About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Documentation / networking / switchdev.txt




Custom Search

Based on kernel version 4.13.3. Page generated on 2017-09-23 13:55 EST.

1	Ethernet switch device driver model (switchdev)
2	===============================================
3	Copyright (c) 2014 Jiri Pirko <jiri@resnulli.us>
4	Copyright (c) 2014-2015 Scott Feldman <sfeldma@gmail.com>
5	
6	
7	The Ethernet switch device driver model (switchdev) is an in-kernel driver
8	model for switch devices which offload the forwarding (data) plane from the
9	kernel.
10	
11	Figure 1 is a block diagram showing the components of the switchdev model for
12	an example setup using a data-center-class switch ASIC chip.  Other setups
13	with SR-IOV or soft switches, such as OVS, are possible.
14	
15	
16	                             User-space tools
17	
18	       user space                   |
19	      +-------------------------------------------------------------------+
20	       kernel                       | Netlink
21	                                    |
22	                     +--------------+-------------------------------+
23	                     |         Network stack                        |
24	                     |           (Linux)                            |
25	                     |                                              |
26	                     +----------------------------------------------+
27	
28	                           sw1p2     sw1p4     sw1p6
29	                      sw1p1  +  sw1p3  +  sw1p5  +          eth1
30	                        +    |    +    |    +    |            +
31	                        |    |    |    |    |    |            |
32	                     +--+----+----+----+-+--+----+---+  +-----+-----+
33	                     |         Switch driver         |  |    mgmt   |
34	                     |        (this document)        |  |   driver  |
35	                     |                               |  |           |
36	                     +--------------+----------------+  +-----------+
37	                                    |
38	       kernel                       | HW bus (eg PCI)
39	      +-------------------------------------------------------------------+
40	       hardware                     |
41	                     +--------------+---+------------+
42	                     |         Switch device (sw1)   |
43	                     |  +----+                       +--------+
44	                     |  |    v offloaded data path   | mgmt port
45	                     |  |    |                       |
46	                     +--|----|----+----+----+----+---+
47	                        |    |    |    |    |    |
48	                        +    +    +    +    +    +
49	                       p1   p2   p3   p4   p5   p6
50	
51	                             front-panel ports
52	
53	
54	                                    Fig 1.
55	
56	
57	Include Files
58	-------------
59	
60	#include <linux/netdevice.h>
61	#include <net/switchdev.h>
62	
63	
64	Configuration
65	-------------
66	
67	Use "depends NET_SWITCHDEV" in driver's Kconfig to ensure switchdev model
68	support is built for driver.
69	
70	
71	Switch Ports
72	------------
73	
74	On switchdev driver initialization, the driver will allocate and register a
75	struct net_device (using register_netdev()) for each enumerated physical switch
76	port, called the port netdev.  A port netdev is the software representation of
77	the physical port and provides a conduit for control traffic to/from the
78	controller (the kernel) and the network, as well as an anchor point for higher
79	level constructs such as bridges, bonds, VLANs, tunnels, and L3 routers.  Using
80	standard netdev tools (iproute2, ethtool, etc), the port netdev can also
81	provide to the user access to the physical properties of the switch port such
82	as PHY link state and I/O statistics.
83	
84	There is (currently) no higher-level kernel object for the switch beyond the
85	port netdevs.  All of the switchdev driver ops are netdev ops or switchdev ops.
86	
87	A switch management port is outside the scope of the switchdev driver model.
88	Typically, the management port is not participating in offloaded data plane and
89	is loaded with a different driver, such as a NIC driver, on the management port
90	device.
91	
92	Switch ID
93	^^^^^^^^^
94	
95	The switchdev driver must implement the switchdev op switchdev_port_attr_get
96	for SWITCHDEV_ATTR_ID_PORT_PARENT_ID for each port netdev, returning the same
97	physical ID for each port of a switch.  The ID must be unique between switches
98	on the same system.  The ID does not need to be unique between switches on
99	different systems.
100	
101	The switch ID is used to locate ports on a switch and to know if aggregated
102	ports belong to the same switch.
103	
104	Port Netdev Naming
105	^^^^^^^^^^^^^^^^^^
106	
107	Udev rules should be used for port netdev naming, using some unique attribute
108	of the port as a key, for example the port MAC address or the port PHYS name.
109	Hard-coding of kernel netdev names within the driver is discouraged; let the
110	kernel pick the default netdev name, and let udev set the final name based on a
111	port attribute.
112	
113	Using port PHYS name (ndo_get_phys_port_name) for the key is particularly
114	useful for dynamically-named ports where the device names its ports based on
115	external configuration.  For example, if a physical 40G port is split logically
116	into 4 10G ports, resulting in 4 port netdevs, the device can give a unique
117	name for each port using port PHYS name.  The udev rule would be:
118	
119	SUBSYSTEM=="net", ACTION=="add", ATTR{phys_switch_id}=="<phys_switch_id>", \
120		ATTR{phys_port_name}!="", NAME="swX$attr{phys_port_name}"
121	
122	Suggested naming convention is "swXpYsZ", where X is the switch name or ID, Y
123	is the port name or ID, and Z is the sub-port name or ID.  For example, sw1p1s0
124	would be sub-port 0 on port 1 on switch 1.
125	
126	Port Features
127	^^^^^^^^^^^^^
128	
129	NETIF_F_NETNS_LOCAL
130	
131	If the switchdev driver (and device) only supports offloading of the default
132	network namespace (netns), the driver should set this feature flag to prevent
133	the port netdev from being moved out of the default netns.  A netns-aware
134	driver/device would not set this flag and be responsible for partitioning
135	hardware to preserve netns containment.  This means hardware cannot forward
136	traffic from a port in one namespace to another port in another namespace.
137	
138	Port Topology
139	^^^^^^^^^^^^^
140	
141	The port netdevs representing the physical switch ports can be organized into
142	higher-level switching constructs.  The default construct is a standalone
143	router port, used to offload L3 forwarding.  Two or more ports can be bonded
144	together to form a LAG.  Two or more ports (or LAGs) can be bridged to bridge
145	L2 networks.  VLANs can be applied to sub-divide L2 networks.  L2-over-L3
146	tunnels can be built on ports.  These constructs are built using standard Linux
147	tools such as the bridge driver, the bonding/team drivers, and netlink-based
148	tools such as iproute2.
149	
150	The switchdev driver can know a particular port's position in the topology by
151	monitoring NETDEV_CHANGEUPPER notifications.  For example, a port moved into a
152	bond will see it's upper master change.  If that bond is moved into a bridge,
153	the bond's upper master will change.  And so on.  The driver will track such
154	movements to know what position a port is in in the overall topology by
155	registering for netdevice events and acting on NETDEV_CHANGEUPPER.
156	
157	L2 Forwarding Offload
158	---------------------
159	
160	The idea is to offload the L2 data forwarding (switching) path from the kernel
161	to the switchdev device by mirroring bridge FDB entries down to the device.  An
162	FDB entry is the {port, MAC, VLAN} tuple forwarding destination.
163	
164	To offloading L2 bridging, the switchdev driver/device should support:
165	
166		- Static FDB entries installed on a bridge port
167		- Notification of learned/forgotten src mac/vlans from device
168		- STP state changes on the port
169		- VLAN flooding of multicast/broadcast and unknown unicast packets
170	
171	Static FDB Entries
172	^^^^^^^^^^^^^^^^^^
173	
174	The switchdev driver should implement ndo_fdb_add, ndo_fdb_del and ndo_fdb_dump
175	to support static FDB entries installed to the device.  Static bridge FDB
176	entries are installed, for example, using iproute2 bridge cmd:
177	
178		bridge fdb add ADDR dev DEV [vlan VID] [self]
179	
180	The driver should use the helper switchdev_port_fdb_xxx ops for ndo_fdb_xxx
181	ops, and handle add/delete/dump of SWITCHDEV_OBJ_ID_PORT_FDB object using
182	switchdev_port_obj_xxx ops.
183	
184	XXX: what should be done if offloading this rule to hardware fails (for
185	example, due to full capacity in hardware tables) ?
186	
187	Note: by default, the bridge does not filter on VLAN and only bridges untagged
188	traffic.  To enable VLAN support, turn on VLAN filtering:
189	
190		echo 1 >/sys/class/net/<bridge>/bridge/vlan_filtering
191	
192	Notification of Learned/Forgotten Source MAC/VLANs
193	^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
194	
195	The switch device will learn/forget source MAC address/VLAN on ingress packets
196	and notify the switch driver of the mac/vlan/port tuples.  The switch driver,
197	in turn, will notify the bridge driver using the switchdev notifier call:
198	
199		err = call_switchdev_notifiers(val, dev, info);
200	
201	Where val is SWITCHDEV_FDB_ADD when learning and SWITCHDEV_FDB_DEL when
202	forgetting, and info points to a struct switchdev_notifier_fdb_info.  On
203	SWITCHDEV_FDB_ADD, the bridge driver will install the FDB entry into the
204	bridge's FDB and mark the entry as NTF_EXT_LEARNED.  The iproute2 bridge
205	command will label these entries "offload":
206	
207		$ bridge fdb
208		52:54:00:12:35:01 dev sw1p1 master br0 permanent
209		00:02:00:00:02:00 dev sw1p1 master br0 offload
210		00:02:00:00:02:00 dev sw1p1 self
211		52:54:00:12:35:02 dev sw1p2 master br0 permanent
212		00:02:00:00:03:00 dev sw1p2 master br0 offload
213		00:02:00:00:03:00 dev sw1p2 self
214		33:33:00:00:00:01 dev eth0 self permanent
215		01:00:5e:00:00:01 dev eth0 self permanent
216		33:33:ff:00:00:00 dev eth0 self permanent
217		01:80:c2:00:00:0e dev eth0 self permanent
218		33:33:00:00:00:01 dev br0 self permanent
219		01:00:5e:00:00:01 dev br0 self permanent
220		33:33:ff:12:35:01 dev br0 self permanent
221	
222	Learning on the port should be disabled on the bridge using the bridge command:
223	
224		bridge link set dev DEV learning off
225	
226	Learning on the device port should be enabled, as well as learning_sync:
227	
228		bridge link set dev DEV learning on self
229		bridge link set dev DEV learning_sync on self
230	
231	Learning_sync attribute enables syncing of the learned/forgotten FDB entry to
232	the bridge's FDB.  It's possible, but not optimal, to enable learning on the
233	device port and on the bridge port, and disable learning_sync.
234	
235	To support learning and learning_sync port attributes, the driver implements
236	switchdev op switchdev_port_attr_get/set for
237	SWITCHDEV_ATTR_PORT_ID_BRIDGE_FLAGS. The driver should initialize the attributes
238	to the hardware defaults.
239	
240	FDB Ageing
241	^^^^^^^^^^
242	
243	The bridge will skip ageing FDB entries marked with NTF_EXT_LEARNED and it is
244	the responsibility of the port driver/device to age out these entries.  If the
245	port device supports ageing, when the FDB entry expires, it will notify the
246	driver which in turn will notify the bridge with SWITCHDEV_FDB_DEL.  If the
247	device does not support ageing, the driver can simulate ageing using a
248	garbage collection timer to monitor FDB entries.  Expired entries will be
249	notified to the bridge using SWITCHDEV_FDB_DEL.  See rocker driver for
250	example of driver running ageing timer.
251	
252	To keep an NTF_EXT_LEARNED entry "alive", the driver should refresh the FDB
253	entry by calling call_switchdev_notifiers(SWITCHDEV_FDB_ADD, ...).  The
254	notification will reset the FDB entry's last-used time to now.  The driver
255	should rate limit refresh notifications, for example, no more than once a
256	second.  (The last-used time is visible using the bridge -s fdb option).
257	
258	STP State Change on Port
259	^^^^^^^^^^^^^^^^^^^^^^^^
260	
261	Internally or with a third-party STP protocol implementation (e.g. mstpd), the
262	bridge driver maintains the STP state for ports, and will notify the switch
263	driver of STP state change on a port using the switchdev op
264	switchdev_attr_port_set for SWITCHDEV_ATTR_PORT_ID_STP_UPDATE.
265	
266	State is one of BR_STATE_*.  The switch driver can use STP state updates to
267	update ingress packet filter list for the port.  For example, if port is
268	DISABLED, no packets should pass, but if port moves to BLOCKED, then STP BPDUs
269	and other IEEE 01:80:c2:xx:xx:xx link-local multicast packets can pass.
270	
271	Note that STP BDPUs are untagged and STP state applies to all VLANs on the port
272	so packet filters should be applied consistently across untagged and tagged
273	VLANs on the port.
274	
275	Flooding L2 domain
276	^^^^^^^^^^^^^^^^^^
277	
278	For a given L2 VLAN domain, the switch device should flood multicast/broadcast
279	and unknown unicast packets to all ports in domain, if allowed by port's
280	current STP state.  The switch driver, knowing which ports are within which
281	vlan L2 domain, can program the switch device for flooding.  The packet may
282	be sent to the port netdev for processing by the bridge driver.  The
283	bridge should not reflood the packet to the same ports the device flooded,
284	otherwise there will be duplicate packets on the wire.
285	
286	To avoid duplicate packets, the switch driver should mark a packet as already
287	forwarded by setting the skb->offload_fwd_mark bit. The bridge driver will mark
288	the skb using the ingress bridge port's mark and prevent it from being forwarded
289	through any bridge port with the same mark.
290	
291	It is possible for the switch device to not handle flooding and push the
292	packets up to the bridge driver for flooding.  This is not ideal as the number
293	of ports scale in the L2 domain as the device is much more efficient at
294	flooding packets that software.
295	
296	If supported by the device, flood control can be offloaded to it, preventing
297	certain netdevs from flooding unicast traffic for which there is no FDB entry.
298	
299	IGMP Snooping
300	^^^^^^^^^^^^^
301	
302	In order to support IGMP snooping, the port netdevs should trap to the bridge
303	driver all IGMP join and leave messages.
304	The bridge multicast module will notify port netdevs on every multicast group
305	changed whether it is static configured or dynamically joined/leave.
306	The hardware implementation should be forwarding all registered multicast
307	traffic groups only to the configured ports.
308	
309	L3 Routing Offload
310	------------------
311	
312	Offloading L3 routing requires that device be programmed with FIB entries from
313	the kernel, with the device doing the FIB lookup and forwarding.  The device
314	does a longest prefix match (LPM) on FIB entries matching route prefix and
315	forwards the packet to the matching FIB entry's nexthop(s) egress ports.
316	
317	To program the device, the driver has to register a FIB notifier handler
318	using register_fib_notifier. The following events are available:
319	FIB_EVENT_ENTRY_ADD: used for both adding a new FIB entry to the device,
320	                     or modifying an existing entry on the device.
321	FIB_EVENT_ENTRY_DEL: used for removing a FIB entry
322	FIB_EVENT_RULE_ADD, FIB_EVENT_RULE_DEL: used to propagate FIB rule changes
323	
324	FIB_EVENT_ENTRY_ADD and FIB_EVENT_ENTRY_DEL events pass:
325	
326		struct fib_entry_notifier_info {
327			struct fib_notifier_info info; /* must be first */
328			u32 dst;
329			int dst_len;
330			struct fib_info *fi;
331			u8 tos;
332			u8 type;
333			u32 tb_id;
334			u32 nlflags;
335		};
336	
337	to add/modify/delete IPv4 dst/dest_len prefix on table tb_id.  The *fi
338	structure holds details on the route and route's nexthops.  *dev is one of the
339	port netdevs mentioned in the route's next hop list.
340	
341	Routes offloaded to the device are labeled with "offload" in the ip route
342	listing:
343	
344		$ ip route show
345		default via 192.168.0.2 dev eth0
346		11.0.0.0/30 dev sw1p1  proto kernel  scope link  src 11.0.0.2 offload
347		11.0.0.4/30 via 11.0.0.1 dev sw1p1  proto zebra  metric 20 offload
348		11.0.0.8/30 dev sw1p2  proto kernel  scope link  src 11.0.0.10 offload
349		11.0.0.12/30 via 11.0.0.9 dev sw1p2  proto zebra  metric 20 offload
350		12.0.0.2  proto zebra  metric 30 offload
351			nexthop via 11.0.0.1  dev sw1p1 weight 1
352			nexthop via 11.0.0.9  dev sw1p2 weight 1
353		12.0.0.3 via 11.0.0.1 dev sw1p1  proto zebra  metric 20 offload
354		12.0.0.4 via 11.0.0.9 dev sw1p2  proto zebra  metric 20 offload
355		192.168.0.0/24 dev eth0  proto kernel  scope link  src 192.168.0.15
356	
357	The "offload" flag is set in case at least one device offloads the FIB entry.
358	
359	XXX: add/mod/del IPv6 FIB API
360	
361	Nexthop Resolution
362	^^^^^^^^^^^^^^^^^^
363	
364	The FIB entry's nexthop list contains the nexthop tuple (gateway, dev), but for
365	the switch device to forward the packet with the correct dst mac address, the
366	nexthop gateways must be resolved to the neighbor's mac address.  Neighbor mac
367	address discovery comes via the ARP (or ND) process and is available via the
368	arp_tbl neighbor table.  To resolve the routes nexthop gateways, the driver
369	should trigger the kernel's neighbor resolution process.  See the rocker
370	driver's rocker_port_ipv4_resolve() for an example.
371	
372	The driver can monitor for updates to arp_tbl using the netevent notifier
373	NETEVENT_NEIGH_UPDATE.  The device can be programmed with resolved nexthops
374	for the routes as arp_tbl updates.  The driver implements ndo_neigh_destroy
375	to know when arp_tbl neighbor entries are purged from the port.
376	
377	Transaction item queue
378	^^^^^^^^^^^^^^^^^^^^^^
379	
380	For switchdev ops attr_set and obj_add, there is a 2 phase transaction model
381	used. First phase is to "prepare" anything needed, including various checks,
382	memory allocation, etc. The goal is to handle the stuff that is not unlikely
383	to fail here. The second phase is to "commit" the actual changes.
384	
385	Switchdev provides an infrastructure for sharing items (for example memory
386	allocations) between the two phases.
387	
388	The object created by a driver in "prepare" phase and it is queued up by:
389	switchdev_trans_item_enqueue()
390	During the "commit" phase, the driver gets the object by:
391	switchdev_trans_item_dequeue()
392	
393	If a transaction is aborted during "prepare" phase, switchdev code will handle
394	cleanup of the queued-up objects.
Hide Line Numbers
About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Information is copyright its respective author. All material is available from the Linux Kernel Source distributed under a GPL License. This page is provided as a free service by mjmwired.net.