1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2016 6WIND S.A.
3 Copyright 2016 Mellanox Technologies, Ltd
5 Generic flow API (rte_flow)
6 ===========================
11 This API provides a generic means to configure hardware to match specific
12 ingress or egress traffic, alter its fate and query related counters
13 according to any number of user-defined rules.
15 It is named *rte_flow* after the prefix used for all its symbols, and is
16 defined in ``rte_flow.h``.
18 - Matching can be performed on packet data (protocol headers, payload) and
19 properties (e.g. associated physical port, virtual device function ID).
21 - Possible operations include dropping traffic, diverting it to specific
22 queues, to virtual/physical device functions or ports, performing tunnel
23 offloads, adding marks and so on.
31 A flow rule is the combination of attributes with a matching pattern and a
32 list of actions. Flow rules form the basis of this API.
34 Flow rules can have several distinct actions (such as counting,
35 encapsulating, decapsulating before redirecting packets to a particular
36 queue, etc.), instead of relying on several rules to achieve this and having
37 applications deal with hardware implementation details regarding their
40 Support for different priority levels on a rule basis is provided, for
41 example in order to force a more specific rule to come before a more generic
42 one for packets matched by both. However hardware support for more than a
43 single priority level cannot be guaranteed. When supported, the number of
44 available priority levels is usually low, which is why they can also be
45 implemented in software by PMDs (e.g. missing priority levels may be
46 emulated by reordering rules).
48 In order to remain as hardware-agnostic as possible, by default all rules
49 are considered to have the same priority, which means that the order between
50 overlapping rules (when a packet is matched by several filters) is
53 PMDs may refuse to create overlapping rules at a given priority level when
54 they can be detected (e.g. if a pattern matches an existing filter).
56 Thus predictable results for a given priority level can only be achieved
57 with non-overlapping rules, using perfect matching on all protocol layers.
59 Flow rules can also be grouped, the flow rule priority is specific to the
60 group they belong to. All flow rules in a given group are thus processed within
61 the context of that group. Groups are not linked by default, so the logical
62 hierarchy of groups must be explicitly defined by flow rules themselves in each
63 group using the JUMP action to define the next group to redirect too. Only flow
64 rules defined in the default group 0 are guarantee to be matched against, this
65 makes group 0 the origin of any group hierarchy defined by an application.
67 Support for multiple actions per rule may be implemented internally on top
68 of non-default hardware priorities, as a result both features may not be
69 simultaneously available to applications.
71 Considering that allowed pattern/actions combinations cannot be known in
72 advance and would result in an impractically large number of capabilities to
73 expose, a method is provided to validate a given rule from the current
74 device configuration state.
76 This enables applications to check if the rule types they need is supported
77 at initialization time, before starting their data path. This method can be
78 used anytime, its only requirement being that the resources needed by a rule
79 should exist (e.g. a target RX queue should be configured first).
81 Each defined rule is associated with an opaque handle managed by the PMD,
82 applications are responsible for keeping it. These can be used for queries
83 and rules management, such as retrieving counters or other data and
86 To avoid resource leaks on the PMD side, handles must be explicitly
87 destroyed by the application before releasing associated resources such as
90 The following sections cover:
92 - **Attributes** (represented by ``struct rte_flow_attr``): properties of a
93 flow rule such as its direction (ingress or egress) and priority.
95 - **Pattern item** (represented by ``struct rte_flow_item``): part of a
96 matching pattern that either matches specific packet data or traffic
97 properties. It can also describe properties of the pattern itself, such as
100 - **Matching pattern**: traffic properties to look for, a combination of any
103 - **Actions** (represented by ``struct rte_flow_action``): operations to
104 perform whenever a packet is matched by a pattern.
112 Flow rules can be grouped by assigning them a common group number. Groups
113 allow a logical hierarchy of flow rule groups (tables) to be defined. These
114 groups can be supported virtually in the PMD or in the physical device.
115 Group 0 is the default group and this is the only group which flows are
116 guarantee to matched against, all subsequent groups can only be reached by
117 way of the JUMP action from a matched flow rule.
119 Although optional, applications are encouraged to group similar rules as
120 much as possible to fully take advantage of hardware capabilities
121 (e.g. optimized matching) and work around limitations (e.g. a single pattern
122 type possibly allowed in a given group), while being aware that the groups
123 hierarchies must be programmed explicitly.
125 Note that support for more than a single group is not guaranteed.
130 A priority level can be assigned to a flow rule, lower values
131 denote higher priority, with 0 as the maximum.
133 Priority levels are arbitrary and up to the application, they do
134 not need to be contiguous nor start from 0, however the maximum number
135 varies between devices and may be affected by existing flow rules.
137 A flow which matches multiple rules in the same group will always matched by
138 the rule with the highest priority in that group.
140 If a packet is matched by several rules of a given group for a given
141 priority level, the outcome is undefined. It can take any path, may be
142 duplicated or even cause unrecoverable errors.
144 Note that support for more than a single priority level is not guaranteed.
146 Attribute: Traffic direction
147 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
149 Flow rule patterns apply to inbound and/or outbound traffic.
151 In the context of this API, **ingress** and **egress** respectively stand
152 for **inbound** and **outbound** based on the standpoint of the application
153 creating a flow rule.
155 There are no exceptions to this definition.
157 Several pattern items and actions are valid and can be used in both
158 directions. At least one direction must be specified.
160 Specifying both directions at once for a given rule is not recommended but
161 may be valid in a few cases (e.g. shared counters).
166 Instead of simply matching the properties of traffic as it would appear on a
167 given DPDK port ID, enabling this attribute transfers a flow rule to the
168 lowest possible level of any device endpoints found in the pattern.
170 When supported, this effectively enables an application to reroute traffic
171 not necessarily intended for it (e.g. coming from or addressed to different
172 physical ports, VFs or applications) at the device level.
174 It complements the behavior of some pattern items such as `Item: PHY_PORT`_
175 and is meaningless without them.
177 When transferring flow rules, **ingress** and **egress** attributes
178 (`Attribute: Traffic direction`_) keep their original meaning, as if
179 processing traffic emitted or received by the application.
184 Pattern items fall in two categories:
186 - Matching protocol headers and packet data, usually associated with a
187 specification structure. These must be stacked in the same order as the
188 protocol layers to match inside packets, starting from the lowest.
190 - Matching meta-data or affecting pattern processing, often without a
191 specification structure. Since they do not match packet contents, their
192 position in the list is usually not relevant.
194 Item specification structures are used to match specific values among
195 protocol fields (or item properties). Documentation describes for each item
196 whether they are associated with one and their type name if so.
198 Up to three structures of the same type can be set for a given item:
200 - ``spec``: values to match (e.g. a given IPv4 address).
202 - ``last``: upper bound for an inclusive range with corresponding fields in
205 - ``mask``: bit-mask applied to both ``spec`` and ``last`` whose purpose is
206 to distinguish the values to take into account and/or partially mask them
207 out (e.g. in order to match an IPv4 address prefix).
209 Usage restrictions and expected behavior:
211 - Setting either ``mask`` or ``last`` without ``spec`` is an error.
213 - Field values in ``last`` which are either 0 or equal to the corresponding
214 values in ``spec`` are ignored; they do not generate a range. Nonzero
215 values lower than those in ``spec`` are not supported.
217 - Setting ``spec`` and optionally ``last`` without ``mask`` causes the PMD
218 to use the default mask defined for that item (defined as
219 ``rte_flow_item_{name}_mask`` constants).
221 - Not setting any of them (assuming item type allows it) is equivalent to
222 providing an empty (zeroed) ``mask`` for broad (nonspecific) matching.
224 - ``mask`` is a simple bit-mask applied before interpreting the contents of
225 ``spec`` and ``last``, which may yield unexpected results if not used
226 carefully. For example, if for an IPv4 address field, ``spec`` provides
227 *10.1.2.3*, ``last`` provides *10.3.4.5* and ``mask`` provides
228 *255.255.0.0*, the effective range becomes *10.1.0.0* to *10.3.255.255*.
230 Example of an item specification matching an Ethernet header:
232 .. _table_rte_flow_pattern_item_example:
234 .. table:: Ethernet item
236 +----------+----------+-----------------------+
237 | Field | Subfield | Value |
238 +==========+==========+=======================+
239 | ``spec`` | ``src`` | ``00:00:01:02:03:04`` |
240 | +----------+-----------------------+
241 | | ``dst`` | ``00:00:2a:66:00:01`` |
242 | +----------+-----------------------+
243 | | ``type`` | ``0x22aa`` |
244 +----------+----------+-----------------------+
245 | ``last`` | unspecified |
246 +----------+----------+-----------------------+
247 | ``mask`` | ``src`` | ``00:00:ff:ff:ff:00`` |
248 | +----------+-----------------------+
249 | | ``dst`` | ``00:00:00:00:00:ff`` |
250 | +----------+-----------------------+
251 | | ``type`` | ``0x0000`` |
252 +----------+----------+-----------------------+
254 Non-masked bits stand for any value (shown as ``?`` below), Ethernet headers
255 with the following properties are thus matched:
257 - ``src``: ``??:??:01:02:03:??``
258 - ``dst``: ``??:??:??:??:??:01``
259 - ``type``: ``0x????``
264 A pattern is formed by stacking items starting from the lowest protocol
265 layer to match. This stacking restriction does not apply to meta items which
266 can be placed anywhere in the stack without affecting the meaning of the
269 Patterns are terminated by END items.
273 .. _table_rte_flow_tcpv4_as_l4:
275 .. table:: TCPv4 as L4
291 .. _table_rte_flow_tcpv6_in_vxlan:
293 .. table:: TCPv6 in VXLAN
295 +-------+------------+
297 +=======+============+
299 +-------+------------+
301 +-------+------------+
303 +-------+------------+
305 +-------+------------+
307 +-------+------------+
309 +-------+------------+
311 +-------+------------+
313 +-------+------------+
317 .. _table_rte_flow_tcpv4_as_l4_meta:
319 .. table:: TCPv4 as L4 with meta items
341 The above example shows how meta items do not affect packet data matching
342 items, as long as those remain stacked properly. The resulting matching
343 pattern is identical to "TCPv4 as L4".
345 .. _table_rte_flow_udpv6_anywhere:
347 .. table:: UDPv6 anywhere
359 If supported by the PMD, omitting one or several protocol layers at the
360 bottom of the stack as in the above example (missing an Ethernet
361 specification) enables looking up anywhere in packets.
363 It is unspecified whether the payload of supported encapsulations
364 (e.g. VXLAN payload) is matched by such a pattern, which may apply to inner,
365 outer or both packets.
367 .. _table_rte_flow_invalid_l3:
369 .. table:: Invalid, missing L3
381 The above pattern is invalid due to a missing L3 specification between L2
382 (Ethernet) and L4 (UDP). Doing so is only allowed at the bottom and at the
388 They match meta-data or affect pattern processing instead of matching packet
389 data directly, most of them do not need a specification structure. This
390 particularity allows them to be specified anywhere in the stack without
391 causing any side effect.
396 End marker for item lists. Prevents further processing of items, thereby
399 - Its numeric value is 0 for convenience.
400 - PMD support is mandatory.
401 - ``spec``, ``last`` and ``mask`` are ignored.
403 .. _table_rte_flow_item_end:
407 +----------+---------+
409 +==========+=========+
410 | ``spec`` | ignored |
411 +----------+---------+
412 | ``last`` | ignored |
413 +----------+---------+
414 | ``mask`` | ignored |
415 +----------+---------+
420 Used as a placeholder for convenience. It is ignored and simply discarded by
423 - PMD support is mandatory.
424 - ``spec``, ``last`` and ``mask`` are ignored.
426 .. _table_rte_flow_item_void:
430 +----------+---------+
432 +==========+=========+
433 | ``spec`` | ignored |
434 +----------+---------+
435 | ``last`` | ignored |
436 +----------+---------+
437 | ``mask`` | ignored |
438 +----------+---------+
440 One usage example for this type is generating rules that share a common
441 prefix quickly without reallocating memory, only by updating item types:
443 .. _table_rte_flow_item_void_example:
445 .. table:: TCP, UDP or ICMP as L4
447 +-------+--------------------+
449 +=======+====================+
451 +-------+--------------------+
453 +-------+------+------+------+
454 | 2 | UDP | VOID | VOID |
455 +-------+------+------+------+
456 | 3 | VOID | TCP | VOID |
457 +-------+------+------+------+
458 | 4 | VOID | VOID | ICMP |
459 +-------+------+------+------+
461 +-------+--------------------+
466 Inverted matching, i.e. process packets that do not match the pattern.
468 - ``spec``, ``last`` and ``mask`` are ignored.
470 .. _table_rte_flow_item_invert:
474 +----------+---------+
476 +==========+=========+
477 | ``spec`` | ignored |
478 +----------+---------+
479 | ``last`` | ignored |
480 +----------+---------+
481 | ``mask`` | ignored |
482 +----------+---------+
484 Usage example, matching non-TCPv4 packets only:
486 .. _table_rte_flow_item_invert_example:
488 .. table:: Anything but TCPv4
507 Matches traffic originating from (ingress) or going to (egress) the physical
508 function of the current device.
510 If supported, should work even if the physical function is not managed by
511 the application and thus not associated with a DPDK port ID.
513 - Can be combined with any number of `Item: VF`_ to match both PF and VF
515 - ``spec``, ``last`` and ``mask`` must not be set.
517 .. _table_rte_flow_item_pf:
534 Matches traffic originating from (ingress) or going to (egress) a given
535 virtual function of the current device.
537 If supported, should work even if the virtual function is not managed by the
538 application and thus not associated with a DPDK port ID.
540 Note this pattern item does not match VF representors traffic which, as
541 separate entities, should be addressed through their own DPDK port IDs.
543 - Can be specified multiple times to match traffic addressed to several VF
545 - Can be combined with a PF item to match both PF and VF traffic.
546 - Default ``mask`` matches any VF ID.
548 .. _table_rte_flow_item_vf:
552 +----------+----------+---------------------------+
553 | Field | Subfield | Value |
554 +==========+==========+===========================+
555 | ``spec`` | ``id`` | destination VF ID |
556 +----------+----------+---------------------------+
557 | ``last`` | ``id`` | upper range value |
558 +----------+----------+---------------------------+
559 | ``mask`` | ``id`` | zeroed to match any VF ID |
560 +----------+----------+---------------------------+
565 Matches traffic originating from (ingress) or going to (egress) a physical
566 port of the underlying device.
568 The first PHY_PORT item overrides the physical port normally associated with
569 the specified DPDK input port (port_id). This item can be provided several
570 times to match additional physical ports.
572 Note that physical ports are not necessarily tied to DPDK input ports
573 (port_id) when those are not under DPDK control. Possible values are
574 specific to each device, they are not necessarily indexed from zero and may
577 As a device property, the list of allowed values as well as the value
578 associated with a port_id should be retrieved by other means.
580 - Default ``mask`` matches any port index.
582 .. _table_rte_flow_item_phy_port:
586 +----------+-----------+--------------------------------+
587 | Field | Subfield | Value |
588 +==========+===========+================================+
589 | ``spec`` | ``index`` | physical port index |
590 +----------+-----------+--------------------------------+
591 | ``last`` | ``index`` | upper range value |
592 +----------+-----------+--------------------------------+
593 | ``mask`` | ``index`` | zeroed to match any port index |
594 +----------+-----------+--------------------------------+
599 Matches traffic originating from (ingress) or going to (egress) a given DPDK
602 Normally only supported if the port ID in question is known by the
603 underlying PMD and related to the device the flow rule is created against.
605 This must not be confused with `Item: PHY_PORT`_ which refers to the
606 physical port of a device, whereas `Item: PORT_ID`_ refers to a ``struct
607 rte_eth_dev`` object on the application side (also known as "port
608 representor" depending on the kind of underlying device).
610 - Default ``mask`` matches the specified DPDK port ID.
612 .. _table_rte_flow_item_port_id:
616 +----------+----------+-----------------------------+
617 | Field | Subfield | Value |
618 +==========+==========+=============================+
619 | ``spec`` | ``id`` | DPDK port ID |
620 +----------+----------+-----------------------------+
621 | ``last`` | ``id`` | upper range value |
622 +----------+----------+-----------------------------+
623 | ``mask`` | ``id`` | zeroed to match any port ID |
624 +----------+----------+-----------------------------+
629 Matches an arbitrary integer value which was set using the ``MARK`` action in
630 a previously matched rule.
632 This item can only specified once as a match criteria as the ``MARK`` action can
633 only be specified once in a flow action.
635 Note the value of MARK field is arbitrary and application defined.
637 Depending on the underlying implementation the MARK item may be supported on
638 the physical device, with virtual groups in the PMD or not at all.
640 - Default ``mask`` matches any integer value.
642 .. _table_rte_flow_item_mark:
646 +----------+----------+---------------------------+
647 | Field | Subfield | Value |
648 +==========+==========+===========================+
649 | ``spec`` | ``id`` | integer value |
650 +----------+--------------------------------------+
651 | ``last`` | ``id`` | upper range value |
652 +----------+----------+---------------------------+
653 | ``mask`` | ``id`` | zeroed to match any value |
654 +----------+----------+---------------------------+
659 Matches tag item set by other flows. Multiple tags are supported by specifying
662 - Default ``mask`` matches the specified tag value and index.
664 .. _table_rte_flow_item_tag:
668 +----------+----------+----------------------------------------+
669 | Field | Subfield | Value |
670 +==========+===========+=======================================+
671 | ``spec`` | ``data`` | 32 bit flow tag value |
672 | +-----------+---------------------------------------+
673 | | ``index`` | index of flow tag |
674 +----------+-----------+---------------------------------------+
675 | ``last`` | ``data`` | upper range value |
676 | +-----------+---------------------------------------+
677 | | ``index`` | field is ignored |
678 +----------+-----------+---------------------------------------+
679 | ``mask`` | ``data`` | bit-mask applies to "spec" and "last" |
680 | +-----------+---------------------------------------+
681 | | ``index`` | field is ignored |
682 +----------+-----------+---------------------------------------+
687 Matches 32 bit metadata item set.
689 On egress, metadata can be set either by mbuf metadata field with
690 PKT_TX_DYNF_METADATA flag or ``SET_META`` action. On ingress, ``SET_META``
691 action sets metadata for a packet and the metadata will be reported via
692 ``metadata`` dynamic field of ``rte_mbuf`` with PKT_RX_DYNF_METADATA flag.
694 - Default ``mask`` matches the specified Rx metadata value.
696 .. _table_rte_flow_item_meta:
700 +----------+----------+---------------------------------------+
701 | Field | Subfield | Value |
702 +==========+==========+=======================================+
703 | ``spec`` | ``data`` | 32 bit metadata value |
704 +----------+----------+---------------------------------------+
705 | ``last`` | ``data`` | upper range value |
706 +----------+----------+---------------------------------------+
707 | ``mask`` | ``data`` | bit-mask applies to "spec" and "last" |
708 +----------+----------+---------------------------------------+
710 Data matching item types
711 ~~~~~~~~~~~~~~~~~~~~~~~~
713 Most of these are basically protocol header definitions with associated
714 bit-masks. They must be specified (stacked) from lowest to highest protocol
715 layer to form a matching pattern.
717 The following list is not exhaustive, new protocols will be added in the
723 Matches any protocol in place of the current layer, a single ANY may also
724 stand for several protocol layers.
726 This is usually specified as the first pattern item when looking for a
727 protocol anywhere in a packet.
729 - Default ``mask`` stands for any number of layers.
731 .. _table_rte_flow_item_any:
735 +----------+----------+--------------------------------------+
736 | Field | Subfield | Value |
737 +==========+==========+======================================+
738 | ``spec`` | ``num`` | number of layers covered |
739 +----------+----------+--------------------------------------+
740 | ``last`` | ``num`` | upper range value |
741 +----------+----------+--------------------------------------+
742 | ``mask`` | ``num`` | zeroed to cover any number of layers |
743 +----------+----------+--------------------------------------+
745 Example for VXLAN TCP payload matching regardless of outer L3 (IPv4 or IPv6)
746 and L4 (UDP) both matched by the first ANY specification, and inner L3 (IPv4
747 or IPv6) matched by the second ANY specification:
749 .. _table_rte_flow_item_any_example:
751 .. table:: TCP in VXLAN with wildcards
753 +-------+------+----------+----------+-------+
754 | Index | Item | Field | Subfield | Value |
755 +=======+======+==========+==========+=======+
757 +-------+------+----------+----------+-------+
758 | 1 | ANY | ``spec`` | ``num`` | 2 |
759 +-------+------+----------+----------+-------+
761 +-------+------------------------------------+
763 +-------+------+----------+----------+-------+
764 | 4 | ANY | ``spec`` | ``num`` | 1 |
765 +-------+------+----------+----------+-------+
767 +-------+------------------------------------+
769 +-------+------------------------------------+
774 Matches a byte string of a given length at a given offset.
776 Offset is either absolute (using the start of the packet) or relative to the
777 end of the previous matched item in the stack, in which case negative values
780 If search is enabled, offset is used as the starting point. The search area
781 can be delimited by setting limit to a nonzero value, which is the maximum
782 number of bytes after offset where the pattern may start.
784 Matching a zero-length pattern is allowed, doing so resets the relative
785 offset for subsequent items.
787 - This type does not support ranges (``last`` field).
788 - Default ``mask`` matches all fields exactly.
790 .. _table_rte_flow_item_raw:
794 +----------+--------------+-------------------------------------------------+
795 | Field | Subfield | Value |
796 +==========+==============+=================================================+
797 | ``spec`` | ``relative`` | look for pattern after the previous item |
798 | +--------------+-------------------------------------------------+
799 | | ``search`` | search pattern from offset (see also ``limit``) |
800 | +--------------+-------------------------------------------------+
801 | | ``reserved`` | reserved, must be set to zero |
802 | +--------------+-------------------------------------------------+
803 | | ``offset`` | absolute or relative offset for ``pattern`` |
804 | +--------------+-------------------------------------------------+
805 | | ``limit`` | search area limit for start of ``pattern`` |
806 | +--------------+-------------------------------------------------+
807 | | ``length`` | ``pattern`` length |
808 | +--------------+-------------------------------------------------+
809 | | ``pattern`` | byte string to look for |
810 +----------+--------------+-------------------------------------------------+
811 | ``last`` | if specified, either all 0 or with the same values as ``spec`` |
812 +----------+----------------------------------------------------------------+
813 | ``mask`` | bit-mask applied to ``spec`` values with usual behavior |
814 +----------+----------------------------------------------------------------+
816 Example pattern looking for several strings at various offsets of a UDP
817 payload, using combined RAW items:
819 .. _table_rte_flow_item_raw_example:
821 .. table:: UDP payload matching
823 +-------+------+----------+--------------+-------+
824 | Index | Item | Field | Subfield | Value |
825 +=======+======+==========+==============+=======+
827 +-------+----------------------------------------+
829 +-------+----------------------------------------+
831 +-------+------+----------+--------------+-------+
832 | 3 | RAW | ``spec`` | ``relative`` | 1 |
833 | | | +--------------+-------+
834 | | | | ``search`` | 1 |
835 | | | +--------------+-------+
836 | | | | ``offset`` | 10 |
837 | | | +--------------+-------+
838 | | | | ``limit`` | 0 |
839 | | | +--------------+-------+
840 | | | | ``length`` | 3 |
841 | | | +--------------+-------+
842 | | | | ``pattern`` | "foo" |
843 +-------+------+----------+--------------+-------+
844 | 4 | RAW | ``spec`` | ``relative`` | 1 |
845 | | | +--------------+-------+
846 | | | | ``search`` | 0 |
847 | | | +--------------+-------+
848 | | | | ``offset`` | 20 |
849 | | | +--------------+-------+
850 | | | | ``limit`` | 0 |
851 | | | +--------------+-------+
852 | | | | ``length`` | 3 |
853 | | | +--------------+-------+
854 | | | | ``pattern`` | "bar" |
855 +-------+------+----------+--------------+-------+
856 | 5 | RAW | ``spec`` | ``relative`` | 1 |
857 | | | +--------------+-------+
858 | | | | ``search`` | 0 |
859 | | | +--------------+-------+
860 | | | | ``offset`` | -29 |
861 | | | +--------------+-------+
862 | | | | ``limit`` | 0 |
863 | | | +--------------+-------+
864 | | | | ``length`` | 3 |
865 | | | +--------------+-------+
866 | | | | ``pattern`` | "baz" |
867 +-------+------+----------+--------------+-------+
869 +-------+----------------------------------------+
873 - Locate "foo" at least 10 bytes deep inside UDP payload.
874 - Locate "bar" after "foo" plus 20 bytes.
875 - Locate "baz" after "bar" minus 29 bytes.
877 Such a packet may be represented as follows (not to scale)::
880 | |<--------->| |<--------->|
882 |-----|------|-----|-----|-----|-----|-----------|-----|------|
883 | ETH | IPv4 | UDP | ... | baz | foo | ......... | bar | .... |
884 |-----|------|-----|-----|-----|-----|-----------|-----|------|
886 |<--------------------------->|
889 Note that matching subsequent pattern items would resume after "baz", not
890 "bar" since matching is always performed after the previous item of the
896 Matches an Ethernet header.
898 The ``type`` field either stands for "EtherType" or "TPID" when followed by
899 so-called layer 2.5 pattern items such as ``RTE_FLOW_ITEM_TYPE_VLAN``. In
900 the latter case, ``type`` refers to that of the outer header, with the inner
901 EtherType/TPID provided by the subsequent pattern item. This is the same
902 order as on the wire.
903 If the ``type`` field contains a TPID value, then only tagged packets with the
904 specified TPID will match the pattern.
905 The field ``has_vlan`` can be used to match any type of tagged packets,
906 instead of using the ``type`` field.
907 If the ``type`` and ``has_vlan`` fields are not specified, then both tagged
908 and untagged packets will match the pattern.
910 - ``dst``: destination MAC.
911 - ``src``: source MAC.
912 - ``type``: EtherType or TPID.
913 - ``has_vlan``: packet header contains at least one VLAN.
914 - Default ``mask`` matches destination and source addresses only.
919 Matches an 802.1Q/ad VLAN tag.
921 The corresponding standard outer EtherType (TPID) values are
922 ``RTE_ETHER_TYPE_VLAN`` or ``RTE_ETHER_TYPE_QINQ``. It can be overridden by the
923 preceding pattern item.
924 If a ``VLAN`` item is present in the pattern, then only tagged packets will
926 The field ``has_more_vlan`` can be used to match any type of tagged packets,
927 instead of using the ``inner_type field``.
928 If the ``inner_type`` and ``has_more_vlan`` fields are not specified,
929 then any tagged packets will match the pattern.
931 - ``tci``: tag control information.
932 - ``inner_type``: inner EtherType or TPID.
933 - ``has_more_vlan``: packet header contains at least one more VLAN, after this VLAN.
934 - Default ``mask`` matches the VID part of TCI only (lower 12 bits).
939 Matches an IPv4 header.
941 Note: IPv4 options are handled by dedicated pattern items.
943 - ``hdr``: IPv4 header definition (``rte_ip.h``).
944 - Default ``mask`` matches source and destination addresses only.
949 Matches an IPv6 header.
951 Dedicated flags indicate if header contains specific extension headers.
952 To match on packets containing a specific extension header, an application
953 should match on the dedicated flag set to 1.
954 To match on packets not containing a specific extension header, an application
955 should match on the dedicated flag clear to 0.
956 In case application doesn't care about the existence of a specific extension
957 header, it should not specify the dedicated flag for matching.
959 - ``hdr``: IPv6 header definition (``rte_ip.h``).
960 - ``has_hop_ext``: header contains Hop-by-Hop Options extension header.
961 - ``has_route_ext``: header contains Routing extension header.
962 - ``has_frag_ext``: header contains Fragment extension header.
963 - ``has_auth_ext``: header contains Authentication extension header.
964 - ``has_esp_ext``: header contains Encapsulation Security Payload extension header.
965 - ``has_dest_ext``: header contains Destination Options extension header.
966 - ``has_mobil_ext``: header contains Mobility extension header.
967 - ``has_hip_ext``: header contains Host Identity Protocol extension header.
968 - ``has_shim6_ext``: header contains Shim6 Protocol extension header.
969 - Default ``mask`` matches ``hdr`` source and destination addresses only.
974 Matches an ICMP header.
976 - ``hdr``: ICMP header definition (``rte_icmp.h``).
977 - Default ``mask`` matches ICMP type and code only.
982 Matches a UDP header.
984 - ``hdr``: UDP header definition (``rte_udp.h``).
985 - Default ``mask`` matches source and destination ports only.
990 Matches a TCP header.
992 - ``hdr``: TCP header definition (``rte_tcp.h``).
993 - Default ``mask`` matches source and destination ports only.
998 Matches a SCTP header.
1000 - ``hdr``: SCTP header definition (``rte_sctp.h``).
1001 - Default ``mask`` matches source and destination ports only.
1006 Matches a VXLAN header (RFC 7348).
1008 - ``flags``: normally 0x08 (I flag).
1009 - ``rsvd0``: reserved, normally 0x000000.
1010 - ``vni``: VXLAN network identifier.
1011 - ``rsvd1``: reserved, normally 0x00.
1012 - Default ``mask`` matches VNI only.
1017 Matches an IEEE 802.1BR E-Tag header.
1019 The corresponding standard outer EtherType (TPID) value is
1020 ``RTE_ETHER_TYPE_ETAG``. It can be overridden by the preceding pattern item.
1022 - ``epcp_edei_in_ecid_b``: E-Tag control information (E-TCI), E-PCP (3b),
1023 E-DEI (1b), ingress E-CID base (12b).
1024 - ``rsvd_grp_ecid_b``: reserved (2b), GRP (2b), E-CID base (12b).
1025 - ``in_ecid_e``: ingress E-CID ext.
1026 - ``ecid_e``: E-CID ext.
1027 - ``inner_type``: inner EtherType or TPID.
1028 - Default ``mask`` simultaneously matches GRP and E-CID base.
1033 Matches a NVGRE header (RFC 7637).
1035 - ``c_k_s_rsvd0_ver``: checksum (1b), undefined (1b), key bit (1b),
1036 sequence number (1b), reserved 0 (9b), version (3b). This field must have
1037 value 0x2000 according to RFC 7637.
1038 - ``protocol``: protocol type (0x6558).
1039 - ``tni``: virtual subnet ID.
1040 - ``flow_id``: flow ID.
1041 - Default ``mask`` matches TNI only.
1046 Matches a MPLS header.
1048 - ``label_tc_s_ttl``: label, TC, Bottom of Stack and TTL.
1049 - Default ``mask`` matches label only.
1054 Matches a GRE header.
1056 - ``c_rsvd0_ver``: checksum, reserved 0 and version.
1057 - ``protocol``: protocol type.
1058 - Default ``mask`` matches protocol only.
1063 Matches a GRE key field.
1064 This should be preceded by item ``GRE``.
1066 - Value to be matched is a big-endian 32 bit integer.
1067 - When this item present it implicitly match K bit in default mask as "1"
1072 Fuzzy pattern match, expect faster than default.
1074 This is for device that support fuzzy match option. Usually a fuzzy match is
1075 fast but the cost is accuracy. i.e. Signature Match only match pattern's hash
1076 value, but it is possible two different patterns have the same hash value.
1078 Matching accuracy level can be configured by threshold. Driver can divide the
1079 range of threshold and map to different accuracy levels that device support.
1081 Threshold 0 means perfect match (no fuzziness), while threshold 0xffffffff
1082 means fuzziest match.
1084 .. _table_rte_flow_item_fuzzy:
1088 +----------+---------------+--------------------------------------------------+
1089 | Field | Subfield | Value |
1090 +==========+===============+==================================================+
1091 | ``spec`` | ``threshold`` | 0 as perfect match, 0xffffffff as fuzziest match |
1092 +----------+---------------+--------------------------------------------------+
1093 | ``last`` | ``threshold`` | upper range value |
1094 +----------+---------------+--------------------------------------------------+
1095 | ``mask`` | ``threshold`` | bit-mask apply to "spec" and "last" |
1096 +----------+---------------+--------------------------------------------------+
1098 Usage example, fuzzy match a TCPv4 packets:
1100 .. _table_rte_flow_item_fuzzy_example:
1102 .. table:: Fuzzy matching
1104 +-------+----------+
1106 +=======+==========+
1108 +-------+----------+
1110 +-------+----------+
1112 +-------+----------+
1114 +-------+----------+
1116 +-------+----------+
1118 Item: ``GTP``, ``GTPC``, ``GTPU``
1119 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1121 Matches a GTPv1 header.
1123 Note: GTP, GTPC and GTPU use the same structure. GTPC and GTPU item
1124 are defined for a user-friendly API when creating GTP-C and GTP-U
1127 - ``v_pt_rsv_flags``: version (3b), protocol type (1b), reserved (1b),
1128 extension header flag (1b), sequence number flag (1b), N-PDU number
1130 - ``msg_type``: message type.
1131 - ``msg_len``: message length.
1132 - ``teid``: tunnel endpoint identifier.
1133 - Default ``mask`` matches teid only.
1138 Matches an ESP header.
1140 - ``hdr``: ESP header definition (``rte_esp.h``).
1141 - Default ``mask`` matches SPI only.
1146 Matches a GENEVE header.
1148 - ``ver_opt_len_o_c_rsvd0``: version (2b), length of the options fields (6b),
1149 OAM packet (1b), critical options present (1b), reserved 0 (6b).
1150 - ``protocol``: protocol type.
1151 - ``vni``: virtual network identifier.
1152 - ``rsvd1``: reserved, normally 0x00.
1153 - Default ``mask`` matches VNI only.
1158 Matches a VXLAN-GPE header (draft-ietf-nvo3-vxlan-gpe-05).
1160 - ``flags``: normally 0x0C (I and P flags).
1161 - ``rsvd0``: reserved, normally 0x0000.
1162 - ``protocol``: protocol type.
1163 - ``vni``: VXLAN network identifier.
1164 - ``rsvd1``: reserved, normally 0x00.
1165 - Default ``mask`` matches VNI only.
1167 Item: ``ARP_ETH_IPV4``
1168 ^^^^^^^^^^^^^^^^^^^^^^
1170 Matches an ARP header for Ethernet/IPv4.
1172 - ``hdr``: hardware type, normally 1.
1173 - ``pro``: protocol type, normally 0x0800.
1174 - ``hln``: hardware address length, normally 6.
1175 - ``pln``: protocol address length, normally 4.
1176 - ``op``: opcode (1 for request, 2 for reply).
1177 - ``sha``: sender hardware address.
1178 - ``spa``: sender IPv4 address.
1179 - ``tha``: target hardware address.
1180 - ``tpa``: target IPv4 address.
1181 - Default ``mask`` matches SHA, SPA, THA and TPA.
1186 Matches the presence of any IPv6 extension header.
1188 - ``next_hdr``: next header.
1189 - Default ``mask`` matches ``next_hdr``.
1191 Normally preceded by any of:
1196 Item: ``IPV6_FRAG_EXT``
1197 ^^^^^^^^^^^^^^^^^^^^^^^
1199 Matches the presence of IPv6 fragment extension header.
1201 - ``hdr``: IPv6 fragment extension header definition (``rte_ip.h``).
1203 Normally preceded by any of:
1211 Matches any ICMPv6 header.
1213 - ``type``: ICMPv6 type.
1214 - ``code``: ICMPv6 code.
1215 - ``checksum``: ICMPv6 checksum.
1216 - Default ``mask`` matches ``type`` and ``code``.
1218 Item: ``ICMP6_ND_NS``
1219 ^^^^^^^^^^^^^^^^^^^^^
1221 Matches an ICMPv6 neighbor discovery solicitation.
1223 - ``type``: ICMPv6 type, normally 135.
1224 - ``code``: ICMPv6 code, normally 0.
1225 - ``checksum``: ICMPv6 checksum.
1226 - ``reserved``: reserved, normally 0.
1227 - ``target_addr``: target address.
1228 - Default ``mask`` matches target address only.
1230 Item: ``ICMP6_ND_NA``
1231 ^^^^^^^^^^^^^^^^^^^^^
1233 Matches an ICMPv6 neighbor discovery advertisement.
1235 - ``type``: ICMPv6 type, normally 136.
1236 - ``code``: ICMPv6 code, normally 0.
1237 - ``checksum``: ICMPv6 checksum.
1238 - ``rso_reserved``: route flag (1b), solicited flag (1b), override flag
1239 (1b), reserved (29b).
1240 - ``target_addr``: target address.
1241 - Default ``mask`` matches target address only.
1243 Item: ``ICMP6_ND_OPT``
1244 ^^^^^^^^^^^^^^^^^^^^^^
1246 Matches the presence of any ICMPv6 neighbor discovery option.
1248 - ``type``: ND option type.
1249 - ``length``: ND option length.
1250 - Default ``mask`` matches type only.
1252 Normally preceded by any of:
1254 - `Item: ICMP6_ND_NA`_
1255 - `Item: ICMP6_ND_NS`_
1256 - `Item: ICMP6_ND_OPT`_
1258 Item: ``ICMP6_ND_OPT_SLA_ETH``
1259 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1261 Matches an ICMPv6 neighbor discovery source Ethernet link-layer address
1264 - ``type``: ND option type, normally 1.
1265 - ``length``: ND option length, normally 1.
1266 - ``sla``: source Ethernet LLA.
1267 - Default ``mask`` matches source link-layer address only.
1269 Normally preceded by any of:
1271 - `Item: ICMP6_ND_NA`_
1272 - `Item: ICMP6_ND_OPT`_
1274 Item: ``ICMP6_ND_OPT_TLA_ETH``
1275 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1277 Matches an ICMPv6 neighbor discovery target Ethernet link-layer address
1280 - ``type``: ND option type, normally 2.
1281 - ``length``: ND option length, normally 1.
1282 - ``tla``: target Ethernet LLA.
1283 - Default ``mask`` matches target link-layer address only.
1285 Normally preceded by any of:
1287 - `Item: ICMP6_ND_NS`_
1288 - `Item: ICMP6_ND_OPT`_
1293 Matches an application specific 32 bit metadata item.
1295 - Default ``mask`` matches the specified metadata value.
1300 Matches a GTP PDU extension header with type 0x85.
1302 - ``pdu_type``: PDU type.
1303 - ``qfi``: QoS flow identifier.
1304 - Default ``mask`` matches QFI only.
1306 Item: ``PPPOES``, ``PPPOED``
1307 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1309 Matches a PPPoE header.
1311 - ``version_type``: version (4b), type (4b).
1312 - ``code``: message type.
1313 - ``session_id``: session identifier.
1314 - ``length``: payload length.
1316 Item: ``PPPOE_PROTO_ID``
1317 ^^^^^^^^^^^^^^^^^^^^^^^^
1319 Matches a PPPoE session protocol identifier.
1321 - ``proto_id``: PPP protocol identifier.
1322 - Default ``mask`` matches proto_id only.
1327 Matches a network service header (RFC 8300).
1329 - ``version``: normally 0x0 (2 bits).
1330 - ``oam_pkt``: indicate oam packet (1 bit).
1331 - ``reserved``: reserved bit (1 bit).
1332 - ``ttl``: maximum SFF hopes (6 bits).
1333 - ``length``: total length in 4 bytes words (6 bits).
1334 - ``reserved1``: reserved1 bits (4 bits).
1335 - ``mdtype``: ndicates format of NSH header (4 bits).
1336 - ``next_proto``: indicates protocol type of encap data (8 bits).
1337 - ``spi``: service path identifier (3 bytes).
1338 - ``sindex``: service index (1 byte).
1339 - Default ``mask`` matches mdtype, next_proto, spi, sindex.
1345 Matches a Internet Group Management Protocol (RFC 2236).
1347 - ``type``: IGMP message type (Query/Report).
1348 - ``max_resp_time``: max time allowed before sending report.
1349 - ``checksum``: checksum, 1s complement of whole IGMP message.
1350 - ``group_addr``: group address, for Query value will be 0.
1351 - Default ``mask`` matches group_addr.
1357 Matches a IP Authentication Header (RFC 4302).
1359 - ``next_hdr``: next payload after AH.
1360 - ``payload_len``: total length of AH in 4B words.
1361 - ``reserved``: reserved bits.
1362 - ``spi``: security parameters index.
1363 - ``seq_num``: counter value increased by 1 on each packet sent.
1364 - Default ``mask`` matches spi.
1369 Matches a HIGIG2 header field. It is layer 2.5 protocol and used in
1372 - Default ``mask`` matches classification and vlan.
1377 Matches a L2TPv3 over IP header.
1379 - ``session_id``: L2TPv3 over IP session identifier.
1380 - Default ``mask`` matches session_id only.
1385 Matches a PFCP Header.
1387 - ``s_field``: S field.
1388 - ``msg_type``: message type.
1389 - ``msg_len``: message length.
1390 - ``seid``: session endpoint identifier.
1391 - Default ``mask`` matches s_field and seid.
1396 Matches a eCPRI header.
1398 - ``hdr``: eCPRI header definition (``rte_ecpri.h``).
1399 - Default ``mask`` matches nothing, for all eCPRI messages.
1404 Each possible action is represented by a type.
1405 An action can have an associated configuration object.
1406 Several actions combined in a list can be assigned
1407 to a flow rule and are performed in order.
1409 They fall in three categories:
1411 - Actions that modify the fate of matching traffic, for instance by dropping
1412 or assigning it a specific destination.
1414 - Actions that modify matching traffic contents or its properties. This
1415 includes adding/removing encapsulation, encryption, compression and marks.
1417 - Actions related to the flow rule itself, such as updating counters or
1418 making it non-terminating.
1420 Flow rules being terminating by default, not specifying any action of the
1421 fate kind results in undefined behavior. This applies to both ingress and
1424 PASSTHRU, when supported, makes a flow rule non-terminating.
1426 Like matching patterns, action lists are terminated by END items.
1428 Example of action that redirects packets to queue index 10:
1430 .. _table_rte_flow_action_example:
1432 .. table:: Queue action
1434 +-----------+-------+
1436 +===========+=======+
1438 +-----------+-------+
1440 Actions are performed in list order:
1442 .. _table_rte_flow_count_then_drop:
1444 .. table:: Count then drop
1458 .. _table_rte_flow_mark_count_redirect:
1460 .. table:: Mark, count then redirect
1462 +-------+--------+------------+-------+
1463 | Index | Action | Field | Value |
1464 +=======+========+============+=======+
1465 | 0 | MARK | ``mark`` | 0x2a |
1466 +-------+--------+------------+-------+
1467 | 1 | COUNT | ``shared`` | 0 |
1468 | | +------------+-------+
1470 +-------+--------+------------+-------+
1471 | 2 | QUEUE | ``queue`` | 10 |
1472 +-------+--------+------------+-------+
1474 +-------+-----------------------------+
1478 .. _table_rte_flow_redirect_queue_5:
1480 .. table:: Redirect to queue 5
1482 +-------+--------+-----------+-------+
1483 | Index | Action | Field | Value |
1484 +=======+========+===========+=======+
1486 +-------+--------+-----------+-------+
1487 | 1 | QUEUE | ``queue`` | 5 |
1488 +-------+--------+-----------+-------+
1490 +-------+----------------------------+
1492 In the above example, while DROP and QUEUE must be performed in order, both
1493 have to happen before reaching END. Only QUEUE has a visible effect.
1495 Note that such a list may be thought as ambiguous and rejected on that
1498 .. _table_rte_flow_redirect_queue_5_3:
1500 .. table:: Redirect to queues 5 and 3
1502 +-------+--------+-----------+-------+
1503 | Index | Action | Field | Value |
1504 +=======+========+===========+=======+
1505 | 0 | QUEUE | ``queue`` | 5 |
1506 +-------+--------+-----------+-------+
1508 +-------+--------+-----------+-------+
1509 | 2 | QUEUE | ``queue`` | 3 |
1510 +-------+--------+-----------+-------+
1512 +-------+----------------------------+
1514 As previously described, all actions must be taken into account. This
1515 effectively duplicates traffic to both queues. The above example also shows
1516 that VOID is ignored.
1521 Common action types are described in this section. Like pattern item types,
1522 this list is not exhaustive as new actions will be added in the future.
1527 End marker for action lists. Prevents further processing of actions, thereby
1530 - Its numeric value is 0 for convenience.
1531 - PMD support is mandatory.
1532 - No configurable properties.
1534 .. _table_rte_flow_action_end:
1547 Used as a placeholder for convenience. It is ignored and simply discarded by
1550 - PMD support is mandatory.
1551 - No configurable properties.
1553 .. _table_rte_flow_action_void:
1563 Action: ``PASSTHRU``
1564 ^^^^^^^^^^^^^^^^^^^^
1566 Leaves traffic up for additional processing by subsequent flow rules; makes
1567 a flow rule non-terminating.
1569 - No configurable properties.
1571 .. _table_rte_flow_action_passthru:
1581 Example to copy a packet to a queue and continue processing by subsequent
1584 .. _table_rte_flow_action_passthru_example:
1586 .. table:: Copy to queue 8
1588 +-------+--------+-----------+-------+
1589 | Index | Action | Field | Value |
1590 +=======+========+===========+=======+
1592 +-------+--------+-----------+-------+
1593 | 1 | QUEUE | ``queue`` | 8 |
1594 +-------+--------+-----------+-------+
1596 +-------+----------------------------+
1601 Redirects packets to a group on the current device.
1603 In a hierarchy of groups, which can be used to represent physical or logical
1604 flow group/tables on the device, this action redirects the matched flow to
1605 the specified group on that device.
1607 If a matched flow is redirected to a table which doesn't contain a matching
1608 rule for that flow then the behavior is undefined and the resulting behavior
1609 is up to the specific device. Best practice when using groups would be define
1610 a default flow rule for each group which a defines the default actions in that
1611 group so a consistent behavior is defined.
1613 Defining an action for matched flow in a group to jump to a group which is
1614 higher in the group hierarchy may not be supported by physical devices,
1615 depending on how groups are mapped to the physical devices. In the
1616 definitions of jump actions, applications should be aware that it may be
1617 possible to define flow rules which trigger an undefined behavior causing
1618 flows to loop between groups.
1620 .. _table_rte_flow_action_jump:
1624 +-----------+------------------------------+
1626 +===========+==============================+
1627 | ``group`` | Group to redirect packets to |
1628 +-----------+------------------------------+
1633 Attaches an integer value to packets and sets ``PKT_RX_FDIR`` and
1634 ``PKT_RX_FDIR_ID`` mbuf flags.
1636 This value is arbitrary and application-defined. Maximum allowed value
1637 depends on the underlying implementation. It is returned in the
1638 ``hash.fdir.hi`` mbuf field.
1640 .. _table_rte_flow_action_mark:
1644 +--------+--------------------------------------+
1646 +========+======================================+
1647 | ``id`` | integer value to return with packets |
1648 +--------+--------------------------------------+
1653 Flags packets. Similar to `Action: MARK`_ without a specific value; only
1654 sets the ``PKT_RX_FDIR`` mbuf flag.
1656 - No configurable properties.
1658 .. _table_rte_flow_action_flag:
1671 Assigns packets to a given queue index.
1673 .. _table_rte_flow_action_queue:
1677 +-----------+--------------------+
1679 +===========+====================+
1680 | ``index`` | queue index to use |
1681 +-----------+--------------------+
1688 - No configurable properties.
1690 .. _table_rte_flow_action_drop:
1703 Adds a counter action to a matched flow.
1705 If more than one count action is specified in a single flow rule, then each
1706 action must specify a unique id.
1708 Counters can be retrieved and reset through ``rte_flow_query()``, see
1709 ``struct rte_flow_query_count``.
1711 The shared flag indicates whether the counter is unique to the flow rule the
1712 action is specified with, or whether it is a shared counter.
1714 For a count action with the shared flag set, then a global device
1715 namespace is assumed for the counter id, so that any matched flow rules using
1716 a count action with the same counter id on the same port will contribute to
1719 For ports within the same switch domain then the counter id namespace extends
1720 to all ports within that switch domain.
1722 The shared flag is DEPRECATED and ``SHARED`` ``COUNT`` action should be used
1723 to make shared counters.
1725 .. _table_rte_flow_action_count:
1729 +------------+---------------------------------+
1731 +============+=================================+
1732 | ``shared`` | DEPRECATED, shared counter flag |
1733 +------------+---------------------------------+
1734 | ``id`` | counter id |
1735 +------------+---------------------------------+
1737 Query structure to retrieve and reset flow rule counters:
1739 .. _table_rte_flow_query_count:
1741 .. table:: COUNT query
1743 +---------------+-----+-----------------------------------+
1744 | Field | I/O | Value |
1745 +===============+=====+===================================+
1746 | ``reset`` | in | reset counter after query |
1747 +---------------+-----+-----------------------------------+
1748 | ``hits_set`` | out | ``hits`` field is set |
1749 +---------------+-----+-----------------------------------+
1750 | ``bytes_set`` | out | ``bytes`` field is set |
1751 +---------------+-----+-----------------------------------+
1752 | ``hits`` | out | number of hits for this rule |
1753 +---------------+-----+-----------------------------------+
1754 | ``bytes`` | out | number of bytes through this rule |
1755 +---------------+-----+-----------------------------------+
1760 Similar to QUEUE, except RSS is additionally performed on packets to spread
1761 them among several queues according to the provided parameters.
1763 Unlike global RSS settings used by other DPDK APIs, unsetting the ``types``
1764 field does not disable RSS in a flow rule. Doing so instead requests safe
1765 unspecified "best-effort" settings from the underlying PMD, which depending
1766 on the flow rule, may result in anything ranging from empty (single queue)
1767 to all-inclusive RSS.
1769 If non-applicable for matching packets RSS types are requested,
1770 these RSS types are simply ignored. For example, it happens if:
1772 - Hashing of both TCP and UDP ports is requested
1773 (only one can be present in a packet).
1775 - Requested RSS types contradict to flow rule pattern
1776 (e.g. pattern has UDP item, but RSS types contain TCP).
1778 If requested RSS hash types are not supported by the Ethernet device at all
1779 (not reported in ``dev_info.flow_type_rss_offloads``),
1780 the flow creation will fail.
1782 Note: RSS hash result is stored in the ``hash.rss`` mbuf field which
1783 overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi``
1784 field only, both can be requested simultaneously.
1786 Also, regarding packet encapsulation ``level``:
1788 - ``0`` requests the default behavior. Depending on the packet type, it can
1789 mean outermost, innermost, anything in between or even no RSS.
1791 It basically stands for the innermost encapsulation level RSS can be
1792 performed on according to PMD and device capabilities.
1794 - ``1`` requests RSS to be performed on the outermost packet encapsulation
1797 - ``2`` and subsequent values request RSS to be performed on the specified
1798 inner packet encapsulation level, from outermost to innermost (lower to
1801 Values other than ``0`` are not necessarily supported.
1803 Requesting a specific RSS level on unrecognized traffic results in undefined
1804 behavior. For predictable results, it is recommended to make the flow rule
1805 pattern match packet headers up to the requested encapsulation level so that
1806 only matching traffic goes through.
1808 .. _table_rte_flow_action_rss:
1812 +---------------+---------------------------------------------+
1814 +===============+=============================================+
1815 | ``func`` | RSS hash function to apply |
1816 +---------------+---------------------------------------------+
1817 | ``level`` | encapsulation level for ``types`` |
1818 +---------------+---------------------------------------------+
1819 | ``types`` | specific RSS hash types (see ``ETH_RSS_*``) |
1820 +---------------+---------------------------------------------+
1821 | ``key_len`` | hash key length in bytes |
1822 +---------------+---------------------------------------------+
1823 | ``queue_num`` | number of entries in ``queue`` |
1824 +---------------+---------------------------------------------+
1825 | ``key`` | hash key |
1826 +---------------+---------------------------------------------+
1827 | ``queue`` | queue indices to use |
1828 +---------------+---------------------------------------------+
1833 Directs matching traffic to the physical function (PF) of the current
1838 - No configurable properties.
1840 .. _table_rte_flow_action_pf:
1853 Directs matching traffic to a given virtual function of the current device.
1855 Packets matched by a VF pattern item can be redirected to their original VF
1856 ID instead of the specified one. This parameter may not be available and is
1857 not guaranteed to work properly if the VF part is matched by a prior flow
1858 rule or if packets are not addressed to a VF in the first place.
1862 .. _table_rte_flow_action_vf:
1866 +--------------+--------------------------------+
1868 +==============+================================+
1869 | ``original`` | use original VF ID if possible |
1870 +--------------+--------------------------------+
1872 +--------------+--------------------------------+
1874 Action: ``PHY_PORT``
1875 ^^^^^^^^^^^^^^^^^^^^
1877 Directs matching traffic to a given physical port index of the underlying
1880 See `Item: PHY_PORT`_.
1882 .. _table_rte_flow_action_phy_port:
1886 +--------------+-------------------------------------+
1888 +==============+=====================================+
1889 | ``original`` | use original port index if possible |
1890 +--------------+-------------------------------------+
1891 | ``index`` | physical port index |
1892 +--------------+-------------------------------------+
1896 Directs matching traffic to a given DPDK port ID.
1898 See `Item: PORT_ID`_.
1900 .. _table_rte_flow_action_port_id:
1904 +--------------+---------------------------------------+
1906 +==============+=======================================+
1907 | ``original`` | use original DPDK port ID if possible |
1908 +--------------+---------------------------------------+
1909 | ``id`` | DPDK port ID |
1910 +--------------+---------------------------------------+
1915 Applies a stage of metering and policing.
1917 The metering and policing (MTR) object has to be first created using the
1918 rte_mtr_create() API function. The ID of the MTR object is specified as
1919 action parameter. More than one flow can use the same MTR object through
1920 the meter action. The MTR object can be further updated or queried using
1923 .. _table_rte_flow_action_meter:
1927 +--------------+---------------+
1929 +==============+===============+
1930 | ``mtr_id`` | MTR object ID |
1931 +--------------+---------------+
1933 Action: ``SECURITY``
1934 ^^^^^^^^^^^^^^^^^^^^
1936 Perform the security action on flows matched by the pattern items
1937 according to the configuration of the security session.
1939 This action modifies the payload of matched flows. For INLINE_CRYPTO, the
1940 security protocol headers and IV are fully provided by the application as
1941 specified in the flow pattern. The payload of matching packets is
1942 encrypted on egress, and decrypted and authenticated on ingress.
1943 For INLINE_PROTOCOL, the security protocol is fully offloaded to HW,
1944 providing full encapsulation and decapsulation of packets in security
1945 protocols. The flow pattern specifies both the outer security header fields
1946 and the inner packet fields. The security session specified in the action
1947 must match the pattern parameters.
1949 The security session specified in the action must be created on the same
1950 port as the flow action that is being specified.
1952 The ingress/egress flow attribute should match that specified in the
1953 security session if the security session supports the definition of the
1956 Multiple flows can be configured to use the same security session.
1958 .. _table_rte_flow_action_security:
1962 +----------------------+--------------------------------------+
1964 +======================+======================================+
1965 | ``security_session`` | security session to apply |
1966 +----------------------+--------------------------------------+
1968 The following is an example of configuring IPsec inline using the
1969 INLINE_CRYPTO security session:
1971 The encryption algorithm, keys and salt are part of the opaque
1972 ``rte_security_session``. The SA is identified according to the IP and ESP
1973 fields in the pattern items.
1975 .. _table_rte_flow_item_esp_inline_example:
1977 .. table:: IPsec inline crypto flow pattern items.
1979 +-------+----------+
1981 +=======+==========+
1983 +-------+----------+
1985 +-------+----------+
1987 +-------+----------+
1989 +-------+----------+
1991 .. _table_rte_flow_action_esp_inline_example:
1993 .. table:: IPsec inline flow actions.
1995 +-------+----------+
1997 +=======+==========+
1999 +-------+----------+
2001 +-------+----------+
2003 Action: ``OF_SET_MPLS_TTL``
2004 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
2006 Implements ``OFPAT_SET_MPLS_TTL`` ("MPLS TTL") as defined by the `OpenFlow
2007 Switch Specification`_.
2009 .. _table_rte_flow_action_of_set_mpls_ttl:
2011 .. table:: OF_SET_MPLS_TTL
2013 +--------------+----------+
2015 +==============+==========+
2016 | ``mpls_ttl`` | MPLS TTL |
2017 +--------------+----------+
2019 Action: ``OF_DEC_MPLS_TTL``
2020 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
2022 Implements ``OFPAT_DEC_MPLS_TTL`` ("decrement MPLS TTL") as defined by the
2023 `OpenFlow Switch Specification`_.
2025 .. _table_rte_flow_action_of_dec_mpls_ttl:
2027 .. table:: OF_DEC_MPLS_TTL
2035 Action: ``OF_SET_NW_TTL``
2036 ^^^^^^^^^^^^^^^^^^^^^^^^^
2038 Implements ``OFPAT_SET_NW_TTL`` ("IP TTL") as defined by the `OpenFlow
2039 Switch Specification`_.
2041 .. _table_rte_flow_action_of_set_nw_ttl:
2043 .. table:: OF_SET_NW_TTL
2045 +------------+--------+
2047 +============+========+
2048 | ``nw_ttl`` | IP TTL |
2049 +------------+--------+
2051 Action: ``OF_DEC_NW_TTL``
2052 ^^^^^^^^^^^^^^^^^^^^^^^^^
2054 Implements ``OFPAT_DEC_NW_TTL`` ("decrement IP TTL") as defined by the
2055 `OpenFlow Switch Specification`_.
2057 .. _table_rte_flow_action_of_dec_nw_ttl:
2059 .. table:: OF_DEC_NW_TTL
2067 Action: ``OF_COPY_TTL_OUT``
2068 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
2070 Implements ``OFPAT_COPY_TTL_OUT`` ("copy TTL "outwards" -- from
2071 next-to-outermost to outermost") as defined by the `OpenFlow Switch
2074 .. _table_rte_flow_action_of_copy_ttl_out:
2076 .. table:: OF_COPY_TTL_OUT
2084 Action: ``OF_COPY_TTL_IN``
2085 ^^^^^^^^^^^^^^^^^^^^^^^^^^
2087 Implements ``OFPAT_COPY_TTL_IN`` ("copy TTL "inwards" -- from outermost to
2088 next-to-outermost") as defined by the `OpenFlow Switch Specification`_.
2090 .. _table_rte_flow_action_of_copy_ttl_in:
2092 .. table:: OF_COPY_TTL_IN
2100 Action: ``OF_POP_VLAN``
2101 ^^^^^^^^^^^^^^^^^^^^^^^
2103 Implements ``OFPAT_POP_VLAN`` ("pop the outer VLAN tag") as defined
2104 by the `OpenFlow Switch Specification`_.
2106 .. _table_rte_flow_action_of_pop_vlan:
2108 .. table:: OF_POP_VLAN
2116 Action: ``OF_PUSH_VLAN``
2117 ^^^^^^^^^^^^^^^^^^^^^^^^
2119 Implements ``OFPAT_PUSH_VLAN`` ("push a new VLAN tag") as defined by the
2120 `OpenFlow Switch Specification`_.
2122 .. _table_rte_flow_action_of_push_vlan:
2124 .. table:: OF_PUSH_VLAN
2126 +---------------+-----------+
2128 +===============+===========+
2129 | ``ethertype`` | EtherType |
2130 +---------------+-----------+
2132 Action: ``OF_SET_VLAN_VID``
2133 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
2135 Implements ``OFPAT_SET_VLAN_VID`` ("set the 802.1q VLAN id") as defined by
2136 the `OpenFlow Switch Specification`_.
2138 .. _table_rte_flow_action_of_set_vlan_vid:
2140 .. table:: OF_SET_VLAN_VID
2142 +--------------+---------+
2144 +==============+=========+
2145 | ``vlan_vid`` | VLAN id |
2146 +--------------+---------+
2148 Action: ``OF_SET_VLAN_PCP``
2149 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
2151 Implements ``OFPAT_SET_LAN_PCP`` ("set the 802.1q priority") as defined by
2152 the `OpenFlow Switch Specification`_.
2154 .. _table_rte_flow_action_of_set_vlan_pcp:
2156 .. table:: OF_SET_VLAN_PCP
2158 +--------------+---------------+
2160 +==============+===============+
2161 | ``vlan_pcp`` | VLAN priority |
2162 +--------------+---------------+
2164 Action: ``OF_POP_MPLS``
2165 ^^^^^^^^^^^^^^^^^^^^^^^
2167 Implements ``OFPAT_POP_MPLS`` ("pop the outer MPLS tag") as defined by the
2168 `OpenFlow Switch Specification`_.
2170 .. _table_rte_flow_action_of_pop_mpls:
2172 .. table:: OF_POP_MPLS
2174 +---------------+-----------+
2176 +===============+===========+
2177 | ``ethertype`` | EtherType |
2178 +---------------+-----------+
2180 Action: ``OF_PUSH_MPLS``
2181 ^^^^^^^^^^^^^^^^^^^^^^^^
2183 Implements ``OFPAT_PUSH_MPLS`` ("push a new MPLS tag") as defined by the
2184 `OpenFlow Switch Specification`_.
2186 .. _table_rte_flow_action_of_push_mpls:
2188 .. table:: OF_PUSH_MPLS
2190 +---------------+-----------+
2192 +===============+===========+
2193 | ``ethertype`` | EtherType |
2194 +---------------+-----------+
2196 Action: ``VXLAN_ENCAP``
2197 ^^^^^^^^^^^^^^^^^^^^^^^
2199 Performs a VXLAN encapsulation action by encapsulating the matched flow in the
2200 VXLAN tunnel as defined in the``rte_flow_action_vxlan_encap`` flow items
2203 This action modifies the payload of matched flows. The flow definition specified
2204 in the ``rte_flow_action_tunnel_encap`` action structure must define a valid
2205 VLXAN network overlay which conforms with RFC 7348 (Virtual eXtensible Local
2206 Area Network (VXLAN): A Framework for Overlaying Virtualized Layer 2 Networks
2207 over Layer 3 Networks). The pattern must be terminated with the
2208 RTE_FLOW_ITEM_TYPE_END item type.
2210 .. _table_rte_flow_action_vxlan_encap:
2212 .. table:: VXLAN_ENCAP
2214 +----------------+-------------------------------------+
2216 +================+=====================================+
2217 | ``definition`` | Tunnel end-point overlay definition |
2218 +----------------+-------------------------------------+
2220 .. _table_rte_flow_action_vxlan_encap_example:
2222 .. table:: IPv4 VxLAN flow pattern example.
2224 +-------+----------+
2226 +=======+==========+
2228 +-------+----------+
2230 +-------+----------+
2232 +-------+----------+
2234 +-------+----------+
2236 +-------+----------+
2238 Action: ``VXLAN_DECAP``
2239 ^^^^^^^^^^^^^^^^^^^^^^^
2241 Performs a decapsulation action by stripping all headers of the VXLAN tunnel
2242 network overlay from the matched flow.
2244 The flow items pattern defined for the flow rule with which a ``VXLAN_DECAP``
2245 action is specified, must define a valid VXLAN tunnel as per RFC7348. If the
2246 flow pattern does not specify a valid VXLAN tunnel then a
2247 RTE_FLOW_ERROR_TYPE_ACTION error should be returned.
2249 This action modifies the payload of matched flows.
2251 Action: ``NVGRE_ENCAP``
2252 ^^^^^^^^^^^^^^^^^^^^^^^
2254 Performs a NVGRE encapsulation action by encapsulating the matched flow in the
2255 NVGRE tunnel as defined in the``rte_flow_action_tunnel_encap`` flow item
2258 This action modifies the payload of matched flows. The flow definition specified
2259 in the ``rte_flow_action_tunnel_encap`` action structure must defined a valid
2260 NVGRE network overlay which conforms with RFC 7637 (NVGRE: Network
2261 Virtualization Using Generic Routing Encapsulation). The pattern must be
2262 terminated with the RTE_FLOW_ITEM_TYPE_END item type.
2264 .. _table_rte_flow_action_nvgre_encap:
2266 .. table:: NVGRE_ENCAP
2268 +----------------+-------------------------------------+
2270 +================+=====================================+
2271 | ``definition`` | NVGRE end-point overlay definition |
2272 +----------------+-------------------------------------+
2274 .. _table_rte_flow_action_nvgre_encap_example:
2276 .. table:: IPv4 NVGRE flow pattern example.
2278 +-------+----------+
2280 +=======+==========+
2282 +-------+----------+
2284 +-------+----------+
2286 +-------+----------+
2288 +-------+----------+
2290 Action: ``NVGRE_DECAP``
2291 ^^^^^^^^^^^^^^^^^^^^^^^
2293 Performs a decapsulation action by stripping all headers of the NVGRE tunnel
2294 network overlay from the matched flow.
2296 The flow items pattern defined for the flow rule with which a ``NVGRE_DECAP``
2297 action is specified, must define a valid NVGRE tunnel as per RFC7637. If the
2298 flow pattern does not specify a valid NVGRE tunnel then a
2299 RTE_FLOW_ERROR_TYPE_ACTION error should be returned.
2301 This action modifies the payload of matched flows.
2303 Action: ``RAW_ENCAP``
2304 ^^^^^^^^^^^^^^^^^^^^^
2306 Adds outer header whose template is provided in its data buffer,
2307 as defined in the ``rte_flow_action_raw_encap`` definition.
2309 This action modifies the payload of matched flows. The data supplied must
2310 be a valid header, either holding layer 2 data in case of adding layer 2 after
2311 decap layer 3 tunnel (for example MPLSoGRE) or complete tunnel definition
2312 starting from layer 2 and moving to the tunnel item itself. When applied to
2313 the original packet the resulting packet must be a valid packet.
2315 .. _table_rte_flow_action_raw_encap:
2317 .. table:: RAW_ENCAP
2319 +----------------+----------------------------------------+
2321 +================+========================================+
2322 | ``data`` | Encapsulation data |
2323 +----------------+----------------------------------------+
2324 | ``preserve`` | Bit-mask of data to preserve on output |
2325 +----------------+----------------------------------------+
2326 | ``size`` | Size of data and preserve |
2327 +----------------+----------------------------------------+
2329 Action: ``RAW_DECAP``
2330 ^^^^^^^^^^^^^^^^^^^^^^^
2332 Remove outer header whose template is provided in its data buffer,
2333 as defined in the ``rte_flow_action_raw_decap``
2335 This action modifies the payload of matched flows. The data supplied must
2336 be a valid header, either holding layer 2 data in case of removing layer 2
2337 before encapsulation of layer 3 tunnel (for example MPLSoGRE) or complete
2338 tunnel definition starting from layer 2 and moving to the tunnel item itself.
2339 When applied to the original packet the resulting packet must be a
2342 .. _table_rte_flow_action_raw_decap:
2344 .. table:: RAW_DECAP
2346 +----------------+----------------------------------------+
2348 +================+========================================+
2349 | ``data`` | Decapsulation data |
2350 +----------------+----------------------------------------+
2351 | ``size`` | Size of data |
2352 +----------------+----------------------------------------+
2354 Action: ``SET_IPV4_SRC``
2355 ^^^^^^^^^^^^^^^^^^^^^^^^
2357 Set a new IPv4 source address in the outermost IPv4 header.
2359 It must be used with a valid RTE_FLOW_ITEM_TYPE_IPV4 flow pattern item.
2360 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2362 .. _table_rte_flow_action_set_ipv4_src:
2364 .. table:: SET_IPV4_SRC
2366 +-----------------------------------------+
2368 +===============+=========================+
2369 | ``ipv4_addr`` | new IPv4 source address |
2370 +---------------+-------------------------+
2372 Action: ``SET_IPV4_DST``
2373 ^^^^^^^^^^^^^^^^^^^^^^^^
2375 Set a new IPv4 destination address in the outermost IPv4 header.
2377 It must be used with a valid RTE_FLOW_ITEM_TYPE_IPV4 flow pattern item.
2378 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2380 .. _table_rte_flow_action_set_ipv4_dst:
2382 .. table:: SET_IPV4_DST
2384 +---------------+------------------------------+
2386 +===============+==============================+
2387 | ``ipv4_addr`` | new IPv4 destination address |
2388 +---------------+------------------------------+
2390 Action: ``SET_IPV6_SRC``
2391 ^^^^^^^^^^^^^^^^^^^^^^^^
2393 Set a new IPv6 source address in the outermost IPv6 header.
2395 It must be used with a valid RTE_FLOW_ITEM_TYPE_IPV6 flow pattern item.
2396 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2398 .. _table_rte_flow_action_set_ipv6_src:
2400 .. table:: SET_IPV6_SRC
2402 +---------------+-------------------------+
2404 +===============+=========================+
2405 | ``ipv6_addr`` | new IPv6 source address |
2406 +---------------+-------------------------+
2408 Action: ``SET_IPV6_DST``
2409 ^^^^^^^^^^^^^^^^^^^^^^^^
2411 Set a new IPv6 destination address in the outermost IPv6 header.
2413 It must be used with a valid RTE_FLOW_ITEM_TYPE_IPV6 flow pattern item.
2414 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2416 .. _table_rte_flow_action_set_ipv6_dst:
2418 .. table:: SET_IPV6_DST
2420 +---------------+------------------------------+
2422 +===============+==============================+
2423 | ``ipv6_addr`` | new IPv6 destination address |
2424 +---------------+------------------------------+
2426 Action: ``SET_TP_SRC``
2427 ^^^^^^^^^^^^^^^^^^^^^^^^^
2429 Set a new source port number in the outermost TCP/UDP header.
2431 It must be used with a valid RTE_FLOW_ITEM_TYPE_TCP or RTE_FLOW_ITEM_TYPE_UDP
2432 flow pattern item. Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2434 .. _table_rte_flow_action_set_tp_src:
2436 .. table:: SET_TP_SRC
2438 +----------+-------------------------+
2440 +==========+=========================+
2441 | ``port`` | new TCP/UDP source port |
2442 +---------------+--------------------+
2444 Action: ``SET_TP_DST``
2445 ^^^^^^^^^^^^^^^^^^^^^^^^^
2447 Set a new destination port number in the outermost TCP/UDP header.
2449 It must be used with a valid RTE_FLOW_ITEM_TYPE_TCP or RTE_FLOW_ITEM_TYPE_UDP
2450 flow pattern item. Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2452 .. _table_rte_flow_action_set_tp_dst:
2454 .. table:: SET_TP_DST
2456 +----------+------------------------------+
2458 +==========+==============================+
2459 | ``port`` | new TCP/UDP destination port |
2460 +---------------+-------------------------+
2462 Action: ``MAC_SWAP``
2463 ^^^^^^^^^^^^^^^^^^^^^^^^^
2465 Swap the source and destination MAC addresses in the outermost Ethernet
2468 It must be used with a valid RTE_FLOW_ITEM_TYPE_ETH flow pattern item.
2469 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2471 .. _table_rte_flow_action_mac_swap:
2486 If there is no valid RTE_FLOW_ITEM_TYPE_IPV4 or RTE_FLOW_ITEM_TYPE_IPV6
2487 in pattern, Some PMDs will reject rule because behavior will be undefined.
2489 .. _table_rte_flow_action_dec_ttl:
2502 Assigns a new TTL value.
2504 If there is no valid RTE_FLOW_ITEM_TYPE_IPV4 or RTE_FLOW_ITEM_TYPE_IPV6
2505 in pattern, Some PMDs will reject rule because behavior will be undefined.
2507 .. _table_rte_flow_action_set_ttl:
2511 +---------------+--------------------+
2513 +===============+====================+
2514 | ``ttl_value`` | new TTL value |
2515 +---------------+--------------------+
2517 Action: ``SET_MAC_SRC``
2518 ^^^^^^^^^^^^^^^^^^^^^^^
2520 Set source MAC address.
2522 It must be used with a valid RTE_FLOW_ITEM_TYPE_ETH flow pattern item.
2523 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2525 .. _table_rte_flow_action_set_mac_src:
2527 .. table:: SET_MAC_SRC
2529 +--------------+---------------+
2531 +==============+===============+
2532 | ``mac_addr`` | MAC address |
2533 +--------------+---------------+
2535 Action: ``SET_MAC_DST``
2536 ^^^^^^^^^^^^^^^^^^^^^^^
2538 Set destination MAC address.
2540 It must be used with a valid RTE_FLOW_ITEM_TYPE_ETH flow pattern item.
2541 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2543 .. _table_rte_flow_action_set_mac_dst:
2545 .. table:: SET_MAC_DST
2547 +--------------+---------------+
2549 +==============+===============+
2550 | ``mac_addr`` | MAC address |
2551 +--------------+---------------+
2553 Action: ``INC_TCP_SEQ``
2554 ^^^^^^^^^^^^^^^^^^^^^^^
2556 Increase sequence number in the outermost TCP header.
2557 Value to increase TCP sequence number by is a big-endian 32 bit integer.
2559 Using this action on non-matching traffic will result in undefined behavior.
2561 Action: ``DEC_TCP_SEQ``
2562 ^^^^^^^^^^^^^^^^^^^^^^^
2564 Decrease sequence number in the outermost TCP header.
2565 Value to decrease TCP sequence number by is a big-endian 32 bit integer.
2567 Using this action on non-matching traffic will result in undefined behavior.
2569 Action: ``INC_TCP_ACK``
2570 ^^^^^^^^^^^^^^^^^^^^^^^
2572 Increase acknowledgment number in the outermost TCP header.
2573 Value to increase TCP acknowledgment number by is a big-endian 32 bit integer.
2575 Using this action on non-matching traffic will result in undefined behavior.
2577 Action: ``DEC_TCP_ACK``
2578 ^^^^^^^^^^^^^^^^^^^^^^^
2580 Decrease acknowledgment number in the outermost TCP header.
2581 Value to decrease TCP acknowledgment number by is a big-endian 32 bit integer.
2583 Using this action on non-matching traffic will result in undefined behavior.
2590 Tag is a transient data used during flow matching. This is not delivered to
2591 application. Multiple tags are supported by specifying index.
2593 .. _table_rte_flow_action_set_tag:
2597 +-----------+----------------------------+
2599 +===========+============================+
2600 | ``data`` | 32 bit tag value |
2601 +-----------+----------------------------+
2602 | ``mask`` | bit-mask applies to "data" |
2603 +-----------+----------------------------+
2604 | ``index`` | index of tag to set |
2605 +-----------+----------------------------+
2607 Action: ``SET_META``
2608 ^^^^^^^^^^^^^^^^^^^^^^^
2610 Set metadata. Item ``META`` matches metadata.
2612 Metadata set by mbuf metadata field with PKT_TX_DYNF_METADATA flag on egress
2613 will be overridden by this action. On ingress, the metadata will be carried by
2614 ``metadata`` dynamic field of ``rte_mbuf`` which can be accessed by
2615 ``RTE_FLOW_DYNF_METADATA()``. PKT_RX_DYNF_METADATA flag will be set along
2618 The mbuf dynamic field must be registered by calling
2619 ``rte_flow_dynf_metadata_register()`` prior to use ``SET_META`` action.
2621 Altering partial bits is supported with ``mask``. For bits which have never been
2622 set, unpredictable value will be seen depending on driver implementation. For
2623 loopback/hairpin packet, metadata set on Rx/Tx may or may not be propagated to
2624 the other path depending on HW capability.
2626 In hairpin case with Tx explicit flow mode, metadata could (not mandatory) be
2627 used to connect the Rx and Tx flows if it can be propagated from Rx to Tx path.
2629 .. _table_rte_flow_action_set_meta:
2633 +----------+----------------------------+
2635 +==========+============================+
2636 | ``data`` | 32 bit metadata value |
2637 +----------+----------------------------+
2638 | ``mask`` | bit-mask applies to "data" |
2639 +----------+----------------------------+
2641 Action: ``SET_IPV4_DSCP``
2642 ^^^^^^^^^^^^^^^^^^^^^^^^^
2646 Modify DSCP in IPv4 header.
2648 It must be used with RTE_FLOW_ITEM_TYPE_IPV4 in pattern.
2649 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2651 .. _table_rte_flow_action_set_ipv4_dscp:
2653 .. table:: SET_IPV4_DSCP
2655 +-----------+---------------------------------+
2657 +===========+=================================+
2658 | ``dscp`` | DSCP in low 6 bits, rest ignore |
2659 +-----------+---------------------------------+
2661 Action: ``SET_IPV6_DSCP``
2662 ^^^^^^^^^^^^^^^^^^^^^^^^^
2666 Modify DSCP in IPv6 header.
2668 It must be used with RTE_FLOW_ITEM_TYPE_IPV6 in pattern.
2669 Otherwise, RTE_FLOW_ERROR_TYPE_ACTION error will be returned.
2671 .. _table_rte_flow_action_set_ipv6_dscp:
2673 .. table:: SET_IPV6_DSCP
2675 +-----------+---------------------------------+
2677 +===========+=================================+
2678 | ``dscp`` | DSCP in low 6 bits, rest ignore |
2679 +-----------+---------------------------------+
2684 Set ageing timeout configuration to a flow.
2686 Event RTE_ETH_EVENT_FLOW_AGED will be reported if
2687 timeout passed without any matching on the flow.
2689 .. _table_rte_flow_action_age:
2693 +--------------+---------------------------------+
2695 +==============+=================================+
2696 | ``timeout`` | 24 bits timeout value |
2697 +--------------+---------------------------------+
2698 | ``reserved`` | 8 bits reserved, must be zero |
2699 +--------------+---------------------------------+
2700 | ``context`` | user input flow context |
2701 +--------------+---------------------------------+
2703 Query structure to retrieve ageing status information of a
2704 shared AGE action, or a flow rule using the AGE action:
2706 .. _table_rte_flow_query_age:
2708 .. table:: AGE query
2710 +------------------------------+-----+----------------------------------------+
2711 | Field | I/O | Value |
2712 +==============================+=====+========================================+
2713 | ``aged`` | out | Aging timeout expired |
2714 +------------------------------+-----+----------------------------------------+
2715 | ``sec_since_last_hit_valid`` | out | ``sec_since_last_hit`` value is valid |
2716 +------------------------------+-----+----------------------------------------+
2717 | ``sec_since_last_hit`` | out | Seconds since last traffic hit |
2718 +------------------------------+-----+----------------------------------------+
2723 Adds a sample action to a matched flow.
2725 The matching packets will be duplicated with the specified ``ratio`` and
2726 applied with own set of actions with a fate action, the packets sampled
2727 equals is '1/ratio'. All the packets continue to the target destination.
2729 When the ``ratio`` is set to 1 then the packets will be 100% mirrored.
2730 ``actions`` represent the different set of actions for the sampled or mirrored
2731 packets, and must have a fate action.
2733 .. _table_rte_flow_action_sample:
2737 +--------------+---------------------------------+
2739 +==============+=================================+
2740 | ``ratio`` | 32 bits sample ratio value |
2741 +--------------+---------------------------------+
2742 | ``actions`` | sub-action list for sampling |
2743 +--------------+---------------------------------+
2748 Flow utilize shared action by handle as returned from
2749 ``rte_flow_shared_action_create()``.
2751 The behaviour of the shared action defined by ``action`` argument of type
2752 ``struct rte_flow_action`` passed to ``rte_flow_shared_action_create()``.
2754 Multiple flows can use the same shared action.
2755 The shared action can be in-place updated by ``rte_flow_shared_action_update()``
2756 without destroying flow and creating flow again.
2758 The shared action specified data (e.g. counter) can be queried by
2759 ``rte_flow_shared_action_query()``.
2761 .. _table_rte_flow_shared_action:
2771 Action: ``MODIFY_FIELD``
2772 ^^^^^^^^^^^^^^^^^^^^^^^^
2774 Modify ``dst`` field according to ``op`` selected (set, addition,
2775 subtraction) with ``width`` bits of data from ``src`` field.
2777 Any arbitrary header field (as well as mark, metadata or tag values)
2778 can be used as both source and destination fields as set by ``field``.
2779 The immediate value ``RTE_FLOW_FIELD_VALUE`` (or a pointer to it
2780 ``RTE_FLOW_FIELD_POINTER``) is allowed as a source only.
2781 ``RTE_FLOW_FIELD_START`` is used to point to the beginning of a packet.
2782 See ``enum rte_flow_field_id`` for the list of supported fields.
2784 ``op`` selects the operation to perform on a destination field.
2785 - ``set`` copies the data from ``src`` field to ``dst`` field.
2786 - ``add`` adds together ``dst`` and ``src`` and stores the result into ``dst``.
2787 - ``sub`` subtracts ``src`` from ``dst`` and stores the result into ``dst``
2789 ``width`` defines a number of bits to use from ``src`` field.
2791 ``level`` is used to access any packet field on any encapsulation level
2792 as well as any tag element in the tag array.
2793 - ``0`` means the default behaviour. Depending on the packet type, it can
2794 mean outermost, innermost or anything in between.
2795 - ``1`` requests access to the outermost packet encapsulation level.
2796 - ``2`` and subsequent values requests access to the specified packet
2797 encapsulation level, from outermost to innermost (lower to higher values).
2798 For the tag array (in case of multiple tags are supported and present)
2799 ``level`` translates directly into the array index.
2801 ``offset`` specifies the number of bits to skip from a field's start.
2802 That allows performing a partial copy of the needed part or to divide a big
2803 packet field into multiple smaller fields. Alternatively, ``offset`` allows
2804 going past the specified packet field boundary to copy a field to an
2805 arbitrary place in a packet, essentially providing a way to copy any part of
2806 a packet to any other part of it.
2808 ``value`` sets an immediate value to be used as a source or points to a
2809 location of the value in memory. It is used instead of ``level`` and ``offset``
2810 for ``RTE_FLOW_FIELD_VALUE`` and ``RTE_FLOW_FIELD_POINTER`` respectively.
2812 .. _table_rte_flow_action_modify_field:
2814 .. table:: MODIFY_FIELD
2816 +---------------+-------------------------+
2818 +===============+=========================+
2819 | ``op`` | operation to perform |
2820 +---------------+-------------------------+
2821 | ``dst`` | destination field |
2822 +---------------+-------------------------+
2823 | ``src`` | source field |
2824 +---------------+-------------------------+
2825 | ``width`` | number of bits to use |
2826 +---------------+-------------------------+
2828 .. _table_rte_flow_action_modify_data:
2830 .. table:: destination/source field definition
2832 +---------------+----------------------------------------------------------+
2834 +===============+==========================================================+
2835 | ``field`` | ID: packet field, mark, meta, tag, immediate, pointer |
2836 +---------------+----------------------------------------------------------+
2837 | ``level`` | encapsulation level of a packet field or tag array index |
2838 +---------------+----------------------------------------------------------+
2839 | ``offset`` | number of bits to skip at the beginning |
2840 +---------------+----------------------------------------------------------+
2841 | ``value`` | immediate value or a pointer to this value |
2842 +---------------+----------------------------------------------------------+
2847 All specified pattern items (``enum rte_flow_item_type``) and actions
2848 (``enum rte_flow_action_type``) use positive identifiers.
2850 The negative space is reserved for dynamic types generated by PMDs during
2851 run-time. PMDs may encounter them as a result but must not accept negative
2852 identifiers they are not aware of.
2854 A method to generate them remains to be defined.
2856 Application may use PMD dynamic items or actions in flow rules. In that case
2857 size of configuration object in dynamic element must be a pointer size.
2862 Pattern item types will be added as new protocols are implemented.
2864 Variable headers support through dedicated pattern items, for example in
2865 order to match specific IPv4 options and IPv6 extension headers would be
2866 stacked after IPv4/IPv6 items.
2868 Other action types are planned but are not defined yet. These include the
2869 ability to alter packet data in several ways, such as performing
2870 encapsulation/decapsulation of tunnel headers.
2875 A rather simple API with few functions is provided to fully manage flow
2878 Each created flow rule is associated with an opaque, PMD-specific handle
2879 pointer. The application is responsible for keeping it until the rule is
2882 Flows rules are represented by ``struct rte_flow`` objects.
2887 Given that expressing a definite set of device capabilities is not
2888 practical, a dedicated function is provided to check if a flow rule is
2889 supported and can be created.
2894 rte_flow_validate(uint16_t port_id,
2895 const struct rte_flow_attr *attr,
2896 const struct rte_flow_item pattern[],
2897 const struct rte_flow_action actions[],
2898 struct rte_flow_error *error);
2900 The flow rule is validated for correctness and whether it could be accepted
2901 by the device given sufficient resources. The rule is checked against the
2902 current device mode and queue configuration. The flow rule may also
2903 optionally be validated against existing flow rules and device resources.
2904 This function has no effect on the target device.
2906 The returned value is guaranteed to remain valid only as long as no
2907 successful calls to ``rte_flow_create()`` or ``rte_flow_destroy()`` are made
2908 in the meantime and no device parameter affecting flow rules in any way are
2909 modified, due to possible collisions or resource limitations (although in
2910 such cases ``EINVAL`` should not be returned).
2914 - ``port_id``: port identifier of Ethernet device.
2915 - ``attr``: flow rule attributes.
2916 - ``pattern``: pattern specification (list terminated by the END pattern
2918 - ``actions``: associated actions (list terminated by the END action).
2919 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
2920 this structure in case of error only.
2924 - 0 if flow rule is valid and can be created. A negative errno value
2925 otherwise (``rte_errno`` is also set), the following errors are defined.
2926 - ``-ENOSYS``: underlying device does not support this functionality.
2927 - ``-EINVAL``: unknown or invalid rule specification.
2928 - ``-ENOTSUP``: valid but unsupported rule specification (e.g. partial
2929 bit-masks are unsupported).
2930 - ``EEXIST``: collision with an existing rule. Only returned if device
2931 supports flow rule collision checking and there was a flow rule
2932 collision. Not receiving this return code is no guarantee that creating
2933 the rule will not fail due to a collision.
2934 - ``ENOMEM``: not enough memory to execute the function, or if the device
2935 supports resource validation, resource limitation on the device.
2936 - ``-EBUSY``: action cannot be performed due to busy device resources, may
2937 succeed if the affected queues or even the entire port are in a stopped
2938 state (see ``rte_eth_dev_rx_queue_stop()`` and ``rte_eth_dev_stop()``).
2943 Creating a flow rule is similar to validating one, except the rule is
2944 actually created and a handle returned.
2949 rte_flow_create(uint16_t port_id,
2950 const struct rte_flow_attr *attr,
2951 const struct rte_flow_item pattern[],
2952 const struct rte_flow_action *actions[],
2953 struct rte_flow_error *error);
2957 - ``port_id``: port identifier of Ethernet device.
2958 - ``attr``: flow rule attributes.
2959 - ``pattern``: pattern specification (list terminated by the END pattern
2961 - ``actions``: associated actions (list terminated by the END action).
2962 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
2963 this structure in case of error only.
2967 A valid handle in case of success, NULL otherwise and ``rte_errno`` is set
2968 to the positive version of one of the error codes defined for
2969 ``rte_flow_validate()``.
2974 Flow rules destruction is not automatic, and a queue or a port should not be
2975 released if any are still attached to them. Applications must take care of
2976 performing this step before releasing resources.
2981 rte_flow_destroy(uint16_t port_id,
2982 struct rte_flow *flow,
2983 struct rte_flow_error *error);
2986 Failure to destroy a flow rule handle may occur when other flow rules depend
2987 on it, and destroying it would result in an inconsistent state.
2989 This function is only guaranteed to succeed if handles are destroyed in
2990 reverse order of their creation.
2994 - ``port_id``: port identifier of Ethernet device.
2995 - ``flow``: flow rule handle to destroy.
2996 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
2997 this structure in case of error only.
3001 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
3006 Convenience function to destroy all flow rule handles associated with a
3007 port. They are released as with successive calls to ``rte_flow_destroy()``.
3012 rte_flow_flush(uint16_t port_id,
3013 struct rte_flow_error *error);
3015 In the unlikely event of failure, handles are still considered destroyed and
3016 no longer valid but the port must be assumed to be in an inconsistent state.
3020 - ``port_id``: port identifier of Ethernet device.
3021 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
3022 this structure in case of error only.
3026 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
3031 Query an existing flow rule.
3033 This function allows retrieving flow-specific data such as counters. Data
3034 is gathered by special actions which must be present in the flow rule
3040 rte_flow_query(uint16_t port_id,
3041 struct rte_flow *flow,
3042 const struct rte_flow_action *action,
3044 struct rte_flow_error *error);
3048 - ``port_id``: port identifier of Ethernet device.
3049 - ``flow``: flow rule handle to query.
3050 - ``action``: action to query, this must match prototype from flow rule.
3051 - ``data``: pointer to storage for the associated query data type.
3052 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
3053 this structure in case of error only.
3057 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
3059 .. _flow_isolated_mode:
3064 The general expectation for ingress traffic is that flow rules process it
3065 first; the remaining unmatched or pass-through traffic usually ends up in a
3066 queue (with or without RSS, locally or in some sub-device instance)
3067 depending on the global configuration settings of a port.
3069 While fine from a compatibility standpoint, this approach makes drivers more
3070 complex as they have to check for possible side effects outside of this API
3071 when creating or destroying flow rules. It results in a more limited set of
3072 available rule types due to the way device resources are assigned (e.g. no
3073 support for the RSS action even on capable hardware).
3075 Given that nonspecific traffic can be handled by flow rules as well,
3076 isolated mode is a means for applications to tell a driver that ingress on
3077 the underlying port must be injected from the defined flow rules only; that
3078 no default traffic is expected outside those rules.
3080 This has the following benefits:
3082 - Applications get finer-grained control over the kind of traffic they want
3083 to receive (no traffic by default).
3085 - More importantly they control at what point nonspecific traffic is handled
3086 relative to other flow rules, by adjusting priority levels.
3088 - Drivers can assign more hardware resources to flow rules and expand the
3089 set of supported rule types.
3091 Because toggling isolated mode may cause profound changes to the ingress
3092 processing path of a driver, it may not be possible to leave it once
3093 entered. Likewise, existing flow rules or global configuration settings may
3094 prevent a driver from entering isolated mode.
3096 Applications relying on this mode are therefore encouraged to toggle it as
3097 soon as possible after device initialization, ideally before the first call
3098 to ``rte_eth_dev_configure()`` to avoid possible failures due to conflicting
3101 Once effective, the following functionality has no effect on the underlying
3102 port and may return errors such as ``ENOTSUP`` ("not supported"):
3104 - Toggling promiscuous mode.
3105 - Toggling allmulticast mode.
3106 - Configuring MAC addresses.
3107 - Configuring multicast addresses.
3108 - Configuring VLAN filters.
3109 - Configuring global RSS settings.
3114 rte_flow_isolate(uint16_t port_id, int set, struct rte_flow_error *error);
3118 - ``port_id``: port identifier of Ethernet device.
3119 - ``set``: nonzero to enter isolated mode, attempt to leave it otherwise.
3120 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
3121 this structure in case of error only.
3125 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
3127 Verbose error reporting
3128 -----------------------
3130 The defined *errno* values may not be accurate enough for users or
3131 application developers who want to investigate issues related to flow rules
3132 management. A dedicated error object is defined for this purpose:
3136 enum rte_flow_error_type {
3137 RTE_FLOW_ERROR_TYPE_NONE, /**< No error. */
3138 RTE_FLOW_ERROR_TYPE_UNSPECIFIED, /**< Cause unspecified. */
3139 RTE_FLOW_ERROR_TYPE_HANDLE, /**< Flow rule (handle). */
3140 RTE_FLOW_ERROR_TYPE_ATTR_GROUP, /**< Group field. */
3141 RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY, /**< Priority field. */
3142 RTE_FLOW_ERROR_TYPE_ATTR_INGRESS, /**< Ingress field. */
3143 RTE_FLOW_ERROR_TYPE_ATTR_EGRESS, /**< Egress field. */
3144 RTE_FLOW_ERROR_TYPE_ATTR, /**< Attributes structure. */
3145 RTE_FLOW_ERROR_TYPE_ITEM_NUM, /**< Pattern length. */
3146 RTE_FLOW_ERROR_TYPE_ITEM, /**< Specific pattern item. */
3147 RTE_FLOW_ERROR_TYPE_ACTION_NUM, /**< Number of actions. */
3148 RTE_FLOW_ERROR_TYPE_ACTION, /**< Specific action. */
3151 struct rte_flow_error {
3152 enum rte_flow_error_type type; /**< Cause field and error types. */
3153 const void *cause; /**< Object responsible for the error. */
3154 const char *message; /**< Human-readable error message. */
3157 Error type ``RTE_FLOW_ERROR_TYPE_NONE`` stands for no error, in which case
3158 remaining fields can be ignored. Other error types describe the type of the
3159 object pointed by ``cause``.
3161 If non-NULL, ``cause`` points to the object responsible for the error. For a
3162 flow rule, this may be a pattern item or an individual action.
3164 If non-NULL, ``message`` provides a human-readable error message.
3166 This object is normally allocated by applications and set by PMDs in case of
3167 error, the message points to a constant string which does not need to be
3168 freed by the application, however its pointer can be considered valid only
3169 as long as its associated DPDK port remains configured. Closing the
3170 underlying device or unloading the PMD invalidates it.
3181 rte_flow_error_set(struct rte_flow_error *error,
3183 enum rte_flow_error_type type,
3185 const char *message);
3187 This function initializes ``error`` (if non-NULL) with the provided
3188 parameters and sets ``rte_errno`` to ``code``. A negative error ``code`` is
3197 rte_flow_conv(enum rte_flow_conv_op op,
3201 struct rte_flow_error *error);
3203 Convert ``src`` to ``dst`` according to operation ``op``. Possible
3206 - Attributes, pattern item or action duplication.
3207 - Duplication of an entire pattern or list of actions.
3208 - Duplication of a complete flow rule description.
3209 - Pattern item or action name retrieval.
3211 Tunneled traffic offload
3212 ~~~~~~~~~~~~~~~~~~~~~~~~
3214 rte_flow API provides the building blocks for vendor-agnostic flow
3215 classification offloads. The rte_flow "patterns" and "actions"
3216 primitives are fine-grained, thus enabling DPDK applications the
3217 flexibility to offload network stacks and complex pipelines.
3218 Applications wishing to offload tunneled traffic are required to use
3219 the rte_flow primitives, such as group, meta, mark, tag, and others to
3220 model their high-level objects. The hardware model design for
3221 high-level software objects is not trivial. Furthermore, an optimal
3222 design is often vendor-specific.
3224 When hardware offloads tunneled traffic in multi-group logic,
3225 partially offloaded packets may arrive to the application after they
3226 were modified in hardware. In this case, the application may need to
3227 restore the original packet headers. Consider the following sequence:
3228 The application decaps a packet in one group and jumps to a second
3229 group where it tries to match on a 5-tuple, that will miss and send
3230 the packet to the application. In this case, the application does not
3231 receive the original packet but a modified one. Also, in this case,
3232 the application cannot match on the outer header fields, such as VXLAN
3235 There are several possible ways to use rte_flow "patterns" and
3236 "actions" to resolve the issues above. For example:
3238 1 Mapping headers to a hardware registers using the
3239 rte_flow_action_mark/rte_flow_action_tag/rte_flow_set_meta objects.
3241 2 Apply the decap only at the last offload stage after all the
3242 "patterns" were matched and the packet will be fully offloaded.
3244 Every approach has its pros and cons and is highly dependent on the
3245 hardware vendor. For example, some hardware may have a limited number
3246 of registers while other hardware could not support inner actions and
3247 must decap before accessing inner headers.
3249 The tunnel offload model resolves these issues. The model goals are:
3251 1 Provide a unified application API to offload tunneled traffic that
3252 is capable to match on outer headers after decap.
3254 2 Allow the application to restore the outer header of partially
3257 The tunnel offload model does not introduce new elements to the
3258 existing RTE flow model and is implemented as a set of helper
3261 For the application to work with the tunnel offload API it
3262 has to adjust flow rules in multi-table tunnel offload in the
3265 1 Remove explicit call to decap action and replace it with PMD actions
3266 obtained from rte_flow_tunnel_decap_and_set() helper.
3268 2 Add PMD items obtained from rte_flow_tunnel_match() helper to all
3269 other rules in the tunnel offload sequence.
3271 The model requirements:
3273 Software application must initialize
3274 rte_tunnel object with tunnel parameters before calling
3275 rte_flow_tunnel_decap_set() & rte_flow_tunnel_match().
3277 PMD actions array obtained in rte_flow_tunnel_decap_set() must be
3278 released by application with rte_flow_action_release() call.
3280 PMD items array obtained with rte_flow_tunnel_match() must be released
3281 by application with rte_flow_item_release() call. Application can
3282 release PMD items and actions after rule was created. However, if the
3283 application needs to create additional rule for the same tunnel it
3284 will need to obtain PMD items again.
3286 Application cannot destroy rte_tunnel object before it releases all
3287 PMD actions & PMD items referencing that tunnel.
3292 - DPDK does not keep track of flow rules definitions or flow rule objects
3293 automatically. Applications may keep track of the former and must keep
3294 track of the latter. PMDs may also do it for internal needs, however this
3295 must not be relied on by applications.
3297 - Flow rules are not maintained between successive port initializations. An
3298 application exiting without releasing them and restarting must re-create
3301 - API operations are synchronous and blocking (``EAGAIN`` cannot be
3304 - Stopping the data path (TX/RX) should not be necessary when managing flow
3305 rules. If this cannot be achieved naturally or with workarounds (such as
3306 temporarily replacing the burst function pointers), an appropriate error
3307 code must be returned (``EBUSY``).
3309 - Applications, not PMDs, are responsible for maintaining flow rules
3310 configuration when closing, stopping or restarting a port or performing other
3311 actions which may affect them.
3312 Applications must assume that after port close, stop or restart all flows
3313 related to that port are not valid, hardware rules are destroyed and relevant
3314 PMD resources are released.
3316 For devices exposing multiple ports sharing global settings affected by flow
3319 - All ports under DPDK control must behave consistently, PMDs are
3320 responsible for making sure that existing flow rules on a port are not
3321 affected by other ports.
3323 - Ports not under DPDK control (unaffected or handled by other applications)
3324 are user's responsibility. They may affect existing flow rules and cause
3325 undefined behavior. PMDs aware of this may prevent flow rules creation
3326 altogether in such cases.
3331 The PMD interface is defined in ``rte_flow_driver.h``. It is not subject to
3332 API/ABI versioning constraints as it is not exposed to applications and may
3333 evolve independently.
3335 The PMD interface is based on callbacks pointed by the ``struct rte_flow_ops``.
3337 - PMD callbacks implement exactly the interface described in `Rules
3338 management`_, except for the port ID argument which has already been
3339 converted to a pointer to the underlying ``struct rte_eth_dev``.
3341 - Public API functions do not process flow rules definitions at all before
3342 calling PMD functions (no basic error checking, no validation
3343 whatsoever). They only make sure these callbacks are non-NULL or return
3344 the ``ENOSYS`` (function not supported) error.
3346 This interface additionally defines the following helper function:
3348 - ``rte_flow_ops_get()``: get generic flow operations structure from a
3351 If PMD interfaces don't support re-entrancy/multi-thread safety,
3352 the rte_flow API functions will protect threads by mutex per port.
3353 The application can check whether ``RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE``
3354 is set in ``dev_flags``, meaning the PMD is thread-safe regarding rte_flow,
3355 so the API level protection is disabled.
3356 Please note that this API-level mutex protects only rte_flow functions,
3357 other control path functions are not in scope.
3359 More will be added over time.
3361 Device compatibility
3362 --------------------
3364 No known implementation supports all the described features.
3366 Unsupported features or combinations are not expected to be fully emulated
3367 in software by PMDs for performance reasons. Partially supported features
3368 may be completed in software as long as hardware performs most of the work
3369 (such as queue redirection and packet recognition).
3371 However PMDs are expected to do their best to satisfy application requests
3372 by working around hardware limitations as long as doing so does not affect
3373 the behavior of existing flow rules.
3375 The following sections provide a few examples of such cases and describe how
3376 PMDs should handle them, they are based on limitations built into the
3382 Each flow rule comes with its own, per-layer bit-masks, while hardware may
3383 support only a single, device-wide bit-mask for a given layer type, so that
3384 two IPv4 rules cannot use different bit-masks.
3386 The expected behavior in this case is that PMDs automatically configure
3387 global bit-masks according to the needs of the first flow rule created.
3389 Subsequent rules are allowed only if their bit-masks match those, the
3390 ``EEXIST`` error code should be returned otherwise.
3392 Unsupported layer types
3393 ~~~~~~~~~~~~~~~~~~~~~~~
3395 Many protocols can be simulated by crafting patterns with the `Item: RAW`_
3398 PMDs can rely on this capability to simulate support for protocols with
3399 headers not directly recognized by hardware.
3401 ``ANY`` pattern item
3402 ~~~~~~~~~~~~~~~~~~~~
3404 This pattern item stands for anything, which can be difficult to translate
3405 to something hardware would understand, particularly if followed by more
3408 Consider the following pattern:
3410 .. _table_rte_flow_unsupported_any:
3412 .. table:: Pattern with ANY as L3
3414 +-------+-----------------------+
3416 +=======+=======================+
3418 +-------+-----+---------+-------+
3419 | 1 | ANY | ``num`` | ``1`` |
3420 +-------+-----+---------+-------+
3422 +-------+-----------------------+
3424 +-------+-----------------------+
3426 Knowing that TCP does not make sense with something other than IPv4 and IPv6
3427 as L3, such a pattern may be translated to two flow rules instead:
3429 .. _table_rte_flow_unsupported_any_ipv4:
3431 .. table:: ANY replaced with IPV4
3433 +-------+--------------------+
3435 +=======+====================+
3437 +-------+--------------------+
3438 | 1 | IPV4 (zeroed mask) |
3439 +-------+--------------------+
3441 +-------+--------------------+
3443 +-------+--------------------+
3447 .. _table_rte_flow_unsupported_any_ipv6:
3449 .. table:: ANY replaced with IPV6
3451 +-------+--------------------+
3453 +=======+====================+
3455 +-------+--------------------+
3456 | 1 | IPV6 (zeroed mask) |
3457 +-------+--------------------+
3459 +-------+--------------------+
3461 +-------+--------------------+
3463 Note that as soon as a ANY rule covers several layers, this approach may
3464 yield a large number of hidden flow rules. It is thus suggested to only
3465 support the most common scenarios (anything as L2 and/or L3).
3470 - When combined with `Action: QUEUE`_, packet counting (`Action: COUNT`_)
3471 and tagging (`Action: MARK`_ or `Action: FLAG`_) may be implemented in
3472 software as long as the target queue is used by a single rule.
3474 - When a single target queue is provided, `Action: RSS`_ can also be
3475 implemented through `Action: QUEUE`_.
3480 While it would naturally make sense, flow rules cannot be assumed to be
3481 processed by hardware in the same order as their creation for several
3484 - They may be managed internally as a tree or a hash table instead of a
3486 - Removing a flow rule before adding another one can either put the new rule
3487 at the end of the list or reuse a freed entry.
3488 - Duplication may occur when packets are matched by several rules.
3490 For overlapping rules (particularly in order to use `Action: PASSTHRU`_)
3491 predictable behavior is only guaranteed by using different priority levels.
3493 Priority levels are not necessarily implemented in hardware, or may be
3494 severely limited (e.g. a single priority bit).
3496 For these reasons, priority levels may be implemented purely in software by
3499 - For devices expecting flow rules to be added in the correct order, PMDs
3500 may destroy and re-create existing rules after adding a new one with
3503 - A configurable number of dummy or empty rules can be created at
3504 initialization time to save high priority slots for later.
3506 - In order to save priority levels, PMDs may evaluate whether rules are
3507 likely to collide and adjust their priority accordingly.
3512 - A device profile selection function which could be used to force a
3513 permanent profile instead of relying on its automatic configuration based
3514 on existing flow rules.
3516 - A method to optimize *rte_flow* rules with specific pattern items and
3517 action types generated on the fly by PMDs. DPDK should assign negative
3518 numbers to these in order to not collide with the existing types. See
3521 - Adding specific egress pattern items and actions as described in
3522 `Attribute: Traffic direction`_.
3524 - Optional software fallback when PMDs are unable to handle requested flow
3525 rules so applications do not have to implement their own.
3527 .. _OpenFlow Switch Specification: https://www.opennetworking.org/software-defined-standards/specifications/