct dp_upcall_info *upcall_info)
 {
 	struct ovs_header *upcall;
@@ -403,11 +393,15 @@ static int queue_userspace_packet(struct net *net, int dp_ifindex,
 	struct sk_buff *user_skb; /* to be queued to userspace */
 	struct nlattr *nla;
 	struct genl_info info = {
-		.dst_sk = net->genl_sock,
+		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
 		.snd_portid = upcall_info->portid,
 	};
 	size_t len;
-	int err;
+	int err, dp_ifindex;
+
+	dp_ifindex = get_dpifindex(dp);
+	if (!dp_ifindex)
+		return -ENODEV;
 
 	if (vlan_tx_tag_present(skb)) {
 		nskb = skb_clone(skb, GFP_ATOMIC);
@@ -452,7 +446,7 @@ static int queue_userspace_packet(struct net *net, int dp_ifindex,
 	skb_copy_and_csum_dev(skb, nla_data(nla));
 
 	genlmsg_end(user_skb, upcall);
-	err = genlmsg_unicast(net, user_skb, upcall_info->portid);
+	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
 
 out:
 	kfree_skb(nskb);
-- 
1.8.3.2

[PATCH net-next 15/17] openvswitch: Use skb_zerocopy() for upcall

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

Use of skb_zerocopy() can avoid the expensive call to memcpy()
when copying the packet data into the Netlink skb. Completes
checksum through skb_checksum_help() if not already done in
GSO segmentation.

Zerocopy is only performed if user space supported unaligned
Netlink messages. memory mapped netlink i/o is preferred over
zerocopy if it is set up.

Cost of upcall is significantly reduced from:
+   7.48%       vhost-8471  [k] memcpy
+   5.57%     ovs-vswitchd  [k] memcpy
+   2.81%       vhost-8471  [k] csum_partial_copy_generic

to:
+   5.72%     ovs-vswitchd  [k] memcpy
+   3.32%       vhost-5153  [k] memcpy
+   0.68%       vhost-5153  [k] skb_zerocopy

(megaflows disabled)

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c | 37 +++++++++++++++++++++++++++++--------
 1 file changed, 29 insertions(+), 8 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 235acae..8557834 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -371,11 +371,11 @@ static size_t key_attr_size(void)
 		+ nla_total_size(28); /* OVS_KEY_ATTR_ND */
 }
 
-static size_t upcall_msg_size(const struct sk_buff *skb,
-			      const struct nlattr *userdata)
+static size_t upcall_msg_size(const struct nlattr *userdata,
+			      unsigned int hdrlen)
 {
 	size_t size = NLMSG_ALIGN(sizeof(struct ovs_header))
-		+ nla_total_size(skb->len) /* OVS_PACKET_ATTR_PACKET */
+		+ nla_total_size(hdrlen) /* OVS_PACKET_ATTR_PACKET */
 		+ nla_total_size(key_attr_size()); /* OVS_PACKET_ATTR_KEY */
 
 	/* OVS_PACKET_ATTR_USERDATA */
@@ -397,6 +397,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 		.snd_portid = upcall_info->portid,
 	};
 	size_t len;
+	unsigned int hlen;
 	int err, dp_ifindex;
 
 	dp_ifindex = get_dpifindex(dp);
@@ -421,7 +422,21 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 		goto out;
 	}
 
-	len = upcall_msg_size(skb, upcall_info->userdata);
+	/* Complete checksum if needed */
+	if (skb->ip_summed == CHECKSUM_PARTIAL &&
+	    (err = skb_checksum_help(skb)))
+		goto out;
+
+	/* Older versions of OVS user space enforce alignment of the last
+	 * Netlink attribute to NLA_ALIGNTO which would require extensive
+	 * padding logic. Only perform zerocopy if padding is not required.
+	 */
+	if (dp->user_features & OVS_DP_F_UNALIGNED)
+		hlen = skb_zerocopy_headlen(skb);
+	else
+		hlen = skb->len;
+
+	len = upcall_msg_size(upcall_info->userdata, hlen);
 	user_skb = genlmsg_new_unicast(len, &info, GFP_ATOMIC);
 	if (!user_skb) {
 		err = -ENOMEM;
@@ -441,13 +456,19 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 			  nla_len(upcall_info->userdata),
 			  nla_data(upcall_info->userdata));
 
-	nla = __nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, skb->len);
+	/* Only reserve room for attribute header, packet data is added
+	 * in skb_zerocopy() */
+	if (!(nla = nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, 0))) {
+		err = -ENOBUFS;
+		goto out;
+	}
+	nla->nla_len = nla_attr_size(skb->len);
 
-	skb_copy_and_csum_dev(skb, nla_data(nla));
+	skb_zerocopy(user_skb, skb, skb->len, hlen);
 
-	genlmsg_end(user_skb, upcall);
-	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
+	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
 
+	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
 out:
 	kfree_skb(nskb);
 	return err;
-- 
1.8.3.2

[PATCH net-next 16/17] openvswitch: Compute checksum in skb_gso_segment() if needed

[Jesse Gross]

From: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>

The copy & csum optimization is no longer present with zerocopy
enabled. Compute the checksum in skb_gso_segment() directly by
dropping the HW CSUM capability from the features passed in.

Signed-off-by: Thomas Graf <tgraf-G/eBtMaohhA@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 8557834..61ae3b8 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -311,7 +311,7 @@ static int queue_gso_packets(struct datapath *dp, struct sk_buff *skb,
 	struct sk_buff *segs, *nskb;
 	int err;
 
-	segs = __skb_gso_segment(skb, NETIF_F_SG | NETIF_F_HW_CSUM, false);
+	segs = __skb_gso_segment(skb, NETIF_F_SG, false);
 	if (IS_ERR(segs))
 		return PTR_ERR(segs);
 
-- 
1.8.3.2

[PATCH net-next 17/17] ovs: make functions local

[Jesse Gross]

From: Stephen Hemminger <stephen-OTpzqLSitTUnbdJkjeBofR2eb7JE58TQ@public.gmane.org>

Several functions and datastructures could be local
Found with 'make namespacecheck'

Signed-off-by: Stephen Hemminger <stephen-OTpzqLSitTUnbdJkjeBofR2eb7JE58TQ@public.gmane.org>
Signed-off-by: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
---
 net/openvswitch/datapath.c | 4 ++--
 net/openvswitch/datapath.h | 2 --
 net/openvswitch/vport.c    | 6 +++++-
 net/openvswitch/vport.h    | 1 -
 4 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 61ae3b8..df46928 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -132,7 +132,7 @@ static struct datapath *get_dp(struct net *net, int dp_ifindex)
 }
 
 /* Must be called with rcu_read_lock or ovs_mutex. */
-const char *ovs_dp_name(const struct datapath *dp)
+static const char *ovs_dp_name(const struct datapath *dp)
 {
 	struct vport *vport = ovs_vport_ovsl_rcu(dp, OVSP_LOCAL);
 	return vport->ops->get_name(vport);
@@ -1466,7 +1466,7 @@ struct genl_family dp_vport_genl_family = {
 	.parallel_ops = true,
 };
 
-struct genl_multicast_group ovs_dp_vport_multicast_group = {
+static struct genl_multicast_group ovs_dp_vport_multicast_group = {
 	.name = OVS_VPORT_MCGROUP
 };
 
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index 62619a4..6be9fbb 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -182,14 +182,12 @@ static inline struct vport *ovs_vport_ovsl(const struct datapath *dp, int port_n
 
 extern struct notifier_block ovs_dp_device_notifier;
 extern struct genl_family dp_vport_genl_family;
-extern struct genl_multicast_group ovs_dp_vport_multicast_group;
 
 void ovs_dp_process_received_packet(struct vport *, struct sk_buff *);
 void ovs_dp_detach_port(struct vport *);
 int ovs_dp_upcall(struct datapath *, struct sk_buff *,
 		  const struct dp_upcall_info *);
 
-const char *ovs_dp_name(const struct datapath *dp);
 struct sk_buff *ovs_vport_cmd_build_info(struct vport *, u32 pid, u32 seq,
 					 u8 cmd);
 
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index d830a95f..f28ff7e 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -33,6 +33,9 @@
 #include "vport.h"
 #include "vport-internal_dev.h"
 
+static void ovs_vport_record_error(struct vport *,
+				   enum vport_err_type err_type);
+
 /* List of statically compiled vport implementations.  Don't forget to also
  * add yours to the list at the bottom of vport.h. */
 static const struct vport_ops *vport_ops_list[] = {
@@ -396,7 +399,8 @@ int ovs_vport_send(struct vport *vport, struct sk_buff *skb)
  * If using the vport generic stats layer indicate that an error of the given
  * type has occurred.
  */
-void ovs_vport_record_error(struct vport *vport, enum vport_err_type err_type)
+static void ovs_vport_record_error(struct vport *vport,
+				   enum vport_err_type err_type)
 {
 	spin_lock(&vport->stats_lock);
 
diff --git a/net/openvswitch/vport.h b/net/openvswitch/vport.h
index 1a9fbce..92137dd 100644
--- a/net/openvswitch/vport.h
+++ b/net/openvswitch/vport.h
@@ -192,7 +192,6 @@ static inline struct vport *vport_from_priv(const void *priv)
 
 void ovs_vport_receive(struct vport *, struct sk_buff *,
 		       struct ovs_key_ipv4_tunnel *);
-void ovs_vport_record_error(struct vport *, enum vport_err_type err_type);
 
 /* List of statically compiled vport implementations.  Don't forget to also
  * add yours to the list at the top of vport.c. */
-- 
1.8.3.2

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Date: Mon,  6 Jan 2014 16:15:59 -0800

> Open vSwitch changes for net-next/3.14. Highlights are:
>  * Performance improvements in the mechanism to get packets to userspace
>    using memory mapped netlink and skb zero copy where appropriate.
>  * Per-cpu flow stats in situations where flows are likely to be shared
>    across CPUs. Standard flow stats are used in other situations to save
>    memory and allocation time.
>  * A handful of code cleanups and rationalization.

Lots of good stuff in here, pulled, thanks Jesse.

Re: [ovs-dev] [GIT net-next] Open vSwitch

[Zoltan Kiss]

Hi,

I've tried the latest net-next on a Xenserver install with 1.9.3 
userspace, and it seems this patch series broke it (at least after 
reverting that locally it works now). I haven't went too far yet 
checking what's the problem, but it seems the xenbrX device doesn't 
really receive too much of the traffic coming through the NIC. Is it 
expected?

Regards,

Zoli

On 07/01/14 00:15, Jesse Gross wrote:
> Open vSwitch changes for net-next/3.14. Highlights are:
>   * Performance improvements in the mechanism to get packets to userspace
>     using memory mapped netlink and skb zero copy where appropriate.
>   * Per-cpu flow stats in situations where flows are likely to be shared
>     across CPUs. Standard flow stats are used in other situations to save
>     memory and allocation time.
>   * A handful of code cleanups and rationalization.
>
> The following changes since commit 6ce4eac1f600b34f2f7f58f9cd8f0503d79e42ae:
>
>    Linux 3.13-rc1 (2013-11-22 11:30:55 -0800)
>
> are available in the git repository at:
>
>    git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master
>
> for you to fetch changes up to 443cd88c8a31379e95326428bbbd40af25c1d440:
>
>    ovs: make functions local (2014-01-06 15:54:39 -0800)
>
> ----------------------------------------------------------------
> Andy Zhou (1):
>        openvswitch: Change ovs_flow_tbl_lookup_xx() APIs
>
> Ben Pfaff (2):
>        openvswitch: Correct comment.
>        openvswitch: Shrink sw_flow_mask by 8 bytes (64-bit) or 4 bytes (32-bit).
>
> Daniel Borkmann (1):
>        net: ovs: use kfree_rcu instead of rcu_free_{sw_flow_mask_cb,acts_callback}
>
> Jesse Gross (1):
>        openvswitch: Silence RCU lockdep checks from flow lookup.
>
> Pravin B Shelar (1):
>        openvswitch: Per cpu flow stats.
>
> Stephen Hemminger (1):
>        ovs: make functions local
>
> Thomas Graf (9):
>        genl: Add genlmsg_new_unicast() for unicast message allocation
>        netlink: Avoid netlink mmap alloc if msg size exceeds frame size
>        openvswitch: Enable memory mapped Netlink i/o
>        net: Export skb_zerocopy() to zerocopy from one skb to another
>        openvswitch: Allow user space to announce ability to accept unaligned Netlink messages
>        openvswitch: Drop user features if old user space attempted to create datapath
>        openvswitch: Pass datapath into userspace queue functions
>        openvswitch: Use skb_zerocopy() for upcall
>        openvswitch: Compute checksum in skb_gso_segment() if needed
>
> Wei Yongjun (1):
>        openvswitch: remove duplicated include from flow_table.c
>
>   include/linux/skbuff.h               |   3 +
>   include/net/genetlink.h              |   4 +
>   include/uapi/linux/openvswitch.h     |  14 ++-
>   net/core/skbuff.c                    |  85 +++++++++++++
>   net/netfilter/nfnetlink_queue_core.c |  59 +--------
>   net/netlink/af_netlink.c             |   4 +
>   net/netlink/genetlink.c              |  21 ++++
>   net/openvswitch/datapath.c           | 231 +++++++++++++++++++----------------
>   net/openvswitch/datapath.h           |   6 +-
>   net/openvswitch/flow.c               |  96 +++++++++++++--
>   net/openvswitch/flow.h               |  33 +++--
>   net/openvswitch/flow_netlink.c       |  66 ++++++++--
>   net/openvswitch/flow_netlink.h       |   1 +
>   net/openvswitch/flow_table.c         |  60 ++++++---
>   net/openvswitch/flow_table.h         |   6 +-
>   net/openvswitch/vport.c              |   6 +-
>   net/openvswitch/vport.h              |   1 -
>   17 files changed, 483 insertions(+), 213 deletions(-)
> _______________________________________________
> dev mailing list
> dev@openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev
>

Re: [GIT net-next] Open vSwitch

[Jesse Gross]

On Wed, Jan 8, 2014 at 9:49 AM, Zoltan Kiss <zoltan.kiss-Sxgqhf6Nn4DQT0dZR+AlfA@public.gmane.org> wrote:
> Hi,
>
> I've tried the latest net-next on a Xenserver install with 1.9.3 userspace,
> and it seems this patch series broke it (at least after reverting that
> locally it works now). I haven't went too far yet checking what's the
> problem, but it seems the xenbrX device doesn't really receive too much of
> the traffic coming through the NIC. Is it expected?

What do you mean by doesn't receive too much traffic? What does it get?

Re: [ovs-dev] [GIT net-next] Open vSwitch

[Zoltan Kiss]

On 08/01/14 15:10, Jesse Gross wrote:
> On Wed, Jan 8, 2014 at 9:49 AM, Zoltan Kiss <zoltan.kiss@citrix.com> wrote:
>> Hi,
>>
>> I've tried the latest net-next on a Xenserver install with 1.9.3 userspace,
>> and it seems this patch series broke it (at least after reverting that
>> locally it works now). I haven't went too far yet checking what's the
>> problem, but it seems the xenbrX device doesn't really receive too much of
>> the traffic coming through the NIC. Is it expected?
>
> What do you mean by doesn't receive too much traffic? What does it get?
>

Sorry for the vague error description, now I had more time to look into 
this. I think the problem boils down to this:

Jan 13 17:55:07 localhost ovs-vswitchd: 
07890|netlink_socket|DBG|nl_sock_recv__ (Success): nl(len:274, 
type=29(ovs_packet), flags=0, seq=0, pid=0,genl(cmd=1,version=1)
Jan 13 17:55:07 localhost ovs-vswitchd: 07891|netlink|DBG|attributes 
followed by garbage
Jan 13 17:55:07 localhost ovs-vswitchd: 07892|dpif|WARN|system@xenbr0: 
recv failed (Invalid argument)

That's just keep repeating. I'm keep looking.

Zoli

Re: [GIT net-next] Open vSwitch

[Zoltan Kiss]

On 13/01/14 18:04, Zoltan Kiss wrote:
> On 08/01/14 15:10, Jesse Gross wrote:
>> On Wed, Jan 8, 2014 at 9:49 AM, Zoltan Kiss <zoltan.kiss-Sxgqhf6Nn4DQT0dZR+AlfA@public.gmane.org> 
>> wrote:
>>> Hi,
>>>
>>> I've tried the latest net-next on a Xenserver install with 1.9.3 
>>> userspace,
>>> and it seems this patch series broke it (at least after reverting that
>>> locally it works now). I haven't went too far yet checking what's the
>>> problem, but it seems the xenbrX device doesn't really receive too 
>>> much of
>>> the traffic coming through the NIC. Is it expected?
>>
>> What do you mean by doesn't receive too much traffic? What does it get?
>>
>
> Sorry for the vague error description, now I had more time to look 
> into this. I think the problem boils down to this:
>
> Jan 13 17:55:07 localhost ovs-vswitchd: 
> 07890|netlink_socket|DBG|nl_sock_recv__ (Success): nl(len:274, 
> type=29(ovs_packet), flags=0, seq=0, pid=0,genl(cmd=1,version=1)
> Jan 13 17:55:07 localhost ovs-vswitchd: 07891|netlink|DBG|attributes 
> followed by garbage
> Jan 13 17:55:07 localhost ovs-vswitchd: 07892|dpif|WARN|system@xenbr0: 
> recv failed (Invalid argument)
>
> That's just keep repeating. I'm keep looking.

Now I reverted these top 3 commits:

ovs: make functions local
openvswitch: Compute checksum in skb_gso_segment() if needed
openvswitch: Use skb_zerocopy() for upcall

And it works. I guess the last one causing the problem. Might be an 
important factor, I'm using 32 bit Dom0.

Zoli

Re: [GIT net-next] Open vSwitch

[Jesse Gross]

On Mon, Jan 13, 2014 at 4:31 PM, Zoltan Kiss <zoltan.kiss-Sxgqhf6Nn4DQT0dZR+AlfA@public.gmane.org> wrote:
> On 13/01/14 18:04, Zoltan Kiss wrote:
>>
>> On 08/01/14 15:10, Jesse Gross wrote:
>>>
>>> On Wed, Jan 8, 2014 at 9:49 AM, Zoltan Kiss <zoltan.kiss-Sxgqhf6Nn4DQT0dZR+AlfA@public.gmane.org>
>>> wrote:
>>>>
>>>> Hi,
>>>>
>>>> I've tried the latest net-next on a Xenserver install with 1.9.3
>>>> userspace,
>>>> and it seems this patch series broke it (at least after reverting that
>>>> locally it works now). I haven't went too far yet checking what's the
>>>> problem, but it seems the xenbrX device doesn't really receive too much
>>>> of
>>>> the traffic coming through the NIC. Is it expected?
>>>
>>>
>>> What do you mean by doesn't receive too much traffic? What does it get?
>>>
>>
>> Sorry for the vague error description, now I had more time to look into
>> this. I think the problem boils down to this:
>>
>> Jan 13 17:55:07 localhost ovs-vswitchd:
>> 07890|netlink_socket|DBG|nl_sock_recv__ (Success): nl(len:274,
>> type=29(ovs_packet), flags=0, seq=0, pid=0,genl(cmd=1,version=1)
>> Jan 13 17:55:07 localhost ovs-vswitchd: 07891|netlink|DBG|attributes
>> followed by garbage
>> Jan 13 17:55:07 localhost ovs-vswitchd: 07892|dpif|WARN|system@xenbr0:
>> recv failed (Invalid argument)
>>
>> That's just keep repeating. I'm keep looking.
>
>
> Now I reverted these top 3 commits:
>
> ovs: make functions local
>
> openvswitch: Compute checksum in skb_gso_segment() if needed
> openvswitch: Use skb_zerocopy() for upcall
>
> And it works. I guess the last one causing the problem. Might be an
> important factor, I'm using 32 bit Dom0.

I think you're probably right. Thomas - can you take a look?

We shouldn't be doing any zerocopy in this situation but it looks to
me like we don't do any padding at all, even in situations where we
are copying the data.

Re: [GIT net-next] Open vSwitch

[Thomas Graf]

On 01/14/2014 02:30 AM, Jesse Gross wrote:
>> And it works. I guess the last one causing the problem. Might be an
>> important factor, I'm using 32 bit Dom0.
>
> I think you're probably right. Thomas - can you take a look?
>
> We shouldn't be doing any zerocopy in this situation but it looks to
> me like we don't do any padding at all, even in situations where we
> are copying the data.

I'm looking into this now. The zerocopy method should only be attempted
if user space has announced the ability to received unaligned messages.

@Zoltan: I assume you are using an unmodified OVS 1.9.3?

Re: [ovs-dev] [GIT net-next] Open vSwitch

[Zoltan Kiss]

On 14/01/14 09:46, Thomas Graf wrote:
> On 01/14/2014 02:30 AM, Jesse Gross wrote:
>>> And it works. I guess the last one causing the problem. Might be an
>>> important factor, I'm using 32 bit Dom0.
>>
>> I think you're probably right. Thomas - can you take a look?
>>
>> We shouldn't be doing any zerocopy in this situation but it looks to
>> me like we don't do any padding at all, even in situations where we
>> are copying the data.
>
> I'm looking into this now. The zerocopy method should only be attempted
> if user space has announced the ability to received unaligned messages.
>
> @Zoltan: I assume you are using an unmodified OVS 1.9.3?
>
Well, there are a few changes, but none of them seems to be significant. 
Most of them are just making OVS work with our build system. Here is the 
patchqueue:

https://github.com/xenserver/openvswitch.pg/tree/master/master

See the series file for the order of the patches.

Zoli

[PATCH net-next] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[Thomas Graf]

While the zerocopy method is correctly omitted if user space
does not support unaligned Netlink messages. The attribute is
still not padded correctly as skb_zerocopy() will not ensure
padding and the attribute size is no longer pre calculated
though nla_reserve() which ensured padding previously.

This patch applies appropriate padding if a linear data copy
was performed in skb_zerocopy().

Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
 net/openvswitch/datapath.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index df46928..24e93f5 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -396,7 +396,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
 		.snd_portid = upcall_info->portid,
 	};
-	size_t len;
+	size_t len, plen;
 	unsigned int hlen;
 	int err, dp_ifindex;
 
@@ -466,6 +466,11 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 
 	skb_zerocopy(user_skb, skb, skb->len, hlen);
 
+	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
+	if (!(dp->user_features & OVS_DP_F_UNALIGNED) &&
+	    (plen = (ALIGN(user_skb->len, NLA_ALIGNTO) - user_skb->len)) > 0)
+		skb_put(user_skb, plen);
+
 	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
 
 	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);

Re: [PATCH net-next] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[Thomas Graf]

On 01/14/14 at 04:16pm, Thomas Graf wrote:
> While the zerocopy method is correctly omitted if user space
> does not support unaligned Netlink messages. The attribute is
> still not padded correctly as skb_zerocopy() will not ensure
> padding and the attribute size is no longer pre calculated
> though nla_reserve() which ensured padding previously.
> 
> This patch applies appropriate padding if a linear data copy
> was performed in skb_zerocopy().
> 
> Signed-off-by: Thomas Graf <tgraf@suug.ch>
> ---
>  net/openvswitch/datapath.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
> index df46928..24e93f5 100644
> --- a/net/openvswitch/datapath.c
> +++ b/net/openvswitch/datapath.c
> @@ -396,7 +396,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
>  		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
>  		.snd_portid = upcall_info->portid,
>  	};
> -	size_t len;
> +	size_t len, plen;
>  	unsigned int hlen;
>  	int err, dp_ifindex;
>  
> @@ -466,6 +466,11 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
>  
>  	skb_zerocopy(user_skb, skb, skb->len, hlen);
>  
> +	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
> +	if (!(dp->user_features & OVS_DP_F_UNALIGNED) &&
> +	    (plen = (ALIGN(user_skb->len, NLA_ALIGNTO) - user_skb->len)) > 0)
> +		skb_put(user_skb, plen);

While this fixes the padding issue, it leaves the padding
uninitialized, I will send v2 with an additional memset().

[PATCH net-next v2] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[Thomas Graf]

While the zerocopy method is correctly omitted if user space
does not support unaligned Netlink messages. The attribute is
still not padded correctly as skb_zerocopy() will not ensure
padding and the attribute size is no longer pre calculated
though nla_reserve() which ensured padding previously.

This patch applies appropriate padding if a linear data copy
was performed in skb_zerocopy().

Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
v2: initialize padding to 0's

 net/openvswitch/datapath.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index df46928..3ca9121 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -396,7 +396,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
 		.snd_portid = upcall_info->portid,
 	};
-	size_t len;
+	size_t len, plen;
 	unsigned int hlen;
 	int err, dp_ifindex;
 
@@ -466,6 +466,11 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 
 	skb_zerocopy(user_skb, skb, skb->len, hlen);
 
+	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
+	if (!(dp->user_features & OVS_DP_F_UNALIGNED) &&
+	    (plen = (ALIGN(user_skb->len, NLA_ALIGNTO) - user_skb->len)) > 0)
+		memset(skb_put(user_skb, plen), 0, plen);
+
 	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
 
 	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);

Re: [PATCH net-next v2] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[Zoltan Kiss]

It works for me, thanks!

Acked-by: Zoltan Kiss <zoltan.kiss@citrix.com>

On 14/01/14 16:27, Thomas Graf wrote:
> While the zerocopy method is correctly omitted if user space
> does not support unaligned Netlink messages. The attribute is
> still not padded correctly as skb_zerocopy() will not ensure
> padding and the attribute size is no longer pre calculated
> though nla_reserve() which ensured padding previously.
>
> This patch applies appropriate padding if a linear data copy
> was performed in skb_zerocopy().
>
> Signed-off-by: Thomas Graf <tgraf@suug.ch>
> ---
> v2: initialize padding to 0's
>
>   net/openvswitch/datapath.c | 7 ++++++-
>   1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
> index df46928..3ca9121 100644
> --- a/net/openvswitch/datapath.c
> +++ b/net/openvswitch/datapath.c
> @@ -396,7 +396,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
>   		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
>   		.snd_portid = upcall_info->portid,
>   	};
> -	size_t len;
> +	size_t len, plen;
>   	unsigned int hlen;
>   	int err, dp_ifindex;
>
> @@ -466,6 +466,11 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
>
>   	skb_zerocopy(user_skb, skb, skb->len, hlen);
>
> +	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
> +	if (!(dp->user_features & OVS_DP_F_UNALIGNED) &&
> +	    (plen = (ALIGN(user_skb->len, NLA_ALIGNTO) - user_skb->len)) > 0)
> +		memset(skb_put(user_skb, plen), 0, plen);
> +
>   	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
>
>   	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
>

Re: [PATCH net-next v2] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[David Miller]

From: Thomas Graf <tgraf@suug.ch>
Date: Tue, 14 Jan 2014 16:27:49 +0000

> While the zerocopy method is correctly omitted if user space
> does not support unaligned Netlink messages. The attribute is
> still not padded correctly as skb_zerocopy() will not ensure
> padding and the attribute size is no longer pre calculated
> though nla_reserve() which ensured padding previously.
> 
> This patch applies appropriate padding if a linear data copy
> was performed in skb_zerocopy().
> 
> Signed-off-by: Thomas Graf <tgraf@suug.ch>
> ---
> v2: initialize padding to 0's

Jesse do you want to queue this up or should I apply it directly?

Thanks.

Re: [PATCH net-next v2] openvswitch: Pad OVS_PACKET_ATTR_PACKET if linear copy was performed

[Jesse Gross]

On Tue, Jan 14, 2014 at 7:03 PM, David Miller <davem@davemloft.net> wrote:
> From: Thomas Graf <tgraf@suug.ch>
> Date: Tue, 14 Jan 2014 16:27:49 +0000
>
>> While the zerocopy method is correctly omitted if user space
>> does not support unaligned Netlink messages. The attribute is
>> still not padded correctly as skb_zerocopy() will not ensure
>> padding and the attribute size is no longer pre calculated
>> though nla_reserve() which ensured padding previously.
>>
>> This patch applies appropriate padding if a linear data copy
>> was performed in skb_zerocopy().
>>
>> Signed-off-by: Thomas Graf <tgraf@suug.ch>
>> ---
>> v2: initialize padding to 0's
>
> Jesse do you want to queue this up or should I apply it directly?

I queued it up, thanks.

[GIT net-next] Open vSwitch

[Pravin B Shelar]

Following batch of patches brings feature parity between upstream
ovs and out of tree ovs module.

Two features are added, first adds support to export egress
tunnel information for a packet. This is used to improve
visibility in network traffic. Second feature allows userspace
vswitchd process to probe ovs module features. Other patches
are optimization and code cleanup.

----------------------------------------------------------------
The following changes since commit c0560b9c523341516eabf0f3b51832256caa7bbb:

  dccp: Convert DCCP_WARN to net_warn_ratelimited (2014-11-08 21:22:54 -0500)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs

for you to fetch changes up to 05da5898a96c05e32aa9850c9cd89eef29471b13:

  openvswitch: Add support for OVS_FLOW_ATTR_PROBE. (2014-11-09 18:58:44 -0800)

----------------------------------------------------------------
Jarno Rajahalme (1):
      openvswitch: Add support for OVS_FLOW_ATTR_PROBE.

Pravin B Shelar (3):
      openvswitch: Export symbols as GPL symbols.
      openvswitch: Optimize recirc action.
      openvswitch: Remove redundant key ref from upcall_info.

Thomas Graf (1):
      openvswitch: Constify various function arguments

Wenyu Zhang (1):
      openvswitch: Extend packet attribute for egress tunnel info

 include/uapi/linux/openvswitch.h |  15 ++
 net/openvswitch/actions.c        | 180 ++++++++++++++------
 net/openvswitch/datapath.c       | 129 ++++++++------
 net/openvswitch/datapath.h       |  22 +--
 net/openvswitch/flow.c           |   8 +-
 net/openvswitch/flow.h           |  71 ++++++--
 net/openvswitch/flow_netlink.c   | 357 +++++++++++++++++++++++----------------
 net/openvswitch/flow_netlink.h   |  13 +-
 net/openvswitch/flow_table.c     |  12 +-
 net/openvswitch/flow_table.h     |   8 +-
 net/openvswitch/vport-geneve.c   |  23 ++-
 net/openvswitch/vport-gre.c      |  12 +-
 net/openvswitch/vport-netdev.c   |   2 +-
 net/openvswitch/vport-vxlan.c    |  24 ++-
 net/openvswitch/vport.c          |  81 +++++++--
 net/openvswitch/vport.h          |  20 ++-
 16 files changed, 664 insertions(+), 313 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Pravin B Shelar <pshelar@nicira.com>
Date: Sun,  9 Nov 2014 19:58:59 -0800

> Following batch of patches brings feature parity between upstream
> ovs and out of tree ovs module.
> 
> Two features are added, first adds support to export egress
> tunnel information for a packet. This is used to improve
> visibility in network traffic. Second feature allows userspace
> vswitchd process to probe ovs module features. Other patches
> are optimization and code cleanup.

Pulled, thanks Pravin.

[GIT net-next] Open vSwitch

[Pravin B Shelar]

First two patches are related to OVS MPLS support. Rest of patches
are various refactoring and minor improvements to openvswitch.

----------------------------------------------------------------

The following changes since commit 30349bdbc4da5ecf0efa25556e3caff9c9b8c5f7:

  net: phy: spi_ks8995: remove sysfs bin file by registered attribute (2014-11-04 17:18:45 -0500)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs

for you to fetch changes up to fb90c8d8d5169d4dfbe5896721367e1904638a91:

  openvswitch: Avoid NULL mask check while building mask (2014-11-04 22:20:33 -0800)

----------------------------------------------------------------
Andy Zhou (2):
      openvswitch: refactor do_output() to move NULL check out of fast path
      openvswitch: Refactor get_dp() function into multiple access APIs.

Chunhe Li (1):
      openvswitch: Drop packets when interdev is not up

Jarno Rajahalme (1):
      openvswitch: Fix the type of struct ovs_key_nd nd_target field.

Jesse Gross (1):
      openvswitch: Additional logging for -EINVAL on flow setups.

Joe Stringer (3):
      openvswitch: Remove redundant tcp_flags code.
      openvswitch: Refactor ovs_flow_cmd_fill_info().
      openvswitch: Move key_attr_size() to flow_netlink.h.

Lorand Jakab (1):
      openvswitch: Remove flow member from struct ovs_skb_cb

Pravin B Shelar (4):
      net: Remove MPLS GSO feature.
      openvswitch: Move table destroy to dp-rcu callback.
      openvswitch: Refactor action alloc and copy api.
      openvswitch: Avoid NULL mask check while building mask

Simon Horman (1):
      openvswitch: Add basic MPLS support to kernel

 include/linux/netdev_features.h      |   7 +-
 include/linux/netdevice.h            |   1 -
 include/linux/skbuff.h               |   3 -
 include/net/mpls.h                   |  39 +++++
 include/uapi/linux/openvswitch.h     |  38 ++++-
 net/core/dev.c                       |   3 +-
 net/core/ethtool.c                   |   1 -
 net/ipv4/af_inet.c                   |   1 -
 net/ipv4/tcp_offload.c               |   1 -
 net/ipv4/udp_offload.c               |   3 +-
 net/ipv6/ip6_offload.c               |   1 -
 net/ipv6/udp_offload.c               |   3 +-
 net/mpls/mpls_gso.c                  |   3 +-
 net/openvswitch/Kconfig              |   1 +
 net/openvswitch/actions.c            | 136 ++++++++++++---
 net/openvswitch/datapath.c           | 215 ++++++++++++-----------
 net/openvswitch/datapath.h           |   4 +-
 net/openvswitch/flow.c               |  30 ++++
 net/openvswitch/flow.h               |  17 +-
 net/openvswitch/flow_netlink.c       | 322 +++++++++++++++++++++++++----------
 net/openvswitch/flow_netlink.h       |   5 +-
 net/openvswitch/flow_table.c         |  11 +-
 net/openvswitch/flow_table.h         |   2 +-
 net/openvswitch/vport-internal_dev.c |   5 +
 24 files changed, 606 insertions(+), 246 deletions(-)
 create mode 100644 include/net/mpls.h

Re: [GIT net-next] Open vSwitch

[David Miller]

Please do not submit your patches such that the email Date: field is
the commit's date.  You're not posting these on Nov. 4th, yet that
is the Date: field on all of the individual patch emails.

I want them to be the date at the time you post the patch to the mailing
list.

Otherwise the ordering in patchwork is not cronological wrt. the list's
postings and this makes my work more difficult than it needs to be.

Thanks.

Re: [GIT net-next] Open vSwitch

[Pravin Shelar]

On Wed, Nov 5, 2014 at 12:10 PM, David Miller <davem@davemloft.net> wrote:
>
> Please do not submit your patches such that the email Date: field is
> the commit's date.  You're not posting these on Nov. 4th, yet that
> is the Date: field on all of the individual patch emails.
>
> I want them to be the date at the time you post the patch to the mailing
> list.
>
> Otherwise the ordering in patchwork is not cronological wrt. the list's
> postings and this makes my work more difficult than it needs to be.
>
Sorry about the Date field. NTP stopped working on my machine thats
why the date got messed up.

[GIT net-next] Open vSwitch

[Pravin B Shelar]

Following patches adds recirculation and hash action to OVS.
First three patches does code restructuring which is required
for last patch.
Recirculation implementation is changed, according to comments from
David Miller, to avoid using recursive calls in OVS. It is using
queue to record recirc action and deferred recirc is executed at
the end of current actions execution.

----------------------------------------------------------------
The following changes since commit b954d83421d51d822c42e5ab7b65069b25ad3005:

  net: bpf: only build bpf_jit_binary_{alloc, free}() when jit selected (2014-09-10 14:05:07 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs

for you to fetch changes up to 9b8ede54a8bd319789e8bceb19789463bb944701:

  openvswitch: Add recirc and hash action. (2014-09-11 13:35:29 -0700)

----------------------------------------------------------------
Andy Zhou (2):
      datapath: simplify sample action implementation
      openvswitch: Add recirc and hash action.

Pravin B Shelar (2):
      datapath: refactor ovs flow extract API.
      datapath: Use tun_key only for egress tunnel path.

 include/uapi/linux/openvswitch.h |  26 +++++
 net/openvswitch/actions.c        | 247 ++++++++++++++++++++++++++++++++++-----
 net/openvswitch/datapath.c       |  52 +++++----
 net/openvswitch/datapath.h       |  17 ++-
 net/openvswitch/flow.c           |  54 +++++++--
 net/openvswitch/flow.h           |  10 +-
 net/openvswitch/flow_netlink.c   |  63 +++++++---
 net/openvswitch/flow_netlink.h   |   4 +-
 net/openvswitch/vport-gre.c      |  22 ++--
 net/openvswitch/vport-vxlan.c    |  20 ++--
 net/openvswitch/vport.c          |  13 ++-
 11 files changed, 419 insertions(+), 109 deletions(-)

Re: [GIT net-next] Open vSwitch

[Pravin Shelar]

Please ignore this series, I have used datapath as subsystem name
rather than openvswitch. I will respin it shortly.

On Thu, Sep 11, 2014 at 3:01 PM, Pravin B Shelar <pshelar@nicira.com> wrote:
> Following patches adds recirculation and hash action to OVS.
> First three patches does code restructuring which is required
> for last patch.
> Recirculation implementation is changed, according to comments from
> David Miller, to avoid using recursive calls in OVS. It is using
> queue to record recirc action and deferred recirc is executed at
> the end of current actions execution.
>
> ----------------------------------------------------------------
> The following changes since commit b954d83421d51d822c42e5ab7b65069b25ad3005:
>
>   net: bpf: only build bpf_jit_binary_{alloc, free}() when jit selected (2014-09-10 14:05:07 -0700)
>
> are available in the git repository at:
>
>   git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs
>
> for you to fetch changes up to 9b8ede54a8bd319789e8bceb19789463bb944701:
>
>   openvswitch: Add recirc and hash action. (2014-09-11 13:35:29 -0700)
>
> ----------------------------------------------------------------
> Andy Zhou (2):
>       datapath: simplify sample action implementation
>       openvswitch: Add recirc and hash action.
>
> Pravin B Shelar (2):
>       datapath: refactor ovs flow extract API.
>       datapath: Use tun_key only for egress tunnel path.
>
>  include/uapi/linux/openvswitch.h |  26 +++++
>  net/openvswitch/actions.c        | 247 ++++++++++++++++++++++++++++++++++-----
>  net/openvswitch/datapath.c       |  52 +++++----
>  net/openvswitch/datapath.h       |  17 ++-
>  net/openvswitch/flow.c           |  54 +++++++--
>  net/openvswitch/flow.h           |  10 +-
>  net/openvswitch/flow_netlink.c   |  63 +++++++---
>  net/openvswitch/flow_netlink.h   |   4 +-
>  net/openvswitch/vport-gre.c      |  22 ++--
>  net/openvswitch/vport-vxlan.c    |  20 ++--
>  net/openvswitch/vport.c          |  13 ++-
>  11 files changed, 419 insertions(+), 109 deletions(-)

[GIT net-next] Open vSwitch

[Pravin B Shelar]

Following patches introduces flow mask cache. To process any packet
OVS need to apply flow mask to the flow and lookup the flow in flow table.
so packet processing performance is directly dependant on number of entries
in mask list.

Following patch adds mask cache so that we do not need to iterate over
all entries in mask list on every packet. We have seen good performance
improvement with this patch.

Before the mask-cache, a single stream which matched the first mask
got a throughput of about 900K pps. A stream which matched the 20th mask
got a throughput of about 400K pps. After the mask-cache patch, all
streams throughput went back up to 900K pps.

----------------------------------------------------------------

The following changes since commit 2f55daa5464e8dfc8787ec863b6d1094522dbd69:

  net: stmmac: Support devicetree configs for mcast and ucast filter entries (2014-07-31 15:31:02 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs

for you to fetch changes up to 4955f0f9cbefa73577cd30ec262538ffc73dd4c2:

  openvswitch: Introduce flow mask cache. (2014-07-31 15:49:55 -0700)

----------------------------------------------------------------
Pravin B Shelar (3):
      openvswitch: Move table destroy to dp-rcu callback.
      openvswitch: Convert mask list into mask array.
      openvswitch: Introduce flow mask cache.

 net/openvswitch/datapath.c   |   8 +-
 net/openvswitch/flow.h       |   1 -
 net/openvswitch/flow_table.c | 293 +++++++++++++++++++++++++++++++++++++------
 net/openvswitch/flow_table.h |  21 +++-
 4 files changed, 275 insertions(+), 48 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Pravin B Shelar <pshelar@nicira.com>
Date: Thu, 31 Jul 2014 16:57:37 -0700

> Following patch adds mask cache so that we do not need to iterate over
> all entries in mask list on every packet. We have seen good performance
> improvement with this patch.

How much have you thought about the DoS'ability of openvswitch's
datastructures?

What are the upper bounds for performance of packet switching?

To be quite honest, a lot of the openvswitch data structures
adjustments that hit my inbox seem to me to only address specific
situations that specific user configurations have run into.

It took us two decades, but we ripped out the ipv4 routing cache
because external entities could provoke unreasonable worst case
behavior in routing lookups.

With openvswitch you guys have a unique opportunity to try and design
all of your features such that they absolutely can use scalable
datastructures from the beginning that provide reasonable performance
in the common case and precise upper bounds for any possible sequence
of incoming packets.

New features tend to blind the developer to the eventual long term
ramifications on performance.  Would you add a new feature if you
could know ahead of time that you'll never be able to design a
datastructure which supports that feature and is not DoS'able by a
remote entity?

Re: [GIT net-next] Open vSwitch

[Pravin Shelar]

On Sat, Aug 2, 2014 at 3:16 PM, David Miller <davem@davemloft.net> wrote:
> From: Pravin B Shelar <pshelar@nicira.com>
> Date: Thu, 31 Jul 2014 16:57:37 -0700
>
>> Following patch adds mask cache so that we do not need to iterate over
>> all entries in mask list on every packet. We have seen good performance
>> improvement with this patch.
>
> How much have you thought about the DoS'ability of openvswitch's
> datastructures?
>
This cache is populated by flow lookup in fast path. The mask cache is
fixed in size. Userspace or number of packets can not change its size.
Memory is statically allocated, no garbage collection. So DoS attack
can not exploit this cache to increase ovs memory footprint.

> What are the upper bounds for performance of packet switching?
>
Cache is keyed on packet RSS. Worst case scenario this cache adds one
extra flow-table lookup for the flow if RSS hash matches but packet
belong to different flow (hash collision).
This is designed to be lightweight, stateless cache (does not take any
reference on other data structures) to have least impact on
DoS'ability of Open vSwitch.

> To be quite honest, a lot of the openvswitch data structures
> adjustments that hit my inbox seem to me to only address specific
> situations that specific user configurations have run into.
>

Overall OVS DoS defense has improved since introduction of mega-flow.
Recently introduced OVS feature allows userspace to set multiple
sockets for upcall processing for given vport. This adds fairness by
separating upcall from different flows to a socket. Userspace process
upcall from these sockets in round-robin fashion. This helps to avoid
one port monopolize upcall communication path.

I agree there is scope for improving DoS defense and we will keep
working on this issue.

> It took us two decades, but we ripped out the ipv4 routing cache
> because external entities could provoke unreasonable worst case
> behavior in routing lookups.
>
> With openvswitch you guys have a unique opportunity to try and design
> all of your features such that they absolutely can use scalable
> datastructures from the beginning that provide reasonable performance
> in the common case and precise upper bounds for any possible sequence
> of incoming packets.
>
> New features tend to blind the developer to the eventual long term
> ramifications on performance.  Would you add a new feature if you
> could know ahead of time that you'll never be able to design a
> datastructure which supports that feature and is not DoS'able by a
> remote entity?
>

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Pravin Shelar <pshelar@nicira.com>
Date: Sun, 3 Aug 2014 12:20:32 -0700

> On Sat, Aug 2, 2014 at 3:16 PM, David Miller <davem@davemloft.net> wrote:
>> From: Pravin B Shelar <pshelar@nicira.com>
>> Date: Thu, 31 Jul 2014 16:57:37 -0700
>>
>>> Following patch adds mask cache so that we do not need to iterate over
>>> all entries in mask list on every packet. We have seen good performance
>>> improvement with this patch.
>>
>> How much have you thought about the DoS'ability of openvswitch's
>> datastructures?
>>
> This cache is populated by flow lookup in fast path. The mask cache is
> fixed in size. Userspace or number of packets can not change its size.
> Memory is statically allocated, no garbage collection. So DoS attack
> can not exploit this cache to increase ovs memory footprint.

An attacker can construct a packet sequence such that every mask cache
lookup misses, making the cache effectively useless.

Re: [GIT net-next] Open vSwitch

[Pravin Shelar]

On Sun, Aug 3, 2014 at 9:21 PM, David Miller <davem@davemloft.net> wrote:
> From: Pravin Shelar <pshelar@nicira.com>
> Date: Sun, 3 Aug 2014 12:20:32 -0700
>
>> On Sat, Aug 2, 2014 at 3:16 PM, David Miller <davem@davemloft.net> wrote:
>>> From: Pravin B Shelar <pshelar@nicira.com>
>>> Date: Thu, 31 Jul 2014 16:57:37 -0700
>>>
>>>> Following patch adds mask cache so that we do not need to iterate over
>>>> all entries in mask list on every packet. We have seen good performance
>>>> improvement with this patch.
>>>
>>> How much have you thought about the DoS'ability of openvswitch's
>>> datastructures?
>>>
>> This cache is populated by flow lookup in fast path. The mask cache is
>> fixed in size. Userspace or number of packets can not change its size.
>> Memory is statically allocated, no garbage collection. So DoS attack
>> can not exploit this cache to increase ovs memory footprint.
>
> An attacker can construct a packet sequence such that every mask cache
> lookup misses, making the cache effectively useless.

Yes, but it does work in normal traffic situations. I have posted
performance numbers in the cover letter.
Under DoS attack as you said attacker need to build sequence of
packets to make cache ineffective. Which results in cache miss and a
full in-kernel flow lookup. Therefore with this cache there is one
more lookup done under DoS. But this is not very different than
current OVS anyways.

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Pravin Shelar <pshelar@nicira.com>
Date: Mon, 4 Aug 2014 12:35:59 -0700

> On Sun, Aug 3, 2014 at 9:21 PM, David Miller <davem@davemloft.net> wrote:
>> From: Pravin Shelar <pshelar@nicira.com>
>> Date: Sun, 3 Aug 2014 12:20:32 -0700
>>
>>> On Sat, Aug 2, 2014 at 3:16 PM, David Miller <davem@davemloft.net> wrote:
>>>> From: Pravin B Shelar <pshelar@nicira.com>
>>>> Date: Thu, 31 Jul 2014 16:57:37 -0700
>>>>
>>>>> Following patch adds mask cache so that we do not need to iterate over
>>>>> all entries in mask list on every packet. We have seen good performance
>>>>> improvement with this patch.
>>>>
>>>> How much have you thought about the DoS'ability of openvswitch's
>>>> datastructures?
>>>>
>>> This cache is populated by flow lookup in fast path. The mask cache is
>>> fixed in size. Userspace or number of packets can not change its size.
>>> Memory is statically allocated, no garbage collection. So DoS attack
>>> can not exploit this cache to increase ovs memory footprint.
>>
>> An attacker can construct a packet sequence such that every mask cache
>> lookup misses, making the cache effectively useless.
> 
> Yes, but it does work in normal traffic situations.

You're basically just reiterating the point I'm trying to make.

Your caches are designed for specific configuration and packet traffic
pattern cases, and can be easily duped into a worse case performance
scenerio by an attacker.

Caches, basically, do not work on the real internet.

Make the fundamental core data structures fast and scalable enough,
rather than bolting caches (which are basically hacks) on top every
time they don't perform to your expectations.

What if you made the full flow lookup fundamentally faster?  Then an
attacker can't do anything about that.  That's a real performance
improvement, one that sustains arbitrary traffic patterns.

Re: [GIT net-next] Open vSwitch

[Alexei Starovoitov]

On Mon, Aug 4, 2014 at 12:42 PM, David Miller <davem@davemloft.net> wrote:
> From: Pravin Shelar <pshelar@nicira.com>
> Date: Mon, 4 Aug 2014 12:35:59 -0700
>
>> On Sun, Aug 3, 2014 at 9:21 PM, David Miller <davem@davemloft.net> wrote:
>>> An attacker can construct a packet sequence such that every mask cache
>>> lookup misses, making the cache effectively useless.
>>
>> Yes, but it does work in normal traffic situations.
>
> You're basically just reiterating the point I'm trying to make.
>
> Your caches are designed for specific configuration and packet traffic
> pattern cases, and can be easily duped into a worse case performance
> scenerio by an attacker.
>
> Caches, basically, do not work on the real internet.
>
> Make the fundamental core data structures fast and scalable enough,
> rather than bolting caches (which are basically hacks) on top every
> time they don't perform to your expectations.
>
> What if you made the full flow lookup fundamentally faster?  Then an

I suspect that the flow lookup in ovs is as fast as it can be, yet
ovs is still dos-able, since kernel datapath (flow lookup and action)
is considered to be first level cache for user space. By design flow
miss is always punted to userspace. Therefore netperf TCP_CRR test
from a VM is not cheap for host userspace component. Mega-flows and
multiple upcall pids are partially addressing this fundamental
problem. Consider simple distributed virtual bridge with VMs
distributed across multiple hosts. Mega-flow mask that selects dmac
can solve CRR case for well behaving VMs, but rogue VM that spams
random dmac will keep taxing host userspace. So we'd need to add
another flow mask to match the rest of traffic unconditionally and
drop it. Now consider virtual bridge-router-bridge topology (two
subnets and router using openstack names). Since VMs on the same
host may be in different subnets their macs can be the same, so
'mega-flow mask dmac' approach won't work and CRR test again is
getting costly to userspace. We can try to use 'in_port + dmac'
mask, but as network topology grows the flow mask tricks get out
of hand. Situation is worse when ovs works as gateway and receives
internet traffic. Since flow miss goes to userspace remote attacker
can find a way to saturate gateway cpu with certain traffic.
I guess none of this is new to ovs and there is probably a solution
that I don't know about.

Re: [GIT net-next] Open vSwitch

[Nicolas Dichtel]

Le 04/08/2014 21:42, David Miller a écrit :
[snip]
> Caches, basically, do not work on the real internet.
A bit late, but I completely agree!

[GIT net-next] Open vSwitch

[Pravin B Shelar]

Following patches adds three features to OVS
1. Add fairness to upcall processing.
2. Recirculation and Hash action.
3. Enable Tunnel GSO features.
Rest of patches are bug fixes related to patches from same series.

The following changes since commit 279f64b7a771d84cbdea51ac2f794becfb06bcd4:

  net/hsr: Remove left-over never-true conditional code. (2014-07-11 15:04:40 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git net_next_ovs

for you to fetch changes up to 3dec4774b6343e5db5d346f1f05c6a883e0069db:

  openvswitch: Add skb_clone NULL check for the sampling action. (2014-07-13 12:02:12 -0700)

----------------------------------------------------------------
Alex Wang (1):
      openvswitch: Allow each vport to have an array of 'port_id's.

Andy Zhou (6):
      openvswitch: Add hash action
      openvswitch: Add recirc action
      openvswitch: Fix key size computation in key_attr_size()
      openvswitch: Avoid memory corruption in queue_userspace_packet()
      openvswitch: Add skb_clone NULL check in the recirc action.
      openvswitch: Add skb_clone NULL check for the sampling action.

Pravin B Shelar (2):
      openvswitch: Enable tunnel GSO for OVS bridge.
      net: Export xmit_recursion

Simon Horman (2):
      openvswitch: Free skb(s) on recirculation error
      openvswitch: Sample action without side effects

 include/linux/netdev_features.h      |   8 +++
 include/linux/netdevice.h            |   3 +
 include/uapi/linux/openvswitch.h     |  39 ++++++++++--
 net/core/dev.c                       |  10 +--
 net/openvswitch/actions.c            | 119 +++++++++++++++++++++++++++++++----
 net/openvswitch/datapath.c           |  79 ++++++++++++++++-------
 net/openvswitch/datapath.h           |   8 ++-
 net/openvswitch/flow.h               |   2 +
 net/openvswitch/flow_netlink.c       |  43 ++++++++++++-
 net/openvswitch/vport-internal_dev.c |   5 +-
 net/openvswitch/vport.c              | 102 +++++++++++++++++++++++++++++-
 net/openvswitch/vport.h              |  27 ++++++--
 12 files changed, 393 insertions(+), 52 deletions(-)

[GIT net-next] Open vSwitch

[Pravin B Shelar]

A set of OVS changes for net-next/3.16.

Most of change are related to improving performance of flow setup by
minimizing critical sections.

The following changes since commit 091b64868b43ed84334c6623ea6a08497529d4ff:

  Merge branch 'mlx4-next' (2014-05-22 17:17:34 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pshelar/openvswitch.git master

for you to fetch changes up to 0c200ef94c9492205e18a18c25650cf27939889c:

  openvswitch: Simplify genetlink code. (2014-05-22 16:27:37 -0700)

----------------------------------------------------------------
Jarno Rajahalme (12):
  openvswitch: Compact sw_flow_key.
  openvswitch: Avoid assigning a NULL pointer to flow actions.
  openvswitch: Clarify locking.
  openvswitch: Build flow cmd netlink reply only if needed.
  openvswitch: Make flow mask removal symmetric.
  openvswitch: Minimize dp and vport critical sections.
  openvswitch: Fix typo.
  openvswitch: Fix ovs_flow_stats_get/clear RCU dereference.
  openvswitch: Reduce locking requirements.
  openvswitch: Minimize ovs_flow_cmd_del critical section.
  openvswitch: Split ovs_flow_cmd_new_or_set().
  openvswitch: Minimize ovs_flow_cmd_new|set critical sections.

Pravin B Shelar (1):
  openvswitch: Simplify genetlink code.

 include/uapi/linux/openvswitch.h |   4 +-
 net/openvswitch/datapath.c       | 771 +++++++++++++++++++++++----------------
 net/openvswitch/flow.c           |  53 ++-
 net/openvswitch/flow.h           |  35 +-
 net/openvswitch/flow_netlink.c   | 112 ++----
 net/openvswitch/flow_table.c     |  46 ++-
 6 files changed, 558 insertions(+), 463 deletions(-)

-- 
1.9.0

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Pravin B Shelar <pshelar@nicira.com>
Date: Tue, 20 May 2014 01:59:38 -0700

> A set of OVS changes for net-next/3.16.
> 
> Most of change are related to improving performance of flow setup by
> minimizing critical sections.

Pulled, thanks Pravin.

In the future please make your postings so that they have the current date
and time when you make the postings, not when the commits when into your
tree.

Otherwise it messed up the order in which your changes appear in patchwork
wrt. other submissions.

Re: [GIT net-next] Open vSwitch

[Pravin Shelar]

On Fri, May 23, 2014 at 11:46 AM, David Miller <davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org> wrote:
> From: Pravin B Shelar <pshelar-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
> Date: Tue, 20 May 2014 01:59:38 -0700
>
>> A set of OVS changes for net-next/3.16.
>>
>> Most of change are related to improving performance of flow setup by
>> minimizing critical sections.
>
> Pulled, thanks Pravin.
>
> In the future please make your postings so that they have the current date
> and time when you make the postings, not when the commits when into your
> tree.
>
> Otherwise it messed up the order in which your changes appear in patchwork
> wrt. other submissions.

ok, I will keep it in mind.

Thanks,
Pravin.

[GIT net-next] Open vSwitch

[Jesse Gross]

A set of OVS changes for net-next/3.16.

The major change here is a switch from per-CPU to per-NUMA flow
statistics. This improves scalability by reducing kernel overhead
in flow setup and maintenance.

The following changes since commit a188a54d11629bef2169052297e61f3767ca8ce5:

  macvlan: simplify the structure port (2014-05-15 23:35:16 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 944df8ae84d88f5e8eb027990dad2cfa4fbe4be5:

  net/openvswitch: Use with RCU_INIT_POINTER(x, NULL) in vport-gre.c (2014-05-16 13:40:29 -0700)

----------------------------------------------------------------
Daniele Di Proietto (4):
      openvswitch: use const in some local vars and casts
      openvswitch: avoid warnings in vport_from_priv
      openvswitch: avoid cast-qual warning in vport_priv
      openvswitch: Added (unsigned long long) cast in printf

Jarno Rajahalme (4):
      openvswitch: Remove 5-tuple optimization.
      openvswitch: Per NUMA node flow stats.
      openvswitch: Fix output of SCTP mask.
      openvswitch: Use TCP flags in the flow key for stats.

Joe Perches (3):
      openvswitch: Use net_ratelimit in OVS_NLERR
      openvswitch: flow_netlink: Use pr_fmt to OVS_NLERR output
      openvswitch: Use ether_addr_copy

Monam Agarwal (1):
      net/openvswitch: Use with RCU_INIT_POINTER(x, NULL) in vport-gre.c

 net/openvswitch/actions.c      |   4 +-
 net/openvswitch/datapath.c     |  11 ++-
 net/openvswitch/datapath.h     |   8 ++-
 net/openvswitch/flow.c         | 149 ++++++++++++++++++++++++-----------------
 net/openvswitch/flow.h         |  18 +++--
 net/openvswitch/flow_netlink.c |  82 +++++------------------
 net/openvswitch/flow_netlink.h |   1 -
 net/openvswitch/flow_table.c   |  75 ++++++++++++---------
 net/openvswitch/flow_table.h   |   4 +-
 net/openvswitch/vport-gre.c    |   2 +-
 net/openvswitch/vport.h        |   6 +-
 11 files changed, 176 insertions(+), 184 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Date: Fri, 16 May 2014 14:07:27 -0700

> A set of OVS changes for net-next/3.16.
> 
> The major change here is a switch from per-CPU to per-NUMA flow
> statistics. This improves scalability by reducing kernel overhead
> in flow setup and maintenance.

Pulled, thanks Jesse.

[GIT net-next] Open vSwitch

[Jesse Gross]

A set of updates for net-next/3.13. Major changes are:
 * Restructure flow handling code to be more logically organized and
   easier to read.
 * Rehashing of the flow table is moved from a workqueue to flow
   installation time. Before, heavy load could block the workqueue for
   excessive periods of time.
 * Additional debugging information is provided to help diagnose megaflows.
 * It's now possible to match on TCP flags.

The following changes since commit 272b98c6455f00884f0350f775c5342358ebb73f:

  Linux 3.12-rc1 (2013-09-16 16:17:51 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 8ddd094675cfd453fc9838caa46ea108a4107183:

  openvswitch: Use flow hash during flow lookup operation. (2013-11-01 18:43:46 -0700)

----------------------------------------------------------------
Andy Zhou (1):
      openvswitch: collect mega flow mask stats

Jarno Rajahalme (2):
      openvswitch: Widen TCP flags handling.
      openvswitch: TCP flags matching support.

Pravin B Shelar (6):
      openvswitch: Move flow table rehashing to flow install.
      openvswitch: Restructure datapath.c and flow.c
      openvswitch: Move mega-flow list out of rehashing struct.
      openvswitch: Simplify mega-flow APIs.
      openvswitch: Enable all GSO features on internal port.
      openvswitch: Use flow hash during flow lookup operation.

Wei Yongjun (2):
      openvswitch: remove duplicated include from vport-vxlan.c
      openvswitch: remove duplicated include from vport-gre.c

 include/uapi/linux/openvswitch.h     |   18 +-
 net/openvswitch/Makefile             |    2 +
 net/openvswitch/datapath.c           |  668 ++------------
 net/openvswitch/datapath.h           |    9 +-
 net/openvswitch/flow.c               | 1605 +--------------------------------
 net/openvswitch/flow.h               |  132 +--
 net/openvswitch/flow_netlink.c       | 1630 ++++++++++++++++++++++++++++++++++
 net/openvswitch/flow_netlink.h       |   60 ++
 net/openvswitch/flow_table.c         |  592 ++++++++++++
 net/openvswitch/flow_table.h         |   81 ++
 net/openvswitch/vport-gre.c          |    2 -
 net/openvswitch/vport-internal_dev.c |    2 +-
 net/openvswitch/vport-vxlan.c        |    1 -
 13 files changed, 2511 insertions(+), 2291 deletions(-)
 create mode 100644 net/openvswitch/flow_netlink.c
 create mode 100644 net/openvswitch/flow_netlink.h
 create mode 100644 net/openvswitch/flow_table.c
 create mode 100644 net/openvswitch/flow_table.h

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Sat,  2 Nov 2013 00:43:39 -0700

> A set of updates for net-next/3.13. Major changes are:
>  * Restructure flow handling code to be more logically organized and
>    easier to read.
>  * Rehashing of the flow table is moved from a workqueue to flow
>    installation time. Before, heavy load could block the workqueue for
>    excessive periods of time.
>  * Additional debugging information is provided to help diagnose megaflows.
>  * It's now possible to match on TCP flags.

Looks good, pulled, thanks Jesse.

[GIT net-next] Open vSwitch

[Jesse Gross]

A set of updates for net-next/3.13. Major changes are:
 * Restructure flow handling code to be more logically organized and
   easier to read.
 * Previously flow state was effectively per-CPU but this is no longer
   true with the addition of wildcarded flows (megaflows). While good
   for flow setup rates, it is bad for stats updates. Stats are now
   per-CPU again to get the best of both worlds.
 * Rehashing of the flow table is moved from a workqueue to flow
   installation time. Before, heavy load could block the workqueue for
   excessive periods of time.
 * Additional debugging information is provided to help diagnose megaflows.
 * It's now possible to match on TCP flags.

The following changes since commit 272b98c6455f00884f0350f775c5342358ebb73f:

  Linux 3.12-rc1 (2013-09-16 16:17:51 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to c7e8b9659587449b21ee68d7aee0cacadf650cce:

  openvswitch: TCP flags matching support. (2013-10-23 02:19:50 -0700)

----------------------------------------------------------------
Andy Zhou (1):
      openvswitch: collect mega flow mask stats

Jarno Rajahalme (2):
      openvswitch: Widen TCP flags handling.
      openvswitch: TCP flags matching support.

Pravin B Shelar (6):
      openvswitch: Move flow table rehashing to flow install.
      openvswitch: Restructure datapath.c and flow.c
      openvswitch: Move mega-flow list out of rehashing struct.
      openvswitch: Simplify mega-flow APIs.
      openvswitch: Per cpu flow stats.
      openvswitch: Enable all GSO features on internal port.

Wei Yongjun (2):
      openvswitch: remove duplicated include from vport-vxlan.c
      openvswitch: remove duplicated include from vport-gre.c

 include/uapi/linux/openvswitch.h     |   18 +-
 net/openvswitch/Makefile             |    2 +
 net/openvswitch/datapath.c           |  721 ++-------------
 net/openvswitch/datapath.h           |    9 +-
 net/openvswitch/flow.c               | 1631 ++--------------------------------
 net/openvswitch/flow.h               |  148 +--
 net/openvswitch/flow_netlink.c       | 1630 +++++++++++++++++++++++++++++++++
 net/openvswitch/flow_netlink.h       |   60 ++
 net/openvswitch/flow_table.c         |  600 +++++++++++++
 net/openvswitch/flow_table.h         |   81 ++
 net/openvswitch/vport-gre.c          |    2 -
 net/openvswitch/vport-internal_dev.c |    2 +-
 net/openvswitch/vport-vxlan.c        |    1 -
 13 files changed, 2590 insertions(+), 2315 deletions(-)
 create mode 100644 net/openvswitch/flow_netlink.c
 create mode 100644 net/openvswitch/flow_netlink.h
 create mode 100644 net/openvswitch/flow_table.c
 create mode 100644 net/openvswitch/flow_table.h

[GIT net-next] Open vSwitch

[Jesse Gross]

A number of significant new features and optimizations for net-next/3.12.
Highlights are:
 * "Megaflows", an optimization that allows userspace to specify which
   flow fields were used to compute the results of the flow lookup.
   This allows for a major reduction in flow setups (the major
   performance bottleneck in Open vSwitch) without reducing flexibility.
 * Converting netlink dump operations to use RCU, allowing for
   additional parallelism in userspace.
 * Matching and modifying SCTP protocol fields.

The following changes since commit 2771399ac9986c75437a83b1c723493cfcdfa439:

  fs_enet: cleanup clock API use (2013-08-22 22:13:54 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 5828cd9a68873df1340b420371c02c47647878fb:

  openvswitch: optimize flow compare and mask functions (2013-08-27 13:13:09 -0700)

----------------------------------------------------------------
Andy Zhou (3):
      openvswitch: Mega flow implementation
      openvswitch: Rename key_len to key_end
      openvswitch: optimize flow compare and mask functions

Cong Wang (1):
      openvswitch: check CONFIG_OPENVSWITCH_GRE in makefile

Jiri Pirko (1):
      openvswitch:: link upper device for port devices

Joe Stringer (2):
      net: Add NEXTHDR_SCTP to ipv6.h
      openvswitch: Add SCTP support

Justin Pettit (1):
      openvswitch: Fix argument descriptions in vport.c.

Pravin B Shelar (3):
      openvswitch: Use RCU lock for flow dump operation.
      openvswitch: Use RCU lock for dp dump operation.
      openvswitch: Use non rcu hlist_del() flow table entry.

 Documentation/networking/openvswitch.txt |   40 +
 include/net/ipv6.h                       |    1 +
 include/uapi/linux/openvswitch.h         |   15 +-
 net/openvswitch/Kconfig                  |    1 +
 net/openvswitch/Makefile                 |    5 +-
 net/openvswitch/actions.c                |   45 +-
 net/openvswitch/datapath.c               |  176 ++--
 net/openvswitch/datapath.h               |    6 +
 net/openvswitch/flow.c                   | 1485 +++++++++++++++++++++---------
 net/openvswitch/flow.h                   |   89 +-
 net/openvswitch/vport-gre.c              |    3 -
 net/openvswitch/vport-netdev.c           |   20 +-
 net/openvswitch/vport.c                  |    3 +-
 13 files changed, 1346 insertions(+), 543 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Tue, 27 Aug 2013 13:20:37 -0700

> A number of significant new features and optimizations for net-next/3.12.
> Highlights are:
>  * "Megaflows", an optimization that allows userspace to specify which
>    flow fields were used to compute the results of the flow lookup.
>    This allows for a major reduction in flow setups (the major
>    performance bottleneck in Open vSwitch) without reducing flexibility.
>  * Converting netlink dump operations to use RCU, allowing for
>    additional parallelism in userspace.
>  * Matching and modifying SCTP protocol fields.

Pulled, thanks Jesse.

[GIT net-next] Open vSwitch

[Jesse Gross]

A few miscellaneous improvements and cleanups before the GRE tunnel
integration series. Intended for net-next/3.11.

The following changes since commit f722406faae2d073cc1d01063d1123c35425939e:

  Linux 3.10-rc1 (2013-05-11 17:14:08 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 93d8fd1514b6862c3370ea92be3f3b4216e0bf8f:

  openvswitch: Simplify interface ovs_flow_metadata_from_nlattrs() (2013-06-14 15:09:12 -0700)

----------------------------------------------------------------
Andy Hill (1):
      openvswitch: Fix misspellings in comments and docs.

Jesse Gross (2):
      openvswitch: Immediately exit on error in ovs_vport_cmd_set().
      openvswitch: Remove unused get_config vport op.

Lorand Jakab (1):
      openvswitch: fix variable names in comment

Pravin B Shelar (4):
      openvswitch: Unify vport error stats handling.
      openvswitch: Fix struct comment.
      openvswitch: make skb->csum consistent with rest of networking stack.
      openvswitch: Simplify interface ovs_flow_metadata_from_nlattrs()

 include/uapi/linux/openvswitch.h     |  1 -
 net/openvswitch/actions.c            |  4 ++++
 net/openvswitch/datapath.c           | 17 ++++++++---------
 net/openvswitch/flow.c               | 29 +++++++++++++++--------------
 net/openvswitch/flow.h               |  4 ++--
 net/openvswitch/vport-internal_dev.c |  1 +
 net/openvswitch/vport-netdev.c       |  7 ++++---
 net/openvswitch/vport-netdev.h       |  1 -
 net/openvswitch/vport.c              | 11 ++++++++---
 net/openvswitch/vport.h              | 13 +++++++++----
 10 files changed, 51 insertions(+), 37 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Fri, 14 Jun 2013 15:28:49 -0700

> A few miscellaneous improvements and cleanups before the GRE tunnel
> integration series. Intended for net-next/3.11.
 ...
>   git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

Pulled, thanks Jesse.

[GIT net-next] Open vSwitch

[Jesse Gross]

A number of improvements for net-next/3.10.

Highlights include:
 * Properly exposing linux/openvswitch.h to userspace after the uapi changes.
 * Simplification of locking. It immediately makes things simpler to reason about and avoids holding RTNL mutex for longer than necessary. In the near future it will also enable tunnel registration and more fine-grained locking.
 * Miscellaneous cleanups and simplifications.

The following changes since commit f498354793d57479d4e1b0f39969acd66737234c:

  qlcnic: Bump up the version to 5.2.39 (2013-03-29 15:51:06 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to e0f0ecf33c3f13401f90bff5afdc3ed1bb40b9af:

  openvswitch: Use generic struct pcpu_tstats. (2013-04-15 14:56:25 -0700)

----------------------------------------------------------------
Andy Zhou (1):
      openvswitch: datapath.h: Fix a stale comment.

Pravin B Shelar (2):
      openvswitch: Simplify datapath locking.
      openvswitch: Use generic struct pcpu_tstats.

Thomas Graf (7):
      openvswitch: Specify the minimal length of OVS_PACKET_ATTR_PACKET in the policy
      openvswitch: Use nla_memcpy() to memcpy() data from attributes
      openvswitch: Refine Netlink message size calculation and kill FLOW_BUFSIZE
      openvswitch: Move common genl notify code into ovs_notify()
      openvswitch: Use ETH_ALEN to define ethernet addresses
      openvswitch: Expose <linux/openvswitch.h> to userspace
      openvswitch: Don't insert empty OVS_VPORT_ATTR_OPTIONS attribute

 include/linux/openvswitch.h          |  432 +-------------------------------
 include/uapi/linux/Kbuild            |    1 +
 include/uapi/linux/openvswitch.h     |  456 ++++++++++++++++++++++++++++++++++
 net/openvswitch/datapath.c           |  393 +++++++++++++++++------------
 net/openvswitch/datapath.h           |   70 ++++--
 net/openvswitch/dp_notify.c          |   82 ++++--
 net/openvswitch/flow.c               |    2 +-
 net/openvswitch/flow.h               |   21 --
 net/openvswitch/vport-internal_dev.c |    6 +
 net/openvswitch/vport-netdev.c       |    8 +-
 net/openvswitch/vport.c              |   58 +++--
 net/openvswitch/vport.h              |   15 +-
 12 files changed, 849 insertions(+), 695 deletions(-)
 create mode 100644 include/uapi/linux/openvswitch.h

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Tue, 16 Apr 2013 14:00:09 -0700

> A number of improvements for net-next/3.10.
> 
> Highlights include:
>  * Properly exposing linux/openvswitch.h to userspace after the uapi changes.
>  * Simplification of locking. It immediately makes things simpler to reason about and avoids holding RTNL mutex for longer than necessary. In the near future it will also enable tunnel registration and more fine-grained locking.
>  * Miscellaneous cleanups and simplifications.

Pulled, but please don't make your email look so silly with those 500
character long lines, make them not exceed 80 columns.

Thanks.

[GIT net-next] Open vSwitch

[Jesse Gross]

A couple of minor enhancements for net-next/3.10.  The largest is an
extension to allow variable length metadata to be passed to userspace
with packets.

There is a merge conflict in net/openvswitch/vport-internal_dev.c:
A existing commit modifies internal_dev_mac_addr() and a new commit
deletes it.  The new one is correct, so you can just remove that function.

The following changes since commit a5a81f0b9025867efb999d14a8dfc1907c5a4c3b:

  ipv6: Fix default route failover when CONFIG_IPV6_ROUTER_PREF=n (2012-12-03 15:34:47 -0500)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 4490108b4a5ada14c7be712260829faecc814ae5:

  openvswitch: Allow OVS_USERSPACE_ATTR_USERDATA to be variable length. (2013-02-22 16:29:22 -0800)

----------------------------------------------------------------
Ben Pfaff (1):
      openvswitch: Allow OVS_USERSPACE_ATTR_USERDATA to be variable length.

Jarno Rajahalme (2):
      linux/openvswitch.h: Make OVSP_LOCAL 32-bit.
      openvswitch: Change ENOENT return value to ENODEV in lookup_vport().

Thomas Graf (2):
      openvswitch: Use eth_mac_addr() instead of duplicating it
      openvswitch: Avoid useless holes in struct vport

 include/linux/openvswitch.h          |   13 +++++++------
 net/openvswitch/datapath.c           |   13 +++++++------
 net/openvswitch/datapath.h           |    2 +-
 net/openvswitch/vport-internal_dev.c |   14 ++------------
 net/openvswitch/vport.h              |    4 ++--
 5 files changed, 19 insertions(+), 27 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Fri, 15 Mar 2013 10:38:46 -0700

> A couple of minor enhancements for net-next/3.10.  The largest is an
> extension to allow variable length metadata to be passed to userspace
> with packets.
> 
> There is a merge conflict in net/openvswitch/vport-internal_dev.c:
> A existing commit modifies internal_dev_mac_addr() and a new commit
> deletes it.  The new one is correct, so you can just remove that function.

Pulled, thanks Jesse.

Thanks, in particular, for the heads up about the merge conflict.

[GIT net-next] Open vSwitch

[Jesse Gross]

This series of improvements for 3.8/net-next contains four components:
 * Support for modifying IPv6 headers
 * Support for matching and setting skb->mark for better integration with
   things like iptables
 * Ability to recognize the EtherType for RARP packets
 * Two small performance enhancements

The movement of ipv6_find_hdr() into exthdrs_core.c causes two small merge
conflicts.  I left it as is but can do the merge if you want.  The conflicts
are:
 * ipv6_find_hdr() and ipv6_find_tlv() were both moved to the bottom of
   exthdrs_core.c.  Both should stay.
 * A new use of ipv6_find_hdr() was added to net/netfilter/ipvs/ip_vs_core.c
   after this patch.  The IPVS user has two instances of the old constant
   name IP6T_FH_F_FRAG which has been renamed to IP6_FH_F_FRAG.

The following changes since commit d04d382980c86bdee9960c3eb157a73f8ed230cc:

  openvswitch: Store flow key len if ARP opcode is not request or reply. (2012-10-30 17:17:09 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 92eb1d477145b2e7780b5002e856f70b8c3d74da:

  openvswitch: Use RCU callback when detaching netdevices. (2012-11-28 14:04:34 -0800)

----------------------------------------------------------------
Ansis Atteka (3):
      ipv6: improve ipv6_find_hdr() to skip empty routing headers
      openvswitch: add ipv6 'set' action
      openvswitch: add skb mark matching and set action

Jesse Gross (2):
      ipv6: Move ipv6_find_hdr() out of Netfilter code.
      openvswitch: Use RCU callback when detaching netdevices.

Mehak Mahajan (1):
      openvswitch: Process RARP packets with ethertype 0x8035 similar to ARP packets.

Shan Wei (1):
      net: openvswitch: use this_cpu_ptr per-cpu helper

 include/linux/netfilter_ipv6/ip6_tables.h |    9 ---
 include/linux/openvswitch.h               |    1 +
 include/net/ipv6.h                        |   10 +++
 net/ipv6/exthdrs_core.c                   |  123 +++++++++++++++++++++++++++++
 net/ipv6/netfilter/ip6_tables.c           |  103 ------------------------
 net/netfilter/xt_HMARK.c                  |    8 +-
 net/openvswitch/actions.c                 |   97 +++++++++++++++++++++++
 net/openvswitch/datapath.c                |   27 ++++++-
 net/openvswitch/flow.c                    |   28 ++++++-
 net/openvswitch/flow.h                    |    8 +-
 net/openvswitch/vport-netdev.c            |   14 +++-
 net/openvswitch/vport-netdev.h            |    3 +
 net/openvswitch/vport.c                   |    5 +-
 13 files changed, 304 insertions(+), 132 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse@nicira.com>
Date: Thu, 29 Nov 2012 10:35:42 -0800

> This series of improvements for 3.8/net-next contains four components:
>  * Support for modifying IPv6 headers
>  * Support for matching and setting skb->mark for better integration with
>    things like iptables
>  * Ability to recognize the EtherType for RARP packets
>  * Two small performance enhancements
> 
> The movement of ipv6_find_hdr() into exthdrs_core.c causes two small merge
> conflicts.  I left it as is but can do the merge if you want.  The conflicts
> are:
>  * ipv6_find_hdr() and ipv6_find_tlv() were both moved to the bottom of
>    exthdrs_core.c.  Both should stay.
>  * A new use of ipv6_find_hdr() was added to net/netfilter/ipvs/ip_vs_core.c
>    after this patch.  The IPVS user has two instances of the old constant
>    name IP6T_FH_F_FRAG which has been renamed to IP6_FH_F_FRAG.

Pulled, thanks Jesse.

The merge conflict directions were particularly helpful.

If you ever do the merge yourself (I'm ambivalent about where you or I
do it), make sure you force the merge commit message to have a
description of the conflict resolution similarly to what you provided
here.

Thanks again.

[GIT net-next] Open vSwitch

[Jesse Gross]

Two feature additions to Open vSwitch for net-next/3.7.

The following changes since commit 0d7614f09c1ebdbaa1599a5aba7593f147bf96ee:

  Linux 3.6-rc1 (2012-08-02 16:38:10 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to 15eac2a74277bc7de68a7c2a64a7c91b4b6f5961:

  openvswitch: Increase maximum number of datapath ports. (2012-09-03 19:20:49 -0700)

----------------------------------------------------------------
Pravin B Shelar (2):
      openvswitch: Add support for network namespaces.
      openvswitch: Increase maximum number of datapath ports.

 net/openvswitch/actions.c            |    2 +-
 net/openvswitch/datapath.c           |  375 +++++++++++++++++++++-------------
 net/openvswitch/datapath.h           |   50 ++++-
 net/openvswitch/dp_notify.c          |    8 +-
 net/openvswitch/flow.c               |   11 +-
 net/openvswitch/flow.h               |    3 +-
 net/openvswitch/vport-internal_dev.c |    7 +-
 net/openvswitch/vport-netdev.c       |    2 +-
 net/openvswitch/vport.c              |   23 ++-
 net/openvswitch/vport.h              |    7 +-
 10 files changed, 317 insertions(+), 171 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Date: Tue,  4 Sep 2012 12:14:07 -0700

> Two feature additions to Open vSwitch for net-next/3.7.
> 
> The following changes since commit 0d7614f09c1ebdbaa1599a5aba7593f147bf96ee:
> 
>   Linux 3.6-rc1 (2012-08-02 16:38:10 -0700)
> 
> are available in the git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master
> 
> for you to fetch changes up to 15eac2a74277bc7de68a7c2a64a7c91b4b6f5961:
> 
>   openvswitch: Increase maximum number of datapath ports. (2012-09-03 19:20:49 -0700)

Pulled, thanks Jesse.

[GIT net-next] Open vSwitch

[Jesse Gross]

A few bug fixes and small enhancements for net-next/3.6.

The following changes since commit bf32fecdc1851ad9ca960f56771b798d17c26cf1:

  openvswitch: Add length check when retrieving TCP flags. (2012-04-02 14:28:57 -0700)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

for you to fetch changes up to efaac3bf087b1a6cec28f2a041e01c874d65390c:

  openvswitch: Fix typo in documentation. (2012-07-20 14:51:07 -0700)

----------------------------------------------------------------
Ansis Atteka (1):
      openvswitch: Do not send notification if ovs_vport_set_options() failed

Ben Pfaff (1):
      openvswitch: Check gso_type for correct sk_buff in queue_gso_packets().

Jesse Gross (2):
      openvswitch: Enable retrieval of TCP flags from IPv6 traffic.
      openvswitch: Reset upper layer protocol info on internal devices.

Leo Alterman (1):
      openvswitch: Fix typo in documentation.

Pravin B Shelar (1):
      openvswitch: Check currect return value from skb_gso_segment()

Raju Subramanian (1):
      openvswitch: Replace Nicira Networks.

 Documentation/networking/openvswitch.txt |    2 +-
 net/openvswitch/actions.c                |    2 +-
 net/openvswitch/datapath.c               |   13 ++++++++-----
 net/openvswitch/datapath.h               |    2 +-
 net/openvswitch/dp_notify.c              |    2 +-
 net/openvswitch/flow.c                   |    5 +++--
 net/openvswitch/flow.h                   |    2 +-
 net/openvswitch/vport-internal_dev.c     |   10 +++++++++-
 net/openvswitch/vport-internal_dev.h     |    2 +-
 net/openvswitch/vport-netdev.c           |    2 +-
 net/openvswitch/vport-netdev.h           |    2 +-
 net/openvswitch/vport.c                  |    2 +-
 net/openvswitch/vport.h                  |    2 +-
 13 files changed, 30 insertions(+), 18 deletions(-)

Re: [GIT net-next] Open vSwitch

[David Miller]

From: Jesse Gross <jesse-l0M0P4e3n4LQT0dZR+AlfA@public.gmane.org>
Date: Fri, 20 Jul 2012 15:26:43 -0700

> A few bug fixes and small enhancements for net-next/3.6.
> 
> The following changes since commit bf32fecdc1851ad9ca960f56771b798d17c26cf1:
> 
>   openvswitch: Add length check when retrieving TCP flags. (2012-04-02 14:28:57 -0700)
> 
> are available in the git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git master

Pulled, thanks Jesse.