1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2010-2017 Intel Corporation
11 * RTE Ethernet Device API
13 * The Ethernet Device API is composed of two parts:
15 * - The application-oriented Ethernet API that includes functions to setup
16 * an Ethernet device (configure it, setup its Rx and Tx queues and start it),
17 * to get its MAC address, the speed and the status of its physical link,
18 * to receive and to transmit packets, and so on.
20 * - The driver-oriented Ethernet API that exports functions allowing
21 * an Ethernet Poll Mode Driver (PMD) to allocate an Ethernet device instance,
22 * create memzone for HW rings and process registered callbacks, and so on.
23 * PMDs should include ethdev_driver.h instead of this header.
25 * By default, all the functions of the Ethernet Device API exported by a PMD
26 * are lock-free functions which assume to not be invoked in parallel on
27 * different logical cores to work on the same target object. For instance,
28 * the receive function of a PMD cannot be invoked in parallel on two logical
29 * cores to poll the same Rx queue [of the same port]. Of course, this function
30 * can be invoked in parallel by different logical cores on different Rx queues.
31 * It is the responsibility of the upper level application to enforce this rule.
33 * If needed, parallel accesses by multiple logical cores to shared queues
34 * shall be explicitly protected by dedicated inline lock-aware functions
35 * built on top of their corresponding lock-free functions of the PMD API.
37 * In all functions of the Ethernet API, the Ethernet device is
38 * designated by an integer >= 0 named the device port identifier.
40 * At the Ethernet driver level, Ethernet devices are represented by a generic
41 * data structure of type *rte_eth_dev*.
43 * Ethernet devices are dynamically registered during the PCI probing phase
44 * performed at EAL initialization time.
45 * When an Ethernet device is being probed, an *rte_eth_dev* structure and
46 * a new port identifier are allocated for that device. Then, the eth_dev_init()
47 * function supplied by the Ethernet driver matching the probed PCI
48 * device is invoked to properly initialize the device.
50 * The role of the device init function consists of resetting the hardware,
51 * checking access to Non-volatile Memory (NVM), reading the MAC address
54 * If the device init operation is successful, the correspondence between
55 * the port identifier assigned to the new device and its associated
56 * *rte_eth_dev* structure is effectively registered.
57 * Otherwise, both the *rte_eth_dev* structure and the port identifier are
60 * The functions exported by the application Ethernet API to setup a device
61 * designated by its port identifier must be invoked in the following order:
62 * - rte_eth_dev_configure()
63 * - rte_eth_tx_queue_setup()
64 * - rte_eth_rx_queue_setup()
65 * - rte_eth_dev_start()
67 * Then, the network application can invoke, in any order, the functions
68 * exported by the Ethernet API to get the MAC address of a given device, to
69 * get the speed and the status of a device physical link, to receive/transmit
70 * [burst of] packets, and so on.
72 * If the application wants to change the configuration (i.e. call
73 * rte_eth_dev_configure(), rte_eth_tx_queue_setup(), or
74 * rte_eth_rx_queue_setup()), it must call rte_eth_dev_stop() first to stop the
75 * device and then do the reconfiguration before calling rte_eth_dev_start()
76 * again. The transmit and receive functions should not be invoked when the
79 * Please note that some configuration is not stored between calls to
80 * rte_eth_dev_stop()/rte_eth_dev_start(). The following configuration will
84 * - flow control settings
85 * - receive mode configuration (promiscuous mode, all-multicast mode,
86 * hardware checksum mode, RSS/VMDQ settings etc.)
87 * - VLAN filtering configuration
88 * - default MAC address
89 * - MAC addresses supplied to MAC address array
90 * - flow director filtering mode (but not filtering rules)
91 * - NIC queue statistics mappings
93 * Any other configuration will not be stored and will need to be re-entered
94 * before a call to rte_eth_dev_start().
96 * Finally, a network application can close an Ethernet device by invoking the
97 * rte_eth_dev_close() function.
99 * Each function of the application Ethernet API invokes a specific function
100 * of the PMD that controls the target device designated by its port
102 * For this purpose, all device-specific functions of an Ethernet driver are
103 * supplied through a set of pointers contained in a generic structure of type
105 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev*
106 * structure by the device init function of the Ethernet driver, which is
107 * invoked during the PCI probing phase, as explained earlier.
109 * In other words, each function of the Ethernet API simply retrieves the
110 * *rte_eth_dev* structure associated with the device port identifier and
111 * performs an indirect invocation of the corresponding driver function
112 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure.
114 * For performance reasons, the address of the burst-oriented Rx and Tx
115 * functions of the Ethernet driver are not contained in the *eth_dev_ops*
116 * structure. Instead, they are directly stored at the beginning of the
117 * *rte_eth_dev* structure to avoid an extra indirect memory access during
120 * RTE Ethernet device drivers do not use interrupts for transmitting or
121 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit
122 * functions to applications.
123 * Both receive and transmit functions are packet-burst oriented to minimize
124 * their cost per packet through the following optimizations:
126 * - Sharing among multiple packets the incompressible cost of the
127 * invocation of receive/transmit functions.
129 * - Enabling receive/transmit functions to take advantage of burst-oriented
130 * hardware features (L1 cache, prefetch instructions, NIC head/tail
131 * registers) to minimize the number of CPU cycles per packet, for instance,
132 * by avoiding useless read memory accesses to ring descriptors, or by
133 * systematically using arrays of pointers that exactly fit L1 cache line
134 * boundaries and sizes.
136 * The burst-oriented receive function does not provide any error notification,
137 * to avoid the corresponding overhead. As a hint, the upper-level application
138 * might check the status of the device link once being systematically returned
139 * a 0 value by the receive function of the driver for a given number of tries.
148 /* Use this macro to check if LRO API is supported */
149 #define RTE_ETHDEV_HAS_LRO_SUPPORT
151 /* Alias RTE_LIBRTE_ETHDEV_DEBUG for backward compatibility. */
152 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
153 #define RTE_ETHDEV_DEBUG_RX
154 #define RTE_ETHDEV_DEBUG_TX
157 #include <rte_compat.h>
159 #include <rte_interrupts.h>
161 #include <rte_devargs.h>
162 #include <rte_bitops.h>
163 #include <rte_errno.h>
164 #include <rte_common.h>
165 #include <rte_config.h>
166 #include <rte_ether.h>
167 #include <rte_power_intrinsics.h>
169 #include "rte_ethdev_trace_fp.h"
170 #include "rte_dev_info.h"
172 extern int rte_eth_dev_logtype;
174 #define RTE_ETHDEV_LOG(level, ...) \
175 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
180 * Initializes a device iterator.
182 * This iterator allows accessing a list of devices matching some devargs.
185 * Device iterator handle initialized by the function.
186 * The fields bus_str and cls_str might be dynamically allocated,
187 * and could be freed by calling rte_eth_iterator_cleanup().
190 * Device description string.
193 * 0 on successful initialization, negative otherwise.
195 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs);
198 * Iterates on devices with devargs filter.
199 * The ownership is not checked.
201 * The next port id is returned, and the iterator is updated.
204 * Device iterator handle initialized by rte_eth_iterator_init().
205 * Some fields bus_str and cls_str might be freed when no more port is found,
206 * by calling rte_eth_iterator_cleanup().
209 * A port id if found, RTE_MAX_ETHPORTS otherwise.
211 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter);
214 * Free some allocated fields of the iterator.
216 * This function is automatically called by rte_eth_iterator_next()
217 * on the last iteration (i.e. when no more matching port is found).
219 * It is safe to call this function twice; it will do nothing more.
222 * Device iterator handle initialized by rte_eth_iterator_init().
223 * The fields bus_str and cls_str are freed if needed.
225 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter);
228 * Macro to iterate over all ethdev ports matching some devargs.
230 * If a break is done before the end of the loop,
231 * the function rte_eth_iterator_cleanup() must be called.
234 * Iterated port id of type uint16_t.
236 * Device parameters input as string of type char*.
238 * Iterator handle of type struct rte_dev_iterator, used internally.
240 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \
241 for (rte_eth_iterator_init(iter, devargs), \
242 id = rte_eth_iterator_next(iter); \
243 id != RTE_MAX_ETHPORTS; \
244 id = rte_eth_iterator_next(iter))
247 * A structure used to retrieve statistics for an Ethernet port.
248 * Not all statistics fields in struct rte_eth_stats are supported
249 * by any type of network interface card (NIC). If any statistics
250 * field is not supported, its value is 0.
251 * All byte-related statistics do not include Ethernet FCS regardless
252 * of whether these bytes have been delivered to the application
253 * (see DEV_RX_OFFLOAD_KEEP_CRC).
255 struct rte_eth_stats {
256 uint64_t ipackets; /**< Total number of successfully received packets. */
257 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
258 uint64_t ibytes; /**< Total number of successfully received bytes. */
259 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
261 * Total of Rx packets dropped by the HW,
262 * because there are no available buffer (i.e. Rx queues are full).
265 uint64_t ierrors; /**< Total number of erroneous received packets. */
266 uint64_t oerrors; /**< Total number of failed transmitted packets. */
267 uint64_t rx_nombuf; /**< Total number of Rx mbuf allocation failures. */
268 /* Queue stats are limited to max 256 queues */
269 /** Total number of queue Rx packets. */
270 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
271 /** Total number of queue Tx packets. */
272 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
273 /** Total number of successfully received queue bytes. */
274 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
275 /** Total number of successfully transmitted queue bytes. */
276 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
277 /** Total number of queue packets received that are dropped. */
278 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
281 /**@{@name Link speed capabilities
282 * Device supported speeds bitmap flags
284 #define ETH_LINK_SPEED_AUTONEG 0 /**< Autonegotiate (all speeds) */
285 #define ETH_LINK_SPEED_FIXED RTE_BIT32(0) /**< Disable autoneg (fixed speed) */
286 #define ETH_LINK_SPEED_10M_HD RTE_BIT32(1) /**< 10 Mbps half-duplex */
287 #define ETH_LINK_SPEED_10M RTE_BIT32(2) /**< 10 Mbps full-duplex */
288 #define ETH_LINK_SPEED_100M_HD RTE_BIT32(3) /**< 100 Mbps half-duplex */
289 #define ETH_LINK_SPEED_100M RTE_BIT32(4) /**< 100 Mbps full-duplex */
290 #define ETH_LINK_SPEED_1G RTE_BIT32(5) /**< 1 Gbps */
291 #define ETH_LINK_SPEED_2_5G RTE_BIT32(6) /**< 2.5 Gbps */
292 #define ETH_LINK_SPEED_5G RTE_BIT32(7) /**< 5 Gbps */
293 #define ETH_LINK_SPEED_10G RTE_BIT32(8) /**< 10 Gbps */
294 #define ETH_LINK_SPEED_20G RTE_BIT32(9) /**< 20 Gbps */
295 #define ETH_LINK_SPEED_25G RTE_BIT32(10) /**< 25 Gbps */
296 #define ETH_LINK_SPEED_40G RTE_BIT32(11) /**< 40 Gbps */
297 #define ETH_LINK_SPEED_50G RTE_BIT32(12) /**< 50 Gbps */
298 #define ETH_LINK_SPEED_56G RTE_BIT32(13) /**< 56 Gbps */
299 #define ETH_LINK_SPEED_100G RTE_BIT32(14) /**< 100 Gbps */
300 #define ETH_LINK_SPEED_200G RTE_BIT32(15) /**< 200 Gbps */
303 /**@{@name Link speed
304 * Ethernet numeric link speeds in Mbps
306 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
307 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
308 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
309 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
310 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
311 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
312 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
313 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
314 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
315 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
316 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
317 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
318 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
319 #define ETH_SPEED_NUM_200G 200000 /**< 200 Gbps */
320 #define ETH_SPEED_NUM_UNKNOWN UINT32_MAX /**< Unknown */
324 * A structure used to retrieve link-level information of an Ethernet port.
327 struct rte_eth_link {
328 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
329 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
330 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
331 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
332 } __rte_aligned(8); /**< aligned for atomic64 read/write */
334 /**@{@name Link negotiation
335 * Constants used in link management.
337 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
338 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
339 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
340 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
341 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
342 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
343 #define RTE_ETH_LINK_MAX_STR_LEN 40 /**< Max length of default link string. */
347 * A structure used to configure the ring threshold registers of an Rx/Tx
348 * queue for an Ethernet port.
350 struct rte_eth_thresh {
351 uint8_t pthresh; /**< Ring prefetch threshold. */
352 uint8_t hthresh; /**< Ring host threshold. */
353 uint8_t wthresh; /**< Ring writeback threshold. */
356 /**@{@name Multi-queue mode
357 * @see rte_eth_conf.rxmode.mq_mode.
359 #define ETH_MQ_RX_RSS_FLAG 0x1 /**< Enable RSS. @see rte_eth_rss_conf */
360 #define ETH_MQ_RX_DCB_FLAG 0x2 /**< Enable DCB. */
361 #define ETH_MQ_RX_VMDQ_FLAG 0x4 /**< Enable VMDq. */
365 * A set of values to identify what method is to be used to route
366 * packets to multiple queues.
368 enum rte_eth_rx_mq_mode {
369 /** None of DCB,RSS or VMDQ mode */
372 /** For Rx side, only RSS is on */
373 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
374 /** For Rx side,only DCB is on. */
375 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
376 /** Both DCB and RSS enable */
377 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
379 /** Only VMDQ, no RSS nor DCB */
380 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
381 /** RSS mode with VMDQ */
382 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
383 /** Use VMDQ+DCB to route traffic to queues */
384 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
385 /** Enable both VMDQ and DCB in VMDq */
386 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
391 * for Rx mq mode backward compatible
393 #define ETH_RSS ETH_MQ_RX_RSS
394 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
395 #define ETH_DCB_RX ETH_MQ_RX_DCB
398 * A set of values to identify what method is to be used to transmit
399 * packets using multi-TCs.
401 enum rte_eth_tx_mq_mode {
402 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
403 ETH_MQ_TX_DCB, /**< For Tx side,only DCB is on. */
404 ETH_MQ_TX_VMDQ_DCB, /**< For Tx side,both DCB and VT is on. */
405 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
409 * for Tx mq mode backward compatible
411 #define ETH_DCB_NONE ETH_MQ_TX_NONE
412 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
413 #define ETH_DCB_TX ETH_MQ_TX_DCB
416 * A structure used to configure the Rx features of an Ethernet port.
418 struct rte_eth_rxmode {
419 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
420 enum rte_eth_rx_mq_mode mq_mode;
421 uint32_t mtu; /**< Requested MTU. */
422 /** Maximum allowed size of LRO aggregated packet. */
423 uint32_t max_lro_pkt_size;
424 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
426 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
427 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
428 * structure are allowed to be set.
432 uint64_t reserved_64s[2]; /**< Reserved for future fields */
433 void *reserved_ptrs[2]; /**< Reserved for future fields */
437 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
438 * Note that single VLAN is treated the same as inner VLAN.
441 ETH_VLAN_TYPE_UNKNOWN = 0,
442 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
443 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
448 * A structure used to describe a vlan filter.
449 * If the bit corresponding to a VID is set, such VID is on.
451 struct rte_vlan_filter_conf {
456 * A structure used to configure the Receive Side Scaling (RSS) feature
457 * of an Ethernet port.
458 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
459 * to an array holding the RSS key to use for hashing specific header
460 * fields of received packets. The length of this array should be indicated
461 * by *rss_key_len* below. Otherwise, a default random hash key is used by
464 * The *rss_key_len* field of the *rss_conf* structure indicates the length
465 * in bytes of the array pointed by *rss_key*. To be compatible, this length
466 * will be checked in i40e only. Others assume 40 bytes to be used as before.
468 * The *rss_hf* field of the *rss_conf* structure indicates the different
469 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
470 * Supplying an *rss_hf* equal to zero disables the RSS feature.
472 struct rte_eth_rss_conf {
473 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
474 uint8_t rss_key_len; /**< hash key length in bytes. */
475 uint64_t rss_hf; /**< Hash functions to apply - see below. */
479 * A packet can be identified by hardware as different flow types. Different
480 * NIC hardware may support different flow types.
481 * Basically, the NIC hardware identifies the flow type as deep protocol as
482 * possible, and exclusively. For example, if a packet is identified as
483 * 'RTE_ETH_FLOW_NONFRAG_IPV4_TCP', it will not be any of other flow types,
484 * though it is an actual IPV4 packet.
486 #define RTE_ETH_FLOW_UNKNOWN 0
487 #define RTE_ETH_FLOW_RAW 1
488 #define RTE_ETH_FLOW_IPV4 2
489 #define RTE_ETH_FLOW_FRAG_IPV4 3
490 #define RTE_ETH_FLOW_NONFRAG_IPV4_TCP 4
491 #define RTE_ETH_FLOW_NONFRAG_IPV4_UDP 5
492 #define RTE_ETH_FLOW_NONFRAG_IPV4_SCTP 6
493 #define RTE_ETH_FLOW_NONFRAG_IPV4_OTHER 7
494 #define RTE_ETH_FLOW_IPV6 8
495 #define RTE_ETH_FLOW_FRAG_IPV6 9
496 #define RTE_ETH_FLOW_NONFRAG_IPV6_TCP 10
497 #define RTE_ETH_FLOW_NONFRAG_IPV6_UDP 11
498 #define RTE_ETH_FLOW_NONFRAG_IPV6_SCTP 12
499 #define RTE_ETH_FLOW_NONFRAG_IPV6_OTHER 13
500 #define RTE_ETH_FLOW_L2_PAYLOAD 14
501 #define RTE_ETH_FLOW_IPV6_EX 15
502 #define RTE_ETH_FLOW_IPV6_TCP_EX 16
503 #define RTE_ETH_FLOW_IPV6_UDP_EX 17
504 /** Consider device port number as a flow differentiator */
505 #define RTE_ETH_FLOW_PORT 18
506 #define RTE_ETH_FLOW_VXLAN 19 /**< VXLAN protocol based flow */
507 #define RTE_ETH_FLOW_GENEVE 20 /**< GENEVE protocol based flow */
508 #define RTE_ETH_FLOW_NVGRE 21 /**< NVGRE protocol based flow */
509 #define RTE_ETH_FLOW_VXLAN_GPE 22 /**< VXLAN-GPE protocol based flow */
510 #define RTE_ETH_FLOW_GTPU 23 /**< GTPU protocol based flow */
511 #define RTE_ETH_FLOW_MAX 24
514 * Below macros are defined for RSS offload types, they can be used to
515 * fill rte_eth_rss_conf.rss_hf or rte_flow_action_rss.types.
517 #define ETH_RSS_IPV4 RTE_BIT64(2)
518 #define ETH_RSS_FRAG_IPV4 RTE_BIT64(3)
519 #define ETH_RSS_NONFRAG_IPV4_TCP RTE_BIT64(4)
520 #define ETH_RSS_NONFRAG_IPV4_UDP RTE_BIT64(5)
521 #define ETH_RSS_NONFRAG_IPV4_SCTP RTE_BIT64(6)
522 #define ETH_RSS_NONFRAG_IPV4_OTHER RTE_BIT64(7)
523 #define ETH_RSS_IPV6 RTE_BIT64(8)
524 #define ETH_RSS_FRAG_IPV6 RTE_BIT64(9)
525 #define ETH_RSS_NONFRAG_IPV6_TCP RTE_BIT64(10)
526 #define ETH_RSS_NONFRAG_IPV6_UDP RTE_BIT64(11)
527 #define ETH_RSS_NONFRAG_IPV6_SCTP RTE_BIT64(12)
528 #define ETH_RSS_NONFRAG_IPV6_OTHER RTE_BIT64(13)
529 #define ETH_RSS_L2_PAYLOAD RTE_BIT64(14)
530 #define ETH_RSS_IPV6_EX RTE_BIT64(15)
531 #define ETH_RSS_IPV6_TCP_EX RTE_BIT64(16)
532 #define ETH_RSS_IPV6_UDP_EX RTE_BIT64(17)
533 #define ETH_RSS_PORT RTE_BIT64(18)
534 #define ETH_RSS_VXLAN RTE_BIT64(19)
535 #define ETH_RSS_GENEVE RTE_BIT64(20)
536 #define ETH_RSS_NVGRE RTE_BIT64(21)
537 #define ETH_RSS_GTPU RTE_BIT64(23)
538 #define ETH_RSS_ETH RTE_BIT64(24)
539 #define ETH_RSS_S_VLAN RTE_BIT64(25)
540 #define ETH_RSS_C_VLAN RTE_BIT64(26)
541 #define ETH_RSS_ESP RTE_BIT64(27)
542 #define ETH_RSS_AH RTE_BIT64(28)
543 #define ETH_RSS_L2TPV3 RTE_BIT64(29)
544 #define ETH_RSS_PFCP RTE_BIT64(30)
545 #define ETH_RSS_PPPOE RTE_BIT64(31)
546 #define ETH_RSS_ECPRI RTE_BIT64(32)
547 #define ETH_RSS_MPLS RTE_BIT64(33)
548 #define ETH_RSS_IPV4_CHKSUM RTE_BIT64(34)
551 * The ETH_RSS_L4_CHKSUM works on checksum field of any L4 header.
552 * It is similar to ETH_RSS_PORT that they don't specify the specific type of
553 * L4 header. This macro is defined to replace some specific L4 (TCP/UDP/SCTP)
554 * checksum type for constructing the use of RSS offload bits.
556 * Due to above reason, some old APIs (and configuration) don't support
557 * ETH_RSS_L4_CHKSUM. The rte_flow RSS API supports it.
559 * For the case that checksum is not used in an UDP header,
560 * it takes the reserved value 0 as input for the hash function.
562 #define ETH_RSS_L4_CHKSUM RTE_BIT64(35)
565 * We use the following macros to combine with above ETH_RSS_* for
566 * more specific input set selection. These bits are defined starting
567 * from the high end of the 64 bits.
568 * Note: If we use above ETH_RSS_* without SRC/DST_ONLY, it represents
569 * both SRC and DST are taken into account. If SRC_ONLY and DST_ONLY of
570 * the same level are used simultaneously, it is the same case as none of
573 #define ETH_RSS_L3_SRC_ONLY RTE_BIT64(63)
574 #define ETH_RSS_L3_DST_ONLY RTE_BIT64(62)
575 #define ETH_RSS_L4_SRC_ONLY RTE_BIT64(61)
576 #define ETH_RSS_L4_DST_ONLY RTE_BIT64(60)
577 #define ETH_RSS_L2_SRC_ONLY RTE_BIT64(59)
578 #define ETH_RSS_L2_DST_ONLY RTE_BIT64(58)
581 * Only select IPV6 address prefix as RSS input set according to
582 * https://tools.ietf.org/html/rfc6052
583 * Must be combined with ETH_RSS_IPV6, ETH_RSS_NONFRAG_IPV6_UDP,
584 * ETH_RSS_NONFRAG_IPV6_TCP, ETH_RSS_NONFRAG_IPV6_SCTP.
586 #define RTE_ETH_RSS_L3_PRE32 RTE_BIT64(57)
587 #define RTE_ETH_RSS_L3_PRE40 RTE_BIT64(56)
588 #define RTE_ETH_RSS_L3_PRE48 RTE_BIT64(55)
589 #define RTE_ETH_RSS_L3_PRE56 RTE_BIT64(54)
590 #define RTE_ETH_RSS_L3_PRE64 RTE_BIT64(53)
591 #define RTE_ETH_RSS_L3_PRE96 RTE_BIT64(52)
594 * Use the following macros to combine with the above layers
595 * to choose inner and outer layers or both for RSS computation.
596 * Bits 50 and 51 are reserved for this.
600 * level 0, requests the default behavior.
601 * Depending on the packet type, it can mean outermost, innermost,
602 * anything in between or even no RSS.
603 * It basically stands for the innermost encapsulation level RSS
604 * can be performed on according to PMD and device capabilities.
606 #define ETH_RSS_LEVEL_PMD_DEFAULT (0ULL << 50)
609 * level 1, requests RSS to be performed on the outermost packet
610 * encapsulation level.
612 #define ETH_RSS_LEVEL_OUTERMOST (1ULL << 50)
615 * level 2, requests RSS to be performed on the specified inner packet
616 * encapsulation level, from outermost to innermost (lower to higher values).
618 #define ETH_RSS_LEVEL_INNERMOST (2ULL << 50)
619 #define ETH_RSS_LEVEL_MASK (3ULL << 50)
621 #define ETH_RSS_LEVEL(rss_hf) ((rss_hf & ETH_RSS_LEVEL_MASK) >> 50)
624 * For input set change of hash filter, if SRC_ONLY and DST_ONLY of
625 * the same level are used simultaneously, it is the same case as
626 * none of them are added.
629 * RSS types with SRC/DST_ONLY.
633 static inline uint64_t
634 rte_eth_rss_hf_refine(uint64_t rss_hf)
636 if ((rss_hf & ETH_RSS_L3_SRC_ONLY) && (rss_hf & ETH_RSS_L3_DST_ONLY))
637 rss_hf &= ~(ETH_RSS_L3_SRC_ONLY | ETH_RSS_L3_DST_ONLY);
639 if ((rss_hf & ETH_RSS_L4_SRC_ONLY) && (rss_hf & ETH_RSS_L4_DST_ONLY))
640 rss_hf &= ~(ETH_RSS_L4_SRC_ONLY | ETH_RSS_L4_DST_ONLY);
645 #define ETH_RSS_IPV6_PRE32 ( \
647 RTE_ETH_RSS_L3_PRE32)
649 #define ETH_RSS_IPV6_PRE40 ( \
651 RTE_ETH_RSS_L3_PRE40)
653 #define ETH_RSS_IPV6_PRE48 ( \
655 RTE_ETH_RSS_L3_PRE48)
657 #define ETH_RSS_IPV6_PRE56 ( \
659 RTE_ETH_RSS_L3_PRE56)
661 #define ETH_RSS_IPV6_PRE64 ( \
663 RTE_ETH_RSS_L3_PRE64)
665 #define ETH_RSS_IPV6_PRE96 ( \
667 RTE_ETH_RSS_L3_PRE96)
669 #define ETH_RSS_IPV6_PRE32_UDP ( \
670 ETH_RSS_NONFRAG_IPV6_UDP | \
671 RTE_ETH_RSS_L3_PRE32)
673 #define ETH_RSS_IPV6_PRE40_UDP ( \
674 ETH_RSS_NONFRAG_IPV6_UDP | \
675 RTE_ETH_RSS_L3_PRE40)
677 #define ETH_RSS_IPV6_PRE48_UDP ( \
678 ETH_RSS_NONFRAG_IPV6_UDP | \
679 RTE_ETH_RSS_L3_PRE48)
681 #define ETH_RSS_IPV6_PRE56_UDP ( \
682 ETH_RSS_NONFRAG_IPV6_UDP | \
683 RTE_ETH_RSS_L3_PRE56)
685 #define ETH_RSS_IPV6_PRE64_UDP ( \
686 ETH_RSS_NONFRAG_IPV6_UDP | \
687 RTE_ETH_RSS_L3_PRE64)
689 #define ETH_RSS_IPV6_PRE96_UDP ( \
690 ETH_RSS_NONFRAG_IPV6_UDP | \
691 RTE_ETH_RSS_L3_PRE96)
693 #define ETH_RSS_IPV6_PRE32_TCP ( \
694 ETH_RSS_NONFRAG_IPV6_TCP | \
695 RTE_ETH_RSS_L3_PRE32)
697 #define ETH_RSS_IPV6_PRE40_TCP ( \
698 ETH_RSS_NONFRAG_IPV6_TCP | \
699 RTE_ETH_RSS_L3_PRE40)
701 #define ETH_RSS_IPV6_PRE48_TCP ( \
702 ETH_RSS_NONFRAG_IPV6_TCP | \
703 RTE_ETH_RSS_L3_PRE48)
705 #define ETH_RSS_IPV6_PRE56_TCP ( \
706 ETH_RSS_NONFRAG_IPV6_TCP | \
707 RTE_ETH_RSS_L3_PRE56)
709 #define ETH_RSS_IPV6_PRE64_TCP ( \
710 ETH_RSS_NONFRAG_IPV6_TCP | \
711 RTE_ETH_RSS_L3_PRE64)
713 #define ETH_RSS_IPV6_PRE96_TCP ( \
714 ETH_RSS_NONFRAG_IPV6_TCP | \
715 RTE_ETH_RSS_L3_PRE96)
717 #define ETH_RSS_IPV6_PRE32_SCTP ( \
718 ETH_RSS_NONFRAG_IPV6_SCTP | \
719 RTE_ETH_RSS_L3_PRE32)
721 #define ETH_RSS_IPV6_PRE40_SCTP ( \
722 ETH_RSS_NONFRAG_IPV6_SCTP | \
723 RTE_ETH_RSS_L3_PRE40)
725 #define ETH_RSS_IPV6_PRE48_SCTP ( \
726 ETH_RSS_NONFRAG_IPV6_SCTP | \
727 RTE_ETH_RSS_L3_PRE48)
729 #define ETH_RSS_IPV6_PRE56_SCTP ( \
730 ETH_RSS_NONFRAG_IPV6_SCTP | \
731 RTE_ETH_RSS_L3_PRE56)
733 #define ETH_RSS_IPV6_PRE64_SCTP ( \
734 ETH_RSS_NONFRAG_IPV6_SCTP | \
735 RTE_ETH_RSS_L3_PRE64)
737 #define ETH_RSS_IPV6_PRE96_SCTP ( \
738 ETH_RSS_NONFRAG_IPV6_SCTP | \
739 RTE_ETH_RSS_L3_PRE96)
741 #define ETH_RSS_IP ( \
743 ETH_RSS_FRAG_IPV4 | \
744 ETH_RSS_NONFRAG_IPV4_OTHER | \
746 ETH_RSS_FRAG_IPV6 | \
747 ETH_RSS_NONFRAG_IPV6_OTHER | \
750 #define ETH_RSS_UDP ( \
751 ETH_RSS_NONFRAG_IPV4_UDP | \
752 ETH_RSS_NONFRAG_IPV6_UDP | \
755 #define ETH_RSS_TCP ( \
756 ETH_RSS_NONFRAG_IPV4_TCP | \
757 ETH_RSS_NONFRAG_IPV6_TCP | \
760 #define ETH_RSS_SCTP ( \
761 ETH_RSS_NONFRAG_IPV4_SCTP | \
762 ETH_RSS_NONFRAG_IPV6_SCTP)
764 #define ETH_RSS_TUNNEL ( \
769 #define ETH_RSS_VLAN ( \
773 /** Mask of valid RSS hash protocols */
774 #define ETH_RSS_PROTO_MASK ( \
776 ETH_RSS_FRAG_IPV4 | \
777 ETH_RSS_NONFRAG_IPV4_TCP | \
778 ETH_RSS_NONFRAG_IPV4_UDP | \
779 ETH_RSS_NONFRAG_IPV4_SCTP | \
780 ETH_RSS_NONFRAG_IPV4_OTHER | \
782 ETH_RSS_FRAG_IPV6 | \
783 ETH_RSS_NONFRAG_IPV6_TCP | \
784 ETH_RSS_NONFRAG_IPV6_UDP | \
785 ETH_RSS_NONFRAG_IPV6_SCTP | \
786 ETH_RSS_NONFRAG_IPV6_OTHER | \
787 ETH_RSS_L2_PAYLOAD | \
789 ETH_RSS_IPV6_TCP_EX | \
790 ETH_RSS_IPV6_UDP_EX | \
798 * Definitions used for redirection table entry size.
799 * Some RSS RETA sizes may not be supported by some drivers, check the
800 * documentation or the description of relevant functions for more details.
802 #define ETH_RSS_RETA_SIZE_64 64
803 #define ETH_RSS_RETA_SIZE_128 128
804 #define ETH_RSS_RETA_SIZE_256 256
805 #define ETH_RSS_RETA_SIZE_512 512
806 #define RTE_RETA_GROUP_SIZE 64
808 /**@{@name VMDq and DCB maximums */
809 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
810 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
811 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
812 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
815 /**@{@name DCB capabilities */
816 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
817 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
820 /**@{@name VLAN offload bits */
821 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
822 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
823 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
824 #define ETH_QINQ_STRIP_OFFLOAD 0x0008 /**< QINQ Strip On/Off */
826 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
827 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
828 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
829 #define ETH_QINQ_STRIP_MASK 0x0008 /**< QINQ Strip setting mask */
830 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
833 /* Definitions used for receive MAC address */
834 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
836 /* Definitions used for unicast hash */
837 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
839 /**@{@name VMDq Rx mode
840 * @see rte_eth_vmdq_rx_conf.rx_mode
842 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
843 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
844 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
845 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
846 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
850 * A structure used to configure 64 entries of Redirection Table of the
851 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
852 * more than 64 entries supported by hardware, an array of this structure
855 struct rte_eth_rss_reta_entry64 {
856 /** Mask bits indicate which entries need to be updated/queried. */
858 /** Group of 64 redirection table entries. */
859 uint16_t reta[RTE_RETA_GROUP_SIZE];
863 * This enum indicates the possible number of traffic classes
864 * in DCB configurations
866 enum rte_eth_nb_tcs {
867 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
868 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
872 * This enum indicates the possible number of queue pools
873 * in VMDQ configurations.
875 enum rte_eth_nb_pools {
876 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
877 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
878 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
879 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
882 /* This structure may be extended in future. */
883 struct rte_eth_dcb_rx_conf {
884 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
885 /** Traffic class each UP mapped to. */
886 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
889 struct rte_eth_vmdq_dcb_tx_conf {
890 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
891 /** Traffic class each UP mapped to. */
892 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
895 struct rte_eth_dcb_tx_conf {
896 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
897 /** Traffic class each UP mapped to. */
898 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
901 struct rte_eth_vmdq_tx_conf {
902 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
906 * A structure used to configure the VMDQ+DCB feature
907 * of an Ethernet port.
909 * Using this feature, packets are routed to a pool of queues, based
910 * on the vlan id in the vlan tag, and then to a specific queue within
911 * that pool, using the user priority vlan tag field.
913 * A default pool may be used, if desired, to route all traffic which
914 * does not match the vlan filter rules.
916 struct rte_eth_vmdq_dcb_conf {
917 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
918 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
919 uint8_t default_pool; /**< The default pool, if applicable */
920 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
922 uint16_t vlan_id; /**< The vlan id of the received frame */
923 uint64_t pools; /**< Bitmask of pools for packet Rx */
924 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
925 /** Selects a queue in a pool */
926 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
930 * A structure used to configure the VMDQ feature of an Ethernet port when
931 * not combined with the DCB feature.
933 * Using this feature, packets are routed to a pool of queues. By default,
934 * the pool selection is based on the MAC address, the vlan id in the
935 * vlan tag as specified in the pool_map array.
936 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
937 * selection using only the MAC address. MAC address to pool mapping is done
938 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
939 * corresponding to the pool id.
941 * Queue selection within the selected pool will be done using RSS when
942 * it is enabled or revert to the first queue of the pool if not.
944 * A default pool may be used, if desired, to route all traffic which
945 * does not match the vlan filter rules or any pool MAC address.
947 struct rte_eth_vmdq_rx_conf {
948 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
949 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
950 uint8_t default_pool; /**< The default pool, if applicable */
951 uint8_t enable_loop_back; /**< Enable VT loop back */
952 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
953 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
955 uint16_t vlan_id; /**< The vlan id of the received frame */
956 uint64_t pools; /**< Bitmask of pools for packet Rx */
957 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
961 * A structure used to configure the Tx features of an Ethernet port.
963 struct rte_eth_txmode {
964 enum rte_eth_tx_mq_mode mq_mode; /**< Tx multi-queues mode. */
966 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
967 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
968 * structure are allowed to be set.
974 uint8_t /** If set, reject sending out tagged pkts */
975 hw_vlan_reject_tagged : 1,
976 /** If set, reject sending out untagged pkts */
977 hw_vlan_reject_untagged : 1,
978 /** If set, enable port based VLAN insertion */
979 hw_vlan_insert_pvid : 1;
981 uint64_t reserved_64s[2]; /**< Reserved for future fields */
982 void *reserved_ptrs[2]; /**< Reserved for future fields */
987 * @b EXPERIMENTAL: this structure may change without prior notice.
989 * A structure used to configure an Rx packet segment to split.
991 * If RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT flag is set in offloads field,
992 * the PMD will split the received packets into multiple segments
993 * according to the specification in the description array:
995 * - The first network buffer will be allocated from the memory pool,
996 * specified in the first array element, the second buffer, from the
997 * pool in the second element, and so on.
999 * - The offsets from the segment description elements specify
1000 * the data offset from the buffer beginning except the first mbuf.
1001 * The first segment offset is added with RTE_PKTMBUF_HEADROOM.
1003 * - The lengths in the elements define the maximal data amount
1004 * being received to each segment. The receiving starts with filling
1005 * up the first mbuf data buffer up to specified length. If the
1006 * there are data remaining (packet is longer than buffer in the first
1007 * mbuf) the following data will be pushed to the next segment
1008 * up to its own length, and so on.
1010 * - If the length in the segment description element is zero
1011 * the actual buffer size will be deduced from the appropriate
1012 * memory pool properties.
1014 * - If there is not enough elements to describe the buffer for entire
1015 * packet of maximal length the following parameters will be used
1016 * for the all remaining segments:
1017 * - pool from the last valid element
1018 * - the buffer size from this pool
1021 struct rte_eth_rxseg_split {
1022 struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
1023 uint16_t length; /**< Segment data length, configures split point. */
1024 uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */
1025 uint32_t reserved; /**< Reserved field. */
1030 * @b EXPERIMENTAL: this structure may change without prior notice.
1032 * A common structure used to describe Rx packet segment properties.
1034 union rte_eth_rxseg {
1035 /* The settings for buffer split offload. */
1036 struct rte_eth_rxseg_split split;
1037 /* The other features settings should be added here. */
1041 * A structure used to configure an Rx ring of an Ethernet port.
1043 struct rte_eth_rxconf {
1044 struct rte_eth_thresh rx_thresh; /**< Rx ring threshold registers. */
1045 uint16_t rx_free_thresh; /**< Drives the freeing of Rx descriptors. */
1046 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
1047 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1048 uint16_t rx_nseg; /**< Number of descriptions in rx_seg array. */
1050 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
1051 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
1052 * fields on rte_eth_dev_info structure are allowed to be set.
1056 * Points to the array of segment descriptions for an entire packet.
1057 * Array elements are properties for consecutive Rx segments.
1059 * The supported capabilities of receiving segmentation is reported
1060 * in rte_eth_dev_info.rx_seg_capa field.
1062 union rte_eth_rxseg *rx_seg;
1064 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1065 void *reserved_ptrs[2]; /**< Reserved for future fields */
1069 * A structure used to configure a Tx ring of an Ethernet port.
1071 struct rte_eth_txconf {
1072 struct rte_eth_thresh tx_thresh; /**< Tx ring threshold registers. */
1073 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
1074 uint16_t tx_free_thresh; /**< Start freeing Tx buffers if there are
1075 less free descriptors than this value. */
1077 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1079 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
1080 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
1081 * fields on rte_eth_dev_info structure are allowed to be set.
1085 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1086 void *reserved_ptrs[2]; /**< Reserved for future fields */
1091 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1093 * A structure used to return the hairpin capabilities that are supported.
1095 struct rte_eth_hairpin_cap {
1096 /** The max number of hairpin queues (different bindings). */
1097 uint16_t max_nb_queues;
1098 /** Max number of Rx queues to be connected to one Tx queue. */
1099 uint16_t max_rx_2_tx;
1100 /** Max number of Tx queues to be connected to one Rx queue. */
1101 uint16_t max_tx_2_rx;
1102 uint16_t max_nb_desc; /**< The max num of descriptors. */
1105 #define RTE_ETH_MAX_HAIRPIN_PEERS 32
1109 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1111 * A structure used to hold hairpin peer data.
1113 struct rte_eth_hairpin_peer {
1114 uint16_t port; /**< Peer port. */
1115 uint16_t queue; /**< Peer queue. */
1120 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1122 * A structure used to configure hairpin binding.
1124 struct rte_eth_hairpin_conf {
1125 uint32_t peer_count:16; /**< The number of peers. */
1128 * Explicit Tx flow rule mode.
1129 * One hairpin pair of queues should have the same attribute.
1131 * - When set, the user should be responsible for inserting the hairpin
1132 * Tx part flows and removing them.
1133 * - When clear, the PMD will try to handle the Tx part of the flows,
1134 * e.g., by splitting one flow into two parts.
1136 uint32_t tx_explicit:1;
1139 * Manually bind hairpin queues.
1140 * One hairpin pair of queues should have the same attribute.
1142 * - When set, to enable hairpin, the user should call the hairpin bind
1143 * function after all the queues are set up properly and the ports are
1144 * started. Also, the hairpin unbind function should be called
1145 * accordingly before stopping a port that with hairpin configured.
1146 * - When clear, the PMD will try to enable the hairpin with the queues
1147 * configured automatically during port start.
1149 uint32_t manual_bind:1;
1150 uint32_t reserved:14; /**< Reserved bits. */
1151 struct rte_eth_hairpin_peer peers[RTE_ETH_MAX_HAIRPIN_PEERS];
1155 * A structure contains information about HW descriptor ring limitations.
1157 struct rte_eth_desc_lim {
1158 uint16_t nb_max; /**< Max allowed number of descriptors. */
1159 uint16_t nb_min; /**< Min allowed number of descriptors. */
1160 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
1163 * Max allowed number of segments per whole packet.
1165 * - For TSO packet this is the total number of data descriptors allowed
1168 * @see nb_mtu_seg_max
1170 uint16_t nb_seg_max;
1173 * Max number of segments per one MTU.
1175 * - For non-TSO packet, this is the maximum allowed number of segments
1176 * in a single transmit packet.
1178 * - For TSO packet each segment within the TSO may span up to this
1183 uint16_t nb_mtu_seg_max;
1187 * This enum indicates the flow control mode
1189 enum rte_eth_fc_mode {
1190 RTE_FC_NONE = 0, /**< Disable flow control. */
1191 RTE_FC_RX_PAUSE, /**< Rx pause frame, enable flowctrl on Tx side. */
1192 RTE_FC_TX_PAUSE, /**< Tx pause frame, enable flowctrl on Rx side. */
1193 RTE_FC_FULL /**< Enable flow control on both side. */
1197 * A structure used to configure Ethernet flow control parameter.
1198 * These parameters will be configured into the register of the NIC.
1199 * Please refer to the corresponding data sheet for proper value.
1201 struct rte_eth_fc_conf {
1202 uint32_t high_water; /**< High threshold value to trigger XOFF */
1203 uint32_t low_water; /**< Low threshold value to trigger XON */
1204 uint16_t pause_time; /**< Pause quota in the Pause frame */
1205 uint16_t send_xon; /**< Is XON frame need be sent */
1206 enum rte_eth_fc_mode mode; /**< Link flow control mode */
1207 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
1208 uint8_t autoneg; /**< Use Pause autoneg */
1212 * A structure used to configure Ethernet priority flow control parameter.
1213 * These parameters will be configured into the register of the NIC.
1214 * Please refer to the corresponding data sheet for proper value.
1216 struct rte_eth_pfc_conf {
1217 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
1218 uint8_t priority; /**< VLAN User Priority. */
1222 * Tunnel type for device-specific classifier configuration.
1223 * @see rte_eth_udp_tunnel
1225 enum rte_eth_tunnel_type {
1226 RTE_TUNNEL_TYPE_NONE = 0,
1227 RTE_TUNNEL_TYPE_VXLAN,
1228 RTE_TUNNEL_TYPE_GENEVE,
1229 RTE_TUNNEL_TYPE_TEREDO,
1230 RTE_TUNNEL_TYPE_NVGRE,
1231 RTE_TUNNEL_TYPE_IP_IN_GRE,
1232 RTE_L2_TUNNEL_TYPE_E_TAG,
1233 RTE_TUNNEL_TYPE_VXLAN_GPE,
1234 RTE_TUNNEL_TYPE_ECPRI,
1235 RTE_TUNNEL_TYPE_MAX,
1238 /* Deprecated API file for rte_eth_dev_filter_* functions */
1239 #include "rte_eth_ctrl.h"
1242 * Memory space that can be configured to store Flow Director filters
1243 * in the board memory.
1245 enum rte_fdir_pballoc_type {
1246 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
1247 RTE_FDIR_PBALLOC_128K, /**< 128k. */
1248 RTE_FDIR_PBALLOC_256K, /**< 256k. */
1252 * Select report mode of FDIR hash information in Rx descriptors.
1254 enum rte_fdir_status_mode {
1255 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
1256 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
1257 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
1261 * A structure used to configure the Flow Director (FDIR) feature
1262 * of an Ethernet port.
1264 * If mode is RTE_FDIR_MODE_NONE, the pballoc value is ignored.
1266 struct rte_fdir_conf {
1267 enum rte_fdir_mode mode; /**< Flow Director mode. */
1268 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
1269 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
1270 /** Rx queue of packets matching a "drop" filter in perfect mode. */
1272 struct rte_eth_fdir_masks mask;
1273 /** Flex payload configuration. */
1274 struct rte_eth_fdir_flex_conf flex_conf;
1278 * UDP tunneling configuration.
1280 * Used to configure the classifier of a device,
1281 * associating an UDP port with a type of tunnel.
1283 * Some NICs may need such configuration to properly parse a tunnel
1284 * with any standard or custom UDP port.
1286 struct rte_eth_udp_tunnel {
1287 uint16_t udp_port; /**< UDP port used for the tunnel. */
1288 uint8_t prot_type; /**< Tunnel type. @see rte_eth_tunnel_type */
1292 * A structure used to enable/disable specific device interrupts.
1294 struct rte_intr_conf {
1295 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
1297 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
1299 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
1304 * A structure used to configure an Ethernet port.
1305 * Depending upon the Rx multi-queue mode, extra advanced
1306 * configuration settings may be needed.
1308 struct rte_eth_conf {
1309 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
1310 used. ETH_LINK_SPEED_FIXED disables link
1311 autonegotiation, and a unique speed shall be
1312 set. Otherwise, the bitmap defines the set of
1313 speeds to be advertised. If the special value
1314 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
1315 supported are advertised. */
1316 struct rte_eth_rxmode rxmode; /**< Port Rx configuration. */
1317 struct rte_eth_txmode txmode; /**< Port Tx configuration. */
1318 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
1319 is 0, meaning the loopback mode is disabled.
1320 Read the datasheet of given Ethernet controller
1321 for details. The possible values of this field
1322 are defined in implementation of each driver. */
1324 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
1325 /** Port vmdq+dcb configuration. */
1326 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
1327 /** Port dcb Rx configuration. */
1328 struct rte_eth_dcb_rx_conf dcb_rx_conf;
1329 /** Port vmdq Rx configuration. */
1330 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
1331 } rx_adv_conf; /**< Port Rx filtering configuration. */
1333 /** Port vmdq+dcb Tx configuration. */
1334 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
1335 /** Port dcb Tx configuration. */
1336 struct rte_eth_dcb_tx_conf dcb_tx_conf;
1337 /** Port vmdq Tx configuration. */
1338 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
1339 } tx_adv_conf; /**< Port Tx DCB configuration (union). */
1340 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
1341 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
1342 uint32_t dcb_capability_en;
1343 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. DEPRECATED */
1344 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
1348 * Rx offload capabilities of a device.
1350 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
1351 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
1352 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
1353 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
1354 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
1355 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
1356 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
1357 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
1358 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
1359 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
1360 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
1361 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
1363 * Timestamp is set by the driver in RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1364 * and RTE_MBUF_DYNFLAG_RX_TIMESTAMP_NAME is set in ol_flags.
1365 * The mbuf field and flag are registered when the offload is configured.
1367 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
1368 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
1369 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
1370 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
1371 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
1372 #define DEV_RX_OFFLOAD_RSS_HASH 0x00080000
1373 #define RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT 0x00100000
1375 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
1376 DEV_RX_OFFLOAD_UDP_CKSUM | \
1377 DEV_RX_OFFLOAD_TCP_CKSUM)
1378 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
1379 DEV_RX_OFFLOAD_VLAN_FILTER | \
1380 DEV_RX_OFFLOAD_VLAN_EXTEND | \
1381 DEV_RX_OFFLOAD_QINQ_STRIP)
1384 * If new Rx offload capabilities are defined, they also must be
1385 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
1389 * Tx offload capabilities of a device.
1391 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
1392 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
1393 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
1394 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
1395 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
1396 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
1397 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
1398 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
1399 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
1400 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
1401 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
1402 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
1403 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
1404 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
1406 * Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
1407 * Tx queue without SW lock.
1409 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
1410 /** Device supports multi segment send. */
1411 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
1413 * Device supports optimization for fast release of mbufs.
1414 * When set application must guarantee that per-queue all mbufs comes from
1415 * the same mempool and has refcnt = 1.
1417 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
1418 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
1420 * Device supports generic UDP tunneled packet TSO.
1421 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
1424 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
1426 * Device supports generic IP tunneled packet TSO.
1427 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
1430 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
1431 /** Device supports outer UDP checksum */
1432 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
1434 * Device sends on time read from RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1435 * if RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME is set in ol_flags.
1436 * The mbuf field and flag are registered when the offload is configured.
1438 #define DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP 0x00200000
1440 * If new Tx offload capabilities are defined, they also must be
1441 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1444 /**@{@name Device capabilities
1445 * Non-offload capabilities reported in rte_eth_dev_info.dev_capa.
1447 /** Device supports Rx queue setup after device started. */
1448 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
1449 /** Device supports Tx queue setup after device started. */
1450 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
1454 * Fallback default preferred Rx/Tx port parameters.
1455 * These are used if an application requests default parameters
1456 * but the PMD does not provide preferred values.
1458 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1459 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1460 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1461 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1464 * Preferred Rx/Tx port parameters.
1465 * There are separate instances of this structure for transmission
1466 * and reception respectively.
1468 struct rte_eth_dev_portconf {
1469 uint16_t burst_size; /**< Device-preferred burst size */
1470 uint16_t ring_size; /**< Device-preferred size of queue rings */
1471 uint16_t nb_queues; /**< Device-preferred number of queues */
1475 * Default values for switch domain id when ethdev does not support switch
1476 * domain definitions.
1478 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (UINT16_MAX)
1481 * Ethernet device associated switch information
1483 struct rte_eth_switch_info {
1484 const char *name; /**< switch name */
1485 uint16_t domain_id; /**< switch domain id */
1487 * Mapping to the devices physical switch port as enumerated from the
1488 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1489 * device this may correspond to the VF_ID of each virtual function,
1490 * but each driver should explicitly define the mapping of switch
1491 * port identifier to that physical interconnect/switch
1498 * @b EXPERIMENTAL: this structure may change without prior notice.
1500 * Ethernet device Rx buffer segmentation capabilities.
1502 struct rte_eth_rxseg_capa {
1504 uint32_t multi_pools:1; /**< Supports receiving to multiple pools.*/
1505 uint32_t offset_allowed:1; /**< Supports buffer offsets. */
1506 uint32_t offset_align_log2:4; /**< Required offset alignment. */
1507 uint16_t max_nseg; /**< Maximum amount of segments to split. */
1508 uint16_t reserved; /**< Reserved field. */
1512 * Ethernet device information
1516 * Ethernet device representor port type.
1518 enum rte_eth_representor_type {
1519 RTE_ETH_REPRESENTOR_NONE, /**< not a representor. */
1520 RTE_ETH_REPRESENTOR_VF, /**< representor of Virtual Function. */
1521 RTE_ETH_REPRESENTOR_SF, /**< representor of Sub Function. */
1522 RTE_ETH_REPRESENTOR_PF, /**< representor of Physical Function. */
1526 * A structure used to retrieve the contextual information of
1527 * an Ethernet device, such as the controlling driver of the
1530 struct rte_eth_dev_info {
1531 struct rte_device *device; /** Generic device information */
1532 const char *driver_name; /**< Device Driver name. */
1533 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1534 Use if_indextoname() to translate into an interface name. */
1535 uint16_t min_mtu; /**< Minimum MTU allowed */
1536 uint16_t max_mtu; /**< Maximum MTU allowed */
1537 const uint32_t *dev_flags; /**< Device flags */
1538 uint32_t min_rx_bufsize; /**< Minimum size of Rx buffer. */
1539 uint32_t max_rx_pktlen; /**< Maximum configurable length of Rx pkt. */
1540 /** Maximum configurable size of LRO aggregated packet. */
1541 uint32_t max_lro_pkt_size;
1542 uint16_t max_rx_queues; /**< Maximum number of Rx queues. */
1543 uint16_t max_tx_queues; /**< Maximum number of Tx queues. */
1544 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1545 uint32_t max_hash_mac_addrs;
1546 /** Maximum number of hash MAC addresses for MTA and UTA. */
1547 uint16_t max_vfs; /**< Maximum number of VFs. */
1548 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1549 struct rte_eth_rxseg_capa rx_seg_capa; /**< Segmentation capability.*/
1550 /** All Rx offload capabilities including all per-queue ones */
1551 uint64_t rx_offload_capa;
1552 /** All Tx offload capabilities including all per-queue ones */
1553 uint64_t tx_offload_capa;
1554 /** Device per-queue Rx offload capabilities. */
1555 uint64_t rx_queue_offload_capa;
1556 /** Device per-queue Tx offload capabilities. */
1557 uint64_t tx_queue_offload_capa;
1558 /** Device redirection table size, the total number of entries. */
1560 uint8_t hash_key_size; /**< Hash key size in bytes */
1561 /** Bit mask of RSS offloads, the bit offset also means flow type */
1562 uint64_t flow_type_rss_offloads;
1563 struct rte_eth_rxconf default_rxconf; /**< Default Rx configuration */
1564 struct rte_eth_txconf default_txconf; /**< Default Tx configuration */
1565 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1566 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1567 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1568 struct rte_eth_desc_lim rx_desc_lim; /**< Rx descriptors limits */
1569 struct rte_eth_desc_lim tx_desc_lim; /**< Tx descriptors limits */
1570 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1571 /** Configured number of Rx/Tx queues */
1572 uint16_t nb_rx_queues; /**< Number of Rx queues. */
1573 uint16_t nb_tx_queues; /**< Number of Tx queues. */
1574 /** Rx parameter recommendations */
1575 struct rte_eth_dev_portconf default_rxportconf;
1576 /** Tx parameter recommendations */
1577 struct rte_eth_dev_portconf default_txportconf;
1578 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1581 * Switching information for ports on a device with a
1582 * embedded managed interconnect/switch.
1584 struct rte_eth_switch_info switch_info;
1586 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1587 void *reserved_ptrs[2]; /**< Reserved for future fields */
1590 /**@{@name Rx/Tx queue states */
1591 #define RTE_ETH_QUEUE_STATE_STOPPED 0 /**< Queue stopped. */
1592 #define RTE_ETH_QUEUE_STATE_STARTED 1 /**< Queue started. */
1593 #define RTE_ETH_QUEUE_STATE_HAIRPIN 2 /**< Queue used for hairpin. */
1597 * Ethernet device Rx queue information structure.
1598 * Used to retrieve information about configured queue.
1600 struct rte_eth_rxq_info {
1601 struct rte_mempool *mp; /**< mempool used by that queue. */
1602 struct rte_eth_rxconf conf; /**< queue config parameters. */
1603 uint8_t scattered_rx; /**< scattered packets Rx supported. */
1604 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */
1605 uint16_t nb_desc; /**< configured number of RXDs. */
1606 uint16_t rx_buf_size; /**< hardware receive buffer size. */
1607 } __rte_cache_min_aligned;
1610 * Ethernet device Tx queue information structure.
1611 * Used to retrieve information about configured queue.
1613 struct rte_eth_txq_info {
1614 struct rte_eth_txconf conf; /**< queue config parameters. */
1615 uint16_t nb_desc; /**< configured number of TXDs. */
1616 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */
1617 } __rte_cache_min_aligned;
1619 /* Generic Burst mode flag definition, values can be ORed. */
1622 * If the queues have different burst mode description, this bit will be set
1623 * by PMD, then the application can iterate to retrieve burst description for
1626 #define RTE_ETH_BURST_FLAG_PER_QUEUE RTE_BIT64(0)
1629 * Ethernet device Rx/Tx queue packet burst mode information structure.
1630 * Used to retrieve information about packet burst mode setting.
1632 struct rte_eth_burst_mode {
1633 uint64_t flags; /**< The ORed values of RTE_ETH_BURST_FLAG_xxx */
1635 #define RTE_ETH_BURST_MODE_INFO_SIZE 1024 /**< Maximum size for information */
1636 char info[RTE_ETH_BURST_MODE_INFO_SIZE]; /**< burst mode information */
1639 /** Maximum name length for extended statistics counters */
1640 #define RTE_ETH_XSTATS_NAME_SIZE 64
1643 * An Ethernet device extended statistic structure
1645 * This structure is used by rte_eth_xstats_get() to provide
1646 * statistics that are not provided in the generic *rte_eth_stats*
1648 * It maps a name id, corresponding to an index in the array returned
1649 * by rte_eth_xstats_get_names(), to a statistic value.
1651 struct rte_eth_xstat {
1652 uint64_t id; /**< The index in xstats name array. */
1653 uint64_t value; /**< The statistic counter value. */
1657 * A name element for extended statistics.
1659 * An array of this structure is returned by rte_eth_xstats_get_names().
1660 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1661 * structure references these names by their array index.
1663 * The xstats should follow a common naming scheme.
1664 * Some names are standardized in rte_stats_strings.
1666 * - rx_missed_errors
1668 * - tx_size_128_to_255_packets
1670 struct rte_eth_xstat_name {
1671 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1674 #define ETH_DCB_NUM_TCS 8
1675 #define ETH_MAX_VMDQ_POOL 64
1678 * A structure used to get the information of queue and
1679 * TC mapping on both Tx and Rx paths.
1681 struct rte_eth_dcb_tc_queue_mapping {
1682 /** Rx queues assigned to tc per Pool */
1686 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1687 /** Rx queues assigned to tc per Pool */
1691 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1695 * A structure used to get the information of DCB.
1696 * It includes TC UP mapping and queue TC mapping.
1698 struct rte_eth_dcb_info {
1699 uint8_t nb_tcs; /**< number of TCs */
1700 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1701 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< Tx BW percentage for each TC */
1702 /** Rx queues assigned to tc */
1703 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1707 * This enum indicates the possible Forward Error Correction (FEC) modes
1708 * of an ethdev port.
1710 enum rte_eth_fec_mode {
1711 RTE_ETH_FEC_NOFEC = 0, /**< FEC is off */
1712 RTE_ETH_FEC_AUTO, /**< FEC autonegotiation modes */
1713 RTE_ETH_FEC_BASER, /**< FEC using common algorithm */
1714 RTE_ETH_FEC_RS, /**< FEC using RS algorithm */
1717 /* Translate from FEC mode to FEC capa */
1718 #define RTE_ETH_FEC_MODE_TO_CAPA(x) RTE_BIT32(x)
1720 /* This macro indicates FEC capa mask */
1721 #define RTE_ETH_FEC_MODE_CAPA_MASK(x) RTE_BIT32(RTE_ETH_FEC_ ## x)
1723 /* A structure used to get capabilities per link speed */
1724 struct rte_eth_fec_capa {
1725 uint32_t speed; /**< Link speed (see ETH_SPEED_NUM_*) */
1726 uint32_t capa; /**< FEC capabilities bitmask */
1729 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1731 /* Macros to check for valid port */
1732 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1733 if (!rte_eth_dev_is_valid_port(port_id)) { \
1734 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1739 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1740 if (!rte_eth_dev_is_valid_port(port_id)) { \
1741 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1746 /**@{@name L2 tunnel configuration */
1747 /** L2 tunnel enable mask */
1748 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1749 /** L2 tunnel insertion mask */
1750 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1751 /** L2 tunnel stripping mask */
1752 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1753 /** L2 tunnel forwarding mask */
1754 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1758 * Function type used for Rx packet processing packet callbacks.
1760 * The callback function is called on Rx with a burst of packets that have
1761 * been received on the given port and queue.
1764 * The Ethernet port on which Rx is being performed.
1766 * The queue on the Ethernet port which is being used to receive the packets.
1768 * The burst of packets that have just been received.
1770 * The number of packets in the burst pointed to by "pkts".
1772 * The max number of packets that can be stored in the "pkts" array.
1774 * The arbitrary user parameter passed in by the application when the callback
1775 * was originally configured.
1777 * The number of packets returned to the user.
1779 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1780 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1784 * Function type used for Tx packet processing packet callbacks.
1786 * The callback function is called on Tx with a burst of packets immediately
1787 * before the packets are put onto the hardware queue for transmission.
1790 * The Ethernet port on which Tx is being performed.
1792 * The queue on the Ethernet port which is being used to transmit the packets.
1794 * The burst of packets that are about to be transmitted.
1796 * The number of packets in the burst pointed to by "pkts".
1798 * The arbitrary user parameter passed in by the application when the callback
1799 * was originally configured.
1801 * The number of packets to be written to the NIC.
1803 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1804 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1807 * Possible states of an ethdev port.
1809 enum rte_eth_dev_state {
1810 /** Device is unused before being probed. */
1811 RTE_ETH_DEV_UNUSED = 0,
1812 /** Device is attached when allocated in probing. */
1813 RTE_ETH_DEV_ATTACHED,
1814 /** Device is in removed state when plug-out is detected. */
1815 RTE_ETH_DEV_REMOVED,
1818 struct rte_eth_dev_sriov {
1819 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1820 uint8_t nb_q_per_pool; /**< Rx queue number per pool */
1821 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1822 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1824 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1826 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1828 #define RTE_ETH_DEV_NO_OWNER 0
1830 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1832 struct rte_eth_dev_owner {
1833 uint64_t id; /**< The owner unique identifier. */
1834 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1837 /**@{@name Device flags
1838 * Flags internally saved in rte_eth_dev_data.dev_flags
1839 * and reported in rte_eth_dev_info.dev_flags.
1841 /** PMD supports thread-safe flow operations */
1842 #define RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE 0x0001
1843 /** Device supports link state interrupt */
1844 #define RTE_ETH_DEV_INTR_LSC 0x0002
1845 /** Device is a bonded slave */
1846 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1847 /** Device supports device removal interrupt */
1848 #define RTE_ETH_DEV_INTR_RMV 0x0008
1849 /** Device is port representor */
1850 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1851 /** Device does not support MAC change after started */
1852 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1854 * Queue xstats filled automatically by ethdev layer.
1855 * PMDs filling the queue xstats themselves should not set this flag
1857 #define RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS 0x0040
1861 * Iterates over valid ethdev ports owned by a specific owner.
1864 * The id of the next possible valid owned port.
1866 * The owner identifier.
1867 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1869 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1871 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1872 const uint64_t owner_id);
1875 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1877 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1878 for (p = rte_eth_find_next_owned_by(0, o); \
1879 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1880 p = rte_eth_find_next_owned_by(p + 1, o))
1883 * Iterates over valid ethdev ports.
1886 * The id of the next possible valid port.
1888 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1890 uint16_t rte_eth_find_next(uint16_t port_id);
1893 * Macro to iterate over all enabled and ownerless ethdev ports.
1895 #define RTE_ETH_FOREACH_DEV(p) \
1896 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1899 * Iterates over ethdev ports of a specified device.
1901 * @param port_id_start
1902 * The id of the next possible valid port.
1904 * The generic device behind the ports to iterate.
1906 * Next port id of the device, possibly port_id_start,
1907 * RTE_MAX_ETHPORTS if there is none.
1910 rte_eth_find_next_of(uint16_t port_id_start,
1911 const struct rte_device *parent);
1914 * Macro to iterate over all ethdev ports of a specified device.
1917 * The id of the matching port being iterated.
1919 * The rte_device pointer matching the iterated ports.
1921 #define RTE_ETH_FOREACH_DEV_OF(port_id, parent) \
1922 for (port_id = rte_eth_find_next_of(0, parent); \
1923 port_id < RTE_MAX_ETHPORTS; \
1924 port_id = rte_eth_find_next_of(port_id + 1, parent))
1927 * Iterates over sibling ethdev ports (i.e. sharing the same rte_device).
1929 * @param port_id_start
1930 * The id of the next possible valid sibling port.
1931 * @param ref_port_id
1932 * The id of a reference port to compare rte_device with.
1934 * Next sibling port id, possibly port_id_start or ref_port_id itself,
1935 * RTE_MAX_ETHPORTS if there is none.
1938 rte_eth_find_next_sibling(uint16_t port_id_start, uint16_t ref_port_id);
1941 * Macro to iterate over all ethdev ports sharing the same rte_device
1942 * as the specified port.
1943 * Note: the specified reference port is part of the loop iterations.
1946 * The id of the matching port being iterated.
1947 * @param ref_port_id
1948 * The id of the port being compared.
1950 #define RTE_ETH_FOREACH_DEV_SIBLING(port_id, ref_port_id) \
1951 for (port_id = rte_eth_find_next_sibling(0, ref_port_id); \
1952 port_id < RTE_MAX_ETHPORTS; \
1953 port_id = rte_eth_find_next_sibling(port_id + 1, ref_port_id))
1957 * @b EXPERIMENTAL: this API may change without prior notice.
1959 * Get a new unique owner identifier.
1960 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1961 * to avoid multiple management of device by different entities.
1964 * Owner identifier pointer.
1966 * Negative errno value on error, 0 on success.
1969 int rte_eth_dev_owner_new(uint64_t *owner_id);
1973 * @b EXPERIMENTAL: this API may change without prior notice.
1975 * Set an Ethernet device owner.
1978 * The identifier of the port to own.
1980 * The owner pointer.
1982 * Negative errno value on error, 0 on success.
1985 int rte_eth_dev_owner_set(const uint16_t port_id,
1986 const struct rte_eth_dev_owner *owner);
1990 * @b EXPERIMENTAL: this API may change without prior notice.
1992 * Unset Ethernet device owner to make the device ownerless.
1995 * The identifier of port to make ownerless.
1997 * The owner identifier.
1999 * 0 on success, negative errno value on error.
2002 int rte_eth_dev_owner_unset(const uint16_t port_id,
2003 const uint64_t owner_id);
2007 * @b EXPERIMENTAL: this API may change without prior notice.
2009 * Remove owner from all Ethernet devices owned by a specific owner.
2012 * The owner identifier.
2014 * 0 on success, negative errno value on error.
2017 int rte_eth_dev_owner_delete(const uint64_t owner_id);
2021 * @b EXPERIMENTAL: this API may change without prior notice.
2023 * Get the owner of an Ethernet device.
2026 * The port identifier.
2028 * The owner structure pointer to fill.
2030 * 0 on success, negative errno value on error..
2033 int rte_eth_dev_owner_get(const uint16_t port_id,
2034 struct rte_eth_dev_owner *owner);
2037 * Get the number of ports which are usable for the application.
2039 * These devices must be iterated by using the macro
2040 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
2041 * to deal with non-contiguous ranges of devices.
2044 * The count of available Ethernet devices.
2046 uint16_t rte_eth_dev_count_avail(void);
2049 * Get the total number of ports which are allocated.
2051 * Some devices may not be available for the application.
2054 * The total count of Ethernet devices.
2056 uint16_t rte_eth_dev_count_total(void);
2059 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
2060 * the bitmap link_speeds of the struct rte_eth_conf
2063 * Numerical speed value in Mbps
2065 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
2067 * 0 if the speed cannot be mapped
2069 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
2072 * Get DEV_RX_OFFLOAD_* flag name.
2077 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2079 const char *rte_eth_dev_rx_offload_name(uint64_t offload);
2082 * Get DEV_TX_OFFLOAD_* flag name.
2087 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2089 const char *rte_eth_dev_tx_offload_name(uint64_t offload);
2092 * Configure an Ethernet device.
2093 * This function must be invoked first before any other function in the
2094 * Ethernet API. This function can also be re-invoked when a device is in the
2098 * The port identifier of the Ethernet device to configure.
2099 * @param nb_rx_queue
2100 * The number of receive queues to set up for the Ethernet device.
2101 * @param nb_tx_queue
2102 * The number of transmit queues to set up for the Ethernet device.
2104 * The pointer to the configuration data to be used for the Ethernet device.
2105 * The *rte_eth_conf* structure includes:
2106 * - the hardware offload features to activate, with dedicated fields for
2107 * each statically configurable offload hardware feature provided by
2108 * Ethernet devices, such as IP checksum or VLAN tag stripping for
2110 * The Rx offload bitfield API is obsolete and will be deprecated.
2111 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
2112 * structure and use offloads field to set per-port offloads instead.
2113 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
2114 * the [rt]x_offload_capa returned from rte_eth_dev_info_get().
2115 * Any type of device supported offloading set in the input argument
2116 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
2117 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
2118 * - the Receive Side Scaling (RSS) configuration when using multiple Rx
2119 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
2120 * must be within the flow_type_rss_offloads provided by drivers via
2121 * rte_eth_dev_info_get() API.
2123 * Embedding all configuration information in a single data structure
2124 * is the more flexible method that allows the addition of new features
2125 * without changing the syntax of the API.
2127 * - 0: Success, device configured.
2128 * - <0: Error code returned by the driver configuration function.
2130 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
2131 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
2135 * @b EXPERIMENTAL: this API may change without prior notice.
2137 * Check if an Ethernet device was physically removed.
2140 * The port identifier of the Ethernet device.
2142 * 1 when the Ethernet device is removed, otherwise 0.
2146 rte_eth_dev_is_removed(uint16_t port_id);
2149 * Allocate and set up a receive queue for an Ethernet device.
2151 * The function allocates a contiguous block of memory for *nb_rx_desc*
2152 * receive descriptors from a memory zone associated with *socket_id*
2153 * and initializes each receive descriptor with a network buffer allocated
2154 * from the memory pool *mb_pool*.
2157 * The port identifier of the Ethernet device.
2158 * @param rx_queue_id
2159 * The index of the receive queue to set up.
2160 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2161 * to rte_eth_dev_configure().
2163 * The number of receive descriptors to allocate for the receive ring.
2165 * The *socket_id* argument is the socket identifier in case of NUMA.
2166 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2167 * the DMA memory allocated for the receive descriptors of the ring.
2169 * The pointer to the configuration data to be used for the receive queue.
2170 * NULL value is allowed, in which case default Rx configuration
2172 * The *rx_conf* structure contains an *rx_thresh* structure with the values
2173 * of the Prefetch, Host, and Write-Back threshold registers of the receive
2175 * In addition it contains the hardware offloads features to activate using
2176 * the DEV_RX_OFFLOAD_* flags.
2177 * If an offloading set in rx_conf->offloads
2178 * hasn't been set in the input argument eth_conf->rxmode.offloads
2179 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2180 * per-queue type and it is enabled for the queue.
2181 * No need to repeat any bit in rx_conf->offloads which has already been
2182 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2183 * at port level can't be disabled at queue level.
2184 * The configuration structure also contains the pointer to the array
2185 * of the receiving buffer segment descriptions, see rx_seg and rx_nseg
2186 * fields, this extended configuration might be used by split offloads like
2187 * RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT. If mb_pool is not NULL,
2188 * the extended configuration fields must be set to NULL and zero.
2190 * The pointer to the memory pool from which to allocate *rte_mbuf* network
2191 * memory buffers to populate each descriptor of the receive ring. There are
2192 * two options to provide Rx buffer configuration:
2194 * mb_pool is not NULL, rx_conf.rx_nseg is 0.
2195 * - multiple segments description:
2196 * mb_pool is NULL, rx_conf.rx_seg is not NULL, rx_conf.rx_nseg is not 0.
2197 * Taken only if flag RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT is set in offloads.
2200 * - 0: Success, receive queue correctly set up.
2201 * - -EIO: if device is removed.
2202 * - -ENODEV: if *port_id* is invalid.
2203 * - -EINVAL: The memory pool pointer is null or the size of network buffers
2204 * which can be allocated from this memory pool does not fit the various
2205 * buffer sizes allowed by the device controller.
2206 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
2207 * allocate network memory buffers from the memory pool when
2208 * initializing receive descriptors.
2210 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
2211 uint16_t nb_rx_desc, unsigned int socket_id,
2212 const struct rte_eth_rxconf *rx_conf,
2213 struct rte_mempool *mb_pool);
2217 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2219 * Allocate and set up a hairpin receive queue for an Ethernet device.
2221 * The function set up the selected queue to be used in hairpin.
2224 * The port identifier of the Ethernet device.
2225 * @param rx_queue_id
2226 * The index of the receive queue to set up.
2227 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2228 * to rte_eth_dev_configure().
2230 * The number of receive descriptors to allocate for the receive ring.
2231 * 0 means the PMD will use default value.
2233 * The pointer to the hairpin configuration.
2236 * - (0) if successful.
2237 * - (-ENODEV) if *port_id* is invalid.
2238 * - (-ENOTSUP) if hardware doesn't support.
2239 * - (-EINVAL) if bad parameter.
2240 * - (-ENOMEM) if unable to allocate the resources.
2243 int rte_eth_rx_hairpin_queue_setup
2244 (uint16_t port_id, uint16_t rx_queue_id, uint16_t nb_rx_desc,
2245 const struct rte_eth_hairpin_conf *conf);
2248 * Allocate and set up a transmit queue for an Ethernet device.
2251 * The port identifier of the Ethernet device.
2252 * @param tx_queue_id
2253 * The index of the transmit queue to set up.
2254 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2255 * to rte_eth_dev_configure().
2257 * The number of transmit descriptors to allocate for the transmit ring.
2259 * The *socket_id* argument is the socket identifier in case of NUMA.
2260 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2261 * the DMA memory allocated for the transmit descriptors of the ring.
2263 * The pointer to the configuration data to be used for the transmit queue.
2264 * NULL value is allowed, in which case default Tx configuration
2266 * The *tx_conf* structure contains the following data:
2267 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
2268 * Write-Back threshold registers of the transmit ring.
2269 * When setting Write-Back threshold to the value greater then zero,
2270 * *tx_rs_thresh* value should be explicitly set to one.
2271 * - The *tx_free_thresh* value indicates the [minimum] number of network
2272 * buffers that must be pending in the transmit ring to trigger their
2273 * [implicit] freeing by the driver transmit function.
2274 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
2275 * descriptors that must be pending in the transmit ring before setting the
2276 * RS bit on a descriptor by the driver transmit function.
2277 * The *tx_rs_thresh* value should be less or equal then
2278 * *tx_free_thresh* value, and both of them should be less then
2280 * - The *offloads* member contains Tx offloads to be enabled.
2281 * If an offloading set in tx_conf->offloads
2282 * hasn't been set in the input argument eth_conf->txmode.offloads
2283 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2284 * per-queue type and it is enabled for the queue.
2285 * No need to repeat any bit in tx_conf->offloads which has already been
2286 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2287 * at port level can't be disabled at queue level.
2289 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
2290 * the transmit function to use default values.
2292 * - 0: Success, the transmit queue is correctly set up.
2293 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
2295 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
2296 uint16_t nb_tx_desc, unsigned int socket_id,
2297 const struct rte_eth_txconf *tx_conf);
2301 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2303 * Allocate and set up a transmit hairpin queue for an Ethernet device.
2306 * The port identifier of the Ethernet device.
2307 * @param tx_queue_id
2308 * The index of the transmit queue to set up.
2309 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2310 * to rte_eth_dev_configure().
2312 * The number of transmit descriptors to allocate for the transmit ring.
2313 * 0 to set default PMD value.
2315 * The hairpin configuration.
2318 * - (0) if successful.
2319 * - (-ENODEV) if *port_id* is invalid.
2320 * - (-ENOTSUP) if hardware doesn't support.
2321 * - (-EINVAL) if bad parameter.
2322 * - (-ENOMEM) if unable to allocate the resources.
2325 int rte_eth_tx_hairpin_queue_setup
2326 (uint16_t port_id, uint16_t tx_queue_id, uint16_t nb_tx_desc,
2327 const struct rte_eth_hairpin_conf *conf);
2331 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2333 * Get all the hairpin peer Rx / Tx ports of the current port.
2334 * The caller should ensure that the array is large enough to save the ports
2338 * The port identifier of the Ethernet device.
2340 * Pointer to the array to store the peer ports list.
2342 * Length of the array to store the port identifiers.
2344 * Current port to peer port direction
2345 * positive - current used as Tx to get all peer Rx ports.
2346 * zero - current used as Rx to get all peer Tx ports.
2349 * - (0 or positive) actual peer ports number.
2350 * - (-EINVAL) if bad parameter.
2351 * - (-ENODEV) if *port_id* invalid
2352 * - (-ENOTSUP) if hardware doesn't support.
2353 * - Others detailed errors from PMD drivers.
2356 int rte_eth_hairpin_get_peer_ports(uint16_t port_id, uint16_t *peer_ports,
2357 size_t len, uint32_t direction);
2361 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2363 * Bind all hairpin Tx queues of one port to the Rx queues of the peer port.
2364 * It is only allowed to call this function after all hairpin queues are
2365 * configured properly and the devices are in started state.
2368 * The identifier of the Tx port.
2370 * The identifier of peer Rx port.
2371 * RTE_MAX_ETHPORTS is allowed for the traversal of all devices.
2372 * Rx port ID could have the same value as Tx port ID.
2375 * - (0) if successful.
2376 * - (-ENODEV) if Tx port ID is invalid.
2377 * - (-EBUSY) if device is not in started state.
2378 * - (-ENOTSUP) if hardware doesn't support.
2379 * - Others detailed errors from PMD drivers.
2382 int rte_eth_hairpin_bind(uint16_t tx_port, uint16_t rx_port);
2386 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2388 * Unbind all hairpin Tx queues of one port from the Rx queues of the peer port.
2389 * This should be called before closing the Tx or Rx devices, if the bind
2390 * function is called before.
2391 * After unbinding the hairpin ports pair, it is allowed to bind them again.
2392 * Changing queues configuration should be after stopping the device(s).
2395 * The identifier of the Tx port.
2397 * The identifier of peer Rx port.
2398 * RTE_MAX_ETHPORTS is allowed for traversal of all devices.
2399 * Rx port ID could have the same value as Tx port ID.
2402 * - (0) if successful.
2403 * - (-ENODEV) if Tx port ID is invalid.
2404 * - (-EBUSY) if device is in stopped state.
2405 * - (-ENOTSUP) if hardware doesn't support.
2406 * - Others detailed errors from PMD drivers.
2409 int rte_eth_hairpin_unbind(uint16_t tx_port, uint16_t rx_port);
2412 * Return the NUMA socket to which an Ethernet device is connected
2415 * The port identifier of the Ethernet device
2417 * The NUMA socket id to which the Ethernet device is connected or
2418 * a default of zero if the socket could not be determined.
2419 * -1 is returned is the port_id value is out of range.
2421 int rte_eth_dev_socket_id(uint16_t port_id);
2424 * Check if port_id of device is attached
2427 * The port identifier of the Ethernet device
2429 * - 0 if port is out of range or not attached
2430 * - 1 if device is attached
2432 int rte_eth_dev_is_valid_port(uint16_t port_id);
2435 * Start specified Rx queue of a port. It is used when rx_deferred_start
2436 * flag of the specified queue is true.
2439 * The port identifier of the Ethernet device
2440 * @param rx_queue_id
2441 * The index of the Rx queue to update the ring.
2442 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2443 * to rte_eth_dev_configure().
2445 * - 0: Success, the receive queue is started.
2446 * - -ENODEV: if *port_id* is invalid.
2447 * - -EINVAL: The queue_id out of range or belong to hairpin.
2448 * - -EIO: if device is removed.
2449 * - -ENOTSUP: The function not supported in PMD driver.
2451 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
2454 * Stop specified Rx queue of a port
2457 * The port identifier of the Ethernet device
2458 * @param rx_queue_id
2459 * The index of the Rx queue to update the ring.
2460 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2461 * to rte_eth_dev_configure().
2463 * - 0: Success, the receive queue is stopped.
2464 * - -ENODEV: if *port_id* is invalid.
2465 * - -EINVAL: The queue_id out of range or belong to hairpin.
2466 * - -EIO: if device is removed.
2467 * - -ENOTSUP: The function not supported in PMD driver.
2469 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
2472 * Start Tx for specified queue of a port. It is used when tx_deferred_start
2473 * flag of the specified queue is true.
2476 * The port identifier of the Ethernet device
2477 * @param tx_queue_id
2478 * The index of the Tx queue to update the ring.
2479 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2480 * to rte_eth_dev_configure().
2482 * - 0: Success, the transmit queue is started.
2483 * - -ENODEV: if *port_id* is invalid.
2484 * - -EINVAL: The queue_id out of range or belong to hairpin.
2485 * - -EIO: if device is removed.
2486 * - -ENOTSUP: The function not supported in PMD driver.
2488 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
2491 * Stop specified Tx queue of a port
2494 * The port identifier of the Ethernet device
2495 * @param tx_queue_id
2496 * The index of the Tx queue to update the ring.
2497 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2498 * to rte_eth_dev_configure().
2500 * - 0: Success, the transmit queue is stopped.
2501 * - -ENODEV: if *port_id* is invalid.
2502 * - -EINVAL: The queue_id out of range or belong to hairpin.
2503 * - -EIO: if device is removed.
2504 * - -ENOTSUP: The function not supported in PMD driver.
2506 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
2509 * Start an Ethernet device.
2511 * The device start step is the last one and consists of setting the configured
2512 * offload features and in starting the transmit and the receive units of the
2515 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
2516 * PMD port start callback function is invoked.
2518 * On success, all basic functions exported by the Ethernet API (link status,
2519 * receive/transmit, and so on) can be invoked.
2522 * The port identifier of the Ethernet device.
2524 * - 0: Success, Ethernet device started.
2525 * - <0: Error code of the driver device start function.
2527 int rte_eth_dev_start(uint16_t port_id);
2530 * Stop an Ethernet device. The device can be restarted with a call to
2531 * rte_eth_dev_start()
2534 * The port identifier of the Ethernet device.
2536 * - 0: Success, Ethernet device stopped.
2537 * - <0: Error code of the driver device stop function.
2539 int rte_eth_dev_stop(uint16_t port_id);
2542 * Link up an Ethernet device.
2544 * Set device link up will re-enable the device Rx/Tx
2545 * functionality after it is previously set device linked down.
2548 * The port identifier of the Ethernet device.
2550 * - 0: Success, Ethernet device linked up.
2551 * - <0: Error code of the driver device link up function.
2553 int rte_eth_dev_set_link_up(uint16_t port_id);
2556 * Link down an Ethernet device.
2557 * The device Rx/Tx functionality will be disabled if success,
2558 * and it can be re-enabled with a call to
2559 * rte_eth_dev_set_link_up()
2562 * The port identifier of the Ethernet device.
2564 int rte_eth_dev_set_link_down(uint16_t port_id);
2567 * Close a stopped Ethernet device. The device cannot be restarted!
2568 * The function frees all port resources.
2571 * The port identifier of the Ethernet device.
2573 * - Zero if the port is closed successfully.
2574 * - Negative if something went wrong.
2576 int rte_eth_dev_close(uint16_t port_id);
2579 * Reset a Ethernet device and keep its port id.
2581 * When a port has to be reset passively, the DPDK application can invoke
2582 * this function. For example when a PF is reset, all its VFs should also
2583 * be reset. Normally a DPDK application can invoke this function when
2584 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
2585 * a port reset in other circumstances.
2587 * When this function is called, it first stops the port and then calls the
2588 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
2589 * state, in which no Tx and Rx queues are setup, as if the port has been
2590 * reset and not started. The port keeps the port id it had before the
2593 * After calling rte_eth_dev_reset( ), the application should use
2594 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
2595 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
2596 * to reconfigure the device as appropriate.
2598 * Note: To avoid unexpected behavior, the application should stop calling
2599 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
2600 * safety, all these controlling functions should be called from the same
2604 * The port identifier of the Ethernet device.
2607 * - (0) if successful.
2608 * - (-ENODEV) if *port_id* is invalid.
2609 * - (-ENOTSUP) if hardware doesn't support this function.
2610 * - (-EPERM) if not ran from the primary process.
2611 * - (-EIO) if re-initialisation failed or device is removed.
2612 * - (-ENOMEM) if the reset failed due to OOM.
2613 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
2615 int rte_eth_dev_reset(uint16_t port_id);
2618 * Enable receipt in promiscuous mode for an Ethernet device.
2621 * The port identifier of the Ethernet device.
2623 * - (0) if successful.
2624 * - (-ENOTSUP) if support for promiscuous_enable() does not exist
2626 * - (-ENODEV) if *port_id* invalid.
2628 int rte_eth_promiscuous_enable(uint16_t port_id);
2631 * Disable receipt in promiscuous mode for an Ethernet device.
2634 * The port identifier of the Ethernet device.
2636 * - (0) if successful.
2637 * - (-ENOTSUP) if support for promiscuous_disable() does not exist
2639 * - (-ENODEV) if *port_id* invalid.
2641 int rte_eth_promiscuous_disable(uint16_t port_id);
2644 * Return the value of promiscuous mode for an Ethernet device.
2647 * The port identifier of the Ethernet device.
2649 * - (1) if promiscuous is enabled
2650 * - (0) if promiscuous is disabled.
2653 int rte_eth_promiscuous_get(uint16_t port_id);
2656 * Enable the receipt of any multicast frame by an Ethernet device.
2659 * The port identifier of the Ethernet device.
2661 * - (0) if successful.
2662 * - (-ENOTSUP) if support for allmulticast_enable() does not exist
2664 * - (-ENODEV) if *port_id* invalid.
2666 int rte_eth_allmulticast_enable(uint16_t port_id);
2669 * Disable the receipt of all multicast frames by an Ethernet device.
2672 * The port identifier of the Ethernet device.
2674 * - (0) if successful.
2675 * - (-ENOTSUP) if support for allmulticast_disable() does not exist
2677 * - (-ENODEV) if *port_id* invalid.
2679 int rte_eth_allmulticast_disable(uint16_t port_id);
2682 * Return the value of allmulticast mode for an Ethernet device.
2685 * The port identifier of the Ethernet device.
2687 * - (1) if allmulticast is enabled
2688 * - (0) if allmulticast is disabled.
2691 int rte_eth_allmulticast_get(uint16_t port_id);
2694 * Retrieve the link status (up/down), the duplex mode (half/full),
2695 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2697 * It might need to wait up to 9 seconds.
2698 * @see rte_eth_link_get_nowait.
2701 * The port identifier of the Ethernet device.
2703 * Link information written back.
2705 * - (0) if successful.
2706 * - (-ENOTSUP) if the function is not supported in PMD driver.
2707 * - (-ENODEV) if *port_id* invalid.
2708 * - (-EINVAL) if bad parameter.
2710 int rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
2713 * Retrieve the link status (up/down), the duplex mode (half/full),
2714 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2717 * The port identifier of the Ethernet device.
2719 * Link information written back.
2721 * - (0) if successful.
2722 * - (-ENOTSUP) if the function is not supported in PMD driver.
2723 * - (-ENODEV) if *port_id* invalid.
2724 * - (-EINVAL) if bad parameter.
2726 int rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
2730 * @b EXPERIMENTAL: this API may change without prior notice.
2732 * The function converts a link_speed to a string. It handles all special
2733 * values like unknown or none speed.
2736 * link_speed of rte_eth_link struct
2738 * Link speed in textual format. It's pointer to immutable memory.
2739 * No free is required.
2742 const char *rte_eth_link_speed_to_str(uint32_t link_speed);
2746 * @b EXPERIMENTAL: this API may change without prior notice.
2748 * The function converts a rte_eth_link struct representing a link status to
2752 * A pointer to a string to be filled with textual representation of
2753 * device status. At least ETH_LINK_MAX_STR_LEN bytes should be allocated to
2754 * store default link status text.
2756 * Length of available memory at 'str' string.
2758 * Link status returned by rte_eth_link_get function
2760 * Number of bytes written to str array or -EINVAL if bad parameter.
2763 int rte_eth_link_to_str(char *str, size_t len,
2764 const struct rte_eth_link *eth_link);
2767 * Retrieve the general I/O statistics of an Ethernet device.
2770 * The port identifier of the Ethernet device.
2772 * A pointer to a structure of type *rte_eth_stats* to be filled with
2773 * the values of device counters for the following set of statistics:
2774 * - *ipackets* with the total of successfully received packets.
2775 * - *opackets* with the total of successfully transmitted packets.
2776 * - *ibytes* with the total of successfully received bytes.
2777 * - *obytes* with the total of successfully transmitted bytes.
2778 * - *ierrors* with the total of erroneous received packets.
2779 * - *oerrors* with the total of failed transmitted packets.
2781 * Zero if successful. Non-zero otherwise.
2783 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
2786 * Reset the general I/O statistics of an Ethernet device.
2789 * The port identifier of the Ethernet device.
2791 * - (0) if device notified to reset stats.
2792 * - (-ENOTSUP) if hardware doesn't support.
2793 * - (-ENODEV) if *port_id* invalid.
2794 * - (<0): Error code of the driver stats reset function.
2796 int rte_eth_stats_reset(uint16_t port_id);
2799 * Retrieve names of extended statistics of an Ethernet device.
2801 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2803 * xstats_names[i].name => xstats[i].value
2805 * And the array index is same with id field of 'struct rte_eth_xstat':
2808 * This assumption makes key-value pair matching less flexible but simpler.
2811 * The port identifier of the Ethernet device.
2812 * @param xstats_names
2813 * An rte_eth_xstat_name array of at least *size* elements to
2814 * be filled. If set to NULL, the function returns the required number
2817 * The size of the xstats_names array (number of elements).
2819 * - A positive value lower or equal to size: success. The return value
2820 * is the number of entries filled in the stats table.
2821 * - A positive value higher than size: error, the given statistics table
2822 * is too small. The return value corresponds to the size that should
2823 * be given to succeed. The entries in the table are not valid and
2824 * shall not be used by the caller.
2825 * - A negative value on error (invalid port id).
2827 int rte_eth_xstats_get_names(uint16_t port_id,
2828 struct rte_eth_xstat_name *xstats_names,
2832 * Retrieve extended statistics of an Ethernet device.
2834 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2836 * xstats_names[i].name => xstats[i].value
2838 * And the array index is same with id field of 'struct rte_eth_xstat':
2841 * This assumption makes key-value pair matching less flexible but simpler.
2844 * The port identifier of the Ethernet device.
2846 * A pointer to a table of structure of type *rte_eth_xstat*
2847 * to be filled with device statistics ids and values.
2848 * This parameter can be set to NULL if n is 0.
2850 * The size of the xstats array (number of elements).
2852 * - A positive value lower or equal to n: success. The return value
2853 * is the number of entries filled in the stats table.
2854 * - A positive value higher than n: error, the given statistics table
2855 * is too small. The return value corresponds to the size that should
2856 * be given to succeed. The entries in the table are not valid and
2857 * shall not be used by the caller.
2858 * - A negative value on error (invalid port id).
2860 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2864 * Retrieve names of extended statistics of an Ethernet device.
2867 * The port identifier of the Ethernet device.
2868 * @param xstats_names
2869 * Array to be filled in with names of requested device statistics.
2870 * Must not be NULL if @p ids are specified (not NULL).
2872 * Number of elements in @p xstats_names array (if not NULL) and in
2873 * @p ids array (if not NULL). Must be 0 if both array pointers are NULL.
2875 * IDs array given by app to retrieve specific statistics. May be NULL to
2876 * retrieve names of all available statistics or, if @p xstats_names is
2877 * NULL as well, just the number of available statistics.
2879 * - A positive value lower or equal to size: success. The return value
2880 * is the number of entries filled in the stats table.
2881 * - A positive value higher than size: success. The given statistics table
2882 * is too small. The return value corresponds to the size that should
2883 * be given to succeed. The entries in the table are not valid and
2884 * shall not be used by the caller.
2885 * - A negative value on error.
2888 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2889 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2893 * Retrieve extended statistics of an Ethernet device.
2896 * The port identifier of the Ethernet device.
2898 * IDs array given by app to retrieve specific statistics. May be NULL to
2899 * retrieve all available statistics or, if @p values is NULL as well,
2900 * just the number of available statistics.
2902 * Array to be filled in with requested device statistics.
2903 * Must not be NULL if ids are specified (not NULL).
2905 * Number of elements in @p values array (if not NULL) and in @p ids
2906 * array (if not NULL). Must be 0 if both array pointers are NULL.
2908 * - A positive value lower or equal to size: success. The return value
2909 * is the number of entries filled in the stats table.
2910 * - A positive value higher than size: success: The given statistics table
2911 * is too small. The return value corresponds to the size that should
2912 * be given to succeed. The entries in the table are not valid and
2913 * shall not be used by the caller.
2914 * - A negative value on error.
2916 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2917 uint64_t *values, unsigned int size);
2920 * Gets the ID of a statistic from its name.
2922 * This function searches for the statistics using string compares, and
2923 * as such should not be used on the fast-path. For fast-path retrieval of
2924 * specific statistics, store the ID as provided in *id* from this function,
2925 * and pass the ID to rte_eth_xstats_get()
2927 * @param port_id The port to look up statistics from
2928 * @param xstat_name The name of the statistic to return
2929 * @param[out] id A pointer to an app-supplied uint64_t which should be
2930 * set to the ID of the stat if the stat exists.
2933 * -ENODEV for invalid port_id,
2934 * -EIO if device is removed,
2935 * -EINVAL if the xstat_name doesn't exist in port_id
2936 * -ENOMEM if bad parameter.
2938 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2942 * Reset extended statistics of an Ethernet device.
2945 * The port identifier of the Ethernet device.
2947 * - (0) if device notified to reset extended stats.
2948 * - (-ENOTSUP) if pmd doesn't support both
2949 * extended stats and basic stats reset.
2950 * - (-ENODEV) if *port_id* invalid.
2951 * - (<0): Error code of the driver xstats reset function.
2953 int rte_eth_xstats_reset(uint16_t port_id);
2956 * Set a mapping for the specified transmit queue to the specified per-queue
2957 * statistics counter.
2960 * The port identifier of the Ethernet device.
2961 * @param tx_queue_id
2962 * The index of the transmit queue for which a queue stats mapping is required.
2963 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2964 * to rte_eth_dev_configure().
2966 * The per-queue packet statistics functionality number that the transmit
2967 * queue is to be assigned.
2968 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2969 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2971 * Zero if successful. Non-zero otherwise.
2973 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2974 uint16_t tx_queue_id, uint8_t stat_idx);
2977 * Set a mapping for the specified receive queue to the specified per-queue
2978 * statistics counter.
2981 * The port identifier of the Ethernet device.
2982 * @param rx_queue_id
2983 * The index of the receive queue for which a queue stats mapping is required.
2984 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2985 * to rte_eth_dev_configure().
2987 * The per-queue packet statistics functionality number that the receive
2988 * queue is to be assigned.
2989 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2990 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2992 * Zero if successful. Non-zero otherwise.
2994 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2995 uint16_t rx_queue_id,
2999 * Retrieve the Ethernet address of an Ethernet device.
3002 * The port identifier of the Ethernet device.
3004 * A pointer to a structure of type *ether_addr* to be filled with
3005 * the Ethernet address of the Ethernet device.
3007 * - (0) if successful
3008 * - (-ENODEV) if *port_id* invalid.
3009 * - (-EINVAL) if bad parameter.
3011 int rte_eth_macaddr_get(uint16_t port_id, struct rte_ether_addr *mac_addr);
3015 * @b EXPERIMENTAL: this API may change without prior notice
3017 * Retrieve the Ethernet addresses of an Ethernet device.
3020 * The port identifier of the Ethernet device.
3022 * A pointer to an array of structures of type *ether_addr* to be filled with
3023 * the Ethernet addresses of the Ethernet device.
3025 * Number of elements in the @p ma array.
3026 * Note that rte_eth_dev_info::max_mac_addrs can be used to retrieve
3027 * max number of Ethernet addresses for given port.
3029 * - number of retrieved addresses if successful
3030 * - (-ENODEV) if *port_id* invalid.
3031 * - (-EINVAL) if bad parameter.
3034 int rte_eth_macaddrs_get(uint16_t port_id, struct rte_ether_addr *ma,
3038 * Retrieve the contextual information of an Ethernet device.
3040 * As part of this function, a number of of fields in dev_info will be
3041 * initialized as follows:
3046 * Where lim is defined within the rte_eth_dev_info_get as
3048 * const struct rte_eth_desc_lim lim = {
3049 * .nb_max = UINT16_MAX,
3052 * .nb_seg_max = UINT16_MAX,
3053 * .nb_mtu_seg_max = UINT16_MAX,
3056 * device = dev->device
3057 * min_mtu = RTE_ETHER_MIN_LEN - RTE_ETHER_HDR_LEN - RTE_ETHER_CRC_LEN
3058 * max_mtu = UINT16_MAX
3060 * The following fields will be populated if support for dev_infos_get()
3061 * exists for the device and the rte_eth_dev 'dev' has been populated
3062 * successfully with a call to it:
3064 * driver_name = dev->device->driver->name
3065 * nb_rx_queues = dev->data->nb_rx_queues
3066 * nb_tx_queues = dev->data->nb_tx_queues
3067 * dev_flags = &dev->data->dev_flags
3070 * The port identifier of the Ethernet device.
3072 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
3073 * the contextual information of the Ethernet device.
3075 * - (0) if successful.
3076 * - (-ENOTSUP) if support for dev_infos_get() does not exist for the device.
3077 * - (-ENODEV) if *port_id* invalid.
3078 * - (-EINVAL) if bad parameter.
3080 int rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
3084 * @b EXPERIMENTAL: this API may change without prior notice.
3086 * Retrieve the configuration of an Ethernet device.
3089 * The port identifier of the Ethernet device.
3091 * Location for Ethernet device configuration to be filled in.
3093 * - (0) if successful.
3094 * - (-ENODEV) if *port_id* invalid.
3095 * - (-EINVAL) if bad parameter.
3098 int rte_eth_dev_conf_get(uint16_t port_id, struct rte_eth_conf *dev_conf);
3101 * Retrieve the firmware version of a device.
3104 * The port identifier of the device.
3106 * A pointer to a string array storing the firmware version of a device,
3107 * the string includes terminating null. This pointer is allocated by caller.
3109 * The size of the string array pointed by fw_version, which should be
3110 * large enough to store firmware version of the device.
3112 * - (0) if successful.
3113 * - (-ENOTSUP) if operation is not supported.
3114 * - (-ENODEV) if *port_id* invalid.
3115 * - (-EIO) if device is removed.
3116 * - (-EINVAL) if bad parameter.
3117 * - (>0) if *fw_size* is not enough to store firmware version, return
3118 * the size of the non truncated string.
3120 int rte_eth_dev_fw_version_get(uint16_t port_id,
3121 char *fw_version, size_t fw_size);
3124 * Retrieve the supported packet types of an Ethernet device.
3126 * When a packet type is announced as supported, it *must* be recognized by
3127 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
3128 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
3129 * packet types for these packets:
3130 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
3131 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
3132 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
3133 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
3135 * When a packet is received by a PMD, the most precise type must be
3136 * returned among the ones supported. However a PMD is allowed to set
3137 * packet type that is not in the supported list, at the condition that it
3138 * is more precise. Therefore, a PMD announcing no supported packet types
3139 * can still set a matching packet type in a received packet.
3142 * Better to invoke this API after the device is already started or Rx burst
3143 * function is decided, to obtain correct supported ptypes.
3145 * if a given PMD does not report what ptypes it supports, then the supported
3146 * ptype count is reported as 0.
3148 * The port identifier of the Ethernet device.
3150 * A hint of what kind of packet type which the caller is interested in.
3152 * An array pointer to store adequate packet types, allocated by caller.
3154 * Size of the array pointed by param ptypes.
3156 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
3157 * only num entries will be filled into the ptypes array, but the full
3158 * count of supported ptypes will be returned.
3159 * - (-ENODEV) if *port_id* invalid.
3160 * - (-EINVAL) if bad parameter.
3162 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
3163 uint32_t *ptypes, int num);
3165 * Inform Ethernet device about reduced range of packet types to handle.
3167 * Application can use this function to set only specific ptypes that it's
3168 * interested. This information can be used by the PMD to optimize Rx path.
3170 * The function accepts an array `set_ptypes` allocated by the caller to
3171 * store the packet types set by the driver, the last element of the array
3172 * is set to RTE_PTYPE_UNKNOWN. The size of the `set_ptype` array should be
3173 * `rte_eth_dev_get_supported_ptypes() + 1` else it might only be filled
3177 * The port identifier of the Ethernet device.
3179 * The ptype family that application is interested in should be bitwise OR of
3180 * RTE_PTYPE_*_MASK or 0.
3182 * An array pointer to store set packet types, allocated by caller. The
3183 * function marks the end of array with RTE_PTYPE_UNKNOWN.
3185 * Size of the array pointed by param ptypes.
3186 * Should be rte_eth_dev_get_supported_ptypes() + 1 to accommodate the
3190 * - (-ENODEV) if *port_id* invalid.
3191 * - (-EINVAL) if *ptype_mask* is invalid (or) set_ptypes is NULL and
3194 int rte_eth_dev_set_ptypes(uint16_t port_id, uint32_t ptype_mask,
3195 uint32_t *set_ptypes, unsigned int num);
3198 * Retrieve the MTU of an Ethernet device.
3201 * The port identifier of the Ethernet device.
3203 * A pointer to a uint16_t where the retrieved MTU is to be stored.
3205 * - (0) if successful.
3206 * - (-ENODEV) if *port_id* invalid.
3207 * - (-EINVAL) if bad parameter.
3209 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
3212 * Change the MTU of an Ethernet device.
3215 * The port identifier of the Ethernet device.
3217 * A uint16_t for the MTU to be applied.
3219 * - (0) if successful.
3220 * - (-ENOTSUP) if operation is not supported.
3221 * - (-ENODEV) if *port_id* invalid.
3222 * - (-EIO) if device is removed.
3223 * - (-EINVAL) if *mtu* invalid, validation of mtu can occur within
3224 * rte_eth_dev_set_mtu if dev_infos_get is supported by the device or
3225 * when the mtu is set using dev->dev_ops->mtu_set.
3226 * - (-EBUSY) if operation is not allowed when the port is running
3228 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
3231 * Enable/Disable hardware filtering by an Ethernet device of received
3232 * VLAN packets tagged with a given VLAN Tag Identifier.
3235 * The port identifier of the Ethernet device.
3237 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
3239 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
3240 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
3242 * - (0) if successful.
3243 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3244 * - (-ENODEV) if *port_id* invalid.
3245 * - (-EIO) if device is removed.
3246 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
3247 * - (-EINVAL) if *vlan_id* > 4095.
3249 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
3252 * Enable/Disable hardware VLAN Strip by a Rx queue of an Ethernet device.
3255 * The port identifier of the Ethernet device.
3256 * @param rx_queue_id
3257 * The index of the receive queue for which a queue stats mapping is required.
3258 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3259 * to rte_eth_dev_configure().
3261 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
3262 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
3264 * - (0) if successful.
3265 * - (-ENOTSUP) if hardware-assisted VLAN stripping not configured.
3266 * - (-ENODEV) if *port_id* invalid.
3267 * - (-EINVAL) if *rx_queue_id* invalid.
3269 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
3273 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
3277 * The port identifier of the Ethernet device.
3281 * The Tag Protocol ID
3283 * - (0) if successful.
3284 * - (-ENOTSUP) if hardware-assisted VLAN TPID setup is not supported.
3285 * - (-ENODEV) if *port_id* invalid.
3286 * - (-EIO) if device is removed.
3288 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
3289 enum rte_vlan_type vlan_type,
3293 * Set VLAN offload configuration on an Ethernet device.
3296 * The port identifier of the Ethernet device.
3297 * @param offload_mask
3298 * The VLAN Offload bit mask can be mixed use with "OR"
3299 * ETH_VLAN_STRIP_OFFLOAD
3300 * ETH_VLAN_FILTER_OFFLOAD
3301 * ETH_VLAN_EXTEND_OFFLOAD
3302 * ETH_QINQ_STRIP_OFFLOAD
3304 * - (0) if successful.
3305 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3306 * - (-ENODEV) if *port_id* invalid.
3307 * - (-EIO) if device is removed.
3309 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
3312 * Read VLAN Offload configuration from an Ethernet device
3315 * The port identifier of the Ethernet device.
3317 * - (>0) if successful. Bit mask to indicate
3318 * ETH_VLAN_STRIP_OFFLOAD
3319 * ETH_VLAN_FILTER_OFFLOAD
3320 * ETH_VLAN_EXTEND_OFFLOAD
3321 * ETH_QINQ_STRIP_OFFLOAD
3322 * - (-ENODEV) if *port_id* invalid.
3324 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
3327 * Set port based Tx VLAN insertion on or off.
3330 * The port identifier of the Ethernet device.
3332 * Port based Tx VLAN identifier together with user priority.
3334 * Turn on or off the port based Tx VLAN insertion.
3337 * - (0) if successful.
3338 * - negative if failed.
3340 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
3342 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
3346 * Structure used to buffer packets for future Tx
3347 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
3349 struct rte_eth_dev_tx_buffer {
3350 buffer_tx_error_fn error_callback;
3351 void *error_userdata;
3352 uint16_t size; /**< Size of buffer for buffered Tx */
3353 uint16_t length; /**< Number of packets in the array */
3354 /** Pending packets to be sent on explicit flush or when full */
3355 struct rte_mbuf *pkts[];
3359 * Calculate the size of the Tx buffer.
3362 * Number of stored packets.
3364 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
3365 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
3368 * Initialize default values for buffered transmitting
3371 * Tx buffer to be initialized.
3378 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
3381 * Configure a callback for buffered packets which cannot be sent
3383 * Register a specific callback to be called when an attempt is made to send
3384 * all packets buffered on an Ethernet port, but not all packets can
3385 * successfully be sent. The callback registered here will be called only
3386 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
3387 * The default callback configured for each queue by default just frees the
3388 * packets back to the calling mempool. If additional behaviour is required,
3389 * for example, to count dropped packets, or to retry transmission of packets
3390 * which cannot be sent, this function should be used to register a suitable
3391 * callback function to implement the desired behaviour.
3392 * The example callback "rte_eth_count_unsent_packet_callback()" is also
3393 * provided as reference.
3396 * The port identifier of the Ethernet device.
3398 * The function to be used as the callback.
3400 * Arbitrary parameter to be passed to the callback function
3402 * 0 on success, or -EINVAL if bad parameter
3405 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
3406 buffer_tx_error_fn callback, void *userdata);
3409 * Callback function for silently dropping unsent buffered packets.
3411 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3412 * adjust the default behavior when buffered packets cannot be sent. This
3413 * function drops any unsent packets silently and is used by Tx buffered
3414 * operations as default behavior.
3416 * NOTE: this function should not be called directly, instead it should be used
3417 * as a callback for packet buffering.
3419 * NOTE: when configuring this function as a callback with
3420 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3421 * should point to an uint64_t value.
3424 * The previously buffered packets which could not be sent
3426 * The number of unsent packets in the pkts array
3431 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
3435 * Callback function for tracking unsent buffered packets.
3437 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3438 * adjust the default behavior when buffered packets cannot be sent. This
3439 * function drops any unsent packets, but also updates a user-supplied counter
3440 * to track the overall number of packets dropped. The counter should be an
3441 * uint64_t variable.
3443 * NOTE: this function should not be called directly, instead it should be used
3444 * as a callback for packet buffering.
3446 * NOTE: when configuring this function as a callback with
3447 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3448 * should point to an uint64_t value.
3451 * The previously buffered packets which could not be sent
3453 * The number of unsent packets in the pkts array
3455 * Pointer to an uint64_t value, which will be incremented by unsent
3458 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
3462 * Request the driver to free mbufs currently cached by the driver. The
3463 * driver will only free the mbuf if it is no longer in use. It is the
3464 * application's responsibility to ensure rte_eth_tx_buffer_flush(..) is
3468 * The port identifier of the Ethernet device.
3470 * The index of the transmit queue through which output packets must be
3472 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
3473 * to rte_eth_dev_configure().
3475 * Maximum number of packets to free. Use 0 to indicate all possible packets
3476 * should be freed. Note that a packet may be using multiple mbufs.
3479 * -ENODEV: Invalid interface
3480 * -EIO: device is removed
3481 * -ENOTSUP: Driver does not support function
3483 * 0-n: Number of packets freed. More packets may still remain in ring that
3487 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
3490 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
3493 enum rte_eth_event_ipsec_subtype {
3494 /** Unknown event type */
3495 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
3496 /** Sequence number overflow */
3497 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
3498 /** Soft time expiry of SA */
3499 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
3500 /** Soft byte expiry of SA */
3501 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
3502 /** Max value of this enum */
3503 RTE_ETH_EVENT_IPSEC_MAX
3507 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
3508 * information of the IPsec offload event.
3510 struct rte_eth_event_ipsec_desc {
3511 /** Type of RTE_ETH_EVENT_IPSEC_* event */
3512 enum rte_eth_event_ipsec_subtype subtype;
3514 * Event specific metadata.
3516 * For the following events, *userdata* registered
3517 * with the *rte_security_session* would be returned
3520 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
3521 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
3522 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
3524 * @see struct rte_security_session_conf
3531 * The eth device event type for interrupt, and maybe others in the future.
3533 enum rte_eth_event_type {
3534 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
3535 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
3536 /** queue state event (enabled/disabled) */
3537 RTE_ETH_EVENT_QUEUE_STATE,
3538 /** reset interrupt event, sent to VF on PF reset */
3539 RTE_ETH_EVENT_INTR_RESET,
3540 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
3541 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
3542 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
3543 RTE_ETH_EVENT_NEW, /**< port is probed */
3544 RTE_ETH_EVENT_DESTROY, /**< port is released */
3545 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
3546 RTE_ETH_EVENT_FLOW_AGED,/**< New aged-out flows is detected */
3547 RTE_ETH_EVENT_MAX /**< max value of this enum */
3550 /** User application callback to be registered for interrupts. */
3551 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
3552 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
3555 * Register a callback function for port event.
3559 * RTE_ETH_ALL means register the event for all port ids.
3563 * User supplied callback function to be called.
3565 * Pointer to the parameters for the registered callback.
3568 * - On success, zero.
3569 * - On failure, a negative value.
3571 int rte_eth_dev_callback_register(uint16_t port_id,
3572 enum rte_eth_event_type event,
3573 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3576 * Unregister a callback function for port event.
3580 * RTE_ETH_ALL means unregister the event for all port ids.
3584 * User supplied callback function to be called.
3586 * Pointer to the parameters for the registered callback. -1 means to
3587 * remove all for the same callback address and same event.
3590 * - On success, zero.
3591 * - On failure, a negative value.
3593 int rte_eth_dev_callback_unregister(uint16_t port_id,
3594 enum rte_eth_event_type event,
3595 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3598 * When there is no Rx packet coming in Rx Queue for a long time, we can
3599 * sleep lcore related to Rx Queue for power saving, and enable Rx interrupt
3600 * to be triggered when Rx packet arrives.
3602 * The rte_eth_dev_rx_intr_enable() function enables Rx queue
3603 * interrupt on specific Rx queue of a port.
3606 * The port identifier of the Ethernet device.
3608 * The index of the receive queue from which to retrieve input packets.
3609 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3610 * to rte_eth_dev_configure().
3612 * - (0) if successful.
3613 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3615 * - (-ENODEV) if *port_id* invalid.
3616 * - (-EIO) if device is removed.
3618 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
3621 * When lcore wakes up from Rx interrupt indicating packet coming, disable Rx
3622 * interrupt and returns to polling mode.
3624 * The rte_eth_dev_rx_intr_disable() function disables Rx queue
3625 * interrupt on specific Rx queue of a port.
3628 * The port identifier of the Ethernet device.
3630 * The index of the receive queue from which to retrieve input packets.
3631 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3632 * to rte_eth_dev_configure().
3634 * - (0) if successful.
3635 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3637 * - (-ENODEV) if *port_id* invalid.
3638 * - (-EIO) if device is removed.
3640 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
3643 * Rx Interrupt control per port.
3646 * The port identifier of the Ethernet device.
3648 * Epoll instance fd which the intr vector associated to.
3649 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3651 * The operation be performed for the vector.
3652 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3656 * - On success, zero.
3657 * - On failure, a negative value.
3659 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
3662 * Rx Interrupt control per queue.
3665 * The port identifier of the Ethernet device.
3667 * The index of the receive queue from which to retrieve input packets.
3668 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3669 * to rte_eth_dev_configure().
3671 * Epoll instance fd which the intr vector associated to.
3672 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3674 * The operation be performed for the vector.
3675 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3679 * - On success, zero.
3680 * - On failure, a negative value.
3682 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
3683 int epfd, int op, void *data);
3686 * Get interrupt fd per Rx queue.
3689 * The port identifier of the Ethernet device.
3691 * The index of the receive queue from which to retrieve input packets.
3692 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3693 * to rte_eth_dev_configure().
3695 * - (>=0) the interrupt fd associated to the requested Rx queue if
3700 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
3703 * Turn on the LED on the Ethernet device.
3704 * This function turns on the LED on the Ethernet device.
3707 * The port identifier of the Ethernet device.
3709 * - (0) if successful.
3710 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3712 * - (-ENODEV) if *port_id* invalid.
3713 * - (-EIO) if device is removed.
3715 int rte_eth_led_on(uint16_t port_id);
3718 * Turn off the LED on the Ethernet device.
3719 * This function turns off the LED on the Ethernet device.
3722 * The port identifier of the Ethernet device.
3724 * - (0) if successful.
3725 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3727 * - (-ENODEV) if *port_id* invalid.
3728 * - (-EIO) if device is removed.
3730 int rte_eth_led_off(uint16_t port_id);
3734 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3736 * Get Forward Error Correction(FEC) capability.
3739 * The port identifier of the Ethernet device.
3740 * @param speed_fec_capa
3741 * speed_fec_capa is out only with per-speed capabilities.
3742 * If set to NULL, the function returns the required number
3743 * of required array entries.
3745 * a number of elements in an speed_fec_capa array.
3748 * - A non-negative value lower or equal to num: success. The return value
3749 * is the number of entries filled in the fec capa array.
3750 * - A non-negative value higher than num: error, the given fec capa array
3751 * is too small. The return value corresponds to the num that should
3752 * be given to succeed. The entries in fec capa array are not valid and
3753 * shall not be used by the caller.
3754 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3756 * - (-EIO) if device is removed.
3757 * - (-ENODEV) if *port_id* invalid.
3758 * - (-EINVAL) if *num* or *speed_fec_capa* invalid
3761 int rte_eth_fec_get_capability(uint16_t port_id,
3762 struct rte_eth_fec_capa *speed_fec_capa,
3767 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3769 * Get current Forward Error Correction(FEC) mode.
3770 * If link is down and AUTO is enabled, AUTO is returned, otherwise,
3771 * configured FEC mode is returned.
3772 * If link is up, current FEC mode is returned.
3775 * The port identifier of the Ethernet device.
3777 * A bitmask of enabled FEC modes. If AUTO bit is set, other
3778 * bits specify FEC modes which may be negotiated. If AUTO
3779 * bit is clear, specify FEC modes to be used (only one valid
3780 * mode per speed may be set).
3782 * - (0) if successful.
3783 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3785 * - (-EIO) if device is removed.
3786 * - (-ENODEV) if *port_id* invalid.
3789 int rte_eth_fec_get(uint16_t port_id, uint32_t *fec_capa);
3793 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3795 * Set Forward Error Correction(FEC) mode.
3798 * The port identifier of the Ethernet device.
3800 * A bitmask of allowed FEC modes. If AUTO bit is set, other
3801 * bits specify FEC modes which may be negotiated. If AUTO
3802 * bit is clear, specify FEC modes to be used (only one valid
3803 * mode per speed may be set).
3805 * - (0) if successful.
3806 * - (-EINVAL) if the FEC mode is not valid.
3807 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3808 * - (-EIO) if device is removed.
3809 * - (-ENODEV) if *port_id* invalid.
3812 int rte_eth_fec_set(uint16_t port_id, uint32_t fec_capa);
3815 * Get current status of the Ethernet link flow control for Ethernet device
3818 * The port identifier of the Ethernet device.
3820 * The pointer to the structure where to store the flow control parameters.
3822 * - (0) if successful.
3823 * - (-ENOTSUP) if hardware doesn't support flow control.
3824 * - (-ENODEV) if *port_id* invalid.
3825 * - (-EIO) if device is removed.
3826 * - (-EINVAL) if bad parameter.
3828 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
3829 struct rte_eth_fc_conf *fc_conf);
3832 * Configure the Ethernet link flow control for Ethernet device
3835 * The port identifier of the Ethernet device.
3837 * The pointer to the structure of the flow control parameters.
3839 * - (0) if successful.
3840 * - (-ENOTSUP) if hardware doesn't support flow control mode.
3841 * - (-ENODEV) if *port_id* invalid.
3842 * - (-EINVAL) if bad parameter
3843 * - (-EIO) if flow control setup failure or device is removed.
3845 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
3846 struct rte_eth_fc_conf *fc_conf);
3849 * Configure the Ethernet priority flow control under DCB environment
3850 * for Ethernet device.
3853 * The port identifier of the Ethernet device.
3855 * The pointer to the structure of the priority flow control parameters.
3857 * - (0) if successful.
3858 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
3859 * - (-ENODEV) if *port_id* invalid.
3860 * - (-EINVAL) if bad parameter
3861 * - (-EIO) if flow control setup failure or device is removed.
3863 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
3864 struct rte_eth_pfc_conf *pfc_conf);
3867 * Add a MAC address to the set used for filtering incoming packets.
3870 * The port identifier of the Ethernet device.
3872 * The MAC address to add.
3874 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
3875 * not enabled, this should be set to 0.
3877 * - (0) if successfully added or *mac_addr* was already added.
3878 * - (-ENOTSUP) if hardware doesn't support this feature.
3879 * - (-ENODEV) if *port* is invalid.
3880 * - (-EIO) if device is removed.
3881 * - (-ENOSPC) if no more MAC addresses can be added.
3882 * - (-EINVAL) if MAC address is invalid.
3884 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct rte_ether_addr *mac_addr,
3888 * Remove a MAC address from the internal array of addresses.
3891 * The port identifier of the Ethernet device.
3893 * MAC address to remove.
3895 * - (0) if successful, or *mac_addr* didn't exist.
3896 * - (-ENOTSUP) if hardware doesn't support.
3897 * - (-ENODEV) if *port* invalid.
3898 * - (-EADDRINUSE) if attempting to remove the default MAC address.
3899 * - (-EINVAL) if MAC address is invalid.
3901 int rte_eth_dev_mac_addr_remove(uint16_t port_id,
3902 struct rte_ether_addr *mac_addr);
3905 * Set the default MAC address.
3908 * The port identifier of the Ethernet device.
3910 * New default MAC address.
3912 * - (0) if successful, or *mac_addr* didn't exist.
3913 * - (-ENOTSUP) if hardware doesn't support.
3914 * - (-ENODEV) if *port* invalid.
3915 * - (-EINVAL) if MAC address is invalid.
3917 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
3918 struct rte_ether_addr *mac_addr);
3921 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3924 * The port identifier of the Ethernet device.
3928 * Redirection table size. The table size can be queried by
3929 * rte_eth_dev_info_get().
3931 * - (0) if successful.
3932 * - (-ENODEV) if *port_id* is invalid.
3933 * - (-ENOTSUP) if hardware doesn't support.
3934 * - (-EINVAL) if bad parameter.
3935 * - (-EIO) if device is removed.
3937 int rte_eth_dev_rss_reta_update(uint16_t port_id,
3938 struct rte_eth_rss_reta_entry64 *reta_conf,
3939 uint16_t reta_size);
3942 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3945 * The port identifier of the Ethernet device.
3947 * RETA to query. For each requested reta entry, corresponding bit
3948 * in mask must be set.
3950 * Redirection table size. The table size can be queried by
3951 * rte_eth_dev_info_get().
3953 * - (0) if successful.
3954 * - (-ENODEV) if *port_id* is invalid.
3955 * - (-ENOTSUP) if hardware doesn't support.
3956 * - (-EINVAL) if bad parameter.
3957 * - (-EIO) if device is removed.
3959 int rte_eth_dev_rss_reta_query(uint16_t port_id,
3960 struct rte_eth_rss_reta_entry64 *reta_conf,
3961 uint16_t reta_size);
3964 * Updates unicast hash table for receiving packet with the given destination
3965 * MAC address, and the packet is routed to all VFs for which the Rx mode is
3966 * accept packets that match the unicast hash table.
3969 * The port identifier of the Ethernet device.
3971 * Unicast MAC address.
3973 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
3974 * 0 - Clear an unicast hash bit.
3976 * - (0) if successful.
3977 * - (-ENOTSUP) if hardware doesn't support.
3978 * - (-ENODEV) if *port_id* invalid.
3979 * - (-EIO) if device is removed.
3980 * - (-EINVAL) if bad parameter.
3982 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct rte_ether_addr *addr,
3986 * Updates all unicast hash bitmaps for receiving packet with any Unicast
3987 * Ethernet MAC addresses,the packet is routed to all VFs for which the Rx
3988 * mode is accept packets that match the unicast hash table.
3991 * The port identifier of the Ethernet device.
3993 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
3995 * 0 - Clear all unicast hash bitmaps
3997 * - (0) if successful.
3998 * - (-ENOTSUP) if hardware doesn't support.
3999 * - (-ENODEV) if *port_id* invalid.
4000 * - (-EIO) if device is removed.
4001 * - (-EINVAL) if bad parameter.
4003 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
4006 * Set the rate limitation for a queue on an Ethernet device.
4009 * The port identifier of the Ethernet device.
4013 * The Tx rate in Mbps. Allocated from the total port link speed.
4015 * - (0) if successful.
4016 * - (-ENOTSUP) if hardware doesn't support this feature.
4017 * - (-ENODEV) if *port_id* invalid.
4018 * - (-EIO) if device is removed.
4019 * - (-EINVAL) if bad parameter.
4021 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
4025 * Configuration of Receive Side Scaling hash computation of Ethernet device.
4028 * The port identifier of the Ethernet device.
4030 * The new configuration to use for RSS hash computation on the port.
4032 * - (0) if successful.
4033 * - (-ENODEV) if port identifier is invalid.
4034 * - (-EIO) if device is removed.
4035 * - (-ENOTSUP) if hardware doesn't support.
4036 * - (-EINVAL) if bad parameter.
4038 int rte_eth_dev_rss_hash_update(uint16_t port_id,
4039 struct rte_eth_rss_conf *rss_conf);
4042 * Retrieve current configuration of Receive Side Scaling hash computation
4043 * of Ethernet device.
4046 * The port identifier of the Ethernet device.
4048 * Where to store the current RSS hash configuration of the Ethernet device.
4050 * - (0) if successful.
4051 * - (-ENODEV) if port identifier is invalid.
4052 * - (-EIO) if device is removed.
4053 * - (-ENOTSUP) if hardware doesn't support RSS.
4054 * - (-EINVAL) if bad parameter.
4057 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
4058 struct rte_eth_rss_conf *rss_conf);
4061 * Add UDP tunneling port for a type of tunnel.
4063 * Some NICs may require such configuration to properly parse a tunnel
4064 * with any standard or custom UDP port.
4065 * The packets with this UDP port will be parsed for this type of tunnel.
4066 * The device parser will also check the rest of the tunnel headers
4067 * before classifying the packet.
4069 * With some devices, this API will affect packet classification, i.e.:
4070 * - mbuf.packet_type reported on Rx
4071 * - rte_flow rules with tunnel items
4074 * The port identifier of the Ethernet device.
4076 * UDP tunneling configuration.
4079 * - (0) if successful.
4080 * - (-ENODEV) if port identifier is invalid.
4081 * - (-EIO) if device is removed.
4082 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4085 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
4086 struct rte_eth_udp_tunnel *tunnel_udp);
4089 * Delete UDP tunneling port for a type of tunnel.
4091 * The packets with this UDP port will not be classified as this type of tunnel
4092 * anymore if the device use such mapping for tunnel packet classification.
4094 * @see rte_eth_dev_udp_tunnel_port_add
4097 * The port identifier of the Ethernet device.
4099 * UDP tunneling configuration.
4102 * - (0) if successful.
4103 * - (-ENODEV) if port identifier is invalid.
4104 * - (-EIO) if device is removed.
4105 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4108 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
4109 struct rte_eth_udp_tunnel *tunnel_udp);
4112 * Get DCB information on an Ethernet device.
4115 * The port identifier of the Ethernet device.
4119 * - (0) if successful.
4120 * - (-ENODEV) if port identifier is invalid.
4121 * - (-EIO) if device is removed.
4122 * - (-ENOTSUP) if hardware doesn't support.
4123 * - (-EINVAL) if bad parameter.
4125 int rte_eth_dev_get_dcb_info(uint16_t port_id,
4126 struct rte_eth_dcb_info *dcb_info);
4128 struct rte_eth_rxtx_callback;
4131 * Add a callback to be called on packet Rx on a given port and queue.
4133 * This API configures a function to be called for each burst of
4134 * packets received on a given NIC port queue. The return value is a pointer
4135 * that can be used to later remove the callback using
4136 * rte_eth_remove_rx_callback().
4138 * Multiple functions are called in the order that they are added.
4141 * The port identifier of the Ethernet device.
4143 * The queue on the Ethernet device on which the callback is to be added.
4145 * The callback function
4147 * A generic pointer parameter which will be passed to each invocation of the
4148 * callback function on this port and queue. Inter-thread synchronization
4149 * of any user data changes is the responsibility of the user.
4153 * On success, a pointer value which can later be used to remove the callback.
4155 const struct rte_eth_rxtx_callback *
4156 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
4157 rte_rx_callback_fn fn, void *user_param);
4160 * Add a callback that must be called first on packet Rx on a given port
4163 * This API configures a first function to be called for each burst of
4164 * packets received on a given NIC port queue. The return value is a pointer
4165 * that can be used to later remove the callback using
4166 * rte_eth_remove_rx_callback().
4168 * Multiple functions are called in the order that they are added.
4171 * The port identifier of the Ethernet device.
4173 * The queue on the Ethernet device on which the callback is to be added.
4175 * The callback function
4177 * A generic pointer parameter which will be passed to each invocation of the
4178 * callback function on this port and queue. Inter-thread synchronization
4179 * of any user data changes is the responsibility of the user.
4183 * On success, a pointer value which can later be used to remove the callback.
4185 const struct rte_eth_rxtx_callback *
4186 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
4187 rte_rx_callback_fn fn, void *user_param);
4190 * Add a callback to be called on packet Tx on a given port and queue.
4192 * This API configures a function to be called for each burst of
4193 * packets sent on a given NIC port queue. The return value is a pointer
4194 * that can be used to later remove the callback using
4195 * rte_eth_remove_tx_callback().
4197 * Multiple functions are called in the order that they are added.
4200 * The port identifier of the Ethernet device.
4202 * The queue on the Ethernet device on which the callback is to be added.
4204 * The callback function
4206 * A generic pointer parameter which will be passed to each invocation of the
4207 * callback function on this port and queue. Inter-thread synchronization
4208 * of any user data changes is the responsibility of the user.
4212 * On success, a pointer value which can later be used to remove the callback.
4214 const struct rte_eth_rxtx_callback *
4215 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
4216 rte_tx_callback_fn fn, void *user_param);
4219 * Remove an Rx packet callback from a given port and queue.
4221 * This function is used to removed callbacks that were added to a NIC port
4222 * queue using rte_eth_add_rx_callback().
4224 * Note: the callback is removed from the callback list but it isn't freed
4225 * since the it may still be in use. The memory for the callback can be
4226 * subsequently freed back by the application by calling rte_free():
4228 * - Immediately - if the port is stopped, or the user knows that no
4229 * callbacks are in flight e.g. if called from the thread doing Rx/Tx
4232 * - After a short delay - where the delay is sufficient to allow any
4233 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4234 * used to detect when data plane threads have ceased referencing the
4238 * The port identifier of the Ethernet device.
4240 * The queue on the Ethernet device from which the callback is to be removed.
4242 * User supplied callback created via rte_eth_add_rx_callback().
4245 * - 0: Success. Callback was removed.
4246 * - -ENODEV: If *port_id* is invalid.
4247 * - -ENOTSUP: Callback support is not available.
4248 * - -EINVAL: The queue_id is out of range, or the callback
4249 * is NULL or not found for the port/queue.
4251 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
4252 const struct rte_eth_rxtx_callback *user_cb);
4255 * Remove a Tx packet callback from a given port and queue.
4257 * This function is used to removed callbacks that were added to a NIC port
4258 * queue using rte_eth_add_tx_callback().
4260 * Note: the callback is removed from the callback list but it isn't freed
4261 * since the it may still be in use. The memory for the callback can be
4262 * subsequently freed back by the application by calling rte_free():
4264 * - Immediately - if the port is stopped, or the user knows that no
4265 * callbacks are in flight e.g. if called from the thread doing Rx/Tx
4268 * - After a short delay - where the delay is sufficient to allow any
4269 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4270 * used to detect when data plane threads have ceased referencing the
4274 * The port identifier of the Ethernet device.
4276 * The queue on the Ethernet device from which the callback is to be removed.
4278 * User supplied callback created via rte_eth_add_tx_callback().
4281 * - 0: Success. Callback was removed.
4282 * - -ENODEV: If *port_id* is invalid.
4283 * - -ENOTSUP: Callback support is not available.
4284 * - -EINVAL: The queue_id is out of range, or the callback
4285 * is NULL or not found for the port/queue.
4287 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
4288 const struct rte_eth_rxtx_callback *user_cb);
4291 * Retrieve information about given port's Rx queue.
4294 * The port identifier of the Ethernet device.
4296 * The Rx queue on the Ethernet device for which information
4297 * will be retrieved.
4299 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
4300 * the information of the Ethernet device.
4304 * - -ENODEV: If *port_id* is invalid.
4305 * - -ENOTSUP: routine is not supported by the device PMD.
4306 * - -EINVAL: The queue_id is out of range, or the queue
4309 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4310 struct rte_eth_rxq_info *qinfo);
4313 * Retrieve information about given port's Tx queue.
4316 * The port identifier of the Ethernet device.
4318 * The Tx queue on the Ethernet device for which information
4319 * will be retrieved.
4321 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
4322 * the information of the Ethernet device.
4326 * - -ENODEV: If *port_id* is invalid.
4327 * - -ENOTSUP: routine is not supported by the device PMD.
4328 * - -EINVAL: The queue_id is out of range, or the queue
4331 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4332 struct rte_eth_txq_info *qinfo);
4335 * Retrieve information about the Rx packet burst mode.
4338 * The port identifier of the Ethernet device.
4340 * The Rx queue on the Ethernet device for which information
4341 * will be retrieved.
4343 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4344 * with the information of the packet burst mode.
4348 * - -ENODEV: If *port_id* is invalid.
4349 * - -ENOTSUP: routine is not supported by the device PMD.
4350 * - -EINVAL: The queue_id is out of range.
4352 int rte_eth_rx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4353 struct rte_eth_burst_mode *mode);
4356 * Retrieve information about the Tx packet burst mode.
4359 * The port identifier of the Ethernet device.
4361 * The Tx queue on the Ethernet device for which information
4362 * will be retrieved.
4364 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4365 * with the information of the packet burst mode.
4369 * - -ENODEV: If *port_id* is invalid.
4370 * - -ENOTSUP: routine is not supported by the device PMD.
4371 * - -EINVAL: The queue_id is out of range.
4373 int rte_eth_tx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4374 struct rte_eth_burst_mode *mode);
4378 * @b EXPERIMENTAL: this API may change without prior notice.
4380 * Retrieve the monitor condition for a given receive queue.
4383 * The port identifier of the Ethernet device.
4385 * The Rx queue on the Ethernet device for which information
4386 * will be retrieved.
4388 * The pointer to power-optimized monitoring condition structure.
4392 * -ENOTSUP: Operation not supported.
4393 * -EINVAL: Invalid parameters.
4394 * -ENODEV: Invalid port ID.
4397 int rte_eth_get_monitor_addr(uint16_t port_id, uint16_t queue_id,
4398 struct rte_power_monitor_cond *pmc);
4401 * Retrieve device registers and register attributes (number of registers and
4405 * The port identifier of the Ethernet device.
4407 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
4408 * NULL the function fills in the width and length fields. If non-NULL
4409 * the registers are put into the buffer pointed at by the data field.
4411 * - (0) if successful.
4412 * - (-ENOTSUP) if hardware doesn't support.
4413 * - (-EINVAL) if bad parameter.
4414 * - (-ENODEV) if *port_id* invalid.
4415 * - (-EIO) if device is removed.
4416 * - others depends on the specific operations implementation.
4418 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
4421 * Retrieve size of device EEPROM
4424 * The port identifier of the Ethernet device.
4426 * - (>=0) EEPROM size if successful.
4427 * - (-ENOTSUP) if hardware doesn't support.
4428 * - (-ENODEV) if *port_id* invalid.
4429 * - (-EIO) if device is removed.
4430 * - others depends on the specific operations implementation.
4432 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
4435 * Retrieve EEPROM and EEPROM attribute
4438 * The port identifier of the Ethernet device.
4440 * The template includes buffer for return EEPROM data and
4441 * EEPROM attributes to be filled.
4443 * - (0) if successful.
4444 * - (-ENOTSUP) if hardware doesn't support.
4445 * - (-EINVAL) if bad parameter.
4446 * - (-ENODEV) if *port_id* invalid.
4447 * - (-EIO) if device is removed.
4448 * - others depends on the specific operations implementation.
4450 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4453 * Program EEPROM with provided data
4456 * The port identifier of the Ethernet device.
4458 * The template includes EEPROM data for programming and
4459 * EEPROM attributes to be filled
4461 * - (0) if successful.
4462 * - (-ENOTSUP) if hardware doesn't support.
4463 * - (-ENODEV) if *port_id* invalid.
4464 * - (-EINVAL) if bad parameter.
4465 * - (-EIO) if device is removed.
4466 * - others depends on the specific operations implementation.
4468 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4472 * @b EXPERIMENTAL: this API may change without prior notice.
4474 * Retrieve the type and size of plugin module EEPROM
4477 * The port identifier of the Ethernet device.
4479 * The type and size of plugin module EEPROM.
4481 * - (0) if successful.
4482 * - (-ENOTSUP) if hardware doesn't support.
4483 * - (-ENODEV) if *port_id* invalid.
4484 * - (-EINVAL) if bad parameter.
4485 * - (-EIO) if device is removed.
4486 * - others depends on the specific operations implementation.
4490 rte_eth_dev_get_module_info(uint16_t port_id,
4491 struct rte_eth_dev_module_info *modinfo);
4495 * @b EXPERIMENTAL: this API may change without prior notice.
4497 * Retrieve the data of plugin module EEPROM
4500 * The port identifier of the Ethernet device.
4502 * The template includes the plugin module EEPROM attributes, and the
4503 * buffer for return plugin module EEPROM data.
4505 * - (0) if successful.
4506 * - (-ENOTSUP) if hardware doesn't support.
4507 * - (-EINVAL) if bad parameter.
4508 * - (-ENODEV) if *port_id* invalid.
4509 * - (-EIO) if device is removed.
4510 * - others depends on the specific operations implementation.
4514 rte_eth_dev_get_module_eeprom(uint16_t port_id,
4515 struct rte_dev_eeprom_info *info);
4518 * Set the list of multicast addresses to filter on an Ethernet device.
4521 * The port identifier of the Ethernet device.
4522 * @param mc_addr_set
4523 * The array of multicast addresses to set. Equal to NULL when the function
4524 * is invoked to flush the set of filtered addresses.
4526 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
4527 * when the function is invoked to flush the set of filtered addresses.
4529 * - (0) if successful.
4530 * - (-ENODEV) if *port_id* invalid.
4531 * - (-EIO) if device is removed.
4532 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
4533 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
4534 * - (-EINVAL) if bad parameter.
4536 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
4537 struct rte_ether_addr *mc_addr_set,
4538 uint32_t nb_mc_addr);
4541 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
4544 * The port identifier of the Ethernet device.
4548 * - -ENODEV: The port ID is invalid.
4549 * - -EIO: if device is removed.
4550 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4552 int rte_eth_timesync_enable(uint16_t port_id);
4555 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
4558 * The port identifier of the Ethernet device.
4562 * - -ENODEV: The port ID is invalid.
4563 * - -EIO: if device is removed.
4564 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4566 int rte_eth_timesync_disable(uint16_t port_id);
4569 * Read an IEEE1588/802.1AS Rx timestamp from an Ethernet device.
4572 * The port identifier of the Ethernet device.
4574 * Pointer to the timestamp struct.
4576 * Device specific flags. Used to pass the Rx timesync register index to
4577 * i40e. Unused in igb/ixgbe, pass 0 instead.
4581 * - -EINVAL: No timestamp is available.
4582 * - -ENODEV: The port ID is invalid.
4583 * - -EIO: if device is removed.
4584 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4586 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
4587 struct timespec *timestamp, uint32_t flags);
4590 * Read an IEEE1588/802.1AS Tx timestamp from an Ethernet device.
4593 * The port identifier of the Ethernet device.
4595 * Pointer to the timestamp struct.
4599 * - -EINVAL: No timestamp is available.
4600 * - -ENODEV: The port ID is invalid.
4601 * - -EIO: if device is removed.
4602 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4604 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
4605 struct timespec *timestamp);
4608 * Adjust the timesync clock on an Ethernet device.
4610 * This is usually used in conjunction with other Ethdev timesync functions to
4611 * synchronize the device time using the IEEE1588/802.1AS protocol.
4614 * The port identifier of the Ethernet device.
4616 * The adjustment in nanoseconds.
4620 * - -ENODEV: The port ID is invalid.
4621 * - -EIO: if device is removed.
4622 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4624 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
4627 * Read the time from the timesync clock on an Ethernet device.
4629 * This is usually used in conjunction with other Ethdev timesync functions to
4630 * synchronize the device time using the IEEE1588/802.1AS protocol.
4633 * The port identifier of the Ethernet device.
4635 * Pointer to the timespec struct that holds the time.
4639 * - -EINVAL: Bad parameter.
4641 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
4644 * Set the time of the timesync clock on an Ethernet device.
4646 * This is usually used in conjunction with other Ethdev timesync functions to
4647 * synchronize the device time using the IEEE1588/802.1AS protocol.
4650 * The port identifier of the Ethernet device.
4652 * Pointer to the timespec struct that holds the time.
4656 * - -EINVAL: No timestamp is available.
4657 * - -ENODEV: The port ID is invalid.
4658 * - -EIO: if device is removed.
4659 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4661 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
4665 * @b EXPERIMENTAL: this API may change without prior notice.
4667 * Read the current clock counter of an Ethernet device
4669 * This returns the current raw clock value of an Ethernet device. It is
4670 * a raw amount of ticks, with no given time reference.
4671 * The value returned here is from the same clock than the one
4672 * filling timestamp field of Rx packets when using hardware timestamp
4673 * offload. Therefore it can be used to compute a precise conversion of
4674 * the device clock to the real time.
4676 * E.g, a simple heuristic to derivate the frequency would be:
4677 * uint64_t start, end;
4678 * rte_eth_read_clock(port, start);
4679 * rte_delay_ms(100);
4680 * rte_eth_read_clock(port, end);
4681 * double freq = (end - start) * 10;
4683 * Compute a common reference with:
4684 * uint64_t base_time_sec = current_time();
4685 * uint64_t base_clock;
4686 * rte_eth_read_clock(port, base_clock);
4688 * Then, convert the raw mbuf timestamp with:
4689 * base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq;
4691 * This simple example will not provide a very good accuracy. One must
4692 * at least measure multiple times the frequency and do a regression.
4693 * To avoid deviation from the system time, the common reference can
4694 * be repeated from time to time. The integer division can also be
4695 * converted by a multiplication and a shift for better performance.
4698 * The port identifier of the Ethernet device.
4700 * Pointer to the uint64_t that holds the raw clock value.
4704 * - -ENODEV: The port ID is invalid.
4705 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4706 * - -EINVAL: if bad parameter.
4710 rte_eth_read_clock(uint16_t port_id, uint64_t *clock);
4713 * Get the port id from device name. The device name should be specified
4715 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
4716 * - SoC device name, for example- fsl-gmac0
4717 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
4720 * pci address or name of the device
4722 * pointer to port identifier of the device
4724 * - (0) if successful and port_id is filled.
4725 * - (-ENODEV or -EINVAL) on failure.
4728 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
4731 * Get the device name from port id. The device name is specified as below:
4732 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
4733 * - SoC device name, for example- fsl-gmac0
4734 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
4737 * Port identifier of the device.
4739 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
4741 * - (0) if successful.
4742 * - (-ENODEV) if *port_id* is invalid.
4743 * - (-EINVAL) on failure.
4746 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
4749 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
4750 * the Ethernet device information, otherwise adjust them to boundaries.
4753 * The port identifier of the Ethernet device.
4755 * A pointer to a uint16_t where the number of receive
4756 * descriptors stored.
4758 * A pointer to a uint16_t where the number of transmit
4759 * descriptors stored.
4761 * - (0) if successful.
4762 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
4764 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
4765 uint16_t *nb_rx_desc,
4766 uint16_t *nb_tx_desc);
4769 * Test if a port supports specific mempool ops.
4772 * Port identifier of the Ethernet device.
4774 * The name of the pool operations to test.
4776 * - 0: best mempool ops choice for this port.
4777 * - 1: mempool ops are supported for this port.
4778 * - -ENOTSUP: mempool ops not supported for this port.
4779 * - -ENODEV: Invalid port Identifier.
4780 * - -EINVAL: Pool param is null.
4783 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
4786 * Get the security context for the Ethernet device.
4789 * Port identifier of the Ethernet device
4792 * - pointer to security context on success.
4795 rte_eth_dev_get_sec_ctx(uint16_t port_id);
4799 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
4801 * Query the device hairpin capabilities.
4804 * The port identifier of the Ethernet device.
4806 * Pointer to a structure that will hold the hairpin capabilities.
4808 * - (0) if successful.
4809 * - (-ENOTSUP) if hardware doesn't support.
4810 * - (-EINVAL) if bad parameter.
4813 int rte_eth_dev_hairpin_capability_get(uint16_t port_id,
4814 struct rte_eth_hairpin_cap *cap);
4818 * @b EXPERIMENTAL: this structure may change without prior notice.
4820 * Ethernet device representor ID range entry
4822 struct rte_eth_representor_range {
4823 enum rte_eth_representor_type type; /**< Representor type */
4824 int controller; /**< Controller index */
4825 int pf; /**< Physical function index */
4828 int vf; /**< VF start index */
4829 int sf; /**< SF start index */
4831 uint32_t id_base; /**< Representor ID start index */
4832 uint32_t id_end; /**< Representor ID end index */
4833 char name[RTE_DEV_NAME_MAX_LEN]; /**< Representor name */
4838 * @b EXPERIMENTAL: this structure may change without prior notice.
4840 * Ethernet device representor information
4842 struct rte_eth_representor_info {
4843 uint16_t controller; /**< Controller ID of caller device. */
4844 uint16_t pf; /**< Physical function ID of caller device. */
4845 uint32_t nb_ranges_alloc; /**< Size of the ranges array. */
4846 uint32_t nb_ranges; /**< Number of initialized ranges. */
4847 struct rte_eth_representor_range ranges[];/**< Representor ID range. */
4851 * Retrieve the representor info of the device.
4853 * Get device representor info to be able to calculate a unique
4854 * representor ID. @see rte_eth_representor_id_get helper.
4857 * The port identifier of the device.
4859 * A pointer to a representor info structure.
4860 * NULL to return number of range entries and allocate memory
4861 * for next call to store detail.
4862 * The number of ranges that were written into this structure
4863 * will be placed into its nb_ranges field. This number cannot be
4864 * larger than the nb_ranges_alloc that by the user before calling
4865 * this function. It can be smaller than the value returned by the
4866 * function, however.
4868 * - (-ENOTSUP) if operation is not supported.
4869 * - (-ENODEV) if *port_id* invalid.
4870 * - (-EIO) if device is removed.
4871 * - (>=0) number of available representor range entries.
4874 int rte_eth_representor_info_get(uint16_t port_id,
4875 struct rte_eth_representor_info *info);
4877 /** The NIC is able to deliver flag (if set) with packets to the PMD. */
4878 #define RTE_ETH_RX_METADATA_USER_FLAG RTE_BIT64(0)
4880 /** The NIC is able to deliver mark ID with packets to the PMD. */
4881 #define RTE_ETH_RX_METADATA_USER_MARK RTE_BIT64(1)
4883 /** The NIC is able to deliver tunnel ID with packets to the PMD. */
4884 #define RTE_ETH_RX_METADATA_TUNNEL_ID RTE_BIT64(2)
4888 * @b EXPERIMENTAL: this API may change without prior notice
4890 * Negotiate the NIC's ability to deliver specific kinds of metadata to the PMD.
4892 * Invoke this API before the first rte_eth_dev_configure() invocation
4893 * to let the PMD make preparations that are inconvenient to do later.
4895 * The negotiation process is as follows:
4897 * - the application requests features intending to use at least some of them;
4898 * - the PMD responds with the guaranteed subset of the requested feature set;
4899 * - the application can retry negotiation with another set of features;
4900 * - the application can pass zero to clear the negotiation result;
4901 * - the last negotiated result takes effect upon
4902 * the ethdev configure and start.
4905 * The PMD is supposed to first consider enabling the requested feature set
4906 * in its entirety. Only if it fails to do so, does it have the right to
4907 * respond with a smaller set of the originally requested features.
4910 * Return code (-ENOTSUP) does not necessarily mean that the requested
4911 * features are unsupported. In this case, the application should just
4912 * assume that these features can be used without prior negotiations.
4915 * Port (ethdev) identifier
4917 * @param[inout] features
4918 * Feature selection buffer
4921 * - (-EBUSY) if the port can't handle this in its current state;
4922 * - (-ENOTSUP) if the method itself is not supported by the PMD;
4923 * - (-ENODEV) if *port_id* is invalid;
4924 * - (-EINVAL) if *features* is NULL;
4925 * - (-EIO) if the device is removed;
4929 int rte_eth_rx_metadata_negotiate(uint16_t port_id, uint64_t *features);
4931 #include <rte_ethdev_core.h>
4935 * Helper routine for rte_eth_rx_burst().
4936 * Should be called at exit from PMD's rte_eth_rx_bulk implementation.
4937 * Does necessary post-processing - invokes Rx callbacks if any, etc.
4940 * The port identifier of the Ethernet device.
4942 * The index of the receive queue from which to retrieve input packets.
4944 * The address of an array of pointers to *rte_mbuf* structures that
4945 * have been retrieved from the device.
4947 * The number of packets that were retrieved from the device.
4949 * The number of elements in @p rx_pkts array.
4951 * Opaque pointer of Rx queue callback related data.
4954 * The number of packets effectively supplied to the @p rx_pkts array.
4956 uint16_t rte_eth_call_rx_callbacks(uint16_t port_id, uint16_t queue_id,
4957 struct rte_mbuf **rx_pkts, uint16_t nb_rx, uint16_t nb_pkts,
4962 * Retrieve a burst of input packets from a receive queue of an Ethernet
4963 * device. The retrieved packets are stored in *rte_mbuf* structures whose
4964 * pointers are supplied in the *rx_pkts* array.
4966 * The rte_eth_rx_burst() function loops, parsing the Rx ring of the
4967 * receive queue, up to *nb_pkts* packets, and for each completed Rx
4968 * descriptor in the ring, it performs the following operations:
4970 * - Initialize the *rte_mbuf* data structure associated with the
4971 * Rx descriptor according to the information provided by the NIC into
4972 * that Rx descriptor.
4974 * - Store the *rte_mbuf* data structure into the next entry of the
4977 * - Replenish the Rx descriptor with a new *rte_mbuf* buffer
4978 * allocated from the memory pool associated with the receive queue at
4979 * initialization time.
4981 * When retrieving an input packet that was scattered by the controller
4982 * into multiple receive descriptors, the rte_eth_rx_burst() function
4983 * appends the associated *rte_mbuf* buffers to the first buffer of the
4986 * The rte_eth_rx_burst() function returns the number of packets
4987 * actually retrieved, which is the number of *rte_mbuf* data structures
4988 * effectively supplied into the *rx_pkts* array.
4989 * A return value equal to *nb_pkts* indicates that the Rx queue contained
4990 * at least *rx_pkts* packets, and this is likely to signify that other
4991 * received packets remain in the input queue. Applications implementing
4992 * a "retrieve as much received packets as possible" policy can check this
4993 * specific case and keep invoking the rte_eth_rx_burst() function until
4994 * a value less than *nb_pkts* is returned.
4996 * This receive method has the following advantages:
4998 * - It allows a run-to-completion network stack engine to retrieve and
4999 * to immediately process received packets in a fast burst-oriented
5000 * approach, avoiding the overhead of unnecessary intermediate packet
5001 * queue/dequeue operations.
5003 * - Conversely, it also allows an asynchronous-oriented processing
5004 * method to retrieve bursts of received packets and to immediately
5005 * queue them for further parallel processing by another logical core,
5006 * for instance. However, instead of having received packets being
5007 * individually queued by the driver, this approach allows the caller
5008 * of the rte_eth_rx_burst() function to queue a burst of retrieved
5009 * packets at a time and therefore dramatically reduce the cost of
5010 * enqueue/dequeue operations per packet.
5012 * - It allows the rte_eth_rx_burst() function of the driver to take
5013 * advantage of burst-oriented hardware features (CPU cache,
5014 * prefetch instructions, and so on) to minimize the number of CPU
5015 * cycles per packet.
5017 * To summarize, the proposed receive API enables many
5018 * burst-oriented optimizations in both synchronous and asynchronous
5019 * packet processing environments with no overhead in both cases.
5022 * Some drivers using vector instructions require that *nb_pkts* is
5023 * divisible by 4 or 8, depending on the driver implementation.
5025 * The rte_eth_rx_burst() function does not provide any error
5026 * notification to avoid the corresponding overhead. As a hint, the
5027 * upper-level application might check the status of the device link once
5028 * being systematically returned a 0 value for a given number of tries.
5031 * The port identifier of the Ethernet device.
5033 * The index of the receive queue from which to retrieve input packets.
5034 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
5035 * to rte_eth_dev_configure().
5037 * The address of an array of pointers to *rte_mbuf* structures that
5038 * must be large enough to store *nb_pkts* pointers in it.
5040 * The maximum number of packets to retrieve.
5041 * The value must be divisible by 8 in order to work with any driver.
5043 * The number of packets actually retrieved, which is the number
5044 * of pointers to *rte_mbuf* structures effectively supplied to the
5047 static inline uint16_t
5048 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
5049 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
5052 struct rte_eth_fp_ops *p;
5055 #ifdef RTE_ETHDEV_DEBUG_RX
5056 if (port_id >= RTE_MAX_ETHPORTS ||
5057 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5059 "Invalid port_id=%u or queue_id=%u\n",
5065 /* fetch pointer to queue data */
5066 p = &rte_eth_fp_ops[port_id];
5067 qd = p->rxq.data[queue_id];
5069 #ifdef RTE_ETHDEV_DEBUG_RX
5070 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
5073 RTE_ETHDEV_LOG(ERR, "Invalid Rx queue_id=%u for port_id=%u\n",
5079 nb_rx = p->rx_pkt_burst(qd, rx_pkts, nb_pkts);
5081 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
5085 /* __ATOMIC_RELEASE memory order was used when the
5086 * call back was inserted into the list.
5087 * Since there is a clear dependency between loading
5088 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5091 cb = __atomic_load_n((void **)&p->rxq.clbk[queue_id],
5093 if (unlikely(cb != NULL))
5094 nb_rx = rte_eth_call_rx_callbacks(port_id, queue_id,
5095 rx_pkts, nb_rx, nb_pkts, cb);
5099 rte_ethdev_trace_rx_burst(port_id, queue_id, (void **)rx_pkts, nb_rx);
5104 * Get the number of used descriptors of a Rx queue
5107 * The port identifier of the Ethernet device.
5109 * The queue id on the specific port.
5111 * The number of used descriptors in the specific queue, or:
5112 * - (-ENODEV) if *port_id* is invalid.
5113 * (-EINVAL) if *queue_id* is invalid
5114 * (-ENOTSUP) if the device does not support this function
5117 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
5119 struct rte_eth_fp_ops *p;
5122 if (port_id >= RTE_MAX_ETHPORTS ||
5123 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5125 "Invalid port_id=%u or queue_id=%u\n",
5130 /* fetch pointer to queue data */
5131 p = &rte_eth_fp_ops[port_id];
5132 qd = p->rxq.data[queue_id];
5134 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5135 RTE_FUNC_PTR_OR_ERR_RET(*p->rx_queue_count, -ENOTSUP);
5139 return (int)(*p->rx_queue_count)(qd);
5142 /**@{@name Rx hardware descriptor states
5143 * @see rte_eth_rx_descriptor_status
5145 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
5146 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
5147 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5151 * Check the status of a Rx descriptor in the queue
5153 * It should be called in a similar context than the Rx function:
5154 * - on a dataplane core
5155 * - not concurrently on the same queue
5157 * Since it's a dataplane function, no check is performed on port_id and
5158 * queue_id. The caller must therefore ensure that the port is enabled
5159 * and the queue is configured and running.
5161 * Note: accessing to a random descriptor in the ring may trigger cache
5162 * misses and have a performance impact.
5165 * A valid port identifier of the Ethernet device which.
5167 * A valid Rx queue identifier on this port.
5169 * The offset of the descriptor starting from tail (0 is the next
5170 * packet to be received by the driver).
5173 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
5175 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
5176 * not yet processed by the driver (i.e. in the receive queue).
5177 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
5178 * the driver and not yet returned to hw, or reserved by the hw.
5179 * - (-EINVAL) bad descriptor offset.
5180 * - (-ENOTSUP) if the device does not support this function.
5181 * - (-ENODEV) bad port or queue (only if compiled with debug).
5184 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
5187 struct rte_eth_fp_ops *p;
5190 #ifdef RTE_ETHDEV_DEBUG_RX
5191 if (port_id >= RTE_MAX_ETHPORTS ||
5192 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5194 "Invalid port_id=%u or queue_id=%u\n",
5200 /* fetch pointer to queue data */
5201 p = &rte_eth_fp_ops[port_id];
5202 qd = p->rxq.data[queue_id];
5204 #ifdef RTE_ETHDEV_DEBUG_RX
5205 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5209 RTE_FUNC_PTR_OR_ERR_RET(*p->rx_descriptor_status, -ENOTSUP);
5210 return (*p->rx_descriptor_status)(qd, offset);
5213 /**@{@name Tx hardware descriptor states
5214 * @see rte_eth_tx_descriptor_status
5216 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
5217 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
5218 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5222 * Check the status of a Tx descriptor in the queue.
5224 * It should be called in a similar context than the Tx function:
5225 * - on a dataplane core
5226 * - not concurrently on the same queue
5228 * Since it's a dataplane function, no check is performed on port_id and
5229 * queue_id. The caller must therefore ensure that the port is enabled
5230 * and the queue is configured and running.
5232 * Note: accessing to a random descriptor in the ring may trigger cache
5233 * misses and have a performance impact.
5236 * A valid port identifier of the Ethernet device which.
5238 * A valid Tx queue identifier on this port.
5240 * The offset of the descriptor starting from tail (0 is the place where
5241 * the next packet will be send).
5244 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
5245 * in the transmit queue.
5246 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
5247 * be reused by the driver.
5248 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
5249 * driver or the hardware.
5250 * - (-EINVAL) bad descriptor offset.
5251 * - (-ENOTSUP) if the device does not support this function.
5252 * - (-ENODEV) bad port or queue (only if compiled with debug).
5254 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
5255 uint16_t queue_id, uint16_t offset)
5257 struct rte_eth_fp_ops *p;
5260 #ifdef RTE_ETHDEV_DEBUG_TX
5261 if (port_id >= RTE_MAX_ETHPORTS ||
5262 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5264 "Invalid port_id=%u or queue_id=%u\n",
5270 /* fetch pointer to queue data */
5271 p = &rte_eth_fp_ops[port_id];
5272 qd = p->txq.data[queue_id];
5274 #ifdef RTE_ETHDEV_DEBUG_TX
5275 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5279 RTE_FUNC_PTR_OR_ERR_RET(*p->tx_descriptor_status, -ENOTSUP);
5280 return (*p->tx_descriptor_status)(qd, offset);
5285 * Helper routine for rte_eth_tx_burst().
5286 * Should be called before entry PMD's rte_eth_tx_bulk implementation.
5287 * Does necessary pre-processing - invokes Tx callbacks if any, etc.
5290 * The port identifier of the Ethernet device.
5292 * The index of the transmit queue through which output packets must be
5295 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5296 * which contain the output packets.
5298 * The maximum number of packets to transmit.
5300 * The number of output packets to transmit.
5302 uint16_t rte_eth_call_tx_callbacks(uint16_t port_id, uint16_t queue_id,
5303 struct rte_mbuf **tx_pkts, uint16_t nb_pkts, void *opaque);
5306 * Send a burst of output packets on a transmit queue of an Ethernet device.
5308 * The rte_eth_tx_burst() function is invoked to transmit output packets
5309 * on the output queue *queue_id* of the Ethernet device designated by its
5311 * The *nb_pkts* parameter is the number of packets to send which are
5312 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5313 * allocated from a pool created with rte_pktmbuf_pool_create().
5314 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
5315 * up to the number of transmit descriptors available in the Tx ring of the
5317 * For each packet to send, the rte_eth_tx_burst() function performs
5318 * the following operations:
5320 * - Pick up the next available descriptor in the transmit ring.
5322 * - Free the network buffer previously sent with that descriptor, if any.
5324 * - Initialize the transmit descriptor with the information provided
5325 * in the *rte_mbuf data structure.
5327 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
5328 * the rte_eth_tx_burst() function uses several transmit descriptors
5331 * The rte_eth_tx_burst() function returns the number of packets it
5332 * actually sent. A return value equal to *nb_pkts* means that all packets
5333 * have been sent, and this is likely to signify that other output packets
5334 * could be immediately transmitted again. Applications that implement a
5335 * "send as many packets to transmit as possible" policy can check this
5336 * specific case and keep invoking the rte_eth_tx_burst() function until
5337 * a value less than *nb_pkts* is returned.
5339 * It is the responsibility of the rte_eth_tx_burst() function to
5340 * transparently free the memory buffers of packets previously sent.
5341 * This feature is driven by the *tx_free_thresh* value supplied to the
5342 * rte_eth_dev_configure() function at device configuration time.
5343 * When the number of free Tx descriptors drops below this threshold, the
5344 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
5345 * of those packets whose transmission was effectively completed.
5347 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
5348 * invoke this function concurrently on the same Tx queue without SW lock.
5349 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
5351 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
5355 * The port identifier of the Ethernet device.
5357 * The index of the transmit queue through which output packets must be
5359 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5360 * to rte_eth_dev_configure().
5362 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5363 * which contain the output packets.
5365 * The maximum number of packets to transmit.
5367 * The number of output packets actually stored in transmit descriptors of
5368 * the transmit ring. The return value can be less than the value of the
5369 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
5371 static inline uint16_t
5372 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
5373 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5375 struct rte_eth_fp_ops *p;
5378 #ifdef RTE_ETHDEV_DEBUG_TX
5379 if (port_id >= RTE_MAX_ETHPORTS ||
5380 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5382 "Invalid port_id=%u or queue_id=%u\n",
5388 /* fetch pointer to queue data */
5389 p = &rte_eth_fp_ops[port_id];
5390 qd = p->txq.data[queue_id];
5392 #ifdef RTE_ETHDEV_DEBUG_TX
5393 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
5396 RTE_ETHDEV_LOG(ERR, "Invalid Tx queue_id=%u for port_id=%u\n",
5402 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
5406 /* __ATOMIC_RELEASE memory order was used when the
5407 * call back was inserted into the list.
5408 * Since there is a clear dependency between loading
5409 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5412 cb = __atomic_load_n((void **)&p->txq.clbk[queue_id],
5414 if (unlikely(cb != NULL))
5415 nb_pkts = rte_eth_call_tx_callbacks(port_id, queue_id,
5416 tx_pkts, nb_pkts, cb);
5420 nb_pkts = p->tx_pkt_burst(qd, tx_pkts, nb_pkts);
5422 rte_ethdev_trace_tx_burst(port_id, queue_id, (void **)tx_pkts, nb_pkts);
5427 * Process a burst of output packets on a transmit queue of an Ethernet device.
5429 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
5430 * transmitted on the output queue *queue_id* of the Ethernet device designated
5432 * The *nb_pkts* parameter is the number of packets to be prepared which are
5433 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5434 * allocated from a pool created with rte_pktmbuf_pool_create().
5435 * For each packet to send, the rte_eth_tx_prepare() function performs
5436 * the following operations:
5438 * - Check if packet meets devices requirements for Tx offloads.
5440 * - Check limitations about number of segments.
5442 * - Check additional requirements when debug is enabled.
5444 * - Update and/or reset required checksums when Tx offload is set for packet.
5446 * Since this function can modify packet data, provided mbufs must be safely
5447 * writable (e.g. modified data cannot be in shared segment).
5449 * The rte_eth_tx_prepare() function returns the number of packets ready to be
5450 * sent. A return value equal to *nb_pkts* means that all packets are valid and
5451 * ready to be sent, otherwise stops processing on the first invalid packet and
5452 * leaves the rest packets untouched.
5454 * When this functionality is not implemented in the driver, all packets are
5455 * are returned untouched.
5458 * The port identifier of the Ethernet device.
5459 * The value must be a valid port id.
5461 * The index of the transmit queue through which output packets must be
5463 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5464 * to rte_eth_dev_configure().
5466 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5467 * which contain the output packets.
5469 * The maximum number of packets to process.
5471 * The number of packets correct and ready to be sent. The return value can be
5472 * less than the value of the *tx_pkts* parameter when some packet doesn't
5473 * meet devices requirements with rte_errno set appropriately:
5474 * - EINVAL: offload flags are not correctly set
5475 * - ENOTSUP: the offload feature is not supported by the hardware
5476 * - ENODEV: if *port_id* is invalid (with debug enabled only)
5480 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
5482 static inline uint16_t
5483 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
5484 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5486 struct rte_eth_fp_ops *p;
5489 #ifdef RTE_ETHDEV_DEBUG_TX
5490 if (port_id >= RTE_MAX_ETHPORTS ||
5491 queue_id >= RTE_MAX_QUEUES_PER_PORT) {
5493 "Invalid port_id=%u or queue_id=%u\n",
5500 /* fetch pointer to queue data */
5501 p = &rte_eth_fp_ops[port_id];
5502 qd = p->txq.data[queue_id];
5504 #ifdef RTE_ETHDEV_DEBUG_TX
5505 if (!rte_eth_dev_is_valid_port(port_id)) {
5506 RTE_ETHDEV_LOG(ERR, "Invalid Tx port_id=%u\n", port_id);
5511 RTE_ETHDEV_LOG(ERR, "Invalid Tx queue_id=%u for port_id=%u\n",
5518 if (!p->tx_pkt_prepare)
5521 return p->tx_pkt_prepare(qd, tx_pkts, nb_pkts);
5527 * Native NOOP operation for compilation targets which doesn't require any
5528 * preparations steps, and functional NOOP may introduce unnecessary performance
5531 * Generally this is not a good idea to turn it on globally and didn't should
5532 * be used if behavior of tx_preparation can change.
5535 static inline uint16_t
5536 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
5537 __rte_unused uint16_t queue_id,
5538 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5546 * Send any packets queued up for transmission on a port and HW queue
5548 * This causes an explicit flush of packets previously buffered via the
5549 * rte_eth_tx_buffer() function. It returns the number of packets successfully
5550 * sent to the NIC, and calls the error callback for any unsent packets. Unless
5551 * explicitly set up otherwise, the default callback simply frees the unsent
5552 * packets back to the owning mempool.
5555 * The port identifier of the Ethernet device.
5557 * The index of the transmit queue through which output packets must be
5559 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5560 * to rte_eth_dev_configure().
5562 * Buffer of packets to be transmit.
5564 * The number of packets successfully sent to the Ethernet device. The error
5565 * callback is called for any packets which could not be sent.
5567 static inline uint16_t
5568 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
5569 struct rte_eth_dev_tx_buffer *buffer)
5572 uint16_t to_send = buffer->length;
5577 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
5581 /* All packets sent, or to be dealt with by callback below */
5582 if (unlikely(sent != to_send))
5583 buffer->error_callback(&buffer->pkts[sent],
5584 (uint16_t)(to_send - sent),
5585 buffer->error_userdata);
5591 * Buffer a single packet for future transmission on a port and queue
5593 * This function takes a single mbuf/packet and buffers it for later
5594 * transmission on the particular port and queue specified. Once the buffer is
5595 * full of packets, an attempt will be made to transmit all the buffered
5596 * packets. In case of error, where not all packets can be transmitted, a
5597 * callback is called with the unsent packets as a parameter. If no callback
5598 * is explicitly set up, the unsent packets are just freed back to the owning
5599 * mempool. The function returns the number of packets actually sent i.e.
5600 * 0 if no buffer flush occurred, otherwise the number of packets successfully
5604 * The port identifier of the Ethernet device.
5606 * The index of the transmit queue through which output packets must be
5608 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5609 * to rte_eth_dev_configure().
5611 * Buffer used to collect packets to be sent.
5613 * Pointer to the packet mbuf to be sent.
5615 * 0 = packet has been buffered for later transmission
5616 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
5617 * causing N packets to be sent, and the error callback to be called for
5620 static __rte_always_inline uint16_t
5621 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
5622 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
5624 buffer->pkts[buffer->length++] = tx_pkt;
5625 if (buffer->length < buffer->size)
5628 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
5635 #endif /* _RTE_ETHDEV_H_ */