Skip to contentSkip to conversation

bridge: vlan: add per-vlan struct and move to rhashtables

PATCH In review Nikolay Aleksandrov 4 participants 4 messages Sep 25, 2015 – Sep 29, 2015

Subsystems (3)
lore
Share & tools
lore

Nikolay Aleksandrov

author maintainer
#0 lore raw

PATCH 1/1 bridge: vlan: add per-vlan struct and move to rhashtables

From: Nikolay Aleksandrov <[email protected]>

This patch changes the bridge vlan implementation to use rhashtables instead of bitmaps. The main motivation behind this change is that we need extensible per-vlan structures (both per-port and global) so more advanced features can be introduced and the vlan support can be extended. I've tried to break this up but the moment net_port_vlans is changed and the whole API goes away, thus this is a larger patch. A few short goals of this patch are:
- Extensible per-vlan structs stored in rhashtables and a sorted list
- Keep user-visible behaviour (compressed vlans etc)
- Keep fastpath ingress/egress logic the same (optimizations to come later)

Here's a brief list of some of the new features we'd like to introduce:
- per-vlan counters
- vlan ingress/egress mapping
- per-vlan igmp configuration
- vlan priorities
- avoid fdb entries replication (e.g. local fdb scaling issues)

The structure is kept single for both global and per-port entries so to avoid code duplication where possible and also because we'll soon introduce "port0 / aka bridge as port" which should simplify things further (thanks to Vlad for the suggestion!).

Now we have per-vlan global rhashtable (bridge-wide) and per-vlan port rhashtable, if an entry is added to a port it'll get a pointer to its global context so it can be quickly accessed later. There's also a sorted vlan list which is used for stable walks and some user-visible behaviour such as the vlan ranges, also for error paths. VLANs are stored in a "vlan group" which currently contains the rhashtable, sorted vlan list and the number of "real" vlan entries. A good side-effect of this change is that it resembles how hw keeps per-vlan data.
One important note after this change is that if a VLAN is being looked up in the bridge's rhashtable for filtering purposes (or to check if it's an existing usable entry, not just a global context) then the new helper br_vlan_should_use() needs to be used if the vlan is found. In case the lookup is done only with a port's vlan group, then this check can be skipped.

Things tested so far:
- basic vlan ingress/egress
- pvids
- untagged vlans
- undef CONFIG_BRIDGE_VLAN_FILTERING
- adding/deleting vlans in different scenarios (with/without global ctx, while transmitting traffic, in ranges etc)
- loading/removing the module while having/adding/deleting vlans
- extracting bridge vlan information (user ABI), compressed requests
- adding/deleting fdbs on vlans
- bridge mac change, promisc mode
- default pvid change
- kmemleak ON during the whole time

---
Things that will be done as follow-ups include fastpath optimization and moving the pvid member into the vlan group which would allow to converge a few functions.

 include/uapi/linux/if_bridge.h |   1 +
 net/bridge/br_device.c         |   2 +-
 net/bridge/br_fdb.c            |  76 ++---
 net/bridge/br_forward.c        |  15 +-
 net/bridge/br_input.c          |  10 +-
 net/bridge/br_mdb.c            |  24 +-
 net/bridge/br_netlink.c        | 174 +++++-----
 net/bridge/br_private.h        | 191 ++++++++---
 net/bridge/br_vlan.c           | 699 +++++++++++++++++++++++++----------------
 9 files changed, 732 insertions(+), 460 deletions(-)
include/uapi/linux/if_bridge.h +1 -0
index 3635b7797508..18db14477bdd 100644
@@ -127,6 +127,7 @@ enum {
 #define BRIDGE_VLAN_INFO_UNTAGGED	(1<<2)	/* VLAN egresses untagged */
 #define BRIDGE_VLAN_INFO_RANGE_BEGIN	(1<<3) /* VLAN is start of vlan range */
 #define BRIDGE_VLAN_INFO_RANGE_END	(1<<4) /* VLAN is end of vlan range */
+#define BRIDGE_VLAN_INFO_BRENTRY	(1<<5) /* Global bridge VLAN entry */
 
 struct bridge_vlan_info {
 	__u16 flags;
net/bridge/br_device.c +1 -1
index 2f81624a8257..c915c5b408ea 100644
@@ -56,7 +56,7 @@ netdev_tx_t br_dev_xmit(struct sk_buff *skb, struct net_device *dev)
 	skb_reset_mac_header(skb);
 	skb_pull(skb, ETH_HLEN);
 
-	if (!br_allowed_ingress(br, br_get_vlan_info(br), skb, &vid))
+	if (!br_allowed_ingress(br, skb, &vid))
 		goto out;
 
 	if (is_broadcast_ether_addr(dest))
net/bridge/br_fdb.c +39 -37

193 lines not on this page · raw message

net/bridge/br_forward.c +11 -4
index 48afca729ed7..6d5ed795c3e2 100644
@@ -30,9 +30,11 @@ static int deliver_clone(const struct net_bridge_port *prev,
 static inline int should_deliver(const struct net_bridge_port *p,
 				 const struct sk_buff *skb)
 {
+	struct net_bridge_vlan_group *vg;
+
+	vg = nbp_vlan_group(p);
 	return ((p->flags & BR_HAIRPIN_MODE) || skb->dev != p->dev) &&
-		br_allowed_egress(p->br, nbp_get_vlan_info(p), skb) &&
-		p->state == BR_STATE_FORWARDING;
+		br_allowed_egress(vg, skb) && p->state == BR_STATE_FORWARDING;
 }
 
 int br_dev_queue_push_xmit(struct net *net, struct sock *sk, struct sk_buff *skb)
@@ -76,7 +78,10 @@ EXPORT_SYMBOL_GPL(br_forward_finish);
 
 static void __br_deliver(const struct net_bridge_port *to, struct sk_buff *skb)
 {
-	skb = br_handle_vlan(to->br, nbp_get_vlan_info(to), skb);
+	struct net_bridge_vlan_group *vg;
+
+	vg = nbp_vlan_group(to);
+	skb = br_handle_vlan(to->br, vg, skb);
 	if (!skb)
 		return;
 
@@ -99,6 +104,7 @@ static void __br_deliver(const struct net_bridge_port *to, struct sk_buff *skb)
 
 static void __br_forward(const struct net_bridge_port *to, struct sk_buff *skb)
 {
+	struct net_bridge_vlan_group *vg;
 	struct net_device *indev;
 
 	if (skb_warn_if_lro(skb)) {
@@ -106,7 +112,8 @@ static void __br_forward(const struct net_bridge_port *to, struct sk_buff *skb)
 		return;
 	}
 
-	skb = br_handle_vlan(to->br, nbp_get_vlan_info(to), skb);
+	vg = nbp_vlan_group(to);
+	skb = br_handle_vlan(to->br, vg, skb);
 	if (!skb)
 		return;
 
net/bridge/br_input.c +5 -5
index 223f4040d9df..e27d0dfd2ee9 100644
@@ -36,28 +36,28 @@ static int br_pass_frame_up(struct sk_buff *skb)
 {
 	struct net_device *indev, *brdev = BR_INPUT_SKB_CB(skb)->brdev;
 	struct net_bridge *br = netdev_priv(brdev);
+	struct net_bridge_vlan_group *vg;
 	struct pcpu_sw_netstats *brstats = this_cpu_ptr(br->stats);
-	struct net_port_vlans *pv;
 
 	u64_stats_update_begin(&brstats->syncp);
 	brstats->rx_packets++;
 	brstats->rx_bytes += skb->len;
 	u64_stats_update_end(&brstats->syncp);
 
+	vg = br_vlan_group(br);
 	/* Bridge is just like any other port.  Make sure the
 	 * packet is allowed except in promisc modue when someone
 	 * may be running packet capture.
 	 */
-	pv = br_get_vlan_info(br);
 	if (!(brdev->flags & IFF_PROMISC) &&
-	    !br_allowed_egress(br, pv, skb)) {
+	    !br_allowed_egress(vg, skb)) {
 		kfree_skb(skb);
 		return NET_RX_DROP;
 	}
 
 	indev = skb->dev;
 	skb->dev = brdev;
-	skb = br_handle_vlan(br, pv, skb);
+	skb = br_handle_vlan(br, vg, skb);
 	if (!skb)
 		return NET_RX_DROP;
 
@@ -140,7 +140,7 @@ int br_handle_frame_finish(struct net *net, struct sock *sk, struct sk_buff *skb
 	if (!p || p->state == BR_STATE_DISABLED)
 		goto drop;
 
-	if (!br_allowed_ingress(p->br, nbp_get_vlan_info(p), skb, &vid))
+	if (!nbp_allowed_ingress(p, skb, &vid))
 		goto out;
 
 	/* insert into forwarding database after filtering to avoid spoofing */
net/bridge/br_mdb.c +12 -12
index d747275fad18..cd8deea2d074 100644
@@ -464,11 +464,11 @@ static int __br_mdb_add(struct net *net, struct net_bridge *br,
 static int br_mdb_add(struct sk_buff *skb, struct nlmsghdr *nlh)
 {
 	struct net *net = sock_net(skb->sk);
-	unsigned short vid = VLAN_N_VID;
+	struct net_bridge_vlan_group *vg;
 	struct net_device *dev, *pdev;
 	struct br_mdb_entry *entry;
 	struct net_bridge_port *p;
-	struct net_port_vlans *pv;
+	struct net_bridge_vlan *v;
 	struct net_bridge *br;
 	int err;
 
@@ -489,10 +489,10 @@ static int br_mdb_add(struct sk_buff *skb, struct nlmsghdr *nlh)
 	if (!p || p->br != br || p->state == BR_STATE_DISABLED)
 		return -EINVAL;
 
-	pv = nbp_get_vlan_info(p);
-	if (br_vlan_enabled(br) && pv && entry->vid == 0) {
-		for_each_set_bit(vid, pv->vlan_bitmap, VLAN_N_VID) {
-			entry->vid = vid;
+	vg = nbp_vlan_group(p);
+	if (br_vlan_enabled(br) && vg && entry->vid == 0) {
+		list_for_each_entry(v, &vg->vlan_list, vlist) {
+			entry->vid = v->vid;
 			err = __br_mdb_add(net, br, entry);
 			if (err)
 				break;
@@ -566,11 +566,11 @@ unlock:
 static int br_mdb_del(struct sk_buff *skb, struct nlmsghdr *nlh)
 {
 	struct net *net = sock_net(skb->sk);
-	unsigned short vid = VLAN_N_VID;
+	struct net_bridge_vlan_group *vg;
 	struct net_device *dev, *pdev;
 	struct br_mdb_entry *entry;
 	struct net_bridge_port *p;
-	struct net_port_vlans *pv;
+	struct net_bridge_vlan *v;
 	struct net_bridge *br;
 	int err;
 
@@ -591,10 +591,10 @@ static int br_mdb_del(struct sk_buff *skb, struct nlmsghdr *nlh)
 	if (!p || p->br != br || p->state == BR_STATE_DISABLED)
 		return -EINVAL;
 
-	pv = nbp_get_vlan_info(p);
-	if (br_vlan_enabled(br) && pv && entry->vid == 0) {
-		for_each_set_bit(vid, pv->vlan_bitmap, VLAN_N_VID) {
-			entry->vid = vid;
+	vg = nbp_vlan_group(p);
+	if (br_vlan_enabled(br) && vg && entry->vid == 0) {
+		list_for_each_entry(v, &vg->vlan_list, vlist) {
+			entry->vid = v->vid;
 			err = __br_mdb_del(br, entry);
 			if (!err)
 				__br_mdb_notify(dev, entry, RTM_DELMDB);
net/bridge/br_netlink.c +98 -76

298 lines not on this page · raw message

net/bridge/br_private.h +142 -49

324 lines not on this page · raw message

net/bridge/br_vlan.c +423 -276

971 lines not on this page · raw message

Signature · 1 line
2.4.3
#1 lore raw
Quoted 56 lines · On Fri, 25 Sep 2015 19:00:11 +0200 Nikolay Aleksandrov <[email protected]>…

From: Nikolay Aleksandrov <[email protected]>

This patch changes the bridge vlan implementation to use rhashtables instead of bitmaps. The main motivation behind this change is that we need extensible per-vlan structures (both per-port and global) so more advanced features can be introduced and the vlan support can be extended. I've tried to break this up but the moment net_port_vlans is changed and the whole API goes away, thus this is a larger patch. A few short goals of this patch are:
- Extensible per-vlan structs stored in rhashtables and a sorted list
- Keep user-visible behaviour (compressed vlans etc)
- Keep fastpath ingress/egress logic the same (optimizations to come later)

Here's a brief list of some of the new features we'd like to introduce:
- per-vlan counters
- vlan ingress/egress mapping
- per-vlan igmp configuration
- vlan priorities
- avoid fdb entries replication (e.g. local fdb scaling issues)

The structure is kept single for both global and per-port entries so to avoid code duplication where possible and also because we'll soon introduce "port0 / aka bridge as port" which should simplify things further (thanks to Vlad for the suggestion!).

Now we have per-vlan global rhashtable (bridge-wide) and per-vlan port rhashtable, if an entry is added to a port it'll get a pointer to its global context so it can be quickly accessed later. There's also a sorted vlan list which is used for stable walks and some user-visible behaviour such as the vlan ranges, also for error paths. VLANs are stored in a "vlan group" which currently contains the rhashtable, sorted vlan list and the number of "real" vlan entries. A good side-effect of this change is that it resembles how hw keeps per-vlan data.
One important note after this change is that if a VLAN is being looked up in the bridge's rhashtable for filtering purposes (or to check if it's an existing usable entry, not just a global context) then the new helper br_vlan_should_use() needs to be used if the vlan is found. In case the lookup is done only with a port's vlan group, then this check can be skipped.

Things tested so far:
- basic vlan ingress/egress
- pvids
- untagged vlans
- undef CONFIG_BRIDGE_VLAN_FILTERING
- adding/deleting vlans in different scenarios (with/without global ctx, while transmitting traffic, in ranges etc)
- loading/removing the module while having/adding/deleting vlans
- extracting bridge vlan information (user ABI), compressed requests
- adding/deleting fdbs on vlans
- bridge mac change, promisc mode
- default pvid change
- kmemleak ON during the whole time

Quoted 1 line

I understand the motivation, but this looks like building infrastructure for something that is not in upstream kernel yet.

It also increases the footprint of bridge when vlan filtering is used, and maybe even for the default case.

#2 lore raw
Quoted 59 lines · Stephen Hemminger
Quoted 56 lines · On Fri, 25 Sep 2015 19:00:11 +0200 Nikolay Aleksandrov <[email protected]>…

From: Nikolay Aleksandrov <[email protected]>

This patch changes the bridge vlan implementation to use rhashtables instead of bitmaps. The main motivation behind this change is that we need extensible per-vlan structures (both per-port and global) so more advanced features can be introduced and the vlan support can be extended. I've tried to break this up but the moment net_port_vlans is changed and the whole API goes away, thus this is a larger patch. A few short goals of this patch are:
- Extensible per-vlan structs stored in rhashtables and a sorted list
- Keep user-visible behaviour (compressed vlans etc)
- Keep fastpath ingress/egress logic the same (optimizations to come later)

Here's a brief list of some of the new features we'd like to introduce:
- per-vlan counters
- vlan ingress/egress mapping
- per-vlan igmp configuration
- vlan priorities
- avoid fdb entries replication (e.g. local fdb scaling issues)

The structure is kept single for both global and per-port entries so to avoid code duplication where possible and also because we'll soon introduce "port0 / aka bridge as port" which should simplify things further (thanks to Vlad for the suggestion!).

Now we have per-vlan global rhashtable (bridge-wide) and per-vlan port rhashtable, if an entry is added to a port it'll get a pointer to its global context so it can be quickly accessed later. There's also a sorted vlan list which is used for stable walks and some user-visible behaviour such as the vlan ranges, also for error paths. VLANs are stored in a "vlan group" which currently contains the rhashtable, sorted vlan list and the number of "real" vlan entries. A good side-effect of this change is that it resembles how hw keeps per-vlan data.
One important note after this change is that if a VLAN is being looked up in the bridge's rhashtable for filtering purposes (or to check if it's an existing usable entry, not just a global context) then the new helper br_vlan_should_use() needs to be used if the vlan is found. In case the lookup is done only with a port's vlan group, then this check can be skipped.

Things tested so far:
- basic vlan ingress/egress
- pvids
- untagged vlans
- undef CONFIG_BRIDGE_VLAN_FILTERING
- adding/deleting vlans in different scenarios (with/without global ctx, while transmitting traffic, in ranges etc)
- loading/removing the module while having/adding/deleting vlans
- extracting bridge vlan information (user ABI), compressed requests
- adding/deleting fdbs on vlans
- bridge mac change, promisc mode
- default pvid change
- kmemleak ON during the whole time

Quoted 5 lines
Quoted 1 line

I understand the motivation, but this looks like building infrastructure for something that is not in upstream kernel yet.

Well, the new features I've listed are not yet in the kernel. :-) I'd like to think of it as ground work for extensible version of the vlan code, we can later extend it to any number of new features which is not possible with the current implementation. Any of the above features would require some (either global or per-port) per-vlan structure. This is the first step.

Quoted 2 lines

It also increases the footprint of bridge when vlan filtering is used, and maybe even for the default case.

The default case is configurable, we can make it start with 4 elements as opposed to the current default 64 slots, that's why I chose to use rhashtable.

Cheers,
Nik

maintainer
#3 lore raw
In reply to Nikolay Aleksandrov

From: Nikolay Aleksandrov <[email protected]>
Date: Fri, 25 Sep 2015 19:00:11 +0200

Quoted 2 lines

This patch changes the bridge vlan implementation to use rhashtables instead of bitmaps.

This seems to be taking the code in a good direction, and I'm kinda happy to see more rhashtable users in the tree as well.

So, applied to net-next, thanks.