1============
2Architecture
3============
4
5This document describes the **Distributed Switch Architecture (DSA)** subsystem
6design principles, limitations, interactions with other subsystems, and how to
7develop drivers for this subsystem as well as a TODO for developers interested
8in joining the effort.
9
10Design principles
11=================
12
13The Distributed Switch Architecture is a subsystem which was primarily designed
14to support Marvell Ethernet switches (MV88E6xxx, a.k.a Linkstreet product line)
15using Linux, but has since evolved to support other vendors as well.
16
17The original philosophy behind this design was to be able to use unmodified
18Linux tools such as bridge, iproute2, ifconfig to work transparently whether
19they configured/queried a switch port network device or a regular network
20device.
21
22An Ethernet switch is typically comprised of multiple front-panel ports, and one
23or more CPU or management port. The DSA subsystem currently relies on the
24presence of a management port connected to an Ethernet controller capable of
25receiving Ethernet frames from the switch. This is a very common setup for all
26kinds of Ethernet switches found in Small Home and Office products: routers,
27gateways, or even top-of-the rack switches. This host Ethernet controller will
28be later referred to as "master" and "cpu" in DSA terminology and code.
29
30The D in DSA stands for Distributed, because the subsystem has been designed
31with the ability to configure and manage cascaded switches on top of each other
32using upstream and downstream Ethernet links between switches. These specific
33ports are referred to as "dsa" ports in DSA terminology and code. A collection
34of multiple switches connected to each other is called a "switch tree".
35
36For each front-panel port, DSA will create specialized network devices which are
37used as controlling and data-flowing endpoints for use by the Linux networking
38stack. These specialized network interfaces are referred to as "slave" network
39interfaces in DSA terminology and code.
40
41The ideal case for using DSA is when an Ethernet switch supports a "switch tag"
42which is a hardware feature making the switch insert a specific tag for each
43Ethernet frames it received to/from specific ports to help the management
44interface figure out:
45
46- what port is this frame coming from
47- what was the reason why this frame got forwarded
48- how to send CPU originated traffic to specific ports
49
50The subsystem does support switches not capable of inserting/stripping tags, but
51the features might be slightly limited in that case (traffic separation relies
52on Port-based VLAN IDs).
53
54Note that DSA does not currently create network interfaces for the "cpu" and
55"dsa" ports because:
56
57- the "cpu" port is the Ethernet switch facing side of the management
58  controller, and as such, would create a duplication of feature, since you
59  would get two interfaces for the same conduit: master netdev, and "cpu" netdev
60
61- the "dsa" port(s) are just conduits between two or more switches, and as such
62  cannot really be used as proper network interfaces either, only the
63  downstream, or the top-most upstream interface makes sense with that model
64
65Switch tagging protocols
66------------------------
67
68DSA supports many vendor-specific tagging protocols, one software-defined
69tagging protocol, and a tag-less mode as well (``DSA_TAG_PROTO_NONE``).
70
71The exact format of the tag protocol is vendor specific, but in general, they
72all contain something which:
73
74- identifies which port the Ethernet frame came from/should be sent to
75- provides a reason why this frame was forwarded to the management interface
76
77All tagging protocols are in ``net/dsa/tag_*.c`` files and implement the
78methods of the ``struct dsa_device_ops`` structure, which are detailed below.
79
80Tagging protocols generally fall in one of three categories:
81
821. The switch-specific frame header is located before the Ethernet header,
83   shifting to the right (from the perspective of the DSA master's frame
84   parser) the MAC DA, MAC SA, EtherType and the entire L2 payload.
852. The switch-specific frame header is located before the EtherType, keeping
86   the MAC DA and MAC SA in place from the DSA master's perspective, but
87   shifting the 'real' EtherType and L2 payload to the right.
883. The switch-specific frame header is located at the tail of the packet,
89   keeping all frame headers in place and not altering the view of the packet
90   that the DSA master's frame parser has.
91
92A tagging protocol may tag all packets with switch tags of the same length, or
93the tag length might vary (for example packets with PTP timestamps might
94require an extended switch tag, or there might be one tag length on TX and a
95different one on RX). Either way, the tagging protocol driver must populate the
96``struct dsa_device_ops::overhead`` with the length in octets of the longest
97switch frame header. The DSA framework will automatically adjust the MTU of the
98master interface to accomodate for this extra size in order for DSA user ports
99to support the standard MTU (L2 payload length) of 1500 octets. The ``overhead``
100is also used to request from the network stack, on a best-effort basis, the
101allocation of packets with a ``needed_headroom`` or ``needed_tailroom``
102sufficient such that the act of pushing the switch tag on transmission of a
103packet does not cause it to reallocate due to lack of memory.
104
105Even though applications are not expected to parse DSA-specific frame headers,
106the format on the wire of the tagging protocol represents an Application Binary
107Interface exposed by the kernel towards user space, for decoders such as
108``libpcap``. The tagging protocol driver must populate the ``proto`` member of
109``struct dsa_device_ops`` with a value that uniquely describes the
110characteristics of the interaction required between the switch hardware and the
111data path driver: the offset of each bit field within the frame header and any
112stateful processing required to deal with the frames (as may be required for
113PTP timestamping).
114
115From the perspective of the network stack, all switches within the same DSA
116switch tree use the same tagging protocol. In case of a packet transiting a
117fabric with more than one switch, the switch-specific frame header is inserted
118by the first switch in the fabric that the packet was received on. This header
119typically contains information regarding its type (whether it is a control
120frame that must be trapped to the CPU, or a data frame to be forwarded).
121Control frames should be decapsulated only by the software data path, whereas
122data frames might also be autonomously forwarded towards other user ports of
123other switches from the same fabric, and in this case, the outermost switch
124ports must decapsulate the packet.
125
126Note that in certain cases, it might be the case that the tagging format used
127by a leaf switch (not connected directly to the CPU) to not be the same as what
128the network stack sees. This can be seen with Marvell switch trees, where the
129CPU port can be configured to use either the DSA or the Ethertype DSA (EDSA)
130format, but the DSA links are configured to use the shorter (without Ethertype)
131DSA frame header, in order to reduce the autonomous packet forwarding overhead.
132It still remains the case that, if the DSA switch tree is configured for the
133EDSA tagging protocol, the operating system sees EDSA-tagged packets from the
134leaf switches that tagged them with the shorter DSA header. This can be done
135because the Marvell switch connected directly to the CPU is configured to
136perform tag translation between DSA and EDSA (which is simply the operation of
137adding or removing the ``ETH_P_EDSA`` EtherType and some padding octets).
138
139It is possible to construct cascaded setups of DSA switches even if their
140tagging protocols are not compatible with one another. In this case, there are
141no DSA links in this fabric, and each switch constitutes a disjoint DSA switch
142tree. The DSA links are viewed as simply a pair of a DSA master (the out-facing
143port of the upstream DSA switch) and a CPU port (the in-facing port of the
144downstream DSA switch).
145
146The tagging protocol of the attached DSA switch tree can be viewed through the
147``dsa/tagging`` sysfs attribute of the DSA master::
148
149    cat /sys/class/net/eth0/dsa/tagging
150
151If the hardware and driver are capable, the tagging protocol of the DSA switch
152tree can be changed at runtime. This is done by writing the new tagging
153protocol name to the same sysfs device attribute as above (the DSA master and
154all attached switch ports must be down while doing this).
155
156It is desirable that all tagging protocols are testable with the ``dsa_loop``
157mockup driver, which can be attached to any network interface. The goal is that
158any network interface should be capable of transmitting the same packet in the
159same way, and the tagger should decode the same received packet in the same way
160regardless of the driver used for the switch control path, and the driver used
161for the DSA master.
162
163The transmission of a packet goes through the tagger's ``xmit`` function.
164The passed ``struct sk_buff *skb`` has ``skb->data`` pointing at
165``skb_mac_header(skb)``, i.e. at the destination MAC address, and the passed
166``struct net_device *dev`` represents the virtual DSA user network interface
167whose hardware counterpart the packet must be steered to (i.e. ``swp0``).
168The job of this method is to prepare the skb in a way that the switch will
169understand what egress port the packet is for (and not deliver it towards other
170ports). Typically this is fulfilled by pushing a frame header. Checking for
171insufficient size in the skb headroom or tailroom is unnecessary provided that
172the ``overhead`` and ``tail_tag`` properties were filled out properly, because
173DSA ensures there is enough space before calling this method.
174
175The reception of a packet goes through the tagger's ``rcv`` function. The
176passed ``struct sk_buff *skb`` has ``skb->data`` pointing at
177``skb_mac_header(skb) + ETH_ALEN`` octets, i.e. to where the first octet after
178the EtherType would have been, were this frame not tagged. The role of this
179method is to consume the frame header, adjust ``skb->data`` to really point at
180the first octet after the EtherType, and to change ``skb->dev`` to point to the
181virtual DSA user network interface corresponding to the physical front-facing
182switch port that the packet was received on.
183
184Since tagging protocols in category 1 and 2 break software (and most often also
185hardware) packet dissection on the DSA master, features such as RPS (Receive
186Packet Steering) on the DSA master would be broken. The DSA framework deals
187with this by hooking into the flow dissector and shifting the offset at which
188the IP header is to be found in the tagged frame as seen by the DSA master.
189This behavior is automatic based on the ``overhead`` value of the tagging
190protocol. If not all packets are of equal size, the tagger can implement the
191``flow_dissect`` method of the ``struct dsa_device_ops`` and override this
192default behavior by specifying the correct offset incurred by each individual
193RX packet. Tail taggers do not cause issues to the flow dissector.
194
195Due to various reasons (most common being category 1 taggers being associated
196with DSA-unaware masters, mangling what the master perceives as MAC DA), the
197tagging protocol may require the DSA master to operate in promiscuous mode, to
198receive all frames regardless of the value of the MAC DA. This can be done by
199setting the ``promisc_on_master`` property of the ``struct dsa_device_ops``.
200Note that this assumes a DSA-unaware master driver, which is the norm.
201
202Hardware manufacturers are strongly discouraged to do this, but some tagging
203protocols might not provide source port information on RX for all packets, but
204e.g. only for control traffic (link-local PDUs). In this case, by implementing
205the ``filter`` method of ``struct dsa_device_ops``, the tagger might select
206which packets are to be redirected on RX towards the virtual DSA user network
207interfaces, and which are to be left in the DSA master's RX data path.
208
209It might also happen (although silicon vendors are strongly discouraged to
210produce hardware like this) that a tagging protocol splits the switch-specific
211information into a header portion and a tail portion, therefore not falling
212cleanly into any of the above 3 categories. DSA does not support this
213configuration.
214
215Master network devices
216----------------------
217
218Master network devices are regular, unmodified Linux network device drivers for
219the CPU/management Ethernet interface. Such a driver might occasionally need to
220know whether DSA is enabled (e.g.: to enable/disable specific offload features),
221but the DSA subsystem has been proven to work with industry standard drivers:
222``e1000e,`` ``mv643xx_eth`` etc. without having to introduce modifications to these
223drivers. Such network devices are also often referred to as conduit network
224devices since they act as a pipe between the host processor and the hardware
225Ethernet switch.
226
227Networking stack hooks
228----------------------
229
230When a master netdev is used with DSA, a small hook is placed in the
231networking stack is in order to have the DSA subsystem process the Ethernet
232switch specific tagging protocol. DSA accomplishes this by registering a
233specific (and fake) Ethernet type (later becoming ``skb->protocol``) with the
234networking stack, this is also known as a ``ptype`` or ``packet_type``. A typical
235Ethernet Frame receive sequence looks like this:
236
237Master network device (e.g.: e1000e):
238
2391. Receive interrupt fires:
240
241        - receive function is invoked
242        - basic packet processing is done: getting length, status etc.
243        - packet is prepared to be processed by the Ethernet layer by calling
244          ``eth_type_trans``
245
2462. net/ethernet/eth.c::
247
248          eth_type_trans(skb, dev)
249                  if (dev->dsa_ptr != NULL)
250                          -> skb->protocol = ETH_P_XDSA
251
2523. drivers/net/ethernet/\*::
253
254          netif_receive_skb(skb)
255                  -> iterate over registered packet_type
256                          -> invoke handler for ETH_P_XDSA, calls dsa_switch_rcv()
257
2584. net/dsa/dsa.c::
259
260          -> dsa_switch_rcv()
261                  -> invoke switch tag specific protocol handler in 'net/dsa/tag_*.c'
262
2635. net/dsa/tag_*.c:
264
265        - inspect and strip switch tag protocol to determine originating port
266        - locate per-port network device
267        - invoke ``eth_type_trans()`` with the DSA slave network device
268        - invoked ``netif_receive_skb()``
269
270Past this point, the DSA slave network devices get delivered regular Ethernet
271frames that can be processed by the networking stack.
272
273Slave network devices
274---------------------
275
276Slave network devices created by DSA are stacked on top of their master network
277device, each of these network interfaces will be responsible for being a
278controlling and data-flowing end-point for each front-panel port of the switch.
279These interfaces are specialized in order to:
280
281- insert/remove the switch tag protocol (if it exists) when sending traffic
282  to/from specific switch ports
283- query the switch for ethtool operations: statistics, link state,
284  Wake-on-LAN, register dumps...
285- external/internal PHY management: link, auto-negotiation etc.
286
287These slave network devices have custom net_device_ops and ethtool_ops function
288pointers which allow DSA to introduce a level of layering between the networking
289stack/ethtool, and the switch driver implementation.
290
291Upon frame transmission from these slave network devices, DSA will look up which
292switch tagging protocol is currently registered with these network devices, and
293invoke a specific transmit routine which takes care of adding the relevant
294switch tag in the Ethernet frames.
295
296These frames are then queued for transmission using the master network device
297``ndo_start_xmit()`` function, since they contain the appropriate switch tag, the
298Ethernet switch will be able to process these incoming frames from the
299management interface and delivers these frames to the physical switch port.
300
301Graphical representation
302------------------------
303
304Summarized, this is basically how DSA looks like from a network device
305perspective::
306
307                Unaware application
308              opens and binds socket
309                       |  ^
310                       |  |
311           +-----------v--|--------------------+
312           |+------+ +------+ +------+ +------+|
313           || swp0 | | swp1 | | swp2 | | swp3 ||
314           |+------+-+------+-+------+-+------+|
315           |          DSA switch driver        |
316           +-----------------------------------+
317                         |        ^
318            Tag added by |        | Tag consumed by
319           switch driver |        | switch driver
320                         v        |
321           +-----------------------------------+
322           | Unmodified host interface driver  | Software
323   --------+-----------------------------------+------------
324           |       Host interface (eth0)       | Hardware
325           +-----------------------------------+
326                         |        ^
327         Tag consumed by |        | Tag added by
328         switch hardware |        | switch hardware
329                         v        |
330           +-----------------------------------+
331           |               Switch              |
332           |+------+ +------+ +------+ +------+|
333           || swp0 | | swp1 | | swp2 | | swp3 ||
334           ++------+-+------+-+------+-+------++
335
336Slave MDIO bus
337--------------
338
339In order to be able to read to/from a switch PHY built into it, DSA creates a
340slave MDIO bus which allows a specific switch driver to divert and intercept
341MDIO reads/writes towards specific PHY addresses. In most MDIO-connected
342switches, these functions would utilize direct or indirect PHY addressing mode
343to return standard MII registers from the switch builtin PHYs, allowing the PHY
344library and/or to return link status, link partner pages, auto-negotiation
345results etc..
346
347For Ethernet switches which have both external and internal MDIO busses, the
348slave MII bus can be utilized to mux/demux MDIO reads and writes towards either
349internal or external MDIO devices this switch might be connected to: internal
350PHYs, external PHYs, or even external switches.
351
352Data structures
353---------------
354
355DSA data structures are defined in ``include/net/dsa.h`` as well as
356``net/dsa/dsa_priv.h``:
357
358- ``dsa_chip_data``: platform data configuration for a given switch device,
359  this structure describes a switch device's parent device, its address, as
360  well as various properties of its ports: names/labels, and finally a routing
361  table indication (when cascading switches)
362
363- ``dsa_platform_data``: platform device configuration data which can reference
364  a collection of dsa_chip_data structure if multiples switches are cascaded,
365  the master network device this switch tree is attached to needs to be
366  referenced
367
368- ``dsa_switch_tree``: structure assigned to the master network device under
369  ``dsa_ptr``, this structure references a dsa_platform_data structure as well as
370  the tagging protocol supported by the switch tree, and which receive/transmit
371  function hooks should be invoked, information about the directly attached
372  switch is also provided: CPU port. Finally, a collection of dsa_switch are
373  referenced to address individual switches in the tree.
374
375- ``dsa_switch``: structure describing a switch device in the tree, referencing
376  a ``dsa_switch_tree`` as a backpointer, slave network devices, master network
377  device, and a reference to the backing``dsa_switch_ops``
378
379- ``dsa_switch_ops``: structure referencing function pointers, see below for a
380  full description.
381
382Design limitations
383==================
384
385Lack of CPU/DSA network devices
386-------------------------------
387
388DSA does not currently create slave network devices for the CPU or DSA ports, as
389described before. This might be an issue in the following cases:
390
391- inability to fetch switch CPU port statistics counters using ethtool, which
392  can make it harder to debug MDIO switch connected using xMII interfaces
393
394- inability to configure the CPU port link parameters based on the Ethernet
395  controller capabilities attached to it: http://patchwork.ozlabs.org/patch/509806/
396
397- inability to configure specific VLAN IDs / trunking VLANs between switches
398  when using a cascaded setup
399
400Common pitfalls using DSA setups
401--------------------------------
402
403Once a master network device is configured to use DSA (dev->dsa_ptr becomes
404non-NULL), and the switch behind it expects a tagging protocol, this network
405interface can only exclusively be used as a conduit interface. Sending packets
406directly through this interface (e.g.: opening a socket using this interface)
407will not make us go through the switch tagging protocol transmit function, so
408the Ethernet switch on the other end, expecting a tag will typically drop this
409frame.
410
411Interactions with other subsystems
412==================================
413
414DSA currently leverages the following subsystems:
415
416- MDIO/PHY library: ``drivers/net/phy/phy.c``, ``mdio_bus.c``
417- Switchdev:``net/switchdev/*``
418- Device Tree for various of_* functions
419- Devlink: ``net/core/devlink.c``
420
421MDIO/PHY library
422----------------
423
424Slave network devices exposed by DSA may or may not be interfacing with PHY
425devices (``struct phy_device`` as defined in ``include/linux/phy.h)``, but the DSA
426subsystem deals with all possible combinations:
427
428- internal PHY devices, built into the Ethernet switch hardware
429- external PHY devices, connected via an internal or external MDIO bus
430- internal PHY devices, connected via an internal MDIO bus
431- special, non-autonegotiated or non MDIO-managed PHY devices: SFPs, MoCA; a.k.a
432  fixed PHYs
433
434The PHY configuration is done by the ``dsa_slave_phy_setup()`` function and the
435logic basically looks like this:
436
437- if Device Tree is used, the PHY device is looked up using the standard
438  "phy-handle" property, if found, this PHY device is created and registered
439  using ``of_phy_connect()``
440
441- if Device Tree is used, and the PHY device is "fixed", that is, conforms to
442  the definition of a non-MDIO managed PHY as defined in
443  ``Documentation/devicetree/bindings/net/fixed-link.txt``, the PHY is registered
444  and connected transparently using the special fixed MDIO bus driver
445
446- finally, if the PHY is built into the switch, as is very common with
447  standalone switch packages, the PHY is probed using the slave MII bus created
448  by DSA
449
450
451SWITCHDEV
452---------
453
454DSA directly utilizes SWITCHDEV when interfacing with the bridge layer, and
455more specifically with its VLAN filtering portion when configuring VLANs on top
456of per-port slave network devices. As of today, the only SWITCHDEV objects
457supported by DSA are the FDB and VLAN objects.
458
459Devlink
460-------
461
462DSA registers one devlink device per physical switch in the fabric.
463For each devlink device, every physical port (i.e. user ports, CPU ports, DSA
464links or unused ports) is exposed as a devlink port.
465
466DSA drivers can make use of the following devlink features:
467
468- Regions: debugging feature which allows user space to dump driver-defined
469  areas of hardware information in a low-level, binary format. Both global
470  regions as well as per-port regions are supported. It is possible to export
471  devlink regions even for pieces of data that are already exposed in some way
472  to the standard iproute2 user space programs (ip-link, bridge), like address
473  tables and VLAN tables. For example, this might be useful if the tables
474  contain additional hardware-specific details which are not visible through
475  the iproute2 abstraction, or it might be useful to inspect these tables on
476  the non-user ports too, which are invisible to iproute2 because no network
477  interface is registered for them.
478- Params: a feature which enables user to configure certain low-level tunable
479  knobs pertaining to the device. Drivers may implement applicable generic
480  devlink params, or may add new device-specific devlink params.
481- Resources: a monitoring feature which enables users to see the degree of
482  utilization of certain hardware tables in the device, such as FDB, VLAN, etc.
483- Shared buffers: a QoS feature for adjusting and partitioning memory and frame
484  reservations per port and per traffic class, in the ingress and egress
485  directions, such that low-priority bulk traffic does not impede the
486  processing of high-priority critical traffic.
487
488For more details, consult ``Documentation/networking/devlink/``.
489
490Device Tree
491-----------
492
493DSA features a standardized binding which is documented in
494``Documentation/devicetree/bindings/net/dsa/dsa.txt``. PHY/MDIO library helper
495functions such as ``of_get_phy_mode()``, ``of_phy_connect()`` are also used to query
496per-port PHY specific details: interface connection, MDIO bus location etc..
497
498Driver development
499==================
500
501DSA switch drivers need to implement a dsa_switch_ops structure which will
502contain the various members described below.
503
504``register_switch_driver()`` registers this dsa_switch_ops in its internal list
505of drivers to probe for. ``unregister_switch_driver()`` does the exact opposite.
506
507Unless requested differently by setting the priv_size member accordingly, DSA
508does not allocate any driver private context space.
509
510Switch configuration
511--------------------
512
513- ``tag_protocol``: this is to indicate what kind of tagging protocol is supported,
514  should be a valid value from the ``dsa_tag_protocol`` enum
515
516- ``probe``: probe routine which will be invoked by the DSA platform device upon
517  registration to test for the presence/absence of a switch device. For MDIO
518  devices, it is recommended to issue a read towards internal registers using
519  the switch pseudo-PHY and return whether this is a supported device. For other
520  buses, return a non-NULL string
521
522- ``setup``: setup function for the switch, this function is responsible for setting
523  up the ``dsa_switch_ops`` private structure with all it needs: register maps,
524  interrupts, mutexes, locks etc.. This function is also expected to properly
525  configure the switch to separate all network interfaces from each other, that
526  is, they should be isolated by the switch hardware itself, typically by creating
527  a Port-based VLAN ID for each port and allowing only the CPU port and the
528  specific port to be in the forwarding vector. Ports that are unused by the
529  platform should be disabled. Past this function, the switch is expected to be
530  fully configured and ready to serve any kind of request. It is recommended
531  to issue a software reset of the switch during this setup function in order to
532  avoid relying on what a previous software agent such as a bootloader/firmware
533  may have previously configured.
534
535PHY devices and link management
536-------------------------------
537
538- ``get_phy_flags``: Some switches are interfaced to various kinds of Ethernet PHYs,
539  if the PHY library PHY driver needs to know about information it cannot obtain
540  on its own (e.g.: coming from switch memory mapped registers), this function
541  should return a 32-bits bitmask of "flags", that is private between the switch
542  driver and the Ethernet PHY driver in ``drivers/net/phy/\*``.
543
544- ``phy_read``: Function invoked by the DSA slave MDIO bus when attempting to read
545  the switch port MDIO registers. If unavailable, return 0xffff for each read.
546  For builtin switch Ethernet PHYs, this function should allow reading the link
547  status, auto-negotiation results, link partner pages etc..
548
549- ``phy_write``: Function invoked by the DSA slave MDIO bus when attempting to write
550  to the switch port MDIO registers. If unavailable return a negative error
551  code.
552
553- ``adjust_link``: Function invoked by the PHY library when a slave network device
554  is attached to a PHY device. This function is responsible for appropriately
555  configuring the switch port link parameters: speed, duplex, pause based on
556  what the ``phy_device`` is providing.
557
558- ``fixed_link_update``: Function invoked by the PHY library, and specifically by
559  the fixed PHY driver asking the switch driver for link parameters that could
560  not be auto-negotiated, or obtained by reading the PHY registers through MDIO.
561  This is particularly useful for specific kinds of hardware such as QSGMII,
562  MoCA or other kinds of non-MDIO managed PHYs where out of band link
563  information is obtained
564
565Ethtool operations
566------------------
567
568- ``get_strings``: ethtool function used to query the driver's strings, will
569  typically return statistics strings, private flags strings etc.
570
571- ``get_ethtool_stats``: ethtool function used to query per-port statistics and
572  return their values. DSA overlays slave network devices general statistics:
573  RX/TX counters from the network device, with switch driver specific statistics
574  per port
575
576- ``get_sset_count``: ethtool function used to query the number of statistics items
577
578- ``get_wol``: ethtool function used to obtain Wake-on-LAN settings per-port, this
579  function may, for certain implementations also query the master network device
580  Wake-on-LAN settings if this interface needs to participate in Wake-on-LAN
581
582- ``set_wol``: ethtool function used to configure Wake-on-LAN settings per-port,
583  direct counterpart to set_wol with similar restrictions
584
585- ``set_eee``: ethtool function which is used to configure a switch port EEE (Green
586  Ethernet) settings, can optionally invoke the PHY library to enable EEE at the
587  PHY level if relevant. This function should enable EEE at the switch port MAC
588  controller and data-processing logic
589
590- ``get_eee``: ethtool function which is used to query a switch port EEE settings,
591  this function should return the EEE state of the switch port MAC controller
592  and data-processing logic as well as query the PHY for its currently configured
593  EEE settings
594
595- ``get_eeprom_len``: ethtool function returning for a given switch the EEPROM
596  length/size in bytes
597
598- ``get_eeprom``: ethtool function returning for a given switch the EEPROM contents
599
600- ``set_eeprom``: ethtool function writing specified data to a given switch EEPROM
601
602- ``get_regs_len``: ethtool function returning the register length for a given
603  switch
604
605- ``get_regs``: ethtool function returning the Ethernet switch internal register
606  contents. This function might require user-land code in ethtool to
607  pretty-print register values and registers
608
609Power management
610----------------
611
612- ``suspend``: function invoked by the DSA platform device when the system goes to
613  suspend, should quiesce all Ethernet switch activities, but keep ports
614  participating in Wake-on-LAN active as well as additional wake-up logic if
615  supported
616
617- ``resume``: function invoked by the DSA platform device when the system resumes,
618  should resume all Ethernet switch activities and re-configure the switch to be
619  in a fully active state
620
621- ``port_enable``: function invoked by the DSA slave network device ndo_open
622  function when a port is administratively brought up, this function should be
623  fully enabling a given switch port. DSA takes care of marking the port with
624  ``BR_STATE_BLOCKING`` if the port is a bridge member, or ``BR_STATE_FORWARDING`` if it
625  was not, and propagating these changes down to the hardware
626
627- ``port_disable``: function invoked by the DSA slave network device ndo_close
628  function when a port is administratively brought down, this function should be
629  fully disabling a given switch port. DSA takes care of marking the port with
630  ``BR_STATE_DISABLED`` and propagating changes to the hardware if this port is
631  disabled while being a bridge member
632
633Bridge layer
634------------
635
636- ``port_bridge_join``: bridge layer function invoked when a given switch port is
637  added to a bridge, this function should be doing the necessary at the switch
638  level to permit the joining port from being added to the relevant logical
639  domain for it to ingress/egress traffic with other members of the bridge.
640
641- ``port_bridge_leave``: bridge layer function invoked when a given switch port is
642  removed from a bridge, this function should be doing the necessary at the
643  switch level to deny the leaving port from ingress/egress traffic from the
644  remaining bridge members. When the port leaves the bridge, it should be aged
645  out at the switch hardware for the switch to (re) learn MAC addresses behind
646  this port.
647
648- ``port_stp_state_set``: bridge layer function invoked when a given switch port STP
649  state is computed by the bridge layer and should be propagated to switch
650  hardware to forward/block/learn traffic. The switch driver is responsible for
651  computing a STP state change based on current and asked parameters and perform
652  the relevant ageing based on the intersection results
653
654- ``port_bridge_flags``: bridge layer function invoked when a port must
655  configure its settings for e.g. flooding of unknown traffic or source address
656  learning. The switch driver is responsible for initial setup of the
657  standalone ports with address learning disabled and egress flooding of all
658  types of traffic, then the DSA core notifies of any change to the bridge port
659  flags when the port joins and leaves a bridge. DSA does not currently manage
660  the bridge port flags for the CPU port. The assumption is that address
661  learning should be statically enabled (if supported by the hardware) on the
662  CPU port, and flooding towards the CPU port should also be enabled, due to a
663  lack of an explicit address filtering mechanism in the DSA core.
664
665Bridge VLAN filtering
666---------------------
667
668- ``port_vlan_filtering``: bridge layer function invoked when the bridge gets
669  configured for turning on or off VLAN filtering. If nothing specific needs to
670  be done at the hardware level, this callback does not need to be implemented.
671  When VLAN filtering is turned on, the hardware must be programmed with
672  rejecting 802.1Q frames which have VLAN IDs outside of the programmed allowed
673  VLAN ID map/rules.  If there is no PVID programmed into the switch port,
674  untagged frames must be rejected as well. When turned off the switch must
675  accept any 802.1Q frames irrespective of their VLAN ID, and untagged frames are
676  allowed.
677
678- ``port_vlan_add``: bridge layer function invoked when a VLAN is configured
679  (tagged or untagged) for the given switch port. If the operation is not
680  supported by the hardware, this function should return ``-EOPNOTSUPP`` to
681  inform the bridge code to fallback to a software implementation.
682
683- ``port_vlan_del``: bridge layer function invoked when a VLAN is removed from the
684  given switch port
685
686- ``port_vlan_dump``: bridge layer function invoked with a switchdev callback
687  function that the driver has to call for each VLAN the given port is a member
688  of. A switchdev object is used to carry the VID and bridge flags.
689
690- ``port_fdb_add``: bridge layer function invoked when the bridge wants to install a
691  Forwarding Database entry, the switch hardware should be programmed with the
692  specified address in the specified VLAN Id in the forwarding database
693  associated with this VLAN ID. If the operation is not supported, this
694  function should return ``-EOPNOTSUPP`` to inform the bridge code to fallback to
695  a software implementation.
696
697.. note:: VLAN ID 0 corresponds to the port private database, which, in the context
698        of DSA, would be its port-based VLAN, used by the associated bridge device.
699
700- ``port_fdb_del``: bridge layer function invoked when the bridge wants to remove a
701  Forwarding Database entry, the switch hardware should be programmed to delete
702  the specified MAC address from the specified VLAN ID if it was mapped into
703  this port forwarding database
704
705- ``port_fdb_dump``: bridge layer function invoked with a switchdev callback
706  function that the driver has to call for each MAC address known to be behind
707  the given port. A switchdev object is used to carry the VID and FDB info.
708
709- ``port_mdb_add``: bridge layer function invoked when the bridge wants to install
710  a multicast database entry. If the operation is not supported, this function
711  should return ``-EOPNOTSUPP`` to inform the bridge code to fallback to a
712  software implementation. The switch hardware should be programmed with the
713  specified address in the specified VLAN ID in the forwarding database
714  associated with this VLAN ID.
715
716.. note:: VLAN ID 0 corresponds to the port private database, which, in the context
717        of DSA, would be its port-based VLAN, used by the associated bridge device.
718
719- ``port_mdb_del``: bridge layer function invoked when the bridge wants to remove a
720  multicast database entry, the switch hardware should be programmed to delete
721  the specified MAC address from the specified VLAN ID if it was mapped into
722  this port forwarding database.
723
724- ``port_mdb_dump``: bridge layer function invoked with a switchdev callback
725  function that the driver has to call for each MAC address known to be behind
726  the given port. A switchdev object is used to carry the VID and MDB info.
727
728Link aggregation
729----------------
730
731Link aggregation is implemented in the Linux networking stack by the bonding
732and team drivers, which are modeled as virtual, stackable network interfaces.
733DSA is capable of offloading a link aggregation group (LAG) to hardware that
734supports the feature, and supports bridging between physical ports and LAGs,
735as well as between LAGs. A bonding/team interface which holds multiple physical
736ports constitutes a logical port, although DSA has no explicit concept of a
737logical port at the moment. Due to this, events where a LAG joins/leaves a
738bridge are treated as if all individual physical ports that are members of that
739LAG join/leave the bridge. Switchdev port attributes (VLAN filtering, STP
740state, etc) and objects (VLANs, MDB entries) offloaded to a LAG as bridge port
741are treated similarly: DSA offloads the same switchdev object / port attribute
742on all members of the LAG. Static bridge FDB entries on a LAG are not yet
743supported, since the DSA driver API does not have the concept of a logical port
744ID.
745
746- ``port_lag_join``: function invoked when a given switch port is added to a
747  LAG. The driver may return ``-EOPNOTSUPP``, and in this case, DSA will fall
748  back to a software implementation where all traffic from this port is sent to
749  the CPU.
750- ``port_lag_leave``: function invoked when a given switch port leaves a LAG
751  and returns to operation as a standalone port.
752- ``port_lag_change``: function invoked when the link state of any member of
753  the LAG changes, and the hashing function needs rebalancing to only make use
754  of the subset of physical LAG member ports that are up.
755
756Drivers that benefit from having an ID associated with each offloaded LAG
757can optionally populate ``ds->num_lag_ids`` from the ``dsa_switch_ops::setup``
758method. The LAG ID associated with a bonding/team interface can then be
759retrieved by a DSA switch driver using the ``dsa_lag_id`` function.
760
761IEC 62439-2 (MRP)
762-----------------
763
764The Media Redundancy Protocol is a topology management protocol optimized for
765fast fault recovery time for ring networks, which has some components
766implemented as a function of the bridge driver. MRP uses management PDUs
767(Test, Topology, LinkDown/Up, Option) sent at a multicast destination MAC
768address range of 01:15:4e:00:00:0x and with an EtherType of 0x88e3.
769Depending on the node's role in the ring (MRM: Media Redundancy Manager,
770MRC: Media Redundancy Client, MRA: Media Redundancy Automanager), certain MRP
771PDUs might need to be terminated locally and others might need to be forwarded.
772An MRM might also benefit from offloading to hardware the creation and
773transmission of certain MRP PDUs (Test).
774
775Normally an MRP instance can be created on top of any network interface,
776however in the case of a device with an offloaded data path such as DSA, it is
777necessary for the hardware, even if it is not MRP-aware, to be able to extract
778the MRP PDUs from the fabric before the driver can proceed with the software
779implementation. DSA today has no driver which is MRP-aware, therefore it only
780listens for the bare minimum switchdev objects required for the software assist
781to work properly. The operations are detailed below.
782
783- ``port_mrp_add`` and ``port_mrp_del``: notifies driver when an MRP instance
784  with a certain ring ID, priority, primary port and secondary port is
785  created/deleted.
786- ``port_mrp_add_ring_role`` and ``port_mrp_del_ring_role``: function invoked
787  when an MRP instance changes ring roles between MRM or MRC. This affects
788  which MRP PDUs should be trapped to software and which should be autonomously
789  forwarded.
790
791IEC 62439-3 (HSR/PRP)
792---------------------
793
794The Parallel Redundancy Protocol (PRP) is a network redundancy protocol which
795works by duplicating and sequence numbering packets through two independent L2
796networks (which are unaware of the PRP tail tags carried in the packets), and
797eliminating the duplicates at the receiver. The High-availability Seamless
798Redundancy (HSR) protocol is similar in concept, except all nodes that carry
799the redundant traffic are aware of the fact that it is HSR-tagged (because HSR
800uses a header with an EtherType of 0x892f) and are physically connected in a
801ring topology. Both HSR and PRP use supervision frames for monitoring the
802health of the network and for discovery of other nodes.
803
804In Linux, both HSR and PRP are implemented in the hsr driver, which
805instantiates a virtual, stackable network interface with two member ports.
806The driver only implements the basic roles of DANH (Doubly Attached Node
807implementing HSR) and DANP (Doubly Attached Node implementing PRP); the roles
808of RedBox and QuadBox are not implemented (therefore, bridging a hsr network
809interface with a physical switch port does not produce the expected result).
810
811A driver which is able of offloading certain functions of a DANP or DANH should
812declare the corresponding netdev features as indicated by the documentation at
813``Documentation/networking/netdev-features.rst``. Additionally, the following
814methods must be implemented:
815
816- ``port_hsr_join``: function invoked when a given switch port is added to a
817  DANP/DANH. The driver may return ``-EOPNOTSUPP`` and in this case, DSA will
818  fall back to a software implementation where all traffic from this port is
819  sent to the CPU.
820- ``port_hsr_leave``: function invoked when a given switch port leaves a
821  DANP/DANH and returns to normal operation as a standalone port.
822
823TODO
824====
825
826Making SWITCHDEV and DSA converge towards an unified codebase
827-------------------------------------------------------------
828
829SWITCHDEV properly takes care of abstracting the networking stack with offload
830capable hardware, but does not enforce a strict switch device driver model. On
831the other DSA enforces a fairly strict device driver model, and deals with most
832of the switch specific. At some point we should envision a merger between these
833two subsystems and get the best of both worlds.
834
835Other hanging fruits
836--------------------
837
838- allowing more than one CPU/management interface:
839  http://comments.gmane.org/gmane.linux.network/365657
840