2 Copyright 2016 6WIND S.A.
3 Copyright 2016 Mellanox.
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
15 * Neither the name of 6WIND S.A. nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
33 Generic flow API (rte_flow)
34 ===========================
39 This API provides a generic means to configure hardware to match specific
40 ingress or egress traffic, alter its fate and query related counters
41 according to any number of user-defined rules.
43 It is named *rte_flow* after the prefix used for all its symbols, and is
44 defined in ``rte_flow.h``.
46 - Matching can be performed on packet data (protocol headers, payload) and
47 properties (e.g. associated physical port, virtual device function ID).
49 - Possible operations include dropping traffic, diverting it to specific
50 queues, to virtual/physical device functions or ports, performing tunnel
51 offloads, adding marks and so on.
53 It is slightly higher-level than the legacy filtering framework which it
54 encompasses and supersedes (including all functions and filter types) in
55 order to expose a single interface with an unambiguous behavior that is
56 common to all poll-mode drivers (PMDs).
58 Several methods to migrate existing applications are described in `API
67 A flow rule is the combination of attributes with a matching pattern and a
68 list of actions. Flow rules form the basis of this API.
70 Flow rules can have several distinct actions (such as counting,
71 encapsulating, decapsulating before redirecting packets to a particular
72 queue, etc.), instead of relying on several rules to achieve this and having
73 applications deal with hardware implementation details regarding their
76 Support for different priority levels on a rule basis is provided, for
77 example in order to force a more specific rule to come before a more generic
78 one for packets matched by both. However hardware support for more than a
79 single priority level cannot be guaranteed. When supported, the number of
80 available priority levels is usually low, which is why they can also be
81 implemented in software by PMDs (e.g. missing priority levels may be
82 emulated by reordering rules).
84 In order to remain as hardware-agnostic as possible, by default all rules
85 are considered to have the same priority, which means that the order between
86 overlapping rules (when a packet is matched by several filters) is
89 PMDs may refuse to create overlapping rules at a given priority level when
90 they can be detected (e.g. if a pattern matches an existing filter).
92 Thus predictable results for a given priority level can only be achieved
93 with non-overlapping rules, using perfect matching on all protocol layers.
95 Flow rules can also be grouped, the flow rule priority is specific to the
96 group they belong to. All flow rules in a given group are thus processed
97 either before or after another group.
99 Support for multiple actions per rule may be implemented internally on top
100 of non-default hardware priorities, as a result both features may not be
101 simultaneously available to applications.
103 Considering that allowed pattern/actions combinations cannot be known in
104 advance and would result in an impractically large number of capabilities to
105 expose, a method is provided to validate a given rule from the current
106 device configuration state.
108 This enables applications to check if the rule types they need is supported
109 at initialization time, before starting their data path. This method can be
110 used anytime, its only requirement being that the resources needed by a rule
111 should exist (e.g. a target RX queue should be configured first).
113 Each defined rule is associated with an opaque handle managed by the PMD,
114 applications are responsible for keeping it. These can be used for queries
115 and rules management, such as retrieving counters or other data and
118 To avoid resource leaks on the PMD side, handles must be explicitly
119 destroyed by the application before releasing associated resources such as
122 The following sections cover:
124 - **Attributes** (represented by ``struct rte_flow_attr``): properties of a
125 flow rule such as its direction (ingress or egress) and priority.
127 - **Pattern item** (represented by ``struct rte_flow_item``): part of a
128 matching pattern that either matches specific packet data or traffic
129 properties. It can also describe properties of the pattern itself, such as
132 - **Matching pattern**: traffic properties to look for, a combination of any
135 - **Actions** (represented by ``struct rte_flow_action``): operations to
136 perform whenever a packet is matched by a pattern.
144 Flow rules can be grouped by assigning them a common group number. Lower
145 values have higher priority. Group 0 has the highest priority.
147 Although optional, applications are encouraged to group similar rules as
148 much as possible to fully take advantage of hardware capabilities
149 (e.g. optimized matching) and work around limitations (e.g. a single pattern
150 type possibly allowed in a given group).
152 Note that support for more than a single group is not guaranteed.
157 A priority level can be assigned to a flow rule. Like groups, lower values
158 denote higher priority, with 0 as the maximum.
160 A rule with priority 0 in group 8 is always matched after a rule with
161 priority 8 in group 0.
163 Group and priority levels are arbitrary and up to the application, they do
164 not need to be contiguous nor start from 0, however the maximum number
165 varies between devices and may be affected by existing flow rules.
167 If a packet is matched by several rules of a given group for a given
168 priority level, the outcome is undefined. It can take any path, may be
169 duplicated or even cause unrecoverable errors.
171 Note that support for more than a single priority level is not guaranteed.
173 Attribute: Traffic direction
174 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
176 Flow rules can apply to inbound and/or outbound traffic (ingress/egress).
178 Several pattern items and actions are valid and can be used in both
179 directions. At least one direction must be specified.
181 Specifying both directions at once for a given rule is not recommended but
182 may be valid in a few cases (e.g. shared counters).
187 Pattern items fall in two categories:
189 - Matching protocol headers and packet data, usually associated with a
190 specification structure. These must be stacked in the same order as the
191 protocol layers to match inside packets, starting from the lowest.
193 - Matching meta-data or affecting pattern processing, often without a
194 specification structure. Since they do not match packet contents, their
195 position in the list is usually not relevant.
197 Item specification structures are used to match specific values among
198 protocol fields (or item properties). Documentation describes for each item
199 whether they are associated with one and their type name if so.
201 Up to three structures of the same type can be set for a given item:
203 - ``spec``: values to match (e.g. a given IPv4 address).
205 - ``last``: upper bound for an inclusive range with corresponding fields in
208 - ``mask``: bit-mask applied to both ``spec`` and ``last`` whose purpose is
209 to distinguish the values to take into account and/or partially mask them
210 out (e.g. in order to match an IPv4 address prefix).
212 Usage restrictions and expected behavior:
214 - Setting either ``mask`` or ``last`` without ``spec`` is an error.
216 - Field values in ``last`` which are either 0 or equal to the corresponding
217 values in ``spec`` are ignored; they do not generate a range. Nonzero
218 values lower than those in ``spec`` are not supported.
220 - Setting ``spec`` and optionally ``last`` without ``mask`` causes the PMD
221 to use the default mask defined for that item (defined as
222 ``rte_flow_item_{name}_mask`` constants).
224 - Not setting any of them (assuming item type allows it) is equivalent to
225 providing an empty (zeroed) ``mask`` for broad (nonspecific) matching.
227 - ``mask`` is a simple bit-mask applied before interpreting the contents of
228 ``spec`` and ``last``, which may yield unexpected results if not used
229 carefully. For example, if for an IPv4 address field, ``spec`` provides
230 *10.1.2.3*, ``last`` provides *10.3.4.5* and ``mask`` provides
231 *255.255.0.0*, the effective range becomes *10.1.0.0* to *10.3.255.255*.
233 Example of an item specification matching an Ethernet header:
235 .. _table_rte_flow_pattern_item_example:
237 .. table:: Ethernet item
239 +----------+----------+--------------------+
240 | Field | Subfield | Value |
241 +==========+==========+====================+
242 | ``spec`` | ``src`` | ``00:01:02:03:04`` |
243 | +----------+--------------------+
244 | | ``dst`` | ``00:2a:66:00:01`` |
245 | +----------+--------------------+
246 | | ``type`` | ``0x22aa`` |
247 +----------+----------+--------------------+
248 | ``last`` | unspecified |
249 +----------+----------+--------------------+
250 | ``mask`` | ``src`` | ``00:ff:ff:ff:00`` |
251 | +----------+--------------------+
252 | | ``dst`` | ``00:00:00:00:ff`` |
253 | +----------+--------------------+
254 | | ``type`` | ``0x0000`` |
255 +----------+----------+--------------------+
257 Non-masked bits stand for any value (shown as ``?`` below), Ethernet headers
258 with the following properties are thus matched:
260 - ``src``: ``??:01:02:03:??``
261 - ``dst``: ``??:??:??:??:01``
262 - ``type``: ``0x????``
267 A pattern is formed by stacking items starting from the lowest protocol
268 layer to match. This stacking restriction does not apply to meta items which
269 can be placed anywhere in the stack without affecting the meaning of the
272 Patterns are terminated by END items.
276 .. _table_rte_flow_tcpv4_as_l4:
278 .. table:: TCPv4 as L4
294 .. _table_rte_flow_tcpv6_in_vxlan:
296 .. table:: TCPv6 in VXLAN
298 +-------+------------+
300 +=======+============+
302 +-------+------------+
304 +-------+------------+
306 +-------+------------+
308 +-------+------------+
310 +-------+------------+
312 +-------+------------+
314 +-------+------------+
316 +-------+------------+
320 .. _table_rte_flow_tcpv4_as_l4_meta:
322 .. table:: TCPv4 as L4 with meta items
344 The above example shows how meta items do not affect packet data matching
345 items, as long as those remain stacked properly. The resulting matching
346 pattern is identical to "TCPv4 as L4".
348 .. _table_rte_flow_udpv6_anywhere:
350 .. table:: UDPv6 anywhere
362 If supported by the PMD, omitting one or several protocol layers at the
363 bottom of the stack as in the above example (missing an Ethernet
364 specification) enables looking up anywhere in packets.
366 It is unspecified whether the payload of supported encapsulations
367 (e.g. VXLAN payload) is matched by such a pattern, which may apply to inner,
368 outer or both packets.
370 .. _table_rte_flow_invalid_l3:
372 .. table:: Invalid, missing L3
384 The above pattern is invalid due to a missing L3 specification between L2
385 (Ethernet) and L4 (UDP). Doing so is only allowed at the bottom and at the
391 They match meta-data or affect pattern processing instead of matching packet
392 data directly, most of them do not need a specification structure. This
393 particularity allows them to be specified anywhere in the stack without
394 causing any side effect.
399 End marker for item lists. Prevents further processing of items, thereby
402 - Its numeric value is 0 for convenience.
403 - PMD support is mandatory.
404 - ``spec``, ``last`` and ``mask`` are ignored.
406 .. _table_rte_flow_item_end:
410 +----------+---------+
412 +==========+=========+
413 | ``spec`` | ignored |
414 +----------+---------+
415 | ``last`` | ignored |
416 +----------+---------+
417 | ``mask`` | ignored |
418 +----------+---------+
423 Used as a placeholder for convenience. It is ignored and simply discarded by
426 - PMD support is mandatory.
427 - ``spec``, ``last`` and ``mask`` are ignored.
429 .. _table_rte_flow_item_void:
433 +----------+---------+
435 +==========+=========+
436 | ``spec`` | ignored |
437 +----------+---------+
438 | ``last`` | ignored |
439 +----------+---------+
440 | ``mask`` | ignored |
441 +----------+---------+
443 One usage example for this type is generating rules that share a common
444 prefix quickly without reallocating memory, only by updating item types:
446 .. _table_rte_flow_item_void_example:
448 .. table:: TCP, UDP or ICMP as L4
450 +-------+--------------------+
452 +=======+====================+
454 +-------+--------------------+
456 +-------+------+------+------+
457 | 2 | UDP | VOID | VOID |
458 +-------+------+------+------+
459 | 3 | VOID | TCP | VOID |
460 +-------+------+------+------+
461 | 4 | VOID | VOID | ICMP |
462 +-------+------+------+------+
464 +-------+--------------------+
469 Inverted matching, i.e. process packets that do not match the pattern.
471 - ``spec``, ``last`` and ``mask`` are ignored.
473 .. _table_rte_flow_item_invert:
477 +----------+---------+
479 +==========+=========+
480 | ``spec`` | ignored |
481 +----------+---------+
482 | ``last`` | ignored |
483 +----------+---------+
484 | ``mask`` | ignored |
485 +----------+---------+
487 Usage example, matching non-TCPv4 packets only:
489 .. _table_rte_flow_item_invert_example:
491 .. table:: Anything but TCPv4
510 Matches packets addressed to the physical function of the device.
512 If the underlying device function differs from the one that would normally
513 receive the matched traffic, specifying this item prevents it from reaching
514 that device unless the flow rule contains a `Action: PF`_. Packets are not
515 duplicated between device instances by default.
517 - Likely to return an error or never match any traffic if applied to a VF
519 - Can be combined with any number of `Item: VF`_ to match both PF and VF
521 - ``spec``, ``last`` and ``mask`` must not be set.
523 .. _table_rte_flow_item_pf:
540 Matches packets addressed to a virtual function ID of the device.
542 If the underlying device function differs from the one that would normally
543 receive the matched traffic, specifying this item prevents it from reaching
544 that device unless the flow rule contains a `Action: VF`_. Packets are not
545 duplicated between device instances by default.
547 - Likely to return an error or never match any traffic if this causes a VF
548 device to match traffic addressed to a different VF.
549 - Can be specified multiple times to match traffic addressed to several VF
551 - Can be combined with a PF item to match both PF and VF traffic.
552 - Default ``mask`` matches any VF ID.
554 .. _table_rte_flow_item_vf:
558 +----------+----------+---------------------------+
559 | Field | Subfield | Value |
560 +==========+==========+===========================+
561 | ``spec`` | ``id`` | destination VF ID |
562 +----------+----------+---------------------------+
563 | ``last`` | ``id`` | upper range value |
564 +----------+----------+---------------------------+
565 | ``mask`` | ``id`` | zeroed to match any VF ID |
566 +----------+----------+---------------------------+
571 Matches packets coming from the specified physical port of the underlying
574 The first PORT item overrides the physical port normally associated with the
575 specified DPDK input port (port_id). This item can be provided several times
576 to match additional physical ports.
578 Note that physical ports are not necessarily tied to DPDK input ports
579 (port_id) when those are not under DPDK control. Possible values are
580 specific to each device, they are not necessarily indexed from zero and may
583 As a device property, the list of allowed values as well as the value
584 associated with a port_id should be retrieved by other means.
586 - Default ``mask`` matches any port index.
588 .. _table_rte_flow_item_port:
592 +----------+-----------+--------------------------------+
593 | Field | Subfield | Value |
594 +==========+===========+================================+
595 | ``spec`` | ``index`` | physical port index |
596 +----------+-----------+--------------------------------+
597 | ``last`` | ``index`` | upper range value |
598 +----------+-----------+--------------------------------+
599 | ``mask`` | ``index`` | zeroed to match any port index |
600 +----------+-----------+--------------------------------+
602 Data matching item types
603 ~~~~~~~~~~~~~~~~~~~~~~~~
605 Most of these are basically protocol header definitions with associated
606 bit-masks. They must be specified (stacked) from lowest to highest protocol
607 layer to form a matching pattern.
609 The following list is not exhaustive, new protocols will be added in the
615 Matches any protocol in place of the current layer, a single ANY may also
616 stand for several protocol layers.
618 This is usually specified as the first pattern item when looking for a
619 protocol anywhere in a packet.
621 - Default ``mask`` stands for any number of layers.
623 .. _table_rte_flow_item_any:
627 +----------+----------+--------------------------------------+
628 | Field | Subfield | Value |
629 +==========+==========+======================================+
630 | ``spec`` | ``num`` | number of layers covered |
631 +----------+----------+--------------------------------------+
632 | ``last`` | ``num`` | upper range value |
633 +----------+----------+--------------------------------------+
634 | ``mask`` | ``num`` | zeroed to cover any number of layers |
635 +----------+----------+--------------------------------------+
637 Example for VXLAN TCP payload matching regardless of outer L3 (IPv4 or IPv6)
638 and L4 (UDP) both matched by the first ANY specification, and inner L3 (IPv4
639 or IPv6) matched by the second ANY specification:
641 .. _table_rte_flow_item_any_example:
643 .. table:: TCP in VXLAN with wildcards
645 +-------+------+----------+----------+-------+
646 | Index | Item | Field | Subfield | Value |
647 +=======+======+==========+==========+=======+
649 +-------+------+----------+----------+-------+
650 | 1 | ANY | ``spec`` | ``num`` | 2 |
651 +-------+------+----------+----------+-------+
653 +-------+------------------------------------+
655 +-------+------+----------+----------+-------+
656 | 4 | ANY | ``spec`` | ``num`` | 1 |
657 +-------+------+----------+----------+-------+
659 +-------+------------------------------------+
661 +-------+------------------------------------+
666 Matches a byte string of a given length at a given offset.
668 Offset is either absolute (using the start of the packet) or relative to the
669 end of the previous matched item in the stack, in which case negative values
672 If search is enabled, offset is used as the starting point. The search area
673 can be delimited by setting limit to a nonzero value, which is the maximum
674 number of bytes after offset where the pattern may start.
676 Matching a zero-length pattern is allowed, doing so resets the relative
677 offset for subsequent items.
679 - This type does not support ranges (``last`` field).
680 - Default ``mask`` matches all fields exactly.
682 .. _table_rte_flow_item_raw:
686 +----------+--------------+-------------------------------------------------+
687 | Field | Subfield | Value |
688 +==========+==============+=================================================+
689 | ``spec`` | ``relative`` | look for pattern after the previous item |
690 | +--------------+-------------------------------------------------+
691 | | ``search`` | search pattern from offset (see also ``limit``) |
692 | +--------------+-------------------------------------------------+
693 | | ``reserved`` | reserved, must be set to zero |
694 | +--------------+-------------------------------------------------+
695 | | ``offset`` | absolute or relative offset for ``pattern`` |
696 | +--------------+-------------------------------------------------+
697 | | ``limit`` | search area limit for start of ``pattern`` |
698 | +--------------+-------------------------------------------------+
699 | | ``length`` | ``pattern`` length |
700 | +--------------+-------------------------------------------------+
701 | | ``pattern`` | byte string to look for |
702 +----------+--------------+-------------------------------------------------+
703 | ``last`` | if specified, either all 0 or with the same values as ``spec`` |
704 +----------+----------------------------------------------------------------+
705 | ``mask`` | bit-mask applied to ``spec`` values with usual behavior |
706 +----------+----------------------------------------------------------------+
708 Example pattern looking for several strings at various offsets of a UDP
709 payload, using combined RAW items:
711 .. _table_rte_flow_item_raw_example:
713 .. table:: UDP payload matching
715 +-------+------+----------+--------------+-------+
716 | Index | Item | Field | Subfield | Value |
717 +=======+======+==========+==============+=======+
719 +-------+----------------------------------------+
721 +-------+----------------------------------------+
723 +-------+------+----------+--------------+-------+
724 | 3 | RAW | ``spec`` | ``relative`` | 1 |
725 | | | +--------------+-------+
726 | | | | ``search`` | 1 |
727 | | | +--------------+-------+
728 | | | | ``offset`` | 10 |
729 | | | +--------------+-------+
730 | | | | ``limit`` | 0 |
731 | | | +--------------+-------+
732 | | | | ``length`` | 3 |
733 | | | +--------------+-------+
734 | | | | ``pattern`` | "foo" |
735 +-------+------+----------+--------------+-------+
736 | 4 | RAW | ``spec`` | ``relative`` | 1 |
737 | | | +--------------+-------+
738 | | | | ``search`` | 0 |
739 | | | +--------------+-------+
740 | | | | ``offset`` | 20 |
741 | | | +--------------+-------+
742 | | | | ``limit`` | 0 |
743 | | | +--------------+-------+
744 | | | | ``length`` | 3 |
745 | | | +--------------+-------+
746 | | | | ``pattern`` | "bar" |
747 +-------+------+----------+--------------+-------+
748 | 5 | RAW | ``spec`` | ``relative`` | 1 |
749 | | | +--------------+-------+
750 | | | | ``search`` | 0 |
751 | | | +--------------+-------+
752 | | | | ``offset`` | -29 |
753 | | | +--------------+-------+
754 | | | | ``limit`` | 0 |
755 | | | +--------------+-------+
756 | | | | ``length`` | 3 |
757 | | | +--------------+-------+
758 | | | | ``pattern`` | "baz" |
759 +-------+------+----------+--------------+-------+
761 +-------+----------------------------------------+
765 - Locate "foo" at least 10 bytes deep inside UDP payload.
766 - Locate "bar" after "foo" plus 20 bytes.
767 - Locate "baz" after "bar" minus 29 bytes.
769 Such a packet may be represented as follows (not to scale)::
772 | |<--------->| |<--------->|
774 |-----|------|-----|-----|-----|-----|-----------|-----|------|
775 | ETH | IPv4 | UDP | ... | baz | foo | ......... | bar | .... |
776 |-----|------|-----|-----|-----|-----|-----------|-----|------|
778 |<--------------------------->|
781 Note that matching subsequent pattern items would resume after "baz", not
782 "bar" since matching is always performed after the previous item of the
788 Matches an Ethernet header.
790 - ``dst``: destination MAC.
791 - ``src``: source MAC.
792 - ``type``: EtherType.
793 - Default ``mask`` matches destination and source addresses only.
798 Matches an 802.1Q/ad VLAN tag.
800 - ``tpid``: tag protocol identifier.
801 - ``tci``: tag control information.
802 - Default ``mask`` matches TCI only.
807 Matches an IPv4 header.
809 Note: IPv4 options are handled by dedicated pattern items.
811 - ``hdr``: IPv4 header definition (``rte_ip.h``).
812 - Default ``mask`` matches source and destination addresses only.
817 Matches an IPv6 header.
819 Note: IPv6 options are handled by dedicated pattern items.
821 - ``hdr``: IPv6 header definition (``rte_ip.h``).
822 - Default ``mask`` matches source and destination addresses only.
827 Matches an ICMP header.
829 - ``hdr``: ICMP header definition (``rte_icmp.h``).
830 - Default ``mask`` matches ICMP type and code only.
835 Matches a UDP header.
837 - ``hdr``: UDP header definition (``rte_udp.h``).
838 - Default ``mask`` matches source and destination ports only.
843 Matches a TCP header.
845 - ``hdr``: TCP header definition (``rte_tcp.h``).
846 - Default ``mask`` matches source and destination ports only.
851 Matches a SCTP header.
853 - ``hdr``: SCTP header definition (``rte_sctp.h``).
854 - Default ``mask`` matches source and destination ports only.
859 Matches a VXLAN header (RFC 7348).
861 - ``flags``: normally 0x08 (I flag).
862 - ``rsvd0``: reserved, normally 0x000000.
863 - ``vni``: VXLAN network identifier.
864 - ``rsvd1``: reserved, normally 0x00.
865 - Default ``mask`` matches VNI only.
870 Matches an IEEE 802.1BR E-Tag header.
872 - ``tpid``: tag protocol identifier (0x893F)
873 - ``epcp_edei_in_ecid_b``: E-Tag control information (E-TCI), E-PCP (3b),
874 E-DEI (1b), ingress E-CID base (12b).
875 - ``rsvd_grp_ecid_b``: reserved (2b), GRP (2b), E-CID base (12b).
876 - ``in_ecid_e``: ingress E-CID ext.
877 - ``ecid_e``: E-CID ext.
878 - Default ``mask`` simultaneously matches GRP and E-CID base.
883 Matches a NVGRE header (RFC 7637).
885 - ``c_k_s_rsvd0_ver``: checksum (1b), undefined (1b), key bit (1b),
886 sequence number (1b), reserved 0 (9b), version (3b). This field must have
887 value 0x2000 according to RFC 7637.
888 - ``protocol``: protocol type (0x6558).
889 - ``tni``: virtual subnet ID.
890 - ``flow_id``: flow ID.
891 - Default ``mask`` matches TNI only.
896 Matches a MPLS header.
898 - ``label_tc_s_ttl``: label, TC, Bottom of Stack and TTL.
899 - Default ``mask`` matches label only.
904 Matches a GRE header.
906 - ``c_rsvd0_ver``: checksum, reserved 0 and version.
907 - ``protocol``: protocol type.
908 - Default ``mask`` matches protocol only.
913 Fuzzy pattern match, expect faster than default.
915 This is for device that support fuzzy match option. Usually a fuzzy match is
916 fast but the cost is accuracy. i.e. Signature Match only match pattern's hash
917 value, but it is possible two different patterns have the same hash value.
919 Matching accuracy level can be configured by threshold. Driver can divide the
920 range of threshold and map to different accuracy levels that device support.
922 Threshold 0 means perfect match (no fuzziness), while threshold 0xffffffff
923 means fuzziest match.
925 .. _table_rte_flow_item_fuzzy:
929 +----------+---------------+--------------------------------------------------+
930 | Field | Subfield | Value |
931 +==========+===============+==================================================+
932 | ``spec`` | ``threshold`` | 0 as perfect match, 0xffffffff as fuzziest match |
933 +----------+---------------+--------------------------------------------------+
934 | ``last`` | ``threshold`` | upper range value |
935 +----------+---------------+--------------------------------------------------+
936 | ``mask`` | ``threshold`` | bit-mask apply to "spec" and "last" |
937 +----------+---------------+--------------------------------------------------+
939 Usage example, fuzzy match a TCPv4 packets:
941 .. _table_rte_flow_item_fuzzy_example:
943 .. table:: Fuzzy matching
959 Item: ``GTP``, ``GTPC``, ``GTPU``
960 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
962 Matches a GTPv1 header.
964 Note: GTP, GTPC and GTPU use the same structure. GTPC and GTPU item
965 are defined for a user-friendly API when creating GTP-C and GTP-U
968 - ``v_pt_rsv_flags``: version (3b), protocol type (1b), reserved (1b),
969 extension header flag (1b), sequence number flag (1b), N-PDU number
971 - ``msg_type``: message type.
972 - ``msg_len``: message length.
973 - ``teid``: tunnel endpoint identifier.
974 - Default ``mask`` matches teid only.
979 Matches an ESP header.
981 - ``hdr``: ESP header definition (``rte_esp.h``).
982 - Default ``mask`` matches SPI only.
987 Matches a GENEVE header.
989 - ``ver_opt_len_o_c_rsvd0``: version (2b), length of the options fields (6b),
990 OAM packet (1b), critical options present (1b), reserved 0 (6b).
991 - ``protocol``: protocol type.
992 - ``vni``: virtual network identifier.
993 - ``rsvd1``: reserved, normally 0x00.
994 - Default ``mask`` matches VNI only.
999 Each possible action is represented by a type. Some have associated
1000 configuration structures. Several actions combined in a list can be assigned
1001 to a flow rule. That list is not ordered.
1003 They fall in three categories:
1005 - Terminating actions that prevent processing matched packets by subsequent
1006 flow rules, unless overridden with PASSTHRU.
1008 - Non-terminating actions that leave matched packets up for additional
1009 processing by subsequent flow rules.
1011 - Other non-terminating meta actions that do not affect the fate of packets.
1013 When several actions are combined in a flow rule, they should all have
1014 different types (e.g. dropping a packet twice is not possible).
1016 Only the last action of a given type is taken into account. PMDs still
1017 perform error checking on the entire list.
1019 Like matching patterns, action lists are terminated by END items.
1021 *Note that PASSTHRU is the only action able to override a terminating rule.*
1023 Example of action that redirects packets to queue index 10:
1025 .. _table_rte_flow_action_example:
1027 .. table:: Queue action
1029 +-----------+-------+
1031 +===========+=======+
1033 +-----------+-------+
1035 Action lists examples, their order is not significant, applications must
1036 consider all actions to be performed simultaneously:
1038 .. _table_rte_flow_count_and_drop:
1040 .. table:: Count and drop
1054 .. _table_rte_flow_mark_count_redirect:
1056 .. table:: Mark, count and redirect
1058 +-------+--------+-----------+-------+
1059 | Index | Action | Field | Value |
1060 +=======+========+===========+=======+
1061 | 0 | MARK | ``mark`` | 0x2a |
1062 +-------+--------+-----------+-------+
1064 +-------+--------+-----------+-------+
1065 | 2 | QUEUE | ``queue`` | 10 |
1066 +-------+--------+-----------+-------+
1068 +-------+----------------------------+
1072 .. _table_rte_flow_redirect_queue_5:
1074 .. table:: Redirect to queue 5
1076 +-------+--------+-----------+-------+
1077 | Index | Action | Field | Value |
1078 +=======+========+===========+=======+
1080 +-------+--------+-----------+-------+
1081 | 1 | QUEUE | ``queue`` | 5 |
1082 +-------+--------+-----------+-------+
1084 +-------+----------------------------+
1086 In the above example, considering both actions are performed simultaneously,
1087 the end result is that only QUEUE has any effect.
1089 .. _table_rte_flow_redirect_queue_3:
1091 .. table:: Redirect to queue 3
1093 +-------+--------+-----------+-------+
1094 | Index | Action | Field | Value |
1095 +=======+========+===========+=======+
1096 | 0 | QUEUE | ``queue`` | 5 |
1097 +-------+--------+-----------+-------+
1099 +-------+--------+-----------+-------+
1100 | 2 | QUEUE | ``queue`` | 3 |
1101 +-------+--------+-----------+-------+
1103 +-------+----------------------------+
1105 As previously described, only the last action of a given type found in the
1106 list is taken into account. The above example also shows that VOID is
1112 Common action types are described in this section. Like pattern item types,
1113 this list is not exhaustive as new actions will be added in the future.
1118 End marker for action lists. Prevents further processing of actions, thereby
1121 - Its numeric value is 0 for convenience.
1122 - PMD support is mandatory.
1123 - No configurable properties.
1125 .. _table_rte_flow_action_end:
1138 Used as a placeholder for convenience. It is ignored and simply discarded by
1141 - PMD support is mandatory.
1142 - No configurable properties.
1144 .. _table_rte_flow_action_void:
1154 Action: ``PASSTHRU``
1155 ^^^^^^^^^^^^^^^^^^^^
1157 Leaves packets up for additional processing by subsequent flow rules. This
1158 is the default when a rule does not contain a terminating action, but can be
1159 specified to force a rule to become non-terminating.
1161 - No configurable properties.
1163 .. _table_rte_flow_action_passthru:
1173 Example to copy a packet to a queue and continue processing by subsequent
1176 .. _table_rte_flow_action_passthru_example:
1178 .. table:: Copy to queue 8
1180 +-------+--------+-----------+-------+
1181 | Index | Action | Field | Value |
1182 +=======+========+===========+=======+
1184 +-------+--------+-----------+-------+
1185 | 1 | QUEUE | ``queue`` | 8 |
1186 +-------+--------+-----------+-------+
1188 +-------+----------------------------+
1193 Attaches an integer value to packets and sets ``PKT_RX_FDIR`` and
1194 ``PKT_RX_FDIR_ID`` mbuf flags.
1196 This value is arbitrary and application-defined. Maximum allowed value
1197 depends on the underlying implementation. It is returned in the
1198 ``hash.fdir.hi`` mbuf field.
1200 .. _table_rte_flow_action_mark:
1204 +--------+--------------------------------------+
1206 +========+======================================+
1207 | ``id`` | integer value to return with packets |
1208 +--------+--------------------------------------+
1213 Flags packets. Similar to `Action: MARK`_ without a specific value; only
1214 sets the ``PKT_RX_FDIR`` mbuf flag.
1216 - No configurable properties.
1218 .. _table_rte_flow_action_flag:
1231 Assigns packets to a given queue index.
1233 - Terminating by default.
1235 .. _table_rte_flow_action_queue:
1239 +-----------+--------------------+
1241 +===========+====================+
1242 | ``index`` | queue index to use |
1243 +-----------+--------------------+
1250 - No configurable properties.
1251 - Terminating by default.
1252 - PASSTHRU overrides this action if both are specified.
1254 .. _table_rte_flow_action_drop:
1267 Enables counters for this rule.
1269 These counters can be retrieved and reset through ``rte_flow_query()``, see
1270 ``struct rte_flow_query_count``.
1272 - Counters can be retrieved with ``rte_flow_query()``.
1273 - No configurable properties.
1275 .. _table_rte_flow_action_count:
1285 Query structure to retrieve and reset flow rule counters:
1287 .. _table_rte_flow_query_count:
1289 .. table:: COUNT query
1291 +---------------+-----+-----------------------------------+
1292 | Field | I/O | Value |
1293 +===============+=====+===================================+
1294 | ``reset`` | in | reset counter after query |
1295 +---------------+-----+-----------------------------------+
1296 | ``hits_set`` | out | ``hits`` field is set |
1297 +---------------+-----+-----------------------------------+
1298 | ``bytes_set`` | out | ``bytes`` field is set |
1299 +---------------+-----+-----------------------------------+
1300 | ``hits`` | out | number of hits for this rule |
1301 +---------------+-----+-----------------------------------+
1302 | ``bytes`` | out | number of bytes through this rule |
1303 +---------------+-----+-----------------------------------+
1308 Duplicates packets to a given queue index.
1310 This is normally combined with QUEUE, however when used alone, it is
1311 actually similar to QUEUE + PASSTHRU.
1313 - Non-terminating by default.
1315 .. _table_rte_flow_action_dup:
1319 +-----------+------------------------------------+
1321 +===========+====================================+
1322 | ``index`` | queue index to duplicate packet to |
1323 +-----------+------------------------------------+
1328 Similar to QUEUE, except RSS is additionally performed on packets to spread
1329 them among several queues according to the provided parameters.
1331 Note: RSS hash result is stored in the ``hash.rss`` mbuf field which
1332 overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi``
1333 field only, both can be requested simultaneously.
1335 - Terminating by default.
1337 .. _table_rte_flow_action_rss:
1341 +--------------+------------------------------+
1343 +==============+==============================+
1344 | ``rss_conf`` | RSS parameters |
1345 +--------------+------------------------------+
1346 | ``num`` | number of entries in queue[] |
1347 +--------------+------------------------------+
1348 | ``queue[]`` | queue indices to use |
1349 +--------------+------------------------------+
1354 Redirects packets to the physical function (PF) of the current device.
1356 - No configurable properties.
1357 - Terminating by default.
1359 .. _table_rte_flow_action_pf:
1372 Redirects packets to a virtual function (VF) of the current device.
1374 Packets matched by a VF pattern item can be redirected to their original VF
1375 ID instead of the specified one. This parameter may not be available and is
1376 not guaranteed to work properly if the VF part is matched by a prior flow
1377 rule or if packets are not addressed to a VF in the first place.
1379 - Terminating by default.
1381 .. _table_rte_flow_action_vf:
1385 +--------------+--------------------------------+
1387 +==============+================================+
1388 | ``original`` | use original VF ID if possible |
1389 +--------------+--------------------------------+
1390 | ``vf`` | VF ID to redirect packets to |
1391 +--------------+--------------------------------+
1396 Applies a stage of metering and policing.
1398 The metering and policing (MTR) object has to be first created using the
1399 rte_mtr_create() API function. The ID of the MTR object is specified as
1400 action parameter. More than one flow can use the same MTR object through
1401 the meter action. The MTR object can be further updated or queried using
1404 - Non-terminating by default.
1406 .. _table_rte_flow_action_meter:
1410 +--------------+---------------+
1412 +==============+===============+
1413 | ``mtr_id`` | MTR object ID |
1414 +--------------+---------------+
1416 Action: ``SECURITY``
1417 ^^^^^^^^^^^^^^^^^^^^
1419 Perform the security action on flows matched by the pattern items
1420 according to the configuration of the security session.
1422 This action modifies the payload of matched flows. For INLINE_CRYPTO, the
1423 security protocol headers and IV are fully provided by the application as
1424 specified in the flow pattern. The payload of matching packets is
1425 encrypted on egress, and decrypted and authenticated on ingress.
1426 For INLINE_PROTOCOL, the security protocol is fully offloaded to HW,
1427 providing full encapsulation and decapsulation of packets in security
1428 protocols. The flow pattern specifies both the outer security header fields
1429 and the inner packet fields. The security session specified in the action
1430 must match the pattern parameters.
1432 The security session specified in the action must be created on the same
1433 port as the flow action that is being specified.
1435 The ingress/egress flow attribute should match that specified in the
1436 security session if the security session supports the definition of the
1439 Multiple flows can be configured to use the same security session.
1441 - Non-terminating by default.
1443 .. _table_rte_flow_action_security:
1447 +----------------------+--------------------------------------+
1449 +======================+======================================+
1450 | ``security_session`` | security session to apply |
1451 +----------------------+--------------------------------------+
1453 The following is an example of configuring IPsec inline using the
1454 INLINE_CRYPTO security session:
1456 The encryption algorithm, keys and salt are part of the opaque
1457 ``rte_security_session``. The SA is identified according to the IP and ESP
1458 fields in the pattern items.
1460 .. _table_rte_flow_item_esp_inline_example:
1462 .. table:: IPsec inline crypto flow pattern items.
1464 +-------+----------+
1466 +=======+==========+
1468 +-------+----------+
1470 +-------+----------+
1472 +-------+----------+
1474 +-------+----------+
1476 .. _table_rte_flow_action_esp_inline_example:
1478 .. table:: IPsec inline flow actions.
1480 +-------+----------+
1482 +=======+==========+
1484 +-------+----------+
1486 +-------+----------+
1491 All specified pattern items (``enum rte_flow_item_type``) and actions
1492 (``enum rte_flow_action_type``) use positive identifiers.
1494 The negative space is reserved for dynamic types generated by PMDs during
1495 run-time. PMDs may encounter them as a result but must not accept negative
1496 identifiers they are not aware of.
1498 A method to generate them remains to be defined.
1503 Pattern item types will be added as new protocols are implemented.
1505 Variable headers support through dedicated pattern items, for example in
1506 order to match specific IPv4 options and IPv6 extension headers would be
1507 stacked after IPv4/IPv6 items.
1509 Other action types are planned but are not defined yet. These include the
1510 ability to alter packet data in several ways, such as performing
1511 encapsulation/decapsulation of tunnel headers.
1516 A rather simple API with few functions is provided to fully manage flow
1519 Each created flow rule is associated with an opaque, PMD-specific handle
1520 pointer. The application is responsible for keeping it until the rule is
1523 Flows rules are represented by ``struct rte_flow`` objects.
1528 Given that expressing a definite set of device capabilities is not
1529 practical, a dedicated function is provided to check if a flow rule is
1530 supported and can be created.
1535 rte_flow_validate(uint16_t port_id,
1536 const struct rte_flow_attr *attr,
1537 const struct rte_flow_item pattern[],
1538 const struct rte_flow_action actions[],
1539 struct rte_flow_error *error);
1541 The flow rule is validated for correctness and whether it could be accepted
1542 by the device given sufficient resources. The rule is checked against the
1543 current device mode and queue configuration. The flow rule may also
1544 optionally be validated against existing flow rules and device resources.
1545 This function has no effect on the target device.
1547 The returned value is guaranteed to remain valid only as long as no
1548 successful calls to ``rte_flow_create()`` or ``rte_flow_destroy()`` are made
1549 in the meantime and no device parameter affecting flow rules in any way are
1550 modified, due to possible collisions or resource limitations (although in
1551 such cases ``EINVAL`` should not be returned).
1555 - ``port_id``: port identifier of Ethernet device.
1556 - ``attr``: flow rule attributes.
1557 - ``pattern``: pattern specification (list terminated by the END pattern
1559 - ``actions``: associated actions (list terminated by the END action).
1560 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1561 this structure in case of error only.
1565 - 0 if flow rule is valid and can be created. A negative errno value
1566 otherwise (``rte_errno`` is also set), the following errors are defined.
1567 - ``-ENOSYS``: underlying device does not support this functionality.
1568 - ``-EINVAL``: unknown or invalid rule specification.
1569 - ``-ENOTSUP``: valid but unsupported rule specification (e.g. partial
1570 bit-masks are unsupported).
1571 - ``EEXIST``: collision with an existing rule. Only returned if device
1572 supports flow rule collision checking and there was a flow rule
1573 collision. Not receiving this return code is no guarantee that creating
1574 the rule will not fail due to a collision.
1575 - ``ENOMEM``: not enough memory to execute the function, or if the device
1576 supports resource validation, resource limitation on the device.
1577 - ``-EBUSY``: action cannot be performed due to busy device resources, may
1578 succeed if the affected queues or even the entire port are in a stopped
1579 state (see ``rte_eth_dev_rx_queue_stop()`` and ``rte_eth_dev_stop()``).
1584 Creating a flow rule is similar to validating one, except the rule is
1585 actually created and a handle returned.
1590 rte_flow_create(uint16_t port_id,
1591 const struct rte_flow_attr *attr,
1592 const struct rte_flow_item pattern[],
1593 const struct rte_flow_action *actions[],
1594 struct rte_flow_error *error);
1598 - ``port_id``: port identifier of Ethernet device.
1599 - ``attr``: flow rule attributes.
1600 - ``pattern``: pattern specification (list terminated by the END pattern
1602 - ``actions``: associated actions (list terminated by the END action).
1603 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1604 this structure in case of error only.
1608 A valid handle in case of success, NULL otherwise and ``rte_errno`` is set
1609 to the positive version of one of the error codes defined for
1610 ``rte_flow_validate()``.
1615 Flow rules destruction is not automatic, and a queue or a port should not be
1616 released if any are still attached to them. Applications must take care of
1617 performing this step before releasing resources.
1622 rte_flow_destroy(uint16_t port_id,
1623 struct rte_flow *flow,
1624 struct rte_flow_error *error);
1627 Failure to destroy a flow rule handle may occur when other flow rules depend
1628 on it, and destroying it would result in an inconsistent state.
1630 This function is only guaranteed to succeed if handles are destroyed in
1631 reverse order of their creation.
1635 - ``port_id``: port identifier of Ethernet device.
1636 - ``flow``: flow rule handle to destroy.
1637 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1638 this structure in case of error only.
1642 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
1647 Convenience function to destroy all flow rule handles associated with a
1648 port. They are released as with successive calls to ``rte_flow_destroy()``.
1653 rte_flow_flush(uint16_t port_id,
1654 struct rte_flow_error *error);
1656 In the unlikely event of failure, handles are still considered destroyed and
1657 no longer valid but the port must be assumed to be in an inconsistent state.
1661 - ``port_id``: port identifier of Ethernet device.
1662 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1663 this structure in case of error only.
1667 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
1672 Query an existing flow rule.
1674 This function allows retrieving flow-specific data such as counters. Data
1675 is gathered by special actions which must be present in the flow rule
1681 rte_flow_query(uint16_t port_id,
1682 struct rte_flow *flow,
1683 enum rte_flow_action_type action,
1685 struct rte_flow_error *error);
1689 - ``port_id``: port identifier of Ethernet device.
1690 - ``flow``: flow rule handle to query.
1691 - ``action``: action type to query.
1692 - ``data``: pointer to storage for the associated query data type.
1693 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1694 this structure in case of error only.
1698 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
1703 The general expectation for ingress traffic is that flow rules process it
1704 first; the remaining unmatched or pass-through traffic usually ends up in a
1705 queue (with or without RSS, locally or in some sub-device instance)
1706 depending on the global configuration settings of a port.
1708 While fine from a compatibility standpoint, this approach makes drivers more
1709 complex as they have to check for possible side effects outside of this API
1710 when creating or destroying flow rules. It results in a more limited set of
1711 available rule types due to the way device resources are assigned (e.g. no
1712 support for the RSS action even on capable hardware).
1714 Given that nonspecific traffic can be handled by flow rules as well,
1715 isolated mode is a means for applications to tell a driver that ingress on
1716 the underlying port must be injected from the defined flow rules only; that
1717 no default traffic is expected outside those rules.
1719 This has the following benefits:
1721 - Applications get finer-grained control over the kind of traffic they want
1722 to receive (no traffic by default).
1724 - More importantly they control at what point nonspecific traffic is handled
1725 relative to other flow rules, by adjusting priority levels.
1727 - Drivers can assign more hardware resources to flow rules and expand the
1728 set of supported rule types.
1730 Because toggling isolated mode may cause profound changes to the ingress
1731 processing path of a driver, it may not be possible to leave it once
1732 entered. Likewise, existing flow rules or global configuration settings may
1733 prevent a driver from entering isolated mode.
1735 Applications relying on this mode are therefore encouraged to toggle it as
1736 soon as possible after device initialization, ideally before the first call
1737 to ``rte_eth_dev_configure()`` to avoid possible failures due to conflicting
1740 Once effective, the following functionality has no effect on the underlying
1741 port and may return errors such as ``ENOTSUP`` ("not supported"):
1743 - Toggling promiscuous mode.
1744 - Toggling allmulticast mode.
1745 - Configuring MAC addresses.
1746 - Configuring multicast addresses.
1747 - Configuring VLAN filters.
1748 - Configuring Rx filters through the legacy API (e.g. FDIR).
1749 - Configuring global RSS settings.
1754 rte_flow_isolate(uint16_t port_id, int set, struct rte_flow_error *error);
1758 - ``port_id``: port identifier of Ethernet device.
1759 - ``set``: nonzero to enter isolated mode, attempt to leave it otherwise.
1760 - ``error``: perform verbose error reporting if not NULL. PMDs initialize
1761 this structure in case of error only.
1765 - 0 on success, a negative errno value otherwise and ``rte_errno`` is set.
1767 Verbose error reporting
1768 -----------------------
1770 The defined *errno* values may not be accurate enough for users or
1771 application developers who want to investigate issues related to flow rules
1772 management. A dedicated error object is defined for this purpose:
1776 enum rte_flow_error_type {
1777 RTE_FLOW_ERROR_TYPE_NONE, /**< No error. */
1778 RTE_FLOW_ERROR_TYPE_UNSPECIFIED, /**< Cause unspecified. */
1779 RTE_FLOW_ERROR_TYPE_HANDLE, /**< Flow rule (handle). */
1780 RTE_FLOW_ERROR_TYPE_ATTR_GROUP, /**< Group field. */
1781 RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY, /**< Priority field. */
1782 RTE_FLOW_ERROR_TYPE_ATTR_INGRESS, /**< Ingress field. */
1783 RTE_FLOW_ERROR_TYPE_ATTR_EGRESS, /**< Egress field. */
1784 RTE_FLOW_ERROR_TYPE_ATTR, /**< Attributes structure. */
1785 RTE_FLOW_ERROR_TYPE_ITEM_NUM, /**< Pattern length. */
1786 RTE_FLOW_ERROR_TYPE_ITEM, /**< Specific pattern item. */
1787 RTE_FLOW_ERROR_TYPE_ACTION_NUM, /**< Number of actions. */
1788 RTE_FLOW_ERROR_TYPE_ACTION, /**< Specific action. */
1791 struct rte_flow_error {
1792 enum rte_flow_error_type type; /**< Cause field and error types. */
1793 const void *cause; /**< Object responsible for the error. */
1794 const char *message; /**< Human-readable error message. */
1797 Error type ``RTE_FLOW_ERROR_TYPE_NONE`` stands for no error, in which case
1798 remaining fields can be ignored. Other error types describe the type of the
1799 object pointed by ``cause``.
1801 If non-NULL, ``cause`` points to the object responsible for the error. For a
1802 flow rule, this may be a pattern item or an individual action.
1804 If non-NULL, ``message`` provides a human-readable error message.
1806 This object is normally allocated by applications and set by PMDs in case of
1807 error, the message points to a constant string which does not need to be
1808 freed by the application, however its pointer can be considered valid only
1809 as long as its associated DPDK port remains configured. Closing the
1810 underlying device or unloading the PMD invalidates it.
1821 rte_flow_error_set(struct rte_flow_error *error,
1823 enum rte_flow_error_type type,
1825 const char *message);
1827 This function initializes ``error`` (if non-NULL) with the provided
1828 parameters and sets ``rte_errno`` to ``code``. A negative error ``code`` is
1834 - DPDK does not keep track of flow rules definitions or flow rule objects
1835 automatically. Applications may keep track of the former and must keep
1836 track of the latter. PMDs may also do it for internal needs, however this
1837 must not be relied on by applications.
1839 - Flow rules are not maintained between successive port initializations. An
1840 application exiting without releasing them and restarting must re-create
1843 - API operations are synchronous and blocking (``EAGAIN`` cannot be
1846 - There is no provision for reentrancy/multi-thread safety, although nothing
1847 should prevent different devices from being configured at the same
1848 time. PMDs may protect their control path functions accordingly.
1850 - Stopping the data path (TX/RX) should not be necessary when managing flow
1851 rules. If this cannot be achieved naturally or with workarounds (such as
1852 temporarily replacing the burst function pointers), an appropriate error
1853 code must be returned (``EBUSY``).
1855 - PMDs, not applications, are responsible for maintaining flow rules
1856 configuration when stopping and restarting a port or performing other
1857 actions which may affect them. They can only be destroyed explicitly by
1860 For devices exposing multiple ports sharing global settings affected by flow
1863 - All ports under DPDK control must behave consistently, PMDs are
1864 responsible for making sure that existing flow rules on a port are not
1865 affected by other ports.
1867 - Ports not under DPDK control (unaffected or handled by other applications)
1868 are user's responsibility. They may affect existing flow rules and cause
1869 undefined behavior. PMDs aware of this may prevent flow rules creation
1870 altogether in such cases.
1875 The PMD interface is defined in ``rte_flow_driver.h``. It is not subject to
1876 API/ABI versioning constraints as it is not exposed to applications and may
1877 evolve independently.
1879 It is currently implemented on top of the legacy filtering framework through
1880 filter type *RTE_ETH_FILTER_GENERIC* that accepts the single operation
1881 *RTE_ETH_FILTER_GET* to return PMD-specific *rte_flow* callbacks wrapped
1882 inside ``struct rte_flow_ops``.
1884 This overhead is temporarily necessary in order to keep compatibility with
1885 the legacy filtering framework, which should eventually disappear.
1887 - PMD callbacks implement exactly the interface described in `Rules
1888 management`_, except for the port ID argument which has already been
1889 converted to a pointer to the underlying ``struct rte_eth_dev``.
1891 - Public API functions do not process flow rules definitions at all before
1892 calling PMD functions (no basic error checking, no validation
1893 whatsoever). They only make sure these callbacks are non-NULL or return
1894 the ``ENOSYS`` (function not supported) error.
1896 This interface additionally defines the following helper function:
1898 - ``rte_flow_ops_get()``: get generic flow operations structure from a
1901 More will be added over time.
1903 Device compatibility
1904 --------------------
1906 No known implementation supports all the described features.
1908 Unsupported features or combinations are not expected to be fully emulated
1909 in software by PMDs for performance reasons. Partially supported features
1910 may be completed in software as long as hardware performs most of the work
1911 (such as queue redirection and packet recognition).
1913 However PMDs are expected to do their best to satisfy application requests
1914 by working around hardware limitations as long as doing so does not affect
1915 the behavior of existing flow rules.
1917 The following sections provide a few examples of such cases and describe how
1918 PMDs should handle them, they are based on limitations built into the
1924 Each flow rule comes with its own, per-layer bit-masks, while hardware may
1925 support only a single, device-wide bit-mask for a given layer type, so that
1926 two IPv4 rules cannot use different bit-masks.
1928 The expected behavior in this case is that PMDs automatically configure
1929 global bit-masks according to the needs of the first flow rule created.
1931 Subsequent rules are allowed only if their bit-masks match those, the
1932 ``EEXIST`` error code should be returned otherwise.
1934 Unsupported layer types
1935 ~~~~~~~~~~~~~~~~~~~~~~~
1937 Many protocols can be simulated by crafting patterns with the `Item: RAW`_
1940 PMDs can rely on this capability to simulate support for protocols with
1941 headers not directly recognized by hardware.
1943 ``ANY`` pattern item
1944 ~~~~~~~~~~~~~~~~~~~~
1946 This pattern item stands for anything, which can be difficult to translate
1947 to something hardware would understand, particularly if followed by more
1950 Consider the following pattern:
1952 .. _table_rte_flow_unsupported_any:
1954 .. table:: Pattern with ANY as L3
1956 +-------+-----------------------+
1958 +=======+=======================+
1960 +-------+-----+---------+-------+
1961 | 1 | ANY | ``num`` | ``1`` |
1962 +-------+-----+---------+-------+
1964 +-------+-----------------------+
1966 +-------+-----------------------+
1968 Knowing that TCP does not make sense with something other than IPv4 and IPv6
1969 as L3, such a pattern may be translated to two flow rules instead:
1971 .. _table_rte_flow_unsupported_any_ipv4:
1973 .. table:: ANY replaced with IPV4
1975 +-------+--------------------+
1977 +=======+====================+
1979 +-------+--------------------+
1980 | 1 | IPV4 (zeroed mask) |
1981 +-------+--------------------+
1983 +-------+--------------------+
1985 +-------+--------------------+
1989 .. _table_rte_flow_unsupported_any_ipv6:
1991 .. table:: ANY replaced with IPV6
1993 +-------+--------------------+
1995 +=======+====================+
1997 +-------+--------------------+
1998 | 1 | IPV6 (zeroed mask) |
1999 +-------+--------------------+
2001 +-------+--------------------+
2003 +-------+--------------------+
2005 Note that as soon as a ANY rule covers several layers, this approach may
2006 yield a large number of hidden flow rules. It is thus suggested to only
2007 support the most common scenarios (anything as L2 and/or L3).
2012 - When combined with `Action: QUEUE`_, packet counting (`Action: COUNT`_)
2013 and tagging (`Action: MARK`_ or `Action: FLAG`_) may be implemented in
2014 software as long as the target queue is used by a single rule.
2016 - A rule specifying both `Action: DUP`_ + `Action: QUEUE`_ may be translated
2017 to two hidden rules combining `Action: QUEUE`_ and `Action: PASSTHRU`_.
2019 - When a single target queue is provided, `Action: RSS`_ can also be
2020 implemented through `Action: QUEUE`_.
2025 While it would naturally make sense, flow rules cannot be assumed to be
2026 processed by hardware in the same order as their creation for several
2029 - They may be managed internally as a tree or a hash table instead of a
2031 - Removing a flow rule before adding another one can either put the new rule
2032 at the end of the list or reuse a freed entry.
2033 - Duplication may occur when packets are matched by several rules.
2035 For overlapping rules (particularly in order to use `Action: PASSTHRU`_)
2036 predictable behavior is only guaranteed by using different priority levels.
2038 Priority levels are not necessarily implemented in hardware, or may be
2039 severely limited (e.g. a single priority bit).
2041 For these reasons, priority levels may be implemented purely in software by
2044 - For devices expecting flow rules to be added in the correct order, PMDs
2045 may destroy and re-create existing rules after adding a new one with
2048 - A configurable number of dummy or empty rules can be created at
2049 initialization time to save high priority slots for later.
2051 - In order to save priority levels, PMDs may evaluate whether rules are
2052 likely to collide and adjust their priority accordingly.
2057 - A device profile selection function which could be used to force a
2058 permanent profile instead of relying on its automatic configuration based
2059 on existing flow rules.
2061 - A method to optimize *rte_flow* rules with specific pattern items and
2062 action types generated on the fly by PMDs. DPDK should assign negative
2063 numbers to these in order to not collide with the existing types. See
2066 - Adding specific egress pattern items and actions as described in
2067 `Attribute: Traffic direction`_.
2069 - Optional software fallback when PMDs are unable to handle requested flow
2070 rules so applications do not have to implement their own.
2075 Exhaustive list of deprecated filter types (normally prefixed with
2076 *RTE_ETH_FILTER_*) found in ``rte_eth_ctrl.h`` and methods to convert them
2077 to *rte_flow* rules.
2079 ``MACVLAN`` to ``ETH`` → ``VF``, ``PF``
2080 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2082 *MACVLAN* can be translated to a basic `Item: ETH`_ flow rule with a
2083 terminating `Action: VF`_ or `Action: PF`_.
2085 .. _table_rte_flow_migration_macvlan:
2087 .. table:: MACVLAN conversion
2089 +--------------------------+---------+
2090 | Pattern | Actions |
2091 +===+=====+==========+=====+=========+
2092 | 0 | ETH | ``spec`` | any | VF, |
2093 | | +----------+-----+ PF |
2094 | | | ``last`` | N/A | |
2095 | | +----------+-----+ |
2096 | | | ``mask`` | any | |
2097 +---+-----+----------+-----+---------+
2099 +---+----------------------+---------+
2101 ``ETHERTYPE`` to ``ETH`` → ``QUEUE``, ``DROP``
2102 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2104 *ETHERTYPE* is basically an `Item: ETH`_ flow rule with a terminating
2105 `Action: QUEUE`_ or `Action: DROP`_.
2107 .. _table_rte_flow_migration_ethertype:
2109 .. table:: ETHERTYPE conversion
2111 +--------------------------+---------+
2112 | Pattern | Actions |
2113 +===+=====+==========+=====+=========+
2114 | 0 | ETH | ``spec`` | any | QUEUE, |
2115 | | +----------+-----+ DROP |
2116 | | | ``last`` | N/A | |
2117 | | +----------+-----+ |
2118 | | | ``mask`` | any | |
2119 +---+-----+----------+-----+---------+
2121 +---+----------------------+---------+
2123 ``FLEXIBLE`` to ``RAW`` → ``QUEUE``
2124 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2126 *FLEXIBLE* can be translated to one `Item: RAW`_ pattern with a terminating
2127 `Action: QUEUE`_ and a defined priority level.
2129 .. _table_rte_flow_migration_flexible:
2131 .. table:: FLEXIBLE conversion
2133 +--------------------------+---------+
2134 | Pattern | Actions |
2135 +===+=====+==========+=====+=========+
2136 | 0 | RAW | ``spec`` | any | QUEUE |
2137 | | +----------+-----+ |
2138 | | | ``last`` | N/A | |
2139 | | +----------+-----+ |
2140 | | | ``mask`` | any | |
2141 +---+-----+----------+-----+---------+
2143 +---+----------------------+---------+
2145 ``SYN`` to ``TCP`` → ``QUEUE``
2146 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2148 *SYN* is a `Item: TCP`_ rule with only the ``syn`` bit enabled and masked,
2149 and a terminating `Action: QUEUE`_.
2151 Priority level can be set to simulate the high priority bit.
2153 .. _table_rte_flow_migration_syn:
2155 .. table:: SYN conversion
2157 +-----------------------------------+---------+
2158 | Pattern | Actions |
2159 +===+======+==========+=============+=========+
2160 | 0 | ETH | ``spec`` | unset | QUEUE |
2161 | | +----------+-------------+ |
2162 | | | ``last`` | unset | |
2163 | | +----------+-------------+ |
2164 | | | ``mask`` | unset | |
2165 +---+------+----------+-------------+---------+
2166 | 1 | IPV4 | ``spec`` | unset | END |
2167 | | +----------+-------------+ |
2168 | | | ``mask`` | unset | |
2169 | | +----------+-------------+ |
2170 | | | ``mask`` | unset | |
2171 +---+------+----------+---------+---+ |
2172 | 2 | TCP | ``spec`` | ``syn`` | 1 | |
2173 | | +----------+---------+---+ |
2174 | | | ``mask`` | ``syn`` | 1 | |
2175 +---+------+----------+---------+---+ |
2177 +---+-------------------------------+---------+
2179 ``NTUPLE`` to ``IPV4``, ``TCP``, ``UDP`` → ``QUEUE``
2180 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2182 *NTUPLE* is similar to specifying an empty L2, `Item: IPV4`_ as L3 with
2183 `Item: TCP`_ or `Item: UDP`_ as L4 and a terminating `Action: QUEUE`_.
2185 A priority level can be specified as well.
2187 .. _table_rte_flow_migration_ntuple:
2189 .. table:: NTUPLE conversion
2191 +-----------------------------+---------+
2192 | Pattern | Actions |
2193 +===+======+==========+=======+=========+
2194 | 0 | ETH | ``spec`` | unset | QUEUE |
2195 | | +----------+-------+ |
2196 | | | ``last`` | unset | |
2197 | | +----------+-------+ |
2198 | | | ``mask`` | unset | |
2199 +---+------+----------+-------+---------+
2200 | 1 | IPV4 | ``spec`` | any | END |
2201 | | +----------+-------+ |
2202 | | | ``last`` | unset | |
2203 | | +----------+-------+ |
2204 | | | ``mask`` | any | |
2205 +---+------+----------+-------+ |
2206 | 2 | TCP, | ``spec`` | any | |
2207 | | UDP +----------+-------+ |
2208 | | | ``last`` | unset | |
2209 | | +----------+-------+ |
2210 | | | ``mask`` | any | |
2211 +---+------+----------+-------+ |
2213 +---+-------------------------+---------+
2215 ``TUNNEL`` to ``ETH``, ``IPV4``, ``IPV6``, ``VXLAN`` (or other) → ``QUEUE``
2216 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2218 *TUNNEL* matches common IPv4 and IPv6 L3/L4-based tunnel types.
2220 In the following table, `Item: ANY`_ is used to cover the optional L4.
2222 .. _table_rte_flow_migration_tunnel:
2224 .. table:: TUNNEL conversion
2226 +-------------------------------------------------------+---------+
2227 | Pattern | Actions |
2228 +===+==========================+==========+=============+=========+
2229 | 0 | ETH | ``spec`` | any | QUEUE |
2230 | | +----------+-------------+ |
2231 | | | ``last`` | unset | |
2232 | | +----------+-------------+ |
2233 | | | ``mask`` | any | |
2234 +---+--------------------------+----------+-------------+---------+
2235 | 1 | IPV4, IPV6 | ``spec`` | any | END |
2236 | | +----------+-------------+ |
2237 | | | ``last`` | unset | |
2238 | | +----------+-------------+ |
2239 | | | ``mask`` | any | |
2240 +---+--------------------------+----------+-------------+ |
2241 | 2 | ANY | ``spec`` | any | |
2242 | | +----------+-------------+ |
2243 | | | ``last`` | unset | |
2244 | | +----------+---------+---+ |
2245 | | | ``mask`` | ``num`` | 0 | |
2246 +---+--------------------------+----------+---------+---+ |
2247 | 3 | VXLAN, GENEVE, TEREDO, | ``spec`` | any | |
2248 | | NVGRE, GRE, ... +----------+-------------+ |
2249 | | | ``last`` | unset | |
2250 | | +----------+-------------+ |
2251 | | | ``mask`` | any | |
2252 +---+--------------------------+----------+-------------+ |
2254 +---+---------------------------------------------------+---------+
2256 ``FDIR`` to most item types → ``QUEUE``, ``DROP``, ``PASSTHRU``
2257 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2259 *FDIR* is more complex than any other type, there are several methods to
2260 emulate its functionality. It is summarized for the most part in the table
2263 A few features are intentionally not supported:
2265 - The ability to configure the matching input set and masks for the entire
2266 device, PMDs should take care of it automatically according to the
2267 requested flow rules.
2269 For example if a device supports only one bit-mask per protocol type,
2270 source/address IPv4 bit-masks can be made immutable by the first created
2271 rule. Subsequent IPv4 or TCPv4 rules can only be created if they are
2274 Note that only protocol bit-masks affected by existing flow rules are
2275 immutable, others can be changed later. They become mutable again after
2276 the related flow rules are destroyed.
2278 - Returning four or eight bytes of matched data when using flex bytes
2279 filtering. Although a specific action could implement it, it conflicts
2280 with the much more useful 32 bits tagging on devices that support it.
2282 - Side effects on RSS processing of the entire device. Flow rules that
2283 conflict with the current device configuration should not be
2284 allowed. Similarly, device configuration should not be allowed when it
2285 affects existing flow rules.
2287 - Device modes of operation. "none" is unsupported since filtering cannot be
2288 disabled as long as a flow rule is present.
2290 - "MAC VLAN" or "tunnel" perfect matching modes should be automatically set
2291 according to the created flow rules.
2293 - Signature mode of operation is not defined but could be handled through
2296 .. _table_rte_flow_migration_fdir:
2298 .. table:: FDIR conversion
2300 +----------------------------------------+-----------------------+
2301 | Pattern | Actions |
2302 +===+===================+==========+=====+=======================+
2303 | 0 | ETH, RAW | ``spec`` | any | QUEUE, DROP, PASSTHRU |
2304 | | +----------+-----+ |
2305 | | | ``last`` | N/A | |
2306 | | +----------+-----+ |
2307 | | | ``mask`` | any | |
2308 +---+-------------------+----------+-----+-----------------------+
2309 | 1 | IPV4, IPv6 | ``spec`` | any | MARK |
2310 | | +----------+-----+ |
2311 | | | ``last`` | N/A | |
2312 | | +----------+-----+ |
2313 | | | ``mask`` | any | |
2314 +---+-------------------+----------+-----+-----------------------+
2315 | 2 | TCP, UDP, SCTP | ``spec`` | any | END |
2316 | | +----------+-----+ |
2317 | | | ``last`` | N/A | |
2318 | | +----------+-----+ |
2319 | | | ``mask`` | any | |
2320 +---+-------------------+----------+-----+ |
2321 | 3 | VF, PF, FUZZY | ``spec`` | any | |
2322 | | (optional) +----------+-----+ |
2323 | | | ``last`` | N/A | |
2324 | | +----------+-----+ |
2325 | | | ``mask`` | any | |
2326 +---+-------------------+----------+-----+ |
2328 +---+------------------------------------+-----------------------+
2333 There is no counterpart to this filter type because it translates to a
2334 global device setting instead of a pattern item. Device settings are
2335 automatically set according to the created flow rules.
2337 ``L2_TUNNEL`` to ``VOID`` → ``VXLAN`` (or others)
2338 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2340 All packets are matched. This type alters incoming packets to encapsulate
2341 them in a chosen tunnel type, optionally redirect them to a VF as well.
2343 The destination pool for tag based forwarding can be emulated with other
2344 flow rules using `Action: DUP`_.
2346 .. _table_rte_flow_migration_l2tunnel:
2348 .. table:: L2_TUNNEL conversion
2350 +---------------------------+--------------------+
2351 | Pattern | Actions |
2352 +===+======+==========+=====+====================+
2353 | 0 | VOID | ``spec`` | N/A | VXLAN, GENEVE, ... |
2356 | | +----------+-----+ |
2357 | | | ``last`` | N/A | |
2358 | | +----------+-----+ |
2359 | | | ``mask`` | N/A | |
2361 +---+------+----------+-----+--------------------+
2362 | 1 | END | VF (optional) |
2363 +---+ +--------------------+
2365 +---+-----------------------+--------------------+