[PATCH net-next v3 0/4] bridge: per-vlan stats

[PATCH net-next v3 0/4] bridge: per-vlan stats

[Nikolay Aleksandrov]

Hi,
This set adds support for bridge per-vlan statistics.
In order to be able to dump statistics for many vlans we need a way to
continue dumping after reaching maximum size, thus patches 01 and 02 extend
the new stats API with a per-device extended link stats attribute and
callback which can save its local state and continue where it left off
afterwards. I considered using the already existing "fill_xstats" callback
but it gets confusing since we need to separate the linkinfo dump from the
new stats api dump and adding a flag/argument to do that just looks messy.
I don't think the rtnl_link_ops size is an issue, so adding these seemed
like the cleaner approach.

Patches 03 and 04 add the stats support and netlink dump support
respectively. The stats accounting is controlled via a bridge option which
is default off, thus the performance impact is kept minimal.
I've tested this set with both old and modified iproute2, kmemleak on and
some traffic stress tests while adding/removing vlans and ports.

v3:
 - drop the RCU pvid patch and remove one pointer fetch as requested
 - make stats accounting optional with default to off, the option is in the
   same cache line as vlan_proto and vlan_enabled, so it is already fetched
   before the fast path check thus the performance impact is minimal, this
   also allows us to avoid one vlan lookup and return early when using pvid
 - rebased and retested

v2:
 - Improve the error checking, rename lidx to prividx and save the current
   idx user instead of restricting it to one in patch 01
 - squash patch 02 into 01 and remove the restriction
 - add callback descriptions, improve the size calculation and change the
   xstats message structure to have an embedding level per rtnl link type
   so we can avoid one call to get the link type (and thus filter on it)
   and also each link type can now have any number of private attributes
   inside
 - fix a problem where the vlan stats are not dumped if the bridge has 0
   vlans on it but has vlans on the ports, add bridge link type private
   attributes and also add paddings for future extensions to avoid at least
   a few netlink attributes and improve struct alignment
 - drop the is_skb_forwardable argument constifying patch as it's not
   needed anymore, but it's a nice cleanup which I'll send separately

Thank you,
 Nik


Nikolay Aleksandrov (4):
  net: rtnetlink: allow rtnl_fill_statsinfo to save private state
    counter
  net: rtnetlink: add linkxstats callbacks and attribute
  bridge: vlan: learn to count
  bridge: netlink: export per-vlan stats

 include/net/rtnetlink.h        |   7 +++
 include/uapi/linux/if_bridge.h |  18 +++++++
 include/uapi/linux/if_link.h   |  14 ++++++
 net/bridge/br_netlink.c        |  78 ++++++++++++++++++++++++++++-
 net/bridge/br_private.h        |  18 +++++++
 net/bridge/br_sysfs_br.c       |  17 +++++++
 net/bridge/br_vlan.c           | 109 +++++++++++++++++++++++++++++++++++------
 net/core/rtnetlink.c           |  74 +++++++++++++++++++++++-----
 8 files changed, 307 insertions(+), 28 deletions(-)

-- 
2.4.11

[PATCH net-next v3 1/4] net: rtnetlink: allow rtnl_fill_statsinfo to save private state counter

[Nikolay Aleksandrov]

The new prividx argument allows the current dumping device to save a
private state counter which would enable it to continue dumping from
where it left off. And the idxattr is used to save the current idx user
so multiple prividx using attributes can be requested at the same time
as suggested by Roopa Prabhu.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
v3: no change
v2: improve the error check in rtnl_fill_statsinfo, rename lidx to
prividx, squash patch 2 into this one and save the current idx user
instead of restricting only one

 net/core/rtnetlink.c | 44 +++++++++++++++++++++++++++++++-------------
 1 file changed, 31 insertions(+), 13 deletions(-)

diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 5503dfe6a050..de529a20cd18 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -3444,13 +3444,21 @@ out:
 	return err;
 }
 
+static bool stats_attr_valid(unsigned int mask, int attrid, int idxattr)
+{
+	return (mask & IFLA_STATS_FILTER_BIT(attrid)) &&
+	       (!idxattr || idxattr == attrid);
+}
+
 static int rtnl_fill_statsinfo(struct sk_buff *skb, struct net_device *dev,
 			       int type, u32 pid, u32 seq, u32 change,
-			       unsigned int flags, unsigned int filter_mask)
+			       unsigned int flags, unsigned int filter_mask,
+			       int *idxattr, int *prividx)
 {
 	struct if_stats_msg *ifsm;
 	struct nlmsghdr *nlh;
 	struct nlattr *attr;
+	int s_prividx = *prividx;
 
 	ASSERT_RTNL();
 
@@ -3462,7 +3470,7 @@ static int rtnl_fill_statsinfo(struct sk_buff *skb, struct net_device *dev,
 	ifsm->ifindex = dev->ifindex;
 	ifsm->filter_mask = filter_mask;
 
-	if (filter_mask & IFLA_STATS_FILTER_BIT(IFLA_STATS_LINK_64)) {
+	if (stats_attr_valid(filter_mask, IFLA_STATS_LINK_64, *idxattr)) {
 		struct rtnl_link_stats64 *sp;
 
 		attr = nla_reserve_64bit(skb, IFLA_STATS_LINK_64,
@@ -3480,7 +3488,11 @@ static int rtnl_fill_statsinfo(struct sk_buff *skb, struct net_device *dev,
 	return 0;
 
 nla_put_failure:
-	nlmsg_cancel(skb, nlh);
+	/* not a multi message or no progress mean a real error */
+	if (!(flags & NLM_F_MULTI) || s_prividx == *prividx)
+		nlmsg_cancel(skb, nlh);
+	else
+		nlmsg_end(skb, nlh);
 
 	return -EMSGSIZE;
 }
@@ -3494,7 +3506,7 @@ static size_t if_nlmsg_stats_size(const struct net_device *dev,
 {
 	size_t size = 0;
 
-	if (filter_mask & IFLA_STATS_FILTER_BIT(IFLA_STATS_LINK_64))
+	if (stats_attr_valid(filter_mask, IFLA_STATS_LINK_64, 0))
 		size += nla_total_size_64bit(sizeof(struct rtnl_link_stats64));
 
 	return size;
@@ -3503,8 +3515,9 @@ static size_t if_nlmsg_stats_size(const struct net_device *dev,
 static int rtnl_stats_get(struct sk_buff *skb, struct nlmsghdr *nlh)
 {
 	struct net *net = sock_net(skb->sk);
-	struct if_stats_msg *ifsm;
 	struct net_device *dev = NULL;
+	int idxattr = 0, prividx = 0;
+	struct if_stats_msg *ifsm;
 	struct sk_buff *nskb;
 	u32 filter_mask;
 	int err;
@@ -3528,7 +3541,7 @@ static int rtnl_stats_get(struct sk_buff *skb, struct nlmsghdr *nlh)
 
 	err = rtnl_fill_statsinfo(nskb, dev, RTM_NEWSTATS,
 				  NETLINK_CB(skb).portid, nlh->nlmsg_seq, 0,
-				  0, filter_mask);
+				  0, filter_mask, &idxattr, &prividx);
 	if (err < 0) {
 		/* -EMSGSIZE implies BUG in if_nlmsg_stats_size */
 		WARN_ON(err == -EMSGSIZE);
@@ -3542,18 +3555,19 @@ static int rtnl_stats_get(struct sk_buff *skb, struct nlmsghdr *nlh)
 
 static int rtnl_stats_dump(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	int h, s_h, err, s_idx, s_idxattr, s_prividx;
 	struct net *net = sock_net(skb->sk);
+	unsigned int flags = NLM_F_MULTI;
 	struct if_stats_msg *ifsm;
-	int h, s_h;
-	int idx = 0, s_idx;
-	struct net_device *dev;
 	struct hlist_head *head;
-	unsigned int flags = NLM_F_MULTI;
+	struct net_device *dev;
 	u32 filter_mask = 0;
-	int err;
+	int idx = 0;
 
 	s_h = cb->args[0];
 	s_idx = cb->args[1];
+	s_idxattr = cb->args[2];
+	s_prividx = cb->args[3];
 
 	cb->seq = net->dev_base_seq;
 
@@ -3571,7 +3585,8 @@ static int rtnl_stats_dump(struct sk_buff *skb, struct netlink_callback *cb)
 			err = rtnl_fill_statsinfo(skb, dev, RTM_NEWSTATS,
 						  NETLINK_CB(cb->skb).portid,
 						  cb->nlh->nlmsg_seq, 0,
-						  flags, filter_mask);
+						  flags, filter_mask,
+						  &s_idxattr, &s_prividx);
 			/* If we ran out of room on the first message,
 			 * we're in trouble
 			 */
@@ -3579,13 +3594,16 @@ static int rtnl_stats_dump(struct sk_buff *skb, struct netlink_callback *cb)
 
 			if (err < 0)
 				goto out;
-
+			s_prividx = 0;
+			s_idxattr = 0;
 			nl_dump_check_consistent(cb, nlmsg_hdr(skb));
 cont:
 			idx++;
 		}
 	}
 out:
+	cb->args[3] = s_prividx;
+	cb->args[2] = s_idxattr;
 	cb->args[1] = idx;
 	cb->args[0] = h;
 
-- 
2.4.11

[PATCH net-next v3 2/4] net: rtnetlink: add linkxstats callbacks and attribute

[Nikolay Aleksandrov]

Add callbacks to calculate the size and fill link extended statistics
which can be split into multiple messages and are dumped via the new
rtnl stats API (RTM_GETSTATS) with the IFLA_STATS_LINK_XSTATS attribute.
Also add that attribute to the idx mask check since it is expected to
be able to save state and resume dumping (e.g. future bridge per-vlan
stats will be dumped via this attribute and callbacks).
Each link type should nest its private attributes under the per-link type
attribute. This allows to have any number of separated private attributes
and to avoid one call to get the dev link type.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
v3: no change
v2: add callback descriptions and make the size calculation more
accurate, change the netlink xstats message structure with one more
level for each rtnl link type which allows for private link type attributes
and also allows us to avoid 1 call to get the dev link type.

 include/net/rtnetlink.h      |  7 +++++++
 include/uapi/linux/if_link.h | 12 ++++++++++++
 net/core/rtnetlink.c         | 30 ++++++++++++++++++++++++++++++
 3 files changed, 49 insertions(+)

diff --git a/include/net/rtnetlink.h b/include/net/rtnetlink.h
index 2f87c1ba13de..006a7b81d758 100644
--- a/include/net/rtnetlink.h
+++ b/include/net/rtnetlink.h
@@ -47,6 +47,9 @@ static inline int rtnl_msg_family(const struct nlmsghdr *nlh)
  *	@get_num_rx_queues: Function to determine number of receive queues
  *			    to create when creating a new device.
  *	@get_link_net: Function to get the i/o netns of the device
+ *	@get_linkxstats_size: Function to calculate the required room for
+ *			      dumping device-specific extended link stats
+ *	@fill_linkxstats: Function to dump device-specific extended link stats
  */
 struct rtnl_link_ops {
 	struct list_head	list;
@@ -95,6 +98,10 @@ struct rtnl_link_ops {
 						   const struct net_device *dev,
 						   const struct net_device *slave_dev);
 	struct net		*(*get_link_net)(const struct net_device *dev);
+	size_t			(*get_linkxstats_size)(const struct net_device *dev);
+	int			(*fill_linkxstats)(struct sk_buff *skb,
+						   const struct net_device *dev,
+						   int *prividx);
 };
 
 int __rtnl_link_register(struct rtnl_link_ops *ops);
diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h
index d82de331bb6b..8577c0e4116f 100644
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h
@@ -802,6 +802,7 @@ struct if_stats_msg {
 enum {
 	IFLA_STATS_UNSPEC, /* also used as 64bit pad attribute */
 	IFLA_STATS_LINK_64,
+	IFLA_STATS_LINK_XSTATS,
 	__IFLA_STATS_MAX,
 };
 
@@ -809,4 +810,15 @@ enum {
 
 #define IFLA_STATS_FILTER_BIT(ATTR)	(1 << (ATTR - 1))
 
+/* These are embedded into IFLA_STATS_LINK_XSTATS:
+ * [IFLA_STATS_LINK_XSTATS]
+ * -> [LINK_XSTATS_TYPE_xxx]
+ *    -> [rtnl link type specific attributes]
+ */
+enum {
+	LINK_XSTATS_TYPE_UNSPEC,
+	__LINK_XSTATS_TYPE_MAX
+};
+#define LINK_XSTATS_TYPE_MAX (__LINK_XSTATS_TYPE_MAX - 1)
+
 #endif /* _UAPI_LINUX_IF_LINK_H */
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index de529a20cd18..d471f097c739 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -3483,6 +3483,26 @@ static int rtnl_fill_statsinfo(struct sk_buff *skb, struct net_device *dev,
 		dev_get_stats(dev, sp);
 	}
 
+	if (stats_attr_valid(filter_mask, IFLA_STATS_LINK_XSTATS, *idxattr)) {
+		const struct rtnl_link_ops *ops = dev->rtnl_link_ops;
+
+		if (ops && ops->fill_linkxstats) {
+			int err;
+
+			*idxattr = IFLA_STATS_LINK_XSTATS;
+			attr = nla_nest_start(skb,
+					      IFLA_STATS_LINK_XSTATS);
+			if (!attr)
+				goto nla_put_failure;
+
+			err = ops->fill_linkxstats(skb, dev, prividx);
+			nla_nest_end(skb, attr);
+			if (err)
+				goto nla_put_failure;
+			*idxattr = 0;
+		}
+	}
+
 	nlmsg_end(skb, nlh);
 
 	return 0;
@@ -3509,6 +3529,16 @@ static size_t if_nlmsg_stats_size(const struct net_device *dev,
 	if (stats_attr_valid(filter_mask, IFLA_STATS_LINK_64, 0))
 		size += nla_total_size_64bit(sizeof(struct rtnl_link_stats64));
 
+	if (stats_attr_valid(filter_mask, IFLA_STATS_LINK_XSTATS, 0)) {
+		const struct rtnl_link_ops *ops = dev->rtnl_link_ops;
+
+		if (ops && ops->get_linkxstats_size) {
+			size += nla_total_size(ops->get_linkxstats_size(dev));
+			/* for IFLA_STATS_LINK_XSTATS */
+			size += nla_total_size(0);
+		}
+	}
+
 	return size;
 }
 
-- 
2.4.11

[PATCH net-next v3 3/4] bridge: vlan: learn to count

[Nikolay Aleksandrov]

Add support for per-VLAN Tx/Rx statistics. Every global vlan context gets
allocated a per-cpu stats which is then set in each per-port vlan context
for quick access. The br_allowed_ingress() common function is used to
account for Rx packets and the br_handle_vlan() common function is used
to account for Tx packets. Stats accounting is performed only if the
bridge-wide vlan_stats_enabled option is set either via sysfs or netlink.
A struct hole between vlan_enabled and vlan_proto is used for the new
option so it is in the same cache line. Currently it is binary (on/off)
but it is intentionally restricted to exactly 0 and 1 since other values
will be used in the future for different purposes (e.g. per-port stats).

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
v3: make stats accounting optional with default to off
v2: no change

 include/uapi/linux/if_link.h |  1 +
 net/bridge/br_netlink.c      | 13 ++++++-
 net/bridge/br_private.h      | 13 ++++++-
 net/bridge/br_sysfs_br.c     | 17 +++++++++
 net/bridge/br_vlan.c         | 82 ++++++++++++++++++++++++++++++++++++--------
 5 files changed, 110 insertions(+), 16 deletions(-)

diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h
index 8577c0e4116f..cc50261baf59 100644
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h
@@ -272,6 +272,7 @@ enum {
 	IFLA_BR_NF_CALL_ARPTABLES,
 	IFLA_BR_VLAN_DEFAULT_PVID,
 	IFLA_BR_PAD,
+	IFLA_BR_VLAN_STATS_ENABLED,
 	__IFLA_BR_MAX,
 };
 
diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
index 6bae1125e36d..7fba1f018bc9 100644
--- a/net/bridge/br_netlink.c
+++ b/net/bridge/br_netlink.c
@@ -850,6 +850,7 @@ static const struct nla_policy br_policy[IFLA_BR_MAX + 1] = {
 	[IFLA_BR_NF_CALL_IP6TABLES] = { .type = NLA_U8 },
 	[IFLA_BR_NF_CALL_ARPTABLES] = { .type = NLA_U8 },
 	[IFLA_BR_VLAN_DEFAULT_PVID] = { .type = NLA_U16 },
+	[IFLA_BR_VLAN_STATS_ENABLED] = { .type = NLA_U8 },
 };
 
 static int br_changelink(struct net_device *brdev, struct nlattr *tb[],
@@ -921,6 +922,14 @@ static int br_changelink(struct net_device *brdev, struct nlattr *tb[],
 		if (err)
 			return err;
 	}
+
+	if (data[IFLA_BR_VLAN_STATS_ENABLED]) {
+		__u8 vlan_stats = nla_get_u8(data[IFLA_BR_VLAN_STATS_ENABLED]);
+
+		err = br_vlan_set_stats(br, vlan_stats);
+		if (err)
+			return err;
+	}
 #endif
 
 	if (data[IFLA_BR_GROUP_FWD_MASK]) {
@@ -1082,6 +1091,7 @@ static size_t br_get_size(const struct net_device *brdev)
 #ifdef CONFIG_BRIDGE_VLAN_FILTERING
 	       nla_total_size(sizeof(__be16)) +	/* IFLA_BR_VLAN_PROTOCOL */
 	       nla_total_size(sizeof(u16)) +    /* IFLA_BR_VLAN_DEFAULT_PVID */
+	       nla_total_size(sizeof(u8)) +     /* IFLA_BR_VLAN_STATS_ENABLED */
 #endif
 	       nla_total_size(sizeof(u16)) +    /* IFLA_BR_GROUP_FWD_MASK */
 	       nla_total_size(sizeof(struct ifla_bridge_id)) +   /* IFLA_BR_ROOT_ID */
@@ -1167,7 +1177,8 @@ static int br_fill_info(struct sk_buff *skb, const struct net_device *brdev)
 
 #ifdef CONFIG_BRIDGE_VLAN_FILTERING
 	if (nla_put_be16(skb, IFLA_BR_VLAN_PROTOCOL, br->vlan_proto) ||
-	    nla_put_u16(skb, IFLA_BR_VLAN_DEFAULT_PVID, br->default_pvid))
+	    nla_put_u16(skb, IFLA_BR_VLAN_DEFAULT_PVID, br->default_pvid) ||
+	    nla_put_u8(skb, IFLA_BR_VLAN_STATS_ENABLED, br->vlan_stats_enabled))
 		return -EMSGSIZE;
 #endif
 #ifdef CONFIG_BRIDGE_IGMP_SNOOPING
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 1b5d145dfcbf..8b644069a1a1 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -77,12 +77,21 @@ struct bridge_mcast_querier {
 };
 #endif
 
+struct br_vlan_stats {
+	u64 rx_bytes;
+	u64 rx_packets;
+	u64 tx_bytes;
+	u64 tx_packets;
+	struct u64_stats_sync syncp;
+};
+
 /**
  * struct net_bridge_vlan - per-vlan entry
  *
  * @vnode: rhashtable member
  * @vid: VLAN id
  * @flags: bridge vlan flags
+ * @stats: per-cpu VLAN statistics
  * @br: if MASTER flag set, this points to a bridge struct
  * @port: if MASTER flag unset, this points to a port struct
  * @refcnt: if MASTER flag set, this is bumped for each port referencing it
@@ -100,6 +109,7 @@ struct net_bridge_vlan {
 	struct rhash_head		vnode;
 	u16				vid;
 	u16				flags;
+	struct br_vlan_stats __percpu	*stats;
 	union {
 		struct net_bridge	*br;
 		struct net_bridge_port	*port;
@@ -342,6 +352,7 @@ struct net_bridge
 #ifdef CONFIG_BRIDGE_VLAN_FILTERING
 	struct net_bridge_vlan_group	__rcu *vlgrp;
 	u8				vlan_enabled;
+	u8				vlan_stats_enabled;
 	__be16				vlan_proto;
 	u16				default_pvid;
 #endif
@@ -691,6 +702,7 @@ int __br_vlan_filter_toggle(struct net_bridge *br, unsigned long val);
 int br_vlan_filter_toggle(struct net_bridge *br, unsigned long val);
 int __br_vlan_set_proto(struct net_bridge *br, __be16 proto);
 int br_vlan_set_proto(struct net_bridge *br, unsigned long val);
+int br_vlan_set_stats(struct net_bridge *br, unsigned long val);
 int br_vlan_init(struct net_bridge *br);
 int br_vlan_set_default_pvid(struct net_bridge *br, unsigned long val);
 int __br_vlan_set_default_pvid(struct net_bridge *br, u16 pvid);
@@ -880,7 +892,6 @@ static inline struct net_bridge_vlan_group *nbp_vlan_group_rcu(
 {
 	return NULL;
 }
-
 #endif
 
 struct nf_br_ops {
diff --git a/net/bridge/br_sysfs_br.c b/net/bridge/br_sysfs_br.c
index 70bddfd0f3e9..beb47071e38d 100644
--- a/net/bridge/br_sysfs_br.c
+++ b/net/bridge/br_sysfs_br.c
@@ -731,6 +731,22 @@ static ssize_t default_pvid_store(struct device *d,
 	return store_bridge_parm(d, buf, len, br_vlan_set_default_pvid);
 }
 static DEVICE_ATTR_RW(default_pvid);
+
+static ssize_t vlan_stats_enabled_show(struct device *d,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	struct net_bridge *br = to_bridge(d);
+	return sprintf(buf, "%u\n", br->vlan_stats_enabled);
+}
+
+static ssize_t vlan_stats_enabled_store(struct device *d,
+					struct device_attribute *attr,
+					const char *buf, size_t len)
+{
+	return store_bridge_parm(d, buf, len, br_vlan_set_stats);
+}
+static DEVICE_ATTR_RW(vlan_stats_enabled);
 #endif
 
 static struct attribute *bridge_attrs[] = {
@@ -778,6 +794,7 @@ static struct attribute *bridge_attrs[] = {
 	&dev_attr_vlan_filtering.attr,
 	&dev_attr_vlan_protocol.attr,
 	&dev_attr_default_pvid.attr,
+	&dev_attr_vlan_stats_enabled.attr,
 #endif
 	NULL
 };
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index e001152d6ad1..065c35351356 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -162,6 +162,17 @@ static struct net_bridge_vlan *br_vlan_get_master(struct net_bridge *br, u16 vid
 	return masterv;
 }
 
+static void br_master_vlan_rcu_free(struct rcu_head *rcu)
+{
+	struct net_bridge_vlan *v;
+
+	v = container_of(rcu, struct net_bridge_vlan, rcu);
+	WARN_ON(!br_vlan_is_master(v));
+	free_percpu(v->stats);
+	v->stats = NULL;
+	kfree(v);
+}
+
 static void br_vlan_put_master(struct net_bridge_vlan *masterv)
 {
 	struct net_bridge_vlan_group *vg;
@@ -174,7 +185,7 @@ static void br_vlan_put_master(struct net_bridge_vlan *masterv)
 		rhashtable_remove_fast(&vg->vlan_hash,
 				       &masterv->vnode, br_vlan_rht_params);
 		__vlan_del_list(masterv);
-		kfree_rcu(masterv, rcu);
+		call_rcu(&masterv->rcu, br_master_vlan_rcu_free);
 	}
 }
 
@@ -230,6 +241,7 @@ static int __vlan_add(struct net_bridge_vlan *v, u16 flags)
 		if (!masterv)
 			goto out_filt;
 		v->brvlan = masterv;
+		v->stats = masterv->stats;
 	}
 
 	/* Add the dev mac and count the vlan only if it's usable */
@@ -329,6 +341,7 @@ struct sk_buff *br_handle_vlan(struct net_bridge *br,
 			       struct net_bridge_vlan_group *vg,
 			       struct sk_buff *skb)
 {
+	struct br_vlan_stats *stats;
 	struct net_bridge_vlan *v;
 	u16 vid;
 
@@ -355,18 +368,27 @@ struct sk_buff *br_handle_vlan(struct net_bridge *br,
 			return NULL;
 		}
 	}
+	if (br->vlan_stats_enabled) {
+		stats = this_cpu_ptr(v->stats);
+		u64_stats_update_begin(&stats->syncp);
+		stats->tx_bytes += skb->len;
+		stats->tx_packets++;
+		u64_stats_update_end(&stats->syncp);
+	}
+
 	if (v->flags & BRIDGE_VLAN_INFO_UNTAGGED)
 		skb->vlan_tci = 0;
-
 out:
 	return skb;
 }
 
 /* Called under RCU */
-static bool __allowed_ingress(struct net_bridge_vlan_group *vg, __be16 proto,
+static bool __allowed_ingress(const struct net_bridge *br,
+			      struct net_bridge_vlan_group *vg,
 			      struct sk_buff *skb, u16 *vid)
 {
-	const struct net_bridge_vlan *v;
+	struct br_vlan_stats *stats;
+	struct net_bridge_vlan *v;
 	bool tagged;
 
 	BR_INPUT_SKB_CB(skb)->vlan_filtered = true;
@@ -375,7 +397,7 @@ static bool __allowed_ingress(struct net_bridge_vlan_group *vg, __be16 proto,
 	 * HW accelerated vlan tag.
 	 */
 	if (unlikely(!skb_vlan_tag_present(skb) &&
-		     skb->protocol == proto)) {
+		     skb->protocol == br->vlan_proto)) {
 		skb = skb_vlan_untag(skb);
 		if (unlikely(!skb))
 			return false;
@@ -383,7 +405,7 @@ static bool __allowed_ingress(struct net_bridge_vlan_group *vg, __be16 proto,
 
 	if (!br_vlan_get_tag(skb, vid)) {
 		/* Tagged frame */
-		if (skb->vlan_proto != proto) {
+		if (skb->vlan_proto != br->vlan_proto) {
 			/* Protocol-mismatch, empty out vlan_tci for new tag */
 			skb_push(skb, ETH_HLEN);
 			skb = vlan_insert_tag_set_proto(skb, skb->vlan_proto,
@@ -419,7 +441,7 @@ static bool __allowed_ingress(struct net_bridge_vlan_group *vg, __be16 proto,
 		*vid = pvid;
 		if (likely(!tagged))
 			/* Untagged Frame. */
-			__vlan_hwaccel_put_tag(skb, proto, pvid);
+			__vlan_hwaccel_put_tag(skb, br->vlan_proto, pvid);
 		else
 			/* Priority-tagged Frame.
 			 * At this point, We know that skb->vlan_tci had
@@ -428,13 +450,24 @@ static bool __allowed_ingress(struct net_bridge_vlan_group *vg, __be16 proto,
 			 */
 			skb->vlan_tci |= pvid;
 
-		return true;
+		/* if stats are disabled we can avoid the lookup */
+		if (!br->vlan_stats_enabled)
+			return true;
 	}
-
-	/* Frame had a valid vlan tag.  See if vlan is allowed */
 	v = br_vlan_find(vg, *vid);
-	if (v && br_vlan_should_use(v))
-		return true;
+	if (!v || !br_vlan_should_use(v))
+		goto drop;
+
+	if (br->vlan_stats_enabled) {
+		stats = this_cpu_ptr(v->stats);
+		u64_stats_update_begin(&stats->syncp);
+		stats->rx_bytes += skb->len;
+		stats->rx_packets++;
+		u64_stats_update_end(&stats->syncp);
+	}
+
+	return true;
+
 drop:
 	kfree_skb(skb);
 	return false;
@@ -452,7 +485,7 @@ bool br_allowed_ingress(const struct net_bridge *br,
 		return true;
 	}
 
-	return __allowed_ingress(vg, br->vlan_proto, skb, vid);
+	return __allowed_ingress(br, vg, skb, vid);
 }
 
 /* Called under RCU. */
@@ -542,6 +575,11 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 	if (!vlan)
 		return -ENOMEM;
 
+	vlan->stats = netdev_alloc_pcpu_stats(struct br_vlan_stats);
+	if (!vlan->stats) {
+		kfree(vlan);
+		return -ENOMEM;
+	}
 	vlan->vid = vid;
 	vlan->flags = flags | BRIDGE_VLAN_INFO_MASTER;
 	vlan->flags &= ~BRIDGE_VLAN_INFO_PVID;
@@ -549,8 +587,10 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 	if (flags & BRIDGE_VLAN_INFO_BRENTRY)
 		atomic_set(&vlan->refcnt, 1);
 	ret = __vlan_add(vlan, flags);
-	if (ret)
+	if (ret) {
+		free_percpu(vlan->stats);
 		kfree(vlan);
+	}
 
 	return ret;
 }
@@ -711,6 +751,20 @@ int br_vlan_set_proto(struct net_bridge *br, unsigned long val)
 	return __br_vlan_set_proto(br, htons(val));
 }
 
+int br_vlan_set_stats(struct net_bridge *br, unsigned long val)
+{
+	switch (val) {
+	case 0:
+	case 1:
+		br->vlan_stats_enabled = val;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static bool vlan_default_pvid(struct net_bridge_vlan_group *vg, u16 vid)
 {
 	struct net_bridge_vlan *v;
-- 
2.4.11

[PATCH net-next v3 4/4] bridge: netlink: export per-vlan stats

[Nikolay Aleksandrov]

Add a new LINK_XSTATS_TYPE_BRIDGE attribute and implement the
RTM_GETSTATS callbacks for IFLA_STATS_LINK_XSTATS (fill_linkxstats and
get_linkxstats_size) in order to export the per-vlan stats.
The paddings were added because soon these fields will be needed for
per-port per-vlan stats (or something else if someone beats me to it) so
avoiding at least a few more netlink attributes.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
v3: no change
v2: remove unused pvid pointer, fix the case where bridge has 0 vlans
but there're global contexts and move to rtnl link type private
attributes nested into a LINK_XSTATS_TYPE_ attribute. The paddings were
added because soon these fields will be needed for per-port per-vlan
stats (or something else if someone beats me to it) so avoiding at least
a few more netlink attributes.

 include/uapi/linux/if_bridge.h | 18 ++++++++++++
 include/uapi/linux/if_link.h   |  1 +
 net/bridge/br_netlink.c        | 65 ++++++++++++++++++++++++++++++++++++++++++
 net/bridge/br_private.h        |  7 +++++
 net/bridge/br_vlan.c           | 27 ++++++++++++++++++
 5 files changed, 118 insertions(+)

diff --git a/include/uapi/linux/if_bridge.h b/include/uapi/linux/if_bridge.h
index 0536eefff9bf..397d503fdedb 100644
--- a/include/uapi/linux/if_bridge.h
+++ b/include/uapi/linux/if_bridge.h
@@ -134,6 +134,16 @@ struct bridge_vlan_info {
 	__u16 vid;
 };
 
+struct bridge_vlan_xstats {
+	__u64 rx_bytes;
+	__u64 rx_packets;
+	__u64 tx_bytes;
+	__u64 tx_packets;
+	__u16 vid;
+	__u16 pad1;
+	__u32 pad2;
+};
+
 /* Bridge multicast database attributes
  * [MDBA_MDB] = {
  *     [MDBA_MDB_ENTRY] = {
@@ -233,4 +243,12 @@ enum {
 };
 #define MDBA_SET_ENTRY_MAX (__MDBA_SET_ENTRY_MAX - 1)
 
+/* Embedded inside LINK_XSTATS_TYPE_BRIDGE */
+enum {
+	BRIDGE_XSTATS_UNSPEC,
+	BRIDGE_XSTATS_VLAN,
+	__BRIDGE_XSTATS_MAX
+};
+#define BRIDGE_XSTATS_MAX (__BRIDGE_XSTATS_MAX - 1)
+
 #endif /* _UAPI_LINUX_IF_BRIDGE_H */
diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h
index cc50261baf59..ed59cbd6d129 100644
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h
@@ -818,6 +818,7 @@ enum {
  */
 enum {
 	LINK_XSTATS_TYPE_UNSPEC,
+	LINK_XSTATS_TYPE_BRIDGE,
 	__LINK_XSTATS_TYPE_MAX
 };
 #define LINK_XSTATS_TYPE_MAX (__LINK_XSTATS_TYPE_MAX - 1)
diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
index 7fba1f018bc9..a5343c7232bf 100644
--- a/net/bridge/br_netlink.c
+++ b/net/bridge/br_netlink.c
@@ -1234,6 +1234,69 @@ static int br_fill_info(struct sk_buff *skb, const struct net_device *brdev)
 	return 0;
 }
 
+static size_t br_get_linkxstats_size(const struct net_device *dev)
+{
+	struct net_bridge *br = netdev_priv(dev);
+	struct net_bridge_vlan_group *vg;
+	struct net_bridge_vlan *v;
+	int numvls = 0;
+
+	vg = br_vlan_group(br);
+	if (!vg)
+		return 0;
+
+	/* we need to count all, even placeholder entries */
+	list_for_each_entry(v, &vg->vlan_list, vlist)
+		numvls++;
+
+	/* account for the vlans and the link xstats type nest attribute */
+	return numvls * nla_total_size(sizeof(struct bridge_vlan_xstats)) +
+	       nla_total_size(0);
+}
+
+static int br_fill_linkxstats(struct sk_buff *skb, const struct net_device *dev,
+			      int *prividx)
+{
+	struct net_bridge *br = netdev_priv(dev);
+	struct net_bridge_vlan_group *vg;
+	struct net_bridge_vlan *v;
+	struct nlattr *nest;
+	int vl_idx = 0;
+
+	vg = br_vlan_group(br);
+	if (!vg)
+		goto out;
+	nest = nla_nest_start(skb, LINK_XSTATS_TYPE_BRIDGE);
+	if (!nest)
+		return -EMSGSIZE;
+	list_for_each_entry(v, &vg->vlan_list, vlist) {
+		struct bridge_vlan_xstats vxi;
+		struct br_vlan_stats stats;
+
+		if (vl_idx++ < *prividx)
+			continue;
+		memset(&vxi, 0, sizeof(vxi));
+		vxi.vid = v->vid;
+		br_vlan_get_stats(v, &stats);
+		vxi.rx_bytes = stats.rx_bytes;
+		vxi.rx_packets = stats.rx_packets;
+		vxi.tx_bytes = stats.tx_bytes;
+		vxi.tx_packets = stats.tx_packets;
+
+		if (nla_put(skb, BRIDGE_XSTATS_VLAN, sizeof(vxi), &vxi))
+			goto nla_put_failure;
+	}
+	nla_nest_end(skb, nest);
+	*prividx = 0;
+out:
+	return 0;
+
+nla_put_failure:
+	nla_nest_end(skb, nest);
+	*prividx = vl_idx;
+
+	return -EMSGSIZE;
+}
 
 static struct rtnl_af_ops br_af_ops __read_mostly = {
 	.family			= AF_BRIDGE,
@@ -1252,6 +1315,8 @@ struct rtnl_link_ops br_link_ops __read_mostly = {
 	.dellink		= br_dev_delete,
 	.get_size		= br_get_size,
 	.fill_info		= br_fill_info,
+	.fill_linkxstats	= br_fill_linkxstats,
+	.get_linkxstats_size	= br_get_linkxstats_size,
 
 	.slave_maxtype		= IFLA_BRPORT_MAX,
 	.slave_policy		= br_port_policy,
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 8b644069a1a1..d71efe57f67d 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -711,6 +711,8 @@ int nbp_vlan_delete(struct net_bridge_port *port, u16 vid);
 void nbp_vlan_flush(struct net_bridge_port *port);
 int nbp_vlan_init(struct net_bridge_port *port);
 int nbp_get_num_vlan_infos(struct net_bridge_port *p, u32 filter_mask);
+void br_vlan_get_stats(const struct net_bridge_vlan *v,
+		       struct br_vlan_stats *stats);
 
 static inline struct net_bridge_vlan_group *br_vlan_group(
 					const struct net_bridge *br)
@@ -892,6 +894,11 @@ static inline struct net_bridge_vlan_group *nbp_vlan_group_rcu(
 {
 	return NULL;
 }
+
+static inline void br_vlan_get_stats(const struct net_bridge_vlan *v,
+				     struct br_vlan_stats *stats)
+{
+}
 #endif
 
 struct nf_br_ops {
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index 065c35351356..b6de4f457161 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -1054,3 +1054,30 @@ void nbp_vlan_flush(struct net_bridge_port *port)
 	synchronize_rcu();
 	__vlan_group_free(vg);
 }
+
+void br_vlan_get_stats(const struct net_bridge_vlan *v,
+		       struct br_vlan_stats *stats)
+{
+	int i;
+
+	memset(stats, 0, sizeof(*stats));
+	for_each_possible_cpu(i) {
+		u64 rxpackets, rxbytes, txpackets, txbytes;
+		struct br_vlan_stats *cpu_stats;
+		unsigned int start;
+
+		cpu_stats = per_cpu_ptr(v->stats, i);
+		do {
+			start = u64_stats_fetch_begin_irq(&cpu_stats->syncp);
+			rxpackets = cpu_stats->rx_packets;
+			rxbytes = cpu_stats->rx_bytes;
+			txbytes = cpu_stats->tx_bytes;
+			txpackets = cpu_stats->tx_packets;
+		} while (u64_stats_fetch_retry_irq(&cpu_stats->syncp, start));
+
+		stats->rx_packets += rxpackets;
+		stats->rx_bytes += rxbytes;
+		stats->tx_bytes += txbytes;
+		stats->tx_packets += txpackets;
+	}
+}
-- 
2.4.11

Re: [PATCH net-next v3 0/4] bridge: per-vlan stats

[David Miller]

From: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Date: Sat, 30 Apr 2016 10:25:25 +0200

> This set adds support for bridge per-vlan statistics.

This looks a lot better, series applied, thanks!