thr3ads.net - Linux Ethernet Bridging - [Bridge] [RFC net-next 0/9] net: bridge: Forward offloading [Apr 2021]

If this information is useful, please help other people find it:
Share via:

Tobias Waldekranz

2021-Apr-26 17:04 UTC

[Bridge] [RFC net-next 0/9] net: bridge: Forward offloading

## Overview

   vlan1   vlan2
       \   /
   .-----------.
   |    br0    |
   '-----------'
   /   /   \   \
swp0 swp1 swp2 eth0
  :   :   :
  (hwdom 1)

Up to this point, switchdevs have been trusted with offloading
forwarding between bridge ports, e.g. forwarding a unicast from swp0
to swp1 or flooding a broadcast from swp2 to swp1 and swp0. This
series extends forward offloading to include some new classes of
traffic:

- Locally originating flows, i.e. packets that ingress on br0 that are
  to be forwarded to one or several of the ports swp{0,1,2}. Notably
  this also includes routed flows, e.g. a packet ingressing swp0 on
  VLAN 1 which is then routed over to VLAN 2 by the CPU and then
  forwarded to swp1 is "locally originating" from br0's point of
view.

- Flows originating from "foreign" interfaces, i.e. an interface that
  is not offloaded by a particular switchdev instance. This includes
  ports belonging to other switchdev instances. A typical example
  would be flows from eth0 towards swp{0,1,2}.

The bridge still looks up its FDB/MDB as usual and then notifies the
switchdev driver that a particular skb should be offloaded if it
matches one of the classes above. It does so by using the _accel
version of dev_queue_xmit, supplying its own netdev as the
"subordinate" device. The driver can react to the presence of the
subordinate in its .ndo_select_queue in what ever way it needs to make
sure to forward the skb in much the same way that it would for packets
ingressing on regular ports.

Hardware domains to which a particular skb has been forwarded are
recorded so that duplicates are avoided.

The main performance benefit is thus seen on multicast flows. Imagine
for example that:

- An IP camera is connected to swp0 (VLAN 1)

- The CPU is acting as a multicast router, routing the group from VLAN
  1 to VLAN 2.

- There are subscribers for the group in question behind both swp1 and
  swp2 (VLAN 2).

With this offloading in place, the bridge need only send a single skb
to the driver, which will send it to the hardware marked in such a way
that the switch will perform the multicast replication according to
the MDB configuration. Naturally, the number of saved skb_clones
increase linearly with the number of subscribed ports.

As an extra benefit, on mv88e6xxx, this also allows the switch to
perform source address learning on these flows, which avoids having to
sync dynamic FDB entries over slow configuration interfaces like MDIO
to avoid flows directed towards the CPU being flooded as unknown
unicast by the switch.


## RFC

- In general, what do you think about this idea?

- hwdom. What do you think about this terminology? Personally I feel
  that we had too many things called offload_fwd_mark, and that as the
  use of the bridge internal ID (nbp->offload_fwd_mark) expands, it
  might be useful to have a separate term for it.

- .dfwd_{add,del}_station. Am I stretching this abstraction too far,
  and if so do you have any suggestion/preference on how to signal the
  offloading from the bridge down to the switchdev driver?

- The way that flooding is implemented in br_forward.c (lazily cloning
  skbs) means that you have to mark the forwarding as completed very
  early (right after should_deliver in maybe_deliver) in order to
  avoid duplicates. Is there some way to move this decision point to a
  later stage that I am missing?

- BR_MULTICAST_TO_UNICAST. Right now, I expect that this series is not
  compatible with unicast-to-multicast being used on a port. Then
  again, I think that this would also be broken for regular switchdev
  bridge offloading as this flag is not offloaded to the switchdev
  port, so there is no way for the driver to refuse it. Any ideas on
  how to handle this?


## mv88e6xxx Specifics

Since we are now only receiving a single skb for both unicast and
multicast flows, we can tag the packets with the FORWARD command
instead of FROM_CPU. The swich(es) will then forward the packet in
accordance with its ATU, VTU, STU, and PVT configuration - just like
for packets ingressing on user ports.

Crucially, FROM_CPU is still used for:

- Ports in standalone mode.

- Flows that are trapped to the CPU and software-forwarded by a
  bridge. Note that these flows match neither of the classes discussed
  in the overview.

- Packets that are sent directly to a port netdev without going
  through the bridge, e.g. lldpd sending out PDU via an AF_PACKET
  socket.

We thus have a pretty clean separation where the data plane uses
FORWARDs and the control plane uses TO_/FROM_CPU.

The barrier between different bridges is enforced by port based VLANs
on mv88e6xxx, which in essence is a mapping from a source device/port
pair to an allowed set of egress ports. In order to have a FORWARD
frame (which carries a _source_ device/port) correctly mapped by the
PVT, we must use a unique pair for each bridge.

Fortunately, there is typically lots of unused address space in most
switch trees. When was the last time you saw an mv88e6xxx product
using more than 4 chips? Even if you found one with 16 (!) devices,
you would still have room to allocate 16*16 virtual ports to software
bridges.

Therefore, the mv88e6xxx driver will allocate a virtual device/port
pair to each bridge that it offloads. All members of the same bridge
are then configured to allow packets from this virtual port in their
PVTs.

Tobias Waldekranz (9):
  net: dfwd: Constrain existing users to macvlan subordinates
  net: bridge: Disambiguate offload_fwd_mark
  net: bridge: switchdev: Recycle unused hwdoms
  net: bridge: switchdev: Forward offloading
  net: dsa: Track port PVIDs
  net: dsa: Forward offloading
  net: dsa: mv88e6xxx: Allocate a virtual DSA port for each bridge
  net: dsa: mv88e6xxx: Map virtual bridge port in PVT
  net: dsa: mv88e6xxx: Forward offloading

 MAINTAINERS                                   |   1 +
 drivers/net/dsa/mv88e6xxx/Makefile            |   1 +
 drivers/net/dsa/mv88e6xxx/chip.c              |  61 ++++++-
 drivers/net/dsa/mv88e6xxx/dst.c               | 160 ++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/dst.h               |  14 ++
 .../net/ethernet/intel/fm10k/fm10k_netdev.c   |   3 +
 drivers/net/ethernet/intel/i40e/i40e_main.c   |   3 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c |   3 +
 include/linux/dsa/mv88e6xxx.h                 |  13 ++
 include/net/dsa.h                             |  13 ++
 net/bridge/br_forward.c                       |  11 +-
 net/bridge/br_if.c                            |   4 +-
 net/bridge/br_private.h                       |  54 +++++-
 net/bridge/br_switchdev.c                     | 141 +++++++++++----
 net/dsa/port.c                                |  16 +-
 net/dsa/slave.c                               |  36 +++-
 net/dsa/tag_dsa.c                             |  33 +++-
 17 files changed, 510 insertions(+), 57 deletions(-)
 create mode 100644 drivers/net/dsa/mv88e6xxx/dst.c
 create mode 100644 drivers/net/dsa/mv88e6xxx/dst.h
 create mode 100644 include/linux/dsa/mv88e6xxx.h

-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 1/9] net: dfwd: Constrain existing users to macvlan subordinates

The dfwd_add/del_station NDOs are currently only used by the macvlan
subsystem to request L2 forwarding offload from lower devices. In
order add support for other types of devices (like bridges), we
constrain the current users to make sure that the subordinate
requesting the offload is in fact a macvlan.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 drivers/net/ethernet/intel/fm10k/fm10k_netdev.c | 3 +++
 drivers/net/ethernet/intel/i40e/i40e_main.c     | 3 +++
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c   | 3 +++
 3 files changed, 9 insertions(+)

diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
index 2fb52bd6fc0e..4dba6e6a282d 100644
--- a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
+++ b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
@@ -1352,6 +1352,9 @@ static void *fm10k_dfwd_add_station(struct net_device
*dev,
 	int size, i;
 	u16 vid, glort;
 
+	if (!netif_is_macvlan(sdev))
+		return ERR_PTR(-EOPNOTSUPP);
+
 	/* The hardware supported by fm10k only filters on the destination MAC
 	 * address. In order to avoid issues we only support offloading modes
 	 * where the hardware can actually provide the functionality.
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c
b/drivers/net/ethernet/intel/i40e/i40e_main.c
index c2d145a56b5e..b90b79f7ee46 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -7663,6 +7663,9 @@ static void *i40e_fwd_add(struct net_device *netdev,
struct net_device *vdev)
 	struct i40e_fwd_adapter *fwd;
 	int avail_macvlan, ret;
 
+	if (!netif_is_macvlan(vdev))
+		return ERR_PTR(-EOPNOTSUPP);
+
 	if ((pf->flags & I40E_FLAG_DCB_ENABLED)) {
 		netdev_info(netdev, "Macvlans are not supported when DCB is
enabled\n");
 		return ERR_PTR(-EINVAL);
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index c5ec17d19c59..ff5334faf6c5 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -9940,6 +9940,9 @@ static void *ixgbe_fwd_add(struct net_device *pdev, struct
net_device *vdev)
 	int tcs = adapter->hw_tcs ? : 1;
 	int pool, err;
 
+	if (!netif_is_macvlan(vdev))
+		return ERR_PTR(-EOPNOTSUPP);
+
 	if (adapter->xdp_prog) {
 		e_warn(probe, "L2FW offload is not supported with XDP\n");
 		return ERR_PTR(-EINVAL);
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 2/9] net: bridge: Disambiguate offload_fwd_mark

Before this change, four related - but distinct - concepts where named
offload_fwd_mark:

- skb->offload_fwd_mark: Set by the switchdev driver if the underlying
  hardware has already forwarded this frame to the other ports in the
  same hardware domain.

- nbp->offload_fwd_mark: An idetifier used to group ports that share
  the same hardware forwarding domain.

- br->offload_fwd_mark: Counter used to make sure that unique IDs are
  used in cases where a bridge contains ports from multiple hardware
  domains.

- skb->cb->offload_fwd_mark: The hardware domain on which the frame
  ingressed and was forwarded.

Introduce the term "hardware forwarding domain" ("hwdom") in
the
bridge to denote a set of ports with the following property:

    If an skb with skb->offload_fwd_mark set, is received on a port
    belonging to hwdom N, that frame has already been forwarded to all
    other ports in hwdom N.

By decoupling the name from "offload_fwd_mark", we can extend the
term's definition in the future - e.g. to add constraints that
describe expected egress behavior - without overloading the meaning of
"offload_fwd_mark".

- nbp->offload_fwd_mark thus becomes nbp->hwdom.

- br->offload_fwd_mark becomes br->last_hwdom.

- skb->cb->offload_fwd_mark becomes skb->cb->src_hwdom. There is a
  slight change here: Whereas previously this was only set for
  offloaded packets, we now always track the incoming hwdom. As all
  uses where already gated behind checks of skb->offload_fwd_mark,
  this will not introduce any functional change, but it paves the way
  for future changes where the ingressing hwdom must be known both for
  offloaded and non-offloaded frames.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 net/bridge/br_if.c        |  2 +-
 net/bridge/br_private.h   | 10 +++++-----
 net/bridge/br_switchdev.c | 16 ++++++++--------
 3 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index f7d2f472ae24..73fa703f8df5 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -643,7 +643,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev,
 	if (err)
 		goto err5;
 
-	err = nbp_switchdev_mark_set(p);
+	err = nbp_switchdev_hwdom_set(p);
 	if (err)
 		goto err6;
 
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 7ce8a77cc6b6..53248715f631 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -327,7 +327,7 @@ struct net_bridge_port {
 	struct netpoll			*np;
 #endif
 #ifdef CONFIG_NET_SWITCHDEV
-	int				offload_fwd_mark;
+	int				hwdom;
 #endif
 	u16				group_fwd_mask;
 	u16				backup_redirected_cnt;
@@ -472,7 +472,7 @@ struct net_bridge {
 	u32				auto_cnt;
 
 #ifdef CONFIG_NET_SWITCHDEV
-	int offload_fwd_mark;
+	int last_hwdom;
 #endif
 	struct hlist_head		fdb_list;
 
@@ -502,7 +502,7 @@ struct br_input_skb_cb {
 #endif
 
 #ifdef CONFIG_NET_SWITCHDEV
-	int offload_fwd_mark;
+	int src_hwdom;
 #endif
 };
 
@@ -1593,7 +1593,7 @@ static inline void br_sysfs_delbr(struct net_device *dev)
{ return; }
 
 /* br_switchdev.c */
 #ifdef CONFIG_NET_SWITCHDEV
-int nbp_switchdev_mark_set(struct net_bridge_port *p);
+int nbp_switchdev_hwdom_set(struct net_bridge_port *p);
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb);
 bool nbp_switchdev_allowed_egress(const struct net_bridge_port *p,
@@ -1613,7 +1613,7 @@ static inline void br_switchdev_frame_unmark(struct
sk_buff *skb)
 	skb->offload_fwd_mark = 0;
 }
 #else
-static inline int nbp_switchdev_mark_set(struct net_bridge_port *p)
+static inline int nbp_switchdev_hwdom_set(struct net_bridge_port *p)
 {
 	return 0;
 }
diff --git a/net/bridge/br_switchdev.c b/net/bridge/br_switchdev.c
index a5e601e41cb9..bc085077ae71 100644
--- a/net/bridge/br_switchdev.c
+++ b/net/bridge/br_switchdev.c
@@ -8,20 +8,20 @@
 
 #include "br_private.h"
 
-static int br_switchdev_mark_get(struct net_bridge *br, struct net_device *dev)
+static int br_switchdev_hwdom_get(struct net_bridge *br, struct net_device
*dev)
 {
 	struct net_bridge_port *p;
 
 	/* dev is yet to be added to the port list. */
 	list_for_each_entry(p, &br->port_list, list) {
 		if (netdev_port_same_parent_id(dev, p->dev))
-			return p->offload_fwd_mark;
+			return p->hwdom;
 	}
 
-	return ++br->offload_fwd_mark;
+	return ++br->last_hwdom;
 }
 
-int nbp_switchdev_mark_set(struct net_bridge_port *p)
+int nbp_switchdev_hwdom_set(struct net_bridge_port *p)
 {
 	struct netdev_phys_item_id ppid = { };
 	int err;
@@ -35,7 +35,7 @@ int nbp_switchdev_mark_set(struct net_bridge_port *p)
 		return err;
 	}
 
-	p->offload_fwd_mark = br_switchdev_mark_get(p->br, p->dev);
+	p->hwdom = br_switchdev_hwdom_get(p->br, p->dev);
 
 	return 0;
 }
@@ -43,15 +43,15 @@ int nbp_switchdev_mark_set(struct net_bridge_port *p)
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb)
 {
-	if (skb->offload_fwd_mark &&
!WARN_ON_ONCE(!p->offload_fwd_mark))
-		BR_INPUT_SKB_CB(skb)->offload_fwd_mark = p->offload_fwd_mark;
+	if (p->hwdom)
+		BR_INPUT_SKB_CB(skb)->src_hwdom = p->hwdom;
 }
 
 bool nbp_switchdev_allowed_egress(const struct net_bridge_port *p,
 				  const struct sk_buff *skb)
 {
 	return !skb->offload_fwd_mark ||
-	       BR_INPUT_SKB_CB(skb)->offload_fwd_mark != p->offload_fwd_mark;
+	       BR_INPUT_SKB_CB(skb)->src_hwdom != p->hwdom;
 }
 
 /* Flags that can be offloaded to hardware */
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 3/9] net: bridge: switchdev: Recycle unused hwdoms

Since hwdoms has thus far only been used for equality comparisons, the
bridge has used the simplest possible assignment policy; using a
counter to keep track of the last value handed out.

With the upcoming transmit offloading, we need to perform set
operations efficiently based on hwdoms, e.g. we want to answer
questions like "has this skb been forwarded to any port within this
hwdom?"

Move to a bitmap-based allocation scheme that recycles hwdoms once all
members leaves the bridge. This means that we can use a single
unsigned long to keep track of the hwdoms that have received an skb.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 net/bridge/br_if.c        |  4 +-
 net/bridge/br_private.h   | 29 +++++++++---
 net/bridge/br_switchdev.c | 94 ++++++++++++++++++++++++++-------------
 3 files changed, 87 insertions(+), 40 deletions(-)

diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 73fa703f8df5..adaf78e45c23 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -349,6 +349,7 @@ static void del_nbp(struct net_bridge_port *p)
 	nbp_backup_clear(p);
 
 	nbp_update_port_count(br);
+	nbp_switchdev_del(p);
 
 	netdev_upper_dev_unlink(dev, br->dev);
 
@@ -643,7 +644,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev,
 	if (err)
 		goto err5;
 
-	err = nbp_switchdev_hwdom_set(p);
+	err = nbp_switchdev_add(p);
 	if (err)
 		goto err6;
 
@@ -704,6 +705,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev,
 	list_del_rcu(&p->list);
 	br_fdb_delete_by_port(br, p, 0, 1);
 	nbp_update_port_count(br);
+	nbp_switchdev_del(p);
 err6:
 	netdev_upper_dev_unlink(dev, br->dev);
 err5:
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 53248715f631..aba92864d285 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -29,6 +29,8 @@
 
 #define BR_MULTICAST_DEFAULT_HASH_MAX 4096
 
+#define BR_HWDOM_MAX BITS_PER_LONG
+
 #define BR_VERSION	"2.3"
 
 /* Control of forwarding link local multicast */
@@ -54,6 +56,8 @@ typedef struct bridge_id bridge_id;
 typedef struct mac_addr mac_addr;
 typedef __u16 port_id;
 
+typedef DECLARE_BITMAP(br_hwdom_map_t, BR_HWDOM_MAX);
+
 struct bridge_id {
 	unsigned char	prio[2];
 	unsigned char	addr[ETH_ALEN];
@@ -472,7 +476,7 @@ struct net_bridge {
 	u32				auto_cnt;
 
 #ifdef CONFIG_NET_SWITCHDEV
-	int last_hwdom;
+	br_hwdom_map_t			busy_hwdoms;
 #endif
 	struct hlist_head		fdb_list;
 
@@ -1593,7 +1597,6 @@ static inline void br_sysfs_delbr(struct net_device *dev)
{ return; }
 
 /* br_switchdev.c */
 #ifdef CONFIG_NET_SWITCHDEV
-int nbp_switchdev_hwdom_set(struct net_bridge_port *p);
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb);
 bool nbp_switchdev_allowed_egress(const struct net_bridge_port *p,
@@ -1607,17 +1610,15 @@ void br_switchdev_fdb_notify(const struct
net_bridge_fdb_entry *fdb,
 int br_switchdev_port_vlan_add(struct net_device *dev, u16 vid, u16 flags,
 			       struct netlink_ext_ack *extack);
 int br_switchdev_port_vlan_del(struct net_device *dev, u16 vid);
+int nbp_switchdev_add(struct net_bridge_port *p);
+void nbp_switchdev_del(struct net_bridge_port *p);
+void br_switchdev_init(struct net_bridge *br);
 
 static inline void br_switchdev_frame_unmark(struct sk_buff *skb)
 {
 	skb->offload_fwd_mark = 0;
 }
 #else
-static inline int nbp_switchdev_hwdom_set(struct net_bridge_port *p)
-{
-	return 0;
-}
-
 static inline void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 					    struct sk_buff *skb)
 {
@@ -1657,6 +1658,20 @@ br_switchdev_fdb_notify(const struct net_bridge_fdb_entry
*fdb, int type)
 static inline void br_switchdev_frame_unmark(struct sk_buff *skb)
 {
 }
+
+static inline int nbp_switchdev_add(struct net_bridge_port *p)
+{
+	return 0;
+}
+
+static inline void nbp_switchdev_del(struct net_bridge_port *p)
+{
+}
+
+static inline void br_switchdev_init(struct net_bridge *br)
+{
+}
+
 #endif /* CONFIG_NET_SWITCHDEV */
 
 /* br_arp_nd_proxy.c */
diff --git a/net/bridge/br_switchdev.c b/net/bridge/br_switchdev.c
index bc085077ae71..54bd7205bfb5 100644
--- a/net/bridge/br_switchdev.c
+++ b/net/bridge/br_switchdev.c
@@ -8,38 +8,6 @@
 
 #include "br_private.h"
 
-static int br_switchdev_hwdom_get(struct net_bridge *br, struct net_device
*dev)
-{
-	struct net_bridge_port *p;
-
-	/* dev is yet to be added to the port list. */
-	list_for_each_entry(p, &br->port_list, list) {
-		if (netdev_port_same_parent_id(dev, p->dev))
-			return p->hwdom;
-	}
-
-	return ++br->last_hwdom;
-}
-
-int nbp_switchdev_hwdom_set(struct net_bridge_port *p)
-{
-	struct netdev_phys_item_id ppid = { };
-	int err;
-
-	ASSERT_RTNL();
-
-	err = dev_get_port_parent_id(p->dev, &ppid, true);
-	if (err) {
-		if (err == -EOPNOTSUPP)
-			return 0;
-		return err;
-	}
-
-	p->hwdom = br_switchdev_hwdom_get(p->br, p->dev);
-
-	return 0;
-}
-
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb)
 {
@@ -156,3 +124,65 @@ int br_switchdev_port_vlan_del(struct net_device *dev, u16
vid)
 
 	return switchdev_port_obj_del(dev, &v.obj);
 }
+
+static int nbp_switchdev_hwdom_set(struct net_bridge_port *joining)
+{
+	struct net_bridge *br = joining->br;
+	struct net_bridge_port *p;
+	int hwdom;
+
+	/* joining is yet to be added to the port list. */
+	list_for_each_entry(p, &br->port_list, list) {
+		if (netdev_port_same_parent_id(joining->dev, p->dev)) {
+			joining->hwdom = p->hwdom;
+			return 0;
+		}
+	}
+
+	hwdom = find_next_zero_bit(br->busy_hwdoms, BR_HWDOM_MAX, 1);
+	if (hwdom >= BR_HWDOM_MAX)
+		return -EBUSY;
+
+	set_bit(hwdom, br->busy_hwdoms);
+	joining->hwdom = hwdom;
+	return 0;
+}
+
+static void nbp_switchdev_hwdom_put(struct net_bridge_port *leaving)
+{
+	struct net_bridge *br = leaving->br;
+	struct net_bridge_port *p;
+
+	/* leaving is no longer in the port list. */
+	list_for_each_entry(p, &br->port_list, list) {
+		if (p->hwdom == leaving->hwdom)
+			return;
+	}
+
+	clear_bit(leaving->hwdom, br->busy_hwdoms);
+}
+
+int nbp_switchdev_add(struct net_bridge_port *p)
+{
+	struct netdev_phys_item_id ppid = { };
+	int err;
+
+	ASSERT_RTNL();
+
+	err = dev_get_port_parent_id(p->dev, &ppid, true);
+	if (err) {
+		if (err == -EOPNOTSUPP)
+			return 0;
+		return err;
+	}
+
+	return nbp_switchdev_hwdom_set(p);
+}
+
+void nbp_switchdev_del(struct net_bridge_port *p)
+{
+	ASSERT_RTNL();
+
+	if (p->hwdom)
+		nbp_switchdev_hwdom_put(p);
+}
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 4/9] net: bridge: switchdev: Forward offloading

Allow switchdevs to forward frames from the CPU in accordance with the
bridge configuration in the same way as is done between bridge
ports. This means that the bridge will only send a single skb towards
one of the ports under the switchdev's control, and expects the driver
to deliver the packet to all eligible ports in its domain.

Primarily this improves the performance of multicast flows with
multiple subscribers, as it allows the hardware to perform the frame
replication.

The basic flow between the driver and the bridge is as follows:

- The switchdev accepts the offload by returning a non-null pointer
  from .ndo_dfwd_add_station when the port is added to the bridge.

- The bridge sends offloadable skbs to one of the ports under the
  switchdev's control using dev_queue_xmit_accel.

- The switchdev notices the offload by checking for a non-NULL
  "sb_dev" in the core's call to .ndo_select_queue.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 net/bridge/br_forward.c   | 11 +++++++-
 net/bridge/br_private.h   | 27 ++++++++++++++++++
 net/bridge/br_switchdev.c | 59 +++++++++++++++++++++++++++++++++++++--
 3 files changed, 93 insertions(+), 4 deletions(-)

diff --git a/net/bridge/br_forward.c b/net/bridge/br_forward.c
index 6e9b049ae521..b4fb3b0bb1ec 100644
--- a/net/bridge/br_forward.c
+++ b/net/bridge/br_forward.c
@@ -32,6 +32,8 @@ static inline int should_deliver(const struct net_bridge_port
*p,
 
 int br_dev_queue_push_xmit(struct net *net, struct sock *sk, struct sk_buff
*skb)
 {
+	struct net_device *sb_dev = NULL;
+
 	skb_push(skb, ETH_HLEN);
 	if (!is_skb_forwardable(skb->dev, skb))
 		goto drop;
@@ -48,7 +50,10 @@ int br_dev_queue_push_xmit(struct net *net, struct sock *sk,
struct sk_buff *skb
 		skb_set_network_header(skb, depth);
 	}
 
-	dev_queue_xmit(skb);
+	if (br_switchdev_accels_skb(skb))
+		sb_dev = BR_INPUT_SKB_CB(skb)->brdev;
+
+	dev_queue_xmit_accel(skb, sb_dev);
 
 	return 0;
 
@@ -105,6 +110,8 @@ static void __br_forward(const struct net_bridge_port *to,
 		indev = NULL;
 	}
 
+	nbp_switchdev_frame_mark_accel(to, skb);
+
 	NF_HOOK(NFPROTO_BRIDGE, br_hook,
 		net, NULL, skb, indev, skb->dev,
 		br_forward_finish);
@@ -174,6 +181,8 @@ static struct net_bridge_port *maybe_deliver(
 	if (!should_deliver(p, skb))
 		return prev;
 
+	nbp_switchdev_frame_mark_fwd(p, skb);
+
 	if (!prev)
 		goto out;
 
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index aba92864d285..933e951b0d7a 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -332,6 +332,7 @@ struct net_bridge_port {
 #endif
 #ifdef CONFIG_NET_SWITCHDEV
 	int				hwdom;
+	void				*accel_priv;
 #endif
 	u16				group_fwd_mask;
 	u16				backup_redirected_cnt;
@@ -506,7 +507,9 @@ struct br_input_skb_cb {
 #endif
 
 #ifdef CONFIG_NET_SWITCHDEV
+	u8 fwd_accel:1;
 	int src_hwdom;
+	br_hwdom_map_t fwd_hwdoms;
 #endif
 };
 
@@ -1597,6 +1600,15 @@ static inline void br_sysfs_delbr(struct net_device *dev)
{ return; }
 
 /* br_switchdev.c */
 #ifdef CONFIG_NET_SWITCHDEV
+static inline bool br_switchdev_accels_skb(struct sk_buff *skb)
+{
+	return BR_INPUT_SKB_CB(skb)->fwd_accel;
+}
+
+void nbp_switchdev_frame_mark_accel(const struct net_bridge_port *p,
+				    struct sk_buff *skb);
+void nbp_switchdev_frame_mark_fwd(const struct net_bridge_port *p,
+				  struct sk_buff *skb);
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb);
 bool nbp_switchdev_allowed_egress(const struct net_bridge_port *p,
@@ -1619,6 +1631,21 @@ static inline void br_switchdev_frame_unmark(struct
sk_buff *skb)
 	skb->offload_fwd_mark = 0;
 }
 #else
+static inline bool br_switchdev_accels_skb(struct sk_buff *skb)
+{
+	return false;
+}
+
+static inline void nbp_switchdev_frame_mark_accel(const struct net_bridge_port
*p,
+						  struct sk_buff *skb)
+{
+}
+
+static inline void nbp_switchdev_frame_mark_fwd(const struct net_bridge_port
*p,
+						struct sk_buff *skb)
+{
+}
+
 static inline void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 					    struct sk_buff *skb)
 {
diff --git a/net/bridge/br_switchdev.c b/net/bridge/br_switchdev.c
index 54bd7205bfb5..c903171ad291 100644
--- a/net/bridge/br_switchdev.c
+++ b/net/bridge/br_switchdev.c
@@ -8,6 +8,26 @@
 
 #include "br_private.h"
 
+static bool nbp_switchdev_can_accel(const struct net_bridge_port *p,
+				    const struct sk_buff *skb)
+{
+	return p->accel_priv && (p->hwdom !=
BR_INPUT_SKB_CB(skb)->src_hwdom);
+}
+
+void nbp_switchdev_frame_mark_accel(const struct net_bridge_port *p,
+				    struct sk_buff *skb)
+{
+	if (nbp_switchdev_can_accel(p, skb))
+		BR_INPUT_SKB_CB(skb)->fwd_accel = true;
+}
+
+void nbp_switchdev_frame_mark_fwd(const struct net_bridge_port *p,
+				  struct sk_buff *skb)
+{
+	if (nbp_switchdev_can_accel(p, skb))
+		set_bit(p->hwdom, BR_INPUT_SKB_CB(skb)->fwd_hwdoms);
+}
+
 void nbp_switchdev_frame_mark(const struct net_bridge_port *p,
 			      struct sk_buff *skb)
 {
@@ -18,8 +38,10 @@ void nbp_switchdev_frame_mark(const struct net_bridge_port
*p,
 bool nbp_switchdev_allowed_egress(const struct net_bridge_port *p,
 				  const struct sk_buff *skb)
 {
-	return !skb->offload_fwd_mark ||
-	       BR_INPUT_SKB_CB(skb)->src_hwdom != p->hwdom;
+	struct br_input_skb_cb *cb = BR_INPUT_SKB_CB(skb);
+
+	return !test_bit(p->hwdom, cb->fwd_hwdoms) &&
+		(!skb->offload_fwd_mark || cb->src_hwdom != p->hwdom);
 }
 
 /* Flags that can be offloaded to hardware */
@@ -125,6 +147,27 @@ int br_switchdev_port_vlan_del(struct net_device *dev, u16
vid)
 	return switchdev_port_obj_del(dev, &v.obj);
 }
 
+static void nbp_switchdev_fwd_offload_add(struct net_bridge_port *p)
+{
+	void *priv;
+
+	if (!(p->dev->features & NETIF_F_HW_L2FW_DOFFLOAD))
+		return;
+
+	priv = p->dev->netdev_ops->ndo_dfwd_add_station(p->dev,
p->br->dev);
+	if (!IS_ERR_OR_NULL(priv))
+		p->accel_priv = priv;
+}
+
+static void nbp_switchdev_fwd_offload_del(struct net_bridge_port *p)
+{
+	if (!p->accel_priv)
+		return;
+
+	p->dev->netdev_ops->ndo_dfwd_del_station(p->dev,
p->accel_priv);
+	p->accel_priv = NULL;
+}
+
 static int nbp_switchdev_hwdom_set(struct net_bridge_port *joining)
 {
 	struct net_bridge *br = joining->br;
@@ -176,13 +219,23 @@ int nbp_switchdev_add(struct net_bridge_port *p)
 		return err;
 	}
 
-	return nbp_switchdev_hwdom_set(p);
+	err = nbp_switchdev_hwdom_set(p);
+	if (err)
+		return err;
+
+	if (p->hwdom)
+		nbp_switchdev_fwd_offload_add(p);
+
+	return 0;
 }
 
 void nbp_switchdev_del(struct net_bridge_port *p)
 {
 	ASSERT_RTNL();
 
+	if (p->accel_priv)
+		nbp_switchdev_fwd_offload_del(p);
+
 	if (p->hwdom)
 		nbp_switchdev_hwdom_put(p);
 }
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 5/9] net: dsa: Track port PVIDs

In some scenarios a tagger must know which VLAN to assign to a packet,
even if the packet is set to egress untagged. Since the VLAN
information in the skb will be removed by the bridge in this case,
track each port's PVID such that the VID of an outgoing frame can
always be determined.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 include/net/dsa.h |  1 +
 net/dsa/port.c    | 16 ++++++++++++++--
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/include/net/dsa.h b/include/net/dsa.h
index 507082959aa4..1f9ba9889034 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -270,6 +270,7 @@ struct dsa_port {
 	unsigned int		ageing_time;
 	bool			vlan_filtering;
 	u8			stp_state;
+	u16			pvid;
 	struct net_device	*bridge_dev;
 	struct devlink_port	devlink_port;
 	bool			devlink_port_setup;
diff --git a/net/dsa/port.c b/net/dsa/port.c
index 6379d66a6bb3..02d96aebfcc6 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -651,8 +651,14 @@ int dsa_port_vlan_add(struct dsa_port *dp,
 		.vlan = vlan,
 		.extack = extack,
 	};
+	int err;
+
+	err = dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info);
 
-	return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info);
+	if (!err && (vlan->flags & BRIDGE_VLAN_INFO_PVID))
+		dp->pvid = vlan->vid;
+
+	return err;
 }
 
 int dsa_port_vlan_del(struct dsa_port *dp,
@@ -663,8 +669,14 @@ int dsa_port_vlan_del(struct dsa_port *dp,
 		.port = dp->index,
 		.vlan = vlan,
 	};
+	int err;
+
+	err = dsa_port_notify(dp, DSA_NOTIFIER_VLAN_DEL, &info);
 
-	return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_DEL, &info);
+	if (!err && vlan->vid == dp->pvid)
+		dp->pvid = 0;
+
+	return err;
 }
 
 int dsa_port_mrp_add(const struct dsa_port *dp,
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 6/9] net: dsa: Forward offloading

Allow DSA drivers to support forward offloading from a bridge by:

- Passing calls to .ndo_dfwd_{add,del}_station to the drivers.

- Recording the subordinate device of offloaded skbs in the control
  buffer so that the tagger can take the appropriate action.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 include/net/dsa.h |  7 +++++++
 net/dsa/slave.c   | 36 ++++++++++++++++++++++++++++++++++--
 2 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/include/net/dsa.h b/include/net/dsa.h
index 1f9ba9889034..77d4df819299 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -119,6 +119,7 @@ struct dsa_netdevice_ops {
 
 struct dsa_skb_cb {
 	struct sk_buff *clone;
+	struct net_device *sb_dev;
 };
 
 struct __dsa_skb_cb {
@@ -828,6 +829,12 @@ struct dsa_switch_ops {
 					  const struct switchdev_obj_ring_role_mrp *mrp);
 	int	(*port_mrp_del_ring_role)(struct dsa_switch *ds, int port,
 					  const struct switchdev_obj_ring_role_mrp *mrp);
+
+	/* L2 forward offloading */
+	void *	(*dfwd_add_station)(struct dsa_switch *ds, int port,
+				    struct net_device *sb_dev);
+	void	(*dfwd_del_station)(struct dsa_switch *ds, int port,
+				    struct net_device *sb_dev);
 };
 
 #define DSA_DEVLINK_PARAM_DRIVER(_id, _name, _type, _cmodes)		\
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index 77b33bd161b8..3689ffa2dbb8 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -657,6 +657,13 @@ static netdev_tx_t dsa_slave_xmit(struct sk_buff *skb,
struct net_device *dev)
 	return dsa_enqueue_skb(nskb, dev);
 }
 
+static u16 dsa_slave_select_queue(struct net_device *dev, struct sk_buff *skb,
+				  struct net_device *sb_dev)
+{
+	DSA_SKB_CB(skb)->sb_dev = sb_dev;
+	return netdev_pick_tx(dev, skb, sb_dev);
+}
+
 /* ethtool operations *******************************************************/
 
 static void dsa_slave_get_drvinfo(struct net_device *dev,
@@ -1708,10 +1715,33 @@ static int dsa_slave_fill_forward_path(struct
net_device_path_ctx *ctx,
 	return 0;
 }
 
+static void *dsa_slave_dfwd_add_station(struct net_device *dev,
+					struct net_device *sb_dev)
+{
+	struct dsa_port *dp = dsa_slave_to_port(dev);
+	struct dsa_switch *ds = dp->ds;
+
+	if (ds->ops->dfwd_add_station)
+		return ds->ops->dfwd_add_station(ds, dp->index, sb_dev);
+
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static void dsa_slave_dfwd_del_station(struct net_device *dev,
+				       void *sb_dev)
+{
+	struct dsa_port *dp = dsa_slave_to_port(dev);
+	struct dsa_switch *ds = dp->ds;
+
+	if (ds->ops->dfwd_del_station)
+		ds->ops->dfwd_del_station(ds, dp->index, sb_dev);
+}
+
 static const struct net_device_ops dsa_slave_netdev_ops = {
 	.ndo_open	 	= dsa_slave_open,
 	.ndo_stop		= dsa_slave_close,
 	.ndo_start_xmit		= dsa_slave_xmit,
+	.ndo_select_queue	= dsa_slave_select_queue,
 	.ndo_change_rx_flags	= dsa_slave_change_rx_flags,
 	.ndo_set_rx_mode	= dsa_slave_set_rx_mode,
 	.ndo_set_mac_address	= dsa_slave_set_mac_address,
@@ -1734,6 +1764,8 @@ static const struct net_device_ops dsa_slave_netdev_ops =
{
 	.ndo_get_devlink_port	= dsa_slave_get_devlink_port,
 	.ndo_change_mtu		= dsa_slave_change_mtu,
 	.ndo_fill_forward_path	= dsa_slave_fill_forward_path,
+	.ndo_dfwd_add_station	= dsa_slave_dfwd_add_station,
+	.ndo_dfwd_del_station	= dsa_slave_dfwd_del_station,
 };
 
 static struct device_type dsa_type = {
@@ -1914,8 +1946,8 @@ int dsa_slave_create(struct dsa_port *port)
 	slave_dev->features = master->vlan_features | NETIF_F_HW_TC;
 	if (ds->ops->port_vlan_add && ds->ops->port_vlan_del)
 		slave_dev->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
-	slave_dev->hw_features |= NETIF_F_HW_TC;
-	slave_dev->features |= NETIF_F_LLTX;
+	slave_dev->hw_features |= NETIF_F_HW_TC | NETIF_F_HW_L2FW_DOFFLOAD;
+	slave_dev->features |= NETIF_F_LLTX | NETIF_F_HW_L2FW_DOFFLOAD;
 	slave_dev->ethtool_ops = &dsa_slave_ethtool_ops;
 	if (!is_zero_ether_addr(port->mac))
 		ether_addr_copy(slave_dev->dev_addr, port->mac);
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 7/9] net: dsa: mv88e6xxx: Allocate a virtual DSA port for each bridge

In the near future we want to offload transmission of both unicasts
and multicasts from a bridge by sending a single FORWARD and use the
switches' config to determine the destination(s). Much in the same way
as we have already relied on them to do between user ports in the
past.

As isolation between bridges must still be maintained, we need to pass
an identifier in the DSA tag that the switches can use to determine
the set of physical ports that make up a particular flooding domain.

Therefore: allocate a DSA device/port tuple that is not used by any
physical device to each bridge we are offloading. We can then in
upcoming changes use this tuple to setup cross-chip port based VLANs
to restrict the set of valid egress ports to only contain the ports
that are offloading the same bridge.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 drivers/net/dsa/mv88e6xxx/Makefile |   1 +
 drivers/net/dsa/mv88e6xxx/chip.c   |  11 +++
 drivers/net/dsa/mv88e6xxx/dst.c    | 127 +++++++++++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/dst.h    |  12 +++
 include/net/dsa.h                  |   5 ++
 5 files changed, 156 insertions(+)
 create mode 100644 drivers/net/dsa/mv88e6xxx/dst.c
 create mode 100644 drivers/net/dsa/mv88e6xxx/dst.h

diff --git a/drivers/net/dsa/mv88e6xxx/Makefile
b/drivers/net/dsa/mv88e6xxx/Makefile
index c8eca2b6f959..20e00695b28d 100644
--- a/drivers/net/dsa/mv88e6xxx/Makefile
+++ b/drivers/net/dsa/mv88e6xxx/Makefile
@@ -2,6 +2,7 @@
 obj-$(CONFIG_NET_DSA_MV88E6XXX) += mv88e6xxx.o
 mv88e6xxx-objs := chip.o
 mv88e6xxx-objs += devlink.o
+mv88e6xxx-objs += dst.o
 mv88e6xxx-objs += global1.o
 mv88e6xxx-objs += global1_atu.o
 mv88e6xxx-objs += global1_vtu.o
diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index eca285aaf72f..06ef654472b7 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -33,6 +33,7 @@
 
 #include "chip.h"
 #include "devlink.h"
+#include "dst.h"
 #include "global1.h"
 #include "global2.h"
 #include "hwtstamp.h"
@@ -2371,6 +2372,10 @@ static int mv88e6xxx_port_bridge_join(struct dsa_switch
*ds, int port,
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err;
 
+	err = mv88e6xxx_dst_bridge_join(ds->dst, br);
+	if (err)
+		return err;
+
 	mv88e6xxx_reg_lock(chip);
 	err = mv88e6xxx_bridge_map(chip, br);
 	mv88e6xxx_reg_unlock(chip);
@@ -2388,6 +2393,8 @@ static void mv88e6xxx_port_bridge_leave(struct dsa_switch
*ds, int port,
 	    mv88e6xxx_port_vlan_map(chip, port))
 		dev_err(ds->dev, "failed to remap in-chip Port VLAN\n");
 	mv88e6xxx_reg_unlock(chip);
+
+	mv88e6xxx_dst_bridge_leave(ds->dst, br);
 }
 
 static int mv88e6xxx_crosschip_bridge_join(struct dsa_switch *ds,
@@ -3027,6 +3034,10 @@ static int mv88e6xxx_setup(struct dsa_switch *ds)
 
 	mv88e6xxx_reg_lock(chip);
 
+	err = mv88e6xxx_dst_add_chip(chip);
+	if (err)
+		goto unlock;
+
 	if (chip->info->ops->setup_errata) {
 		err = chip->info->ops->setup_errata(chip);
 		if (err)
diff --git a/drivers/net/dsa/mv88e6xxx/dst.c b/drivers/net/dsa/mv88e6xxx/dst.c
new file mode 100644
index 000000000000..399a818063bf
--- /dev/null
+++ b/drivers/net/dsa/mv88e6xxx/dst.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * mv88e6xxx global DSA switch tree state
+ */
+
+#include <linux/bitmap.h>
+#include <linux/dsa/mv88e6xxx.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/netdevice.h>
+#include <net/dsa.h>
+
+#include "chip.h"
+#include "dst.h"
+#include "global2.h"
+
+struct mv88e6xxx_br {
+	struct list_head list;
+
+	struct net_device *brdev;
+	u8 dev;
+	u8 port;
+};
+
+struct mv88e6xxx_dst {
+	struct list_head bridges;
+
+	DECLARE_BITMAP(busy_ports, MV88E6XXX_MAX_PVT_ENTRIES);
+
+#define DEV_PORT_TO_BIT(_dev, _port)			\
+	((_dev) * MV88E6XXX_MAX_PVT_PORTS + (_port))
+#define DEV_FROM_BIT(_bit) ((_bit) / MV88E6XXX_MAX_PVT_PORTS)
+#define PORT_FROM_BIT(_bit) ((_bit) % (MV88E6XXX_MAX_PVT_PORTS))
+};
+
+int mv88e6xxx_dst_bridge_join(struct dsa_switch_tree *dst,
+			      struct net_device *brdev)
+{
+	struct mv88e6xxx_dst *mvdst = dst->priv;
+	struct mv88e6xxx_br *mvbr;
+	unsigned int bit;
+
+	list_for_each_entry(mvbr, &mvdst->bridges, list) {
+		if (mvbr->brdev == brdev)
+			return 0;
+	}
+
+	bit = find_first_zero_bit(mvdst->busy_ports,
+				  MV88E6XXX_MAX_PVT_ENTRIES);
+
+	if (bit >= MV88E6XXX_MAX_PVT_ENTRIES) {
+		pr_err("Unable to allocate virtual port for %s in DSA tree %d\n",
+		       netdev_name(brdev), dst->index);
+		return -ENOSPC;
+	}
+
+	mvbr = kzalloc(sizeof(*mvbr), GFP_KERNEL);
+	if (!mvbr)
+		return -ENOMEM;
+
+	mvbr->brdev = brdev;
+	mvbr->dev = DEV_FROM_BIT(bit);
+	mvbr->port = PORT_FROM_BIT(bit);
+
+	INIT_LIST_HEAD(&mvbr->list);
+	list_add_tail(&mvbr->list, &mvdst->bridges);
+	set_bit(bit, mvdst->busy_ports);
+	return 0;
+}
+
+void mv88e6xxx_dst_bridge_leave(struct dsa_switch_tree *dst,
+				struct net_device *brdev)
+{
+	struct mv88e6xxx_dst *mvdst = dst->priv;
+	struct mv88e6xxx_br *mvbr;
+	struct dsa_port *dp;
+
+	list_for_each_entry(dp, &dst->ports, list) {
+		if (dp->bridge_dev == brdev)
+			return;
+	}
+
+	list_for_each_entry(mvbr, &mvdst->bridges, list) {
+		if (mvbr->brdev == brdev) {
+			clear_bit(DEV_PORT_TO_BIT(mvbr->dev, mvbr->port),
+				  mvdst->busy_ports);
+			list_del(&mvbr->list);
+			kfree(mvbr);
+			return;
+		}
+	}
+}
+
+static struct mv88e6xxx_dst *mv88e6xxx_dst_get(struct dsa_switch_tree *dst)
+{
+	struct mv88e6xxx_dst *mvdst;
+
+	if (dst->priv)
+		return dst->priv;
+
+	mvdst = kzalloc(sizeof(*mvdst), GFP_KERNEL);
+	if (!mvdst)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&mvdst->bridges);
+
+	bitmap_set(mvdst->busy_ports,
+		   DEV_PORT_TO_BIT(MV88E6XXX_G2_PVT_ADDR_DEV_TRUNK, 0),
+		   MV88E6XXX_MAX_PVT_PORTS);
+
+	dst->priv = mvdst;
+	return mvdst;
+}
+
+int mv88e6xxx_dst_add_chip(struct mv88e6xxx_chip *chip)
+{
+	struct dsa_switch_tree *dst = chip->ds->dst;
+	struct mv88e6xxx_dst *mvdst;
+
+	mvdst = mv88e6xxx_dst_get(dst);
+	if (IS_ERR(mvdst))
+		return PTR_ERR(mvdst);
+
+	bitmap_set(mvdst->busy_ports, DEV_PORT_TO_BIT(chip->ds->index, 0),
+		   MV88E6XXX_MAX_PVT_PORTS);
+	return 0;
+}
diff --git a/drivers/net/dsa/mv88e6xxx/dst.h b/drivers/net/dsa/mv88e6xxx/dst.h
new file mode 100644
index 000000000000..3845a19192ef
--- /dev/null
+++ b/drivers/net/dsa/mv88e6xxx/dst.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef _MV88E6XXX_DST_H
+#define _MV88E6XXX_DST_H
+
+int mv88e6xxx_dst_bridge_join(struct dsa_switch_tree *dst,
+			      struct net_device *brdev);
+void mv88e6xxx_dst_bridge_leave(struct dsa_switch_tree *dst,
+				struct net_device *brdev);
+int mv88e6xxx_dst_add_chip(struct mv88e6xxx_chip *chip);
+
+#endif /* _MV88E6XXX_DST_H */
diff --git a/include/net/dsa.h b/include/net/dsa.h
index 77d4df819299..c01e74d6e134 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -172,6 +172,11 @@ struct dsa_switch_tree {
 	 */
 	struct net_device **lags;
 	unsigned int lags_len;
+
+	/* Give the switch driver somewhere to hang its tree-wide
+	 * private data structure.
+	 */
+	void *priv;
 };
 
 #define dsa_lags_foreach_id(_id, _dst)				\
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 8/9] net: dsa: mv88e6xxx: Map virtual bridge port in PVT

Now that each bridge has a unique DSA device/port tuple, make sure
that each chip limits forwarding from the bridge to only include
fabric ports and local ports that are members of the same bridge.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 MAINTAINERS                      |  1 +
 drivers/net/dsa/mv88e6xxx/chip.c | 33 ++++++++++++++++++++++++--------
 drivers/net/dsa/mv88e6xxx/dst.c  | 33 ++++++++++++++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/dst.h  |  2 ++
 include/linux/dsa/mv88e6xxx.h    | 13 +++++++++++++
 5 files changed, 74 insertions(+), 8 deletions(-)
 create mode 100644 include/linux/dsa/mv88e6xxx.h

diff --git a/MAINTAINERS b/MAINTAINERS
index c3c8fa572580..8794b05793b2 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -10647,6 +10647,7 @@ S:	Maintained
 F:	Documentation/devicetree/bindings/net/dsa/marvell.txt
 F:	Documentation/networking/devlink/mv88e6xxx.rst
 F:	drivers/net/dsa/mv88e6xxx/
+F:	include/linux/dsa/mv88e6xxx.h
 F:	include/linux/platform_data/mv88e6xxx.h
 
 MARVELL ARMADA 3700 PHY DRIVERS
diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index 06ef654472b7..6975cf16da65 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -12,6 +12,7 @@
 
 #include <linux/bitfield.h>
 #include <linux/delay.h>
+#include <linux/dsa/mv88e6xxx.h>
 #include <linux/etherdevice.h>
 #include <linux/ethtool.h>
 #include <linux/if_bridge.h>
@@ -1229,15 +1230,25 @@ static u16 mv88e6xxx_port_vlan(struct mv88e6xxx_chip
*chip, int dev, int port)
 		}
 	}
 
-	/* Prevent frames from unknown switch or port */
-	if (!found)
-		return 0;
+	if (found) {
+		/* Frames from DSA links and CPU ports can egress any
+		 * local port.
+		 */
+		if (dp->type == DSA_PORT_TYPE_CPU ||
+		    dp->type == DSA_PORT_TYPE_DSA)
+			return mv88e6xxx_port_mask(chip);
 
-	/* Frames from DSA links and CPU ports can egress any local port */
-	if (dp->type == DSA_PORT_TYPE_CPU || dp->type == DSA_PORT_TYPE_DSA)
-		return mv88e6xxx_port_mask(chip);
+		br = dp->bridge_dev;
+	} else {
+		br = mv88e6xxx_dst_bridge_from_dsa(dst, dev, port);
+
+		/* Reject frames from ports that are neither physical
+		 * nor virtual bridge ports.
+		 */
+		if (!br)
+			return 0;
+	}
 
-	br = dp->bridge_dev;
 	pvlan = 0;
 
 	/* Frames from user ports can egress any local DSA links and CPU ports,
@@ -2340,6 +2351,7 @@ static int mv88e6xxx_bridge_map(struct mv88e6xxx_chip
*chip,
 	struct dsa_switch *ds = chip->ds;
 	struct dsa_switch_tree *dst = ds->dst;
 	struct dsa_port *dp;
+	u8 dev, port;
 	int err;
 
 	list_for_each_entry(dp, &dst->ports, list) {
@@ -2363,7 +2375,12 @@ static int mv88e6xxx_bridge_map(struct mv88e6xxx_chip
*chip,
 		}
 	}
 
-	return 0;
+	/* Map the virtual bridge port if one is assigned. */
+	err = mv88e6xxx_dst_bridge_to_dsa(dst, br, &dev, &port);
+	if (!err)
+		err = mv88e6xxx_pvt_map(chip, dev, port);
+
+	return err;
 }
 
 static int mv88e6xxx_port_bridge_join(struct dsa_switch *ds, int port,
diff --git a/drivers/net/dsa/mv88e6xxx/dst.c b/drivers/net/dsa/mv88e6xxx/dst.c
index 399a818063bf..a5f9077e5b3f 100644
--- a/drivers/net/dsa/mv88e6xxx/dst.c
+++ b/drivers/net/dsa/mv88e6xxx/dst.c
@@ -33,6 +33,39 @@ struct mv88e6xxx_dst {
 #define PORT_FROM_BIT(_bit) ((_bit) % (MV88E6XXX_MAX_PVT_PORTS))
 };
 
+struct net_device *mv88e6xxx_dst_bridge_from_dsa(struct dsa_switch_tree *dst,
+						 u8 dev, u8 port)
+{
+	struct mv88e6xxx_dst *mvdst = dst->priv;
+	struct mv88e6xxx_br *mvbr;
+
+	list_for_each_entry(mvbr, &mvdst->bridges, list) {
+		if (mvbr->dev == dev && mvbr->port == port)
+			return mvbr->brdev;
+	}
+
+	return NULL;
+}
+
+int mv88e6xxx_dst_bridge_to_dsa(const struct dsa_switch_tree *dst,
+				const struct net_device *brdev,
+				u8 *dev, u8 *port)
+{
+	struct mv88e6xxx_dst *mvdst = dst->priv;
+	struct mv88e6xxx_br *mvbr;
+
+	list_for_each_entry(mvbr, &mvdst->bridges, list) {
+		if (mvbr->brdev == brdev) {
+			*dev = mvbr->dev;
+			*port = mvbr->port;
+			return 0;
+		}
+	}
+
+	return -ENODEV;
+}
+EXPORT_SYMBOL_GPL(mv88e6xxx_dst_bridge_to_dsa);
+
 int mv88e6xxx_dst_bridge_join(struct dsa_switch_tree *dst,
 			      struct net_device *brdev)
 {
diff --git a/drivers/net/dsa/mv88e6xxx/dst.h b/drivers/net/dsa/mv88e6xxx/dst.h
index 3845a19192ef..911890ec4792 100644
--- a/drivers/net/dsa/mv88e6xxx/dst.h
+++ b/drivers/net/dsa/mv88e6xxx/dst.h
@@ -3,6 +3,8 @@
 #ifndef _MV88E6XXX_DST_H
 #define _MV88E6XXX_DST_H
 
+struct net_device *mv88e6xxx_dst_bridge_from_dsa(struct dsa_switch_tree *dst,
+						 u8 dev, u8 port);
 int mv88e6xxx_dst_bridge_join(struct dsa_switch_tree *dst,
 			      struct net_device *brdev);
 void mv88e6xxx_dst_bridge_leave(struct dsa_switch_tree *dst,
diff --git a/include/linux/dsa/mv88e6xxx.h b/include/linux/dsa/mv88e6xxx.h
new file mode 100644
index 000000000000..fa486dfe9808
--- /dev/null
+++ b/include/linux/dsa/mv88e6xxx.h
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef _NET_DSA_MV88E6XXX_H
+#define _NET_DSA_MV88E6XXX_H
+
+#include <linux/netdevice.h>
+#include <net/dsa.h>
+
+int mv88e6xxx_dst_bridge_to_dsa(const struct dsa_switch_tree *dst,
+				const struct net_device *brdev,
+				u8 *dev, u8 *port);
+
+#endif /* _NET_DSA_MV88E6XXX_H */
-- 
2.25.1

Tobias Waldekranz

2021-Apr-26 17:04 UTC

head link

[Bridge] [RFC net-next 9/9] net: dsa: mv88e6xxx: Forward offloading

Allow the DSA tagger to generate FORWARD frames for offloaded skbs
sent from a bridge that we offload, allowing the switch to handle any
frame replication that may be required. This also means that source
address learning takes place on packets sent from the CPU, meaning
that return traffic no longer needs to be flooded as unknown unicast.

Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
---
 drivers/net/dsa/mv88e6xxx/chip.c | 17 ++++++++++++++++
 net/dsa/tag_dsa.c                | 33 ++++++++++++++++++++++++--------
 2 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index 6975cf16da65..00ed1aa2a55a 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -6077,6 +6077,21 @@ static int mv88e6xxx_crosschip_lag_leave(struct
dsa_switch *ds, int sw_index,
 	return err_sync ? : err_pvt;
 }
 
+static void *mv88e6xxx_dfwd_add_station(struct dsa_switch *ds, int port,
+					struct net_device *sb_dev)
+{
+	struct dsa_port *dp = dsa_to_port(ds, port);
+	struct mv88e6xxx_chip *chip = ds->priv;
+
+	if (!mv88e6xxx_has_pvt(chip))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	if (sb_dev == dp->bridge_dev)
+		return sb_dev;
+
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
 static const struct dsa_switch_ops mv88e6xxx_switch_ops = {
 	.get_tag_protocol	= mv88e6xxx_get_tag_protocol,
 	.change_tag_protocol	= mv88e6xxx_change_tag_protocol,
@@ -6138,6 +6153,7 @@ static const struct dsa_switch_ops mv88e6xxx_switch_ops =
{
 	.crosschip_lag_change	= mv88e6xxx_crosschip_lag_change,
 	.crosschip_lag_join	= mv88e6xxx_crosschip_lag_join,
 	.crosschip_lag_leave	= mv88e6xxx_crosschip_lag_leave,
+	.dfwd_add_station	= mv88e6xxx_dfwd_add_station,
 };
 
 static int mv88e6xxx_register_switch(struct mv88e6xxx_chip *chip)
@@ -6156,6 +6172,7 @@ static int mv88e6xxx_register_switch(struct mv88e6xxx_chip
*chip)
 	ds->ops = &mv88e6xxx_switch_ops;
 	ds->ageing_time_min = chip->info->age_time_coeff;
 	ds->ageing_time_max = chip->info->age_time_coeff * U8_MAX;
+	ds->num_tx_queues = 2;
 
 	/* Some chips support up to 32, but that requires enabling the
 	 * 5-bit port mode, which we do not support. 640k^W16 ought to
diff --git a/net/dsa/tag_dsa.c b/net/dsa/tag_dsa.c
index 7e7b7decdf39..09cdf77697b2 100644
--- a/net/dsa/tag_dsa.c
+++ b/net/dsa/tag_dsa.c
@@ -46,6 +46,7 @@
  */
 
 #include <linux/etherdevice.h>
+#include <linux/dsa/mv88e6xxx.h>
 #include <linux/list.h>
 #include <linux/slab.h>
 
@@ -126,7 +127,22 @@ static struct sk_buff *dsa_xmit_ll(struct sk_buff *skb,
struct net_device *dev,
 				   u8 extra)
 {
 	struct dsa_port *dp = dsa_slave_to_port(dev);
+	u16 pvid = dp->pvid;
+	enum dsa_cmd cmd;
 	u8 *dsa_header;
+	u8 tag_dev, tag_port;
+
+	if (DSA_SKB_CB(skb)->sb_dev) {
+		cmd = DSA_CMD_FORWARD;
+		if (mv88e6xxx_dst_bridge_to_dsa(dp->ds->dst,
+						DSA_SKB_CB(skb)->sb_dev,
+						&tag_dev, &tag_port))
+			return NULL;
+	} else {
+		cmd = DSA_CMD_FROM_CPU;
+		tag_dev = dp->ds->index;
+		tag_port = dp->index;
+	}
 
 	if (skb->protocol == htons(ETH_P_8021Q)) {
 		if (extra) {
@@ -134,10 +150,10 @@ static struct sk_buff *dsa_xmit_ll(struct sk_buff *skb,
struct net_device *dev,
 			memmove(skb->data, skb->data + extra, 2 * ETH_ALEN);
 		}
 
-		/* Construct tagged FROM_CPU DSA tag from 802.1Q tag. */
+		/* Construct tagged DSA tag from 802.1Q tag. */
 		dsa_header = skb->data + 2 * ETH_ALEN + extra;
-		dsa_header[0] = (DSA_CMD_FROM_CPU << 6) | 0x20 | dp->ds->index;
-		dsa_header[1] = dp->index << 3;
+		dsa_header[0] = (cmd << 6) | 0x20 | tag_dev;
+		dsa_header[1] = tag_port << 3;
 
 		/* Move CFI field from byte 2 to byte 1. */
 		if (dsa_header[2] & 0x10) {
@@ -148,12 +164,13 @@ static struct sk_buff *dsa_xmit_ll(struct sk_buff *skb,
struct net_device *dev,
 		skb_push(skb, DSA_HLEN + extra);
 		memmove(skb->data, skb->data + DSA_HLEN + extra, 2 * ETH_ALEN);
 
-		/* Construct untagged FROM_CPU DSA tag. */
+		/* Construct untagged DSA tag. */
 		dsa_header = skb->data + 2 * ETH_ALEN + extra;
-		dsa_header[0] = (DSA_CMD_FROM_CPU << 6) | dp->ds->index;
-		dsa_header[1] = dp->index << 3;
-		dsa_header[2] = 0x00;
-		dsa_header[3] = 0x00;
+
+		dsa_header[0] = (cmd << 6) | tag_dev;
+		dsa_header[1] = tag_port << 3;
+		dsa_header[2] = pvid >> 8;
+		dsa_header[3] = pvid & 0xff;
 	}
 
 	return skb;
-- 
2.25.1

Vladimir Oltean

2021-Apr-26 19:40 UTC

head link

[Bridge] [RFC net-next 5/9] net: dsa: Track port PVIDs

Hi Tobias,

On Mon, Apr 26, 2021 at 07:04:07PM +0200, Tobias Waldekranz
wrote:> In some scenarios a tagger must know which VLAN to assign to a packet,
> even if the packet is set to egress untagged. Since the VLAN
> information in the skb will be removed by the bridge in this case,
> track each port's PVID such that the VID of an outgoing frame can
> always be determined.
> 
> Signed-off-by: Tobias Waldekranz <tobias at waldekranz.com>
> ---
Let me give you this real-life example:

#!/bin/bash

ip link add br0 type bridge vlan_filtering 1
for eth in eth0 eth1 swp2 swp3 swp4 swp5; do
	ip link set $eth up
	ip link set $eth master br0
done
ip link set br0 up

bridge vlan add dev eth0 vid 100 pvid untagged
bridge vlan del dev swp2 vid 1
bridge vlan del dev swp3 vid 1
bridge vlan add dev swp2 vid 100
bridge vlan add dev swp3 vid 100 untagged

reproducible on the NXP LS1021A-TSN board.
The bridge receives an untagged packet on eth0 and floods it.
It should reach swp2 and swp3, and be tagged on swp2, and untagged on
swp3 respectively.

With your idea of sending untagged frames towards the port's pvid,
wouldn't we be leaking this packet to VLAN 1, therefore towards ports
swp4 and swp5, and the real destination ports would not get this packet?

Ido Schimmel

2021-May-02 14:58 UTC

head link

[Bridge] [RFC net-next 0/9] net: bridge: Forward offloading

On Mon, Apr 26, 2021 at 07:04:02PM +0200, Tobias Waldekranz
wrote:> ## Overview
> 
>    vlan1   vlan2
>        \   /
>    .-----------.
>    |    br0    |
>    '-----------'
>    /   /   \   \
> swp0 swp1 swp2 eth0
>   :   :   :
>   (hwdom 1)
> 
> Up to this point, switchdevs have been trusted with offloading
> forwarding between bridge ports, e.g. forwarding a unicast from swp0
> to swp1 or flooding a broadcast from swp2 to swp1 and swp0. This
> series extends forward offloading to include some new classes of
> traffic:
> 
> - Locally originating flows, i.e. packets that ingress on br0 that are
>   to be forwarded to one or several of the ports swp{0,1,2}. Notably
>   this also includes routed flows, e.g. a packet ingressing swp0 on
>   VLAN 1 which is then routed over to VLAN 2 by the CPU and then
>   forwarded to swp1 is "locally originating" from br0's point
of view.
> 
> - Flows originating from "foreign" interfaces, i.e. an interface
that
>   is not offloaded by a particular switchdev instance. This includes
>   ports belonging to other switchdev instances. A typical example
>   would be flows from eth0 towards swp{0,1,2}.
> 
> The bridge still looks up its FDB/MDB as usual and then notifies the
> switchdev driver that a particular skb should be offloaded if it
> matches one of the classes above. It does so by using the _accel
> version of dev_queue_xmit, supplying its own netdev as the
> "subordinate" device. The driver can react to the presence of the
> subordinate in its .ndo_select_queue in what ever way it needs to make
> sure to forward the skb in much the same way that it would for packets
> ingressing on regular ports.
> 
> Hardware domains to which a particular skb has been forwarded are
> recorded so that duplicates are avoided.
> 
> The main performance benefit is thus seen on multicast flows. Imagine
> for example that:
> 
> - An IP camera is connected to swp0 (VLAN 1)
> 
> - The CPU is acting as a multicast router, routing the group from VLAN
>   1 to VLAN 2.
> 
> - There are subscribers for the group in question behind both swp1 and
>   swp2 (VLAN 2).
IIUC, this falls under the first use case ("Locally originating
flows").
Do you have a need for this optimization in the forwarding case? Asking
because it might allow us to avoid unnecessary modifications to the
forwarding path. I have yet to look at the code, so maybe it's not a big
deal.
> 
> With this offloading in place, the bridge need only send a single skb
> to the driver, which will send it to the hardware marked in such a way
> that the switch will perform the multicast replication according to
> the MDB configuration. Naturally, the number of saved skb_clones
> increase linearly with the number of subscribed ports.
Yes, this is clear. FWIW, Spectrum has something similar. You can send
packets as either "data" or "control". Data packets are
injected via the
CPU port and forwarded according to the hardware database. Control
packets are sent as-is via the specified front panel port, bypassing the
hardware data path. mlxsw is always sending packets as "control".
> 
> As an extra benefit, on mv88e6xxx, this also allows the switch to
> perform source address learning on these flows, which avoids having to
> sync dynamic FDB entries over slow configuration interfaces like MDIO
> to avoid flows directed towards the CPU being flooded as unknown
> unicast by the switch.
Since you are not syncing FDBs, it is possible that you are needlessly
flooding locally generated packets. This optimization avoids it.
> 
> 
> ## RFC
> 
> - In general, what do you think about this idea?
Looks sane to me
> 
> - hwdom. What do you think about this terminology? Personally I feel
>   that we had too many things called offload_fwd_mark, and that as the
>   use of the bridge internal ID (nbp->offload_fwd_mark) expands, it
>   might be useful to have a separate term for it.
Sounds OK
> 
> - .dfwd_{add,del}_station. Am I stretching this abstraction too far,
>   and if so do you have any suggestion/preference on how to signal the
>   offloading from the bridge down to the switchdev driver?
I was not aware of this interface before the RFC, but your use case
seems to fit the kdoc: "Called by upper layer devices to accelerate
switching or other station functionality into hardware".

Do you expect this optimization to only work when physical netdevs are
enslaved to the bridge? What about LAG/VLANs?
> 
> - The way that flooding is implemented in br_forward.c (lazily cloning
>   skbs) means that you have to mark the forwarding as completed very
>   early (right after should_deliver in maybe_deliver) in order to
>   avoid duplicates. Is there some way to move this decision point to a
>   later stage that I am missing?
> 
> - BR_MULTICAST_TO_UNICAST. Right now, I expect that this series is not
>   compatible with unicast-to-multicast being used on a port. Then
>   again, I think that this would also be broken for regular switchdev
>   bridge offloading as this flag is not offloaded to the switchdev
>   port, so there is no way for the driver to refuse it. Any ideas on
>   how to handle this?
> 
> 
> ## mv88e6xxx Specifics
> 
> Since we are now only receiving a single skb for both unicast and
> multicast flows, we can tag the packets with the FORWARD command
> instead of FROM_CPU. The swich(es) will then forward the packet in
> accordance with its ATU, VTU, STU, and PVT configuration - just like
> for packets ingressing on user ports.
> 
> Crucially, FROM_CPU is still used for:
> 
> - Ports in standalone mode.
> 
> - Flows that are trapped to the CPU and software-forwarded by a
>   bridge. Note that these flows match neither of the classes discussed
>   in the overview.
> 
> - Packets that are sent directly to a port netdev without going
>   through the bridge, e.g. lldpd sending out PDU via an AF_PACKET
>   socket.
> 
> We thus have a pretty clean separation where the data plane uses
> FORWARDs and the control plane uses TO_/FROM_CPU.
> 
> The barrier between different bridges is enforced by port based VLANs
> on mv88e6xxx, which in essence is a mapping from a source device/port
> pair to an allowed set of egress ports. In order to have a FORWARD
> frame (which carries a _source_ device/port) correctly mapped by the
> PVT, we must use a unique pair for each bridge.
> 
> Fortunately, there is typically lots of unused address space in most
> switch trees. When was the last time you saw an mv88e6xxx product
> using more than 4 chips? Even if you found one with 16 (!) devices,
> you would still have room to allocate 16*16 virtual ports to software
> bridges.
> 
> Therefore, the mv88e6xxx driver will allocate a virtual device/port
> pair to each bridge that it offloads. All members of the same bridge
> are then configured to allow packets from this virtual port in their
> PVTs.
> 
> Tobias Waldekranz (9):
>   net: dfwd: Constrain existing users to macvlan subordinates
>   net: bridge: Disambiguate offload_fwd_mark
>   net: bridge: switchdev: Recycle unused hwdoms
>   net: bridge: switchdev: Forward offloading
>   net: dsa: Track port PVIDs
>   net: dsa: Forward offloading
>   net: dsa: mv88e6xxx: Allocate a virtual DSA port for each bridge
>   net: dsa: mv88e6xxx: Map virtual bridge port in PVT
>   net: dsa: mv88e6xxx: Forward offloading
> 
>  MAINTAINERS                                   |   1 +
>  drivers/net/dsa/mv88e6xxx/Makefile            |   1 +
>  drivers/net/dsa/mv88e6xxx/chip.c              |  61 ++++++-
>  drivers/net/dsa/mv88e6xxx/dst.c               | 160 ++++++++++++++++++
>  drivers/net/dsa/mv88e6xxx/dst.h               |  14 ++
>  .../net/ethernet/intel/fm10k/fm10k_netdev.c   |   3 +
>  drivers/net/ethernet/intel/i40e/i40e_main.c   |   3 +
>  drivers/net/ethernet/intel/ixgbe/ixgbe_main.c |   3 +
>  include/linux/dsa/mv88e6xxx.h                 |  13 ++
>  include/net/dsa.h                             |  13 ++
>  net/bridge/br_forward.c                       |  11 +-
>  net/bridge/br_if.c                            |   4 +-
>  net/bridge/br_private.h                       |  54 +++++-
>  net/bridge/br_switchdev.c                     | 141 +++++++++++----
>  net/dsa/port.c                                |  16 +-
>  net/dsa/slave.c                               |  36 +++-
>  net/dsa/tag_dsa.c                             |  33 +++-
>  17 files changed, 510 insertions(+), 57 deletions(-)
>  create mode 100644 drivers/net/dsa/mv88e6xxx/dst.c
>  create mode 100644 drivers/net/dsa/mv88e6xxx/dst.h
>  create mode 100644 include/linux/dsa/mv88e6xxx.h
> 
> -- 
> 2.25.1
>

Ido Schimmel

2021-May-02 15:04 UTC

head link

[Bridge] [RFC net-next 4/9] net: bridge: switchdev: Forward offloading

On Mon, Apr 26, 2021 at 07:04:06PM +0200, Tobias Waldekranz
wrote:> +static void nbp_switchdev_fwd_offload_add(struct net_bridge_port *p)
> +{
> +	void *priv;
> +
> +	if (!(p->dev->features & NETIF_F_HW_L2FW_DOFFLOAD))
> +		return;
> +
> +	priv = p->dev->netdev_ops->ndo_dfwd_add_station(p->dev,
p->br->dev);
Some changes to team/bond/8021q will be needed in order to get this
optimization to work when they are enslaved to the bridge instead of the
front panel port itself?
> +	if (!IS_ERR_OR_NULL(priv))
> +		p->accel_priv = priv;
> +}
> +
> +static void nbp_switchdev_fwd_offload_del(struct net_bridge_port *p)
> +{
> +	if (!p->accel_priv)
> +		return;
> +
> +	p->dev->netdev_ops->ndo_dfwd_del_station(p->dev,
p->accel_priv);
> +	p->accel_priv = NULL;
> +}

Linux Ethernet Bridging - Apr 2021 - [Bridge] [RFC net-next 0/9] net: bridge: Forward offloading

[Bridge] [RFC net-next 0/9] net: bridge: Forward offloading

[Bridge] [RFC net-next 1/9] net: dfwd: Constrain existing users to macvlan subordinates

[Bridge] [RFC net-next 2/9] net: bridge: Disambiguate offload_fwd_mark

[Bridge] [RFC net-next 3/9] net: bridge: switchdev: Recycle unused hwdoms

[Bridge] [RFC net-next 4/9] net: bridge: switchdev: Forward offloading

[Bridge] [RFC net-next 5/9] net: dsa: Track port PVIDs

[Bridge] [RFC net-next 6/9] net: dsa: Forward offloading

[Bridge] [RFC net-next 7/9] net: dsa: mv88e6xxx: Allocate a virtual DSA port for each bridge

[Bridge] [RFC net-next 8/9] net: dsa: mv88e6xxx: Map virtual bridge port in PVT

[Bridge] [RFC net-next 9/9] net: dsa: mv88e6xxx: Forward offloading

[Bridge] [RFC net-next 5/9] net: dsa: Track port PVIDs

[Bridge] [RFC net-next 0/9] net: bridge: Forward offloading

[Bridge] [RFC net-next 4/9] net: bridge: switchdev: Forward offloading