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 #include <rte_compat.h>
153 #include <rte_interrupts.h>
155 #include <rte_devargs.h>
156 #include <rte_errno.h>
157 #include <rte_common.h>
158 #include <rte_config.h>
159 #include <rte_ether.h>
160 #include <rte_power_intrinsics.h>
162 #include "rte_ethdev_trace_fp.h"
163 #include "rte_dev_info.h"
165 extern int rte_eth_dev_logtype;
167 #define RTE_ETHDEV_LOG(level, ...) \
168 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
173 * Initializes a device iterator.
175 * This iterator allows accessing a list of devices matching some devargs.
178 * Device iterator handle initialized by the function.
179 * The fields bus_str and cls_str might be dynamically allocated,
180 * and could be freed by calling rte_eth_iterator_cleanup().
183 * Device description string.
186 * 0 on successful initialization, negative otherwise.
188 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs);
191 * Iterates on devices with devargs filter.
192 * The ownership is not checked.
194 * The next port id is returned, and the iterator is updated.
197 * Device iterator handle initialized by rte_eth_iterator_init().
198 * Some fields bus_str and cls_str might be freed when no more port is found,
199 * by calling rte_eth_iterator_cleanup().
202 * A port id if found, RTE_MAX_ETHPORTS otherwise.
204 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter);
207 * Free some allocated fields of the iterator.
209 * This function is automatically called by rte_eth_iterator_next()
210 * on the last iteration (i.e. when no more matching port is found).
212 * It is safe to call this function twice; it will do nothing more.
215 * Device iterator handle initialized by rte_eth_iterator_init().
216 * The fields bus_str and cls_str are freed if needed.
218 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter);
221 * Macro to iterate over all ethdev ports matching some devargs.
223 * If a break is done before the end of the loop,
224 * the function rte_eth_iterator_cleanup() must be called.
227 * Iterated port id of type uint16_t.
229 * Device parameters input as string of type char*.
231 * Iterator handle of type struct rte_dev_iterator, used internally.
233 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \
234 for (rte_eth_iterator_init(iter, devargs), \
235 id = rte_eth_iterator_next(iter); \
236 id != RTE_MAX_ETHPORTS; \
237 id = rte_eth_iterator_next(iter))
240 * A structure used to retrieve statistics for an Ethernet port.
241 * Not all statistics fields in struct rte_eth_stats are supported
242 * by any type of network interface card (NIC). If any statistics
243 * field is not supported, its value is 0.
244 * All byte-related statistics do not include Ethernet FCS regardless
245 * of whether these bytes have been delivered to the application
246 * (see DEV_RX_OFFLOAD_KEEP_CRC).
248 struct rte_eth_stats {
249 uint64_t ipackets; /**< Total number of successfully received packets. */
250 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
251 uint64_t ibytes; /**< Total number of successfully received bytes. */
252 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
254 /**< Total of RX packets dropped by the HW,
255 * because there are no available buffer (i.e. RX queues are full).
257 uint64_t ierrors; /**< Total number of erroneous received packets. */
258 uint64_t oerrors; /**< Total number of failed transmitted packets. */
259 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
260 /* Queue stats are limited to max 256 queues */
261 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
262 /**< Total number of queue RX packets. */
263 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
264 /**< Total number of queue TX packets. */
265 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
266 /**< Total number of successfully received queue bytes. */
267 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
268 /**< Total number of successfully transmitted queue bytes. */
269 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
270 /**< Total number of queue packets received that are dropped. */
274 * Device supported speeds bitmap flags
276 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
277 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
278 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
279 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
280 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
281 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
282 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
283 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
284 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
285 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
286 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
287 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
288 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
289 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
290 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
291 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
292 #define ETH_LINK_SPEED_200G (1 << 15) /**< 200 Gbps */
295 * Ethernet numeric link speeds in Mbps
297 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
298 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
299 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
300 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
301 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
302 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
303 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
304 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
305 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
306 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
307 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
308 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
309 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
310 #define ETH_SPEED_NUM_200G 200000 /**< 200 Gbps */
311 #define ETH_SPEED_NUM_UNKNOWN UINT32_MAX /**< Unknown */
314 * A structure used to retrieve link-level information of an Ethernet port.
317 struct rte_eth_link {
318 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
319 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
320 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
321 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
322 } __rte_aligned(8); /**< aligned for atomic64 read/write */
324 /* Utility constants */
325 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
326 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
327 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
328 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
329 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
330 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
331 #define RTE_ETH_LINK_MAX_STR_LEN 40 /**< Max length of default link string. */
334 * A structure used to configure the ring threshold registers of an RX/TX
335 * queue for an Ethernet port.
337 struct rte_eth_thresh {
338 uint8_t pthresh; /**< Ring prefetch threshold. */
339 uint8_t hthresh; /**< Ring host threshold. */
340 uint8_t wthresh; /**< Ring writeback threshold. */
344 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
346 #define ETH_MQ_RX_RSS_FLAG 0x1
347 #define ETH_MQ_RX_DCB_FLAG 0x2
348 #define ETH_MQ_RX_VMDQ_FLAG 0x4
351 * A set of values to identify what method is to be used to route
352 * packets to multiple queues.
354 enum rte_eth_rx_mq_mode {
355 /** None of DCB,RSS or VMDQ mode */
358 /** For RX side, only RSS is on */
359 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
360 /** For RX side,only DCB is on. */
361 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
362 /** Both DCB and RSS enable */
363 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
365 /** Only VMDQ, no RSS nor DCB */
366 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
367 /** RSS mode with VMDQ */
368 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
369 /** Use VMDQ+DCB to route traffic to queues */
370 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
371 /** Enable both VMDQ and DCB in VMDq */
372 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
377 * for rx mq mode backward compatible
379 #define ETH_RSS ETH_MQ_RX_RSS
380 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
381 #define ETH_DCB_RX ETH_MQ_RX_DCB
384 * A set of values to identify what method is to be used to transmit
385 * packets using multi-TCs.
387 enum rte_eth_tx_mq_mode {
388 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
389 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
390 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
391 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
395 * for tx mq mode backward compatible
397 #define ETH_DCB_NONE ETH_MQ_TX_NONE
398 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
399 #define ETH_DCB_TX ETH_MQ_TX_DCB
402 * A structure used to configure the RX features of an Ethernet port.
404 struct rte_eth_rxmode {
405 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
406 enum rte_eth_rx_mq_mode mq_mode;
407 uint32_t max_rx_pkt_len; /**< Only used if JUMBO_FRAME enabled. */
408 /** Maximum allowed size of LRO aggregated packet. */
409 uint32_t max_lro_pkt_size;
410 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
412 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
413 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
414 * structure are allowed to be set.
418 uint64_t reserved_64s[2]; /**< Reserved for future fields */
419 void *reserved_ptrs[2]; /**< Reserved for future fields */
423 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
424 * Note that single VLAN is treated the same as inner VLAN.
427 ETH_VLAN_TYPE_UNKNOWN = 0,
428 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
429 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
434 * A structure used to describe a vlan filter.
435 * If the bit corresponding to a VID is set, such VID is on.
437 struct rte_vlan_filter_conf {
442 * A structure used to configure the Receive Side Scaling (RSS) feature
443 * of an Ethernet port.
444 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
445 * to an array holding the RSS key to use for hashing specific header
446 * fields of received packets. The length of this array should be indicated
447 * by *rss_key_len* below. Otherwise, a default random hash key is used by
450 * The *rss_key_len* field of the *rss_conf* structure indicates the length
451 * in bytes of the array pointed by *rss_key*. To be compatible, this length
452 * will be checked in i40e only. Others assume 40 bytes to be used as before.
454 * The *rss_hf* field of the *rss_conf* structure indicates the different
455 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
456 * Supplying an *rss_hf* equal to zero disables the RSS feature.
458 struct rte_eth_rss_conf {
459 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
460 uint8_t rss_key_len; /**< hash key length in bytes. */
461 uint64_t rss_hf; /**< Hash functions to apply - see below. */
465 * A packet can be identified by hardware as different flow types. Different
466 * NIC hardware may support different flow types.
467 * Basically, the NIC hardware identifies the flow type as deep protocol as
468 * possible, and exclusively. For example, if a packet is identified as
469 * 'RTE_ETH_FLOW_NONFRAG_IPV4_TCP', it will not be any of other flow types,
470 * though it is an actual IPV4 packet.
472 #define RTE_ETH_FLOW_UNKNOWN 0
473 #define RTE_ETH_FLOW_RAW 1
474 #define RTE_ETH_FLOW_IPV4 2
475 #define RTE_ETH_FLOW_FRAG_IPV4 3
476 #define RTE_ETH_FLOW_NONFRAG_IPV4_TCP 4
477 #define RTE_ETH_FLOW_NONFRAG_IPV4_UDP 5
478 #define RTE_ETH_FLOW_NONFRAG_IPV4_SCTP 6
479 #define RTE_ETH_FLOW_NONFRAG_IPV4_OTHER 7
480 #define RTE_ETH_FLOW_IPV6 8
481 #define RTE_ETH_FLOW_FRAG_IPV6 9
482 #define RTE_ETH_FLOW_NONFRAG_IPV6_TCP 10
483 #define RTE_ETH_FLOW_NONFRAG_IPV6_UDP 11
484 #define RTE_ETH_FLOW_NONFRAG_IPV6_SCTP 12
485 #define RTE_ETH_FLOW_NONFRAG_IPV6_OTHER 13
486 #define RTE_ETH_FLOW_L2_PAYLOAD 14
487 #define RTE_ETH_FLOW_IPV6_EX 15
488 #define RTE_ETH_FLOW_IPV6_TCP_EX 16
489 #define RTE_ETH_FLOW_IPV6_UDP_EX 17
490 #define RTE_ETH_FLOW_PORT 18
491 /**< Consider device port number as a flow differentiator */
492 #define RTE_ETH_FLOW_VXLAN 19 /**< VXLAN protocol based flow */
493 #define RTE_ETH_FLOW_GENEVE 20 /**< GENEVE protocol based flow */
494 #define RTE_ETH_FLOW_NVGRE 21 /**< NVGRE protocol based flow */
495 #define RTE_ETH_FLOW_VXLAN_GPE 22 /**< VXLAN-GPE protocol based flow */
496 #define RTE_ETH_FLOW_GTPU 23 /**< GTPU protocol based flow */
497 #define RTE_ETH_FLOW_MAX 24
500 * Below macros are defined for RSS offload types, they can be used to
501 * fill rte_eth_rss_conf.rss_hf or rte_flow_action_rss.types.
503 #define ETH_RSS_IPV4 (1ULL << 2)
504 #define ETH_RSS_FRAG_IPV4 (1ULL << 3)
505 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << 4)
506 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << 5)
507 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << 6)
508 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << 7)
509 #define ETH_RSS_IPV6 (1ULL << 8)
510 #define ETH_RSS_FRAG_IPV6 (1ULL << 9)
511 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << 10)
512 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << 11)
513 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << 12)
514 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << 13)
515 #define ETH_RSS_L2_PAYLOAD (1ULL << 14)
516 #define ETH_RSS_IPV6_EX (1ULL << 15)
517 #define ETH_RSS_IPV6_TCP_EX (1ULL << 16)
518 #define ETH_RSS_IPV6_UDP_EX (1ULL << 17)
519 #define ETH_RSS_PORT (1ULL << 18)
520 #define ETH_RSS_VXLAN (1ULL << 19)
521 #define ETH_RSS_GENEVE (1ULL << 20)
522 #define ETH_RSS_NVGRE (1ULL << 21)
523 #define ETH_RSS_GTPU (1ULL << 23)
524 #define ETH_RSS_ETH (1ULL << 24)
525 #define ETH_RSS_S_VLAN (1ULL << 25)
526 #define ETH_RSS_C_VLAN (1ULL << 26)
527 #define ETH_RSS_ESP (1ULL << 27)
528 #define ETH_RSS_AH (1ULL << 28)
529 #define ETH_RSS_L2TPV3 (1ULL << 29)
530 #define ETH_RSS_PFCP (1ULL << 30)
531 #define ETH_RSS_PPPOE (1ULL << 31)
532 #define ETH_RSS_ECPRI (1ULL << 32)
533 #define ETH_RSS_MPLS (1ULL << 33)
536 * We use the following macros to combine with above ETH_RSS_* for
537 * more specific input set selection. These bits are defined starting
538 * from the high end of the 64 bits.
539 * Note: If we use above ETH_RSS_* without SRC/DST_ONLY, it represents
540 * both SRC and DST are taken into account. If SRC_ONLY and DST_ONLY of
541 * the same level are used simultaneously, it is the same case as none of
544 #define ETH_RSS_L3_SRC_ONLY (1ULL << 63)
545 #define ETH_RSS_L3_DST_ONLY (1ULL << 62)
546 #define ETH_RSS_L4_SRC_ONLY (1ULL << 61)
547 #define ETH_RSS_L4_DST_ONLY (1ULL << 60)
548 #define ETH_RSS_L2_SRC_ONLY (1ULL << 59)
549 #define ETH_RSS_L2_DST_ONLY (1ULL << 58)
552 * Only select IPV6 address prefix as RSS input set according to
553 * https://tools.ietf.org/html/rfc6052
554 * Must be combined with ETH_RSS_IPV6, ETH_RSS_NONFRAG_IPV6_UDP,
555 * ETH_RSS_NONFRAG_IPV6_TCP, ETH_RSS_NONFRAG_IPV6_SCTP.
557 #define RTE_ETH_RSS_L3_PRE32 (1ULL << 57)
558 #define RTE_ETH_RSS_L3_PRE40 (1ULL << 56)
559 #define RTE_ETH_RSS_L3_PRE48 (1ULL << 55)
560 #define RTE_ETH_RSS_L3_PRE56 (1ULL << 54)
561 #define RTE_ETH_RSS_L3_PRE64 (1ULL << 53)
562 #define RTE_ETH_RSS_L3_PRE96 (1ULL << 52)
565 * Use the following macros to combine with the above layers
566 * to choose inner and outer layers or both for RSS computation.
567 * Bits 50 and 51 are reserved for this.
571 * level 0, requests the default behavior.
572 * Depending on the packet type, it can mean outermost, innermost,
573 * anything in between or even no RSS.
574 * It basically stands for the innermost encapsulation level RSS
575 * can be performed on according to PMD and device capabilities.
577 #define ETH_RSS_LEVEL_PMD_DEFAULT (0ULL << 50)
580 * level 1, requests RSS to be performed on the outermost packet
581 * encapsulation level.
583 #define ETH_RSS_LEVEL_OUTERMOST (1ULL << 50)
586 * level 2, requests RSS to be performed on the specified inner packet
587 * encapsulation level, from outermost to innermost (lower to higher values).
589 #define ETH_RSS_LEVEL_INNERMOST (2ULL << 50)
590 #define ETH_RSS_LEVEL_MASK (3ULL << 50)
592 #define ETH_RSS_LEVEL(rss_hf) ((rss_hf & ETH_RSS_LEVEL_MASK) >> 50)
595 * For input set change of hash filter, if SRC_ONLY and DST_ONLY of
596 * the same level are used simultaneously, it is the same case as
597 * none of them are added.
600 * RSS types with SRC/DST_ONLY.
604 static inline uint64_t
605 rte_eth_rss_hf_refine(uint64_t rss_hf)
607 if ((rss_hf & ETH_RSS_L3_SRC_ONLY) && (rss_hf & ETH_RSS_L3_DST_ONLY))
608 rss_hf &= ~(ETH_RSS_L3_SRC_ONLY | ETH_RSS_L3_DST_ONLY);
610 if ((rss_hf & ETH_RSS_L4_SRC_ONLY) && (rss_hf & ETH_RSS_L4_DST_ONLY))
611 rss_hf &= ~(ETH_RSS_L4_SRC_ONLY | ETH_RSS_L4_DST_ONLY);
616 #define ETH_RSS_IPV6_PRE32 ( \
618 RTE_ETH_RSS_L3_PRE32)
620 #define ETH_RSS_IPV6_PRE40 ( \
622 RTE_ETH_RSS_L3_PRE40)
624 #define ETH_RSS_IPV6_PRE48 ( \
626 RTE_ETH_RSS_L3_PRE48)
628 #define ETH_RSS_IPV6_PRE56 ( \
630 RTE_ETH_RSS_L3_PRE56)
632 #define ETH_RSS_IPV6_PRE64 ( \
634 RTE_ETH_RSS_L3_PRE64)
636 #define ETH_RSS_IPV6_PRE96 ( \
638 RTE_ETH_RSS_L3_PRE96)
640 #define ETH_RSS_IPV6_PRE32_UDP ( \
641 ETH_RSS_NONFRAG_IPV6_UDP | \
642 RTE_ETH_RSS_L3_PRE32)
644 #define ETH_RSS_IPV6_PRE40_UDP ( \
645 ETH_RSS_NONFRAG_IPV6_UDP | \
646 RTE_ETH_RSS_L3_PRE40)
648 #define ETH_RSS_IPV6_PRE48_UDP ( \
649 ETH_RSS_NONFRAG_IPV6_UDP | \
650 RTE_ETH_RSS_L3_PRE48)
652 #define ETH_RSS_IPV6_PRE56_UDP ( \
653 ETH_RSS_NONFRAG_IPV6_UDP | \
654 RTE_ETH_RSS_L3_PRE56)
656 #define ETH_RSS_IPV6_PRE64_UDP ( \
657 ETH_RSS_NONFRAG_IPV6_UDP | \
658 RTE_ETH_RSS_L3_PRE64)
660 #define ETH_RSS_IPV6_PRE96_UDP ( \
661 ETH_RSS_NONFRAG_IPV6_UDP | \
662 RTE_ETH_RSS_L3_PRE96)
664 #define ETH_RSS_IPV6_PRE32_TCP ( \
665 ETH_RSS_NONFRAG_IPV6_TCP | \
666 RTE_ETH_RSS_L3_PRE32)
668 #define ETH_RSS_IPV6_PRE40_TCP ( \
669 ETH_RSS_NONFRAG_IPV6_TCP | \
670 RTE_ETH_RSS_L3_PRE40)
672 #define ETH_RSS_IPV6_PRE48_TCP ( \
673 ETH_RSS_NONFRAG_IPV6_TCP | \
674 RTE_ETH_RSS_L3_PRE48)
676 #define ETH_RSS_IPV6_PRE56_TCP ( \
677 ETH_RSS_NONFRAG_IPV6_TCP | \
678 RTE_ETH_RSS_L3_PRE56)
680 #define ETH_RSS_IPV6_PRE64_TCP ( \
681 ETH_RSS_NONFRAG_IPV6_TCP | \
682 RTE_ETH_RSS_L3_PRE64)
684 #define ETH_RSS_IPV6_PRE96_TCP ( \
685 ETH_RSS_NONFRAG_IPV6_TCP | \
686 RTE_ETH_RSS_L3_PRE96)
688 #define ETH_RSS_IPV6_PRE32_SCTP ( \
689 ETH_RSS_NONFRAG_IPV6_SCTP | \
690 RTE_ETH_RSS_L3_PRE32)
692 #define ETH_RSS_IPV6_PRE40_SCTP ( \
693 ETH_RSS_NONFRAG_IPV6_SCTP | \
694 RTE_ETH_RSS_L3_PRE40)
696 #define ETH_RSS_IPV6_PRE48_SCTP ( \
697 ETH_RSS_NONFRAG_IPV6_SCTP | \
698 RTE_ETH_RSS_L3_PRE48)
700 #define ETH_RSS_IPV6_PRE56_SCTP ( \
701 ETH_RSS_NONFRAG_IPV6_SCTP | \
702 RTE_ETH_RSS_L3_PRE56)
704 #define ETH_RSS_IPV6_PRE64_SCTP ( \
705 ETH_RSS_NONFRAG_IPV6_SCTP | \
706 RTE_ETH_RSS_L3_PRE64)
708 #define ETH_RSS_IPV6_PRE96_SCTP ( \
709 ETH_RSS_NONFRAG_IPV6_SCTP | \
710 RTE_ETH_RSS_L3_PRE96)
712 #define ETH_RSS_IP ( \
714 ETH_RSS_FRAG_IPV4 | \
715 ETH_RSS_NONFRAG_IPV4_OTHER | \
717 ETH_RSS_FRAG_IPV6 | \
718 ETH_RSS_NONFRAG_IPV6_OTHER | \
721 #define ETH_RSS_UDP ( \
722 ETH_RSS_NONFRAG_IPV4_UDP | \
723 ETH_RSS_NONFRAG_IPV6_UDP | \
726 #define ETH_RSS_TCP ( \
727 ETH_RSS_NONFRAG_IPV4_TCP | \
728 ETH_RSS_NONFRAG_IPV6_TCP | \
731 #define ETH_RSS_SCTP ( \
732 ETH_RSS_NONFRAG_IPV4_SCTP | \
733 ETH_RSS_NONFRAG_IPV6_SCTP)
735 #define ETH_RSS_TUNNEL ( \
740 #define ETH_RSS_VLAN ( \
744 /**< Mask of valid RSS hash protocols */
745 #define ETH_RSS_PROTO_MASK ( \
747 ETH_RSS_FRAG_IPV4 | \
748 ETH_RSS_NONFRAG_IPV4_TCP | \
749 ETH_RSS_NONFRAG_IPV4_UDP | \
750 ETH_RSS_NONFRAG_IPV4_SCTP | \
751 ETH_RSS_NONFRAG_IPV4_OTHER | \
753 ETH_RSS_FRAG_IPV6 | \
754 ETH_RSS_NONFRAG_IPV6_TCP | \
755 ETH_RSS_NONFRAG_IPV6_UDP | \
756 ETH_RSS_NONFRAG_IPV6_SCTP | \
757 ETH_RSS_NONFRAG_IPV6_OTHER | \
758 ETH_RSS_L2_PAYLOAD | \
760 ETH_RSS_IPV6_TCP_EX | \
761 ETH_RSS_IPV6_UDP_EX | \
769 * Definitions used for redirection table entry size.
770 * Some RSS RETA sizes may not be supported by some drivers, check the
771 * documentation or the description of relevant functions for more details.
773 #define ETH_RSS_RETA_SIZE_64 64
774 #define ETH_RSS_RETA_SIZE_128 128
775 #define ETH_RSS_RETA_SIZE_256 256
776 #define ETH_RSS_RETA_SIZE_512 512
777 #define RTE_RETA_GROUP_SIZE 64
779 /* Definitions used for VMDQ and DCB functionality */
780 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
781 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
782 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
783 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
785 /* DCB capability defines */
786 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
787 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
789 /* Definitions used for VLAN Offload functionality */
790 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
791 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
792 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
793 #define ETH_QINQ_STRIP_OFFLOAD 0x0008 /**< QINQ Strip On/Off */
795 /* Definitions used for mask VLAN setting */
796 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
797 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
798 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
799 #define ETH_QINQ_STRIP_MASK 0x0008 /**< QINQ Strip setting mask */
800 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
802 /* Definitions used for receive MAC address */
803 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
805 /* Definitions used for unicast hash */
806 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
808 /* Definitions used for VMDQ pool rx mode setting */
809 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
810 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
811 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
812 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
813 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
815 /** Maximum nb. of vlan per mirror rule */
816 #define ETH_MIRROR_MAX_VLANS 64
818 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
819 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
820 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
821 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
822 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
825 * A structure used to configure VLAN traffic mirror of an Ethernet port.
827 struct rte_eth_vlan_mirror {
828 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
829 /** VLAN ID list for vlan mirroring. */
830 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
834 * A structure used to configure traffic mirror of an Ethernet port.
836 struct rte_eth_mirror_conf {
837 uint8_t rule_type; /**< Mirroring rule type */
838 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
839 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
840 /** VLAN ID setting for VLAN mirroring. */
841 struct rte_eth_vlan_mirror vlan;
845 * A structure used to configure 64 entries of Redirection Table of the
846 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
847 * more than 64 entries supported by hardware, an array of this structure
850 struct rte_eth_rss_reta_entry64 {
852 /**< Mask bits indicate which entries need to be updated/queried. */
853 uint16_t reta[RTE_RETA_GROUP_SIZE];
854 /**< Group of 64 redirection table entries. */
858 * This enum indicates the possible number of traffic classes
859 * in DCB configurations
861 enum rte_eth_nb_tcs {
862 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
863 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
867 * This enum indicates the possible number of queue pools
868 * in VMDQ configurations.
870 enum rte_eth_nb_pools {
871 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
872 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
873 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
874 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
877 /* This structure may be extended in future. */
878 struct rte_eth_dcb_rx_conf {
879 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
880 /** Traffic class each UP mapped to. */
881 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
884 struct rte_eth_vmdq_dcb_tx_conf {
885 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
886 /** Traffic class each UP mapped to. */
887 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
890 struct rte_eth_dcb_tx_conf {
891 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
892 /** Traffic class each UP mapped to. */
893 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
896 struct rte_eth_vmdq_tx_conf {
897 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
901 * A structure used to configure the VMDQ+DCB feature
902 * of an Ethernet port.
904 * Using this feature, packets are routed to a pool of queues, based
905 * on the vlan id in the vlan tag, and then to a specific queue within
906 * that pool, using the user priority vlan tag field.
908 * A default pool may be used, if desired, to route all traffic which
909 * does not match the vlan filter rules.
911 struct rte_eth_vmdq_dcb_conf {
912 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
913 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
914 uint8_t default_pool; /**< The default pool, if applicable */
915 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
917 uint16_t vlan_id; /**< The vlan id of the received frame */
918 uint64_t pools; /**< Bitmask of pools for packet rx */
919 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
920 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
921 /**< Selects a queue in a pool */
925 * A structure used to configure the VMDQ feature of an Ethernet port when
926 * not combined with the DCB feature.
928 * Using this feature, packets are routed to a pool of queues. By default,
929 * the pool selection is based on the MAC address, the vlan id in the
930 * vlan tag as specified in the pool_map array.
931 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
932 * selection using only the MAC address. MAC address to pool mapping is done
933 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
934 * corresponding to the pool id.
936 * Queue selection within the selected pool will be done using RSS when
937 * it is enabled or revert to the first queue of the pool if not.
939 * A default pool may be used, if desired, to route all traffic which
940 * does not match the vlan filter rules or any pool MAC address.
942 struct rte_eth_vmdq_rx_conf {
943 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
944 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
945 uint8_t default_pool; /**< The default pool, if applicable */
946 uint8_t enable_loop_back; /**< Enable VT loop back */
947 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
948 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
950 uint16_t vlan_id; /**< The vlan id of the received frame */
951 uint64_t pools; /**< Bitmask of pools for packet rx */
952 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
956 * A structure used to configure the TX features of an Ethernet port.
958 struct rte_eth_txmode {
959 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
961 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
962 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
963 * structure are allowed to be set.
969 uint8_t hw_vlan_reject_tagged : 1,
970 /**< If set, reject sending out tagged pkts */
971 hw_vlan_reject_untagged : 1,
972 /**< If set, reject sending out untagged pkts */
973 hw_vlan_insert_pvid : 1;
974 /**< If set, enable port based VLAN insertion */
976 uint64_t reserved_64s[2]; /**< Reserved for future fields */
977 void *reserved_ptrs[2]; /**< Reserved for future fields */
982 * @b EXPERIMENTAL: this structure may change without prior notice.
984 * A structure used to configure an Rx packet segment to split.
986 * If RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT flag is set in offloads field,
987 * the PMD will split the received packets into multiple segments
988 * according to the specification in the description array:
990 * - The first network buffer will be allocated from the memory pool,
991 * specified in the first array element, the second buffer, from the
992 * pool in the second element, and so on.
994 * - The offsets from the segment description elements specify
995 * the data offset from the buffer beginning except the first mbuf.
996 * The first segment offset is added with RTE_PKTMBUF_HEADROOM.
998 * - The lengths in the elements define the maximal data amount
999 * being received to each segment. The receiving starts with filling
1000 * up the first mbuf data buffer up to specified length. If the
1001 * there are data remaining (packet is longer than buffer in the first
1002 * mbuf) the following data will be pushed to the next segment
1003 * up to its own length, and so on.
1005 * - If the length in the segment description element is zero
1006 * the actual buffer size will be deduced from the appropriate
1007 * memory pool properties.
1009 * - If there is not enough elements to describe the buffer for entire
1010 * packet of maximal length the following parameters will be used
1011 * for the all remaining segments:
1012 * - pool from the last valid element
1013 * - the buffer size from this pool
1016 struct rte_eth_rxseg_split {
1017 struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
1018 uint16_t length; /**< Segment data length, configures split point. */
1019 uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */
1020 uint32_t reserved; /**< Reserved field. */
1025 * @b EXPERIMENTAL: this structure may change without prior notice.
1027 * A common structure used to describe Rx packet segment properties.
1029 union rte_eth_rxseg {
1030 /* The settings for buffer split offload. */
1031 struct rte_eth_rxseg_split split;
1032 /* The other features settings should be added here. */
1036 * A structure used to configure an RX ring of an Ethernet port.
1038 struct rte_eth_rxconf {
1039 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
1040 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
1041 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
1042 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1043 uint16_t rx_nseg; /**< Number of descriptions in rx_seg array. */
1045 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
1046 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
1047 * fields on rte_eth_dev_info structure are allowed to be set.
1051 * Points to the array of segment descriptions for an entire packet.
1052 * Array elements are properties for consecutive Rx segments.
1054 * The supported capabilities of receiving segmentation is reported
1055 * in rte_eth_dev_info.rx_seg_capa field.
1057 union rte_eth_rxseg *rx_seg;
1059 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1060 void *reserved_ptrs[2]; /**< Reserved for future fields */
1064 * A structure used to configure a TX ring of an Ethernet port.
1066 struct rte_eth_txconf {
1067 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
1068 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
1069 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
1070 less free descriptors than this value. */
1072 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1074 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
1075 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
1076 * fields on rte_eth_dev_info structure are allowed to be set.
1080 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1081 void *reserved_ptrs[2]; /**< Reserved for future fields */
1086 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1088 * A structure used to return the hairpin capabilities that are supported.
1090 struct rte_eth_hairpin_cap {
1091 /** The max number of hairpin queues (different bindings). */
1092 uint16_t max_nb_queues;
1093 /** Max number of Rx queues to be connected to one Tx queue. */
1094 uint16_t max_rx_2_tx;
1095 /** Max number of Tx queues to be connected to one Rx queue. */
1096 uint16_t max_tx_2_rx;
1097 uint16_t max_nb_desc; /**< The max num of descriptors. */
1100 #define RTE_ETH_MAX_HAIRPIN_PEERS 32
1104 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1106 * A structure used to hold hairpin peer data.
1108 struct rte_eth_hairpin_peer {
1109 uint16_t port; /**< Peer port. */
1110 uint16_t queue; /**< Peer queue. */
1115 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1117 * A structure used to configure hairpin binding.
1119 struct rte_eth_hairpin_conf {
1120 uint32_t peer_count:16; /**< The number of peers. */
1123 * Explicit Tx flow rule mode.
1124 * One hairpin pair of queues should have the same attribute.
1126 * - When set, the user should be responsible for inserting the hairpin
1127 * Tx part flows and removing them.
1128 * - When clear, the PMD will try to handle the Tx part of the flows,
1129 * e.g., by splitting one flow into two parts.
1131 uint32_t tx_explicit:1;
1134 * Manually bind hairpin queues.
1135 * One hairpin pair of queues should have the same attribute.
1137 * - When set, to enable hairpin, the user should call the hairpin bind
1138 * function after all the queues are set up properly and the ports are
1139 * started. Also, the hairpin unbind function should be called
1140 * accordingly before stopping a port that with hairpin configured.
1141 * - When clear, the PMD will try to enable the hairpin with the queues
1142 * configured automatically during port start.
1144 uint32_t manual_bind:1;
1145 uint32_t reserved:14; /**< Reserved bits. */
1146 struct rte_eth_hairpin_peer peers[RTE_ETH_MAX_HAIRPIN_PEERS];
1150 * A structure contains information about HW descriptor ring limitations.
1152 struct rte_eth_desc_lim {
1153 uint16_t nb_max; /**< Max allowed number of descriptors. */
1154 uint16_t nb_min; /**< Min allowed number of descriptors. */
1155 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
1158 * Max allowed number of segments per whole packet.
1160 * - For TSO packet this is the total number of data descriptors allowed
1163 * @see nb_mtu_seg_max
1165 uint16_t nb_seg_max;
1168 * Max number of segments per one MTU.
1170 * - For non-TSO packet, this is the maximum allowed number of segments
1171 * in a single transmit packet.
1173 * - For TSO packet each segment within the TSO may span up to this
1178 uint16_t nb_mtu_seg_max;
1182 * This enum indicates the flow control mode
1184 enum rte_eth_fc_mode {
1185 RTE_FC_NONE = 0, /**< Disable flow control. */
1186 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
1187 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
1188 RTE_FC_FULL /**< Enable flow control on both side. */
1192 * A structure used to configure Ethernet flow control parameter.
1193 * These parameters will be configured into the register of the NIC.
1194 * Please refer to the corresponding data sheet for proper value.
1196 struct rte_eth_fc_conf {
1197 uint32_t high_water; /**< High threshold value to trigger XOFF */
1198 uint32_t low_water; /**< Low threshold value to trigger XON */
1199 uint16_t pause_time; /**< Pause quota in the Pause frame */
1200 uint16_t send_xon; /**< Is XON frame need be sent */
1201 enum rte_eth_fc_mode mode; /**< Link flow control mode */
1202 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
1203 uint8_t autoneg; /**< Use Pause autoneg */
1207 * A structure used to configure Ethernet priority flow control parameter.
1208 * These parameters will be configured into the register of the NIC.
1209 * Please refer to the corresponding data sheet for proper value.
1211 struct rte_eth_pfc_conf {
1212 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
1213 uint8_t priority; /**< VLAN User Priority. */
1217 * Tunnel type for device-specific classifier configuration.
1218 * @see rte_eth_udp_tunnel
1220 enum rte_eth_tunnel_type {
1221 RTE_TUNNEL_TYPE_NONE = 0,
1222 RTE_TUNNEL_TYPE_VXLAN,
1223 RTE_TUNNEL_TYPE_GENEVE,
1224 RTE_TUNNEL_TYPE_TEREDO,
1225 RTE_TUNNEL_TYPE_NVGRE,
1226 RTE_TUNNEL_TYPE_IP_IN_GRE,
1227 RTE_L2_TUNNEL_TYPE_E_TAG,
1228 RTE_TUNNEL_TYPE_VXLAN_GPE,
1229 RTE_TUNNEL_TYPE_ECPRI,
1230 RTE_TUNNEL_TYPE_MAX,
1233 /* Deprecated API file for rte_eth_dev_filter_* functions */
1234 #include "rte_eth_ctrl.h"
1237 * Memory space that can be configured to store Flow Director filters
1238 * in the board memory.
1240 enum rte_fdir_pballoc_type {
1241 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
1242 RTE_FDIR_PBALLOC_128K, /**< 128k. */
1243 RTE_FDIR_PBALLOC_256K, /**< 256k. */
1247 * Select report mode of FDIR hash information in RX descriptors.
1249 enum rte_fdir_status_mode {
1250 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
1251 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
1252 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
1256 * A structure used to configure the Flow Director (FDIR) feature
1257 * of an Ethernet port.
1259 * If mode is RTE_FDIR_MODE_NONE, the pballoc value is ignored.
1261 struct rte_fdir_conf {
1262 enum rte_fdir_mode mode; /**< Flow Director mode. */
1263 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
1264 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
1265 /** RX queue of packets matching a "drop" filter in perfect mode. */
1267 struct rte_eth_fdir_masks mask;
1268 struct rte_eth_fdir_flex_conf flex_conf;
1269 /**< Flex payload configuration. */
1273 * UDP tunneling configuration.
1275 * Used to configure the classifier of a device,
1276 * associating an UDP port with a type of tunnel.
1278 * Some NICs may need such configuration to properly parse a tunnel
1279 * with any standard or custom UDP port.
1281 struct rte_eth_udp_tunnel {
1282 uint16_t udp_port; /**< UDP port used for the tunnel. */
1283 uint8_t prot_type; /**< Tunnel type. @see rte_eth_tunnel_type */
1287 * A structure used to enable/disable specific device interrupts.
1289 struct rte_intr_conf {
1290 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
1292 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
1294 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
1299 * A structure used to configure an Ethernet port.
1300 * Depending upon the RX multi-queue mode, extra advanced
1301 * configuration settings may be needed.
1303 struct rte_eth_conf {
1304 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
1305 used. ETH_LINK_SPEED_FIXED disables link
1306 autonegotiation, and a unique speed shall be
1307 set. Otherwise, the bitmap defines the set of
1308 speeds to be advertised. If the special value
1309 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
1310 supported are advertised. */
1311 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
1312 struct rte_eth_txmode txmode; /**< Port TX configuration. */
1313 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
1314 is 0, meaning the loopback mode is disabled.
1315 Read the datasheet of given ethernet controller
1316 for details. The possible values of this field
1317 are defined in implementation of each driver. */
1319 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
1320 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
1321 /**< Port vmdq+dcb configuration. */
1322 struct rte_eth_dcb_rx_conf dcb_rx_conf;
1323 /**< Port dcb RX configuration. */
1324 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
1325 /**< Port vmdq RX configuration. */
1326 } rx_adv_conf; /**< Port RX filtering configuration. */
1328 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
1329 /**< Port vmdq+dcb TX configuration. */
1330 struct rte_eth_dcb_tx_conf dcb_tx_conf;
1331 /**< Port dcb TX configuration. */
1332 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
1333 /**< Port vmdq TX configuration. */
1334 } tx_adv_conf; /**< Port TX DCB configuration (union). */
1335 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
1336 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
1337 uint32_t dcb_capability_en;
1338 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. DEPRECATED */
1339 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
1343 * RX offload capabilities of a device.
1345 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
1346 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
1347 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
1348 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
1349 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
1350 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
1351 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
1352 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
1353 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
1354 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
1355 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
1356 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
1357 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
1359 * Timestamp is set by the driver in RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1360 * and RTE_MBUF_DYNFLAG_RX_TIMESTAMP_NAME is set in ol_flags.
1361 * The mbuf field and flag are registered when the offload is configured.
1363 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
1364 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
1365 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
1366 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
1367 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
1368 #define DEV_RX_OFFLOAD_RSS_HASH 0x00080000
1369 #define RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT 0x00100000
1371 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
1372 DEV_RX_OFFLOAD_UDP_CKSUM | \
1373 DEV_RX_OFFLOAD_TCP_CKSUM)
1374 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
1375 DEV_RX_OFFLOAD_VLAN_FILTER | \
1376 DEV_RX_OFFLOAD_VLAN_EXTEND | \
1377 DEV_RX_OFFLOAD_QINQ_STRIP)
1380 * If new Rx offload capabilities are defined, they also must be
1381 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
1385 * TX offload capabilities of a device.
1387 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
1388 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
1389 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
1390 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
1391 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
1392 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
1393 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
1394 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
1395 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
1396 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
1397 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
1398 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
1399 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
1400 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
1401 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
1402 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
1403 * tx queue without SW lock.
1405 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
1406 /**< Device supports multi segment send. */
1407 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
1408 /**< Device supports optimization for fast release of mbufs.
1409 * When set application must guarantee that per-queue all mbufs comes from
1410 * the same mempool and has refcnt = 1.
1412 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
1414 * Device supports generic UDP tunneled packet TSO.
1415 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
1418 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
1420 * Device supports generic IP tunneled packet TSO.
1421 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
1424 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
1425 /** Device supports outer UDP checksum */
1426 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
1428 * Device sends on time read from RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1429 * if RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME is set in ol_flags.
1430 * The mbuf field and flag are registered when the offload is configured.
1432 #define DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP 0x00200000
1434 * If new Tx offload capabilities are defined, they also must be
1435 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1438 /**@{@name Device capabilities
1439 * Non-offload capabilities reported in rte_eth_dev_info.dev_capa.
1441 /** Device supports Rx queue setup after device started. */
1442 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
1443 /** Device supports Tx queue setup after device started. */
1444 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
1448 * Fallback default preferred Rx/Tx port parameters.
1449 * These are used if an application requests default parameters
1450 * but the PMD does not provide preferred values.
1452 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1453 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1454 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1455 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1458 * Preferred Rx/Tx port parameters.
1459 * There are separate instances of this structure for transmission
1460 * and reception respectively.
1462 struct rte_eth_dev_portconf {
1463 uint16_t burst_size; /**< Device-preferred burst size */
1464 uint16_t ring_size; /**< Device-preferred size of queue rings */
1465 uint16_t nb_queues; /**< Device-preferred number of queues */
1469 * Default values for switch domain id when ethdev does not support switch
1470 * domain definitions.
1472 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (UINT16_MAX)
1475 * Ethernet device associated switch information
1477 struct rte_eth_switch_info {
1478 const char *name; /**< switch name */
1479 uint16_t domain_id; /**< switch domain id */
1482 * mapping to the devices physical switch port as enumerated from the
1483 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1484 * device this may correspond to the VF_ID of each virtual function,
1485 * but each driver should explicitly define the mapping of switch
1486 * port identifier to that physical interconnect/switch
1492 * @b EXPERIMENTAL: this structure may change without prior notice.
1494 * Ethernet device Rx buffer segmentation capabilities.
1496 struct rte_eth_rxseg_capa {
1498 uint32_t multi_pools:1; /**< Supports receiving to multiple pools.*/
1499 uint32_t offset_allowed:1; /**< Supports buffer offsets. */
1500 uint32_t offset_align_log2:4; /**< Required offset alignment. */
1501 uint16_t max_nseg; /**< Maximum amount of segments to split. */
1502 uint16_t reserved; /**< Reserved field. */
1506 * Ethernet device information
1510 * Ethernet device representor port type.
1512 enum rte_eth_representor_type {
1513 RTE_ETH_REPRESENTOR_NONE, /**< not a representor. */
1514 RTE_ETH_REPRESENTOR_VF, /**< representor of Virtual Function. */
1515 RTE_ETH_REPRESENTOR_SF, /**< representor of Sub Function. */
1516 RTE_ETH_REPRESENTOR_PF, /**< representor of Physical Function. */
1520 * A structure used to retrieve the contextual information of
1521 * an Ethernet device, such as the controlling driver of the
1524 struct rte_eth_dev_info {
1525 struct rte_device *device; /** Generic device information */
1526 const char *driver_name; /**< Device Driver name. */
1527 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1528 Use if_indextoname() to translate into an interface name. */
1529 uint16_t min_mtu; /**< Minimum MTU allowed */
1530 uint16_t max_mtu; /**< Maximum MTU allowed */
1531 const uint32_t *dev_flags; /**< Device flags */
1532 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1533 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1534 /** Maximum configurable size of LRO aggregated packet. */
1535 uint32_t max_lro_pkt_size;
1536 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1537 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1538 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1539 uint32_t max_hash_mac_addrs;
1540 /** Maximum number of hash MAC addresses for MTA and UTA. */
1541 uint16_t max_vfs; /**< Maximum number of VFs. */
1542 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1543 struct rte_eth_rxseg_capa rx_seg_capa; /**< Segmentation capability.*/
1544 uint64_t rx_offload_capa;
1545 /**< All RX offload capabilities including all per-queue ones */
1546 uint64_t tx_offload_capa;
1547 /**< All TX offload capabilities including all per-queue ones */
1548 uint64_t rx_queue_offload_capa;
1549 /**< Device per-queue RX offload capabilities. */
1550 uint64_t tx_queue_offload_capa;
1551 /**< Device per-queue TX offload capabilities. */
1553 /**< Device redirection table size, the total number of entries. */
1554 uint8_t hash_key_size; /**< Hash key size in bytes */
1555 /** Bit mask of RSS offloads, the bit offset also means flow type */
1556 uint64_t flow_type_rss_offloads;
1557 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1558 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1559 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1560 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1561 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1562 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1563 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1564 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1565 /** Configured number of rx/tx queues */
1566 uint16_t nb_rx_queues; /**< Number of RX queues. */
1567 uint16_t nb_tx_queues; /**< Number of TX queues. */
1568 /** Rx parameter recommendations */
1569 struct rte_eth_dev_portconf default_rxportconf;
1570 /** Tx parameter recommendations */
1571 struct rte_eth_dev_portconf default_txportconf;
1572 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1575 * Switching information for ports on a device with a
1576 * embedded managed interconnect/switch.
1578 struct rte_eth_switch_info switch_info;
1580 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1581 void *reserved_ptrs[2]; /**< Reserved for future fields */
1585 * Ethernet device RX queue information structure.
1586 * Used to retrieve information about configured queue.
1588 struct rte_eth_rxq_info {
1589 struct rte_mempool *mp; /**< mempool used by that queue. */
1590 struct rte_eth_rxconf conf; /**< queue config parameters. */
1591 uint8_t scattered_rx; /**< scattered packets RX supported. */
1592 uint16_t nb_desc; /**< configured number of RXDs. */
1593 uint16_t rx_buf_size; /**< hardware receive buffer size. */
1594 } __rte_cache_min_aligned;
1597 * Ethernet device TX queue information structure.
1598 * Used to retrieve information about configured queue.
1600 struct rte_eth_txq_info {
1601 struct rte_eth_txconf conf; /**< queue config parameters. */
1602 uint16_t nb_desc; /**< configured number of TXDs. */
1603 } __rte_cache_min_aligned;
1605 /* Generic Burst mode flag definition, values can be ORed. */
1608 * If the queues have different burst mode description, this bit will be set
1609 * by PMD, then the application can iterate to retrieve burst description for
1612 #define RTE_ETH_BURST_FLAG_PER_QUEUE (1ULL << 0)
1615 * Ethernet device RX/TX queue packet burst mode information structure.
1616 * Used to retrieve information about packet burst mode setting.
1618 struct rte_eth_burst_mode {
1619 uint64_t flags; /**< The ORed values of RTE_ETH_BURST_FLAG_xxx */
1621 #define RTE_ETH_BURST_MODE_INFO_SIZE 1024 /**< Maximum size for information */
1622 char info[RTE_ETH_BURST_MODE_INFO_SIZE]; /**< burst mode information */
1625 /** Maximum name length for extended statistics counters */
1626 #define RTE_ETH_XSTATS_NAME_SIZE 64
1629 * An Ethernet device extended statistic structure
1631 * This structure is used by rte_eth_xstats_get() to provide
1632 * statistics that are not provided in the generic *rte_eth_stats*
1634 * It maps a name id, corresponding to an index in the array returned
1635 * by rte_eth_xstats_get_names(), to a statistic value.
1637 struct rte_eth_xstat {
1638 uint64_t id; /**< The index in xstats name array. */
1639 uint64_t value; /**< The statistic counter value. */
1643 * A name element for extended statistics.
1645 * An array of this structure is returned by rte_eth_xstats_get_names().
1646 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1647 * structure references these names by their array index.
1649 * The xstats should follow a common naming scheme.
1650 * Some names are standardized in rte_stats_strings.
1652 * - rx_missed_errors
1654 * - tx_size_128_to_255_packets
1656 struct rte_eth_xstat_name {
1657 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1660 #define ETH_DCB_NUM_TCS 8
1661 #define ETH_MAX_VMDQ_POOL 64
1664 * A structure used to get the information of queue and
1665 * TC mapping on both TX and RX paths.
1667 struct rte_eth_dcb_tc_queue_mapping {
1668 /** rx queues assigned to tc per Pool */
1672 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1673 /** rx queues assigned to tc per Pool */
1677 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1681 * A structure used to get the information of DCB.
1682 * It includes TC UP mapping and queue TC mapping.
1684 struct rte_eth_dcb_info {
1685 uint8_t nb_tcs; /**< number of TCs */
1686 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1687 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1688 /** rx queues assigned to tc */
1689 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1693 * This enum indicates the possible Forward Error Correction (FEC) modes
1694 * of an ethdev port.
1696 enum rte_eth_fec_mode {
1697 RTE_ETH_FEC_NOFEC = 0, /**< FEC is off */
1698 RTE_ETH_FEC_AUTO, /**< FEC autonegotiation modes */
1699 RTE_ETH_FEC_BASER, /**< FEC using common algorithm */
1700 RTE_ETH_FEC_RS, /**< FEC using RS algorithm */
1703 /* Translate from FEC mode to FEC capa */
1704 #define RTE_ETH_FEC_MODE_TO_CAPA(x) (1U << (x))
1706 /* This macro indicates FEC capa mask */
1707 #define RTE_ETH_FEC_MODE_CAPA_MASK(x) (1U << (RTE_ETH_FEC_ ## x))
1709 /* A structure used to get capabilities per link speed */
1710 struct rte_eth_fec_capa {
1711 uint32_t speed; /**< Link speed (see ETH_SPEED_NUM_*) */
1712 uint32_t capa; /**< FEC capabilities bitmask */
1715 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1717 /* Macros to check for valid port */
1718 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1719 if (!rte_eth_dev_is_valid_port(port_id)) { \
1720 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1725 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1726 if (!rte_eth_dev_is_valid_port(port_id)) { \
1727 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1733 * l2 tunnel configuration.
1736 /**< l2 tunnel enable mask */
1737 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1738 /**< l2 tunnel insertion mask */
1739 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1740 /**< l2 tunnel stripping mask */
1741 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1742 /**< l2 tunnel forwarding mask */
1743 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1746 * Function type used for RX packet processing packet callbacks.
1748 * The callback function is called on RX with a burst of packets that have
1749 * been received on the given port and queue.
1752 * The Ethernet port on which RX is being performed.
1754 * The queue on the Ethernet port which is being used to receive the packets.
1756 * The burst of packets that have just been received.
1758 * The number of packets in the burst pointed to by "pkts".
1760 * The max number of packets that can be stored in the "pkts" array.
1762 * The arbitrary user parameter passed in by the application when the callback
1763 * was originally configured.
1765 * The number of packets returned to the user.
1767 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1768 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1772 * Function type used for TX packet processing packet callbacks.
1774 * The callback function is called on TX with a burst of packets immediately
1775 * before the packets are put onto the hardware queue for transmission.
1778 * The Ethernet port on which TX is being performed.
1780 * The queue on the Ethernet port which is being used to transmit the packets.
1782 * The burst of packets that are about to be transmitted.
1784 * The number of packets in the burst pointed to by "pkts".
1786 * The arbitrary user parameter passed in by the application when the callback
1787 * was originally configured.
1789 * The number of packets to be written to the NIC.
1791 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1792 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1795 * Possible states of an ethdev port.
1797 enum rte_eth_dev_state {
1798 /** Device is unused before being probed. */
1799 RTE_ETH_DEV_UNUSED = 0,
1800 /** Device is attached when allocated in probing. */
1801 RTE_ETH_DEV_ATTACHED,
1802 /** Device is in removed state when plug-out is detected. */
1803 RTE_ETH_DEV_REMOVED,
1806 struct rte_eth_dev_sriov {
1807 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1808 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1809 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1810 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1812 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1814 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1816 #define RTE_ETH_DEV_NO_OWNER 0
1818 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1820 struct rte_eth_dev_owner {
1821 uint64_t id; /**< The owner unique identifier. */
1822 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1825 /** PMD supports thread-safe flow operations */
1826 #define RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE 0x0001
1827 /** Device supports link state interrupt */
1828 #define RTE_ETH_DEV_INTR_LSC 0x0002
1829 /** Device is a bonded slave */
1830 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1831 /** Device supports device removal interrupt */
1832 #define RTE_ETH_DEV_INTR_RMV 0x0008
1833 /** Device is port representor */
1834 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1835 /** Device does not support MAC change after started */
1836 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1838 * Queue xstats filled automatically by ethdev layer.
1839 * PMDs filling the queue xstats themselves should not set this flag
1841 #define RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS 0x0040
1844 * Iterates over valid ethdev ports owned by a specific owner.
1847 * The id of the next possible valid owned port.
1849 * The owner identifier.
1850 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1852 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1854 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1855 const uint64_t owner_id);
1858 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1860 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1861 for (p = rte_eth_find_next_owned_by(0, o); \
1862 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1863 p = rte_eth_find_next_owned_by(p + 1, o))
1866 * Iterates over valid ethdev ports.
1869 * The id of the next possible valid port.
1871 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1873 uint16_t rte_eth_find_next(uint16_t port_id);
1876 * Macro to iterate over all enabled and ownerless ethdev ports.
1878 #define RTE_ETH_FOREACH_DEV(p) \
1879 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1883 * @b EXPERIMENTAL: this API may change without prior notice.
1885 * Iterates over ethdev ports of a specified device.
1887 * @param port_id_start
1888 * The id of the next possible valid port.
1890 * The generic device behind the ports to iterate.
1892 * Next port id of the device, possibly port_id_start,
1893 * RTE_MAX_ETHPORTS if there is none.
1897 rte_eth_find_next_of(uint16_t port_id_start,
1898 const struct rte_device *parent);
1901 * Macro to iterate over all ethdev ports of a specified device.
1904 * The id of the matching port being iterated.
1906 * The rte_device pointer matching the iterated ports.
1908 #define RTE_ETH_FOREACH_DEV_OF(port_id, parent) \
1909 for (port_id = rte_eth_find_next_of(0, parent); \
1910 port_id < RTE_MAX_ETHPORTS; \
1911 port_id = rte_eth_find_next_of(port_id + 1, parent))
1915 * @b EXPERIMENTAL: this API may change without prior notice.
1917 * Iterates over sibling ethdev ports (i.e. sharing the same rte_device).
1919 * @param port_id_start
1920 * The id of the next possible valid sibling port.
1921 * @param ref_port_id
1922 * The id of a reference port to compare rte_device with.
1924 * Next sibling port id, possibly port_id_start or ref_port_id itself,
1925 * RTE_MAX_ETHPORTS if there is none.
1929 rte_eth_find_next_sibling(uint16_t port_id_start, uint16_t ref_port_id);
1932 * Macro to iterate over all ethdev ports sharing the same rte_device
1933 * as the specified port.
1934 * Note: the specified reference port is part of the loop iterations.
1937 * The id of the matching port being iterated.
1938 * @param ref_port_id
1939 * The id of the port being compared.
1941 #define RTE_ETH_FOREACH_DEV_SIBLING(port_id, ref_port_id) \
1942 for (port_id = rte_eth_find_next_sibling(0, ref_port_id); \
1943 port_id < RTE_MAX_ETHPORTS; \
1944 port_id = rte_eth_find_next_sibling(port_id + 1, ref_port_id))
1948 * @b EXPERIMENTAL: this API may change without prior notice.
1950 * Get a new unique owner identifier.
1951 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1952 * to avoid multiple management of device by different entities.
1955 * Owner identifier pointer.
1957 * Negative errno value on error, 0 on success.
1960 int rte_eth_dev_owner_new(uint64_t *owner_id);
1964 * @b EXPERIMENTAL: this API may change without prior notice.
1966 * Set an Ethernet device owner.
1969 * The identifier of the port to own.
1971 * The owner pointer.
1973 * Negative errno value on error, 0 on success.
1976 int rte_eth_dev_owner_set(const uint16_t port_id,
1977 const struct rte_eth_dev_owner *owner);
1981 * @b EXPERIMENTAL: this API may change without prior notice.
1983 * Unset Ethernet device owner to make the device ownerless.
1986 * The identifier of port to make ownerless.
1988 * The owner identifier.
1990 * 0 on success, negative errno value on error.
1993 int rte_eth_dev_owner_unset(const uint16_t port_id,
1994 const uint64_t owner_id);
1998 * @b EXPERIMENTAL: this API may change without prior notice.
2000 * Remove owner from all Ethernet devices owned by a specific owner.
2003 * The owner identifier.
2005 * 0 on success, negative errno value on error.
2008 int rte_eth_dev_owner_delete(const uint64_t owner_id);
2012 * @b EXPERIMENTAL: this API may change without prior notice.
2014 * Get the owner of an Ethernet device.
2017 * The port identifier.
2019 * The owner structure pointer to fill.
2021 * 0 on success, negative errno value on error..
2024 int rte_eth_dev_owner_get(const uint16_t port_id,
2025 struct rte_eth_dev_owner *owner);
2028 * Get the number of ports which are usable for the application.
2030 * These devices must be iterated by using the macro
2031 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
2032 * to deal with non-contiguous ranges of devices.
2035 * The count of available Ethernet devices.
2037 uint16_t rte_eth_dev_count_avail(void);
2040 * Get the total number of ports which are allocated.
2042 * Some devices may not be available for the application.
2045 * The total count of Ethernet devices.
2047 uint16_t rte_eth_dev_count_total(void);
2050 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
2051 * the bitmap link_speeds of the struct rte_eth_conf
2054 * Numerical speed value in Mbps
2056 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
2058 * 0 if the speed cannot be mapped
2060 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
2063 * Get DEV_RX_OFFLOAD_* flag name.
2068 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2070 const char *rte_eth_dev_rx_offload_name(uint64_t offload);
2073 * Get DEV_TX_OFFLOAD_* flag name.
2078 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2080 const char *rte_eth_dev_tx_offload_name(uint64_t offload);
2083 * Configure an Ethernet device.
2084 * This function must be invoked first before any other function in the
2085 * Ethernet API. This function can also be re-invoked when a device is in the
2089 * The port identifier of the Ethernet device to configure.
2090 * @param nb_rx_queue
2091 * The number of receive queues to set up for the Ethernet device.
2092 * @param nb_tx_queue
2093 * The number of transmit queues to set up for the Ethernet device.
2095 * The pointer to the configuration data to be used for the Ethernet device.
2096 * The *rte_eth_conf* structure includes:
2097 * - the hardware offload features to activate, with dedicated fields for
2098 * each statically configurable offload hardware feature provided by
2099 * Ethernet devices, such as IP checksum or VLAN tag stripping for
2101 * The Rx offload bitfield API is obsolete and will be deprecated.
2102 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
2103 * structure and use offloads field to set per-port offloads instead.
2104 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
2105 * the [rt]x_offload_capa returned from rte_eth_dev_info_get().
2106 * Any type of device supported offloading set in the input argument
2107 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
2108 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
2109 * - the Receive Side Scaling (RSS) configuration when using multiple RX
2110 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
2111 * must be within the flow_type_rss_offloads provided by drivers via
2112 * rte_eth_dev_info_get() API.
2114 * Embedding all configuration information in a single data structure
2115 * is the more flexible method that allows the addition of new features
2116 * without changing the syntax of the API.
2118 * - 0: Success, device configured.
2119 * - <0: Error code returned by the driver configuration function.
2121 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
2122 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
2126 * @b EXPERIMENTAL: this API may change without prior notice.
2128 * Check if an Ethernet device was physically removed.
2131 * The port identifier of the Ethernet device.
2133 * 1 when the Ethernet device is removed, otherwise 0.
2137 rte_eth_dev_is_removed(uint16_t port_id);
2140 * Allocate and set up a receive queue for an Ethernet device.
2142 * The function allocates a contiguous block of memory for *nb_rx_desc*
2143 * receive descriptors from a memory zone associated with *socket_id*
2144 * and initializes each receive descriptor with a network buffer allocated
2145 * from the memory pool *mb_pool*.
2148 * The port identifier of the Ethernet device.
2149 * @param rx_queue_id
2150 * The index of the receive queue to set up.
2151 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2152 * to rte_eth_dev_configure().
2154 * The number of receive descriptors to allocate for the receive ring.
2156 * The *socket_id* argument is the socket identifier in case of NUMA.
2157 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2158 * the DMA memory allocated for the receive descriptors of the ring.
2160 * The pointer to the configuration data to be used for the receive queue.
2161 * NULL value is allowed, in which case default RX configuration
2163 * The *rx_conf* structure contains an *rx_thresh* structure with the values
2164 * of the Prefetch, Host, and Write-Back threshold registers of the receive
2166 * In addition it contains the hardware offloads features to activate using
2167 * the DEV_RX_OFFLOAD_* flags.
2168 * If an offloading set in rx_conf->offloads
2169 * hasn't been set in the input argument eth_conf->rxmode.offloads
2170 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2171 * per-queue type and it is enabled for the queue.
2172 * No need to repeat any bit in rx_conf->offloads which has already been
2173 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2174 * at port level can't be disabled at queue level.
2175 * The configuration structure also contains the pointer to the array
2176 * of the receiving buffer segment descriptions, see rx_seg and rx_nseg
2177 * fields, this extended configuration might be used by split offloads like
2178 * RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT. If mp_pool is not NULL,
2179 * the extended configuration fields must be set to NULL and zero.
2181 * The pointer to the memory pool from which to allocate *rte_mbuf* network
2182 * memory buffers to populate each descriptor of the receive ring. There are
2183 * two options to provide Rx buffer configuration:
2185 * mb_pool is not NULL, rx_conf.rx_nseg is 0.
2186 * - multiple segments description:
2187 * mb_pool is NULL, rx_conf.rx_seg is not NULL, rx_conf.rx_nseg is not 0.
2188 * Taken only if flag RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT is set in offloads.
2191 * - 0: Success, receive queue correctly set up.
2192 * - -EIO: if device is removed.
2193 * - -ENODEV: if *port_id* is invalid.
2194 * - -EINVAL: The memory pool pointer is null or the size of network buffers
2195 * which can be allocated from this memory pool does not fit the various
2196 * buffer sizes allowed by the device controller.
2197 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
2198 * allocate network memory buffers from the memory pool when
2199 * initializing receive descriptors.
2201 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
2202 uint16_t nb_rx_desc, unsigned int socket_id,
2203 const struct rte_eth_rxconf *rx_conf,
2204 struct rte_mempool *mb_pool);
2208 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2210 * Allocate and set up a hairpin receive queue for an Ethernet device.
2212 * The function set up the selected queue to be used in hairpin.
2215 * The port identifier of the Ethernet device.
2216 * @param rx_queue_id
2217 * The index of the receive queue to set up.
2218 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2219 * to rte_eth_dev_configure().
2221 * The number of receive descriptors to allocate for the receive ring.
2222 * 0 means the PMD will use default value.
2224 * The pointer to the hairpin configuration.
2227 * - (0) if successful.
2228 * - (-ENODEV) if *port_id* is invalid.
2229 * - (-ENOTSUP) if hardware doesn't support.
2230 * - (-EINVAL) if bad parameter.
2231 * - (-ENOMEM) if unable to allocate the resources.
2234 int rte_eth_rx_hairpin_queue_setup
2235 (uint16_t port_id, uint16_t rx_queue_id, uint16_t nb_rx_desc,
2236 const struct rte_eth_hairpin_conf *conf);
2239 * Allocate and set up a transmit queue for an Ethernet device.
2242 * The port identifier of the Ethernet device.
2243 * @param tx_queue_id
2244 * The index of the transmit queue to set up.
2245 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2246 * to rte_eth_dev_configure().
2248 * The number of transmit descriptors to allocate for the transmit ring.
2250 * The *socket_id* argument is the socket identifier in case of NUMA.
2251 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2252 * the DMA memory allocated for the transmit descriptors of the ring.
2254 * The pointer to the configuration data to be used for the transmit queue.
2255 * NULL value is allowed, in which case default TX configuration
2257 * The *tx_conf* structure contains the following data:
2258 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
2259 * Write-Back threshold registers of the transmit ring.
2260 * When setting Write-Back threshold to the value greater then zero,
2261 * *tx_rs_thresh* value should be explicitly set to one.
2262 * - The *tx_free_thresh* value indicates the [minimum] number of network
2263 * buffers that must be pending in the transmit ring to trigger their
2264 * [implicit] freeing by the driver transmit function.
2265 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
2266 * descriptors that must be pending in the transmit ring before setting the
2267 * RS bit on a descriptor by the driver transmit function.
2268 * The *tx_rs_thresh* value should be less or equal then
2269 * *tx_free_thresh* value, and both of them should be less then
2271 * - The *offloads* member contains Tx offloads to be enabled.
2272 * If an offloading set in tx_conf->offloads
2273 * hasn't been set in the input argument eth_conf->txmode.offloads
2274 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2275 * per-queue type and it is enabled for the queue.
2276 * No need to repeat any bit in tx_conf->offloads which has already been
2277 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2278 * at port level can't be disabled at queue level.
2280 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
2281 * the transmit function to use default values.
2283 * - 0: Success, the transmit queue is correctly set up.
2284 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
2286 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
2287 uint16_t nb_tx_desc, unsigned int socket_id,
2288 const struct rte_eth_txconf *tx_conf);
2292 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2294 * Allocate and set up a transmit hairpin queue for an Ethernet device.
2297 * The port identifier of the Ethernet device.
2298 * @param tx_queue_id
2299 * The index of the transmit queue to set up.
2300 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2301 * to rte_eth_dev_configure().
2303 * The number of transmit descriptors to allocate for the transmit ring.
2304 * 0 to set default PMD value.
2306 * The hairpin configuration.
2309 * - (0) if successful.
2310 * - (-ENODEV) if *port_id* is invalid.
2311 * - (-ENOTSUP) if hardware doesn't support.
2312 * - (-EINVAL) if bad parameter.
2313 * - (-ENOMEM) if unable to allocate the resources.
2316 int rte_eth_tx_hairpin_queue_setup
2317 (uint16_t port_id, uint16_t tx_queue_id, uint16_t nb_tx_desc,
2318 const struct rte_eth_hairpin_conf *conf);
2322 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2324 * Get all the hairpin peer Rx / Tx ports of the current port.
2325 * The caller should ensure that the array is large enough to save the ports
2329 * The port identifier of the Ethernet device.
2331 * Pointer to the array to store the peer ports list.
2333 * Length of the array to store the port identifiers.
2335 * Current port to peer port direction
2336 * positive - current used as Tx to get all peer Rx ports.
2337 * zero - current used as Rx to get all peer Tx ports.
2340 * - (0 or positive) actual peer ports number.
2341 * - (-EINVAL) if bad parameter.
2342 * - (-ENODEV) if *port_id* invalid
2343 * - (-ENOTSUP) if hardware doesn't support.
2344 * - Others detailed errors from PMD drivers.
2347 int rte_eth_hairpin_get_peer_ports(uint16_t port_id, uint16_t *peer_ports,
2348 size_t len, uint32_t direction);
2352 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2354 * Bind all hairpin Tx queues of one port to the Rx queues of the peer port.
2355 * It is only allowed to call this function after all hairpin queues are
2356 * configured properly and the devices are in started state.
2359 * The identifier of the Tx port.
2361 * The identifier of peer Rx port.
2362 * RTE_MAX_ETHPORTS is allowed for the traversal of all devices.
2363 * Rx port ID could have the same value as Tx port ID.
2366 * - (0) if successful.
2367 * - (-ENODEV) if Tx port ID is invalid.
2368 * - (-EBUSY) if device is not in started state.
2369 * - (-ENOTSUP) if hardware doesn't support.
2370 * - Others detailed errors from PMD drivers.
2373 int rte_eth_hairpin_bind(uint16_t tx_port, uint16_t rx_port);
2377 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2379 * Unbind all hairpin Tx queues of one port from the Rx queues of the peer port.
2380 * This should be called before closing the Tx or Rx devices, if the bind
2381 * function is called before.
2382 * After unbinding the hairpin ports pair, it is allowed to bind them again.
2383 * Changing queues configuration should be after stopping the device(s).
2386 * The identifier of the Tx port.
2388 * The identifier of peer Rx port.
2389 * RTE_MAX_ETHPORTS is allowed for traversal of all devices.
2390 * Rx port ID could have the same value as Tx port ID.
2393 * - (0) if successful.
2394 * - (-ENODEV) if Tx port ID is invalid.
2395 * - (-EBUSY) if device is in stopped state.
2396 * - (-ENOTSUP) if hardware doesn't support.
2397 * - Others detailed errors from PMD drivers.
2400 int rte_eth_hairpin_unbind(uint16_t tx_port, uint16_t rx_port);
2403 * Return the NUMA socket to which an Ethernet device is connected
2406 * The port identifier of the Ethernet device
2408 * The NUMA socket id to which the Ethernet device is connected or
2409 * a default of zero if the socket could not be determined.
2410 * -1 is returned is the port_id value is out of range.
2412 int rte_eth_dev_socket_id(uint16_t port_id);
2415 * Check if port_id of device is attached
2418 * The port identifier of the Ethernet device
2420 * - 0 if port is out of range or not attached
2421 * - 1 if device is attached
2423 int rte_eth_dev_is_valid_port(uint16_t port_id);
2426 * Start specified RX queue of a port. It is used when rx_deferred_start
2427 * flag of the specified queue is true.
2430 * The port identifier of the Ethernet device
2431 * @param rx_queue_id
2432 * The index of the rx queue to update the ring.
2433 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2434 * to rte_eth_dev_configure().
2436 * - 0: Success, the receive queue is started.
2437 * - -ENODEV: if *port_id* is invalid.
2438 * - -EINVAL: The queue_id out of range or belong to hairpin.
2439 * - -EIO: if device is removed.
2440 * - -ENOTSUP: The function not supported in PMD driver.
2442 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
2445 * Stop specified RX queue of a port
2448 * The port identifier of the Ethernet device
2449 * @param rx_queue_id
2450 * The index of the rx queue to update the ring.
2451 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2452 * to rte_eth_dev_configure().
2454 * - 0: Success, the receive queue is stopped.
2455 * - -ENODEV: if *port_id* is invalid.
2456 * - -EINVAL: The queue_id out of range or belong to hairpin.
2457 * - -EIO: if device is removed.
2458 * - -ENOTSUP: The function not supported in PMD driver.
2460 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
2463 * Start TX for specified queue of a port. It is used when tx_deferred_start
2464 * flag of the specified queue is true.
2467 * The port identifier of the Ethernet device
2468 * @param tx_queue_id
2469 * The index of the tx queue to update the ring.
2470 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2471 * to rte_eth_dev_configure().
2473 * - 0: Success, the transmit queue is started.
2474 * - -ENODEV: if *port_id* is invalid.
2475 * - -EINVAL: The queue_id out of range or belong to hairpin.
2476 * - -EIO: if device is removed.
2477 * - -ENOTSUP: The function not supported in PMD driver.
2479 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
2482 * Stop specified TX queue of a port
2485 * The port identifier of the Ethernet device
2486 * @param tx_queue_id
2487 * The index of the tx queue to update the ring.
2488 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2489 * to rte_eth_dev_configure().
2491 * - 0: Success, the transmit queue is stopped.
2492 * - -ENODEV: if *port_id* is invalid.
2493 * - -EINVAL: The queue_id out of range or belong to hairpin.
2494 * - -EIO: if device is removed.
2495 * - -ENOTSUP: The function not supported in PMD driver.
2497 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
2500 * Start an Ethernet device.
2502 * The device start step is the last one and consists of setting the configured
2503 * offload features and in starting the transmit and the receive units of the
2506 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
2507 * PMD port start callback function is invoked.
2509 * On success, all basic functions exported by the Ethernet API (link status,
2510 * receive/transmit, and so on) can be invoked.
2513 * The port identifier of the Ethernet device.
2515 * - 0: Success, Ethernet device started.
2516 * - <0: Error code of the driver device start function.
2518 int rte_eth_dev_start(uint16_t port_id);
2521 * Stop an Ethernet device. The device can be restarted with a call to
2522 * rte_eth_dev_start()
2525 * The port identifier of the Ethernet device.
2527 * - 0: Success, Ethernet device stopped.
2528 * - <0: Error code of the driver device stop function.
2530 int rte_eth_dev_stop(uint16_t port_id);
2533 * Link up an Ethernet device.
2535 * Set device link up will re-enable the device rx/tx
2536 * functionality after it is previously set device linked down.
2539 * The port identifier of the Ethernet device.
2541 * - 0: Success, Ethernet device linked up.
2542 * - <0: Error code of the driver device link up function.
2544 int rte_eth_dev_set_link_up(uint16_t port_id);
2547 * Link down an Ethernet device.
2548 * The device rx/tx functionality will be disabled if success,
2549 * and it can be re-enabled with a call to
2550 * rte_eth_dev_set_link_up()
2553 * The port identifier of the Ethernet device.
2555 int rte_eth_dev_set_link_down(uint16_t port_id);
2558 * Close a stopped Ethernet device. The device cannot be restarted!
2559 * The function frees all port resources.
2562 * The port identifier of the Ethernet device.
2564 * - Zero if the port is closed successfully.
2565 * - Negative if something went wrong.
2567 int rte_eth_dev_close(uint16_t port_id);
2570 * Reset a Ethernet device and keep its port id.
2572 * When a port has to be reset passively, the DPDK application can invoke
2573 * this function. For example when a PF is reset, all its VFs should also
2574 * be reset. Normally a DPDK application can invoke this function when
2575 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
2576 * a port reset in other circumstances.
2578 * When this function is called, it first stops the port and then calls the
2579 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
2580 * state, in which no Tx and Rx queues are setup, as if the port has been
2581 * reset and not started. The port keeps the port id it had before the
2584 * After calling rte_eth_dev_reset( ), the application should use
2585 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
2586 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
2587 * to reconfigure the device as appropriate.
2589 * Note: To avoid unexpected behavior, the application should stop calling
2590 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
2591 * safety, all these controlling functions should be called from the same
2595 * The port identifier of the Ethernet device.
2598 * - (0) if successful.
2599 * - (-ENODEV) if *port_id* is invalid.
2600 * - (-ENOTSUP) if hardware doesn't support this function.
2601 * - (-EPERM) if not ran from the primary process.
2602 * - (-EIO) if re-initialisation failed or device is removed.
2603 * - (-ENOMEM) if the reset failed due to OOM.
2604 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
2606 int rte_eth_dev_reset(uint16_t port_id);
2609 * Enable receipt in promiscuous mode for an Ethernet device.
2612 * The port identifier of the Ethernet device.
2614 * - (0) if successful.
2615 * - (-ENOTSUP) if support for promiscuous_enable() does not exist
2617 * - (-ENODEV) if *port_id* invalid.
2619 int rte_eth_promiscuous_enable(uint16_t port_id);
2622 * Disable receipt in promiscuous mode for an Ethernet device.
2625 * The port identifier of the Ethernet device.
2627 * - (0) if successful.
2628 * - (-ENOTSUP) if support for promiscuous_disable() does not exist
2630 * - (-ENODEV) if *port_id* invalid.
2632 int rte_eth_promiscuous_disable(uint16_t port_id);
2635 * Return the value of promiscuous mode for an Ethernet device.
2638 * The port identifier of the Ethernet device.
2640 * - (1) if promiscuous is enabled
2641 * - (0) if promiscuous is disabled.
2644 int rte_eth_promiscuous_get(uint16_t port_id);
2647 * Enable the receipt of any multicast frame by an Ethernet device.
2650 * The port identifier of the Ethernet device.
2652 * - (0) if successful.
2653 * - (-ENOTSUP) if support for allmulticast_enable() does not exist
2655 * - (-ENODEV) if *port_id* invalid.
2657 int rte_eth_allmulticast_enable(uint16_t port_id);
2660 * Disable the receipt of all multicast frames by an Ethernet device.
2663 * The port identifier of the Ethernet device.
2665 * - (0) if successful.
2666 * - (-ENOTSUP) if support for allmulticast_disable() does not exist
2668 * - (-ENODEV) if *port_id* invalid.
2670 int rte_eth_allmulticast_disable(uint16_t port_id);
2673 * Return the value of allmulticast mode for an Ethernet device.
2676 * The port identifier of the Ethernet device.
2678 * - (1) if allmulticast is enabled
2679 * - (0) if allmulticast is disabled.
2682 int rte_eth_allmulticast_get(uint16_t port_id);
2685 * Retrieve the link status (up/down), the duplex mode (half/full),
2686 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2688 * It might need to wait up to 9 seconds.
2689 * @see rte_eth_link_get_nowait.
2692 * The port identifier of the Ethernet device.
2694 * Link information written back.
2696 * - (0) if successful.
2697 * - (-ENOTSUP) if the function is not supported in PMD driver.
2698 * - (-ENODEV) if *port_id* invalid.
2700 int rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
2703 * Retrieve the link status (up/down), the duplex mode (half/full),
2704 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2707 * The port identifier of the Ethernet device.
2709 * Link information written back.
2711 * - (0) if successful.
2712 * - (-ENOTSUP) if the function is not supported in PMD driver.
2713 * - (-ENODEV) if *port_id* invalid.
2715 int rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
2719 * @b EXPERIMENTAL: this API may change without prior notice.
2721 * The function converts a link_speed to a string. It handles all special
2722 * values like unknown or none speed.
2725 * link_speed of rte_eth_link struct
2727 * Link speed in textual format. It's pointer to immutable memory.
2728 * No free is required.
2731 const char *rte_eth_link_speed_to_str(uint32_t link_speed);
2735 * @b EXPERIMENTAL: this API may change without prior notice.
2737 * The function converts a rte_eth_link struct representing a link status to
2741 * A pointer to a string to be filled with textual representation of
2742 * device status. At least ETH_LINK_MAX_STR_LEN bytes should be allocated to
2743 * store default link status text.
2745 * Length of available memory at 'str' string.
2747 * Link status returned by rte_eth_link_get function
2749 * Number of bytes written to str array.
2752 int rte_eth_link_to_str(char *str, size_t len,
2753 const struct rte_eth_link *eth_link);
2756 * Retrieve the general I/O statistics of an Ethernet device.
2759 * The port identifier of the Ethernet device.
2761 * A pointer to a structure of type *rte_eth_stats* to be filled with
2762 * the values of device counters for the following set of statistics:
2763 * - *ipackets* with the total of successfully received packets.
2764 * - *opackets* with the total of successfully transmitted packets.
2765 * - *ibytes* with the total of successfully received bytes.
2766 * - *obytes* with the total of successfully transmitted bytes.
2767 * - *ierrors* with the total of erroneous received packets.
2768 * - *oerrors* with the total of failed transmitted packets.
2770 * Zero if successful. Non-zero otherwise.
2772 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
2775 * Reset the general I/O statistics of an Ethernet device.
2778 * The port identifier of the Ethernet device.
2780 * - (0) if device notified to reset stats.
2781 * - (-ENOTSUP) if hardware doesn't support.
2782 * - (-ENODEV) if *port_id* invalid.
2783 * - (<0): Error code of the driver stats reset function.
2785 int rte_eth_stats_reset(uint16_t port_id);
2788 * Retrieve names of extended statistics of an Ethernet device.
2790 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2792 * xstats_names[i].name => xstats[i].value
2794 * And the array index is same with id field of 'struct rte_eth_xstat':
2797 * This assumption makes key-value pair matching less flexible but simpler.
2800 * The port identifier of the Ethernet device.
2801 * @param xstats_names
2802 * An rte_eth_xstat_name array of at least *size* elements to
2803 * be filled. If set to NULL, the function returns the required number
2806 * The size of the xstats_names array (number of elements).
2808 * - A positive value lower or equal to size: success. The return value
2809 * is the number of entries filled in the stats table.
2810 * - A positive value higher than size: error, the given statistics table
2811 * is too small. The return value corresponds to the size that should
2812 * be given to succeed. The entries in the table are not valid and
2813 * shall not be used by the caller.
2814 * - A negative value on error (invalid port id).
2816 int rte_eth_xstats_get_names(uint16_t port_id,
2817 struct rte_eth_xstat_name *xstats_names,
2821 * Retrieve extended statistics of an Ethernet device.
2823 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2825 * xstats_names[i].name => xstats[i].value
2827 * And the array index is same with id field of 'struct rte_eth_xstat':
2830 * This assumption makes key-value pair matching less flexible but simpler.
2833 * The port identifier of the Ethernet device.
2835 * A pointer to a table of structure of type *rte_eth_xstat*
2836 * to be filled with device statistics ids and values.
2837 * This parameter can be set to NULL if n is 0.
2839 * The size of the xstats array (number of elements).
2841 * - A positive value lower or equal to n: success. The return value
2842 * is the number of entries filled in the stats table.
2843 * - A positive value higher than n: error, the given statistics table
2844 * is too small. The return value corresponds to the size that should
2845 * be given to succeed. The entries in the table are not valid and
2846 * shall not be used by the caller.
2847 * - A negative value on error (invalid port id).
2849 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2853 * Retrieve names of extended statistics of an Ethernet device.
2856 * The port identifier of the Ethernet device.
2857 * @param xstats_names
2858 * An rte_eth_xstat_name array of at least *size* elements to
2859 * be filled. If set to NULL, the function returns the required number
2862 * IDs array given by app to retrieve specific statistics
2864 * The size of the xstats_names array (number of elements).
2866 * - A positive value lower or equal to size: success. The return value
2867 * is the number of entries filled in the stats table.
2868 * - A positive value higher than size: error, the given statistics table
2869 * is too small. The return value corresponds to the size that should
2870 * be given to succeed. The entries in the table are not valid and
2871 * shall not be used by the caller.
2872 * - A negative value on error (invalid port id).
2875 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2876 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2880 * Retrieve extended statistics of an Ethernet device.
2883 * The port identifier of the Ethernet device.
2885 * A pointer to an ids array passed by application. This tells which
2886 * statistics values function should retrieve. This parameter
2887 * can be set to NULL if size is 0. In this case function will retrieve
2888 * all available statistics.
2890 * A pointer to a table to be filled with device statistics values.
2892 * The size of the ids array (number of elements).
2894 * - A positive value lower or equal to size: success. The return value
2895 * is the number of entries filled in the stats table.
2896 * - A positive value higher than size: error, the given statistics table
2897 * is too small. The return value corresponds to the size that should
2898 * be given to succeed. The entries in the table are not valid and
2899 * shall not be used by the caller.
2900 * - A negative value on error (invalid port id).
2902 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2903 uint64_t *values, unsigned int size);
2906 * Gets the ID of a statistic from its name.
2908 * This function searches for the statistics using string compares, and
2909 * as such should not be used on the fast-path. For fast-path retrieval of
2910 * specific statistics, store the ID as provided in *id* from this function,
2911 * and pass the ID to rte_eth_xstats_get()
2913 * @param port_id The port to look up statistics from
2914 * @param xstat_name The name of the statistic to return
2915 * @param[out] id A pointer to an app-supplied uint64_t which should be
2916 * set to the ID of the stat if the stat exists.
2919 * -ENODEV for invalid port_id,
2920 * -EIO if device is removed,
2921 * -EINVAL if the xstat_name doesn't exist in port_id
2923 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2927 * Reset extended statistics of an Ethernet device.
2930 * The port identifier of the Ethernet device.
2932 * - (0) if device notified to reset extended stats.
2933 * - (-ENOTSUP) if pmd doesn't support both
2934 * extended stats and basic stats reset.
2935 * - (-ENODEV) if *port_id* invalid.
2936 * - (<0): Error code of the driver xstats reset function.
2938 int rte_eth_xstats_reset(uint16_t port_id);
2941 * Set a mapping for the specified transmit queue to the specified per-queue
2942 * statistics counter.
2945 * The port identifier of the Ethernet device.
2946 * @param tx_queue_id
2947 * The index of the transmit queue for which a queue stats mapping is required.
2948 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2949 * to rte_eth_dev_configure().
2951 * The per-queue packet statistics functionality number that the transmit
2952 * queue is to be assigned.
2953 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2954 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2956 * Zero if successful. Non-zero otherwise.
2958 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2959 uint16_t tx_queue_id, uint8_t stat_idx);
2962 * Set a mapping for the specified receive queue to the specified per-queue
2963 * statistics counter.
2966 * The port identifier of the Ethernet device.
2967 * @param rx_queue_id
2968 * The index of the receive queue for which a queue stats mapping is required.
2969 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2970 * to rte_eth_dev_configure().
2972 * The per-queue packet statistics functionality number that the receive
2973 * queue is to be assigned.
2974 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2975 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2977 * Zero if successful. Non-zero otherwise.
2979 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2980 uint16_t rx_queue_id,
2984 * Retrieve the Ethernet address of an Ethernet device.
2987 * The port identifier of the Ethernet device.
2989 * A pointer to a structure of type *ether_addr* to be filled with
2990 * the Ethernet address of the Ethernet device.
2992 * - (0) if successful
2993 * - (-ENODEV) if *port_id* invalid.
2995 int rte_eth_macaddr_get(uint16_t port_id, struct rte_ether_addr *mac_addr);
2998 * Retrieve the contextual information of an Ethernet device.
3000 * As part of this function, a number of of fields in dev_info will be
3001 * initialized as follows:
3006 * Where lim is defined within the rte_eth_dev_info_get as
3008 * const struct rte_eth_desc_lim lim = {
3009 * .nb_max = UINT16_MAX,
3012 * .nb_seg_max = UINT16_MAX,
3013 * .nb_mtu_seg_max = UINT16_MAX,
3016 * device = dev->device
3017 * min_mtu = RTE_ETHER_MIN_MTU
3018 * max_mtu = UINT16_MAX
3020 * The following fields will be populated if support for dev_infos_get()
3021 * exists for the device and the rte_eth_dev 'dev' has been populated
3022 * successfully with a call to it:
3024 * driver_name = dev->device->driver->name
3025 * nb_rx_queues = dev->data->nb_rx_queues
3026 * nb_tx_queues = dev->data->nb_tx_queues
3027 * dev_flags = &dev->data->dev_flags
3030 * The port identifier of the Ethernet device.
3032 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
3033 * the contextual information of the Ethernet device.
3035 * - (0) if successful.
3036 * - (-ENOTSUP) if support for dev_infos_get() does not exist for the device.
3037 * - (-ENODEV) if *port_id* invalid.
3039 int rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
3042 * Retrieve the firmware version of a device.
3045 * The port identifier of the device.
3047 * A pointer to a string array storing the firmware version of a device,
3048 * the string includes terminating null. This pointer is allocated by caller.
3050 * The size of the string array pointed by fw_version, which should be
3051 * large enough to store firmware version of the device.
3053 * - (0) if successful.
3054 * - (-ENOTSUP) if operation is not supported.
3055 * - (-ENODEV) if *port_id* invalid.
3056 * - (-EIO) if device is removed.
3057 * - (>0) if *fw_size* is not enough to store firmware version, return
3058 * the size of the non truncated string.
3060 int rte_eth_dev_fw_version_get(uint16_t port_id,
3061 char *fw_version, size_t fw_size);
3064 * Retrieve the supported packet types of an Ethernet device.
3066 * When a packet type is announced as supported, it *must* be recognized by
3067 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
3068 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
3069 * packet types for these packets:
3070 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
3071 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
3072 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
3073 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
3075 * When a packet is received by a PMD, the most precise type must be
3076 * returned among the ones supported. However a PMD is allowed to set
3077 * packet type that is not in the supported list, at the condition that it
3078 * is more precise. Therefore, a PMD announcing no supported packet types
3079 * can still set a matching packet type in a received packet.
3082 * Better to invoke this API after the device is already started or rx burst
3083 * function is decided, to obtain correct supported ptypes.
3085 * if a given PMD does not report what ptypes it supports, then the supported
3086 * ptype count is reported as 0.
3088 * The port identifier of the Ethernet device.
3090 * A hint of what kind of packet type which the caller is interested in.
3092 * An array pointer to store adequate packet types, allocated by caller.
3094 * Size of the array pointed by param ptypes.
3096 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
3097 * only num entries will be filled into the ptypes array, but the full
3098 * count of supported ptypes will be returned.
3099 * - (-ENODEV) if *port_id* invalid.
3101 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
3102 uint32_t *ptypes, int num);
3105 * @b EXPERIMENTAL: this API may change without prior notice.
3107 * Inform Ethernet device about reduced range of packet types to handle.
3109 * Application can use this function to set only specific ptypes that it's
3110 * interested. This information can be used by the PMD to optimize Rx path.
3112 * The function accepts an array `set_ptypes` allocated by the caller to
3113 * store the packet types set by the driver, the last element of the array
3114 * is set to RTE_PTYPE_UNKNOWN. The size of the `set_ptype` array should be
3115 * `rte_eth_dev_get_supported_ptypes() + 1` else it might only be filled
3119 * The port identifier of the Ethernet device.
3121 * The ptype family that application is interested in should be bitwise OR of
3122 * RTE_PTYPE_*_MASK or 0.
3124 * An array pointer to store set packet types, allocated by caller. The
3125 * function marks the end of array with RTE_PTYPE_UNKNOWN.
3127 * Size of the array pointed by param ptypes.
3128 * Should be rte_eth_dev_get_supported_ptypes() + 1 to accommodate the
3132 * - (-ENODEV) if *port_id* invalid.
3133 * - (-EINVAL) if *ptype_mask* is invalid (or) set_ptypes is NULL and
3137 int rte_eth_dev_set_ptypes(uint16_t port_id, uint32_t ptype_mask,
3138 uint32_t *set_ptypes, unsigned int num);
3141 * Retrieve the MTU of an Ethernet device.
3144 * The port identifier of the Ethernet device.
3146 * A pointer to a uint16_t where the retrieved MTU is to be stored.
3148 * - (0) if successful.
3149 * - (-ENODEV) if *port_id* invalid.
3151 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
3154 * Change the MTU of an Ethernet device.
3157 * The port identifier of the Ethernet device.
3159 * A uint16_t for the MTU to be applied.
3161 * - (0) if successful.
3162 * - (-ENOTSUP) if operation is not supported.
3163 * - (-ENODEV) if *port_id* invalid.
3164 * - (-EIO) if device is removed.
3165 * - (-EINVAL) if *mtu* invalid, validation of mtu can occur within
3166 * rte_eth_dev_set_mtu if dev_infos_get is supported by the device or
3167 * when the mtu is set using dev->dev_ops->mtu_set.
3168 * - (-EBUSY) if operation is not allowed when the port is running
3170 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
3173 * Enable/Disable hardware filtering by an Ethernet device of received
3174 * VLAN packets tagged with a given VLAN Tag Identifier.
3177 * The port identifier of the Ethernet device.
3179 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
3181 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
3182 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
3184 * - (0) if successful.
3185 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3186 * - (-ENODEV) if *port_id* invalid.
3187 * - (-EIO) if device is removed.
3188 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
3189 * - (-EINVAL) if *vlan_id* > 4095.
3191 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
3194 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
3197 * The port identifier of the Ethernet device.
3198 * @param rx_queue_id
3199 * The index of the receive queue for which a queue stats mapping is required.
3200 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3201 * to rte_eth_dev_configure().
3203 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
3204 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
3206 * - (0) if successful.
3207 * - (-ENOTSUP) if hardware-assisted VLAN stripping not configured.
3208 * - (-ENODEV) if *port_id* invalid.
3209 * - (-EINVAL) if *rx_queue_id* invalid.
3211 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
3215 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
3219 * The port identifier of the Ethernet device.
3223 * The Tag Protocol ID
3225 * - (0) if successful.
3226 * - (-ENOTSUP) if hardware-assisted VLAN TPID setup is not supported.
3227 * - (-ENODEV) if *port_id* invalid.
3228 * - (-EIO) if device is removed.
3230 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
3231 enum rte_vlan_type vlan_type,
3235 * Set VLAN offload configuration on an Ethernet device.
3238 * The port identifier of the Ethernet device.
3239 * @param offload_mask
3240 * The VLAN Offload bit mask can be mixed use with "OR"
3241 * ETH_VLAN_STRIP_OFFLOAD
3242 * ETH_VLAN_FILTER_OFFLOAD
3243 * ETH_VLAN_EXTEND_OFFLOAD
3244 * ETH_QINQ_STRIP_OFFLOAD
3246 * - (0) if successful.
3247 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3248 * - (-ENODEV) if *port_id* invalid.
3249 * - (-EIO) if device is removed.
3251 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
3254 * Read VLAN Offload configuration from an Ethernet device
3257 * The port identifier of the Ethernet device.
3259 * - (>0) if successful. Bit mask to indicate
3260 * ETH_VLAN_STRIP_OFFLOAD
3261 * ETH_VLAN_FILTER_OFFLOAD
3262 * ETH_VLAN_EXTEND_OFFLOAD
3263 * ETH_QINQ_STRIP_OFFLOAD
3264 * - (-ENODEV) if *port_id* invalid.
3266 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
3269 * Set port based TX VLAN insertion on or off.
3272 * The port identifier of the Ethernet device.
3274 * Port based TX VLAN identifier together with user priority.
3276 * Turn on or off the port based TX VLAN insertion.
3279 * - (0) if successful.
3280 * - negative if failed.
3282 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
3284 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
3288 * Structure used to buffer packets for future TX
3289 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
3291 struct rte_eth_dev_tx_buffer {
3292 buffer_tx_error_fn error_callback;
3293 void *error_userdata;
3294 uint16_t size; /**< Size of buffer for buffered tx */
3295 uint16_t length; /**< Number of packets in the array */
3296 struct rte_mbuf *pkts[];
3297 /**< Pending packets to be sent on explicit flush or when full */
3301 * Calculate the size of the tx buffer.
3304 * Number of stored packets.
3306 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
3307 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
3310 * Initialize default values for buffered transmitting
3313 * Tx buffer to be initialized.
3320 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
3323 * Configure a callback for buffered packets which cannot be sent
3325 * Register a specific callback to be called when an attempt is made to send
3326 * all packets buffered on an ethernet port, but not all packets can
3327 * successfully be sent. The callback registered here will be called only
3328 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
3329 * The default callback configured for each queue by default just frees the
3330 * packets back to the calling mempool. If additional behaviour is required,
3331 * for example, to count dropped packets, or to retry transmission of packets
3332 * which cannot be sent, this function should be used to register a suitable
3333 * callback function to implement the desired behaviour.
3334 * The example callback "rte_eth_count_unsent_packet_callback()" is also
3335 * provided as reference.
3338 * The port identifier of the Ethernet device.
3340 * The function to be used as the callback.
3342 * Arbitrary parameter to be passed to the callback function
3344 * 0 on success, or -1 on error with rte_errno set appropriately
3347 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
3348 buffer_tx_error_fn callback, void *userdata);
3351 * Callback function for silently dropping unsent buffered packets.
3353 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3354 * adjust the default behavior when buffered packets cannot be sent. This
3355 * function drops any unsent packets silently and is used by tx buffered
3356 * operations as default behavior.
3358 * NOTE: this function should not be called directly, instead it should be used
3359 * as a callback for packet buffering.
3361 * NOTE: when configuring this function as a callback with
3362 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3363 * should point to an uint64_t value.
3366 * The previously buffered packets which could not be sent
3368 * The number of unsent packets in the pkts array
3373 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
3377 * Callback function for tracking unsent buffered packets.
3379 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3380 * adjust the default behavior when buffered packets cannot be sent. This
3381 * function drops any unsent packets, but also updates a user-supplied counter
3382 * to track the overall number of packets dropped. The counter should be an
3383 * uint64_t variable.
3385 * NOTE: this function should not be called directly, instead it should be used
3386 * as a callback for packet buffering.
3388 * NOTE: when configuring this function as a callback with
3389 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3390 * should point to an uint64_t value.
3393 * The previously buffered packets which could not be sent
3395 * The number of unsent packets in the pkts array
3397 * Pointer to an uint64_t value, which will be incremented by unsent
3400 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
3404 * Request the driver to free mbufs currently cached by the driver. The
3405 * driver will only free the mbuf if it is no longer in use. It is the
3406 * application's responsibility to ensure rte_eth_tx_buffer_flush(..) is
3410 * The port identifier of the Ethernet device.
3412 * The index of the transmit queue through which output packets must be
3414 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
3415 * to rte_eth_dev_configure().
3417 * Maximum number of packets to free. Use 0 to indicate all possible packets
3418 * should be freed. Note that a packet may be using multiple mbufs.
3421 * -ENODEV: Invalid interface
3422 * -EIO: device is removed
3423 * -ENOTSUP: Driver does not support function
3425 * 0-n: Number of packets freed. More packets may still remain in ring that
3429 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
3432 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
3435 enum rte_eth_event_ipsec_subtype {
3436 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
3437 /**< Unknown event type */
3438 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
3439 /**< Sequence number overflow */
3440 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
3441 /**< Soft time expiry of SA */
3442 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
3443 /**< Soft byte expiry of SA */
3444 RTE_ETH_EVENT_IPSEC_MAX
3445 /**< Max value of this enum */
3449 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
3450 * information of the IPsec offload event.
3452 struct rte_eth_event_ipsec_desc {
3453 enum rte_eth_event_ipsec_subtype subtype;
3454 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
3456 /**< Event specific metadata
3458 * For the following events, *userdata* registered
3459 * with the *rte_security_session* would be returned
3462 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
3463 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
3464 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
3466 * @see struct rte_security_session_conf
3472 * The eth device event type for interrupt, and maybe others in the future.
3474 enum rte_eth_event_type {
3475 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
3476 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
3477 RTE_ETH_EVENT_QUEUE_STATE,
3478 /**< queue state event (enabled/disabled) */
3479 RTE_ETH_EVENT_INTR_RESET,
3480 /**< reset interrupt event, sent to VF on PF reset */
3481 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
3482 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
3483 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
3484 RTE_ETH_EVENT_NEW, /**< port is probed */
3485 RTE_ETH_EVENT_DESTROY, /**< port is released */
3486 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
3487 RTE_ETH_EVENT_FLOW_AGED,/**< New aged-out flows is detected */
3488 RTE_ETH_EVENT_MAX /**< max value of this enum */
3491 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
3492 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
3493 /**< user application callback to be registered for interrupts */
3496 * Register a callback function for port event.
3500 * RTE_ETH_ALL means register the event for all port ids.
3504 * User supplied callback function to be called.
3506 * Pointer to the parameters for the registered callback.
3509 * - On success, zero.
3510 * - On failure, a negative value.
3512 int rte_eth_dev_callback_register(uint16_t port_id,
3513 enum rte_eth_event_type event,
3514 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3517 * Unregister a callback function for port event.
3521 * RTE_ETH_ALL means unregister the event for all port ids.
3525 * User supplied callback function to be called.
3527 * Pointer to the parameters for the registered callback. -1 means to
3528 * remove all for the same callback address and same event.
3531 * - On success, zero.
3532 * - On failure, a negative value.
3534 int rte_eth_dev_callback_unregister(uint16_t port_id,
3535 enum rte_eth_event_type event,
3536 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3539 * When there is no rx packet coming in Rx Queue for a long time, we can
3540 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
3541 * to be triggered when Rx packet arrives.
3543 * The rte_eth_dev_rx_intr_enable() function enables rx queue
3544 * interrupt on specific rx queue of a port.
3547 * The port identifier of the Ethernet device.
3549 * The index of the receive queue from which to retrieve input packets.
3550 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3551 * to rte_eth_dev_configure().
3553 * - (0) if successful.
3554 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3556 * - (-ENODEV) if *port_id* invalid.
3557 * - (-EIO) if device is removed.
3559 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
3562 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
3563 * interrupt and returns to polling mode.
3565 * The rte_eth_dev_rx_intr_disable() function disables rx queue
3566 * interrupt on specific rx queue of a port.
3569 * The port identifier of the Ethernet device.
3571 * The index of the receive queue from which to retrieve input packets.
3572 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3573 * to rte_eth_dev_configure().
3575 * - (0) if successful.
3576 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3578 * - (-ENODEV) if *port_id* invalid.
3579 * - (-EIO) if device is removed.
3581 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
3584 * RX Interrupt control per port.
3587 * The port identifier of the Ethernet device.
3589 * Epoll instance fd which the intr vector associated to.
3590 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3592 * The operation be performed for the vector.
3593 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3597 * - On success, zero.
3598 * - On failure, a negative value.
3600 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
3603 * RX Interrupt control per queue.
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 * Epoll instance fd which the intr vector associated to.
3613 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3615 * The operation be performed for the vector.
3616 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3620 * - On success, zero.
3621 * - On failure, a negative value.
3623 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
3624 int epfd, int op, void *data);
3628 * @b EXPERIMENTAL: this API may change without prior notice.
3630 * Get interrupt fd per Rx queue.
3633 * The port identifier of the Ethernet device.
3635 * The index of the receive queue from which to retrieve input packets.
3636 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3637 * to rte_eth_dev_configure().
3639 * - (>=0) the interrupt fd associated to the requested Rx queue if
3645 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
3648 * Turn on the LED on the Ethernet device.
3649 * This function turns on the LED on the Ethernet device.
3652 * The port identifier of the Ethernet device.
3654 * - (0) if successful.
3655 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3657 * - (-ENODEV) if *port_id* invalid.
3658 * - (-EIO) if device is removed.
3660 int rte_eth_led_on(uint16_t port_id);
3663 * Turn off the LED on the Ethernet device.
3664 * This function turns off the LED on the Ethernet device.
3667 * The port identifier of the Ethernet device.
3669 * - (0) if successful.
3670 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3672 * - (-ENODEV) if *port_id* invalid.
3673 * - (-EIO) if device is removed.
3675 int rte_eth_led_off(uint16_t port_id);
3679 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3681 * Get Forward Error Correction(FEC) capability.
3684 * The port identifier of the Ethernet device.
3685 * @param speed_fec_capa
3686 * speed_fec_capa is out only with per-speed capabilities.
3687 * If set to NULL, the function returns the required number
3688 * of required array entries.
3690 * a number of elements in an speed_fec_capa array.
3693 * - A non-negative value lower or equal to num: success. The return value
3694 * is the number of entries filled in the fec capa array.
3695 * - A non-negative value higher than num: error, the given fec capa array
3696 * is too small. The return value corresponds to the num that should
3697 * be given to succeed. The entries in fec capa array are not valid and
3698 * shall not be used by the caller.
3699 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3701 * - (-EIO) if device is removed.
3702 * - (-ENODEV) if *port_id* invalid.
3703 * - (-EINVAL) if *num* or *speed_fec_capa* invalid
3706 int rte_eth_fec_get_capability(uint16_t port_id,
3707 struct rte_eth_fec_capa *speed_fec_capa,
3712 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3714 * Get current Forward Error Correction(FEC) mode.
3715 * If link is down and AUTO is enabled, AUTO is returned, otherwise,
3716 * configured FEC mode is returned.
3717 * If link is up, current FEC mode is returned.
3720 * The port identifier of the Ethernet device.
3722 * A bitmask of enabled FEC modes. If AUTO bit is set, other
3723 * bits specify FEC modes which may be negotiated. If AUTO
3724 * bit is clear, specify FEC modes to be used (only one valid
3725 * mode per speed may be set).
3727 * - (0) if successful.
3728 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3730 * - (-EIO) if device is removed.
3731 * - (-ENODEV) if *port_id* invalid.
3734 int rte_eth_fec_get(uint16_t port_id, uint32_t *fec_capa);
3738 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3740 * Set Forward Error Correction(FEC) mode.
3743 * The port identifier of the Ethernet device.
3745 * A bitmask of allowed FEC modes. If AUTO bit is set, other
3746 * bits specify FEC modes which may be negotiated. If AUTO
3747 * bit is clear, specify FEC modes to be used (only one valid
3748 * mode per speed may be set).
3750 * - (0) if successful.
3751 * - (-EINVAL) if the FEC mode is not valid.
3752 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3753 * - (-EIO) if device is removed.
3754 * - (-ENODEV) if *port_id* invalid.
3757 int rte_eth_fec_set(uint16_t port_id, uint32_t fec_capa);
3760 * Get current status of the Ethernet link flow control for Ethernet device
3763 * The port identifier of the Ethernet device.
3765 * The pointer to the structure where to store the flow control parameters.
3767 * - (0) if successful.
3768 * - (-ENOTSUP) if hardware doesn't support flow control.
3769 * - (-ENODEV) if *port_id* invalid.
3770 * - (-EIO) if device is removed.
3772 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
3773 struct rte_eth_fc_conf *fc_conf);
3776 * Configure the Ethernet link flow control for Ethernet device
3779 * The port identifier of the Ethernet device.
3781 * The pointer to the structure of the flow control parameters.
3783 * - (0) if successful.
3784 * - (-ENOTSUP) if hardware doesn't support flow control mode.
3785 * - (-ENODEV) if *port_id* invalid.
3786 * - (-EINVAL) if bad parameter
3787 * - (-EIO) if flow control setup failure or device is removed.
3789 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
3790 struct rte_eth_fc_conf *fc_conf);
3793 * Configure the Ethernet priority flow control under DCB environment
3794 * for Ethernet device.
3797 * The port identifier of the Ethernet device.
3799 * The pointer to the structure of the priority flow control parameters.
3801 * - (0) if successful.
3802 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
3803 * - (-ENODEV) if *port_id* invalid.
3804 * - (-EINVAL) if bad parameter
3805 * - (-EIO) if flow control setup failure or device is removed.
3807 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
3808 struct rte_eth_pfc_conf *pfc_conf);
3811 * Add a MAC address to the set used for filtering incoming packets.
3814 * The port identifier of the Ethernet device.
3816 * The MAC address to add.
3818 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
3819 * not enabled, this should be set to 0.
3821 * - (0) if successfully added or *mac_addr* was already added.
3822 * - (-ENOTSUP) if hardware doesn't support this feature.
3823 * - (-ENODEV) if *port* is invalid.
3824 * - (-EIO) if device is removed.
3825 * - (-ENOSPC) if no more MAC addresses can be added.
3826 * - (-EINVAL) if MAC address is invalid.
3828 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct rte_ether_addr *mac_addr,
3832 * Remove a MAC address from the internal array of addresses.
3835 * The port identifier of the Ethernet device.
3837 * MAC address to remove.
3839 * - (0) if successful, or *mac_addr* didn't exist.
3840 * - (-ENOTSUP) if hardware doesn't support.
3841 * - (-ENODEV) if *port* invalid.
3842 * - (-EADDRINUSE) if attempting to remove the default MAC address
3844 int rte_eth_dev_mac_addr_remove(uint16_t port_id,
3845 struct rte_ether_addr *mac_addr);
3848 * Set the default MAC address.
3851 * The port identifier of the Ethernet device.
3853 * New default MAC address.
3855 * - (0) if successful, or *mac_addr* didn't exist.
3856 * - (-ENOTSUP) if hardware doesn't support.
3857 * - (-ENODEV) if *port* invalid.
3858 * - (-EINVAL) if MAC address is invalid.
3860 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
3861 struct rte_ether_addr *mac_addr);
3864 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3867 * The port identifier of the Ethernet device.
3871 * Redirection table size. The table size can be queried by
3872 * rte_eth_dev_info_get().
3874 * - (0) if successful.
3875 * - (-ENODEV) if *port_id* is invalid.
3876 * - (-ENOTSUP) if hardware doesn't support.
3877 * - (-EINVAL) if bad parameter.
3878 * - (-EIO) if device is removed.
3880 int rte_eth_dev_rss_reta_update(uint16_t port_id,
3881 struct rte_eth_rss_reta_entry64 *reta_conf,
3882 uint16_t reta_size);
3885 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3888 * The port identifier of the Ethernet device.
3890 * RETA to query. For each requested reta entry, corresponding bit
3891 * in mask must be set.
3893 * Redirection table size. The table size can be queried by
3894 * rte_eth_dev_info_get().
3896 * - (0) if successful.
3897 * - (-ENODEV) if *port_id* is invalid.
3898 * - (-ENOTSUP) if hardware doesn't support.
3899 * - (-EINVAL) if bad parameter.
3900 * - (-EIO) if device is removed.
3902 int rte_eth_dev_rss_reta_query(uint16_t port_id,
3903 struct rte_eth_rss_reta_entry64 *reta_conf,
3904 uint16_t reta_size);
3907 * Updates unicast hash table for receiving packet with the given destination
3908 * MAC address, and the packet is routed to all VFs for which the RX mode is
3909 * accept packets that match the unicast hash table.
3912 * The port identifier of the Ethernet device.
3914 * Unicast MAC address.
3916 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
3917 * 0 - Clear an unicast hash bit.
3919 * - (0) if successful.
3920 * - (-ENOTSUP) if hardware doesn't support.
3921 * - (-ENODEV) if *port_id* invalid.
3922 * - (-EIO) if device is removed.
3923 * - (-EINVAL) if bad parameter.
3925 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct rte_ether_addr *addr,
3929 * Updates all unicast hash bitmaps for receiving packet with any Unicast
3930 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
3931 * mode is accept packets that match the unicast hash table.
3934 * The port identifier of the Ethernet device.
3936 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
3938 * 0 - Clear all unicast hash bitmaps
3940 * - (0) if successful.
3941 * - (-ENOTSUP) if hardware doesn't support.
3942 * - (-ENODEV) if *port_id* invalid.
3943 * - (-EIO) if device is removed.
3944 * - (-EINVAL) if bad parameter.
3946 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
3949 * Set a traffic mirroring rule on an Ethernet device
3952 * The port identifier of the Ethernet device.
3953 * @param mirror_conf
3954 * The pointer to the traffic mirroring structure describing the mirroring rule.
3955 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
3956 * destination pool and the value of rule if enable vlan or pool mirroring.
3959 * The index of traffic mirroring rule, we support four separated rules.
3961 * 1 - Enable a mirroring rule.
3962 * 0 - Disable a mirroring rule.
3964 * - (0) if successful.
3965 * - (-ENOTSUP) if hardware doesn't support this feature.
3966 * - (-ENODEV) if *port_id* invalid.
3967 * - (-EIO) if device is removed.
3968 * - (-EINVAL) if the mr_conf information is not correct.
3970 int rte_eth_mirror_rule_set(uint16_t port_id,
3971 struct rte_eth_mirror_conf *mirror_conf,
3976 * Reset a traffic mirroring rule on an Ethernet device.
3979 * The port identifier of the Ethernet device.
3981 * The index of traffic mirroring rule, we support four separated rules.
3983 * - (0) if successful.
3984 * - (-ENOTSUP) if hardware doesn't support this feature.
3985 * - (-ENODEV) if *port_id* invalid.
3986 * - (-EIO) if device is removed.
3987 * - (-EINVAL) if bad parameter.
3989 int rte_eth_mirror_rule_reset(uint16_t port_id,
3993 * Set the rate limitation for a queue on an Ethernet device.
3996 * The port identifier of the Ethernet device.
4000 * The tx rate in Mbps. Allocated from the total port link speed.
4002 * - (0) if successful.
4003 * - (-ENOTSUP) if hardware doesn't support this feature.
4004 * - (-ENODEV) if *port_id* invalid.
4005 * - (-EIO) if device is removed.
4006 * - (-EINVAL) if bad parameter.
4008 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
4012 * Configuration of Receive Side Scaling hash computation of Ethernet device.
4015 * The port identifier of the Ethernet device.
4017 * The new configuration to use for RSS hash computation on the port.
4019 * - (0) if successful.
4020 * - (-ENODEV) if port identifier is invalid.
4021 * - (-EIO) if device is removed.
4022 * - (-ENOTSUP) if hardware doesn't support.
4023 * - (-EINVAL) if bad parameter.
4025 int rte_eth_dev_rss_hash_update(uint16_t port_id,
4026 struct rte_eth_rss_conf *rss_conf);
4029 * Retrieve current configuration of Receive Side Scaling hash computation
4030 * of Ethernet device.
4033 * The port identifier of the Ethernet device.
4035 * Where to store the current RSS hash configuration of the Ethernet device.
4037 * - (0) if successful.
4038 * - (-ENODEV) if port identifier is invalid.
4039 * - (-EIO) if device is removed.
4040 * - (-ENOTSUP) if hardware doesn't support RSS.
4043 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
4044 struct rte_eth_rss_conf *rss_conf);
4047 * Add UDP tunneling port for a type of tunnel.
4049 * Some NICs may require such configuration to properly parse a tunnel
4050 * with any standard or custom UDP port.
4051 * The packets with this UDP port will be parsed for this type of tunnel.
4052 * The device parser will also check the rest of the tunnel headers
4053 * before classifying the packet.
4055 * With some devices, this API will affect packet classification, i.e.:
4056 * - mbuf.packet_type reported on Rx
4057 * - rte_flow rules with tunnel items
4060 * The port identifier of the Ethernet device.
4062 * UDP tunneling configuration.
4065 * - (0) if successful.
4066 * - (-ENODEV) if port identifier is invalid.
4067 * - (-EIO) if device is removed.
4068 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4071 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
4072 struct rte_eth_udp_tunnel *tunnel_udp);
4075 * Delete UDP tunneling port for a type of tunnel.
4077 * The packets with this UDP port will not be classified as this type of tunnel
4078 * anymore if the device use such mapping for tunnel packet classification.
4080 * @see rte_eth_dev_udp_tunnel_port_add
4083 * The port identifier of the Ethernet device.
4085 * UDP tunneling configuration.
4088 * - (0) if successful.
4089 * - (-ENODEV) if port identifier is invalid.
4090 * - (-EIO) if device is removed.
4091 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4094 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
4095 struct rte_eth_udp_tunnel *tunnel_udp);
4098 * Get DCB information on an Ethernet device.
4101 * The port identifier of the Ethernet device.
4105 * - (0) if successful.
4106 * - (-ENODEV) if port identifier is invalid.
4107 * - (-EIO) if device is removed.
4108 * - (-ENOTSUP) if hardware doesn't support.
4110 int rte_eth_dev_get_dcb_info(uint16_t port_id,
4111 struct rte_eth_dcb_info *dcb_info);
4113 struct rte_eth_rxtx_callback;
4116 * Add a callback to be called on packet RX on a given port and queue.
4118 * This API configures a function to be called for each burst of
4119 * packets received on a given NIC port queue. The return value is a pointer
4120 * that can be used to later remove the callback using
4121 * rte_eth_remove_rx_callback().
4123 * Multiple functions are called in the order that they are added.
4126 * The port identifier of the Ethernet device.
4128 * The queue on the Ethernet device on which the callback is to be added.
4130 * The callback function
4132 * A generic pointer parameter which will be passed to each invocation of the
4133 * callback function on this port and queue. Inter-thread synchronization
4134 * of any user data changes is the responsibility of the user.
4138 * On success, a pointer value which can later be used to remove the callback.
4140 const struct rte_eth_rxtx_callback *
4141 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
4142 rte_rx_callback_fn fn, void *user_param);
4145 * Add a callback that must be called first on packet RX on a given port
4148 * This API configures a first function to be called for each burst of
4149 * packets received on a given NIC port queue. The return value is a pointer
4150 * that can be used to later remove the callback using
4151 * rte_eth_remove_rx_callback().
4153 * Multiple functions are called in the order that they are added.
4156 * The port identifier of the Ethernet device.
4158 * The queue on the Ethernet device on which the callback is to be added.
4160 * The callback function
4162 * A generic pointer parameter which will be passed to each invocation of the
4163 * callback function on this port and queue. Inter-thread synchronization
4164 * of any user data changes is the responsibility of the user.
4168 * On success, a pointer value which can later be used to remove the callback.
4170 const struct rte_eth_rxtx_callback *
4171 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
4172 rte_rx_callback_fn fn, void *user_param);
4175 * Add a callback to be called on packet TX on a given port and queue.
4177 * This API configures a function to be called for each burst of
4178 * packets sent on a given NIC port queue. The return value is a pointer
4179 * that can be used to later remove the callback using
4180 * rte_eth_remove_tx_callback().
4182 * Multiple functions are called in the order that they are added.
4185 * The port identifier of the Ethernet device.
4187 * The queue on the Ethernet device on which the callback is to be added.
4189 * The callback function
4191 * A generic pointer parameter which will be passed to each invocation of the
4192 * callback function on this port and queue. Inter-thread synchronization
4193 * of any user data changes is the responsibility of the user.
4197 * On success, a pointer value which can later be used to remove the callback.
4199 const struct rte_eth_rxtx_callback *
4200 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
4201 rte_tx_callback_fn fn, void *user_param);
4204 * Remove an RX packet callback from a given port and queue.
4206 * This function is used to removed callbacks that were added to a NIC port
4207 * queue using rte_eth_add_rx_callback().
4209 * Note: the callback is removed from the callback list but it isn't freed
4210 * since the it may still be in use. The memory for the callback can be
4211 * subsequently freed back by the application by calling rte_free():
4213 * - Immediately - if the port is stopped, or the user knows that no
4214 * callbacks are in flight e.g. if called from the thread doing RX/TX
4217 * - After a short delay - where the delay is sufficient to allow any
4218 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4219 * used to detect when data plane threads have ceased referencing the
4223 * The port identifier of the Ethernet device.
4225 * The queue on the Ethernet device from which the callback is to be removed.
4227 * User supplied callback created via rte_eth_add_rx_callback().
4230 * - 0: Success. Callback was removed.
4231 * - -ENODEV: If *port_id* is invalid.
4232 * - -ENOTSUP: Callback support is not available.
4233 * - -EINVAL: The queue_id is out of range, or the callback
4234 * is NULL or not found for the port/queue.
4236 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
4237 const struct rte_eth_rxtx_callback *user_cb);
4240 * Remove a TX packet callback from a given port and queue.
4242 * This function is used to removed callbacks that were added to a NIC port
4243 * queue using rte_eth_add_tx_callback().
4245 * Note: the callback is removed from the callback list but it isn't freed
4246 * since the it may still be in use. The memory for the callback can be
4247 * subsequently freed back by the application by calling rte_free():
4249 * - Immediately - if the port is stopped, or the user knows that no
4250 * callbacks are in flight e.g. if called from the thread doing RX/TX
4253 * - After a short delay - where the delay is sufficient to allow any
4254 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4255 * used to detect when data plane threads have ceased referencing the
4259 * The port identifier of the Ethernet device.
4261 * The queue on the Ethernet device from which the callback is to be removed.
4263 * User supplied callback created via rte_eth_add_tx_callback().
4266 * - 0: Success. Callback was removed.
4267 * - -ENODEV: If *port_id* is invalid.
4268 * - -ENOTSUP: Callback support is not available.
4269 * - -EINVAL: The queue_id is out of range, or the callback
4270 * is NULL or not found for the port/queue.
4272 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
4273 const struct rte_eth_rxtx_callback *user_cb);
4276 * Retrieve information about given port's RX queue.
4279 * The port identifier of the Ethernet device.
4281 * The RX queue on the Ethernet device for which information
4282 * will be retrieved.
4284 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
4285 * the information of the Ethernet device.
4289 * - -ENODEV: If *port_id* is invalid.
4290 * - -ENOTSUP: routine is not supported by the device PMD.
4291 * - -EINVAL: The queue_id is out of range, or the queue
4294 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4295 struct rte_eth_rxq_info *qinfo);
4298 * Retrieve information about given port's TX queue.
4301 * The port identifier of the Ethernet device.
4303 * The TX queue on the Ethernet device for which information
4304 * will be retrieved.
4306 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
4307 * the information of the Ethernet device.
4311 * - -ENODEV: If *port_id* is invalid.
4312 * - -ENOTSUP: routine is not supported by the device PMD.
4313 * - -EINVAL: The queue_id is out of range, or the queue
4316 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4317 struct rte_eth_txq_info *qinfo);
4320 * Retrieve information about the Rx packet burst mode.
4323 * The port identifier of the Ethernet device.
4325 * The Rx queue on the Ethernet device for which information
4326 * will be retrieved.
4328 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4329 * with the information of the packet burst mode.
4333 * - -ENODEV: If *port_id* is invalid.
4334 * - -ENOTSUP: routine is not supported by the device PMD.
4335 * - -EINVAL: The queue_id is out of range.
4338 int rte_eth_rx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4339 struct rte_eth_burst_mode *mode);
4342 * Retrieve information about the Tx packet burst mode.
4345 * The port identifier of the Ethernet device.
4347 * The Tx queue on the Ethernet device for which information
4348 * will be retrieved.
4350 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4351 * with the information of the packet burst mode.
4355 * - -ENODEV: If *port_id* is invalid.
4356 * - -ENOTSUP: routine is not supported by the device PMD.
4357 * - -EINVAL: The queue_id is out of range.
4360 int rte_eth_tx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4361 struct rte_eth_burst_mode *mode);
4365 * @b EXPERIMENTAL: this API may change without prior notice.
4367 * Retrieve the monitor condition for a given receive queue.
4370 * The port identifier of the Ethernet device.
4372 * The Rx queue on the Ethernet device for which information
4373 * will be retrieved.
4375 * The pointer to power-optimized monitoring condition structure.
4379 * -ENOTSUP: Operation not supported.
4380 * -EINVAL: Invalid parameters.
4381 * -ENODEV: Invalid port ID.
4384 int rte_eth_get_monitor_addr(uint16_t port_id, uint16_t queue_id,
4385 struct rte_power_monitor_cond *pmc);
4388 * Retrieve device registers and register attributes (number of registers and
4392 * The port identifier of the Ethernet device.
4394 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
4395 * NULL the function fills in the width and length fields. If non-NULL
4396 * the registers are put into the buffer pointed at by the data field.
4398 * - (0) if successful.
4399 * - (-ENOTSUP) if hardware doesn't support.
4400 * - (-ENODEV) if *port_id* invalid.
4401 * - (-EIO) if device is removed.
4402 * - others depends on the specific operations implementation.
4404 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
4407 * Retrieve size of device EEPROM
4410 * The port identifier of the Ethernet device.
4412 * - (>=0) EEPROM size if successful.
4413 * - (-ENOTSUP) if hardware doesn't support.
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_eeprom_length(uint16_t port_id);
4421 * Retrieve EEPROM and EEPROM attribute
4424 * The port identifier of the Ethernet device.
4426 * The template includes buffer for return EEPROM data and
4427 * EEPROM attributes to be filled.
4429 * - (0) if successful.
4430 * - (-ENOTSUP) if hardware doesn't support.
4431 * - (-ENODEV) if *port_id* invalid.
4432 * - (-EIO) if device is removed.
4433 * - others depends on the specific operations implementation.
4435 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4438 * Program EEPROM with provided data
4441 * The port identifier of the Ethernet device.
4443 * The template includes EEPROM data for programming and
4444 * EEPROM attributes to be filled
4446 * - (0) if successful.
4447 * - (-ENOTSUP) if hardware doesn't support.
4448 * - (-ENODEV) if *port_id* invalid.
4449 * - (-EIO) if device is removed.
4450 * - others depends on the specific operations implementation.
4452 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4456 * @b EXPERIMENTAL: this API may change without prior notice.
4458 * Retrieve the type and size of plugin module EEPROM
4461 * The port identifier of the Ethernet device.
4463 * The type and size of plugin module EEPROM.
4465 * - (0) if successful.
4466 * - (-ENOTSUP) if hardware doesn't support.
4467 * - (-ENODEV) if *port_id* invalid.
4468 * - (-EIO) if device is removed.
4469 * - others depends on the specific operations implementation.
4473 rte_eth_dev_get_module_info(uint16_t port_id,
4474 struct rte_eth_dev_module_info *modinfo);
4478 * @b EXPERIMENTAL: this API may change without prior notice.
4480 * Retrieve the data of plugin module EEPROM
4483 * The port identifier of the Ethernet device.
4485 * The template includes the plugin module EEPROM attributes, and the
4486 * buffer for return plugin module EEPROM data.
4488 * - (0) if successful.
4489 * - (-ENOTSUP) if hardware doesn't support.
4490 * - (-ENODEV) if *port_id* invalid.
4491 * - (-EIO) if device is removed.
4492 * - others depends on the specific operations implementation.
4496 rte_eth_dev_get_module_eeprom(uint16_t port_id,
4497 struct rte_dev_eeprom_info *info);
4500 * Set the list of multicast addresses to filter on an Ethernet device.
4503 * The port identifier of the Ethernet device.
4504 * @param mc_addr_set
4505 * The array of multicast addresses to set. Equal to NULL when the function
4506 * is invoked to flush the set of filtered addresses.
4508 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
4509 * when the function is invoked to flush the set of filtered addresses.
4511 * - (0) if successful.
4512 * - (-ENODEV) if *port_id* invalid.
4513 * - (-EIO) if device is removed.
4514 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
4515 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
4517 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
4518 struct rte_ether_addr *mc_addr_set,
4519 uint32_t nb_mc_addr);
4522 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
4525 * The port identifier of the Ethernet device.
4529 * - -ENODEV: The port ID is invalid.
4530 * - -EIO: if device is removed.
4531 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4533 int rte_eth_timesync_enable(uint16_t port_id);
4536 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
4539 * The port identifier of the Ethernet device.
4543 * - -ENODEV: The port ID is invalid.
4544 * - -EIO: if device is removed.
4545 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4547 int rte_eth_timesync_disable(uint16_t port_id);
4550 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
4553 * The port identifier of the Ethernet device.
4555 * Pointer to the timestamp struct.
4557 * Device specific flags. Used to pass the RX timesync register index to
4558 * i40e. Unused in igb/ixgbe, pass 0 instead.
4562 * - -EINVAL: No timestamp is available.
4563 * - -ENODEV: The port ID is invalid.
4564 * - -EIO: if device is removed.
4565 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4567 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
4568 struct timespec *timestamp, uint32_t flags);
4571 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
4574 * The port identifier of the Ethernet device.
4576 * Pointer to the timestamp struct.
4580 * - -EINVAL: No timestamp is available.
4581 * - -ENODEV: The port ID is invalid.
4582 * - -EIO: if device is removed.
4583 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4585 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
4586 struct timespec *timestamp);
4589 * Adjust the timesync clock on an Ethernet device.
4591 * This is usually used in conjunction with other Ethdev timesync functions to
4592 * synchronize the device time using the IEEE1588/802.1AS protocol.
4595 * The port identifier of the Ethernet device.
4597 * The adjustment in nanoseconds.
4601 * - -ENODEV: The port ID is invalid.
4602 * - -EIO: if device is removed.
4603 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4605 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
4608 * Read the time from 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 * Pointer to the timespec struct that holds the time.
4621 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
4624 * Set the time of the timesync clock on an Ethernet device.
4626 * This is usually used in conjunction with other Ethdev timesync functions to
4627 * synchronize the device time using the IEEE1588/802.1AS protocol.
4630 * The port identifier of the Ethernet device.
4632 * Pointer to the timespec struct that holds the time.
4636 * - -EINVAL: No timestamp is available.
4637 * - -ENODEV: The port ID is invalid.
4638 * - -EIO: if device is removed.
4639 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4641 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
4645 * @b EXPERIMENTAL: this API may change without prior notice.
4647 * Read the current clock counter of an Ethernet device
4649 * This returns the current raw clock value of an Ethernet device. It is
4650 * a raw amount of ticks, with no given time reference.
4651 * The value returned here is from the same clock than the one
4652 * filling timestamp field of Rx packets when using hardware timestamp
4653 * offload. Therefore it can be used to compute a precise conversion of
4654 * the device clock to the real time.
4656 * E.g, a simple heuristic to derivate the frequency would be:
4657 * uint64_t start, end;
4658 * rte_eth_read_clock(port, start);
4659 * rte_delay_ms(100);
4660 * rte_eth_read_clock(port, end);
4661 * double freq = (end - start) * 10;
4663 * Compute a common reference with:
4664 * uint64_t base_time_sec = current_time();
4665 * uint64_t base_clock;
4666 * rte_eth_read_clock(port, base_clock);
4668 * Then, convert the raw mbuf timestamp with:
4669 * base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq;
4671 * This simple example will not provide a very good accuracy. One must
4672 * at least measure multiple times the frequency and do a regression.
4673 * To avoid deviation from the system time, the common reference can
4674 * be repeated from time to time. The integer division can also be
4675 * converted by a multiplication and a shift for better performance.
4678 * The port identifier of the Ethernet device.
4680 * Pointer to the uint64_t that holds the raw clock value.
4684 * - -ENODEV: The port ID is invalid.
4685 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4689 rte_eth_read_clock(uint16_t port_id, uint64_t *clock);
4692 * Get the port id from device name. The device name should be specified
4694 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
4695 * - SoC device name, for example- fsl-gmac0
4696 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
4699 * pci address or name of the device
4701 * pointer to port identifier of the device
4703 * - (0) if successful and port_id is filled.
4704 * - (-ENODEV or -EINVAL) on failure.
4707 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
4710 * Get the device name from port id. The device name is specified as below:
4711 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
4712 * - SoC device name, for example- fsl-gmac0
4713 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
4716 * Port identifier of the device.
4718 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
4720 * - (0) if successful.
4721 * - (-ENODEV) if *port_id* is invalid.
4722 * - (-EINVAL) on failure.
4725 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
4728 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
4729 * the ethernet device information, otherwise adjust them to boundaries.
4732 * The port identifier of the Ethernet device.
4734 * A pointer to a uint16_t where the number of receive
4735 * descriptors stored.
4737 * A pointer to a uint16_t where the number of transmit
4738 * descriptors stored.
4740 * - (0) if successful.
4741 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
4743 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
4744 uint16_t *nb_rx_desc,
4745 uint16_t *nb_tx_desc);
4748 * Test if a port supports specific mempool ops.
4751 * Port identifier of the Ethernet device.
4753 * The name of the pool operations to test.
4755 * - 0: best mempool ops choice for this port.
4756 * - 1: mempool ops are supported for this port.
4757 * - -ENOTSUP: mempool ops not supported for this port.
4758 * - -ENODEV: Invalid port Identifier.
4759 * - -EINVAL: Pool param is null.
4762 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
4765 * Get the security context for the Ethernet device.
4768 * Port identifier of the Ethernet device
4771 * - pointer to security context on success.
4774 rte_eth_dev_get_sec_ctx(uint16_t port_id);
4778 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
4780 * Query the device hairpin capabilities.
4783 * The port identifier of the Ethernet device.
4785 * Pointer to a structure that will hold the hairpin capabilities.
4787 * - (0) if successful.
4788 * - (-ENOTSUP) if hardware doesn't support.
4791 int rte_eth_dev_hairpin_capability_get(uint16_t port_id,
4792 struct rte_eth_hairpin_cap *cap);
4796 * @b EXPERIMENTAL: this structure may change without prior notice.
4798 * ethernet device representor ID range entry
4800 struct rte_eth_representor_range {
4801 enum rte_eth_representor_type type; /**< Representor type */
4802 int controller; /**< Controller index */
4803 int pf; /**< Physical function index */
4806 int vf; /**< VF start index */
4807 int sf; /**< SF start index */
4809 uint32_t id_base; /**< Representor ID start index */
4810 uint32_t id_end; /**< Representor ID end index */
4811 char name[RTE_DEV_NAME_MAX_LEN]; /**< Representor name */
4816 * @b EXPERIMENTAL: this structure may change without prior notice.
4818 * Ethernet device representor information
4820 struct rte_eth_representor_info {
4821 uint16_t controller; /**< Controller ID of caller device. */
4822 uint16_t pf; /**< Physical function ID of caller device. */
4823 struct rte_eth_representor_range ranges[];/**< Representor ID range. */
4827 * Retrieve the representor info of the device.
4829 * Get device representor info to be able to calculate a unique
4830 * representor ID. @see rte_eth_representor_id_get helper.
4833 * The port identifier of the device.
4835 * A pointer to a representor info structure.
4836 * NULL to return number of range entries and allocate memory
4837 * for next call to store detail.
4839 * - (-ENOTSUP) if operation is not supported.
4840 * - (-ENODEV) if *port_id* invalid.
4841 * - (-EIO) if device is removed.
4842 * - (>=0) number of representor range entries supported by device.
4845 int rte_eth_representor_info_get(uint16_t port_id,
4846 struct rte_eth_representor_info *info);
4848 #include <rte_ethdev_core.h>
4852 * Retrieve a burst of input packets from a receive queue of an Ethernet
4853 * device. The retrieved packets are stored in *rte_mbuf* structures whose
4854 * pointers are supplied in the *rx_pkts* array.
4856 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
4857 * receive queue, up to *nb_pkts* packets, and for each completed RX
4858 * descriptor in the ring, it performs the following operations:
4860 * - Initialize the *rte_mbuf* data structure associated with the
4861 * RX descriptor according to the information provided by the NIC into
4862 * that RX descriptor.
4864 * - Store the *rte_mbuf* data structure into the next entry of the
4867 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
4868 * allocated from the memory pool associated with the receive queue at
4869 * initialization time.
4871 * When retrieving an input packet that was scattered by the controller
4872 * into multiple receive descriptors, the rte_eth_rx_burst() function
4873 * appends the associated *rte_mbuf* buffers to the first buffer of the
4876 * The rte_eth_rx_burst() function returns the number of packets
4877 * actually retrieved, which is the number of *rte_mbuf* data structures
4878 * effectively supplied into the *rx_pkts* array.
4879 * A return value equal to *nb_pkts* indicates that the RX queue contained
4880 * at least *rx_pkts* packets, and this is likely to signify that other
4881 * received packets remain in the input queue. Applications implementing
4882 * a "retrieve as much received packets as possible" policy can check this
4883 * specific case and keep invoking the rte_eth_rx_burst() function until
4884 * a value less than *nb_pkts* is returned.
4886 * This receive method has the following advantages:
4888 * - It allows a run-to-completion network stack engine to retrieve and
4889 * to immediately process received packets in a fast burst-oriented
4890 * approach, avoiding the overhead of unnecessary intermediate packet
4891 * queue/dequeue operations.
4893 * - Conversely, it also allows an asynchronous-oriented processing
4894 * method to retrieve bursts of received packets and to immediately
4895 * queue them for further parallel processing by another logical core,
4896 * for instance. However, instead of having received packets being
4897 * individually queued by the driver, this approach allows the caller
4898 * of the rte_eth_rx_burst() function to queue a burst of retrieved
4899 * packets at a time and therefore dramatically reduce the cost of
4900 * enqueue/dequeue operations per packet.
4902 * - It allows the rte_eth_rx_burst() function of the driver to take
4903 * advantage of burst-oriented hardware features (CPU cache,
4904 * prefetch instructions, and so on) to minimize the number of CPU
4905 * cycles per packet.
4907 * To summarize, the proposed receive API enables many
4908 * burst-oriented optimizations in both synchronous and asynchronous
4909 * packet processing environments with no overhead in both cases.
4912 * Some drivers using vector instructions require that *nb_pkts* is
4913 * divisible by 4 or 8, depending on the driver implementation.
4915 * The rte_eth_rx_burst() function does not provide any error
4916 * notification to avoid the corresponding overhead. As a hint, the
4917 * upper-level application might check the status of the device link once
4918 * being systematically returned a 0 value for a given number of tries.
4921 * The port identifier of the Ethernet device.
4923 * The index of the receive queue from which to retrieve input packets.
4924 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
4925 * to rte_eth_dev_configure().
4927 * The address of an array of pointers to *rte_mbuf* structures that
4928 * must be large enough to store *nb_pkts* pointers in it.
4930 * The maximum number of packets to retrieve.
4931 * The value must be divisible by 8 in order to work with any driver.
4933 * The number of packets actually retrieved, which is the number
4934 * of pointers to *rte_mbuf* structures effectively supplied to the
4937 static inline uint16_t
4938 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
4939 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
4941 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4944 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4945 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
4946 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
4948 if (queue_id >= dev->data->nb_rx_queues) {
4949 RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n", queue_id);
4953 nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
4956 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4957 struct rte_eth_rxtx_callback *cb;
4959 /* __ATOMIC_RELEASE memory order was used when the
4960 * call back was inserted into the list.
4961 * Since there is a clear dependency between loading
4962 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
4965 cb = __atomic_load_n(&dev->post_rx_burst_cbs[queue_id],
4968 if (unlikely(cb != NULL)) {
4970 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
4971 nb_pkts, cb->param);
4973 } while (cb != NULL);
4977 rte_ethdev_trace_rx_burst(port_id, queue_id, (void **)rx_pkts, nb_rx);
4982 * Get the number of used descriptors of a rx queue
4985 * The port identifier of the Ethernet device.
4987 * The queue id on the specific port.
4989 * The number of used descriptors in the specific queue, or:
4990 * - (-ENODEV) if *port_id* is invalid.
4991 * (-EINVAL) if *queue_id* is invalid
4992 * (-ENOTSUP) if the device does not support this function
4995 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
4997 struct rte_eth_dev *dev;
4999 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5000 dev = &rte_eth_devices[port_id];
5001 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_queue_count, -ENOTSUP);
5002 if (queue_id >= dev->data->nb_rx_queues ||
5003 dev->data->rx_queues[queue_id] == NULL)
5006 return (int)(*dev->rx_queue_count)(dev, queue_id);
5010 * Check if the DD bit of the specific RX descriptor in the queue has been set
5013 * The port identifier of the Ethernet device.
5015 * The queue id on the specific port.
5017 * The offset of the descriptor ID from tail.
5019 * - (1) if the specific DD bit is set.
5020 * - (0) if the specific DD bit is not set.
5021 * - (-ENODEV) if *port_id* invalid.
5022 * - (-ENOTSUP) if the device does not support this function
5026 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
5028 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
5029 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5030 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_done, -ENOTSUP);
5031 return (*dev->rx_descriptor_done)(dev->data->rx_queues[queue_id], offset);
5034 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
5035 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
5036 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5039 * Check the status of a Rx descriptor in the queue
5041 * It should be called in a similar context than the Rx function:
5042 * - on a dataplane core
5043 * - not concurrently on the same queue
5045 * Since it's a dataplane function, no check is performed on port_id and
5046 * queue_id. The caller must therefore ensure that the port is enabled
5047 * and the queue is configured and running.
5049 * Note: accessing to a random descriptor in the ring may trigger cache
5050 * misses and have a performance impact.
5053 * A valid port identifier of the Ethernet device which.
5055 * A valid Rx queue identifier on this port.
5057 * The offset of the descriptor starting from tail (0 is the next
5058 * packet to be received by the driver).
5061 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
5063 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
5064 * not yet processed by the driver (i.e. in the receive queue).
5065 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
5066 * the driver and not yet returned to hw, or reserved by the hw.
5067 * - (-EINVAL) bad descriptor offset.
5068 * - (-ENOTSUP) if the device does not support this function.
5069 * - (-ENODEV) bad port or queue (only if compiled with debug).
5072 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
5075 struct rte_eth_dev *dev;
5078 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5079 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5081 dev = &rte_eth_devices[port_id];
5082 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5083 if (queue_id >= dev->data->nb_rx_queues)
5086 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_status, -ENOTSUP);
5087 rxq = dev->data->rx_queues[queue_id];
5089 return (*dev->rx_descriptor_status)(rxq, offset);
5092 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
5093 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
5094 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5097 * Check the status of a Tx descriptor in the queue.
5099 * It should be called in a similar context than the Tx function:
5100 * - on a dataplane core
5101 * - not concurrently on the same queue
5103 * Since it's a dataplane function, no check is performed on port_id and
5104 * queue_id. The caller must therefore ensure that the port is enabled
5105 * and the queue is configured and running.
5107 * Note: accessing to a random descriptor in the ring may trigger cache
5108 * misses and have a performance impact.
5111 * A valid port identifier of the Ethernet device which.
5113 * A valid Tx queue identifier on this port.
5115 * The offset of the descriptor starting from tail (0 is the place where
5116 * the next packet will be send).
5119 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
5120 * in the transmit queue.
5121 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
5122 * be reused by the driver.
5123 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
5124 * driver or the hardware.
5125 * - (-EINVAL) bad descriptor offset.
5126 * - (-ENOTSUP) if the device does not support this function.
5127 * - (-ENODEV) bad port or queue (only if compiled with debug).
5129 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
5130 uint16_t queue_id, uint16_t offset)
5132 struct rte_eth_dev *dev;
5135 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5136 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5138 dev = &rte_eth_devices[port_id];
5139 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5140 if (queue_id >= dev->data->nb_tx_queues)
5143 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_descriptor_status, -ENOTSUP);
5144 txq = dev->data->tx_queues[queue_id];
5146 return (*dev->tx_descriptor_status)(txq, offset);
5150 * Send a burst of output packets on a transmit queue of an Ethernet device.
5152 * The rte_eth_tx_burst() function is invoked to transmit output packets
5153 * on the output queue *queue_id* of the Ethernet device designated by its
5155 * The *nb_pkts* parameter is the number of packets to send which are
5156 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5157 * allocated from a pool created with rte_pktmbuf_pool_create().
5158 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
5159 * up to the number of transmit descriptors available in the TX ring of the
5161 * For each packet to send, the rte_eth_tx_burst() function performs
5162 * the following operations:
5164 * - Pick up the next available descriptor in the transmit ring.
5166 * - Free the network buffer previously sent with that descriptor, if any.
5168 * - Initialize the transmit descriptor with the information provided
5169 * in the *rte_mbuf data structure.
5171 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
5172 * the rte_eth_tx_burst() function uses several transmit descriptors
5175 * The rte_eth_tx_burst() function returns the number of packets it
5176 * actually sent. A return value equal to *nb_pkts* means that all packets
5177 * have been sent, and this is likely to signify that other output packets
5178 * could be immediately transmitted again. Applications that implement a
5179 * "send as many packets to transmit as possible" policy can check this
5180 * specific case and keep invoking the rte_eth_tx_burst() function until
5181 * a value less than *nb_pkts* is returned.
5183 * It is the responsibility of the rte_eth_tx_burst() function to
5184 * transparently free the memory buffers of packets previously sent.
5185 * This feature is driven by the *tx_free_thresh* value supplied to the
5186 * rte_eth_dev_configure() function at device configuration time.
5187 * When the number of free TX descriptors drops below this threshold, the
5188 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
5189 * of those packets whose transmission was effectively completed.
5191 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
5192 * invoke this function concurrently on the same tx queue without SW lock.
5193 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
5195 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
5199 * The port identifier of the Ethernet device.
5201 * The index of the transmit queue through which output packets must be
5203 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5204 * to rte_eth_dev_configure().
5206 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5207 * which contain the output packets.
5209 * The maximum number of packets to transmit.
5211 * The number of output packets actually stored in transmit descriptors of
5212 * the transmit ring. The return value can be less than the value of the
5213 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
5215 static inline uint16_t
5216 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
5217 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5219 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
5221 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5222 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
5223 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
5225 if (queue_id >= dev->data->nb_tx_queues) {
5226 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5231 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
5232 struct rte_eth_rxtx_callback *cb;
5234 /* __ATOMIC_RELEASE memory order was used when the
5235 * call back was inserted into the list.
5236 * Since there is a clear dependency between loading
5237 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5240 cb = __atomic_load_n(&dev->pre_tx_burst_cbs[queue_id],
5243 if (unlikely(cb != NULL)) {
5245 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
5248 } while (cb != NULL);
5252 rte_ethdev_trace_tx_burst(port_id, queue_id, (void **)tx_pkts,
5254 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
5258 * Process a burst of output packets on a transmit queue of an Ethernet device.
5260 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
5261 * transmitted on the output queue *queue_id* of the Ethernet device designated
5263 * The *nb_pkts* parameter is the number of packets to be prepared which are
5264 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5265 * allocated from a pool created with rte_pktmbuf_pool_create().
5266 * For each packet to send, the rte_eth_tx_prepare() function performs
5267 * the following operations:
5269 * - Check if packet meets devices requirements for tx offloads.
5271 * - Check limitations about number of segments.
5273 * - Check additional requirements when debug is enabled.
5275 * - Update and/or reset required checksums when tx offload is set for packet.
5277 * Since this function can modify packet data, provided mbufs must be safely
5278 * writable (e.g. modified data cannot be in shared segment).
5280 * The rte_eth_tx_prepare() function returns the number of packets ready to be
5281 * sent. A return value equal to *nb_pkts* means that all packets are valid and
5282 * ready to be sent, otherwise stops processing on the first invalid packet and
5283 * leaves the rest packets untouched.
5285 * When this functionality is not implemented in the driver, all packets are
5286 * are returned untouched.
5289 * The port identifier of the Ethernet device.
5290 * The value must be a valid port id.
5292 * The index of the transmit queue through which output packets must be
5294 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5295 * to rte_eth_dev_configure().
5297 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5298 * which contain the output packets.
5300 * The maximum number of packets to process.
5302 * The number of packets correct and ready to be sent. The return value can be
5303 * less than the value of the *tx_pkts* parameter when some packet doesn't
5304 * meet devices requirements with rte_errno set appropriately:
5305 * - EINVAL: offload flags are not correctly set
5306 * - ENOTSUP: the offload feature is not supported by the hardware
5307 * - ENODEV: if *port_id* is invalid (with debug enabled only)
5311 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
5313 static inline uint16_t
5314 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
5315 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5317 struct rte_eth_dev *dev;
5319 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5320 if (!rte_eth_dev_is_valid_port(port_id)) {
5321 RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
5327 dev = &rte_eth_devices[port_id];
5329 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5330 if (queue_id >= dev->data->nb_tx_queues) {
5331 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5337 if (!dev->tx_pkt_prepare)
5340 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
5347 * Native NOOP operation for compilation targets which doesn't require any
5348 * preparations steps, and functional NOOP may introduce unnecessary performance
5351 * Generally this is not a good idea to turn it on globally and didn't should
5352 * be used if behavior of tx_preparation can change.
5355 static inline uint16_t
5356 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
5357 __rte_unused uint16_t queue_id,
5358 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5366 * Send any packets queued up for transmission on a port and HW queue
5368 * This causes an explicit flush of packets previously buffered via the
5369 * rte_eth_tx_buffer() function. It returns the number of packets successfully
5370 * sent to the NIC, and calls the error callback for any unsent packets. Unless
5371 * explicitly set up otherwise, the default callback simply frees the unsent
5372 * packets back to the owning mempool.
5375 * The port identifier of the Ethernet device.
5377 * The index of the transmit queue through which output packets must be
5379 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5380 * to rte_eth_dev_configure().
5382 * Buffer of packets to be transmit.
5384 * The number of packets successfully sent to the Ethernet device. The error
5385 * callback is called for any packets which could not be sent.
5387 static inline uint16_t
5388 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
5389 struct rte_eth_dev_tx_buffer *buffer)
5392 uint16_t to_send = buffer->length;
5397 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
5401 /* All packets sent, or to be dealt with by callback below */
5402 if (unlikely(sent != to_send))
5403 buffer->error_callback(&buffer->pkts[sent],
5404 (uint16_t)(to_send - sent),
5405 buffer->error_userdata);
5411 * Buffer a single packet for future transmission on a port and queue
5413 * This function takes a single mbuf/packet and buffers it for later
5414 * transmission on the particular port and queue specified. Once the buffer is
5415 * full of packets, an attempt will be made to transmit all the buffered
5416 * packets. In case of error, where not all packets can be transmitted, a
5417 * callback is called with the unsent packets as a parameter. If no callback
5418 * is explicitly set up, the unsent packets are just freed back to the owning
5419 * mempool. The function returns the number of packets actually sent i.e.
5420 * 0 if no buffer flush occurred, otherwise the number of packets successfully
5424 * The port identifier of the Ethernet device.
5426 * The index of the transmit queue through which output packets must be
5428 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5429 * to rte_eth_dev_configure().
5431 * Buffer used to collect packets to be sent.
5433 * Pointer to the packet mbuf to be sent.
5435 * 0 = packet has been buffered for later transmission
5436 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
5437 * causing N packets to be sent, and the error callback to be called for
5440 static __rte_always_inline uint16_t
5441 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
5442 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
5444 buffer->pkts[buffer->length++] = tx_pkt;
5445 if (buffer->length < buffer->size)
5448 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
5455 #endif /* _RTE_ETHDEV_H_ */