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 rte_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>
161 #include "rte_ethdev_trace_fp.h"
162 #include "rte_dev_info.h"
164 extern int rte_eth_dev_logtype;
166 #define RTE_ETHDEV_LOG(level, ...) \
167 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
172 * Initializes a device iterator.
174 * This iterator allows accessing a list of devices matching some devargs.
177 * Device iterator handle initialized by the function.
178 * The fields bus_str and cls_str might be dynamically allocated,
179 * and could be freed by calling rte_eth_iterator_cleanup().
182 * Device description string.
185 * 0 on successful initialization, negative otherwise.
187 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs);
190 * Iterates on devices with devargs filter.
191 * The ownership is not checked.
193 * The next port id is returned, and the iterator is updated.
196 * Device iterator handle initialized by rte_eth_iterator_init().
197 * Some fields bus_str and cls_str might be freed when no more port is found,
198 * by calling rte_eth_iterator_cleanup().
201 * A port id if found, RTE_MAX_ETHPORTS otherwise.
203 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter);
206 * Free some allocated fields of the iterator.
208 * This function is automatically called by rte_eth_iterator_next()
209 * on the last iteration (i.e. when no more matching port is found).
211 * It is safe to call this function twice; it will do nothing more.
214 * Device iterator handle initialized by rte_eth_iterator_init().
215 * The fields bus_str and cls_str are freed if needed.
217 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter);
220 * Macro to iterate over all ethdev ports matching some devargs.
222 * If a break is done before the end of the loop,
223 * the function rte_eth_iterator_cleanup() must be called.
226 * Iterated port id of type uint16_t.
228 * Device parameters input as string of type char*.
230 * Iterator handle of type struct rte_dev_iterator, used internally.
232 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \
233 for (rte_eth_iterator_init(iter, devargs), \
234 id = rte_eth_iterator_next(iter); \
235 id != RTE_MAX_ETHPORTS; \
236 id = rte_eth_iterator_next(iter))
239 * A structure used to retrieve statistics for an Ethernet port.
240 * Not all statistics fields in struct rte_eth_stats are supported
241 * by any type of network interface card (NIC). If any statistics
242 * field is not supported, its value is 0.
244 struct rte_eth_stats {
245 uint64_t ipackets; /**< Total number of successfully received packets. */
246 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
247 uint64_t ibytes; /**< Total number of successfully received bytes. */
248 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
250 /**< Total of RX packets dropped by the HW,
251 * because there are no available buffer (i.e. RX queues are full).
253 uint64_t ierrors; /**< Total number of erroneous received packets. */
254 uint64_t oerrors; /**< Total number of failed transmitted packets. */
255 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
256 /* Queue stats are limited to max 256 queues */
257 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
258 /**< Total number of queue RX packets. */
259 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
260 /**< Total number of queue TX packets. */
261 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
262 /**< Total number of successfully received queue bytes. */
263 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
264 /**< Total number of successfully transmitted queue bytes. */
265 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
266 /**< Total number of queue packets received that are dropped. */
270 * Device supported speeds bitmap flags
272 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
273 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
274 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
275 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
276 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
277 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
278 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
279 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
280 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
281 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
282 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
283 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
284 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
285 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
286 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
287 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
288 #define ETH_LINK_SPEED_200G (1 << 15) /**< 200 Gbps */
291 * Ethernet numeric link speeds in Mbps
293 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
294 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
295 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
296 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
297 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
298 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
299 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
300 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
301 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
302 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
303 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
304 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
305 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
306 #define ETH_SPEED_NUM_200G 200000 /**< 200 Gbps */
307 #define ETH_SPEED_NUM_UNKNOWN UINT32_MAX /**< Unknown */
310 * A structure used to retrieve link-level information of an Ethernet port.
313 struct rte_eth_link {
314 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
315 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
316 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
317 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
318 } __rte_aligned(8); /**< aligned for atomic64 read/write */
320 /* Utility constants */
321 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
322 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
323 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
324 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
325 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
326 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
327 #define RTE_ETH_LINK_MAX_STR_LEN 40 /**< Max length of default link string. */
330 * A structure used to configure the ring threshold registers of an RX/TX
331 * queue for an Ethernet port.
333 struct rte_eth_thresh {
334 uint8_t pthresh; /**< Ring prefetch threshold. */
335 uint8_t hthresh; /**< Ring host threshold. */
336 uint8_t wthresh; /**< Ring writeback threshold. */
340 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
342 #define ETH_MQ_RX_RSS_FLAG 0x1
343 #define ETH_MQ_RX_DCB_FLAG 0x2
344 #define ETH_MQ_RX_VMDQ_FLAG 0x4
347 * A set of values to identify what method is to be used to route
348 * packets to multiple queues.
350 enum rte_eth_rx_mq_mode {
351 /** None of DCB,RSS or VMDQ mode */
354 /** For RX side, only RSS is on */
355 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
356 /** For RX side,only DCB is on. */
357 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
358 /** Both DCB and RSS enable */
359 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
361 /** Only VMDQ, no RSS nor DCB */
362 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
363 /** RSS mode with VMDQ */
364 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
365 /** Use VMDQ+DCB to route traffic to queues */
366 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
367 /** Enable both VMDQ and DCB in VMDq */
368 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
373 * for rx mq mode backward compatible
375 #define ETH_RSS ETH_MQ_RX_RSS
376 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
377 #define ETH_DCB_RX ETH_MQ_RX_DCB
380 * A set of values to identify what method is to be used to transmit
381 * packets using multi-TCs.
383 enum rte_eth_tx_mq_mode {
384 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
385 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
386 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
387 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
391 * for tx mq mode backward compatible
393 #define ETH_DCB_NONE ETH_MQ_TX_NONE
394 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
395 #define ETH_DCB_TX ETH_MQ_TX_DCB
398 * A structure used to configure the RX features of an Ethernet port.
400 struct rte_eth_rxmode {
401 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
402 enum rte_eth_rx_mq_mode mq_mode;
403 uint32_t max_rx_pkt_len; /**< Only used if JUMBO_FRAME enabled. */
404 /** Maximum allowed size of LRO aggregated packet. */
405 uint32_t max_lro_pkt_size;
406 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
408 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
409 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
410 * structure are allowed to be set.
414 uint64_t reserved_64s[2]; /**< Reserved for future fields */
415 void *reserved_ptrs[2]; /**< Reserved for future fields */
419 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
420 * Note that single VLAN is treated the same as inner VLAN.
423 ETH_VLAN_TYPE_UNKNOWN = 0,
424 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
425 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
430 * A structure used to describe a vlan filter.
431 * If the bit corresponding to a VID is set, such VID is on.
433 struct rte_vlan_filter_conf {
438 * A structure used to configure the Receive Side Scaling (RSS) feature
439 * of an Ethernet port.
440 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
441 * to an array holding the RSS key to use for hashing specific header
442 * fields of received packets. The length of this array should be indicated
443 * by *rss_key_len* below. Otherwise, a default random hash key is used by
446 * The *rss_key_len* field of the *rss_conf* structure indicates the length
447 * in bytes of the array pointed by *rss_key*. To be compatible, this length
448 * will be checked in i40e only. Others assume 40 bytes to be used as before.
450 * The *rss_hf* field of the *rss_conf* structure indicates the different
451 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
452 * Supplying an *rss_hf* equal to zero disables the RSS feature.
454 struct rte_eth_rss_conf {
455 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
456 uint8_t rss_key_len; /**< hash key length in bytes. */
457 uint64_t rss_hf; /**< Hash functions to apply - see below. */
461 * A packet can be identified by hardware as different flow types. Different
462 * NIC hardware may support different flow types.
463 * Basically, the NIC hardware identifies the flow type as deep protocol as
464 * possible, and exclusively. For example, if a packet is identified as
465 * 'RTE_ETH_FLOW_NONFRAG_IPV4_TCP', it will not be any of other flow types,
466 * though it is an actual IPV4 packet.
468 #define RTE_ETH_FLOW_UNKNOWN 0
469 #define RTE_ETH_FLOW_RAW 1
470 #define RTE_ETH_FLOW_IPV4 2
471 #define RTE_ETH_FLOW_FRAG_IPV4 3
472 #define RTE_ETH_FLOW_NONFRAG_IPV4_TCP 4
473 #define RTE_ETH_FLOW_NONFRAG_IPV4_UDP 5
474 #define RTE_ETH_FLOW_NONFRAG_IPV4_SCTP 6
475 #define RTE_ETH_FLOW_NONFRAG_IPV4_OTHER 7
476 #define RTE_ETH_FLOW_IPV6 8
477 #define RTE_ETH_FLOW_FRAG_IPV6 9
478 #define RTE_ETH_FLOW_NONFRAG_IPV6_TCP 10
479 #define RTE_ETH_FLOW_NONFRAG_IPV6_UDP 11
480 #define RTE_ETH_FLOW_NONFRAG_IPV6_SCTP 12
481 #define RTE_ETH_FLOW_NONFRAG_IPV6_OTHER 13
482 #define RTE_ETH_FLOW_L2_PAYLOAD 14
483 #define RTE_ETH_FLOW_IPV6_EX 15
484 #define RTE_ETH_FLOW_IPV6_TCP_EX 16
485 #define RTE_ETH_FLOW_IPV6_UDP_EX 17
486 #define RTE_ETH_FLOW_PORT 18
487 /**< Consider device port number as a flow differentiator */
488 #define RTE_ETH_FLOW_VXLAN 19 /**< VXLAN protocol based flow */
489 #define RTE_ETH_FLOW_GENEVE 20 /**< GENEVE protocol based flow */
490 #define RTE_ETH_FLOW_NVGRE 21 /**< NVGRE protocol based flow */
491 #define RTE_ETH_FLOW_VXLAN_GPE 22 /**< VXLAN-GPE protocol based flow */
492 #define RTE_ETH_FLOW_GTPU 23 /**< GTPU protocol based flow */
493 #define RTE_ETH_FLOW_MAX 24
496 * Below macros are defined for RSS offload types, they can be used to
497 * fill rte_eth_rss_conf.rss_hf or rte_flow_action_rss.types.
499 #define ETH_RSS_IPV4 (1ULL << 2)
500 #define ETH_RSS_FRAG_IPV4 (1ULL << 3)
501 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << 4)
502 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << 5)
503 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << 6)
504 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << 7)
505 #define ETH_RSS_IPV6 (1ULL << 8)
506 #define ETH_RSS_FRAG_IPV6 (1ULL << 9)
507 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << 10)
508 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << 11)
509 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << 12)
510 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << 13)
511 #define ETH_RSS_L2_PAYLOAD (1ULL << 14)
512 #define ETH_RSS_IPV6_EX (1ULL << 15)
513 #define ETH_RSS_IPV6_TCP_EX (1ULL << 16)
514 #define ETH_RSS_IPV6_UDP_EX (1ULL << 17)
515 #define ETH_RSS_PORT (1ULL << 18)
516 #define ETH_RSS_VXLAN (1ULL << 19)
517 #define ETH_RSS_GENEVE (1ULL << 20)
518 #define ETH_RSS_NVGRE (1ULL << 21)
519 #define ETH_RSS_GTPU (1ULL << 23)
520 #define ETH_RSS_ETH (1ULL << 24)
521 #define ETH_RSS_S_VLAN (1ULL << 25)
522 #define ETH_RSS_C_VLAN (1ULL << 26)
523 #define ETH_RSS_ESP (1ULL << 27)
524 #define ETH_RSS_AH (1ULL << 28)
525 #define ETH_RSS_L2TPV3 (1ULL << 29)
526 #define ETH_RSS_PFCP (1ULL << 30)
527 #define ETH_RSS_PPPOE (1ULL << 31)
528 #define ETH_RSS_ECPRI (1ULL << 32)
531 * We use the following macros to combine with above ETH_RSS_* for
532 * more specific input set selection. These bits are defined starting
533 * from the high end of the 64 bits.
534 * Note: If we use above ETH_RSS_* without SRC/DST_ONLY, it represents
535 * both SRC and DST are taken into account. If SRC_ONLY and DST_ONLY of
536 * the same level are used simultaneously, it is the same case as none of
539 #define ETH_RSS_L3_SRC_ONLY (1ULL << 63)
540 #define ETH_RSS_L3_DST_ONLY (1ULL << 62)
541 #define ETH_RSS_L4_SRC_ONLY (1ULL << 61)
542 #define ETH_RSS_L4_DST_ONLY (1ULL << 60)
543 #define ETH_RSS_L2_SRC_ONLY (1ULL << 59)
544 #define ETH_RSS_L2_DST_ONLY (1ULL << 58)
547 * Only select IPV6 address prefix as RSS input set according to
548 * https://tools.ietf.org/html/rfc6052
549 * Must be combined with ETH_RSS_IPV6, ETH_RSS_NONFRAG_IPV6_UDP,
550 * ETH_RSS_NONFRAG_IPV6_TCP, ETH_RSS_NONFRAG_IPV6_SCTP.
552 #define RTE_ETH_RSS_L3_PRE32 (1ULL << 57)
553 #define RTE_ETH_RSS_L3_PRE40 (1ULL << 56)
554 #define RTE_ETH_RSS_L3_PRE48 (1ULL << 55)
555 #define RTE_ETH_RSS_L3_PRE56 (1ULL << 54)
556 #define RTE_ETH_RSS_L3_PRE64 (1ULL << 53)
557 #define RTE_ETH_RSS_L3_PRE96 (1ULL << 52)
560 * Use the following macros to combine with the above layers
561 * to choose inner and outer layers or both for RSS computation.
562 * Bits 50 and 51 are reserved for this.
566 * level 0, requests the default behavior.
567 * Depending on the packet type, it can mean outermost, innermost,
568 * anything in between or even no RSS.
569 * It basically stands for the innermost encapsulation level RSS
570 * can be performed on according to PMD and device capabilities.
572 #define ETH_RSS_LEVEL_PMD_DEFAULT (0ULL << 50)
575 * level 1, requests RSS to be performed on the outermost packet
576 * encapsulation level.
578 #define ETH_RSS_LEVEL_OUTERMOST (1ULL << 50)
581 * level 2, requests RSS to be performed on the specified inner packet
582 * encapsulation level, from outermost to innermost (lower to higher values).
584 #define ETH_RSS_LEVEL_INNERMOST (2ULL << 50)
585 #define ETH_RSS_LEVEL_MASK (3ULL << 50)
587 #define ETH_RSS_LEVEL(rss_hf) ((rss_hf & ETH_RSS_LEVEL_MASK) >> 50)
590 * For input set change of hash filter, if SRC_ONLY and DST_ONLY of
591 * the same level are used simultaneously, it is the same case as
592 * none of them are added.
595 * RSS types with SRC/DST_ONLY.
599 static inline uint64_t
600 rte_eth_rss_hf_refine(uint64_t rss_hf)
602 if ((rss_hf & ETH_RSS_L3_SRC_ONLY) && (rss_hf & ETH_RSS_L3_DST_ONLY))
603 rss_hf &= ~(ETH_RSS_L3_SRC_ONLY | ETH_RSS_L3_DST_ONLY);
605 if ((rss_hf & ETH_RSS_L4_SRC_ONLY) && (rss_hf & ETH_RSS_L4_DST_ONLY))
606 rss_hf &= ~(ETH_RSS_L4_SRC_ONLY | ETH_RSS_L4_DST_ONLY);
611 #define ETH_RSS_IPV6_PRE32 ( \
613 RTE_ETH_RSS_L3_PRE32)
615 #define ETH_RSS_IPV6_PRE40 ( \
617 RTE_ETH_RSS_L3_PRE40)
619 #define ETH_RSS_IPV6_PRE48 ( \
621 RTE_ETH_RSS_L3_PRE48)
623 #define ETH_RSS_IPV6_PRE56 ( \
625 RTE_ETH_RSS_L3_PRE56)
627 #define ETH_RSS_IPV6_PRE64 ( \
629 RTE_ETH_RSS_L3_PRE64)
631 #define ETH_RSS_IPV6_PRE96 ( \
633 RTE_ETH_RSS_L3_PRE96)
635 #define ETH_RSS_IPV6_PRE32_UDP ( \
636 ETH_RSS_NONFRAG_IPV6_UDP | \
637 RTE_ETH_RSS_L3_PRE32)
639 #define ETH_RSS_IPV6_PRE40_UDP ( \
640 ETH_RSS_NONFRAG_IPV6_UDP | \
641 RTE_ETH_RSS_L3_PRE40)
643 #define ETH_RSS_IPV6_PRE48_UDP ( \
644 ETH_RSS_NONFRAG_IPV6_UDP | \
645 RTE_ETH_RSS_L3_PRE48)
647 #define ETH_RSS_IPV6_PRE56_UDP ( \
648 ETH_RSS_NONFRAG_IPV6_UDP | \
649 RTE_ETH_RSS_L3_PRE56)
651 #define ETH_RSS_IPV6_PRE64_UDP ( \
652 ETH_RSS_NONFRAG_IPV6_UDP | \
653 RTE_ETH_RSS_L3_PRE64)
655 #define ETH_RSS_IPV6_PRE96_UDP ( \
656 ETH_RSS_NONFRAG_IPV6_UDP | \
657 RTE_ETH_RSS_L3_PRE96)
659 #define ETH_RSS_IPV6_PRE32_TCP ( \
660 ETH_RSS_NONFRAG_IPV6_TCP | \
661 RTE_ETH_RSS_L3_PRE32)
663 #define ETH_RSS_IPV6_PRE40_TCP ( \
664 ETH_RSS_NONFRAG_IPV6_TCP | \
665 RTE_ETH_RSS_L3_PRE40)
667 #define ETH_RSS_IPV6_PRE48_TCP ( \
668 ETH_RSS_NONFRAG_IPV6_TCP | \
669 RTE_ETH_RSS_L3_PRE48)
671 #define ETH_RSS_IPV6_PRE56_TCP ( \
672 ETH_RSS_NONFRAG_IPV6_TCP | \
673 RTE_ETH_RSS_L3_PRE56)
675 #define ETH_RSS_IPV6_PRE64_TCP ( \
676 ETH_RSS_NONFRAG_IPV6_TCP | \
677 RTE_ETH_RSS_L3_PRE64)
679 #define ETH_RSS_IPV6_PRE96_TCP ( \
680 ETH_RSS_NONFRAG_IPV6_TCP | \
681 RTE_ETH_RSS_L3_PRE96)
683 #define ETH_RSS_IPV6_PRE32_SCTP ( \
684 ETH_RSS_NONFRAG_IPV6_SCTP | \
685 RTE_ETH_RSS_L3_PRE32)
687 #define ETH_RSS_IPV6_PRE40_SCTP ( \
688 ETH_RSS_NONFRAG_IPV6_SCTP | \
689 RTE_ETH_RSS_L3_PRE40)
691 #define ETH_RSS_IPV6_PRE48_SCTP ( \
692 ETH_RSS_NONFRAG_IPV6_SCTP | \
693 RTE_ETH_RSS_L3_PRE48)
695 #define ETH_RSS_IPV6_PRE56_SCTP ( \
696 ETH_RSS_NONFRAG_IPV6_SCTP | \
697 RTE_ETH_RSS_L3_PRE56)
699 #define ETH_RSS_IPV6_PRE64_SCTP ( \
700 ETH_RSS_NONFRAG_IPV6_SCTP | \
701 RTE_ETH_RSS_L3_PRE64)
703 #define ETH_RSS_IPV6_PRE96_SCTP ( \
704 ETH_RSS_NONFRAG_IPV6_SCTP | \
705 RTE_ETH_RSS_L3_PRE96)
707 #define ETH_RSS_IP ( \
709 ETH_RSS_FRAG_IPV4 | \
710 ETH_RSS_NONFRAG_IPV4_OTHER | \
712 ETH_RSS_FRAG_IPV6 | \
713 ETH_RSS_NONFRAG_IPV6_OTHER | \
716 #define ETH_RSS_UDP ( \
717 ETH_RSS_NONFRAG_IPV4_UDP | \
718 ETH_RSS_NONFRAG_IPV6_UDP | \
721 #define ETH_RSS_TCP ( \
722 ETH_RSS_NONFRAG_IPV4_TCP | \
723 ETH_RSS_NONFRAG_IPV6_TCP | \
726 #define ETH_RSS_SCTP ( \
727 ETH_RSS_NONFRAG_IPV4_SCTP | \
728 ETH_RSS_NONFRAG_IPV6_SCTP)
730 #define ETH_RSS_TUNNEL ( \
735 #define ETH_RSS_VLAN ( \
739 /**< Mask of valid RSS hash protocols */
740 #define ETH_RSS_PROTO_MASK ( \
742 ETH_RSS_FRAG_IPV4 | \
743 ETH_RSS_NONFRAG_IPV4_TCP | \
744 ETH_RSS_NONFRAG_IPV4_UDP | \
745 ETH_RSS_NONFRAG_IPV4_SCTP | \
746 ETH_RSS_NONFRAG_IPV4_OTHER | \
748 ETH_RSS_FRAG_IPV6 | \
749 ETH_RSS_NONFRAG_IPV6_TCP | \
750 ETH_RSS_NONFRAG_IPV6_UDP | \
751 ETH_RSS_NONFRAG_IPV6_SCTP | \
752 ETH_RSS_NONFRAG_IPV6_OTHER | \
753 ETH_RSS_L2_PAYLOAD | \
755 ETH_RSS_IPV6_TCP_EX | \
756 ETH_RSS_IPV6_UDP_EX | \
763 * Definitions used for redirection table entry size.
764 * Some RSS RETA sizes may not be supported by some drivers, check the
765 * documentation or the description of relevant functions for more details.
767 #define ETH_RSS_RETA_SIZE_64 64
768 #define ETH_RSS_RETA_SIZE_128 128
769 #define ETH_RSS_RETA_SIZE_256 256
770 #define ETH_RSS_RETA_SIZE_512 512
771 #define RTE_RETA_GROUP_SIZE 64
773 /* Definitions used for VMDQ and DCB functionality */
774 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
775 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
776 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
777 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
779 /* DCB capability defines */
780 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
781 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
783 /* Definitions used for VLAN Offload functionality */
784 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
785 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
786 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
787 #define ETH_QINQ_STRIP_OFFLOAD 0x0008 /**< QINQ Strip On/Off */
789 /* Definitions used for mask VLAN setting */
790 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
791 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
792 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
793 #define ETH_QINQ_STRIP_MASK 0x0008 /**< QINQ Strip setting mask */
794 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
796 /* Definitions used for receive MAC address */
797 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
799 /* Definitions used for unicast hash */
800 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
802 /* Definitions used for VMDQ pool rx mode setting */
803 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
804 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
805 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
806 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
807 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
809 /** Maximum nb. of vlan per mirror rule */
810 #define ETH_MIRROR_MAX_VLANS 64
812 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
813 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
814 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
815 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
816 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
819 * A structure used to configure VLAN traffic mirror of an Ethernet port.
821 struct rte_eth_vlan_mirror {
822 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
823 /** VLAN ID list for vlan mirroring. */
824 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
828 * A structure used to configure traffic mirror of an Ethernet port.
830 struct rte_eth_mirror_conf {
831 uint8_t rule_type; /**< Mirroring rule type */
832 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
833 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
834 /** VLAN ID setting for VLAN mirroring. */
835 struct rte_eth_vlan_mirror vlan;
839 * A structure used to configure 64 entries of Redirection Table of the
840 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
841 * more than 64 entries supported by hardware, an array of this structure
844 struct rte_eth_rss_reta_entry64 {
846 /**< Mask bits indicate which entries need to be updated/queried. */
847 uint16_t reta[RTE_RETA_GROUP_SIZE];
848 /**< Group of 64 redirection table entries. */
852 * This enum indicates the possible number of traffic classes
853 * in DCB configurations
855 enum rte_eth_nb_tcs {
856 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
857 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
861 * This enum indicates the possible number of queue pools
862 * in VMDQ configurations.
864 enum rte_eth_nb_pools {
865 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
866 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
867 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
868 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
871 /* This structure may be extended in future. */
872 struct rte_eth_dcb_rx_conf {
873 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
874 /** Traffic class each UP mapped to. */
875 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
878 struct rte_eth_vmdq_dcb_tx_conf {
879 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
880 /** Traffic class each UP mapped to. */
881 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
884 struct rte_eth_dcb_tx_conf {
885 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
886 /** Traffic class each UP mapped to. */
887 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
890 struct rte_eth_vmdq_tx_conf {
891 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
895 * A structure used to configure the VMDQ+DCB feature
896 * of an Ethernet port.
898 * Using this feature, packets are routed to a pool of queues, based
899 * on the vlan id in the vlan tag, and then to a specific queue within
900 * that pool, using the user priority vlan tag field.
902 * A default pool may be used, if desired, to route all traffic which
903 * does not match the vlan filter rules.
905 struct rte_eth_vmdq_dcb_conf {
906 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
907 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
908 uint8_t default_pool; /**< The default pool, if applicable */
909 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
911 uint16_t vlan_id; /**< The vlan id of the received frame */
912 uint64_t pools; /**< Bitmask of pools for packet rx */
913 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
914 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
915 /**< Selects a queue in a pool */
919 * A structure used to configure the VMDQ feature of an Ethernet port when
920 * not combined with the DCB feature.
922 * Using this feature, packets are routed to a pool of queues. By default,
923 * the pool selection is based on the MAC address, the vlan id in the
924 * vlan tag as specified in the pool_map array.
925 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
926 * selection using only the MAC address. MAC address to pool mapping is done
927 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
928 * corresponding to the pool id.
930 * Queue selection within the selected pool will be done using RSS when
931 * it is enabled or revert to the first queue of the pool if not.
933 * A default pool may be used, if desired, to route all traffic which
934 * does not match the vlan filter rules or any pool MAC address.
936 struct rte_eth_vmdq_rx_conf {
937 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
938 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
939 uint8_t default_pool; /**< The default pool, if applicable */
940 uint8_t enable_loop_back; /**< Enable VT loop back */
941 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
942 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
944 uint16_t vlan_id; /**< The vlan id of the received frame */
945 uint64_t pools; /**< Bitmask of pools for packet rx */
946 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
950 * A structure used to configure the TX features of an Ethernet port.
952 struct rte_eth_txmode {
953 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
955 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
956 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
957 * structure are allowed to be set.
963 uint8_t hw_vlan_reject_tagged : 1,
964 /**< If set, reject sending out tagged pkts */
965 hw_vlan_reject_untagged : 1,
966 /**< If set, reject sending out untagged pkts */
967 hw_vlan_insert_pvid : 1;
968 /**< If set, enable port based VLAN insertion */
970 uint64_t reserved_64s[2]; /**< Reserved for future fields */
971 void *reserved_ptrs[2]; /**< Reserved for future fields */
976 * @b EXPERIMENTAL: this structure may change without prior notice.
978 * A structure used to configure an Rx packet segment to split.
980 * If RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT flag is set in offloads field,
981 * the PMD will split the received packets into multiple segments
982 * according to the specification in the description array:
984 * - The first network buffer will be allocated from the memory pool,
985 * specified in the first array element, the second buffer, from the
986 * pool in the second element, and so on.
988 * - The offsets from the segment description elements specify
989 * the data offset from the buffer beginning except the first mbuf.
990 * The first segment offset is added with RTE_PKTMBUF_HEADROOM.
992 * - The lengths in the elements define the maximal data amount
993 * being received to each segment. The receiving starts with filling
994 * up the first mbuf data buffer up to specified length. If the
995 * there are data remaining (packet is longer than buffer in the first
996 * mbuf) the following data will be pushed to the next segment
997 * up to its own length, and so on.
999 * - If the length in the segment description element is zero
1000 * the actual buffer size will be deduced from the appropriate
1001 * memory pool properties.
1003 * - If there is not enough elements to describe the buffer for entire
1004 * packet of maximal length the following parameters will be used
1005 * for the all remaining segments:
1006 * - pool from the last valid element
1007 * - the buffer size from this pool
1010 struct rte_eth_rxseg_split {
1011 struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
1012 uint16_t length; /**< Segment data length, configures split point. */
1013 uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */
1014 uint32_t reserved; /**< Reserved field. */
1019 * @b EXPERIMENTAL: this structure may change without prior notice.
1021 * A common structure used to describe Rx packet segment properties.
1023 union rte_eth_rxseg {
1024 /* The settings for buffer split offload. */
1025 struct rte_eth_rxseg_split split;
1026 /* The other features settings should be added here. */
1030 * A structure used to configure an RX ring of an Ethernet port.
1032 struct rte_eth_rxconf {
1033 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
1034 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
1035 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
1036 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1037 uint16_t rx_nseg; /**< Number of descriptions in rx_seg array. */
1039 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
1040 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
1041 * fields on rte_eth_dev_info structure are allowed to be set.
1045 * Points to the array of segment descriptions for an entire packet.
1046 * Array elements are properties for consecutive Rx segments.
1048 * The supported capabilities of receiving segmentation is reported
1049 * in rte_eth_dev_info.rx_seg_capa field.
1051 union rte_eth_rxseg *rx_seg;
1053 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1054 void *reserved_ptrs[2]; /**< Reserved for future fields */
1058 * A structure used to configure a TX ring of an Ethernet port.
1060 struct rte_eth_txconf {
1061 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
1062 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
1063 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
1064 less free descriptors than this value. */
1066 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1068 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
1069 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
1070 * fields on rte_eth_dev_info structure are allowed to be set.
1074 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1075 void *reserved_ptrs[2]; /**< Reserved for future fields */
1080 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1082 * A structure used to return the hairpin capabilities that are supported.
1084 struct rte_eth_hairpin_cap {
1085 /** The max number of hairpin queues (different bindings). */
1086 uint16_t max_nb_queues;
1087 /** Max number of Rx queues to be connected to one Tx queue. */
1088 uint16_t max_rx_2_tx;
1089 /** Max number of Tx queues to be connected to one Rx queue. */
1090 uint16_t max_tx_2_rx;
1091 uint16_t max_nb_desc; /**< The max num of descriptors. */
1094 #define RTE_ETH_MAX_HAIRPIN_PEERS 32
1098 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1100 * A structure used to hold hairpin peer data.
1102 struct rte_eth_hairpin_peer {
1103 uint16_t port; /**< Peer port. */
1104 uint16_t queue; /**< Peer queue. */
1109 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1111 * A structure used to configure hairpin binding.
1113 struct rte_eth_hairpin_conf {
1114 uint32_t peer_count:16; /**< The number of peers. */
1117 * Explicit Tx flow rule mode.
1118 * One hairpin pair of queues should have the same attribute.
1120 * - When set, the user should be responsible for inserting the hairpin
1121 * Tx part flows and removing them.
1122 * - When clear, the PMD will try to handle the Tx part of the flows,
1123 * e.g., by splitting one flow into two parts.
1125 uint32_t tx_explicit:1;
1128 * Manually bind hairpin queues.
1129 * One hairpin pair of queues should have the same attribute.
1131 * - When set, to enable hairpin, the user should call the hairpin bind
1132 * function after all the queues are set up properly and the ports are
1133 * started. Also, the hairpin unbind function should be called
1134 * accordingly before stopping a port that with hairpin configured.
1135 * - When clear, the PMD will try to enable the hairpin with the queues
1136 * configured automatically during port start.
1138 uint32_t manual_bind:1;
1139 uint32_t reserved:14; /**< Reserved bits. */
1140 struct rte_eth_hairpin_peer peers[RTE_ETH_MAX_HAIRPIN_PEERS];
1144 * A structure contains information about HW descriptor ring limitations.
1146 struct rte_eth_desc_lim {
1147 uint16_t nb_max; /**< Max allowed number of descriptors. */
1148 uint16_t nb_min; /**< Min allowed number of descriptors. */
1149 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
1152 * Max allowed number of segments per whole packet.
1154 * - For TSO packet this is the total number of data descriptors allowed
1157 * @see nb_mtu_seg_max
1159 uint16_t nb_seg_max;
1162 * Max number of segments per one MTU.
1164 * - For non-TSO packet, this is the maximum allowed number of segments
1165 * in a single transmit packet.
1167 * - For TSO packet each segment within the TSO may span up to this
1172 uint16_t nb_mtu_seg_max;
1176 * This enum indicates the flow control mode
1178 enum rte_eth_fc_mode {
1179 RTE_FC_NONE = 0, /**< Disable flow control. */
1180 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
1181 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
1182 RTE_FC_FULL /**< Enable flow control on both side. */
1186 * A structure used to configure Ethernet flow control parameter.
1187 * These parameters will be configured into the register of the NIC.
1188 * Please refer to the corresponding data sheet for proper value.
1190 struct rte_eth_fc_conf {
1191 uint32_t high_water; /**< High threshold value to trigger XOFF */
1192 uint32_t low_water; /**< Low threshold value to trigger XON */
1193 uint16_t pause_time; /**< Pause quota in the Pause frame */
1194 uint16_t send_xon; /**< Is XON frame need be sent */
1195 enum rte_eth_fc_mode mode; /**< Link flow control mode */
1196 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
1197 uint8_t autoneg; /**< Use Pause autoneg */
1201 * A structure used to configure Ethernet priority flow control parameter.
1202 * These parameters will be configured into the register of the NIC.
1203 * Please refer to the corresponding data sheet for proper value.
1205 struct rte_eth_pfc_conf {
1206 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
1207 uint8_t priority; /**< VLAN User Priority. */
1213 enum rte_eth_tunnel_type {
1214 RTE_TUNNEL_TYPE_NONE = 0,
1215 RTE_TUNNEL_TYPE_VXLAN,
1216 RTE_TUNNEL_TYPE_GENEVE,
1217 RTE_TUNNEL_TYPE_TEREDO,
1218 RTE_TUNNEL_TYPE_NVGRE,
1219 RTE_TUNNEL_TYPE_IP_IN_GRE,
1220 RTE_L2_TUNNEL_TYPE_E_TAG,
1221 RTE_TUNNEL_TYPE_VXLAN_GPE,
1222 RTE_TUNNEL_TYPE_MAX,
1225 /* Deprecated API file for rte_eth_dev_filter_* functions */
1226 #include "rte_eth_ctrl.h"
1229 * Memory space that can be configured to store Flow Director filters
1230 * in the board memory.
1232 enum rte_fdir_pballoc_type {
1233 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
1234 RTE_FDIR_PBALLOC_128K, /**< 128k. */
1235 RTE_FDIR_PBALLOC_256K, /**< 256k. */
1239 * Select report mode of FDIR hash information in RX descriptors.
1241 enum rte_fdir_status_mode {
1242 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
1243 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
1244 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
1248 * A structure used to configure the Flow Director (FDIR) feature
1249 * of an Ethernet port.
1251 * If mode is RTE_FDIR_MODE_NONE, the pballoc value is ignored.
1253 struct rte_fdir_conf {
1254 enum rte_fdir_mode mode; /**< Flow Director mode. */
1255 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
1256 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
1257 /** RX queue of packets matching a "drop" filter in perfect mode. */
1259 struct rte_eth_fdir_masks mask;
1260 struct rte_eth_fdir_flex_conf flex_conf;
1261 /**< Flex payload configuration. */
1265 * UDP tunneling configuration.
1266 * Used to config the UDP port for a type of tunnel.
1267 * NICs need the UDP port to identify the tunnel type.
1268 * Normally a type of tunnel has a default UDP port, this structure can be used
1269 * in case if the users want to change or support more UDP port.
1271 struct rte_eth_udp_tunnel {
1272 uint16_t udp_port; /**< UDP port used for the tunnel. */
1273 uint8_t prot_type; /**< Tunnel type. Defined in rte_eth_tunnel_type. */
1277 * A structure used to enable/disable specific device interrupts.
1279 struct rte_intr_conf {
1280 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
1282 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
1284 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
1289 * A structure used to configure an Ethernet port.
1290 * Depending upon the RX multi-queue mode, extra advanced
1291 * configuration settings may be needed.
1293 struct rte_eth_conf {
1294 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
1295 used. ETH_LINK_SPEED_FIXED disables link
1296 autonegotiation, and a unique speed shall be
1297 set. Otherwise, the bitmap defines the set of
1298 speeds to be advertised. If the special value
1299 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
1300 supported are advertised. */
1301 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
1302 struct rte_eth_txmode txmode; /**< Port TX configuration. */
1303 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
1304 is 0, meaning the loopback mode is disabled.
1305 Read the datasheet of given ethernet controller
1306 for details. The possible values of this field
1307 are defined in implementation of each driver. */
1309 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
1310 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
1311 /**< Port vmdq+dcb configuration. */
1312 struct rte_eth_dcb_rx_conf dcb_rx_conf;
1313 /**< Port dcb RX configuration. */
1314 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
1315 /**< Port vmdq RX configuration. */
1316 } rx_adv_conf; /**< Port RX filtering configuration. */
1318 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
1319 /**< Port vmdq+dcb TX configuration. */
1320 struct rte_eth_dcb_tx_conf dcb_tx_conf;
1321 /**< Port dcb TX configuration. */
1322 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
1323 /**< Port vmdq TX configuration. */
1324 } tx_adv_conf; /**< Port TX DCB configuration (union). */
1325 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
1326 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
1327 uint32_t dcb_capability_en;
1328 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. DEPRECATED */
1329 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
1333 * RX offload capabilities of a device.
1335 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
1336 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
1337 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
1338 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
1339 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
1340 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
1341 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
1342 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
1343 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
1344 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
1345 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
1346 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
1347 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
1349 * Timestamp is set by the driver in RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1350 * and RTE_MBUF_DYNFLAG_RX_TIMESTAMP_NAME is set in ol_flags.
1351 * The mbuf field and flag are registered when the offload is configured.
1353 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
1354 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
1355 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
1356 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
1357 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
1358 #define DEV_RX_OFFLOAD_RSS_HASH 0x00080000
1359 #define RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT 0x00100000
1361 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
1362 DEV_RX_OFFLOAD_UDP_CKSUM | \
1363 DEV_RX_OFFLOAD_TCP_CKSUM)
1364 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
1365 DEV_RX_OFFLOAD_VLAN_FILTER | \
1366 DEV_RX_OFFLOAD_VLAN_EXTEND | \
1367 DEV_RX_OFFLOAD_QINQ_STRIP)
1370 * If new Rx offload capabilities are defined, they also must be
1371 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
1375 * TX offload capabilities of a device.
1377 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
1378 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
1379 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
1380 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
1381 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
1382 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
1383 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
1384 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
1385 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
1386 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
1387 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
1388 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
1389 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
1390 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
1391 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
1392 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
1393 * tx queue without SW lock.
1395 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
1396 /**< Device supports multi segment send. */
1397 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
1398 /**< Device supports optimization for fast release of mbufs.
1399 * When set application must guarantee that per-queue all mbufs comes from
1400 * the same mempool and has refcnt = 1.
1402 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
1404 * Device supports generic UDP tunneled packet TSO.
1405 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
1408 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
1410 * Device supports generic IP tunneled packet TSO.
1411 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
1414 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
1415 /** Device supports outer UDP checksum */
1416 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
1418 * Device sends on time read from RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1419 * if RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME is set in ol_flags.
1420 * The mbuf field and flag are registered when the offload is configured.
1422 #define DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP 0x00200000
1424 * If new Tx offload capabilities are defined, they also must be
1425 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1428 /**@{@name Device capabilities
1429 * Non-offload capabilities reported in rte_eth_dev_info.dev_capa.
1431 /** Device supports Rx queue setup after device started. */
1432 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
1433 /** Device supports Tx queue setup after device started. */
1434 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
1438 * Fallback default preferred Rx/Tx port parameters.
1439 * These are used if an application requests default parameters
1440 * but the PMD does not provide preferred values.
1442 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1443 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1444 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1445 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1448 * Preferred Rx/Tx port parameters.
1449 * There are separate instances of this structure for transmission
1450 * and reception respectively.
1452 struct rte_eth_dev_portconf {
1453 uint16_t burst_size; /**< Device-preferred burst size */
1454 uint16_t ring_size; /**< Device-preferred size of queue rings */
1455 uint16_t nb_queues; /**< Device-preferred number of queues */
1459 * Default values for switch domain id when ethdev does not support switch
1460 * domain definitions.
1462 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (UINT16_MAX)
1465 * Ethernet device associated switch information
1467 struct rte_eth_switch_info {
1468 const char *name; /**< switch name */
1469 uint16_t domain_id; /**< switch domain id */
1472 * mapping to the devices physical switch port as enumerated from the
1473 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1474 * device this may correspond to the VF_ID of each virtual function,
1475 * but each driver should explicitly define the mapping of switch
1476 * port identifier to that physical interconnect/switch
1482 * @b EXPERIMENTAL: this structure may change without prior notice.
1484 * Ethernet device Rx buffer segmentation capabilities.
1486 struct rte_eth_rxseg_capa {
1488 uint32_t multi_pools:1; /**< Supports receiving to multiple pools.*/
1489 uint32_t offset_allowed:1; /**< Supports buffer offsets. */
1490 uint32_t offset_align_log2:4; /**< Required offset alignment. */
1491 uint16_t max_nseg; /**< Maximum amount of segments to split. */
1492 uint16_t reserved; /**< Reserved field. */
1496 * Ethernet device information
1500 * A structure used to retrieve the contextual information of
1501 * an Ethernet device, such as the controlling driver of the
1504 struct rte_eth_dev_info {
1505 struct rte_device *device; /** Generic device information */
1506 const char *driver_name; /**< Device Driver name. */
1507 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1508 Use if_indextoname() to translate into an interface name. */
1509 uint16_t min_mtu; /**< Minimum MTU allowed */
1510 uint16_t max_mtu; /**< Maximum MTU allowed */
1511 const uint32_t *dev_flags; /**< Device flags */
1512 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1513 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1514 /** Maximum configurable size of LRO aggregated packet. */
1515 uint32_t max_lro_pkt_size;
1516 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1517 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1518 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1519 uint32_t max_hash_mac_addrs;
1520 /** Maximum number of hash MAC addresses for MTA and UTA. */
1521 uint16_t max_vfs; /**< Maximum number of VFs. */
1522 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1523 struct rte_eth_rxseg_capa rx_seg_capa; /**< Segmentation capability.*/
1524 uint64_t rx_offload_capa;
1525 /**< All RX offload capabilities including all per-queue ones */
1526 uint64_t tx_offload_capa;
1527 /**< All TX offload capabilities including all per-queue ones */
1528 uint64_t rx_queue_offload_capa;
1529 /**< Device per-queue RX offload capabilities. */
1530 uint64_t tx_queue_offload_capa;
1531 /**< Device per-queue TX offload capabilities. */
1533 /**< Device redirection table size, the total number of entries. */
1534 uint8_t hash_key_size; /**< Hash key size in bytes */
1535 /** Bit mask of RSS offloads, the bit offset also means flow type */
1536 uint64_t flow_type_rss_offloads;
1537 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1538 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1539 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1540 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1541 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1542 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1543 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1544 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1545 /** Configured number of rx/tx queues */
1546 uint16_t nb_rx_queues; /**< Number of RX queues. */
1547 uint16_t nb_tx_queues; /**< Number of TX queues. */
1548 /** Rx parameter recommendations */
1549 struct rte_eth_dev_portconf default_rxportconf;
1550 /** Tx parameter recommendations */
1551 struct rte_eth_dev_portconf default_txportconf;
1552 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1555 * Switching information for ports on a device with a
1556 * embedded managed interconnect/switch.
1558 struct rte_eth_switch_info switch_info;
1560 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1561 void *reserved_ptrs[2]; /**< Reserved for future fields */
1565 * Ethernet device RX queue information structure.
1566 * Used to retrieve information about configured queue.
1568 struct rte_eth_rxq_info {
1569 struct rte_mempool *mp; /**< mempool used by that queue. */
1570 struct rte_eth_rxconf conf; /**< queue config parameters. */
1571 uint8_t scattered_rx; /**< scattered packets RX supported. */
1572 uint16_t nb_desc; /**< configured number of RXDs. */
1573 uint16_t rx_buf_size; /**< hardware receive buffer size. */
1574 } __rte_cache_min_aligned;
1577 * Ethernet device TX queue information structure.
1578 * Used to retrieve information about configured queue.
1580 struct rte_eth_txq_info {
1581 struct rte_eth_txconf conf; /**< queue config parameters. */
1582 uint16_t nb_desc; /**< configured number of TXDs. */
1583 } __rte_cache_min_aligned;
1585 /* Generic Burst mode flag definition, values can be ORed. */
1588 * If the queues have different burst mode description, this bit will be set
1589 * by PMD, then the application can iterate to retrieve burst description for
1592 #define RTE_ETH_BURST_FLAG_PER_QUEUE (1ULL << 0)
1595 * Ethernet device RX/TX queue packet burst mode information structure.
1596 * Used to retrieve information about packet burst mode setting.
1598 struct rte_eth_burst_mode {
1599 uint64_t flags; /**< The ORed values of RTE_ETH_BURST_FLAG_xxx */
1601 #define RTE_ETH_BURST_MODE_INFO_SIZE 1024 /**< Maximum size for information */
1602 char info[RTE_ETH_BURST_MODE_INFO_SIZE]; /**< burst mode information */
1605 /** Maximum name length for extended statistics counters */
1606 #define RTE_ETH_XSTATS_NAME_SIZE 64
1609 * An Ethernet device extended statistic structure
1611 * This structure is used by rte_eth_xstats_get() to provide
1612 * statistics that are not provided in the generic *rte_eth_stats*
1614 * It maps a name id, corresponding to an index in the array returned
1615 * by rte_eth_xstats_get_names(), to a statistic value.
1617 struct rte_eth_xstat {
1618 uint64_t id; /**< The index in xstats name array. */
1619 uint64_t value; /**< The statistic counter value. */
1623 * A name element for extended statistics.
1625 * An array of this structure is returned by rte_eth_xstats_get_names().
1626 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1627 * structure references these names by their array index.
1629 * The xstats should follow a common naming scheme.
1630 * Some names are standardized in rte_stats_strings.
1632 * - rx_missed_errors
1634 * - tx_size_128_to_255_packets
1636 struct rte_eth_xstat_name {
1637 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1640 #define ETH_DCB_NUM_TCS 8
1641 #define ETH_MAX_VMDQ_POOL 64
1644 * A structure used to get the information of queue and
1645 * TC mapping on both TX and RX paths.
1647 struct rte_eth_dcb_tc_queue_mapping {
1648 /** rx queues assigned to tc per Pool */
1652 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1653 /** rx queues assigned to tc per Pool */
1657 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1661 * A structure used to get the information of DCB.
1662 * It includes TC UP mapping and queue TC mapping.
1664 struct rte_eth_dcb_info {
1665 uint8_t nb_tcs; /**< number of TCs */
1666 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1667 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1668 /** rx queues assigned to tc */
1669 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1673 * This enum indicates the possible Forward Error Correction (FEC) modes
1674 * of an ethdev port.
1676 enum rte_eth_fec_mode {
1677 RTE_ETH_FEC_NOFEC = 0, /**< FEC is off */
1678 RTE_ETH_FEC_AUTO, /**< FEC autonegotiation modes */
1679 RTE_ETH_FEC_BASER, /**< FEC using common algorithm */
1680 RTE_ETH_FEC_RS, /**< FEC using RS algorithm */
1683 /* Translate from FEC mode to FEC capa */
1684 #define RTE_ETH_FEC_MODE_TO_CAPA(x) (1U << (x))
1686 /* This macro indicates FEC capa mask */
1687 #define RTE_ETH_FEC_MODE_CAPA_MASK(x) (1U << (RTE_ETH_FEC_ ## x))
1689 /* A structure used to get capabilities per link speed */
1690 struct rte_eth_fec_capa {
1691 uint32_t speed; /**< Link speed (see ETH_SPEED_NUM_*) */
1692 uint32_t capa; /**< FEC capabilities bitmask */
1695 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1697 /* Macros to check for valid port */
1698 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1699 if (!rte_eth_dev_is_valid_port(port_id)) { \
1700 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1705 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1706 if (!rte_eth_dev_is_valid_port(port_id)) { \
1707 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1713 * l2 tunnel configuration.
1716 /**< l2 tunnel enable mask */
1717 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1718 /**< l2 tunnel insertion mask */
1719 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1720 /**< l2 tunnel stripping mask */
1721 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1722 /**< l2 tunnel forwarding mask */
1723 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1726 * Function type used for RX packet processing packet callbacks.
1728 * The callback function is called on RX with a burst of packets that have
1729 * been received on the given port and queue.
1732 * The Ethernet port on which RX is being performed.
1734 * The queue on the Ethernet port which is being used to receive the packets.
1736 * The burst of packets that have just been received.
1738 * The number of packets in the burst pointed to by "pkts".
1740 * The max number of packets that can be stored in the "pkts" array.
1742 * The arbitrary user parameter passed in by the application when the callback
1743 * was originally configured.
1745 * The number of packets returned to the user.
1747 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1748 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1752 * Function type used for TX packet processing packet callbacks.
1754 * The callback function is called on TX with a burst of packets immediately
1755 * before the packets are put onto the hardware queue for transmission.
1758 * The Ethernet port on which TX is being performed.
1760 * The queue on the Ethernet port which is being used to transmit the packets.
1762 * The burst of packets that are about to be transmitted.
1764 * The number of packets in the burst pointed to by "pkts".
1766 * The arbitrary user parameter passed in by the application when the callback
1767 * was originally configured.
1769 * The number of packets to be written to the NIC.
1771 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1772 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1775 * Possible states of an ethdev port.
1777 enum rte_eth_dev_state {
1778 /** Device is unused before being probed. */
1779 RTE_ETH_DEV_UNUSED = 0,
1780 /** Device is attached when allocated in probing. */
1781 RTE_ETH_DEV_ATTACHED,
1782 /** Device is in removed state when plug-out is detected. */
1783 RTE_ETH_DEV_REMOVED,
1786 struct rte_eth_dev_sriov {
1787 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1788 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1789 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1790 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1792 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1794 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1796 #define RTE_ETH_DEV_NO_OWNER 0
1798 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1800 struct rte_eth_dev_owner {
1801 uint64_t id; /**< The owner unique identifier. */
1802 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1805 /** PMD supports thread-safe flow operations */
1806 #define RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE 0x0001
1807 /** Device supports link state interrupt */
1808 #define RTE_ETH_DEV_INTR_LSC 0x0002
1809 /** Device is a bonded slave */
1810 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1811 /** Device supports device removal interrupt */
1812 #define RTE_ETH_DEV_INTR_RMV 0x0008
1813 /** Device is port representor */
1814 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1815 /** Device does not support MAC change after started */
1816 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1818 * Queue xstats filled automatically by ethdev layer.
1819 * PMDs filling the queue xstats themselves should not set this flag
1821 #define RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS 0x0040
1824 * Iterates over valid ethdev ports owned by a specific owner.
1827 * The id of the next possible valid owned port.
1829 * The owner identifier.
1830 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1832 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1834 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1835 const uint64_t owner_id);
1838 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1840 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1841 for (p = rte_eth_find_next_owned_by(0, o); \
1842 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1843 p = rte_eth_find_next_owned_by(p + 1, o))
1846 * Iterates over valid ethdev ports.
1849 * The id of the next possible valid port.
1851 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1853 uint16_t rte_eth_find_next(uint16_t port_id);
1856 * Macro to iterate over all enabled and ownerless ethdev ports.
1858 #define RTE_ETH_FOREACH_DEV(p) \
1859 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1863 * @b EXPERIMENTAL: this API may change without prior notice.
1865 * Iterates over ethdev ports of a specified device.
1867 * @param port_id_start
1868 * The id of the next possible valid port.
1870 * The generic device behind the ports to iterate.
1872 * Next port id of the device, possibly port_id_start,
1873 * RTE_MAX_ETHPORTS if there is none.
1877 rte_eth_find_next_of(uint16_t port_id_start,
1878 const struct rte_device *parent);
1881 * Macro to iterate over all ethdev ports of a specified device.
1884 * The id of the matching port being iterated.
1886 * The rte_device pointer matching the iterated ports.
1888 #define RTE_ETH_FOREACH_DEV_OF(port_id, parent) \
1889 for (port_id = rte_eth_find_next_of(0, parent); \
1890 port_id < RTE_MAX_ETHPORTS; \
1891 port_id = rte_eth_find_next_of(port_id + 1, parent))
1895 * @b EXPERIMENTAL: this API may change without prior notice.
1897 * Iterates over sibling ethdev ports (i.e. sharing the same rte_device).
1899 * @param port_id_start
1900 * The id of the next possible valid sibling port.
1901 * @param ref_port_id
1902 * The id of a reference port to compare rte_device with.
1904 * Next sibling port id, possibly port_id_start or ref_port_id itself,
1905 * RTE_MAX_ETHPORTS if there is none.
1909 rte_eth_find_next_sibling(uint16_t port_id_start, uint16_t ref_port_id);
1912 * Macro to iterate over all ethdev ports sharing the same rte_device
1913 * as the specified port.
1914 * Note: the specified reference port is part of the loop iterations.
1917 * The id of the matching port being iterated.
1918 * @param ref_port_id
1919 * The id of the port being compared.
1921 #define RTE_ETH_FOREACH_DEV_SIBLING(port_id, ref_port_id) \
1922 for (port_id = rte_eth_find_next_sibling(0, ref_port_id); \
1923 port_id < RTE_MAX_ETHPORTS; \
1924 port_id = rte_eth_find_next_sibling(port_id + 1, ref_port_id))
1928 * @b EXPERIMENTAL: this API may change without prior notice.
1930 * Get a new unique owner identifier.
1931 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1932 * to avoid multiple management of device by different entities.
1935 * Owner identifier pointer.
1937 * Negative errno value on error, 0 on success.
1940 int rte_eth_dev_owner_new(uint64_t *owner_id);
1944 * @b EXPERIMENTAL: this API may change without prior notice.
1946 * Set an Ethernet device owner.
1949 * The identifier of the port to own.
1951 * The owner pointer.
1953 * Negative errno value on error, 0 on success.
1956 int rte_eth_dev_owner_set(const uint16_t port_id,
1957 const struct rte_eth_dev_owner *owner);
1961 * @b EXPERIMENTAL: this API may change without prior notice.
1963 * Unset Ethernet device owner to make the device ownerless.
1966 * The identifier of port to make ownerless.
1968 * The owner identifier.
1970 * 0 on success, negative errno value on error.
1973 int rte_eth_dev_owner_unset(const uint16_t port_id,
1974 const uint64_t owner_id);
1978 * @b EXPERIMENTAL: this API may change without prior notice.
1980 * Remove owner from all Ethernet devices owned by a specific owner.
1983 * The owner identifier.
1985 * 0 on success, negative errno value on error.
1988 int rte_eth_dev_owner_delete(const uint64_t owner_id);
1992 * @b EXPERIMENTAL: this API may change without prior notice.
1994 * Get the owner of an Ethernet device.
1997 * The port identifier.
1999 * The owner structure pointer to fill.
2001 * 0 on success, negative errno value on error..
2004 int rte_eth_dev_owner_get(const uint16_t port_id,
2005 struct rte_eth_dev_owner *owner);
2008 * Get the number of ports which are usable for the application.
2010 * These devices must be iterated by using the macro
2011 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
2012 * to deal with non-contiguous ranges of devices.
2015 * The count of available Ethernet devices.
2017 uint16_t rte_eth_dev_count_avail(void);
2020 * Get the total number of ports which are allocated.
2022 * Some devices may not be available for the application.
2025 * The total count of Ethernet devices.
2027 uint16_t rte_eth_dev_count_total(void);
2030 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
2031 * the bitmap link_speeds of the struct rte_eth_conf
2034 * Numerical speed value in Mbps
2036 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
2038 * 0 if the speed cannot be mapped
2040 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
2043 * Get DEV_RX_OFFLOAD_* flag name.
2048 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2050 const char *rte_eth_dev_rx_offload_name(uint64_t offload);
2053 * Get DEV_TX_OFFLOAD_* flag name.
2058 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2060 const char *rte_eth_dev_tx_offload_name(uint64_t offload);
2063 * Configure an Ethernet device.
2064 * This function must be invoked first before any other function in the
2065 * Ethernet API. This function can also be re-invoked when a device is in the
2069 * The port identifier of the Ethernet device to configure.
2070 * @param nb_rx_queue
2071 * The number of receive queues to set up for the Ethernet device.
2072 * @param nb_tx_queue
2073 * The number of transmit queues to set up for the Ethernet device.
2075 * The pointer to the configuration data to be used for the Ethernet device.
2076 * The *rte_eth_conf* structure includes:
2077 * - the hardware offload features to activate, with dedicated fields for
2078 * each statically configurable offload hardware feature provided by
2079 * Ethernet devices, such as IP checksum or VLAN tag stripping for
2081 * The Rx offload bitfield API is obsolete and will be deprecated.
2082 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
2083 * structure and use offloads field to set per-port offloads instead.
2084 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
2085 * the [rt]x_offload_capa returned from rte_eth_dev_info_get().
2086 * Any type of device supported offloading set in the input argument
2087 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
2088 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
2089 * - the Receive Side Scaling (RSS) configuration when using multiple RX
2090 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
2091 * must be within the flow_type_rss_offloads provided by drivers via
2092 * rte_eth_dev_info_get() API.
2094 * Embedding all configuration information in a single data structure
2095 * is the more flexible method that allows the addition of new features
2096 * without changing the syntax of the API.
2098 * - 0: Success, device configured.
2099 * - <0: Error code returned by the driver configuration function.
2101 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
2102 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
2106 * @b EXPERIMENTAL: this API may change without prior notice.
2108 * Check if an Ethernet device was physically removed.
2111 * The port identifier of the Ethernet device.
2113 * 1 when the Ethernet device is removed, otherwise 0.
2117 rte_eth_dev_is_removed(uint16_t port_id);
2120 * Allocate and set up a receive queue for an Ethernet device.
2122 * The function allocates a contiguous block of memory for *nb_rx_desc*
2123 * receive descriptors from a memory zone associated with *socket_id*
2124 * and initializes each receive descriptor with a network buffer allocated
2125 * from the memory pool *mb_pool*.
2128 * The port identifier of the Ethernet device.
2129 * @param rx_queue_id
2130 * The index of the receive queue to set up.
2131 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2132 * to rte_eth_dev_configure().
2134 * The number of receive descriptors to allocate for the receive ring.
2136 * The *socket_id* argument is the socket identifier in case of NUMA.
2137 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2138 * the DMA memory allocated for the receive descriptors of the ring.
2140 * The pointer to the configuration data to be used for the receive queue.
2141 * NULL value is allowed, in which case default RX configuration
2143 * The *rx_conf* structure contains an *rx_thresh* structure with the values
2144 * of the Prefetch, Host, and Write-Back threshold registers of the receive
2146 * In addition it contains the hardware offloads features to activate using
2147 * the DEV_RX_OFFLOAD_* flags.
2148 * If an offloading set in rx_conf->offloads
2149 * hasn't been set in the input argument eth_conf->rxmode.offloads
2150 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2151 * per-queue type and it is enabled for the queue.
2152 * No need to repeat any bit in rx_conf->offloads which has already been
2153 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2154 * at port level can't be disabled at queue level.
2155 * The configuration structure also contains the pointer to the array
2156 * of the receiving buffer segment descriptions, see rx_seg and rx_nseg
2157 * fields, this extended configuration might be used by split offloads like
2158 * RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT. If mp_pool is not NULL,
2159 * the extended configuration fields must be set to NULL and zero.
2161 * The pointer to the memory pool from which to allocate *rte_mbuf* network
2162 * memory buffers to populate each descriptor of the receive ring. There are
2163 * two options to provide Rx buffer configuration:
2165 * mb_pool is not NULL, rx_conf.rx_nseg is 0.
2166 * - multiple segments description:
2167 * mb_pool is NULL, rx_conf.rx_seg is not NULL, rx_conf.rx_nseg is not 0.
2168 * Taken only if flag RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT is set in offloads.
2171 * - 0: Success, receive queue correctly set up.
2172 * - -EIO: if device is removed.
2173 * - -ENODEV: if *port_id* is invalid.
2174 * - -EINVAL: The memory pool pointer is null or the size of network buffers
2175 * which can be allocated from this memory pool does not fit the various
2176 * buffer sizes allowed by the device controller.
2177 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
2178 * allocate network memory buffers from the memory pool when
2179 * initializing receive descriptors.
2181 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
2182 uint16_t nb_rx_desc, unsigned int socket_id,
2183 const struct rte_eth_rxconf *rx_conf,
2184 struct rte_mempool *mb_pool);
2188 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2190 * Allocate and set up a hairpin receive queue for an Ethernet device.
2192 * The function set up the selected queue to be used in hairpin.
2195 * The port identifier of the Ethernet device.
2196 * @param rx_queue_id
2197 * The index of the receive queue to set up.
2198 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2199 * to rte_eth_dev_configure().
2201 * The number of receive descriptors to allocate for the receive ring.
2202 * 0 means the PMD will use default value.
2204 * The pointer to the hairpin configuration.
2207 * - (0) if successful.
2208 * - (-ENODEV) if *port_id* is invalid.
2209 * - (-ENOTSUP) if hardware doesn't support.
2210 * - (-EINVAL) if bad parameter.
2211 * - (-ENOMEM) if unable to allocate the resources.
2214 int rte_eth_rx_hairpin_queue_setup
2215 (uint16_t port_id, uint16_t rx_queue_id, uint16_t nb_rx_desc,
2216 const struct rte_eth_hairpin_conf *conf);
2219 * Allocate and set up a transmit queue for an Ethernet device.
2222 * The port identifier of the Ethernet device.
2223 * @param tx_queue_id
2224 * The index of the transmit queue to set up.
2225 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2226 * to rte_eth_dev_configure().
2228 * The number of transmit descriptors to allocate for the transmit ring.
2230 * The *socket_id* argument is the socket identifier in case of NUMA.
2231 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2232 * the DMA memory allocated for the transmit descriptors of the ring.
2234 * The pointer to the configuration data to be used for the transmit queue.
2235 * NULL value is allowed, in which case default TX configuration
2237 * The *tx_conf* structure contains the following data:
2238 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
2239 * Write-Back threshold registers of the transmit ring.
2240 * When setting Write-Back threshold to the value greater then zero,
2241 * *tx_rs_thresh* value should be explicitly set to one.
2242 * - The *tx_free_thresh* value indicates the [minimum] number of network
2243 * buffers that must be pending in the transmit ring to trigger their
2244 * [implicit] freeing by the driver transmit function.
2245 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
2246 * descriptors that must be pending in the transmit ring before setting the
2247 * RS bit on a descriptor by the driver transmit function.
2248 * The *tx_rs_thresh* value should be less or equal then
2249 * *tx_free_thresh* value, and both of them should be less then
2251 * - The *offloads* member contains Tx offloads to be enabled.
2252 * If an offloading set in tx_conf->offloads
2253 * hasn't been set in the input argument eth_conf->txmode.offloads
2254 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2255 * per-queue type and it is enabled for the queue.
2256 * No need to repeat any bit in tx_conf->offloads which has already been
2257 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2258 * at port level can't be disabled at queue level.
2260 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
2261 * the transmit function to use default values.
2263 * - 0: Success, the transmit queue is correctly set up.
2264 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
2266 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
2267 uint16_t nb_tx_desc, unsigned int socket_id,
2268 const struct rte_eth_txconf *tx_conf);
2272 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2274 * Allocate and set up a transmit hairpin queue for an Ethernet device.
2277 * The port identifier of the Ethernet device.
2278 * @param tx_queue_id
2279 * The index of the transmit queue to set up.
2280 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2281 * to rte_eth_dev_configure().
2283 * The number of transmit descriptors to allocate for the transmit ring.
2284 * 0 to set default PMD value.
2286 * The hairpin configuration.
2289 * - (0) if successful.
2290 * - (-ENODEV) if *port_id* is invalid.
2291 * - (-ENOTSUP) if hardware doesn't support.
2292 * - (-EINVAL) if bad parameter.
2293 * - (-ENOMEM) if unable to allocate the resources.
2296 int rte_eth_tx_hairpin_queue_setup
2297 (uint16_t port_id, uint16_t tx_queue_id, uint16_t nb_tx_desc,
2298 const struct rte_eth_hairpin_conf *conf);
2302 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2304 * Get all the hairpin peer Rx / Tx ports of the current port.
2305 * The caller should ensure that the array is large enough to save the ports
2309 * The port identifier of the Ethernet device.
2311 * Pointer to the array to store the peer ports list.
2313 * Length of the array to store the port identifiers.
2315 * Current port to peer port direction
2316 * positive - current used as Tx to get all peer Rx ports.
2317 * zero - current used as Rx to get all peer Tx ports.
2320 * - (0 or positive) actual peer ports number.
2321 * - (-EINVAL) if bad parameter.
2322 * - (-ENODEV) if *port_id* invalid
2323 * - (-ENOTSUP) if hardware doesn't support.
2324 * - Others detailed errors from PMD drivers.
2327 int rte_eth_hairpin_get_peer_ports(uint16_t port_id, uint16_t *peer_ports,
2328 size_t len, uint32_t direction);
2332 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2334 * Bind all hairpin Tx queues of one port to the Rx queues of the peer port.
2335 * It is only allowed to call this function after all hairpin queues are
2336 * configured properly and the devices are in started state.
2339 * The identifier of the Tx port.
2341 * The identifier of peer Rx port.
2342 * RTE_MAX_ETHPORTS is allowed for the traversal of all devices.
2343 * Rx port ID could have the same value as Tx port ID.
2346 * - (0) if successful.
2347 * - (-ENODEV) if Tx port ID is invalid.
2348 * - (-EBUSY) if device is not in started state.
2349 * - (-ENOTSUP) if hardware doesn't support.
2350 * - Others detailed errors from PMD drivers.
2353 int rte_eth_hairpin_bind(uint16_t tx_port, uint16_t rx_port);
2357 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2359 * Unbind all hairpin Tx queues of one port from the Rx queues of the peer port.
2360 * This should be called before closing the Tx or Rx devices, if the bind
2361 * function is called before.
2362 * After unbinding the hairpin ports pair, it is allowed to bind them again.
2363 * Changing queues configuration should be after stopping the device(s).
2366 * The identifier of the Tx port.
2368 * The identifier of peer Rx port.
2369 * RTE_MAX_ETHPORTS is allowed for traversal of all devices.
2370 * Rx port ID could have the same value as Tx port ID.
2373 * - (0) if successful.
2374 * - (-ENODEV) if Tx port ID is invalid.
2375 * - (-EBUSY) if device is in stopped state.
2376 * - (-ENOTSUP) if hardware doesn't support.
2377 * - Others detailed errors from PMD drivers.
2380 int rte_eth_hairpin_unbind(uint16_t tx_port, uint16_t rx_port);
2383 * Return the NUMA socket to which an Ethernet device is connected
2386 * The port identifier of the Ethernet device
2388 * The NUMA socket id to which the Ethernet device is connected or
2389 * a default of zero if the socket could not be determined.
2390 * -1 is returned is the port_id value is out of range.
2392 int rte_eth_dev_socket_id(uint16_t port_id);
2395 * Check if port_id of device is attached
2398 * The port identifier of the Ethernet device
2400 * - 0 if port is out of range or not attached
2401 * - 1 if device is attached
2403 int rte_eth_dev_is_valid_port(uint16_t port_id);
2406 * Start specified RX queue of a port. It is used when rx_deferred_start
2407 * flag of the specified queue is true.
2410 * The port identifier of the Ethernet device
2411 * @param rx_queue_id
2412 * The index of the rx queue to update the ring.
2413 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2414 * to rte_eth_dev_configure().
2416 * - 0: Success, the receive queue is started.
2417 * - -ENODEV: if *port_id* is invalid.
2418 * - -EINVAL: The queue_id out of range or belong to hairpin.
2419 * - -EIO: if device is removed.
2420 * - -ENOTSUP: The function not supported in PMD driver.
2422 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
2425 * Stop specified RX queue of a port
2428 * The port identifier of the Ethernet device
2429 * @param rx_queue_id
2430 * The index of the rx queue to update the ring.
2431 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2432 * to rte_eth_dev_configure().
2434 * - 0: Success, the receive queue is stopped.
2435 * - -ENODEV: if *port_id* is invalid.
2436 * - -EINVAL: The queue_id out of range or belong to hairpin.
2437 * - -EIO: if device is removed.
2438 * - -ENOTSUP: The function not supported in PMD driver.
2440 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
2443 * Start TX for specified queue of a port. It is used when tx_deferred_start
2444 * flag of the specified queue is true.
2447 * The port identifier of the Ethernet device
2448 * @param tx_queue_id
2449 * The index of the tx queue to update the ring.
2450 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2451 * to rte_eth_dev_configure().
2453 * - 0: Success, the transmit queue is started.
2454 * - -ENODEV: if *port_id* is invalid.
2455 * - -EINVAL: The queue_id out of range or belong to hairpin.
2456 * - -EIO: if device is removed.
2457 * - -ENOTSUP: The function not supported in PMD driver.
2459 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
2462 * Stop specified TX queue of a port
2465 * The port identifier of the Ethernet device
2466 * @param tx_queue_id
2467 * The index of the tx queue to update the ring.
2468 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2469 * to rte_eth_dev_configure().
2471 * - 0: Success, the transmit queue is stopped.
2472 * - -ENODEV: if *port_id* is invalid.
2473 * - -EINVAL: The queue_id out of range or belong to hairpin.
2474 * - -EIO: if device is removed.
2475 * - -ENOTSUP: The function not supported in PMD driver.
2477 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
2480 * Start an Ethernet device.
2482 * The device start step is the last one and consists of setting the configured
2483 * offload features and in starting the transmit and the receive units of the
2486 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
2487 * PMD port start callback function is invoked.
2489 * On success, all basic functions exported by the Ethernet API (link status,
2490 * receive/transmit, and so on) can be invoked.
2493 * The port identifier of the Ethernet device.
2495 * - 0: Success, Ethernet device started.
2496 * - <0: Error code of the driver device start function.
2498 int rte_eth_dev_start(uint16_t port_id);
2501 * Stop an Ethernet device. The device can be restarted with a call to
2502 * rte_eth_dev_start()
2505 * The port identifier of the Ethernet device.
2507 * - 0: Success, Ethernet device stopped.
2508 * - <0: Error code of the driver device stop function.
2510 int rte_eth_dev_stop(uint16_t port_id);
2513 * Link up an Ethernet device.
2515 * Set device link up will re-enable the device rx/tx
2516 * functionality after it is previously set device linked down.
2519 * The port identifier of the Ethernet device.
2521 * - 0: Success, Ethernet device linked up.
2522 * - <0: Error code of the driver device link up function.
2524 int rte_eth_dev_set_link_up(uint16_t port_id);
2527 * Link down an Ethernet device.
2528 * The device rx/tx functionality will be disabled if success,
2529 * and it can be re-enabled with a call to
2530 * rte_eth_dev_set_link_up()
2533 * The port identifier of the Ethernet device.
2535 int rte_eth_dev_set_link_down(uint16_t port_id);
2538 * Close a stopped Ethernet device. The device cannot be restarted!
2539 * The function frees all port resources.
2542 * The port identifier of the Ethernet device.
2544 * - Zero if the port is closed successfully.
2545 * - Negative if something went wrong.
2547 int rte_eth_dev_close(uint16_t port_id);
2550 * Reset a Ethernet device and keep its port id.
2552 * When a port has to be reset passively, the DPDK application can invoke
2553 * this function. For example when a PF is reset, all its VFs should also
2554 * be reset. Normally a DPDK application can invoke this function when
2555 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
2556 * a port reset in other circumstances.
2558 * When this function is called, it first stops the port and then calls the
2559 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
2560 * state, in which no Tx and Rx queues are setup, as if the port has been
2561 * reset and not started. The port keeps the port id it had before the
2564 * After calling rte_eth_dev_reset( ), the application should use
2565 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
2566 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
2567 * to reconfigure the device as appropriate.
2569 * Note: To avoid unexpected behavior, the application should stop calling
2570 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
2571 * safety, all these controlling functions should be called from the same
2575 * The port identifier of the Ethernet device.
2578 * - (0) if successful.
2579 * - (-ENODEV) if *port_id* is invalid.
2580 * - (-ENOTSUP) if hardware doesn't support this function.
2581 * - (-EPERM) if not ran from the primary process.
2582 * - (-EIO) if re-initialisation failed or device is removed.
2583 * - (-ENOMEM) if the reset failed due to OOM.
2584 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
2586 int rte_eth_dev_reset(uint16_t port_id);
2589 * Enable receipt in promiscuous mode for an Ethernet device.
2592 * The port identifier of the Ethernet device.
2594 * - (0) if successful.
2595 * - (-ENOTSUP) if support for promiscuous_enable() does not exist
2597 * - (-ENODEV) if *port_id* invalid.
2599 int rte_eth_promiscuous_enable(uint16_t port_id);
2602 * Disable receipt in promiscuous mode for an Ethernet device.
2605 * The port identifier of the Ethernet device.
2607 * - (0) if successful.
2608 * - (-ENOTSUP) if support for promiscuous_disable() does not exist
2610 * - (-ENODEV) if *port_id* invalid.
2612 int rte_eth_promiscuous_disable(uint16_t port_id);
2615 * Return the value of promiscuous mode for an Ethernet device.
2618 * The port identifier of the Ethernet device.
2620 * - (1) if promiscuous is enabled
2621 * - (0) if promiscuous is disabled.
2624 int rte_eth_promiscuous_get(uint16_t port_id);
2627 * Enable the receipt of any multicast frame by an Ethernet device.
2630 * The port identifier of the Ethernet device.
2632 * - (0) if successful.
2633 * - (-ENOTSUP) if support for allmulticast_enable() does not exist
2635 * - (-ENODEV) if *port_id* invalid.
2637 int rte_eth_allmulticast_enable(uint16_t port_id);
2640 * Disable the receipt of all multicast frames by an Ethernet device.
2643 * The port identifier of the Ethernet device.
2645 * - (0) if successful.
2646 * - (-ENOTSUP) if support for allmulticast_disable() does not exist
2648 * - (-ENODEV) if *port_id* invalid.
2650 int rte_eth_allmulticast_disable(uint16_t port_id);
2653 * Return the value of allmulticast mode for an Ethernet device.
2656 * The port identifier of the Ethernet device.
2658 * - (1) if allmulticast is enabled
2659 * - (0) if allmulticast is disabled.
2662 int rte_eth_allmulticast_get(uint16_t port_id);
2665 * Retrieve the link status (up/down), the duplex mode (half/full),
2666 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2668 * It might need to wait up to 9 seconds.
2669 * @see rte_eth_link_get_nowait.
2672 * The port identifier of the Ethernet device.
2674 * Link information written back.
2676 * - (0) if successful.
2677 * - (-ENOTSUP) if the function is not supported in PMD driver.
2678 * - (-ENODEV) if *port_id* invalid.
2680 int rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
2683 * Retrieve the link status (up/down), the duplex mode (half/full),
2684 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2687 * The port identifier of the Ethernet device.
2689 * Link information written back.
2691 * - (0) if successful.
2692 * - (-ENOTSUP) if the function is not supported in PMD driver.
2693 * - (-ENODEV) if *port_id* invalid.
2695 int rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
2699 * @b EXPERIMENTAL: this API may change without prior notice.
2701 * The function converts a link_speed to a string. It handles all special
2702 * values like unknown or none speed.
2705 * link_speed of rte_eth_link struct
2707 * Link speed in textual format. It's pointer to immutable memory.
2708 * No free is required.
2711 const char *rte_eth_link_speed_to_str(uint32_t link_speed);
2715 * @b EXPERIMENTAL: this API may change without prior notice.
2717 * The function converts a rte_eth_link struct representing a link status to
2721 * A pointer to a string to be filled with textual representation of
2722 * device status. At least ETH_LINK_MAX_STR_LEN bytes should be allocated to
2723 * store default link status text.
2725 * Length of available memory at 'str' string.
2727 * Link status returned by rte_eth_link_get function
2729 * Number of bytes written to str array.
2732 int rte_eth_link_to_str(char *str, size_t len,
2733 const struct rte_eth_link *eth_link);
2736 * Retrieve the general I/O statistics of an Ethernet device.
2739 * The port identifier of the Ethernet device.
2741 * A pointer to a structure of type *rte_eth_stats* to be filled with
2742 * the values of device counters for the following set of statistics:
2743 * - *ipackets* with the total of successfully received packets.
2744 * - *opackets* with the total of successfully transmitted packets.
2745 * - *ibytes* with the total of successfully received bytes.
2746 * - *obytes* with the total of successfully transmitted bytes.
2747 * - *ierrors* with the total of erroneous received packets.
2748 * - *oerrors* with the total of failed transmitted packets.
2750 * Zero if successful. Non-zero otherwise.
2752 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
2755 * Reset the general I/O statistics of an Ethernet device.
2758 * The port identifier of the Ethernet device.
2760 * - (0) if device notified to reset stats.
2761 * - (-ENOTSUP) if hardware doesn't support.
2762 * - (-ENODEV) if *port_id* invalid.
2763 * - (<0): Error code of the driver stats reset function.
2765 int rte_eth_stats_reset(uint16_t port_id);
2768 * Retrieve names of extended statistics of an Ethernet device.
2770 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2772 * xstats_names[i].name => xstats[i].value
2774 * And the array index is same with id field of 'struct rte_eth_xstat':
2777 * This assumption makes key-value pair matching less flexible but simpler.
2780 * The port identifier of the Ethernet device.
2781 * @param xstats_names
2782 * An rte_eth_xstat_name array of at least *size* elements to
2783 * be filled. If set to NULL, the function returns the required number
2786 * The size of the xstats_names array (number of elements).
2788 * - A positive value lower or equal to size: success. The return value
2789 * is the number of entries filled in the stats table.
2790 * - A positive value higher than size: error, the given statistics table
2791 * is too small. The return value corresponds to the size that should
2792 * be given to succeed. The entries in the table are not valid and
2793 * shall not be used by the caller.
2794 * - A negative value on error (invalid port id).
2796 int rte_eth_xstats_get_names(uint16_t port_id,
2797 struct rte_eth_xstat_name *xstats_names,
2801 * Retrieve extended statistics of an Ethernet device.
2803 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2805 * xstats_names[i].name => xstats[i].value
2807 * And the array index is same with id field of 'struct rte_eth_xstat':
2810 * This assumption makes key-value pair matching less flexible but simpler.
2813 * The port identifier of the Ethernet device.
2815 * A pointer to a table of structure of type *rte_eth_xstat*
2816 * to be filled with device statistics ids and values.
2817 * This parameter can be set to NULL if n is 0.
2819 * The size of the xstats array (number of elements).
2821 * - A positive value lower or equal to n: success. The return value
2822 * is the number of entries filled in the stats table.
2823 * - A positive value higher than n: error, the given statistics table
2824 * is too small. The return value corresponds to the size that should
2825 * be given to succeed. The entries in the table are not valid and
2826 * shall not be used by the caller.
2827 * - A negative value on error (invalid port id).
2829 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2833 * Retrieve names of extended statistics of an Ethernet device.
2836 * The port identifier of the Ethernet device.
2837 * @param xstats_names
2838 * An rte_eth_xstat_name array of at least *size* elements to
2839 * be filled. If set to NULL, the function returns the required number
2842 * IDs array given by app to retrieve specific statistics
2844 * The size of the xstats_names array (number of elements).
2846 * - A positive value lower or equal to size: success. The return value
2847 * is the number of entries filled in the stats table.
2848 * - A positive value higher than size: error, the given statistics table
2849 * is too small. The return value corresponds to the size that should
2850 * be given to succeed. The entries in the table are not valid and
2851 * shall not be used by the caller.
2852 * - A negative value on error (invalid port id).
2855 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2856 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2860 * Retrieve extended statistics of an Ethernet device.
2863 * The port identifier of the Ethernet device.
2865 * A pointer to an ids array passed by application. This tells which
2866 * statistics values function should retrieve. This parameter
2867 * can be set to NULL if size is 0. In this case function will retrieve
2868 * all available statistics.
2870 * A pointer to a table to be filled with device statistics values.
2872 * The size of the ids array (number of elements).
2874 * - A positive value lower or equal to size: success. The return value
2875 * is the number of entries filled in the stats table.
2876 * - A positive value higher than size: error, the given statistics table
2877 * is too small. The return value corresponds to the size that should
2878 * be given to succeed. The entries in the table are not valid and
2879 * shall not be used by the caller.
2880 * - A negative value on error (invalid port id).
2882 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2883 uint64_t *values, unsigned int size);
2886 * Gets the ID of a statistic from its name.
2888 * This function searches for the statistics using string compares, and
2889 * as such should not be used on the fast-path. For fast-path retrieval of
2890 * specific statistics, store the ID as provided in *id* from this function,
2891 * and pass the ID to rte_eth_xstats_get()
2893 * @param port_id The port to look up statistics from
2894 * @param xstat_name The name of the statistic to return
2895 * @param[out] id A pointer to an app-supplied uint64_t which should be
2896 * set to the ID of the stat if the stat exists.
2899 * -ENODEV for invalid port_id,
2900 * -EIO if device is removed,
2901 * -EINVAL if the xstat_name doesn't exist in port_id
2903 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2907 * Reset extended statistics of an Ethernet device.
2910 * The port identifier of the Ethernet device.
2912 * - (0) if device notified to reset extended stats.
2913 * - (-ENOTSUP) if pmd doesn't support both
2914 * extended stats and basic stats reset.
2915 * - (-ENODEV) if *port_id* invalid.
2916 * - (<0): Error code of the driver xstats reset function.
2918 int rte_eth_xstats_reset(uint16_t port_id);
2921 * Set a mapping for the specified transmit queue to the specified per-queue
2922 * statistics counter.
2925 * The port identifier of the Ethernet device.
2926 * @param tx_queue_id
2927 * The index of the transmit queue for which a queue stats mapping is required.
2928 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2929 * to rte_eth_dev_configure().
2931 * The per-queue packet statistics functionality number that the transmit
2932 * queue is to be assigned.
2933 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2934 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2936 * Zero if successful. Non-zero otherwise.
2938 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2939 uint16_t tx_queue_id, uint8_t stat_idx);
2942 * Set a mapping for the specified receive queue to the specified per-queue
2943 * statistics counter.
2946 * The port identifier of the Ethernet device.
2947 * @param rx_queue_id
2948 * The index of the receive queue for which a queue stats mapping is required.
2949 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2950 * to rte_eth_dev_configure().
2952 * The per-queue packet statistics functionality number that the receive
2953 * queue is to be assigned.
2954 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2955 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2957 * Zero if successful. Non-zero otherwise.
2959 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2960 uint16_t rx_queue_id,
2964 * Retrieve the Ethernet address of an Ethernet device.
2967 * The port identifier of the Ethernet device.
2969 * A pointer to a structure of type *ether_addr* to be filled with
2970 * the Ethernet address of the Ethernet device.
2972 * - (0) if successful
2973 * - (-ENODEV) if *port_id* invalid.
2975 int rte_eth_macaddr_get(uint16_t port_id, struct rte_ether_addr *mac_addr);
2978 * Retrieve the contextual information of an Ethernet device.
2980 * As part of this function, a number of of fields in dev_info will be
2981 * initialized as follows:
2986 * Where lim is defined within the rte_eth_dev_info_get as
2988 * const struct rte_eth_desc_lim lim = {
2989 * .nb_max = UINT16_MAX,
2992 * .nb_seg_max = UINT16_MAX,
2993 * .nb_mtu_seg_max = UINT16_MAX,
2996 * device = dev->device
2997 * min_mtu = RTE_ETHER_MIN_MTU
2998 * max_mtu = UINT16_MAX
3000 * The following fields will be populated if support for dev_infos_get()
3001 * exists for the device and the rte_eth_dev 'dev' has been populated
3002 * successfully with a call to it:
3004 * driver_name = dev->device->driver->name
3005 * nb_rx_queues = dev->data->nb_rx_queues
3006 * nb_tx_queues = dev->data->nb_tx_queues
3007 * dev_flags = &dev->data->dev_flags
3010 * The port identifier of the Ethernet device.
3012 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
3013 * the contextual information of the Ethernet device.
3015 * - (0) if successful.
3016 * - (-ENOTSUP) if support for dev_infos_get() does not exist for the device.
3017 * - (-ENODEV) if *port_id* invalid.
3019 int rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
3022 * Retrieve the firmware version of a device.
3025 * The port identifier of the device.
3027 * A pointer to a string array storing the firmware version of a device,
3028 * the string includes terminating null. This pointer is allocated by caller.
3030 * The size of the string array pointed by fw_version, which should be
3031 * large enough to store firmware version of the device.
3033 * - (0) if successful.
3034 * - (-ENOTSUP) if operation is not supported.
3035 * - (-ENODEV) if *port_id* invalid.
3036 * - (-EIO) if device is removed.
3037 * - (>0) if *fw_size* is not enough to store firmware version, return
3038 * the size of the non truncated string.
3040 int rte_eth_dev_fw_version_get(uint16_t port_id,
3041 char *fw_version, size_t fw_size);
3044 * Retrieve the supported packet types of an Ethernet device.
3046 * When a packet type is announced as supported, it *must* be recognized by
3047 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
3048 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
3049 * packet types for these packets:
3050 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
3051 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
3052 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
3053 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
3055 * When a packet is received by a PMD, the most precise type must be
3056 * returned among the ones supported. However a PMD is allowed to set
3057 * packet type that is not in the supported list, at the condition that it
3058 * is more precise. Therefore, a PMD announcing no supported packet types
3059 * can still set a matching packet type in a received packet.
3062 * Better to invoke this API after the device is already started or rx burst
3063 * function is decided, to obtain correct supported ptypes.
3065 * if a given PMD does not report what ptypes it supports, then the supported
3066 * ptype count is reported as 0.
3068 * The port identifier of the Ethernet device.
3070 * A hint of what kind of packet type which the caller is interested in.
3072 * An array pointer to store adequate packet types, allocated by caller.
3074 * Size of the array pointed by param ptypes.
3076 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
3077 * only num entries will be filled into the ptypes array, but the full
3078 * count of supported ptypes will be returned.
3079 * - (-ENODEV) if *port_id* invalid.
3081 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
3082 uint32_t *ptypes, int num);
3085 * @b EXPERIMENTAL: this API may change without prior notice.
3087 * Inform Ethernet device about reduced range of packet types to handle.
3089 * Application can use this function to set only specific ptypes that it's
3090 * interested. This information can be used by the PMD to optimize Rx path.
3092 * The function accepts an array `set_ptypes` allocated by the caller to
3093 * store the packet types set by the driver, the last element of the array
3094 * is set to RTE_PTYPE_UNKNOWN. The size of the `set_ptype` array should be
3095 * `rte_eth_dev_get_supported_ptypes() + 1` else it might only be filled
3099 * The port identifier of the Ethernet device.
3101 * The ptype family that application is interested in should be bitwise OR of
3102 * RTE_PTYPE_*_MASK or 0.
3104 * An array pointer to store set packet types, allocated by caller. The
3105 * function marks the end of array with RTE_PTYPE_UNKNOWN.
3107 * Size of the array pointed by param ptypes.
3108 * Should be rte_eth_dev_get_supported_ptypes() + 1 to accommodate the
3112 * - (-ENODEV) if *port_id* invalid.
3113 * - (-EINVAL) if *ptype_mask* is invalid (or) set_ptypes is NULL and
3117 int rte_eth_dev_set_ptypes(uint16_t port_id, uint32_t ptype_mask,
3118 uint32_t *set_ptypes, unsigned int num);
3121 * Retrieve the MTU of an Ethernet device.
3124 * The port identifier of the Ethernet device.
3126 * A pointer to a uint16_t where the retrieved MTU is to be stored.
3128 * - (0) if successful.
3129 * - (-ENODEV) if *port_id* invalid.
3131 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
3134 * Change the MTU of an Ethernet device.
3137 * The port identifier of the Ethernet device.
3139 * A uint16_t for the MTU to be applied.
3141 * - (0) if successful.
3142 * - (-ENOTSUP) if operation is not supported.
3143 * - (-ENODEV) if *port_id* invalid.
3144 * - (-EIO) if device is removed.
3145 * - (-EINVAL) if *mtu* invalid, validation of mtu can occur within
3146 * rte_eth_dev_set_mtu if dev_infos_get is supported by the device or
3147 * when the mtu is set using dev->dev_ops->mtu_set.
3148 * - (-EBUSY) if operation is not allowed when the port is running
3150 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
3153 * Enable/Disable hardware filtering by an Ethernet device of received
3154 * VLAN packets tagged with a given VLAN Tag Identifier.
3157 * The port identifier of the Ethernet device.
3159 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
3161 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
3162 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
3164 * - (0) if successful.
3165 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3166 * - (-ENODEV) if *port_id* invalid.
3167 * - (-EIO) if device is removed.
3168 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
3169 * - (-EINVAL) if *vlan_id* > 4095.
3171 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
3174 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
3177 * The port identifier of the Ethernet device.
3178 * @param rx_queue_id
3179 * The index of the receive queue for which a queue stats mapping is required.
3180 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3181 * to rte_eth_dev_configure().
3183 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
3184 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
3186 * - (0) if successful.
3187 * - (-ENOTSUP) if hardware-assisted VLAN stripping not configured.
3188 * - (-ENODEV) if *port_id* invalid.
3189 * - (-EINVAL) if *rx_queue_id* invalid.
3191 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
3195 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
3199 * The port identifier of the Ethernet device.
3203 * The Tag Protocol ID
3205 * - (0) if successful.
3206 * - (-ENOTSUP) if hardware-assisted VLAN TPID setup is not supported.
3207 * - (-ENODEV) if *port_id* invalid.
3208 * - (-EIO) if device is removed.
3210 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
3211 enum rte_vlan_type vlan_type,
3215 * Set VLAN offload configuration on an Ethernet device.
3218 * The port identifier of the Ethernet device.
3219 * @param offload_mask
3220 * The VLAN Offload bit mask can be mixed use with "OR"
3221 * ETH_VLAN_STRIP_OFFLOAD
3222 * ETH_VLAN_FILTER_OFFLOAD
3223 * ETH_VLAN_EXTEND_OFFLOAD
3224 * ETH_QINQ_STRIP_OFFLOAD
3226 * - (0) if successful.
3227 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3228 * - (-ENODEV) if *port_id* invalid.
3229 * - (-EIO) if device is removed.
3231 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
3234 * Read VLAN Offload configuration from an Ethernet device
3237 * The port identifier of the Ethernet device.
3239 * - (>0) if successful. Bit mask to indicate
3240 * ETH_VLAN_STRIP_OFFLOAD
3241 * ETH_VLAN_FILTER_OFFLOAD
3242 * ETH_VLAN_EXTEND_OFFLOAD
3243 * ETH_QINQ_STRIP_OFFLOAD
3244 * - (-ENODEV) if *port_id* invalid.
3246 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
3249 * Set port based TX VLAN insertion on or off.
3252 * The port identifier of the Ethernet device.
3254 * Port based TX VLAN identifier together with user priority.
3256 * Turn on or off the port based TX VLAN insertion.
3259 * - (0) if successful.
3260 * - negative if failed.
3262 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
3264 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
3268 * Structure used to buffer packets for future TX
3269 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
3271 struct rte_eth_dev_tx_buffer {
3272 buffer_tx_error_fn error_callback;
3273 void *error_userdata;
3274 uint16_t size; /**< Size of buffer for buffered tx */
3275 uint16_t length; /**< Number of packets in the array */
3276 struct rte_mbuf *pkts[];
3277 /**< Pending packets to be sent on explicit flush or when full */
3281 * Calculate the size of the tx buffer.
3284 * Number of stored packets.
3286 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
3287 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
3290 * Initialize default values for buffered transmitting
3293 * Tx buffer to be initialized.
3300 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
3303 * Configure a callback for buffered packets which cannot be sent
3305 * Register a specific callback to be called when an attempt is made to send
3306 * all packets buffered on an ethernet port, but not all packets can
3307 * successfully be sent. The callback registered here will be called only
3308 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
3309 * The default callback configured for each queue by default just frees the
3310 * packets back to the calling mempool. If additional behaviour is required,
3311 * for example, to count dropped packets, or to retry transmission of packets
3312 * which cannot be sent, this function should be used to register a suitable
3313 * callback function to implement the desired behaviour.
3314 * The example callback "rte_eth_count_unsent_packet_callback()" is also
3315 * provided as reference.
3318 * The port identifier of the Ethernet device.
3320 * The function to be used as the callback.
3322 * Arbitrary parameter to be passed to the callback function
3324 * 0 on success, or -1 on error with rte_errno set appropriately
3327 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
3328 buffer_tx_error_fn callback, void *userdata);
3331 * Callback function for silently dropping unsent buffered packets.
3333 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3334 * adjust the default behavior when buffered packets cannot be sent. This
3335 * function drops any unsent packets silently and is used by tx buffered
3336 * operations as default behavior.
3338 * NOTE: this function should not be called directly, instead it should be used
3339 * as a callback for packet buffering.
3341 * NOTE: when configuring this function as a callback with
3342 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3343 * should point to an uint64_t value.
3346 * The previously buffered packets which could not be sent
3348 * The number of unsent packets in the pkts array
3353 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
3357 * Callback function for tracking unsent buffered packets.
3359 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3360 * adjust the default behavior when buffered packets cannot be sent. This
3361 * function drops any unsent packets, but also updates a user-supplied counter
3362 * to track the overall number of packets dropped. The counter should be an
3363 * uint64_t variable.
3365 * NOTE: this function should not be called directly, instead it should be used
3366 * as a callback for packet buffering.
3368 * NOTE: when configuring this function as a callback with
3369 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3370 * should point to an uint64_t value.
3373 * The previously buffered packets which could not be sent
3375 * The number of unsent packets in the pkts array
3377 * Pointer to an uint64_t value, which will be incremented by unsent
3380 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
3384 * Request the driver to free mbufs currently cached by the driver. The
3385 * driver will only free the mbuf if it is no longer in use. It is the
3386 * application's responsibility to ensure rte_eth_tx_buffer_flush(..) is
3390 * The port identifier of the Ethernet device.
3392 * The index of the transmit queue through which output packets must be
3394 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
3395 * to rte_eth_dev_configure().
3397 * Maximum number of packets to free. Use 0 to indicate all possible packets
3398 * should be freed. Note that a packet may be using multiple mbufs.
3401 * -ENODEV: Invalid interface
3402 * -EIO: device is removed
3403 * -ENOTSUP: Driver does not support function
3405 * 0-n: Number of packets freed. More packets may still remain in ring that
3409 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
3412 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
3415 enum rte_eth_event_ipsec_subtype {
3416 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
3417 /**< Unknown event type */
3418 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
3419 /**< Sequence number overflow */
3420 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
3421 /**< Soft time expiry of SA */
3422 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
3423 /**< Soft byte expiry of SA */
3424 RTE_ETH_EVENT_IPSEC_MAX
3425 /**< Max value of this enum */
3429 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
3430 * information of the IPsec offload event.
3432 struct rte_eth_event_ipsec_desc {
3433 enum rte_eth_event_ipsec_subtype subtype;
3434 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
3436 /**< Event specific metadata
3438 * For the following events, *userdata* registered
3439 * with the *rte_security_session* would be returned
3442 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
3443 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
3444 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
3446 * @see struct rte_security_session_conf
3452 * The eth device event type for interrupt, and maybe others in the future.
3454 enum rte_eth_event_type {
3455 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
3456 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
3457 RTE_ETH_EVENT_QUEUE_STATE,
3458 /**< queue state event (enabled/disabled) */
3459 RTE_ETH_EVENT_INTR_RESET,
3460 /**< reset interrupt event, sent to VF on PF reset */
3461 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
3462 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
3463 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
3464 RTE_ETH_EVENT_NEW, /**< port is probed */
3465 RTE_ETH_EVENT_DESTROY, /**< port is released */
3466 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
3467 RTE_ETH_EVENT_FLOW_AGED,/**< New aged-out flows is detected */
3468 RTE_ETH_EVENT_MAX /**< max value of this enum */
3471 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
3472 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
3473 /**< user application callback to be registered for interrupts */
3476 * Register a callback function for port event.
3480 * RTE_ETH_ALL means register the event for all port ids.
3484 * User supplied callback function to be called.
3486 * Pointer to the parameters for the registered callback.
3489 * - On success, zero.
3490 * - On failure, a negative value.
3492 int rte_eth_dev_callback_register(uint16_t port_id,
3493 enum rte_eth_event_type event,
3494 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3497 * Unregister a callback function for port event.
3501 * RTE_ETH_ALL means unregister the event for all port ids.
3505 * User supplied callback function to be called.
3507 * Pointer to the parameters for the registered callback. -1 means to
3508 * remove all for the same callback address and same event.
3511 * - On success, zero.
3512 * - On failure, a negative value.
3514 int rte_eth_dev_callback_unregister(uint16_t port_id,
3515 enum rte_eth_event_type event,
3516 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3519 * When there is no rx packet coming in Rx Queue for a long time, we can
3520 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
3521 * to be triggered when Rx packet arrives.
3523 * The rte_eth_dev_rx_intr_enable() function enables rx queue
3524 * interrupt on specific rx queue of a port.
3527 * The port identifier of the Ethernet device.
3529 * The index of the receive queue from which to retrieve input packets.
3530 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3531 * to rte_eth_dev_configure().
3533 * - (0) if successful.
3534 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3536 * - (-ENODEV) if *port_id* invalid.
3537 * - (-EIO) if device is removed.
3539 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
3542 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
3543 * interrupt and returns to polling mode.
3545 * The rte_eth_dev_rx_intr_disable() function disables rx queue
3546 * interrupt on specific rx queue of a port.
3549 * The port identifier of the Ethernet device.
3551 * The index of the receive queue from which to retrieve input packets.
3552 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3553 * to rte_eth_dev_configure().
3555 * - (0) if successful.
3556 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3558 * - (-ENODEV) if *port_id* invalid.
3559 * - (-EIO) if device is removed.
3561 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
3564 * RX Interrupt control per port.
3567 * The port identifier of the Ethernet device.
3569 * Epoll instance fd which the intr vector associated to.
3570 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3572 * The operation be performed for the vector.
3573 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3577 * - On success, zero.
3578 * - On failure, a negative value.
3580 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
3583 * RX Interrupt control per queue.
3586 * The port identifier of the Ethernet device.
3588 * The index of the receive queue from which to retrieve input packets.
3589 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3590 * to rte_eth_dev_configure().
3592 * Epoll instance fd which the intr vector associated to.
3593 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3595 * The operation be performed for the vector.
3596 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3600 * - On success, zero.
3601 * - On failure, a negative value.
3603 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
3604 int epfd, int op, void *data);
3608 * @b EXPERIMENTAL: this API may change without prior notice.
3610 * Get interrupt fd per Rx queue.
3613 * The port identifier of the Ethernet device.
3615 * The index of the receive queue from which to retrieve input packets.
3616 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3617 * to rte_eth_dev_configure().
3619 * - (>=0) the interrupt fd associated to the requested Rx queue if
3625 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
3628 * Turn on the LED on the Ethernet device.
3629 * This function turns on the LED on the Ethernet device.
3632 * The port identifier of the Ethernet device.
3634 * - (0) if successful.
3635 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3637 * - (-ENODEV) if *port_id* invalid.
3638 * - (-EIO) if device is removed.
3640 int rte_eth_led_on(uint16_t port_id);
3643 * Turn off the LED on the Ethernet device.
3644 * This function turns off the LED on the Ethernet device.
3647 * The port identifier of the Ethernet device.
3649 * - (0) if successful.
3650 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3652 * - (-ENODEV) if *port_id* invalid.
3653 * - (-EIO) if device is removed.
3655 int rte_eth_led_off(uint16_t port_id);
3659 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3661 * Get Forward Error Correction(FEC) capability.
3664 * The port identifier of the Ethernet device.
3665 * @param speed_fec_capa
3666 * speed_fec_capa is out only with per-speed capabilities.
3667 * If set to NULL, the function returns the required number
3668 * of required array entries.
3670 * a number of elements in an speed_fec_capa array.
3673 * - A non-negative value lower or equal to num: success. The return value
3674 * is the number of entries filled in the fec capa array.
3675 * - A non-negative value higher than num: error, the given fec capa array
3676 * is too small. The return value corresponds to the num that should
3677 * be given to succeed. The entries in fec capa array are not valid and
3678 * shall not be used by the caller.
3679 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3681 * - (-EIO) if device is removed.
3682 * - (-ENODEV) if *port_id* invalid.
3683 * - (-EINVAL) if *num* or *speed_fec_capa* invalid
3686 int rte_eth_fec_get_capability(uint16_t port_id,
3687 struct rte_eth_fec_capa *speed_fec_capa,
3692 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3694 * Get current Forward Error Correction(FEC) mode.
3695 * If link is down and AUTO is enabled, AUTO is returned, otherwise,
3696 * configured FEC mode is returned.
3697 * If link is up, current FEC mode is returned.
3700 * The port identifier of the Ethernet device.
3702 * A bitmask of enabled FEC modes. If AUTO bit is set, other
3703 * bits specify FEC modes which may be negotiated. If AUTO
3704 * bit is clear, specify FEC modes to be used (only one valid
3705 * mode per speed may be set).
3707 * - (0) if successful.
3708 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3710 * - (-EIO) if device is removed.
3711 * - (-ENODEV) if *port_id* invalid.
3714 int rte_eth_fec_get(uint16_t port_id, uint32_t *fec_capa);
3718 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3720 * Set Forward Error Correction(FEC) mode.
3723 * The port identifier of the Ethernet device.
3725 * A bitmask of allowed FEC modes. If AUTO bit is set, other
3726 * bits specify FEC modes which may be negotiated. If AUTO
3727 * bit is clear, specify FEC modes to be used (only one valid
3728 * mode per speed may be set).
3730 * - (0) if successful.
3731 * - (-EINVAL) if the FEC mode is not valid.
3732 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3733 * - (-EIO) if device is removed.
3734 * - (-ENODEV) if *port_id* invalid.
3737 int rte_eth_fec_set(uint16_t port_id, uint32_t fec_capa);
3740 * Get current status of the Ethernet link flow control for Ethernet device
3743 * The port identifier of the Ethernet device.
3745 * The pointer to the structure where to store the flow control parameters.
3747 * - (0) if successful.
3748 * - (-ENOTSUP) if hardware doesn't support flow control.
3749 * - (-ENODEV) if *port_id* invalid.
3750 * - (-EIO) if device is removed.
3752 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
3753 struct rte_eth_fc_conf *fc_conf);
3756 * Configure the Ethernet link flow control for Ethernet device
3759 * The port identifier of the Ethernet device.
3761 * The pointer to the structure of the flow control parameters.
3763 * - (0) if successful.
3764 * - (-ENOTSUP) if hardware doesn't support flow control mode.
3765 * - (-ENODEV) if *port_id* invalid.
3766 * - (-EINVAL) if bad parameter
3767 * - (-EIO) if flow control setup failure or device is removed.
3769 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
3770 struct rte_eth_fc_conf *fc_conf);
3773 * Configure the Ethernet priority flow control under DCB environment
3774 * for Ethernet device.
3777 * The port identifier of the Ethernet device.
3779 * The pointer to the structure of the priority flow control parameters.
3781 * - (0) if successful.
3782 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
3783 * - (-ENODEV) if *port_id* invalid.
3784 * - (-EINVAL) if bad parameter
3785 * - (-EIO) if flow control setup failure or device is removed.
3787 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
3788 struct rte_eth_pfc_conf *pfc_conf);
3791 * Add a MAC address to the set used for filtering incoming packets.
3794 * The port identifier of the Ethernet device.
3796 * The MAC address to add.
3798 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
3799 * not enabled, this should be set to 0.
3801 * - (0) if successfully added or *mac_addr* was already added.
3802 * - (-ENOTSUP) if hardware doesn't support this feature.
3803 * - (-ENODEV) if *port* is invalid.
3804 * - (-EIO) if device is removed.
3805 * - (-ENOSPC) if no more MAC addresses can be added.
3806 * - (-EINVAL) if MAC address is invalid.
3808 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct rte_ether_addr *mac_addr,
3812 * Remove a MAC address from the internal array of addresses.
3815 * The port identifier of the Ethernet device.
3817 * MAC address to remove.
3819 * - (0) if successful, or *mac_addr* didn't exist.
3820 * - (-ENOTSUP) if hardware doesn't support.
3821 * - (-ENODEV) if *port* invalid.
3822 * - (-EADDRINUSE) if attempting to remove the default MAC address
3824 int rte_eth_dev_mac_addr_remove(uint16_t port_id,
3825 struct rte_ether_addr *mac_addr);
3828 * Set the default MAC address.
3831 * The port identifier of the Ethernet device.
3833 * New default MAC address.
3835 * - (0) if successful, or *mac_addr* didn't exist.
3836 * - (-ENOTSUP) if hardware doesn't support.
3837 * - (-ENODEV) if *port* invalid.
3838 * - (-EINVAL) if MAC address is invalid.
3840 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
3841 struct rte_ether_addr *mac_addr);
3844 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3847 * The port identifier of the Ethernet device.
3851 * Redirection table size. The table size can be queried by
3852 * rte_eth_dev_info_get().
3854 * - (0) if successful.
3855 * - (-ENODEV) if *port_id* is invalid.
3856 * - (-ENOTSUP) if hardware doesn't support.
3857 * - (-EINVAL) if bad parameter.
3858 * - (-EIO) if device is removed.
3860 int rte_eth_dev_rss_reta_update(uint16_t port_id,
3861 struct rte_eth_rss_reta_entry64 *reta_conf,
3862 uint16_t reta_size);
3865 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3868 * The port identifier of the Ethernet device.
3870 * RETA to query. For each requested reta entry, corresponding bit
3871 * in mask must be set.
3873 * Redirection table size. The table size can be queried by
3874 * rte_eth_dev_info_get().
3876 * - (0) if successful.
3877 * - (-ENODEV) if *port_id* is invalid.
3878 * - (-ENOTSUP) if hardware doesn't support.
3879 * - (-EINVAL) if bad parameter.
3880 * - (-EIO) if device is removed.
3882 int rte_eth_dev_rss_reta_query(uint16_t port_id,
3883 struct rte_eth_rss_reta_entry64 *reta_conf,
3884 uint16_t reta_size);
3887 * Updates unicast hash table for receiving packet with the given destination
3888 * MAC address, and the packet is routed to all VFs for which the RX mode is
3889 * accept packets that match the unicast hash table.
3892 * The port identifier of the Ethernet device.
3894 * Unicast MAC address.
3896 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
3897 * 0 - Clear an unicast hash bit.
3899 * - (0) if successful.
3900 * - (-ENOTSUP) if hardware doesn't support.
3901 * - (-ENODEV) if *port_id* invalid.
3902 * - (-EIO) if device is removed.
3903 * - (-EINVAL) if bad parameter.
3905 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct rte_ether_addr *addr,
3909 * Updates all unicast hash bitmaps for receiving packet with any Unicast
3910 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
3911 * mode is accept packets that match the unicast hash table.
3914 * The port identifier of the Ethernet device.
3916 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
3918 * 0 - Clear all unicast hash bitmaps
3920 * - (0) if successful.
3921 * - (-ENOTSUP) if hardware doesn't support.
3922 * - (-ENODEV) if *port_id* invalid.
3923 * - (-EIO) if device is removed.
3924 * - (-EINVAL) if bad parameter.
3926 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
3929 * Set a traffic mirroring rule on an Ethernet device
3932 * The port identifier of the Ethernet device.
3933 * @param mirror_conf
3934 * The pointer to the traffic mirroring structure describing the mirroring rule.
3935 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
3936 * destination pool and the value of rule if enable vlan or pool mirroring.
3939 * The index of traffic mirroring rule, we support four separated rules.
3941 * 1 - Enable a mirroring rule.
3942 * 0 - Disable a mirroring rule.
3944 * - (0) if successful.
3945 * - (-ENOTSUP) if hardware doesn't support this feature.
3946 * - (-ENODEV) if *port_id* invalid.
3947 * - (-EIO) if device is removed.
3948 * - (-EINVAL) if the mr_conf information is not correct.
3950 int rte_eth_mirror_rule_set(uint16_t port_id,
3951 struct rte_eth_mirror_conf *mirror_conf,
3956 * Reset a traffic mirroring rule on an Ethernet device.
3959 * The port identifier of the Ethernet device.
3961 * The index of traffic mirroring rule, we support four separated rules.
3963 * - (0) if successful.
3964 * - (-ENOTSUP) if hardware doesn't support this feature.
3965 * - (-ENODEV) if *port_id* invalid.
3966 * - (-EIO) if device is removed.
3967 * - (-EINVAL) if bad parameter.
3969 int rte_eth_mirror_rule_reset(uint16_t port_id,
3973 * Set the rate limitation for a queue on an Ethernet device.
3976 * The port identifier of the Ethernet device.
3980 * The tx rate in Mbps. Allocated from the total port link speed.
3982 * - (0) if successful.
3983 * - (-ENOTSUP) if hardware doesn't support this feature.
3984 * - (-ENODEV) if *port_id* invalid.
3985 * - (-EIO) if device is removed.
3986 * - (-EINVAL) if bad parameter.
3988 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
3992 * Configuration of Receive Side Scaling hash computation of Ethernet device.
3995 * The port identifier of the Ethernet device.
3997 * The new configuration to use for RSS hash computation on the port.
3999 * - (0) if successful.
4000 * - (-ENODEV) if port identifier is invalid.
4001 * - (-EIO) if device is removed.
4002 * - (-ENOTSUP) if hardware doesn't support.
4003 * - (-EINVAL) if bad parameter.
4005 int rte_eth_dev_rss_hash_update(uint16_t port_id,
4006 struct rte_eth_rss_conf *rss_conf);
4009 * Retrieve current configuration of Receive Side Scaling hash computation
4010 * of Ethernet device.
4013 * The port identifier of the Ethernet device.
4015 * Where to store the current RSS hash configuration of the Ethernet device.
4017 * - (0) if successful.
4018 * - (-ENODEV) if port identifier is invalid.
4019 * - (-EIO) if device is removed.
4020 * - (-ENOTSUP) if hardware doesn't support RSS.
4023 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
4024 struct rte_eth_rss_conf *rss_conf);
4027 * Add UDP tunneling port for a specific type of tunnel.
4028 * The packets with this UDP port will be identified as this type of tunnel.
4029 * Before enabling any offloading function for a tunnel, users can call this API
4030 * to change or add more UDP port for the tunnel. So the offloading function
4031 * can take effect on the packets with the specific UDP port.
4034 * The port identifier of the Ethernet device.
4036 * UDP tunneling configuration.
4039 * - (0) if successful.
4040 * - (-ENODEV) if port identifier is invalid.
4041 * - (-EIO) if device is removed.
4042 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4045 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
4046 struct rte_eth_udp_tunnel *tunnel_udp);
4049 * Delete UDP tunneling port a specific type of tunnel.
4050 * The packets with this UDP port will not be identified as this type of tunnel
4052 * Before enabling any offloading function for a tunnel, users can call this API
4053 * to delete a UDP port for the tunnel. So the offloading function will not take
4054 * effect on the packets with the specific UDP port.
4057 * The port identifier of the Ethernet device.
4059 * UDP tunneling configuration.
4062 * - (0) if successful.
4063 * - (-ENODEV) if port identifier is invalid.
4064 * - (-EIO) if device is removed.
4065 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4068 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
4069 struct rte_eth_udp_tunnel *tunnel_udp);
4072 * Get DCB information on an Ethernet device.
4075 * The port identifier of the Ethernet device.
4079 * - (0) if successful.
4080 * - (-ENODEV) if port identifier is invalid.
4081 * - (-EIO) if device is removed.
4082 * - (-ENOTSUP) if hardware doesn't support.
4084 int rte_eth_dev_get_dcb_info(uint16_t port_id,
4085 struct rte_eth_dcb_info *dcb_info);
4087 struct rte_eth_rxtx_callback;
4090 * Add a callback to be called on packet RX on a given port and queue.
4092 * This API configures a function to be called for each burst of
4093 * packets received on a given NIC port queue. The return value is a pointer
4094 * that can be used to later remove the callback using
4095 * rte_eth_remove_rx_callback().
4097 * Multiple functions are called in the order that they are added.
4100 * The port identifier of the Ethernet device.
4102 * The queue on the Ethernet device on which the callback is to be added.
4104 * The callback function
4106 * A generic pointer parameter which will be passed to each invocation of the
4107 * callback function on this port and queue. Inter-thread synchronization
4108 * of any user data changes is the responsibility of the user.
4112 * On success, a pointer value which can later be used to remove the callback.
4114 const struct rte_eth_rxtx_callback *
4115 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
4116 rte_rx_callback_fn fn, void *user_param);
4119 * Add a callback that must be called first on packet RX on a given port
4122 * This API configures a first function to be called for each burst of
4123 * packets received on a given NIC port queue. The return value is a pointer
4124 * that can be used to later remove the callback using
4125 * rte_eth_remove_rx_callback().
4127 * Multiple functions are called in the order that they are added.
4130 * The port identifier of the Ethernet device.
4132 * The queue on the Ethernet device on which the callback is to be added.
4134 * The callback function
4136 * A generic pointer parameter which will be passed to each invocation of the
4137 * callback function on this port and queue. Inter-thread synchronization
4138 * of any user data changes is the responsibility of the user.
4142 * On success, a pointer value which can later be used to remove the callback.
4144 const struct rte_eth_rxtx_callback *
4145 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
4146 rte_rx_callback_fn fn, void *user_param);
4149 * Add a callback to be called on packet TX on a given port and queue.
4151 * This API configures a function to be called for each burst of
4152 * packets sent on a given NIC port queue. The return value is a pointer
4153 * that can be used to later remove the callback using
4154 * rte_eth_remove_tx_callback().
4156 * Multiple functions are called in the order that they are added.
4159 * The port identifier of the Ethernet device.
4161 * The queue on the Ethernet device on which the callback is to be added.
4163 * The callback function
4165 * A generic pointer parameter which will be passed to each invocation of the
4166 * callback function on this port and queue. Inter-thread synchronization
4167 * of any user data changes is the responsibility of the user.
4171 * On success, a pointer value which can later be used to remove the callback.
4173 const struct rte_eth_rxtx_callback *
4174 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
4175 rte_tx_callback_fn fn, void *user_param);
4178 * Remove an RX packet callback from a given port and queue.
4180 * This function is used to removed callbacks that were added to a NIC port
4181 * queue using rte_eth_add_rx_callback().
4183 * Note: the callback is removed from the callback list but it isn't freed
4184 * since the it may still be in use. The memory for the callback can be
4185 * subsequently freed back by the application by calling rte_free():
4187 * - Immediately - if the port is stopped, or the user knows that no
4188 * callbacks are in flight e.g. if called from the thread doing RX/TX
4191 * - After a short delay - where the delay is sufficient to allow any
4192 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4193 * used to detect when data plane threads have ceased referencing the
4197 * The port identifier of the Ethernet device.
4199 * The queue on the Ethernet device from which the callback is to be removed.
4201 * User supplied callback created via rte_eth_add_rx_callback().
4204 * - 0: Success. Callback was removed.
4205 * - -ENODEV: If *port_id* is invalid.
4206 * - -ENOTSUP: Callback support is not available.
4207 * - -EINVAL: The queue_id is out of range, or the callback
4208 * is NULL or not found for the port/queue.
4210 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
4211 const struct rte_eth_rxtx_callback *user_cb);
4214 * Remove a TX packet callback from a given port and queue.
4216 * This function is used to removed callbacks that were added to a NIC port
4217 * queue using rte_eth_add_tx_callback().
4219 * Note: the callback is removed from the callback list but it isn't freed
4220 * since the it may still be in use. The memory for the callback can be
4221 * subsequently freed back by the application by calling rte_free():
4223 * - Immediately - if the port is stopped, or the user knows that no
4224 * callbacks are in flight e.g. if called from the thread doing RX/TX
4227 * - After a short delay - where the delay is sufficient to allow any
4228 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4229 * used to detect when data plane threads have ceased referencing the
4233 * The port identifier of the Ethernet device.
4235 * The queue on the Ethernet device from which the callback is to be removed.
4237 * User supplied callback created via rte_eth_add_tx_callback().
4240 * - 0: Success. Callback was removed.
4241 * - -ENODEV: If *port_id* is invalid.
4242 * - -ENOTSUP: Callback support is not available.
4243 * - -EINVAL: The queue_id is out of range, or the callback
4244 * is NULL or not found for the port/queue.
4246 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
4247 const struct rte_eth_rxtx_callback *user_cb);
4250 * Retrieve information about given port's RX queue.
4253 * The port identifier of the Ethernet device.
4255 * The RX queue on the Ethernet device for which information
4256 * will be retrieved.
4258 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
4259 * the information of the Ethernet device.
4263 * - -ENODEV: If *port_id* is invalid.
4264 * - -ENOTSUP: routine is not supported by the device PMD.
4265 * - -EINVAL: The queue_id is out of range, or the queue
4268 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4269 struct rte_eth_rxq_info *qinfo);
4272 * Retrieve information about given port's TX queue.
4275 * The port identifier of the Ethernet device.
4277 * The TX queue on the Ethernet device for which information
4278 * will be retrieved.
4280 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
4281 * the information of the Ethernet device.
4285 * - -ENODEV: If *port_id* is invalid.
4286 * - -ENOTSUP: routine is not supported by the device PMD.
4287 * - -EINVAL: The queue_id is out of range, or the queue
4290 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4291 struct rte_eth_txq_info *qinfo);
4294 * Retrieve information about the Rx packet burst mode.
4297 * The port identifier of the Ethernet device.
4299 * The Rx queue on the Ethernet device for which information
4300 * will be retrieved.
4302 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4303 * with the information of the packet burst mode.
4307 * - -ENODEV: If *port_id* is invalid.
4308 * - -ENOTSUP: routine is not supported by the device PMD.
4309 * - -EINVAL: The queue_id is out of range.
4312 int rte_eth_rx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4313 struct rte_eth_burst_mode *mode);
4316 * Retrieve information about the Tx packet burst mode.
4319 * The port identifier of the Ethernet device.
4321 * The Tx queue on the Ethernet device for which information
4322 * will be retrieved.
4324 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4325 * with the information of the packet burst mode.
4329 * - -ENODEV: If *port_id* is invalid.
4330 * - -ENOTSUP: routine is not supported by the device PMD.
4331 * - -EINVAL: The queue_id is out of range.
4334 int rte_eth_tx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4335 struct rte_eth_burst_mode *mode);
4338 * Retrieve device registers and register attributes (number of registers and
4342 * The port identifier of the Ethernet device.
4344 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
4345 * NULL the function fills in the width and length fields. If non-NULL
4346 * the registers are put into the buffer pointed at by the data field.
4348 * - (0) if successful.
4349 * - (-ENOTSUP) if hardware doesn't support.
4350 * - (-ENODEV) if *port_id* invalid.
4351 * - (-EIO) if device is removed.
4352 * - others depends on the specific operations implementation.
4354 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
4357 * Retrieve size of device EEPROM
4360 * The port identifier of the Ethernet device.
4362 * - (>=0) EEPROM size if successful.
4363 * - (-ENOTSUP) if hardware doesn't support.
4364 * - (-ENODEV) if *port_id* invalid.
4365 * - (-EIO) if device is removed.
4366 * - others depends on the specific operations implementation.
4368 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
4371 * Retrieve EEPROM and EEPROM attribute
4374 * The port identifier of the Ethernet device.
4376 * The template includes buffer for return EEPROM data and
4377 * EEPROM attributes to be filled.
4379 * - (0) if successful.
4380 * - (-ENOTSUP) if hardware doesn't support.
4381 * - (-ENODEV) if *port_id* invalid.
4382 * - (-EIO) if device is removed.
4383 * - others depends on the specific operations implementation.
4385 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4388 * Program EEPROM with provided data
4391 * The port identifier of the Ethernet device.
4393 * The template includes EEPROM data for programming and
4394 * EEPROM attributes to be filled
4396 * - (0) if successful.
4397 * - (-ENOTSUP) if hardware doesn't support.
4398 * - (-ENODEV) if *port_id* invalid.
4399 * - (-EIO) if device is removed.
4400 * - others depends on the specific operations implementation.
4402 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4406 * @b EXPERIMENTAL: this API may change without prior notice.
4408 * Retrieve the type and size of plugin module EEPROM
4411 * The port identifier of the Ethernet device.
4413 * The type and size of plugin module EEPROM.
4415 * - (0) if successful.
4416 * - (-ENOTSUP) if hardware doesn't support.
4417 * - (-ENODEV) if *port_id* invalid.
4418 * - (-EIO) if device is removed.
4419 * - others depends on the specific operations implementation.
4423 rte_eth_dev_get_module_info(uint16_t port_id,
4424 struct rte_eth_dev_module_info *modinfo);
4428 * @b EXPERIMENTAL: this API may change without prior notice.
4430 * Retrieve the data of plugin module EEPROM
4433 * The port identifier of the Ethernet device.
4435 * The template includes the plugin module EEPROM attributes, and the
4436 * buffer for return plugin module EEPROM data.
4438 * - (0) if successful.
4439 * - (-ENOTSUP) if hardware doesn't support.
4440 * - (-ENODEV) if *port_id* invalid.
4441 * - (-EIO) if device is removed.
4442 * - others depends on the specific operations implementation.
4446 rte_eth_dev_get_module_eeprom(uint16_t port_id,
4447 struct rte_dev_eeprom_info *info);
4450 * Set the list of multicast addresses to filter on an Ethernet device.
4453 * The port identifier of the Ethernet device.
4454 * @param mc_addr_set
4455 * The array of multicast addresses to set. Equal to NULL when the function
4456 * is invoked to flush the set of filtered addresses.
4458 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
4459 * when the function is invoked to flush the set of filtered addresses.
4461 * - (0) if successful.
4462 * - (-ENODEV) if *port_id* invalid.
4463 * - (-EIO) if device is removed.
4464 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
4465 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
4467 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
4468 struct rte_ether_addr *mc_addr_set,
4469 uint32_t nb_mc_addr);
4472 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
4475 * The port identifier of the Ethernet device.
4479 * - -ENODEV: The port ID is invalid.
4480 * - -EIO: if device is removed.
4481 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4483 int rte_eth_timesync_enable(uint16_t port_id);
4486 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
4489 * The port identifier of the Ethernet device.
4493 * - -ENODEV: The port ID is invalid.
4494 * - -EIO: if device is removed.
4495 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4497 int rte_eth_timesync_disable(uint16_t port_id);
4500 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
4503 * The port identifier of the Ethernet device.
4505 * Pointer to the timestamp struct.
4507 * Device specific flags. Used to pass the RX timesync register index to
4508 * i40e. Unused in igb/ixgbe, pass 0 instead.
4512 * - -EINVAL: No timestamp is available.
4513 * - -ENODEV: The port ID is invalid.
4514 * - -EIO: if device is removed.
4515 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4517 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
4518 struct timespec *timestamp, uint32_t flags);
4521 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
4524 * The port identifier of the Ethernet device.
4526 * Pointer to the timestamp struct.
4530 * - -EINVAL: No timestamp is available.
4531 * - -ENODEV: The port ID is invalid.
4532 * - -EIO: if device is removed.
4533 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4535 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
4536 struct timespec *timestamp);
4539 * Adjust the timesync clock on an Ethernet device.
4541 * This is usually used in conjunction with other Ethdev timesync functions to
4542 * synchronize the device time using the IEEE1588/802.1AS protocol.
4545 * The port identifier of the Ethernet device.
4547 * The adjustment in nanoseconds.
4551 * - -ENODEV: The port ID is invalid.
4552 * - -EIO: if device is removed.
4553 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4555 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
4558 * Read the time from the timesync clock on an Ethernet device.
4560 * This is usually used in conjunction with other Ethdev timesync functions to
4561 * synchronize the device time using the IEEE1588/802.1AS protocol.
4564 * The port identifier of the Ethernet device.
4566 * Pointer to the timespec struct that holds the time.
4571 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
4574 * Set the time of the timesync clock on an Ethernet device.
4576 * This is usually used in conjunction with other Ethdev timesync functions to
4577 * synchronize the device time using the IEEE1588/802.1AS protocol.
4580 * The port identifier of the Ethernet device.
4582 * Pointer to the timespec struct that holds the time.
4586 * - -EINVAL: No timestamp is available.
4587 * - -ENODEV: The port ID is invalid.
4588 * - -EIO: if device is removed.
4589 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4591 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
4595 * @b EXPERIMENTAL: this API may change without prior notice.
4597 * Read the current clock counter of an Ethernet device
4599 * This returns the current raw clock value of an Ethernet device. It is
4600 * a raw amount of ticks, with no given time reference.
4601 * The value returned here is from the same clock than the one
4602 * filling timestamp field of Rx packets when using hardware timestamp
4603 * offload. Therefore it can be used to compute a precise conversion of
4604 * the device clock to the real time.
4606 * E.g, a simple heuristic to derivate the frequency would be:
4607 * uint64_t start, end;
4608 * rte_eth_read_clock(port, start);
4609 * rte_delay_ms(100);
4610 * rte_eth_read_clock(port, end);
4611 * double freq = (end - start) * 10;
4613 * Compute a common reference with:
4614 * uint64_t base_time_sec = current_time();
4615 * uint64_t base_clock;
4616 * rte_eth_read_clock(port, base_clock);
4618 * Then, convert the raw mbuf timestamp with:
4619 * base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq;
4621 * This simple example will not provide a very good accuracy. One must
4622 * at least measure multiple times the frequency and do a regression.
4623 * To avoid deviation from the system time, the common reference can
4624 * be repeated from time to time. The integer division can also be
4625 * converted by a multiplication and a shift for better performance.
4628 * The port identifier of the Ethernet device.
4630 * Pointer to the uint64_t that holds the raw clock value.
4634 * - -ENODEV: The port ID is invalid.
4635 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4639 rte_eth_read_clock(uint16_t port_id, uint64_t *clock);
4642 * Get the port id from device name. The device name should be specified
4644 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
4645 * - SoC device name, for example- fsl-gmac0
4646 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
4649 * pci address or name of the device
4651 * pointer to port identifier of the device
4653 * - (0) if successful and port_id is filled.
4654 * - (-ENODEV or -EINVAL) on failure.
4657 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
4660 * Get the device name from port id. The device name is specified as below:
4661 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
4662 * - SoC device name, for example- fsl-gmac0
4663 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
4666 * Port identifier of the device.
4668 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
4670 * - (0) if successful.
4671 * - (-ENODEV) if *port_id* is invalid.
4672 * - (-EINVAL) on failure.
4675 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
4678 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
4679 * the ethernet device information, otherwise adjust them to boundaries.
4682 * The port identifier of the Ethernet device.
4684 * A pointer to a uint16_t where the number of receive
4685 * descriptors stored.
4687 * A pointer to a uint16_t where the number of transmit
4688 * descriptors stored.
4690 * - (0) if successful.
4691 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
4693 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
4694 uint16_t *nb_rx_desc,
4695 uint16_t *nb_tx_desc);
4698 * Test if a port supports specific mempool ops.
4701 * Port identifier of the Ethernet device.
4703 * The name of the pool operations to test.
4705 * - 0: best mempool ops choice for this port.
4706 * - 1: mempool ops are supported for this port.
4707 * - -ENOTSUP: mempool ops not supported for this port.
4708 * - -ENODEV: Invalid port Identifier.
4709 * - -EINVAL: Pool param is null.
4712 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
4715 * Get the security context for the Ethernet device.
4718 * Port identifier of the Ethernet device
4721 * - pointer to security context on success.
4724 rte_eth_dev_get_sec_ctx(uint16_t port_id);
4728 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
4730 * Query the device hairpin capabilities.
4733 * The port identifier of the Ethernet device.
4735 * Pointer to a structure that will hold the hairpin capabilities.
4737 * - (0) if successful.
4738 * - (-ENOTSUP) if hardware doesn't support.
4741 int rte_eth_dev_hairpin_capability_get(uint16_t port_id,
4742 struct rte_eth_hairpin_cap *cap);
4744 #include <rte_ethdev_core.h>
4748 * Retrieve a burst of input packets from a receive queue of an Ethernet
4749 * device. The retrieved packets are stored in *rte_mbuf* structures whose
4750 * pointers are supplied in the *rx_pkts* array.
4752 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
4753 * receive queue, up to *nb_pkts* packets, and for each completed RX
4754 * descriptor in the ring, it performs the following operations:
4756 * - Initialize the *rte_mbuf* data structure associated with the
4757 * RX descriptor according to the information provided by the NIC into
4758 * that RX descriptor.
4760 * - Store the *rte_mbuf* data structure into the next entry of the
4763 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
4764 * allocated from the memory pool associated with the receive queue at
4765 * initialization time.
4767 * When retrieving an input packet that was scattered by the controller
4768 * into multiple receive descriptors, the rte_eth_rx_burst() function
4769 * appends the associated *rte_mbuf* buffers to the first buffer of the
4772 * The rte_eth_rx_burst() function returns the number of packets
4773 * actually retrieved, which is the number of *rte_mbuf* data structures
4774 * effectively supplied into the *rx_pkts* array.
4775 * A return value equal to *nb_pkts* indicates that the RX queue contained
4776 * at least *rx_pkts* packets, and this is likely to signify that other
4777 * received packets remain in the input queue. Applications implementing
4778 * a "retrieve as much received packets as possible" policy can check this
4779 * specific case and keep invoking the rte_eth_rx_burst() function until
4780 * a value less than *nb_pkts* is returned.
4782 * This receive method has the following advantages:
4784 * - It allows a run-to-completion network stack engine to retrieve and
4785 * to immediately process received packets in a fast burst-oriented
4786 * approach, avoiding the overhead of unnecessary intermediate packet
4787 * queue/dequeue operations.
4789 * - Conversely, it also allows an asynchronous-oriented processing
4790 * method to retrieve bursts of received packets and to immediately
4791 * queue them for further parallel processing by another logical core,
4792 * for instance. However, instead of having received packets being
4793 * individually queued by the driver, this approach allows the caller
4794 * of the rte_eth_rx_burst() function to queue a burst of retrieved
4795 * packets at a time and therefore dramatically reduce the cost of
4796 * enqueue/dequeue operations per packet.
4798 * - It allows the rte_eth_rx_burst() function of the driver to take
4799 * advantage of burst-oriented hardware features (CPU cache,
4800 * prefetch instructions, and so on) to minimize the number of CPU
4801 * cycles per packet.
4803 * To summarize, the proposed receive API enables many
4804 * burst-oriented optimizations in both synchronous and asynchronous
4805 * packet processing environments with no overhead in both cases.
4808 * Some drivers using vector instructions require that *nb_pkts* is
4809 * divisible by 4 or 8, depending on the driver implementation.
4811 * The rte_eth_rx_burst() function does not provide any error
4812 * notification to avoid the corresponding overhead. As a hint, the
4813 * upper-level application might check the status of the device link once
4814 * being systematically returned a 0 value for a given number of tries.
4817 * The port identifier of the Ethernet device.
4819 * The index of the receive queue from which to retrieve input packets.
4820 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
4821 * to rte_eth_dev_configure().
4823 * The address of an array of pointers to *rte_mbuf* structures that
4824 * must be large enough to store *nb_pkts* pointers in it.
4826 * The maximum number of packets to retrieve.
4827 * The value must be divisible by 8 in order to work with any driver.
4829 * The number of packets actually retrieved, which is the number
4830 * of pointers to *rte_mbuf* structures effectively supplied to the
4833 static inline uint16_t
4834 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
4835 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
4837 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4840 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4841 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
4842 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
4844 if (queue_id >= dev->data->nb_rx_queues) {
4845 RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n", queue_id);
4849 nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
4852 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4853 struct rte_eth_rxtx_callback *cb;
4855 /* __ATOMIC_RELEASE memory order was used when the
4856 * call back was inserted into the list.
4857 * Since there is a clear dependency between loading
4858 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
4861 cb = __atomic_load_n(&dev->post_rx_burst_cbs[queue_id],
4864 if (unlikely(cb != NULL)) {
4866 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
4867 nb_pkts, cb->param);
4869 } while (cb != NULL);
4873 rte_ethdev_trace_rx_burst(port_id, queue_id, (void **)rx_pkts, nb_rx);
4878 * Get the number of used descriptors of a rx queue
4881 * The port identifier of the Ethernet device.
4883 * The queue id on the specific port.
4885 * The number of used descriptors in the specific queue, or:
4886 * - (-ENODEV) if *port_id* is invalid.
4887 * (-EINVAL) if *queue_id* is invalid
4888 * (-ENOTSUP) if the device does not support this function
4891 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
4893 struct rte_eth_dev *dev;
4895 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4896 dev = &rte_eth_devices[port_id];
4897 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_queue_count, -ENOTSUP);
4898 if (queue_id >= dev->data->nb_rx_queues ||
4899 dev->data->rx_queues[queue_id] == NULL)
4902 return (int)(*dev->rx_queue_count)(dev, queue_id);
4906 * Check if the DD bit of the specific RX descriptor in the queue has been set
4909 * The port identifier of the Ethernet device.
4911 * The queue id on the specific port.
4913 * The offset of the descriptor ID from tail.
4915 * - (1) if the specific DD bit is set.
4916 * - (0) if the specific DD bit is not set.
4917 * - (-ENODEV) if *port_id* invalid.
4918 * - (-ENOTSUP) if the device does not support this function
4922 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
4924 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4925 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4926 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_done, -ENOTSUP);
4927 return (*dev->rx_descriptor_done)(dev->data->rx_queues[queue_id], offset);
4930 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
4931 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
4932 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
4935 * Check the status of a Rx descriptor in the queue
4937 * It should be called in a similar context than the Rx function:
4938 * - on a dataplane core
4939 * - not concurrently on the same queue
4941 * Since it's a dataplane function, no check is performed on port_id and
4942 * queue_id. The caller must therefore ensure that the port is enabled
4943 * and the queue is configured and running.
4945 * Note: accessing to a random descriptor in the ring may trigger cache
4946 * misses and have a performance impact.
4949 * A valid port identifier of the Ethernet device which.
4951 * A valid Rx queue identifier on this port.
4953 * The offset of the descriptor starting from tail (0 is the next
4954 * packet to be received by the driver).
4957 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
4959 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
4960 * not yet processed by the driver (i.e. in the receive queue).
4961 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
4962 * the driver and not yet returned to hw, or reserved by the hw.
4963 * - (-EINVAL) bad descriptor offset.
4964 * - (-ENOTSUP) if the device does not support this function.
4965 * - (-ENODEV) bad port or queue (only if compiled with debug).
4968 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
4971 struct rte_eth_dev *dev;
4974 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4975 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4977 dev = &rte_eth_devices[port_id];
4978 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4979 if (queue_id >= dev->data->nb_rx_queues)
4982 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_status, -ENOTSUP);
4983 rxq = dev->data->rx_queues[queue_id];
4985 return (*dev->rx_descriptor_status)(rxq, offset);
4988 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
4989 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
4990 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
4993 * Check the status of a Tx descriptor in the queue.
4995 * It should be called in a similar context than the Tx function:
4996 * - on a dataplane core
4997 * - not concurrently on the same queue
4999 * Since it's a dataplane function, no check is performed on port_id and
5000 * queue_id. The caller must therefore ensure that the port is enabled
5001 * and the queue is configured and running.
5003 * Note: accessing to a random descriptor in the ring may trigger cache
5004 * misses and have a performance impact.
5007 * A valid port identifier of the Ethernet device which.
5009 * A valid Tx queue identifier on this port.
5011 * The offset of the descriptor starting from tail (0 is the place where
5012 * the next packet will be send).
5015 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
5016 * in the transmit queue.
5017 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
5018 * be reused by the driver.
5019 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
5020 * driver or the hardware.
5021 * - (-EINVAL) bad descriptor offset.
5022 * - (-ENOTSUP) if the device does not support this function.
5023 * - (-ENODEV) bad port or queue (only if compiled with debug).
5025 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
5026 uint16_t queue_id, uint16_t offset)
5028 struct rte_eth_dev *dev;
5031 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5032 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5034 dev = &rte_eth_devices[port_id];
5035 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5036 if (queue_id >= dev->data->nb_tx_queues)
5039 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_descriptor_status, -ENOTSUP);
5040 txq = dev->data->tx_queues[queue_id];
5042 return (*dev->tx_descriptor_status)(txq, offset);
5046 * Send a burst of output packets on a transmit queue of an Ethernet device.
5048 * The rte_eth_tx_burst() function is invoked to transmit output packets
5049 * on the output queue *queue_id* of the Ethernet device designated by its
5051 * The *nb_pkts* parameter is the number of packets to send which are
5052 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5053 * allocated from a pool created with rte_pktmbuf_pool_create().
5054 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
5055 * up to the number of transmit descriptors available in the TX ring of the
5057 * For each packet to send, the rte_eth_tx_burst() function performs
5058 * the following operations:
5060 * - Pick up the next available descriptor in the transmit ring.
5062 * - Free the network buffer previously sent with that descriptor, if any.
5064 * - Initialize the transmit descriptor with the information provided
5065 * in the *rte_mbuf data structure.
5067 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
5068 * the rte_eth_tx_burst() function uses several transmit descriptors
5071 * The rte_eth_tx_burst() function returns the number of packets it
5072 * actually sent. A return value equal to *nb_pkts* means that all packets
5073 * have been sent, and this is likely to signify that other output packets
5074 * could be immediately transmitted again. Applications that implement a
5075 * "send as many packets to transmit as possible" policy can check this
5076 * specific case and keep invoking the rte_eth_tx_burst() function until
5077 * a value less than *nb_pkts* is returned.
5079 * It is the responsibility of the rte_eth_tx_burst() function to
5080 * transparently free the memory buffers of packets previously sent.
5081 * This feature is driven by the *tx_free_thresh* value supplied to the
5082 * rte_eth_dev_configure() function at device configuration time.
5083 * When the number of free TX descriptors drops below this threshold, the
5084 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
5085 * of those packets whose transmission was effectively completed.
5087 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
5088 * invoke this function concurrently on the same tx queue without SW lock.
5089 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
5091 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
5095 * The port identifier of the Ethernet device.
5097 * The index of the transmit queue through which output packets must be
5099 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5100 * to rte_eth_dev_configure().
5102 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5103 * which contain the output packets.
5105 * The maximum number of packets to transmit.
5107 * The number of output packets actually stored in transmit descriptors of
5108 * the transmit ring. The return value can be less than the value of the
5109 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
5111 static inline uint16_t
5112 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
5113 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5115 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
5117 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5118 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
5119 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
5121 if (queue_id >= dev->data->nb_tx_queues) {
5122 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5127 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
5128 struct rte_eth_rxtx_callback *cb;
5130 /* __ATOMIC_RELEASE memory order was used when the
5131 * call back was inserted into the list.
5132 * Since there is a clear dependency between loading
5133 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5136 cb = __atomic_load_n(&dev->pre_tx_burst_cbs[queue_id],
5139 if (unlikely(cb != NULL)) {
5141 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
5144 } while (cb != NULL);
5148 rte_ethdev_trace_tx_burst(port_id, queue_id, (void **)tx_pkts,
5150 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
5154 * Process a burst of output packets on a transmit queue of an Ethernet device.
5156 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
5157 * transmitted on the output queue *queue_id* of the Ethernet device designated
5159 * The *nb_pkts* parameter is the number of packets to be prepared which are
5160 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5161 * allocated from a pool created with rte_pktmbuf_pool_create().
5162 * For each packet to send, the rte_eth_tx_prepare() function performs
5163 * the following operations:
5165 * - Check if packet meets devices requirements for tx offloads.
5167 * - Check limitations about number of segments.
5169 * - Check additional requirements when debug is enabled.
5171 * - Update and/or reset required checksums when tx offload is set for packet.
5173 * Since this function can modify packet data, provided mbufs must be safely
5174 * writable (e.g. modified data cannot be in shared segment).
5176 * The rte_eth_tx_prepare() function returns the number of packets ready to be
5177 * sent. A return value equal to *nb_pkts* means that all packets are valid and
5178 * ready to be sent, otherwise stops processing on the first invalid packet and
5179 * leaves the rest packets untouched.
5181 * When this functionality is not implemented in the driver, all packets are
5182 * are returned untouched.
5185 * The port identifier of the Ethernet device.
5186 * The value must be a valid port id.
5188 * The index of the transmit queue through which output packets must be
5190 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5191 * to rte_eth_dev_configure().
5193 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5194 * which contain the output packets.
5196 * The maximum number of packets to process.
5198 * The number of packets correct and ready to be sent. The return value can be
5199 * less than the value of the *tx_pkts* parameter when some packet doesn't
5200 * meet devices requirements with rte_errno set appropriately:
5201 * - EINVAL: offload flags are not correctly set
5202 * - ENOTSUP: the offload feature is not supported by the hardware
5203 * - ENODEV: if *port_id* is invalid (with debug enabled only)
5207 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
5209 static inline uint16_t
5210 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
5211 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5213 struct rte_eth_dev *dev;
5215 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5216 if (!rte_eth_dev_is_valid_port(port_id)) {
5217 RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
5223 dev = &rte_eth_devices[port_id];
5225 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
5226 if (queue_id >= dev->data->nb_tx_queues) {
5227 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5233 if (!dev->tx_pkt_prepare)
5236 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
5243 * Native NOOP operation for compilation targets which doesn't require any
5244 * preparations steps, and functional NOOP may introduce unnecessary performance
5247 * Generally this is not a good idea to turn it on globally and didn't should
5248 * be used if behavior of tx_preparation can change.
5251 static inline uint16_t
5252 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
5253 __rte_unused uint16_t queue_id,
5254 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5262 * Send any packets queued up for transmission on a port and HW queue
5264 * This causes an explicit flush of packets previously buffered via the
5265 * rte_eth_tx_buffer() function. It returns the number of packets successfully
5266 * sent to the NIC, and calls the error callback for any unsent packets. Unless
5267 * explicitly set up otherwise, the default callback simply frees the unsent
5268 * packets back to the owning mempool.
5271 * The port identifier of the Ethernet device.
5273 * The index of the transmit queue through which output packets must be
5275 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5276 * to rte_eth_dev_configure().
5278 * Buffer of packets to be transmit.
5280 * The number of packets successfully sent to the Ethernet device. The error
5281 * callback is called for any packets which could not be sent.
5283 static inline uint16_t
5284 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
5285 struct rte_eth_dev_tx_buffer *buffer)
5288 uint16_t to_send = buffer->length;
5293 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
5297 /* All packets sent, or to be dealt with by callback below */
5298 if (unlikely(sent != to_send))
5299 buffer->error_callback(&buffer->pkts[sent],
5300 (uint16_t)(to_send - sent),
5301 buffer->error_userdata);
5307 * Buffer a single packet for future transmission on a port and queue
5309 * This function takes a single mbuf/packet and buffers it for later
5310 * transmission on the particular port and queue specified. Once the buffer is
5311 * full of packets, an attempt will be made to transmit all the buffered
5312 * packets. In case of error, where not all packets can be transmitted, a
5313 * callback is called with the unsent packets as a parameter. If no callback
5314 * is explicitly set up, the unsent packets are just freed back to the owning
5315 * mempool. The function returns the number of packets actually sent i.e.
5316 * 0 if no buffer flush occurred, otherwise the number of packets successfully
5320 * The port identifier of the Ethernet device.
5322 * The index of the transmit queue through which output packets must be
5324 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5325 * to rte_eth_dev_configure().
5327 * Buffer used to collect packets to be sent.
5329 * Pointer to the packet mbuf to be sent.
5331 * 0 = packet has been buffered for later transmission
5332 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
5333 * causing N packets to be sent, and the error callback to be called for
5336 static __rte_always_inline uint16_t
5337 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
5338 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
5340 buffer->pkts[buffer->length++] = tx_pkt;
5341 if (buffer->length < buffer->size)
5344 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
5351 #endif /* _RTE_ETHDEV_H_ */