1134 lines · plain
1============2Architecture3============4 5This document describes the **Distributed Switch Architecture (DSA)** subsystem6design principles, limitations, interactions with other subsystems, and how to7develop drivers for this subsystem as well as a TODO for developers interested8in joining the effort.9 10Design principles11=================12 13The Distributed Switch Architecture subsystem was primarily designed to14support Marvell Ethernet switches (MV88E6xxx, a.k.a. Link Street product15line) using Linux, but has since evolved to support other vendors as well.16 17The original philosophy behind this design was to be able to use unmodified18Linux tools such as bridge, iproute2, ifconfig to work transparently whether19they configured/queried a switch port network device or a regular network20device.21 22An Ethernet switch typically comprises multiple front-panel ports and one23or more CPU or management ports. The DSA subsystem currently relies on the24presence of a management port connected to an Ethernet controller capable of25receiving Ethernet frames from the switch. This is a very common setup for all26kinds of Ethernet switches found in Small Home and Office products: routers,27gateways, or even top-of-rack switches. This host Ethernet controller will28be later referred to as "conduit" and "cpu" in DSA terminology and code.29 30The D in DSA stands for Distributed, because the subsystem has been designed31with the ability to configure and manage cascaded switches on top of each other32using upstream and downstream Ethernet links between switches. These specific33ports are referred to as "dsa" ports in DSA terminology and code. A collection34of multiple switches connected to each other is called a "switch tree".35 36For each front-panel port, DSA creates specialized network devices which are37used as controlling and data-flowing endpoints for use by the Linux networking38stack. These specialized network interfaces are referred to as "user" network39interfaces 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 each43Ethernet frame it receives to/from specific ports to help the management44interface figure out:45 46- what port is this frame coming from47- what was the reason why this frame got forwarded48- how to send CPU originated traffic to specific ports49 50The subsystem does support switches not capable of inserting/stripping tags, but51the features might be slightly limited in that case (traffic separation relies52on Port-based VLAN IDs).53 54Note that DSA does not currently create network interfaces for the "cpu" and55"dsa" ports because:56 57- the "cpu" port is the Ethernet switch facing side of the management58 controller, and as such, would create a duplication of feature, since you59 would get two interfaces for the same conduit: conduit netdev, and "cpu" netdev60 61- the "dsa" port(s) are just conduits between two or more switches, and as such62 cannot really be used as proper network interfaces either, only the63 downstream, or the top-most upstream interface makes sense with that model64 65NB: for the past 15 years, the DSA subsystem had been making use of the terms66"master" (rather than "conduit") and "slave" (rather than "user"). These terms67have been removed from the DSA codebase and phased out of the uAPI.68 69Switch tagging protocols70------------------------71 72DSA supports many vendor-specific tagging protocols, one software-defined73tagging protocol, and a tag-less mode as well (``DSA_TAG_PROTO_NONE``).74 75The exact format of the tag protocol is vendor specific, but in general, they76all contain something which:77 78- identifies which port the Ethernet frame came from/should be sent to79- provides a reason why this frame was forwarded to the management interface80 81All tagging protocols are in ``net/dsa/tag_*.c`` files and implement the82methods of the ``struct dsa_device_ops`` structure, which are detailed below.83 84Tagging protocols generally fall in one of three categories:85 861. The switch-specific frame header is located before the Ethernet header,87 shifting to the right (from the perspective of the DSA conduit's frame88 parser) the MAC DA, MAC SA, EtherType and the entire L2 payload.892. The switch-specific frame header is located before the EtherType, keeping90 the MAC DA and MAC SA in place from the DSA conduit's perspective, but91 shifting the 'real' EtherType and L2 payload to the right.923. The switch-specific frame header is located at the tail of the packet,93 keeping all frame headers in place and not altering the view of the packet94 that the DSA conduit's frame parser has.95 96A tagging protocol may tag all packets with switch tags of the same length, or97the tag length might vary (for example packets with PTP timestamps might98require an extended switch tag, or there might be one tag length on TX and a99different one on RX). Either way, the tagging protocol driver must populate the100``struct dsa_device_ops::needed_headroom`` and/or ``struct dsa_device_ops::needed_tailroom``101with the length in octets of the longest switch frame header/trailer. The DSA102framework will automatically adjust the MTU of the conduit interface to103accommodate for this extra size in order for DSA user ports to support the104standard MTU (L2 payload length) of 1500 octets. The ``needed_headroom`` and105``needed_tailroom`` properties are also used to request from the network stack,106on a best-effort basis, the allocation of packets with enough extra space such107that the act of pushing the switch tag on transmission of a packet does not108cause it to reallocate due to lack of memory.109 110Even though applications are not expected to parse DSA-specific frame headers,111the format on the wire of the tagging protocol represents an Application Binary112Interface exposed by the kernel towards user space, for decoders such as113``libpcap``. The tagging protocol driver must populate the ``proto`` member of114``struct dsa_device_ops`` with a value that uniquely describes the115characteristics of the interaction required between the switch hardware and the116data path driver: the offset of each bit field within the frame header and any117stateful processing required to deal with the frames (as may be required for118PTP timestamping).119 120From the perspective of the network stack, all switches within the same DSA121switch tree use the same tagging protocol. In case of a packet transiting a122fabric with more than one switch, the switch-specific frame header is inserted123by the first switch in the fabric that the packet was received on. This header124typically contains information regarding its type (whether it is a control125frame that must be trapped to the CPU, or a data frame to be forwarded).126Control frames should be decapsulated only by the software data path, whereas127data frames might also be autonomously forwarded towards other user ports of128other switches from the same fabric, and in this case, the outermost switch129ports must decapsulate the packet.130 131Note that in certain cases, it might be the case that the tagging format used132by a leaf switch (not connected directly to the CPU) is not the same as what133the network stack sees. This can be seen with Marvell switch trees, where the134CPU port can be configured to use either the DSA or the Ethertype DSA (EDSA)135format, but the DSA links are configured to use the shorter (without Ethertype)136DSA frame header, in order to reduce the autonomous packet forwarding overhead.137It still remains the case that, if the DSA switch tree is configured for the138EDSA tagging protocol, the operating system sees EDSA-tagged packets from the139leaf switches that tagged them with the shorter DSA header. This can be done140because the Marvell switch connected directly to the CPU is configured to141perform tag translation between DSA and EDSA (which is simply the operation of142adding or removing the ``ETH_P_EDSA`` EtherType and some padding octets).143 144It is possible to construct cascaded setups of DSA switches even if their145tagging protocols are not compatible with one another. In this case, there are146no DSA links in this fabric, and each switch constitutes a disjoint DSA switch147tree. The DSA links are viewed as simply a pair of a DSA conduit (the out-facing148port of the upstream DSA switch) and a CPU port (the in-facing port of the149downstream DSA switch).150 151The tagging protocol of the attached DSA switch tree can be viewed through the152``dsa/tagging`` sysfs attribute of the DSA conduit::153 154 cat /sys/class/net/eth0/dsa/tagging155 156If the hardware and driver are capable, the tagging protocol of the DSA switch157tree can be changed at runtime. This is done by writing the new tagging158protocol name to the same sysfs device attribute as above (the DSA conduit and159all attached switch ports must be down while doing this).160 161It is desirable that all tagging protocols are testable with the ``dsa_loop``162mockup driver, which can be attached to any network interface. The goal is that163any network interface should be capable of transmitting the same packet in the164same way, and the tagger should decode the same received packet in the same way165regardless of the driver used for the switch control path, and the driver used166for the DSA conduit.167 168The transmission of a packet goes through the tagger's ``xmit`` function.169The passed ``struct sk_buff *skb`` has ``skb->data`` pointing at170``skb_mac_header(skb)``, i.e. at the destination MAC address, and the passed171``struct net_device *dev`` represents the virtual DSA user network interface172whose hardware counterpart the packet must be steered to (i.e. ``swp0``).173The job of this method is to prepare the skb in a way that the switch will174understand what egress port the packet is for (and not deliver it towards other175ports). Typically this is fulfilled by pushing a frame header. Checking for176insufficient size in the skb headroom or tailroom is unnecessary provided that177the ``needed_headroom`` and ``needed_tailroom`` properties were filled out178properly, because DSA ensures there is enough space before calling this method.179 180The reception of a packet goes through the tagger's ``rcv`` function. The181passed ``struct sk_buff *skb`` has ``skb->data`` pointing at182``skb_mac_header(skb) + ETH_ALEN`` octets, i.e. to where the first octet after183the EtherType would have been, were this frame not tagged. The role of this184method is to consume the frame header, adjust ``skb->data`` to really point at185the first octet after the EtherType, and to change ``skb->dev`` to point to the186virtual DSA user network interface corresponding to the physical front-facing187switch port that the packet was received on.188 189Since tagging protocols in category 1 and 2 break software (and most often also190hardware) packet dissection on the DSA conduit, features such as RPS (Receive191Packet Steering) on the DSA conduit would be broken. The DSA framework deals192with this by hooking into the flow dissector and shifting the offset at which193the IP header is to be found in the tagged frame as seen by the DSA conduit.194This behavior is automatic based on the ``overhead`` value of the tagging195protocol. If not all packets are of equal size, the tagger can implement the196``flow_dissect`` method of the ``struct dsa_device_ops`` and override this197default behavior by specifying the correct offset incurred by each individual198RX packet. Tail taggers do not cause issues to the flow dissector.199 200Checksum offload should work with category 1 and 2 taggers when the DSA conduit201driver declares NETIF_F_HW_CSUM in vlan_features and looks at csum_start and202csum_offset. For those cases, DSA will shift the checksum start and offset by203the tag size. If the DSA conduit driver still uses the legacy NETIF_F_IP_CSUM204or NETIF_F_IPV6_CSUM in vlan_features, the offload might only work if the205offload hardware already expects that specific tag (perhaps due to matching206vendors). DSA user ports inherit those flags from the conduit, and it is up to207the driver to correctly fall back to software checksum when the IP header is not208where the hardware expects. If that check is ineffective, the packets might go209to the network without a proper checksum (the checksum field will have the210pseudo IP header sum). For category 3, when the offload hardware does not211already expect the switch tag in use, the checksum must be calculated before any212tag is inserted (i.e. inside the tagger). Otherwise, the DSA conduit would213include the tail tag in the (software or hardware) checksum calculation. Then,214when the tag gets stripped by the switch during transmission, it will leave an215incorrect IP checksum in place.216 217Due to various reasons (most common being category 1 taggers being associated218with DSA-unaware conduits, mangling what the conduit perceives as MAC DA), the219tagging protocol may require the DSA conduit to operate in promiscuous mode, to220receive all frames regardless of the value of the MAC DA. This can be done by221setting the ``promisc_on_conduit`` property of the ``struct dsa_device_ops``.222Note that this assumes a DSA-unaware conduit driver, which is the norm.223 224Conduit network devices225-----------------------226 227Conduit network devices are regular, unmodified Linux network device drivers for228the CPU/management Ethernet interface. Such a driver might occasionally need to229know whether DSA is enabled (e.g.: to enable/disable specific offload features),230but the DSA subsystem has been proven to work with industry standard drivers:231``e1000e,`` ``mv643xx_eth`` etc. without having to introduce modifications to these232drivers. Such network devices are also often referred to as conduit network233devices since they act as a pipe between the host processor and the hardware234Ethernet switch.235 236Networking stack hooks237----------------------238 239When a conduit netdev is used with DSA, a small hook is placed in the240networking stack is in order to have the DSA subsystem process the Ethernet241switch specific tagging protocol. DSA accomplishes this by registering a242specific (and fake) Ethernet type (later becoming ``skb->protocol``) with the243networking stack, this is also known as a ``ptype`` or ``packet_type``. A typical244Ethernet Frame receive sequence looks like this:245 246Conduit network device (e.g.: e1000e):247 2481. Receive interrupt fires:249 250 - receive function is invoked251 - basic packet processing is done: getting length, status etc.252 - packet is prepared to be processed by the Ethernet layer by calling253 ``eth_type_trans``254 2552. net/ethernet/eth.c::256 257 eth_type_trans(skb, dev)258 if (dev->dsa_ptr != NULL)259 -> skb->protocol = ETH_P_XDSA260 2613. drivers/net/ethernet/\*::262 263 netif_receive_skb(skb)264 -> iterate over registered packet_type265 -> invoke handler for ETH_P_XDSA, calls dsa_switch_rcv()266 2674. net/dsa/dsa.c::268 269 -> dsa_switch_rcv()270 -> invoke switch tag specific protocol handler in 'net/dsa/tag_*.c'271 2725. net/dsa/tag_*.c:273 274 - inspect and strip switch tag protocol to determine originating port275 - locate per-port network device276 - invoke ``eth_type_trans()`` with the DSA user network device277 - invoked ``netif_receive_skb()``278 279Past this point, the DSA user network devices get delivered regular Ethernet280frames that can be processed by the networking stack.281 282User network devices283--------------------284 285User network devices created by DSA are stacked on top of their conduit network286device, each of these network interfaces will be responsible for being a287controlling and data-flowing end-point for each front-panel port of the switch.288These interfaces are specialized in order to:289 290- insert/remove the switch tag protocol (if it exists) when sending traffic291 to/from specific switch ports292- query the switch for ethtool operations: statistics, link state,293 Wake-on-LAN, register dumps...294- manage external/internal PHY: link, auto-negotiation, etc.295 296These user network devices have custom net_device_ops and ethtool_ops function297pointers which allow DSA to introduce a level of layering between the networking298stack/ethtool and the switch driver implementation.299 300Upon frame transmission from these user network devices, DSA will look up which301switch tagging protocol is currently registered with these network devices and302invoke a specific transmit routine which takes care of adding the relevant303switch tag in the Ethernet frames.304 305These frames are then queued for transmission using the conduit network device306``ndo_start_xmit()`` function. Since they contain the appropriate switch tag, the307Ethernet switch will be able to process these incoming frames from the308management interface and deliver them to the physical switch port.309 310When using multiple CPU ports, it is possible to stack a LAG (bonding/team)311device between the DSA user devices and the physical DSA conduits. The LAG312device is thus also a DSA conduit, but the LAG slave devices continue to be DSA313conduits as well (just with no user port assigned to them; this is needed for314recovery in case the LAG DSA conduit disappears). Thus, the data path of the LAG315DSA conduit is used asymmetrically. On RX, the ``ETH_P_XDSA`` handler, which316calls ``dsa_switch_rcv()``, is invoked early (on the physical DSA conduit;317LAG slave). Therefore, the RX data path of the LAG DSA conduit is not used.318On the other hand, TX takes place linearly: ``dsa_user_xmit`` calls319``dsa_enqueue_skb``, which calls ``dev_queue_xmit`` towards the LAG DSA conduit.320The latter calls ``dev_queue_xmit`` towards one physical DSA conduit or the321other, and in both cases, the packet exits the system through a hardware path322towards the switch.323 324Graphical representation325------------------------326 327Summarized, this is basically how DSA looks like from a network device328perspective::329 330 Unaware application331 opens and binds socket332 | ^333 | |334 +-----------v--|--------------------+335 |+------+ +------+ +------+ +------+|336 || swp0 | | swp1 | | swp2 | | swp3 ||337 |+------+-+------+-+------+-+------+|338 | DSA switch driver |339 +-----------------------------------+340 | ^341 Tag added by | | Tag consumed by342 switch driver | | switch driver343 v |344 +-----------------------------------+345 | Unmodified host interface driver | Software346 --------+-----------------------------------+------------347 | Host interface (eth0) | Hardware348 +-----------------------------------+349 | ^350 Tag consumed by | | Tag added by351 switch hardware | | switch hardware352 v |353 +-----------------------------------+354 | Switch |355 |+------+ +------+ +------+ +------+|356 || swp0 | | swp1 | | swp2 | | swp3 ||357 ++------+-+------+-+------+-+------++358 359User MDIO bus360-------------361 362In order to be able to read to/from a switch PHY built into it, DSA creates an363user MDIO bus which allows a specific switch driver to divert and intercept364MDIO reads/writes towards specific PHY addresses. In most MDIO-connected365switches, these functions would utilize direct or indirect PHY addressing mode366to return standard MII registers from the switch builtin PHYs, allowing the PHY367library and/or to return link status, link partner pages, auto-negotiation368results, etc.369 370For Ethernet switches which have both external and internal MDIO buses, the371user MII bus can be utilized to mux/demux MDIO reads and writes towards either372internal or external MDIO devices this switch might be connected to: internal373PHYs, external PHYs, or even external switches.374 375Data structures376---------------377 378DSA data structures are defined in ``include/net/dsa.h`` as well as379``net/dsa/dsa_priv.h``:380 381- ``dsa_chip_data``: platform data configuration for a given switch device,382 this structure describes a switch device's parent device, its address, as383 well as various properties of its ports: names/labels, and finally a routing384 table indication (when cascading switches)385 386- ``dsa_platform_data``: platform device configuration data which can reference387 a collection of dsa_chip_data structures if multiple switches are cascaded,388 the conduit network device this switch tree is attached to needs to be389 referenced390 391- ``dsa_switch_tree``: structure assigned to the conduit network device under392 ``dsa_ptr``, this structure references a dsa_platform_data structure as well as393 the tagging protocol supported by the switch tree, and which receive/transmit394 function hooks should be invoked, information about the directly attached395 switch is also provided: CPU port. Finally, a collection of dsa_switch are396 referenced to address individual switches in the tree.397 398- ``dsa_switch``: structure describing a switch device in the tree, referencing399 a ``dsa_switch_tree`` as a backpointer, user network devices, conduit network400 device, and a reference to the backing``dsa_switch_ops``401 402- ``dsa_switch_ops``: structure referencing function pointers, see below for a403 full description.404 405Design limitations406==================407 408Lack of CPU/DSA network devices409-------------------------------410 411DSA does not currently create user network devices for the CPU or DSA ports, as412described before. This might be an issue in the following cases:413 414- inability to fetch switch CPU port statistics counters using ethtool, which415 can make it harder to debug MDIO switch connected using xMII interfaces416 417- inability to configure the CPU port link parameters based on the Ethernet418 controller capabilities attached to it: http://patchwork.ozlabs.org/patch/509806/419 420- inability to configure specific VLAN IDs / trunking VLANs between switches421 when using a cascaded setup422 423Common pitfalls using DSA setups424--------------------------------425 426Once a conduit network device is configured to use DSA (dev->dsa_ptr becomes427non-NULL), and the switch behind it expects a tagging protocol, this network428interface can only exclusively be used as a conduit interface. Sending packets429directly through this interface (e.g.: opening a socket using this interface)430will not make us go through the switch tagging protocol transmit function, so431the Ethernet switch on the other end, expecting a tag will typically drop this432frame.433 434Interactions with other subsystems435==================================436 437DSA currently leverages the following subsystems:438 439- MDIO/PHY library: ``drivers/net/phy/phy.c``, ``mdio_bus.c``440- Switchdev:``net/switchdev/*``441- Device Tree for various of_* functions442- Devlink: ``net/core/devlink.c``443 444MDIO/PHY library445----------------446 447User network devices exposed by DSA may or may not be interfacing with PHY448devices (``struct phy_device`` as defined in ``include/linux/phy.h)``, but the DSA449subsystem deals with all possible combinations:450 451- internal PHY devices, built into the Ethernet switch hardware452- external PHY devices, connected via an internal or external MDIO bus453- internal PHY devices, connected via an internal MDIO bus454- special, non-autonegotiated or non MDIO-managed PHY devices: SFPs, MoCA; a.k.a455 fixed PHYs456 457The PHY configuration is done by the ``dsa_user_phy_setup()`` function and the458logic basically looks like this:459 460- if Device Tree is used, the PHY device is looked up using the standard461 "phy-handle" property, if found, this PHY device is created and registered462 using ``of_phy_connect()``463 464- if Device Tree is used and the PHY device is "fixed", that is, conforms to465 the definition of a non-MDIO managed PHY as defined in466 ``Documentation/devicetree/bindings/net/fixed-link.txt``, the PHY is registered467 and connected transparently using the special fixed MDIO bus driver468 469- finally, if the PHY is built into the switch, as is very common with470 standalone switch packages, the PHY is probed using the user MII bus created471 by DSA472 473 474SWITCHDEV475---------476 477DSA directly utilizes SWITCHDEV when interfacing with the bridge layer, and478more specifically with its VLAN filtering portion when configuring VLANs on top479of per-port user network devices. As of today, the only SWITCHDEV objects480supported by DSA are the FDB and VLAN objects.481 482Devlink483-------484 485DSA registers one devlink device per physical switch in the fabric.486For each devlink device, every physical port (i.e. user ports, CPU ports, DSA487links or unused ports) is exposed as a devlink port.488 489DSA drivers can make use of the following devlink features:490 491- Regions: debugging feature which allows user space to dump driver-defined492 areas of hardware information in a low-level, binary format. Both global493 regions as well as per-port regions are supported. It is possible to export494 devlink regions even for pieces of data that are already exposed in some way495 to the standard iproute2 user space programs (ip-link, bridge), like address496 tables and VLAN tables. For example, this might be useful if the tables497 contain additional hardware-specific details which are not visible through498 the iproute2 abstraction, or it might be useful to inspect these tables on499 the non-user ports too, which are invisible to iproute2 because no network500 interface is registered for them.501- Params: a feature which enables user to configure certain low-level tunable502 knobs pertaining to the device. Drivers may implement applicable generic503 devlink params, or may add new device-specific devlink params.504- Resources: a monitoring feature which enables users to see the degree of505 utilization of certain hardware tables in the device, such as FDB, VLAN, etc.506- Shared buffers: a QoS feature for adjusting and partitioning memory and frame507 reservations per port and per traffic class, in the ingress and egress508 directions, such that low-priority bulk traffic does not impede the509 processing of high-priority critical traffic.510 511For more details, consult ``Documentation/networking/devlink/``.512 513Device Tree514-----------515 516DSA features a standardized binding which is documented in517``Documentation/devicetree/bindings/net/dsa/dsa.txt``. PHY/MDIO library helper518functions such as ``of_get_phy_mode()``, ``of_phy_connect()`` are also used to query519per-port PHY specific details: interface connection, MDIO bus location, etc.520 521Driver development522==================523 524DSA switch drivers need to implement a ``dsa_switch_ops`` structure which will525contain the various members described below.526 527Probing, registration and device lifetime528-----------------------------------------529 530DSA switches are regular ``device`` structures on buses (be they platform, SPI,531I2C, MDIO or otherwise). The DSA framework is not involved in their probing532with the device core.533 534Switch registration from the perspective of a driver means passing a valid535``struct dsa_switch`` pointer to ``dsa_register_switch()``, usually from the536switch driver's probing function. The following members must be valid in the537provided structure:538 539- ``ds->dev``: will be used to parse the switch's OF node or platform data.540 541- ``ds->num_ports``: will be used to create the port list for this switch, and542 to validate the port indices provided in the OF node.543 544- ``ds->ops``: a pointer to the ``dsa_switch_ops`` structure holding the DSA545 method implementations.546 547- ``ds->priv``: backpointer to a driver-private data structure which can be548 retrieved in all further DSA method callbacks.549 550In addition, the following flags in the ``dsa_switch`` structure may optionally551be configured to obtain driver-specific behavior from the DSA core. Their552behavior when set is documented through comments in ``include/net/dsa.h``.553 554- ``ds->vlan_filtering_is_global``555 556- ``ds->needs_standalone_vlan_filtering``557 558- ``ds->configure_vlan_while_not_filtering``559 560- ``ds->untag_bridge_pvid``561 562- ``ds->assisted_learning_on_cpu_port``563 564- ``ds->mtu_enforcement_ingress``565 566- ``ds->fdb_isolation``567 568Internally, DSA keeps an array of switch trees (group of switches) global to569the kernel, and attaches a ``dsa_switch`` structure to a tree on registration.570The tree ID to which the switch is attached is determined by the first u32571number of the ``dsa,member`` property of the switch's OF node (0 if missing).572The switch ID within the tree is determined by the second u32 number of the573same OF property (0 if missing). Registering multiple switches with the same574switch ID and tree ID is illegal and will cause an error. Using platform data,575a single switch and a single switch tree is permitted.576 577In case of a tree with multiple switches, probing takes place asymmetrically.578The first N-1 callers of ``dsa_register_switch()`` only add their ports to the579port list of the tree (``dst->ports``), each port having a backpointer to its580associated switch (``dp->ds``). Then, these switches exit their581``dsa_register_switch()`` call early, because ``dsa_tree_setup_routing_table()``582has determined that the tree is not yet complete (not all ports referenced by583DSA links are present in the tree's port list). The tree becomes complete when584the last switch calls ``dsa_register_switch()``, and this triggers the effective585continuation of initialization (including the call to ``ds->ops->setup()``) for586all switches within that tree, all as part of the calling context of the last587switch's probe function.588 589The opposite of registration takes place when calling ``dsa_unregister_switch()``,590which removes a switch's ports from the port list of the tree. The entire tree591is torn down when the first switch unregisters.592 593It is mandatory for DSA switch drivers to implement the ``shutdown()`` callback594of their respective bus, and call ``dsa_switch_shutdown()`` from it (a minimal595version of the full teardown performed by ``dsa_unregister_switch()``).596The reason is that DSA keeps a reference on the conduit net device, and if the597driver for the conduit device decides to unbind on shutdown, DSA's reference598will block that operation from finalizing.599 600Either ``dsa_switch_shutdown()`` or ``dsa_unregister_switch()`` must be called,601but not both, and the device driver model permits the bus' ``remove()`` method602to be called even if ``shutdown()`` was already called. Therefore, drivers are603expected to implement a mutual exclusion method between ``remove()`` and604``shutdown()`` by setting their drvdata to NULL after any of these has run, and605checking whether the drvdata is NULL before proceeding to take any action.606 607After ``dsa_switch_shutdown()`` or ``dsa_unregister_switch()`` was called, no608further callbacks via the provided ``dsa_switch_ops`` may take place, and the609driver may free the data structures associated with the ``dsa_switch``.610 611Switch configuration612--------------------613 614- ``get_tag_protocol``: this is to indicate what kind of tagging protocol is615 supported, should be a valid value from the ``dsa_tag_protocol`` enum.616 The returned information does not have to be static; the driver is passed the617 CPU port number, as well as the tagging protocol of a possibly stacked618 upstream switch, in case there are hardware limitations in terms of supported619 tag formats.620 621- ``change_tag_protocol``: when the default tagging protocol has compatibility622 problems with the conduit or other issues, the driver may support changing it623 at runtime, either through a device tree property or through sysfs. In that624 case, further calls to ``get_tag_protocol`` should report the protocol in625 current use.626 627- ``setup``: setup function for the switch, this function is responsible for setting628 up the ``dsa_switch_ops`` private structure with all it needs: register maps,629 interrupts, mutexes, locks, etc. This function is also expected to properly630 configure the switch to separate all network interfaces from each other, that631 is, they should be isolated by the switch hardware itself, typically by creating632 a Port-based VLAN ID for each port and allowing only the CPU port and the633 specific port to be in the forwarding vector. Ports that are unused by the634 platform should be disabled. Past this function, the switch is expected to be635 fully configured and ready to serve any kind of request. It is recommended636 to issue a software reset of the switch during this setup function in order to637 avoid relying on what a previous software agent such as a bootloader/firmware638 may have previously configured. The method responsible for undoing any639 applicable allocations or operations done here is ``teardown``.640 641- ``port_setup`` and ``port_teardown``: methods for initialization and642 destruction of per-port data structures. It is mandatory for some operations643 such as registering and unregistering devlink port regions to be done from644 these methods, otherwise they are optional. A port will be torn down only if645 it has been previously set up. It is possible for a port to be set up during646 probing only to be torn down immediately afterwards, for example in case its647 PHY cannot be found. In this case, probing of the DSA switch continues648 without that particular port.649 650- ``port_change_conduit``: method through which the affinity (association used651 for traffic termination purposes) between a user port and a CPU port can be652 changed. By default all user ports from a tree are assigned to the first653 available CPU port that makes sense for them (most of the times this means654 the user ports of a tree are all assigned to the same CPU port, except for H655 topologies as described in commit 2c0b03258b8b). The ``port`` argument656 represents the index of the user port, and the ``conduit`` argument represents657 the new DSA conduit ``net_device``. The CPU port associated with the new658 conduit can be retrieved by looking at ``struct dsa_port *cpu_dp =659 conduit->dsa_ptr``. Additionally, the conduit can also be a LAG device where660 all the slave devices are physical DSA conduits. LAG DSA also have a661 valid ``conduit->dsa_ptr`` pointer, however this is not unique, but rather a662 duplicate of the first physical DSA conduit's (LAG slave) ``dsa_ptr``. In case663 of a LAG DSA conduit, a further call to ``port_lag_join`` will be emitted664 separately for the physical CPU ports associated with the physical DSA665 conduits, requesting them to create a hardware LAG associated with the LAG666 interface.667 668PHY devices and link management669-------------------------------670 671- ``get_phy_flags``: Some switches are interfaced to various kinds of Ethernet PHYs,672 if the PHY library PHY driver needs to know about information it cannot obtain673 on its own (e.g.: coming from switch memory mapped registers), this function674 should return a 32-bit bitmask of "flags" that is private between the switch675 driver and the Ethernet PHY driver in ``drivers/net/phy/\*``.676 677- ``phy_read``: Function invoked by the DSA user MDIO bus when attempting to read678 the switch port MDIO registers. If unavailable, return 0xffff for each read.679 For builtin switch Ethernet PHYs, this function should allow reading the link680 status, auto-negotiation results, link partner pages, etc.681 682- ``phy_write``: Function invoked by the DSA user MDIO bus when attempting to write683 to the switch port MDIO registers. If unavailable return a negative error684 code.685 686- ``adjust_link``: Function invoked by the PHY library when a user network device687 is attached to a PHY device. This function is responsible for appropriately688 configuring the switch port link parameters: speed, duplex, pause based on689 what the ``phy_device`` is providing.690 691- ``fixed_link_update``: Function invoked by the PHY library, and specifically by692 the fixed PHY driver asking the switch driver for link parameters that could693 not be auto-negotiated, or obtained by reading the PHY registers through MDIO.694 This is particularly useful for specific kinds of hardware such as QSGMII,695 MoCA or other kinds of non-MDIO managed PHYs where out of band link696 information is obtained697 698Ethtool operations699------------------700 701- ``get_strings``: ethtool function used to query the driver's strings, will702 typically return statistics strings, private flags strings, etc.703 704- ``get_ethtool_stats``: ethtool function used to query per-port statistics and705 return their values. DSA overlays user network devices general statistics:706 RX/TX counters from the network device, with switch driver specific statistics707 per port708 709- ``get_sset_count``: ethtool function used to query the number of statistics items710 711- ``get_wol``: ethtool function used to obtain Wake-on-LAN settings per-port, this712 function may for certain implementations also query the conduit network device713 Wake-on-LAN settings if this interface needs to participate in Wake-on-LAN714 715- ``set_wol``: ethtool function used to configure Wake-on-LAN settings per-port,716 direct counterpart to set_wol with similar restrictions717 718- ``set_eee``: ethtool function which is used to configure a switch port EEE (Green719 Ethernet) settings, can optionally invoke the PHY library to enable EEE at the720 PHY level if relevant. This function should enable EEE at the switch port MAC721 controller and data-processing logic722 723- ``get_eee``: ethtool function which is used to query a switch port EEE settings,724 this function should return the EEE state of the switch port MAC controller725 and data-processing logic as well as query the PHY for its currently configured726 EEE settings727 728- ``get_eeprom_len``: ethtool function returning for a given switch the EEPROM729 length/size in bytes730 731- ``get_eeprom``: ethtool function returning for a given switch the EEPROM contents732 733- ``set_eeprom``: ethtool function writing specified data to a given switch EEPROM734 735- ``get_regs_len``: ethtool function returning the register length for a given736 switch737 738- ``get_regs``: ethtool function returning the Ethernet switch internal register739 contents. This function might require user-land code in ethtool to740 pretty-print register values and registers741 742Power management743----------------744 745- ``suspend``: function invoked by the DSA platform device when the system goes to746 suspend, should quiesce all Ethernet switch activities, but keep ports747 participating in Wake-on-LAN active as well as additional wake-up logic if748 supported749 750- ``resume``: function invoked by the DSA platform device when the system resumes,751 should resume all Ethernet switch activities and re-configure the switch to be752 in a fully active state753 754- ``port_enable``: function invoked by the DSA user network device ndo_open755 function when a port is administratively brought up, this function should756 fully enable a given switch port. DSA takes care of marking the port with757 ``BR_STATE_BLOCKING`` if the port is a bridge member, or ``BR_STATE_FORWARDING`` if it758 was not, and propagating these changes down to the hardware759 760- ``port_disable``: function invoked by the DSA user network device ndo_close761 function when a port is administratively brought down, this function should762 fully disable a given switch port. DSA takes care of marking the port with763 ``BR_STATE_DISABLED`` and propagating changes to the hardware if this port is764 disabled while being a bridge member765 766Address databases767-----------------768 769Switching hardware is expected to have a table for FDB entries, however not all770of them are active at the same time. An address database is the subset (partition)771of FDB entries that is active (can be matched by address learning on RX, or FDB772lookup on TX) depending on the state of the port. An address database may773occasionally be called "FID" (Filtering ID) in this document, although the774underlying implementation may choose whatever is available to the hardware.775 776For example, all ports that belong to a VLAN-unaware bridge (which is777*currently* VLAN-unaware) are expected to learn source addresses in the778database associated by the driver with that bridge (and not with other779VLAN-unaware bridges). During forwarding and FDB lookup, a packet received on a780VLAN-unaware bridge port should be able to find a VLAN-unaware FDB entry having781the same MAC DA as the packet, which is present on another port member of the782same bridge. At the same time, the FDB lookup process must be able to not find783an FDB entry having the same MAC DA as the packet, if that entry points towards784a port which is a member of a different VLAN-unaware bridge (and is therefore785associated with a different address database).786 787Similarly, each VLAN of each offloaded VLAN-aware bridge should have an788associated address database, which is shared by all ports which are members of789that VLAN, but not shared by ports belonging to different bridges that are790members of the same VID.791 792In this context, a VLAN-unaware database means that all packets are expected to793match on it irrespective of VLAN ID (only MAC address lookup), whereas a794VLAN-aware database means that packets are supposed to match based on the VLAN795ID from the classified 802.1Q header (or the pvid if untagged).796 797At the bridge layer, VLAN-unaware FDB entries have the special VID value of 0,798whereas VLAN-aware FDB entries have non-zero VID values. Note that a799VLAN-unaware bridge may have VLAN-aware (non-zero VID) FDB entries, and a800VLAN-aware bridge may have VLAN-unaware FDB entries. As in hardware, the801software bridge keeps separate address databases, and offloads to hardware the802FDB entries belonging to these databases, through switchdev, asynchronously803relative to the moment when the databases become active or inactive.804 805When a user port operates in standalone mode, its driver should configure it to806use a separate database called a port private database. This is different from807the databases described above, and should impede operation as standalone port808(packet in, packet out to the CPU port) as little as possible. For example,809on ingress, it should not attempt to learn the MAC SA of ingress traffic, since810learning is a bridging layer service and this is a standalone port, therefore811it would consume useless space. With no address learning, the port private812database should be empty in a naive implementation, and in this case, all813received packets should be trivially flooded to the CPU port.814 815DSA (cascade) and CPU ports are also called "shared" ports because they service816multiple address databases, and the database that a packet should be associated817to is usually embedded in the DSA tag. This means that the CPU port may818simultaneously transport packets coming from a standalone port (which were819classified by hardware in one address database), and from a bridge port (which820were classified to a different address database).821 822Switch drivers which satisfy certain criteria are able to optimize the naive823configuration by removing the CPU port from the flooding domain of the switch,824and just program the hardware with FDB entries pointing towards the CPU port825for which it is known that software is interested in those MAC addresses.826Packets which do not match a known FDB entry will not be delivered to the CPU,827which will save CPU cycles required for creating an skb just to drop it.828 829DSA is able to perform host address filtering for the following kinds of830addresses:831 832- Primary unicast MAC addresses of ports (``dev->dev_addr``). These are833 associated with the port private database of the respective user port,834 and the driver is notified to install them through ``port_fdb_add`` towards835 the CPU port.836 837- Secondary unicast and multicast MAC addresses of ports (addresses added838 through ``dev_uc_add()`` and ``dev_mc_add()``). These are also associated839 with the port private database of the respective user port.840 841- Local/permanent bridge FDB entries (``BR_FDB_LOCAL``). These are the MAC842 addresses of the bridge ports, for which packets must be terminated locally843 and not forwarded. They are associated with the address database for that844 bridge.845 846- Static bridge FDB entries installed towards foreign (non-DSA) interfaces847 present in the same bridge as some DSA switch ports. These are also848 associated with the address database for that bridge.849 850- Dynamically learned FDB entries on foreign interfaces present in the same851 bridge as some DSA switch ports, only if ``ds->assisted_learning_on_cpu_port``852 is set to true by the driver. These are associated with the address database853 for that bridge.854 855For various operations detailed below, DSA provides a ``dsa_db`` structure856which can be of the following types:857 858- ``DSA_DB_PORT``: the FDB (or MDB) entry to be installed or deleted belongs to859 the port private database of user port ``db->dp``.860- ``DSA_DB_BRIDGE``: the entry belongs to one of the address databases of bridge861 ``db->bridge``. Separation between the VLAN-unaware database and the per-VID862 databases of this bridge is expected to be done by the driver.863- ``DSA_DB_LAG``: the entry belongs to the address database of LAG ``db->lag``.864 Note: ``DSA_DB_LAG`` is currently unused and may be removed in the future.865 866The drivers which act upon the ``dsa_db`` argument in ``port_fdb_add``,867``port_mdb_add`` etc should declare ``ds->fdb_isolation`` as true.868 869DSA associates each offloaded bridge and each offloaded LAG with a one-based ID870(``struct dsa_bridge :: num``, ``struct dsa_lag :: id``) for the purposes of871refcounting addresses on shared ports. Drivers may piggyback on DSA's numbering872scheme (the ID is readable through ``db->bridge.num`` and ``db->lag.id`` or may873implement their own.874 875Only the drivers which declare support for FDB isolation are notified of FDB876entries on the CPU port belonging to ``DSA_DB_PORT`` databases.877For compatibility/legacy reasons, ``DSA_DB_BRIDGE`` addresses are notified to878drivers even if they do not support FDB isolation. However, ``db->bridge.num``879and ``db->lag.id`` are always set to 0 in that case (to denote the lack of880isolation, for refcounting purposes).881 882Note that it is not mandatory for a switch driver to implement physically883separate address databases for each standalone user port. Since FDB entries in884the port private databases will always point to the CPU port, there is no risk885for incorrect forwarding decisions. In this case, all standalone ports may886share the same database, but the reference counting of host-filtered addresses887(not deleting the FDB entry for a port's MAC address if it's still in use by888another port) becomes the responsibility of the driver, because DSA is unaware889that the port databases are in fact shared. This can be achieved by calling890``dsa_fdb_present_in_other_db()`` and ``dsa_mdb_present_in_other_db()``.891The down side is that the RX filtering lists of each user port are in fact892shared, which means that user port A may accept a packet with a MAC DA it893shouldn't have, only because that MAC address was in the RX filtering list of894user port B. These packets will still be dropped in software, however.895 896Bridge layer897------------898 899Offloading the bridge forwarding plane is optional and handled by the methods900below. They may be absent, return -EOPNOTSUPP, or ``ds->max_num_bridges`` may901be non-zero and exceeded, and in this case, joining a bridge port is still902possible, but the packet forwarding will take place in software, and the ports903under a software bridge must remain configured in the same way as for904standalone operation, i.e. have all bridging service functions (address905learning etc) disabled, and send all received packets to the CPU port only.906 907Concretely, a port starts offloading the forwarding plane of a bridge once it908returns success to the ``port_bridge_join`` method, and stops doing so after909``port_bridge_leave`` has been called. Offloading the bridge means autonomously910learning FDB entries in accordance with the software bridge port's state, and911autonomously forwarding (or flooding) received packets without CPU intervention.912This is optional even when offloading a bridge port. Tagging protocol drivers913are expected to call ``dsa_default_offload_fwd_mark(skb)`` for packets which914have already been autonomously forwarded in the forwarding domain of the915ingress switch port. DSA, through ``dsa_port_devlink_setup()``, considers all916switch ports part of the same tree ID to be part of the same bridge forwarding917domain (capable of autonomous forwarding to each other).918 919Offloading the TX forwarding process of a bridge is a distinct concept from920simply offloading its forwarding plane, and refers to the ability of certain921driver and tag protocol combinations to transmit a single skb coming from the922bridge device's transmit function to potentially multiple egress ports (and923thereby avoid its cloning in software).924 925Packets for which the bridge requests this behavior are called data plane926packets and have ``skb->offload_fwd_mark`` set to true in the tag protocol927driver's ``xmit`` function. Data plane packets are subject to FDB lookup,928hardware learning on the CPU port, and do not override the port STP state.929Additionally, replication of data plane packets (multicast, flooding) is930handled in hardware and the bridge driver will transmit a single skb for each931packet that may or may not need replication.932 933When the TX forwarding offload is enabled, the tag protocol driver is934responsible to inject packets into the data plane of the hardware towards the935correct bridging domain (FID) that the port is a part of. The port may be936VLAN-unaware, and in this case the FID must be equal to the FID used by the937driver for its VLAN-unaware address database associated with that bridge.938Alternatively, the bridge may be VLAN-aware, and in that case, it is guaranteed939that the packet is also VLAN-tagged with the VLAN ID that the bridge processed940this packet in. It is the responsibility of the hardware to untag the VID on941the egress-untagged ports, or keep the tag on the egress-tagged ones.942 943- ``port_bridge_join``: bridge layer function invoked when a given switch port is944 added to a bridge, this function should do what's necessary at the switch945 level to permit the joining port to be added to the relevant logical946 domain for it to ingress/egress traffic with other members of the bridge.947 By setting the ``tx_fwd_offload`` argument to true, the TX forwarding process948 of this bridge is also offloaded.949 950- ``port_bridge_leave``: bridge layer function invoked when a given switch port is951 removed from a bridge, this function should do what's necessary at the952 switch level to deny the leaving port from ingress/egress traffic from the953 remaining bridge members.954 955- ``port_stp_state_set``: bridge layer function invoked when a given switch port STP956 state is computed by the bridge layer and should be propagated to switch957 hardware to forward/block/learn traffic.958 959- ``port_bridge_flags``: bridge layer function invoked when a port must960 configure its settings for e.g. flooding of unknown traffic or source address961 learning. The switch driver is responsible for initial setup of the962 standalone ports with address learning disabled and egress flooding of all963 types of traffic, then the DSA core notifies of any change to the bridge port964 flags when the port joins and leaves a bridge. DSA does not currently manage965 the bridge port flags for the CPU port. The assumption is that address966 learning should be statically enabled (if supported by the hardware) on the967 CPU port, and flooding towards the CPU port should also be enabled, due to a968 lack of an explicit address filtering mechanism in the DSA core.969 970- ``port_fast_age``: bridge layer function invoked when flushing the971 dynamically learned FDB entries on the port is necessary. This is called when972 transitioning from an STP state where learning should take place to an STP973 state where it shouldn't, or when leaving a bridge, or when address learning974 is turned off via ``port_bridge_flags``.975 976Bridge VLAN filtering977---------------------978 979- ``port_vlan_filtering``: bridge layer function invoked when the bridge gets980 configured for turning on or off VLAN filtering. If nothing specific needs to981 be done at the hardware level, this callback does not need to be implemented.982 When VLAN filtering is turned on, the hardware must be programmed with983 rejecting 802.1Q frames which have VLAN IDs outside of the programmed allowed984 VLAN ID map/rules. If there is no PVID programmed into the switch port,985 untagged frames must be rejected as well. When turned off the switch must986 accept any 802.1Q frames irrespective of their VLAN ID, and untagged frames are987 allowed.988 989- ``port_vlan_add``: bridge layer function invoked when a VLAN is configured990 (tagged or untagged) for the given switch port. The CPU port becomes a member991 of a VLAN only if a foreign bridge port is also a member of it (and992 forwarding needs to take place in software), or the VLAN is installed to the993 VLAN group of the bridge device itself, for termination purposes994 (``bridge vlan add dev br0 vid 100 self``). VLANs on shared ports are995 reference counted and removed when there is no user left. Drivers do not need996 to manually install a VLAN on the CPU port.997 998- ``port_vlan_del``: bridge layer function invoked when a VLAN is removed from the999 given switch port1000 1001- ``port_fdb_add``: bridge layer function invoked when the bridge wants to install a1002 Forwarding Database entry, the switch hardware should be programmed with the1003 specified address in the specified VLAN Id in the forwarding database1004 associated with this VLAN ID.1005 1006- ``port_fdb_del``: bridge layer function invoked when the bridge wants to remove a1007 Forwarding Database entry, the switch hardware should be programmed to delete1008 the specified MAC address from the specified VLAN ID if it was mapped into1009 this port forwarding database1010 1011- ``port_fdb_dump``: bridge bypass function invoked by ``ndo_fdb_dump`` on the1012 physical DSA port interfaces. Since DSA does not attempt to keep in sync its1013 hardware FDB entries with the software bridge, this method is implemented as1014 a means to view the entries visible on user ports in the hardware database.1015 The entries reported by this function have the ``self`` flag in the output of1016 the ``bridge fdb show`` command.1017 1018- ``port_mdb_add``: bridge layer function invoked when the bridge wants to install1019 a multicast database entry. The switch hardware should be programmed with the1020 specified address in the specified VLAN ID in the forwarding database1021 associated with this VLAN ID.1022 1023- ``port_mdb_del``: bridge layer function invoked when the bridge wants to remove a1024 multicast database entry, the switch hardware should be programmed to delete1025 the specified MAC address from the specified VLAN ID if it was mapped into1026 this port forwarding database.1027 1028Link aggregation1029----------------1030 1031Link aggregation is implemented in the Linux networking stack by the bonding1032and team drivers, which are modeled as virtual, stackable network interfaces.1033DSA is capable of offloading a link aggregation group (LAG) to hardware that1034supports the feature, and supports bridging between physical ports and LAGs,1035as well as between LAGs. A bonding/team interface which holds multiple physical1036ports constitutes a logical port, although DSA has no explicit concept of a1037logical port at the moment. Due to this, events where a LAG joins/leaves a1038bridge are treated as if all individual physical ports that are members of that1039LAG join/leave the bridge. Switchdev port attributes (VLAN filtering, STP1040state, etc) and objects (VLANs, MDB entries) offloaded to a LAG as bridge port1041are treated similarly: DSA offloads the same switchdev object / port attribute1042on all members of the LAG. Static bridge FDB entries on a LAG are not yet1043supported, since the DSA driver API does not have the concept of a logical port1044ID.1045 1046- ``port_lag_join``: function invoked when a given switch port is added to a1047 LAG. The driver may return ``-EOPNOTSUPP``, and in this case, DSA will fall1048 back to a software implementation where all traffic from this port is sent to1049 the CPU.1050- ``port_lag_leave``: function invoked when a given switch port leaves a LAG1051 and returns to operation as a standalone port.1052- ``port_lag_change``: function invoked when the link state of any member of1053 the LAG changes, and the hashing function needs rebalancing to only make use1054 of the subset of physical LAG member ports that are up.1055 1056Drivers that benefit from having an ID associated with each offloaded LAG1057can optionally populate ``ds->num_lag_ids`` from the ``dsa_switch_ops::setup``1058method. The LAG ID associated with a bonding/team interface can then be1059retrieved by a DSA switch driver using the ``dsa_lag_id`` function.1060 1061IEC 62439-2 (MRP)1062-----------------1063 1064The Media Redundancy Protocol is a topology management protocol optimized for1065fast fault recovery time for ring networks, which has some components1066implemented as a function of the bridge driver. MRP uses management PDUs1067(Test, Topology, LinkDown/Up, Option) sent at a multicast destination MAC1068address range of 01:15:4e:00:00:0x and with an EtherType of 0x88e3.1069Depending on the node's role in the ring (MRM: Media Redundancy Manager,1070MRC: Media Redundancy Client, MRA: Media Redundancy Automanager), certain MRP1071PDUs might need to be terminated locally and others might need to be forwarded.1072An MRM might also benefit from offloading to hardware the creation and1073transmission of certain MRP PDUs (Test).1074 1075Normally an MRP instance can be created on top of any network interface,1076however in the case of a device with an offloaded data path such as DSA, it is1077necessary for the hardware, even if it is not MRP-aware, to be able to extract1078the MRP PDUs from the fabric before the driver can proceed with the software1079implementation. DSA today has no driver which is MRP-aware, therefore it only1080listens for the bare minimum switchdev objects required for the software assist1081to work properly. The operations are detailed below.1082 1083- ``port_mrp_add`` and ``port_mrp_del``: notifies driver when an MRP instance1084 with a certain ring ID, priority, primary port and secondary port is1085 created/deleted.1086- ``port_mrp_add_ring_role`` and ``port_mrp_del_ring_role``: function invoked1087 when an MRP instance changes ring roles between MRM or MRC. This affects1088 which MRP PDUs should be trapped to software and which should be autonomously1089 forwarded.1090 1091IEC 62439-3 (HSR/PRP)1092---------------------1093 1094The Parallel Redundancy Protocol (PRP) is a network redundancy protocol which1095works by duplicating and sequence numbering packets through two independent L21096networks (which are unaware of the PRP tail tags carried in the packets), and1097eliminating the duplicates at the receiver. The High-availability Seamless1098Redundancy (HSR) protocol is similar in concept, except all nodes that carry1099the redundant traffic are aware of the fact that it is HSR-tagged (because HSR1100uses a header with an EtherType of 0x892f) and are physically connected in a1101ring topology. Both HSR and PRP use supervision frames for monitoring the1102health of the network and for discovery of other nodes.1103 1104In Linux, both HSR and PRP are implemented in the hsr driver, which1105instantiates a virtual, stackable network interface with two member ports.1106The driver only implements the basic roles of DANH (Doubly Attached Node1107implementing HSR) and DANP (Doubly Attached Node implementing PRP); the roles1108of RedBox and QuadBox are not implemented (therefore, bridging a hsr network1109interface with a physical switch port does not produce the expected result).1110 1111A driver which is able of offloading certain functions of a DANP or DANH should1112declare the corresponding netdev features as indicated by the documentation at1113``Documentation/networking/netdev-features.rst``. Additionally, the following1114methods must be implemented:1115 1116- ``port_hsr_join``: function invoked when a given switch port is added to a1117 DANP/DANH. The driver may return ``-EOPNOTSUPP`` and in this case, DSA will1118 fall back to a software implementation where all traffic from this port is1119 sent to the CPU.1120- ``port_hsr_leave``: function invoked when a given switch port leaves a1121 DANP/DANH and returns to normal operation as a standalone port.1122 1123TODO1124====1125 1126Making SWITCHDEV and DSA converge towards an unified codebase1127-------------------------------------------------------------1128 1129SWITCHDEV properly takes care of abstracting the networking stack with offload1130capable hardware, but does not enforce a strict switch device driver model. On1131the other DSA enforces a fairly strict device driver model, and deals with most1132of the switch specific. At some point we should envision a merger between these1133two subsystems and get the best of both worlds.1134