1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2010-2017 Intel Corporation
11 * RTE Ethernet Device API
13 * The Ethernet Device API is composed of two parts:
15 * - The application-oriented Ethernet API that includes functions to setup
16 * an Ethernet device (configure it, setup its RX and TX queues and start it),
17 * to get its MAC address, the speed and the status of its physical link,
18 * to receive and to transmit packets, and so on.
20 * - The driver-oriented Ethernet API that exports functions allowing
21 * an Ethernet Poll Mode Driver (PMD) to allocate an Ethernet device instance,
22 * create memzone for HW rings and process registered callbacks, and so on.
23 * PMDs should include ethdev_driver.h instead of this header.
25 * By default, all the functions of the Ethernet Device API exported by a PMD
26 * are lock-free functions which assume to not be invoked in parallel on
27 * different logical cores to work on the same target object. For instance,
28 * the receive function of a PMD cannot be invoked in parallel on two logical
29 * cores to poll the same RX queue [of the same port]. Of course, this function
30 * can be invoked in parallel by different logical cores on different RX queues.
31 * It is the responsibility of the upper level application to enforce this rule.
33 * If needed, parallel accesses by multiple logical cores to shared queues
34 * shall be explicitly protected by dedicated inline lock-aware functions
35 * built on top of their corresponding lock-free functions of the PMD API.
37 * In all functions of the Ethernet API, the Ethernet device is
38 * designated by an integer >= 0 named the device port identifier.
40 * At the Ethernet driver level, Ethernet devices are represented by a generic
41 * data structure of type *rte_eth_dev*.
43 * Ethernet devices are dynamically registered during the PCI probing phase
44 * performed at EAL initialization time.
45 * When an Ethernet device is being probed, an *rte_eth_dev* structure and
46 * a new port identifier are allocated for that device. Then, the eth_dev_init()
47 * function supplied by the Ethernet driver matching the probed PCI
48 * device is invoked to properly initialize the device.
50 * The role of the device init function consists of resetting the hardware,
51 * checking access to Non-volatile Memory (NVM), reading the MAC address
54 * If the device init operation is successful, the correspondence between
55 * the port identifier assigned to the new device and its associated
56 * *rte_eth_dev* structure is effectively registered.
57 * Otherwise, both the *rte_eth_dev* structure and the port identifier are
60 * The functions exported by the application Ethernet API to setup a device
61 * designated by its port identifier must be invoked in the following order:
62 * - rte_eth_dev_configure()
63 * - rte_eth_tx_queue_setup()
64 * - rte_eth_rx_queue_setup()
65 * - rte_eth_dev_start()
67 * Then, the network application can invoke, in any order, the functions
68 * exported by the Ethernet API to get the MAC address of a given device, to
69 * get the speed and the status of a device physical link, to receive/transmit
70 * [burst of] packets, and so on.
72 * If the application wants to change the configuration (i.e. call
73 * rte_eth_dev_configure(), rte_eth_tx_queue_setup(), or
74 * rte_eth_rx_queue_setup()), it must call rte_eth_dev_stop() first to stop the
75 * device and then do the reconfiguration before calling rte_eth_dev_start()
76 * again. The transmit and receive functions should not be invoked when the
79 * Please note that some configuration is not stored between calls to
80 * rte_eth_dev_stop()/rte_eth_dev_start(). The following configuration will
84 * - flow control settings
85 * - receive mode configuration (promiscuous mode, all-multicast mode,
86 * hardware checksum mode, RSS/VMDQ settings etc.)
87 * - VLAN filtering configuration
88 * - default MAC address
89 * - MAC addresses supplied to MAC address array
90 * - flow director filtering mode (but not filtering rules)
91 * - NIC queue statistics mappings
93 * Any other configuration will not be stored and will need to be re-entered
94 * before a call to rte_eth_dev_start().
96 * Finally, a network application can close an Ethernet device by invoking the
97 * rte_eth_dev_close() function.
99 * Each function of the application Ethernet API invokes a specific function
100 * of the PMD that controls the target device designated by its port
102 * For this purpose, all device-specific functions of an Ethernet driver are
103 * supplied through a set of pointers contained in a generic structure of type
105 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev*
106 * structure by the device init function of the Ethernet driver, which is
107 * invoked during the PCI probing phase, as explained earlier.
109 * In other words, each function of the Ethernet API simply retrieves the
110 * *rte_eth_dev* structure associated with the device port identifier and
111 * performs an indirect invocation of the corresponding driver function
112 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure.
114 * For performance reasons, the address of the burst-oriented RX and TX
115 * functions of the Ethernet driver are not contained in the *eth_dev_ops*
116 * structure. Instead, they are directly stored at the beginning of the
117 * *rte_eth_dev* structure to avoid an extra indirect memory access during
120 * RTE ethernet device drivers do not use interrupts for transmitting or
121 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit
122 * functions to applications.
123 * Both receive and transmit functions are packet-burst oriented to minimize
124 * their cost per packet through the following optimizations:
126 * - Sharing among multiple packets the incompressible cost of the
127 * invocation of receive/transmit functions.
129 * - Enabling receive/transmit functions to take advantage of burst-oriented
130 * hardware features (L1 cache, prefetch instructions, NIC head/tail
131 * registers) to minimize the number of CPU cycles per packet, for instance,
132 * by avoiding useless read memory accesses to ring descriptors, or by
133 * systematically using arrays of pointers that exactly fit L1 cache line
134 * boundaries and sizes.
136 * The burst-oriented receive function does not provide any error notification,
137 * to avoid the corresponding overhead. As a hint, the upper-level application
138 * might check the status of the device link once being systematically returned
139 * a 0 value by the receive function of the driver for a given number of tries.
148 /* Use this macro to check if LRO API is supported */
149 #define RTE_ETHDEV_HAS_LRO_SUPPORT
151 /* Alias RTE_LIBRTE_ETHDEV_DEBUG for backward compatibility. */
152 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
153 #define RTE_ETHDEV_DEBUG_RX
154 #define RTE_ETHDEV_DEBUG_TX
157 #include <rte_compat.h>
159 #include <rte_interrupts.h>
161 #include <rte_devargs.h>
162 #include <rte_errno.h>
163 #include <rte_common.h>
164 #include <rte_config.h>
165 #include <rte_ether.h>
166 #include <rte_power_intrinsics.h>
168 #include "rte_ethdev_trace_fp.h"
169 #include "rte_dev_info.h"
171 extern int rte_eth_dev_logtype;
173 #define RTE_ETHDEV_LOG(level, ...) \
174 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
179 * Initializes a device iterator.
181 * This iterator allows accessing a list of devices matching some devargs.
184 * Device iterator handle initialized by the function.
185 * The fields bus_str and cls_str might be dynamically allocated,
186 * and could be freed by calling rte_eth_iterator_cleanup().
189 * Device description string.
192 * 0 on successful initialization, negative otherwise.
194 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs);
197 * Iterates on devices with devargs filter.
198 * The ownership is not checked.
200 * The next port id is returned, and the iterator is updated.
203 * Device iterator handle initialized by rte_eth_iterator_init().
204 * Some fields bus_str and cls_str might be freed when no more port is found,
205 * by calling rte_eth_iterator_cleanup().
208 * A port id if found, RTE_MAX_ETHPORTS otherwise.
210 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter);
213 * Free some allocated fields of the iterator.
215 * This function is automatically called by rte_eth_iterator_next()
216 * on the last iteration (i.e. when no more matching port is found).
218 * It is safe to call this function twice; it will do nothing more.
221 * Device iterator handle initialized by rte_eth_iterator_init().
222 * The fields bus_str and cls_str are freed if needed.
224 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter);
227 * Macro to iterate over all ethdev ports matching some devargs.
229 * If a break is done before the end of the loop,
230 * the function rte_eth_iterator_cleanup() must be called.
233 * Iterated port id of type uint16_t.
235 * Device parameters input as string of type char*.
237 * Iterator handle of type struct rte_dev_iterator, used internally.
239 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \
240 for (rte_eth_iterator_init(iter, devargs), \
241 id = rte_eth_iterator_next(iter); \
242 id != RTE_MAX_ETHPORTS; \
243 id = rte_eth_iterator_next(iter))
246 * A structure used to retrieve statistics for an Ethernet port.
247 * Not all statistics fields in struct rte_eth_stats are supported
248 * by any type of network interface card (NIC). If any statistics
249 * field is not supported, its value is 0.
250 * All byte-related statistics do not include Ethernet FCS regardless
251 * of whether these bytes have been delivered to the application
252 * (see DEV_RX_OFFLOAD_KEEP_CRC).
254 struct rte_eth_stats {
255 uint64_t ipackets; /**< Total number of successfully received packets. */
256 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
257 uint64_t ibytes; /**< Total number of successfully received bytes. */
258 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
260 /**< Total of RX packets dropped by the HW,
261 * because there are no available buffer (i.e. RX queues are full).
263 uint64_t ierrors; /**< Total number of erroneous received packets. */
264 uint64_t oerrors; /**< Total number of failed transmitted packets. */
265 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
266 /* Queue stats are limited to max 256 queues */
267 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
268 /**< Total number of queue RX packets. */
269 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
270 /**< Total number of queue TX packets. */
271 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
272 /**< Total number of successfully received queue bytes. */
273 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
274 /**< Total number of successfully transmitted queue bytes. */
275 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
276 /**< Total number of queue packets received that are dropped. */
280 * Device supported speeds bitmap flags
282 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
283 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
284 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
285 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
286 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
287 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
288 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
289 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
290 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
291 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
292 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
293 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
294 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
295 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
296 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
297 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
298 #define ETH_LINK_SPEED_200G (1 << 15) /**< 200 Gbps */
301 * Ethernet numeric link speeds in Mbps
303 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
304 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
305 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
306 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
307 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
308 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
309 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
310 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
311 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
312 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
313 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
314 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
315 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
316 #define ETH_SPEED_NUM_200G 200000 /**< 200 Gbps */
317 #define ETH_SPEED_NUM_UNKNOWN UINT32_MAX /**< Unknown */
320 * A structure used to retrieve link-level information of an Ethernet port.
323 struct rte_eth_link {
324 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
325 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
326 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
327 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
328 } __rte_aligned(8); /**< aligned for atomic64 read/write */
330 /* Utility constants */
331 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
332 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
333 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
334 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
335 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
336 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
337 #define RTE_ETH_LINK_MAX_STR_LEN 40 /**< Max length of default link string. */
340 * A structure used to configure the ring threshold registers of an RX/TX
341 * queue for an Ethernet port.
343 struct rte_eth_thresh {
344 uint8_t pthresh; /**< Ring prefetch threshold. */
345 uint8_t hthresh; /**< Ring host threshold. */
346 uint8_t wthresh; /**< Ring writeback threshold. */
350 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
352 #define ETH_MQ_RX_RSS_FLAG 0x1
353 #define ETH_MQ_RX_DCB_FLAG 0x2
354 #define ETH_MQ_RX_VMDQ_FLAG 0x4
357 * A set of values to identify what method is to be used to route
358 * packets to multiple queues.
360 enum rte_eth_rx_mq_mode {
361 /** None of DCB,RSS or VMDQ mode */
364 /** For RX side, only RSS is on */
365 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
366 /** For RX side,only DCB is on. */
367 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
368 /** Both DCB and RSS enable */
369 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
371 /** Only VMDQ, no RSS nor DCB */
372 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
373 /** RSS mode with VMDQ */
374 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
375 /** Use VMDQ+DCB to route traffic to queues */
376 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
377 /** Enable both VMDQ and DCB in VMDq */
378 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
383 * for rx mq mode backward compatible
385 #define ETH_RSS ETH_MQ_RX_RSS
386 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
387 #define ETH_DCB_RX ETH_MQ_RX_DCB
390 * A set of values to identify what method is to be used to transmit
391 * packets using multi-TCs.
393 enum rte_eth_tx_mq_mode {
394 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
395 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
396 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
397 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
401 * for tx mq mode backward compatible
403 #define ETH_DCB_NONE ETH_MQ_TX_NONE
404 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
405 #define ETH_DCB_TX ETH_MQ_TX_DCB
408 * A structure used to configure the RX features of an Ethernet port.
410 struct rte_eth_rxmode {
411 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
412 enum rte_eth_rx_mq_mode mq_mode;
413 uint32_t max_rx_pkt_len; /**< Only used if JUMBO_FRAME enabled. */
414 /** Maximum allowed size of LRO aggregated packet. */
415 uint32_t max_lro_pkt_size;
416 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
418 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
419 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
420 * structure are allowed to be set.
424 uint64_t reserved_64s[2]; /**< Reserved for future fields */
425 void *reserved_ptrs[2]; /**< Reserved for future fields */
429 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
430 * Note that single VLAN is treated the same as inner VLAN.
433 ETH_VLAN_TYPE_UNKNOWN = 0,
434 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
435 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
440 * A structure used to describe a vlan filter.
441 * If the bit corresponding to a VID is set, such VID is on.
443 struct rte_vlan_filter_conf {
448 * A structure used to configure the Receive Side Scaling (RSS) feature
449 * of an Ethernet port.
450 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
451 * to an array holding the RSS key to use for hashing specific header
452 * fields of received packets. The length of this array should be indicated
453 * by *rss_key_len* below. Otherwise, a default random hash key is used by
456 * The *rss_key_len* field of the *rss_conf* structure indicates the length
457 * in bytes of the array pointed by *rss_key*. To be compatible, this length
458 * will be checked in i40e only. Others assume 40 bytes to be used as before.
460 * The *rss_hf* field of the *rss_conf* structure indicates the different
461 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
462 * Supplying an *rss_hf* equal to zero disables the RSS feature.
464 struct rte_eth_rss_conf {
465 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
466 uint8_t rss_key_len; /**< hash key length in bytes. */
467 uint64_t rss_hf; /**< Hash functions to apply - see below. */
471 * A packet can be identified by hardware as different flow types. Different
472 * NIC hardware may support different flow types.
473 * Basically, the NIC hardware identifies the flow type as deep protocol as
474 * possible, and exclusively. For example, if a packet is identified as
475 * 'RTE_ETH_FLOW_NONFRAG_IPV4_TCP', it will not be any of other flow types,
476 * though it is an actual IPV4 packet.
478 #define RTE_ETH_FLOW_UNKNOWN 0
479 #define RTE_ETH_FLOW_RAW 1
480 #define RTE_ETH_FLOW_IPV4 2
481 #define RTE_ETH_FLOW_FRAG_IPV4 3
482 #define RTE_ETH_FLOW_NONFRAG_IPV4_TCP 4
483 #define RTE_ETH_FLOW_NONFRAG_IPV4_UDP 5
484 #define RTE_ETH_FLOW_NONFRAG_IPV4_SCTP 6
485 #define RTE_ETH_FLOW_NONFRAG_IPV4_OTHER 7
486 #define RTE_ETH_FLOW_IPV6 8
487 #define RTE_ETH_FLOW_FRAG_IPV6 9
488 #define RTE_ETH_FLOW_NONFRAG_IPV6_TCP 10
489 #define RTE_ETH_FLOW_NONFRAG_IPV6_UDP 11
490 #define RTE_ETH_FLOW_NONFRAG_IPV6_SCTP 12
491 #define RTE_ETH_FLOW_NONFRAG_IPV6_OTHER 13
492 #define RTE_ETH_FLOW_L2_PAYLOAD 14
493 #define RTE_ETH_FLOW_IPV6_EX 15
494 #define RTE_ETH_FLOW_IPV6_TCP_EX 16
495 #define RTE_ETH_FLOW_IPV6_UDP_EX 17
496 #define RTE_ETH_FLOW_PORT 18
497 /**< Consider device port number as a flow differentiator */
498 #define RTE_ETH_FLOW_VXLAN 19 /**< VXLAN protocol based flow */
499 #define RTE_ETH_FLOW_GENEVE 20 /**< GENEVE protocol based flow */
500 #define RTE_ETH_FLOW_NVGRE 21 /**< NVGRE protocol based flow */
501 #define RTE_ETH_FLOW_VXLAN_GPE 22 /**< VXLAN-GPE protocol based flow */
502 #define RTE_ETH_FLOW_GTPU 23 /**< GTPU protocol based flow */
503 #define RTE_ETH_FLOW_MAX 24
506 * Below macros are defined for RSS offload types, they can be used to
507 * fill rte_eth_rss_conf.rss_hf or rte_flow_action_rss.types.
509 #define ETH_RSS_IPV4 (1ULL << 2)
510 #define ETH_RSS_FRAG_IPV4 (1ULL << 3)
511 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << 4)
512 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << 5)
513 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << 6)
514 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << 7)
515 #define ETH_RSS_IPV6 (1ULL << 8)
516 #define ETH_RSS_FRAG_IPV6 (1ULL << 9)
517 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << 10)
518 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << 11)
519 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << 12)
520 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << 13)
521 #define ETH_RSS_L2_PAYLOAD (1ULL << 14)
522 #define ETH_RSS_IPV6_EX (1ULL << 15)
523 #define ETH_RSS_IPV6_TCP_EX (1ULL << 16)
524 #define ETH_RSS_IPV6_UDP_EX (1ULL << 17)
525 #define ETH_RSS_PORT (1ULL << 18)
526 #define ETH_RSS_VXLAN (1ULL << 19)
527 #define ETH_RSS_GENEVE (1ULL << 20)
528 #define ETH_RSS_NVGRE (1ULL << 21)
529 #define ETH_RSS_GTPU (1ULL << 23)
530 #define ETH_RSS_ETH (1ULL << 24)
531 #define ETH_RSS_S_VLAN (1ULL << 25)
532 #define ETH_RSS_C_VLAN (1ULL << 26)
533 #define ETH_RSS_ESP (1ULL << 27)
534 #define ETH_RSS_AH (1ULL << 28)
535 #define ETH_RSS_L2TPV3 (1ULL << 29)
536 #define ETH_RSS_PFCP (1ULL << 30)
537 #define ETH_RSS_PPPOE (1ULL << 31)
538 #define ETH_RSS_ECPRI (1ULL << 32)
539 #define ETH_RSS_MPLS (1ULL << 33)
540 #define ETH_RSS_IPV4_CHKSUM (1ULL << 34)
543 * The ETH_RSS_L4_CHKSUM works on checksum field of any L4 header.
544 * It is similar to ETH_RSS_PORT that they don't specify the specific type of
545 * L4 header. This macro is defined to replace some specific L4 (TCP/UDP/SCTP)
546 * checksum type for constructing the use of RSS offload bits.
548 * Due to above reason, some old APIs (and configuration) don't support
549 * ETH_RSS_L4_CHKSUM. The rte_flow RSS API supports it.
551 * For the case that checksum is not used in an UDP header,
552 * it takes the reserved value 0 as input for the hash function.
554 #define ETH_RSS_L4_CHKSUM (1ULL << 35)
557 * We use the following macros to combine with above ETH_RSS_* for
558 * more specific input set selection. These bits are defined starting
559 * from the high end of the 64 bits.
560 * Note: If we use above ETH_RSS_* without SRC/DST_ONLY, it represents
561 * both SRC and DST are taken into account. If SRC_ONLY and DST_ONLY of
562 * the same level are used simultaneously, it is the same case as none of
565 #define ETH_RSS_L3_SRC_ONLY (1ULL << 63)
566 #define ETH_RSS_L3_DST_ONLY (1ULL << 62)
567 #define ETH_RSS_L4_SRC_ONLY (1ULL << 61)
568 #define ETH_RSS_L4_DST_ONLY (1ULL << 60)
569 #define ETH_RSS_L2_SRC_ONLY (1ULL << 59)
570 #define ETH_RSS_L2_DST_ONLY (1ULL << 58)
573 * Only select IPV6 address prefix as RSS input set according to
574 * https://tools.ietf.org/html/rfc6052
575 * Must be combined with ETH_RSS_IPV6, ETH_RSS_NONFRAG_IPV6_UDP,
576 * ETH_RSS_NONFRAG_IPV6_TCP, ETH_RSS_NONFRAG_IPV6_SCTP.
578 #define RTE_ETH_RSS_L3_PRE32 (1ULL << 57)
579 #define RTE_ETH_RSS_L3_PRE40 (1ULL << 56)
580 #define RTE_ETH_RSS_L3_PRE48 (1ULL << 55)
581 #define RTE_ETH_RSS_L3_PRE56 (1ULL << 54)
582 #define RTE_ETH_RSS_L3_PRE64 (1ULL << 53)
583 #define RTE_ETH_RSS_L3_PRE96 (1ULL << 52)
586 * Use the following macros to combine with the above layers
587 * to choose inner and outer layers or both for RSS computation.
588 * Bits 50 and 51 are reserved for this.
592 * level 0, requests the default behavior.
593 * Depending on the packet type, it can mean outermost, innermost,
594 * anything in between or even no RSS.
595 * It basically stands for the innermost encapsulation level RSS
596 * can be performed on according to PMD and device capabilities.
598 #define ETH_RSS_LEVEL_PMD_DEFAULT (0ULL << 50)
601 * level 1, requests RSS to be performed on the outermost packet
602 * encapsulation level.
604 #define ETH_RSS_LEVEL_OUTERMOST (1ULL << 50)
607 * level 2, requests RSS to be performed on the specified inner packet
608 * encapsulation level, from outermost to innermost (lower to higher values).
610 #define ETH_RSS_LEVEL_INNERMOST (2ULL << 50)
611 #define ETH_RSS_LEVEL_MASK (3ULL << 50)
613 #define ETH_RSS_LEVEL(rss_hf) ((rss_hf & ETH_RSS_LEVEL_MASK) >> 50)
616 * For input set change of hash filter, if SRC_ONLY and DST_ONLY of
617 * the same level are used simultaneously, it is the same case as
618 * none of them are added.
621 * RSS types with SRC/DST_ONLY.
625 static inline uint64_t
626 rte_eth_rss_hf_refine(uint64_t rss_hf)
628 if ((rss_hf & ETH_RSS_L3_SRC_ONLY) && (rss_hf & ETH_RSS_L3_DST_ONLY))
629 rss_hf &= ~(ETH_RSS_L3_SRC_ONLY | ETH_RSS_L3_DST_ONLY);
631 if ((rss_hf & ETH_RSS_L4_SRC_ONLY) && (rss_hf & ETH_RSS_L4_DST_ONLY))
632 rss_hf &= ~(ETH_RSS_L4_SRC_ONLY | ETH_RSS_L4_DST_ONLY);
637 #define ETH_RSS_IPV6_PRE32 ( \
639 RTE_ETH_RSS_L3_PRE32)
641 #define ETH_RSS_IPV6_PRE40 ( \
643 RTE_ETH_RSS_L3_PRE40)
645 #define ETH_RSS_IPV6_PRE48 ( \
647 RTE_ETH_RSS_L3_PRE48)
649 #define ETH_RSS_IPV6_PRE56 ( \
651 RTE_ETH_RSS_L3_PRE56)
653 #define ETH_RSS_IPV6_PRE64 ( \
655 RTE_ETH_RSS_L3_PRE64)
657 #define ETH_RSS_IPV6_PRE96 ( \
659 RTE_ETH_RSS_L3_PRE96)
661 #define ETH_RSS_IPV6_PRE32_UDP ( \
662 ETH_RSS_NONFRAG_IPV6_UDP | \
663 RTE_ETH_RSS_L3_PRE32)
665 #define ETH_RSS_IPV6_PRE40_UDP ( \
666 ETH_RSS_NONFRAG_IPV6_UDP | \
667 RTE_ETH_RSS_L3_PRE40)
669 #define ETH_RSS_IPV6_PRE48_UDP ( \
670 ETH_RSS_NONFRAG_IPV6_UDP | \
671 RTE_ETH_RSS_L3_PRE48)
673 #define ETH_RSS_IPV6_PRE56_UDP ( \
674 ETH_RSS_NONFRAG_IPV6_UDP | \
675 RTE_ETH_RSS_L3_PRE56)
677 #define ETH_RSS_IPV6_PRE64_UDP ( \
678 ETH_RSS_NONFRAG_IPV6_UDP | \
679 RTE_ETH_RSS_L3_PRE64)
681 #define ETH_RSS_IPV6_PRE96_UDP ( \
682 ETH_RSS_NONFRAG_IPV6_UDP | \
683 RTE_ETH_RSS_L3_PRE96)
685 #define ETH_RSS_IPV6_PRE32_TCP ( \
686 ETH_RSS_NONFRAG_IPV6_TCP | \
687 RTE_ETH_RSS_L3_PRE32)
689 #define ETH_RSS_IPV6_PRE40_TCP ( \
690 ETH_RSS_NONFRAG_IPV6_TCP | \
691 RTE_ETH_RSS_L3_PRE40)
693 #define ETH_RSS_IPV6_PRE48_TCP ( \
694 ETH_RSS_NONFRAG_IPV6_TCP | \
695 RTE_ETH_RSS_L3_PRE48)
697 #define ETH_RSS_IPV6_PRE56_TCP ( \
698 ETH_RSS_NONFRAG_IPV6_TCP | \
699 RTE_ETH_RSS_L3_PRE56)
701 #define ETH_RSS_IPV6_PRE64_TCP ( \
702 ETH_RSS_NONFRAG_IPV6_TCP | \
703 RTE_ETH_RSS_L3_PRE64)
705 #define ETH_RSS_IPV6_PRE96_TCP ( \
706 ETH_RSS_NONFRAG_IPV6_TCP | \
707 RTE_ETH_RSS_L3_PRE96)
709 #define ETH_RSS_IPV6_PRE32_SCTP ( \
710 ETH_RSS_NONFRAG_IPV6_SCTP | \
711 RTE_ETH_RSS_L3_PRE32)
713 #define ETH_RSS_IPV6_PRE40_SCTP ( \
714 ETH_RSS_NONFRAG_IPV6_SCTP | \
715 RTE_ETH_RSS_L3_PRE40)
717 #define ETH_RSS_IPV6_PRE48_SCTP ( \
718 ETH_RSS_NONFRAG_IPV6_SCTP | \
719 RTE_ETH_RSS_L3_PRE48)
721 #define ETH_RSS_IPV6_PRE56_SCTP ( \
722 ETH_RSS_NONFRAG_IPV6_SCTP | \
723 RTE_ETH_RSS_L3_PRE56)
725 #define ETH_RSS_IPV6_PRE64_SCTP ( \
726 ETH_RSS_NONFRAG_IPV6_SCTP | \
727 RTE_ETH_RSS_L3_PRE64)
729 #define ETH_RSS_IPV6_PRE96_SCTP ( \
730 ETH_RSS_NONFRAG_IPV6_SCTP | \
731 RTE_ETH_RSS_L3_PRE96)
733 #define ETH_RSS_IP ( \
735 ETH_RSS_FRAG_IPV4 | \
736 ETH_RSS_NONFRAG_IPV4_OTHER | \
738 ETH_RSS_FRAG_IPV6 | \
739 ETH_RSS_NONFRAG_IPV6_OTHER | \
742 #define ETH_RSS_UDP ( \
743 ETH_RSS_NONFRAG_IPV4_UDP | \
744 ETH_RSS_NONFRAG_IPV6_UDP | \
747 #define ETH_RSS_TCP ( \
748 ETH_RSS_NONFRAG_IPV4_TCP | \
749 ETH_RSS_NONFRAG_IPV6_TCP | \
752 #define ETH_RSS_SCTP ( \
753 ETH_RSS_NONFRAG_IPV4_SCTP | \
754 ETH_RSS_NONFRAG_IPV6_SCTP)
756 #define ETH_RSS_TUNNEL ( \
761 #define ETH_RSS_VLAN ( \
765 /**< Mask of valid RSS hash protocols */
766 #define ETH_RSS_PROTO_MASK ( \
768 ETH_RSS_FRAG_IPV4 | \
769 ETH_RSS_NONFRAG_IPV4_TCP | \
770 ETH_RSS_NONFRAG_IPV4_UDP | \
771 ETH_RSS_NONFRAG_IPV4_SCTP | \
772 ETH_RSS_NONFRAG_IPV4_OTHER | \
774 ETH_RSS_FRAG_IPV6 | \
775 ETH_RSS_NONFRAG_IPV6_TCP | \
776 ETH_RSS_NONFRAG_IPV6_UDP | \
777 ETH_RSS_NONFRAG_IPV6_SCTP | \
778 ETH_RSS_NONFRAG_IPV6_OTHER | \
779 ETH_RSS_L2_PAYLOAD | \
781 ETH_RSS_IPV6_TCP_EX | \
782 ETH_RSS_IPV6_UDP_EX | \
790 * Definitions used for redirection table entry size.
791 * Some RSS RETA sizes may not be supported by some drivers, check the
792 * documentation or the description of relevant functions for more details.
794 #define ETH_RSS_RETA_SIZE_64 64
795 #define ETH_RSS_RETA_SIZE_128 128
796 #define ETH_RSS_RETA_SIZE_256 256
797 #define ETH_RSS_RETA_SIZE_512 512
798 #define RTE_RETA_GROUP_SIZE 64
800 /* Definitions used for VMDQ and DCB functionality */
801 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
802 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
803 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
804 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
806 /* DCB capability defines */
807 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
808 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
810 /* Definitions used for VLAN Offload functionality */
811 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
812 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
813 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
814 #define ETH_QINQ_STRIP_OFFLOAD 0x0008 /**< QINQ Strip On/Off */
816 /* Definitions used for mask VLAN setting */
817 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
818 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
819 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
820 #define ETH_QINQ_STRIP_MASK 0x0008 /**< QINQ Strip setting mask */
821 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
823 /* Definitions used for receive MAC address */
824 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
826 /* Definitions used for unicast hash */
827 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
829 /* Definitions used for VMDQ pool rx mode setting */
830 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
831 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
832 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
833 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
834 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
836 /** Maximum nb. of vlan per mirror rule */
837 #define ETH_MIRROR_MAX_VLANS 64
839 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
840 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
841 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
842 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
843 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
846 * A structure used to configure VLAN traffic mirror of an Ethernet port.
848 struct rte_eth_vlan_mirror {
849 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
850 /** VLAN ID list for vlan mirroring. */
851 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
855 * A structure used to configure traffic mirror of an Ethernet port.
857 struct rte_eth_mirror_conf {
858 uint8_t rule_type; /**< Mirroring rule type */
859 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
860 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
861 /** VLAN ID setting for VLAN mirroring. */
862 struct rte_eth_vlan_mirror vlan;
866 * A structure used to configure 64 entries of Redirection Table of the
867 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
868 * more than 64 entries supported by hardware, an array of this structure
871 struct rte_eth_rss_reta_entry64 {
873 /**< Mask bits indicate which entries need to be updated/queried. */
874 uint16_t reta[RTE_RETA_GROUP_SIZE];
875 /**< Group of 64 redirection table entries. */
879 * This enum indicates the possible number of traffic classes
880 * in DCB configurations
882 enum rte_eth_nb_tcs {
883 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
884 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
888 * This enum indicates the possible number of queue pools
889 * in VMDQ configurations.
891 enum rte_eth_nb_pools {
892 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
893 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
894 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
895 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
898 /* This structure may be extended in future. */
899 struct rte_eth_dcb_rx_conf {
900 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
901 /** Traffic class each UP mapped to. */
902 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
905 struct rte_eth_vmdq_dcb_tx_conf {
906 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
907 /** Traffic class each UP mapped to. */
908 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
911 struct rte_eth_dcb_tx_conf {
912 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
913 /** Traffic class each UP mapped to. */
914 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
917 struct rte_eth_vmdq_tx_conf {
918 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
922 * A structure used to configure the VMDQ+DCB feature
923 * of an Ethernet port.
925 * Using this feature, packets are routed to a pool of queues, based
926 * on the vlan id in the vlan tag, and then to a specific queue within
927 * that pool, using the user priority vlan tag field.
929 * A default pool may be used, if desired, to route all traffic which
930 * does not match the vlan filter rules.
932 struct rte_eth_vmdq_dcb_conf {
933 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
934 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
935 uint8_t default_pool; /**< The default pool, if applicable */
936 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
938 uint16_t vlan_id; /**< The vlan id of the received frame */
939 uint64_t pools; /**< Bitmask of pools for packet rx */
940 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
941 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
942 /**< Selects a queue in a pool */
946 * A structure used to configure the VMDQ feature of an Ethernet port when
947 * not combined with the DCB feature.
949 * Using this feature, packets are routed to a pool of queues. By default,
950 * the pool selection is based on the MAC address, the vlan id in the
951 * vlan tag as specified in the pool_map array.
952 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
953 * selection using only the MAC address. MAC address to pool mapping is done
954 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
955 * corresponding to the pool id.
957 * Queue selection within the selected pool will be done using RSS when
958 * it is enabled or revert to the first queue of the pool if not.
960 * A default pool may be used, if desired, to route all traffic which
961 * does not match the vlan filter rules or any pool MAC address.
963 struct rte_eth_vmdq_rx_conf {
964 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
965 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
966 uint8_t default_pool; /**< The default pool, if applicable */
967 uint8_t enable_loop_back; /**< Enable VT loop back */
968 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
969 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
971 uint16_t vlan_id; /**< The vlan id of the received frame */
972 uint64_t pools; /**< Bitmask of pools for packet rx */
973 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
977 * A structure used to configure the TX features of an Ethernet port.
979 struct rte_eth_txmode {
980 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
982 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
983 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
984 * structure are allowed to be set.
990 uint8_t hw_vlan_reject_tagged : 1,
991 /**< If set, reject sending out tagged pkts */
992 hw_vlan_reject_untagged : 1,
993 /**< If set, reject sending out untagged pkts */
994 hw_vlan_insert_pvid : 1;
995 /**< If set, enable port based VLAN insertion */
997 uint64_t reserved_64s[2]; /**< Reserved for future fields */
998 void *reserved_ptrs[2]; /**< Reserved for future fields */
1003 * @b EXPERIMENTAL: this structure may change without prior notice.
1005 * A structure used to configure an Rx packet segment to split.
1007 * If RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT flag is set in offloads field,
1008 * the PMD will split the received packets into multiple segments
1009 * according to the specification in the description array:
1011 * - The first network buffer will be allocated from the memory pool,
1012 * specified in the first array element, the second buffer, from the
1013 * pool in the second element, and so on.
1015 * - The offsets from the segment description elements specify
1016 * the data offset from the buffer beginning except the first mbuf.
1017 * The first segment offset is added with RTE_PKTMBUF_HEADROOM.
1019 * - The lengths in the elements define the maximal data amount
1020 * being received to each segment. The receiving starts with filling
1021 * up the first mbuf data buffer up to specified length. If the
1022 * there are data remaining (packet is longer than buffer in the first
1023 * mbuf) the following data will be pushed to the next segment
1024 * up to its own length, and so on.
1026 * - If the length in the segment description element is zero
1027 * the actual buffer size will be deduced from the appropriate
1028 * memory pool properties.
1030 * - If there is not enough elements to describe the buffer for entire
1031 * packet of maximal length the following parameters will be used
1032 * for the all remaining segments:
1033 * - pool from the last valid element
1034 * - the buffer size from this pool
1037 struct rte_eth_rxseg_split {
1038 struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
1039 uint16_t length; /**< Segment data length, configures split point. */
1040 uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */
1041 uint32_t reserved; /**< Reserved field. */
1046 * @b EXPERIMENTAL: this structure may change without prior notice.
1048 * A common structure used to describe Rx packet segment properties.
1050 union rte_eth_rxseg {
1051 /* The settings for buffer split offload. */
1052 struct rte_eth_rxseg_split split;
1053 /* The other features settings should be added here. */
1057 * A structure used to configure an RX ring of an Ethernet port.
1059 struct rte_eth_rxconf {
1060 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
1061 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
1062 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
1063 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1064 uint16_t rx_nseg; /**< Number of descriptions in rx_seg array. */
1066 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
1067 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
1068 * fields on rte_eth_dev_info structure are allowed to be set.
1072 * Points to the array of segment descriptions for an entire packet.
1073 * Array elements are properties for consecutive Rx segments.
1075 * The supported capabilities of receiving segmentation is reported
1076 * in rte_eth_dev_info.rx_seg_capa field.
1078 union rte_eth_rxseg *rx_seg;
1080 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1081 void *reserved_ptrs[2]; /**< Reserved for future fields */
1085 * A structure used to configure a TX ring of an Ethernet port.
1087 struct rte_eth_txconf {
1088 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
1089 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
1090 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
1091 less free descriptors than this value. */
1093 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
1095 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
1096 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
1097 * fields on rte_eth_dev_info structure are allowed to be set.
1101 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1102 void *reserved_ptrs[2]; /**< Reserved for future fields */
1107 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1109 * A structure used to return the hairpin capabilities that are supported.
1111 struct rte_eth_hairpin_cap {
1112 /** The max number of hairpin queues (different bindings). */
1113 uint16_t max_nb_queues;
1114 /** Max number of Rx queues to be connected to one Tx queue. */
1115 uint16_t max_rx_2_tx;
1116 /** Max number of Tx queues to be connected to one Rx queue. */
1117 uint16_t max_tx_2_rx;
1118 uint16_t max_nb_desc; /**< The max num of descriptors. */
1121 #define RTE_ETH_MAX_HAIRPIN_PEERS 32
1125 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1127 * A structure used to hold hairpin peer data.
1129 struct rte_eth_hairpin_peer {
1130 uint16_t port; /**< Peer port. */
1131 uint16_t queue; /**< Peer queue. */
1136 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
1138 * A structure used to configure hairpin binding.
1140 struct rte_eth_hairpin_conf {
1141 uint32_t peer_count:16; /**< The number of peers. */
1144 * Explicit Tx flow rule mode.
1145 * One hairpin pair of queues should have the same attribute.
1147 * - When set, the user should be responsible for inserting the hairpin
1148 * Tx part flows and removing them.
1149 * - When clear, the PMD will try to handle the Tx part of the flows,
1150 * e.g., by splitting one flow into two parts.
1152 uint32_t tx_explicit:1;
1155 * Manually bind hairpin queues.
1156 * One hairpin pair of queues should have the same attribute.
1158 * - When set, to enable hairpin, the user should call the hairpin bind
1159 * function after all the queues are set up properly and the ports are
1160 * started. Also, the hairpin unbind function should be called
1161 * accordingly before stopping a port that with hairpin configured.
1162 * - When clear, the PMD will try to enable the hairpin with the queues
1163 * configured automatically during port start.
1165 uint32_t manual_bind:1;
1166 uint32_t reserved:14; /**< Reserved bits. */
1167 struct rte_eth_hairpin_peer peers[RTE_ETH_MAX_HAIRPIN_PEERS];
1171 * A structure contains information about HW descriptor ring limitations.
1173 struct rte_eth_desc_lim {
1174 uint16_t nb_max; /**< Max allowed number of descriptors. */
1175 uint16_t nb_min; /**< Min allowed number of descriptors. */
1176 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
1179 * Max allowed number of segments per whole packet.
1181 * - For TSO packet this is the total number of data descriptors allowed
1184 * @see nb_mtu_seg_max
1186 uint16_t nb_seg_max;
1189 * Max number of segments per one MTU.
1191 * - For non-TSO packet, this is the maximum allowed number of segments
1192 * in a single transmit packet.
1194 * - For TSO packet each segment within the TSO may span up to this
1199 uint16_t nb_mtu_seg_max;
1203 * This enum indicates the flow control mode
1205 enum rte_eth_fc_mode {
1206 RTE_FC_NONE = 0, /**< Disable flow control. */
1207 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
1208 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
1209 RTE_FC_FULL /**< Enable flow control on both side. */
1213 * A structure used to configure Ethernet flow control parameter.
1214 * These parameters will be configured into the register of the NIC.
1215 * Please refer to the corresponding data sheet for proper value.
1217 struct rte_eth_fc_conf {
1218 uint32_t high_water; /**< High threshold value to trigger XOFF */
1219 uint32_t low_water; /**< Low threshold value to trigger XON */
1220 uint16_t pause_time; /**< Pause quota in the Pause frame */
1221 uint16_t send_xon; /**< Is XON frame need be sent */
1222 enum rte_eth_fc_mode mode; /**< Link flow control mode */
1223 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
1224 uint8_t autoneg; /**< Use Pause autoneg */
1228 * A structure used to configure Ethernet priority flow control parameter.
1229 * These parameters will be configured into the register of the NIC.
1230 * Please refer to the corresponding data sheet for proper value.
1232 struct rte_eth_pfc_conf {
1233 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
1234 uint8_t priority; /**< VLAN User Priority. */
1238 * Tunnel type for device-specific classifier configuration.
1239 * @see rte_eth_udp_tunnel
1241 enum rte_eth_tunnel_type {
1242 RTE_TUNNEL_TYPE_NONE = 0,
1243 RTE_TUNNEL_TYPE_VXLAN,
1244 RTE_TUNNEL_TYPE_GENEVE,
1245 RTE_TUNNEL_TYPE_TEREDO,
1246 RTE_TUNNEL_TYPE_NVGRE,
1247 RTE_TUNNEL_TYPE_IP_IN_GRE,
1248 RTE_L2_TUNNEL_TYPE_E_TAG,
1249 RTE_TUNNEL_TYPE_VXLAN_GPE,
1250 RTE_TUNNEL_TYPE_ECPRI,
1251 RTE_TUNNEL_TYPE_MAX,
1254 /* Deprecated API file for rte_eth_dev_filter_* functions */
1255 #include "rte_eth_ctrl.h"
1258 * Memory space that can be configured to store Flow Director filters
1259 * in the board memory.
1261 enum rte_fdir_pballoc_type {
1262 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
1263 RTE_FDIR_PBALLOC_128K, /**< 128k. */
1264 RTE_FDIR_PBALLOC_256K, /**< 256k. */
1268 * Select report mode of FDIR hash information in RX descriptors.
1270 enum rte_fdir_status_mode {
1271 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
1272 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
1273 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
1277 * A structure used to configure the Flow Director (FDIR) feature
1278 * of an Ethernet port.
1280 * If mode is RTE_FDIR_MODE_NONE, the pballoc value is ignored.
1282 struct rte_fdir_conf {
1283 enum rte_fdir_mode mode; /**< Flow Director mode. */
1284 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
1285 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
1286 /** RX queue of packets matching a "drop" filter in perfect mode. */
1288 struct rte_eth_fdir_masks mask;
1289 struct rte_eth_fdir_flex_conf flex_conf;
1290 /**< Flex payload configuration. */
1294 * UDP tunneling configuration.
1296 * Used to configure the classifier of a device,
1297 * associating an UDP port with a type of tunnel.
1299 * Some NICs may need such configuration to properly parse a tunnel
1300 * with any standard or custom UDP port.
1302 struct rte_eth_udp_tunnel {
1303 uint16_t udp_port; /**< UDP port used for the tunnel. */
1304 uint8_t prot_type; /**< Tunnel type. @see rte_eth_tunnel_type */
1308 * A structure used to enable/disable specific device interrupts.
1310 struct rte_intr_conf {
1311 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
1313 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
1315 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
1320 * A structure used to configure an Ethernet port.
1321 * Depending upon the RX multi-queue mode, extra advanced
1322 * configuration settings may be needed.
1324 struct rte_eth_conf {
1325 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
1326 used. ETH_LINK_SPEED_FIXED disables link
1327 autonegotiation, and a unique speed shall be
1328 set. Otherwise, the bitmap defines the set of
1329 speeds to be advertised. If the special value
1330 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
1331 supported are advertised. */
1332 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
1333 struct rte_eth_txmode txmode; /**< Port TX configuration. */
1334 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
1335 is 0, meaning the loopback mode is disabled.
1336 Read the datasheet of given ethernet controller
1337 for details. The possible values of this field
1338 are defined in implementation of each driver. */
1340 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
1341 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
1342 /**< Port vmdq+dcb configuration. */
1343 struct rte_eth_dcb_rx_conf dcb_rx_conf;
1344 /**< Port dcb RX configuration. */
1345 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
1346 /**< Port vmdq RX configuration. */
1347 } rx_adv_conf; /**< Port RX filtering configuration. */
1349 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
1350 /**< Port vmdq+dcb TX configuration. */
1351 struct rte_eth_dcb_tx_conf dcb_tx_conf;
1352 /**< Port dcb TX configuration. */
1353 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
1354 /**< Port vmdq TX configuration. */
1355 } tx_adv_conf; /**< Port TX DCB configuration (union). */
1356 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
1357 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
1358 uint32_t dcb_capability_en;
1359 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. DEPRECATED */
1360 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
1364 * RX offload capabilities of a device.
1366 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
1367 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
1368 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
1369 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
1370 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
1371 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
1372 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
1373 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
1374 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
1375 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
1376 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
1377 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
1378 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
1380 * Timestamp is set by the driver in RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1381 * and RTE_MBUF_DYNFLAG_RX_TIMESTAMP_NAME is set in ol_flags.
1382 * The mbuf field and flag are registered when the offload is configured.
1384 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
1385 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
1386 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
1387 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
1388 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
1389 #define DEV_RX_OFFLOAD_RSS_HASH 0x00080000
1390 #define RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT 0x00100000
1392 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
1393 DEV_RX_OFFLOAD_UDP_CKSUM | \
1394 DEV_RX_OFFLOAD_TCP_CKSUM)
1395 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
1396 DEV_RX_OFFLOAD_VLAN_FILTER | \
1397 DEV_RX_OFFLOAD_VLAN_EXTEND | \
1398 DEV_RX_OFFLOAD_QINQ_STRIP)
1401 * If new Rx offload capabilities are defined, they also must be
1402 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
1406 * TX offload capabilities of a device.
1408 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
1409 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
1410 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
1411 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
1412 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
1413 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
1414 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
1415 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
1416 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
1417 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
1418 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
1419 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
1420 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
1421 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
1422 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
1423 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
1424 * tx queue without SW lock.
1426 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
1427 /**< Device supports multi segment send. */
1428 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
1429 /**< Device supports optimization for fast release of mbufs.
1430 * When set application must guarantee that per-queue all mbufs comes from
1431 * the same mempool and has refcnt = 1.
1433 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
1435 * Device supports generic UDP tunneled packet TSO.
1436 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
1439 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
1441 * Device supports generic IP tunneled packet TSO.
1442 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
1445 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
1446 /** Device supports outer UDP checksum */
1447 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
1449 * Device sends on time read from RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
1450 * if RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME is set in ol_flags.
1451 * The mbuf field and flag are registered when the offload is configured.
1453 #define DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP 0x00200000
1455 * If new Tx offload capabilities are defined, they also must be
1456 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1459 /**@{@name Device capabilities
1460 * Non-offload capabilities reported in rte_eth_dev_info.dev_capa.
1462 /** Device supports Rx queue setup after device started. */
1463 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
1464 /** Device supports Tx queue setup after device started. */
1465 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
1469 * Fallback default preferred Rx/Tx port parameters.
1470 * These are used if an application requests default parameters
1471 * but the PMD does not provide preferred values.
1473 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1474 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1475 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1476 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1479 * Preferred Rx/Tx port parameters.
1480 * There are separate instances of this structure for transmission
1481 * and reception respectively.
1483 struct rte_eth_dev_portconf {
1484 uint16_t burst_size; /**< Device-preferred burst size */
1485 uint16_t ring_size; /**< Device-preferred size of queue rings */
1486 uint16_t nb_queues; /**< Device-preferred number of queues */
1490 * Default values for switch domain id when ethdev does not support switch
1491 * domain definitions.
1493 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (UINT16_MAX)
1496 * Ethernet device associated switch information
1498 struct rte_eth_switch_info {
1499 const char *name; /**< switch name */
1500 uint16_t domain_id; /**< switch domain id */
1503 * mapping to the devices physical switch port as enumerated from the
1504 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1505 * device this may correspond to the VF_ID of each virtual function,
1506 * but each driver should explicitly define the mapping of switch
1507 * port identifier to that physical interconnect/switch
1513 * @b EXPERIMENTAL: this structure may change without prior notice.
1515 * Ethernet device Rx buffer segmentation capabilities.
1517 struct rte_eth_rxseg_capa {
1519 uint32_t multi_pools:1; /**< Supports receiving to multiple pools.*/
1520 uint32_t offset_allowed:1; /**< Supports buffer offsets. */
1521 uint32_t offset_align_log2:4; /**< Required offset alignment. */
1522 uint16_t max_nseg; /**< Maximum amount of segments to split. */
1523 uint16_t reserved; /**< Reserved field. */
1527 * Ethernet device information
1531 * Ethernet device representor port type.
1533 enum rte_eth_representor_type {
1534 RTE_ETH_REPRESENTOR_NONE, /**< not a representor. */
1535 RTE_ETH_REPRESENTOR_VF, /**< representor of Virtual Function. */
1536 RTE_ETH_REPRESENTOR_SF, /**< representor of Sub Function. */
1537 RTE_ETH_REPRESENTOR_PF, /**< representor of Physical Function. */
1541 * A structure used to retrieve the contextual information of
1542 * an Ethernet device, such as the controlling driver of the
1545 struct rte_eth_dev_info {
1546 struct rte_device *device; /** Generic device information */
1547 const char *driver_name; /**< Device Driver name. */
1548 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1549 Use if_indextoname() to translate into an interface name. */
1550 uint16_t min_mtu; /**< Minimum MTU allowed */
1551 uint16_t max_mtu; /**< Maximum MTU allowed */
1552 const uint32_t *dev_flags; /**< Device flags */
1553 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1554 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1555 /** Maximum configurable size of LRO aggregated packet. */
1556 uint32_t max_lro_pkt_size;
1557 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1558 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1559 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1560 uint32_t max_hash_mac_addrs;
1561 /** Maximum number of hash MAC addresses for MTA and UTA. */
1562 uint16_t max_vfs; /**< Maximum number of VFs. */
1563 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1564 struct rte_eth_rxseg_capa rx_seg_capa; /**< Segmentation capability.*/
1565 uint64_t rx_offload_capa;
1566 /**< All RX offload capabilities including all per-queue ones */
1567 uint64_t tx_offload_capa;
1568 /**< All TX offload capabilities including all per-queue ones */
1569 uint64_t rx_queue_offload_capa;
1570 /**< Device per-queue RX offload capabilities. */
1571 uint64_t tx_queue_offload_capa;
1572 /**< Device per-queue TX offload capabilities. */
1574 /**< Device redirection table size, the total number of entries. */
1575 uint8_t hash_key_size; /**< Hash key size in bytes */
1576 /** Bit mask of RSS offloads, the bit offset also means flow type */
1577 uint64_t flow_type_rss_offloads;
1578 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1579 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1580 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1581 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1582 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1583 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1584 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1585 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1586 /** Configured number of rx/tx queues */
1587 uint16_t nb_rx_queues; /**< Number of RX queues. */
1588 uint16_t nb_tx_queues; /**< Number of TX queues. */
1589 /** Rx parameter recommendations */
1590 struct rte_eth_dev_portconf default_rxportconf;
1591 /** Tx parameter recommendations */
1592 struct rte_eth_dev_portconf default_txportconf;
1593 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1596 * Switching information for ports on a device with a
1597 * embedded managed interconnect/switch.
1599 struct rte_eth_switch_info switch_info;
1601 uint64_t reserved_64s[2]; /**< Reserved for future fields */
1602 void *reserved_ptrs[2]; /**< Reserved for future fields */
1606 * RX/TX queue states
1608 #define RTE_ETH_QUEUE_STATE_STOPPED 0
1609 #define RTE_ETH_QUEUE_STATE_STARTED 1
1610 #define RTE_ETH_QUEUE_STATE_HAIRPIN 2
1613 * Ethernet device RX queue information structure.
1614 * Used to retrieve information about configured queue.
1616 struct rte_eth_rxq_info {
1617 struct rte_mempool *mp; /**< mempool used by that queue. */
1618 struct rte_eth_rxconf conf; /**< queue config parameters. */
1619 uint8_t scattered_rx; /**< scattered packets RX supported. */
1620 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */
1621 uint16_t nb_desc; /**< configured number of RXDs. */
1622 uint16_t rx_buf_size; /**< hardware receive buffer size. */
1623 } __rte_cache_min_aligned;
1626 * Ethernet device TX queue information structure.
1627 * Used to retrieve information about configured queue.
1629 struct rte_eth_txq_info {
1630 struct rte_eth_txconf conf; /**< queue config parameters. */
1631 uint16_t nb_desc; /**< configured number of TXDs. */
1632 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */
1633 } __rte_cache_min_aligned;
1635 /* Generic Burst mode flag definition, values can be ORed. */
1638 * If the queues have different burst mode description, this bit will be set
1639 * by PMD, then the application can iterate to retrieve burst description for
1642 #define RTE_ETH_BURST_FLAG_PER_QUEUE (1ULL << 0)
1645 * Ethernet device RX/TX queue packet burst mode information structure.
1646 * Used to retrieve information about packet burst mode setting.
1648 struct rte_eth_burst_mode {
1649 uint64_t flags; /**< The ORed values of RTE_ETH_BURST_FLAG_xxx */
1651 #define RTE_ETH_BURST_MODE_INFO_SIZE 1024 /**< Maximum size for information */
1652 char info[RTE_ETH_BURST_MODE_INFO_SIZE]; /**< burst mode information */
1655 /** Maximum name length for extended statistics counters */
1656 #define RTE_ETH_XSTATS_NAME_SIZE 64
1659 * An Ethernet device extended statistic structure
1661 * This structure is used by rte_eth_xstats_get() to provide
1662 * statistics that are not provided in the generic *rte_eth_stats*
1664 * It maps a name id, corresponding to an index in the array returned
1665 * by rte_eth_xstats_get_names(), to a statistic value.
1667 struct rte_eth_xstat {
1668 uint64_t id; /**< The index in xstats name array. */
1669 uint64_t value; /**< The statistic counter value. */
1673 * A name element for extended statistics.
1675 * An array of this structure is returned by rte_eth_xstats_get_names().
1676 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1677 * structure references these names by their array index.
1679 * The xstats should follow a common naming scheme.
1680 * Some names are standardized in rte_stats_strings.
1682 * - rx_missed_errors
1684 * - tx_size_128_to_255_packets
1686 struct rte_eth_xstat_name {
1687 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1690 #define ETH_DCB_NUM_TCS 8
1691 #define ETH_MAX_VMDQ_POOL 64
1694 * A structure used to get the information of queue and
1695 * TC mapping on both TX and RX paths.
1697 struct rte_eth_dcb_tc_queue_mapping {
1698 /** rx queues assigned to tc per Pool */
1702 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1703 /** rx queues assigned to tc per Pool */
1707 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1711 * A structure used to get the information of DCB.
1712 * It includes TC UP mapping and queue TC mapping.
1714 struct rte_eth_dcb_info {
1715 uint8_t nb_tcs; /**< number of TCs */
1716 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1717 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1718 /** rx queues assigned to tc */
1719 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1723 * This enum indicates the possible Forward Error Correction (FEC) modes
1724 * of an ethdev port.
1726 enum rte_eth_fec_mode {
1727 RTE_ETH_FEC_NOFEC = 0, /**< FEC is off */
1728 RTE_ETH_FEC_AUTO, /**< FEC autonegotiation modes */
1729 RTE_ETH_FEC_BASER, /**< FEC using common algorithm */
1730 RTE_ETH_FEC_RS, /**< FEC using RS algorithm */
1733 /* Translate from FEC mode to FEC capa */
1734 #define RTE_ETH_FEC_MODE_TO_CAPA(x) (1U << (x))
1736 /* This macro indicates FEC capa mask */
1737 #define RTE_ETH_FEC_MODE_CAPA_MASK(x) (1U << (RTE_ETH_FEC_ ## x))
1739 /* A structure used to get capabilities per link speed */
1740 struct rte_eth_fec_capa {
1741 uint32_t speed; /**< Link speed (see ETH_SPEED_NUM_*) */
1742 uint32_t capa; /**< FEC capabilities bitmask */
1745 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1747 /* Macros to check for valid port */
1748 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1749 if (!rte_eth_dev_is_valid_port(port_id)) { \
1750 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1755 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1756 if (!rte_eth_dev_is_valid_port(port_id)) { \
1757 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1763 * l2 tunnel configuration.
1766 /**< l2 tunnel enable mask */
1767 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1768 /**< l2 tunnel insertion mask */
1769 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1770 /**< l2 tunnel stripping mask */
1771 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1772 /**< l2 tunnel forwarding mask */
1773 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1776 * Function type used for RX packet processing packet callbacks.
1778 * The callback function is called on RX with a burst of packets that have
1779 * been received on the given port and queue.
1782 * The Ethernet port on which RX is being performed.
1784 * The queue on the Ethernet port which is being used to receive the packets.
1786 * The burst of packets that have just been received.
1788 * The number of packets in the burst pointed to by "pkts".
1790 * The max number of packets that can be stored in the "pkts" array.
1792 * The arbitrary user parameter passed in by the application when the callback
1793 * was originally configured.
1795 * The number of packets returned to the user.
1797 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1798 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1802 * Function type used for TX packet processing packet callbacks.
1804 * The callback function is called on TX with a burst of packets immediately
1805 * before the packets are put onto the hardware queue for transmission.
1808 * The Ethernet port on which TX is being performed.
1810 * The queue on the Ethernet port which is being used to transmit the packets.
1812 * The burst of packets that are about to be transmitted.
1814 * The number of packets in the burst pointed to by "pkts".
1816 * The arbitrary user parameter passed in by the application when the callback
1817 * was originally configured.
1819 * The number of packets to be written to the NIC.
1821 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1822 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1825 * Possible states of an ethdev port.
1827 enum rte_eth_dev_state {
1828 /** Device is unused before being probed. */
1829 RTE_ETH_DEV_UNUSED = 0,
1830 /** Device is attached when allocated in probing. */
1831 RTE_ETH_DEV_ATTACHED,
1832 /** Device is in removed state when plug-out is detected. */
1833 RTE_ETH_DEV_REMOVED,
1836 struct rte_eth_dev_sriov {
1837 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1838 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1839 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1840 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1842 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1844 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1846 #define RTE_ETH_DEV_NO_OWNER 0
1848 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1850 struct rte_eth_dev_owner {
1851 uint64_t id; /**< The owner unique identifier. */
1852 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1855 /** PMD supports thread-safe flow operations */
1856 #define RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE 0x0001
1857 /** Device supports link state interrupt */
1858 #define RTE_ETH_DEV_INTR_LSC 0x0002
1859 /** Device is a bonded slave */
1860 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1861 /** Device supports device removal interrupt */
1862 #define RTE_ETH_DEV_INTR_RMV 0x0008
1863 /** Device is port representor */
1864 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1865 /** Device does not support MAC change after started */
1866 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1868 * Queue xstats filled automatically by ethdev layer.
1869 * PMDs filling the queue xstats themselves should not set this flag
1871 #define RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS 0x0040
1874 * Iterates over valid ethdev ports owned by a specific owner.
1877 * The id of the next possible valid owned port.
1879 * The owner identifier.
1880 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1882 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1884 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1885 const uint64_t owner_id);
1888 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1890 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1891 for (p = rte_eth_find_next_owned_by(0, o); \
1892 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1893 p = rte_eth_find_next_owned_by(p + 1, o))
1896 * Iterates over valid ethdev ports.
1899 * The id of the next possible valid port.
1901 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1903 uint16_t rte_eth_find_next(uint16_t port_id);
1906 * Macro to iterate over all enabled and ownerless ethdev ports.
1908 #define RTE_ETH_FOREACH_DEV(p) \
1909 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1912 * Iterates over ethdev ports of a specified device.
1914 * @param port_id_start
1915 * The id of the next possible valid port.
1917 * The generic device behind the ports to iterate.
1919 * Next port id of the device, possibly port_id_start,
1920 * RTE_MAX_ETHPORTS if there is none.
1923 rte_eth_find_next_of(uint16_t port_id_start,
1924 const struct rte_device *parent);
1927 * Macro to iterate over all ethdev ports of a specified device.
1930 * The id of the matching port being iterated.
1932 * The rte_device pointer matching the iterated ports.
1934 #define RTE_ETH_FOREACH_DEV_OF(port_id, parent) \
1935 for (port_id = rte_eth_find_next_of(0, parent); \
1936 port_id < RTE_MAX_ETHPORTS; \
1937 port_id = rte_eth_find_next_of(port_id + 1, parent))
1940 * Iterates over sibling ethdev ports (i.e. sharing the same rte_device).
1942 * @param port_id_start
1943 * The id of the next possible valid sibling port.
1944 * @param ref_port_id
1945 * The id of a reference port to compare rte_device with.
1947 * Next sibling port id, possibly port_id_start or ref_port_id itself,
1948 * RTE_MAX_ETHPORTS if there is none.
1951 rte_eth_find_next_sibling(uint16_t port_id_start, uint16_t ref_port_id);
1954 * Macro to iterate over all ethdev ports sharing the same rte_device
1955 * as the specified port.
1956 * Note: the specified reference port is part of the loop iterations.
1959 * The id of the matching port being iterated.
1960 * @param ref_port_id
1961 * The id of the port being compared.
1963 #define RTE_ETH_FOREACH_DEV_SIBLING(port_id, ref_port_id) \
1964 for (port_id = rte_eth_find_next_sibling(0, ref_port_id); \
1965 port_id < RTE_MAX_ETHPORTS; \
1966 port_id = rte_eth_find_next_sibling(port_id + 1, ref_port_id))
1970 * @b EXPERIMENTAL: this API may change without prior notice.
1972 * Get a new unique owner identifier.
1973 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1974 * to avoid multiple management of device by different entities.
1977 * Owner identifier pointer.
1979 * Negative errno value on error, 0 on success.
1982 int rte_eth_dev_owner_new(uint64_t *owner_id);
1986 * @b EXPERIMENTAL: this API may change without prior notice.
1988 * Set an Ethernet device owner.
1991 * The identifier of the port to own.
1993 * The owner pointer.
1995 * Negative errno value on error, 0 on success.
1998 int rte_eth_dev_owner_set(const uint16_t port_id,
1999 const struct rte_eth_dev_owner *owner);
2003 * @b EXPERIMENTAL: this API may change without prior notice.
2005 * Unset Ethernet device owner to make the device ownerless.
2008 * The identifier of port to make ownerless.
2010 * The owner identifier.
2012 * 0 on success, negative errno value on error.
2015 int rte_eth_dev_owner_unset(const uint16_t port_id,
2016 const uint64_t owner_id);
2020 * @b EXPERIMENTAL: this API may change without prior notice.
2022 * Remove owner from all Ethernet devices owned by a specific owner.
2025 * The owner identifier.
2027 * 0 on success, negative errno value on error.
2030 int rte_eth_dev_owner_delete(const uint64_t owner_id);
2034 * @b EXPERIMENTAL: this API may change without prior notice.
2036 * Get the owner of an Ethernet device.
2039 * The port identifier.
2041 * The owner structure pointer to fill.
2043 * 0 on success, negative errno value on error..
2046 int rte_eth_dev_owner_get(const uint16_t port_id,
2047 struct rte_eth_dev_owner *owner);
2050 * Get the number of ports which are usable for the application.
2052 * These devices must be iterated by using the macro
2053 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
2054 * to deal with non-contiguous ranges of devices.
2057 * The count of available Ethernet devices.
2059 uint16_t rte_eth_dev_count_avail(void);
2062 * Get the total number of ports which are allocated.
2064 * Some devices may not be available for the application.
2067 * The total count of Ethernet devices.
2069 uint16_t rte_eth_dev_count_total(void);
2072 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
2073 * the bitmap link_speeds of the struct rte_eth_conf
2076 * Numerical speed value in Mbps
2078 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
2080 * 0 if the speed cannot be mapped
2082 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
2085 * Get DEV_RX_OFFLOAD_* flag name.
2090 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2092 const char *rte_eth_dev_rx_offload_name(uint64_t offload);
2095 * Get DEV_TX_OFFLOAD_* flag name.
2100 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
2102 const char *rte_eth_dev_tx_offload_name(uint64_t offload);
2105 * Configure an Ethernet device.
2106 * This function must be invoked first before any other function in the
2107 * Ethernet API. This function can also be re-invoked when a device is in the
2111 * The port identifier of the Ethernet device to configure.
2112 * @param nb_rx_queue
2113 * The number of receive queues to set up for the Ethernet device.
2114 * @param nb_tx_queue
2115 * The number of transmit queues to set up for the Ethernet device.
2117 * The pointer to the configuration data to be used for the Ethernet device.
2118 * The *rte_eth_conf* structure includes:
2119 * - the hardware offload features to activate, with dedicated fields for
2120 * each statically configurable offload hardware feature provided by
2121 * Ethernet devices, such as IP checksum or VLAN tag stripping for
2123 * The Rx offload bitfield API is obsolete and will be deprecated.
2124 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
2125 * structure and use offloads field to set per-port offloads instead.
2126 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
2127 * the [rt]x_offload_capa returned from rte_eth_dev_info_get().
2128 * Any type of device supported offloading set in the input argument
2129 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
2130 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
2131 * - the Receive Side Scaling (RSS) configuration when using multiple RX
2132 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
2133 * must be within the flow_type_rss_offloads provided by drivers via
2134 * rte_eth_dev_info_get() API.
2136 * Embedding all configuration information in a single data structure
2137 * is the more flexible method that allows the addition of new features
2138 * without changing the syntax of the API.
2140 * - 0: Success, device configured.
2141 * - <0: Error code returned by the driver configuration function.
2143 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
2144 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
2148 * @b EXPERIMENTAL: this API may change without prior notice.
2150 * Check if an Ethernet device was physically removed.
2153 * The port identifier of the Ethernet device.
2155 * 1 when the Ethernet device is removed, otherwise 0.
2159 rte_eth_dev_is_removed(uint16_t port_id);
2162 * Allocate and set up a receive queue for an Ethernet device.
2164 * The function allocates a contiguous block of memory for *nb_rx_desc*
2165 * receive descriptors from a memory zone associated with *socket_id*
2166 * and initializes each receive descriptor with a network buffer allocated
2167 * from the memory pool *mb_pool*.
2170 * The port identifier of the Ethernet device.
2171 * @param rx_queue_id
2172 * The index of the receive queue to set up.
2173 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2174 * to rte_eth_dev_configure().
2176 * The number of receive descriptors to allocate for the receive ring.
2178 * The *socket_id* argument is the socket identifier in case of NUMA.
2179 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2180 * the DMA memory allocated for the receive descriptors of the ring.
2182 * The pointer to the configuration data to be used for the receive queue.
2183 * NULL value is allowed, in which case default RX configuration
2185 * The *rx_conf* structure contains an *rx_thresh* structure with the values
2186 * of the Prefetch, Host, and Write-Back threshold registers of the receive
2188 * In addition it contains the hardware offloads features to activate using
2189 * the DEV_RX_OFFLOAD_* flags.
2190 * If an offloading set in rx_conf->offloads
2191 * hasn't been set in the input argument eth_conf->rxmode.offloads
2192 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2193 * per-queue type and it is enabled for the queue.
2194 * No need to repeat any bit in rx_conf->offloads which has already been
2195 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2196 * at port level can't be disabled at queue level.
2197 * The configuration structure also contains the pointer to the array
2198 * of the receiving buffer segment descriptions, see rx_seg and rx_nseg
2199 * fields, this extended configuration might be used by split offloads like
2200 * RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT. If mb_pool is not NULL,
2201 * the extended configuration fields must be set to NULL and zero.
2203 * The pointer to the memory pool from which to allocate *rte_mbuf* network
2204 * memory buffers to populate each descriptor of the receive ring. There are
2205 * two options to provide Rx buffer configuration:
2207 * mb_pool is not NULL, rx_conf.rx_nseg is 0.
2208 * - multiple segments description:
2209 * mb_pool is NULL, rx_conf.rx_seg is not NULL, rx_conf.rx_nseg is not 0.
2210 * Taken only if flag RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT is set in offloads.
2213 * - 0: Success, receive queue correctly set up.
2214 * - -EIO: if device is removed.
2215 * - -ENODEV: if *port_id* is invalid.
2216 * - -EINVAL: The memory pool pointer is null or the size of network buffers
2217 * which can be allocated from this memory pool does not fit the various
2218 * buffer sizes allowed by the device controller.
2219 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
2220 * allocate network memory buffers from the memory pool when
2221 * initializing receive descriptors.
2223 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
2224 uint16_t nb_rx_desc, unsigned int socket_id,
2225 const struct rte_eth_rxconf *rx_conf,
2226 struct rte_mempool *mb_pool);
2230 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2232 * Allocate and set up a hairpin receive queue for an Ethernet device.
2234 * The function set up the selected queue to be used in hairpin.
2237 * The port identifier of the Ethernet device.
2238 * @param rx_queue_id
2239 * The index of the receive queue to set up.
2240 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2241 * to rte_eth_dev_configure().
2243 * The number of receive descriptors to allocate for the receive ring.
2244 * 0 means the PMD will use default value.
2246 * The pointer to the hairpin configuration.
2249 * - (0) if successful.
2250 * - (-ENODEV) if *port_id* is invalid.
2251 * - (-ENOTSUP) if hardware doesn't support.
2252 * - (-EINVAL) if bad parameter.
2253 * - (-ENOMEM) if unable to allocate the resources.
2256 int rte_eth_rx_hairpin_queue_setup
2257 (uint16_t port_id, uint16_t rx_queue_id, uint16_t nb_rx_desc,
2258 const struct rte_eth_hairpin_conf *conf);
2261 * Allocate and set up a transmit queue for an Ethernet device.
2264 * The port identifier of the Ethernet device.
2265 * @param tx_queue_id
2266 * The index of the transmit queue to set up.
2267 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2268 * to rte_eth_dev_configure().
2270 * The number of transmit descriptors to allocate for the transmit ring.
2272 * The *socket_id* argument is the socket identifier in case of NUMA.
2273 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
2274 * the DMA memory allocated for the transmit descriptors of the ring.
2276 * The pointer to the configuration data to be used for the transmit queue.
2277 * NULL value is allowed, in which case default TX configuration
2279 * The *tx_conf* structure contains the following data:
2280 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
2281 * Write-Back threshold registers of the transmit ring.
2282 * When setting Write-Back threshold to the value greater then zero,
2283 * *tx_rs_thresh* value should be explicitly set to one.
2284 * - The *tx_free_thresh* value indicates the [minimum] number of network
2285 * buffers that must be pending in the transmit ring to trigger their
2286 * [implicit] freeing by the driver transmit function.
2287 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
2288 * descriptors that must be pending in the transmit ring before setting the
2289 * RS bit on a descriptor by the driver transmit function.
2290 * The *tx_rs_thresh* value should be less or equal then
2291 * *tx_free_thresh* value, and both of them should be less then
2293 * - The *offloads* member contains Tx offloads to be enabled.
2294 * If an offloading set in tx_conf->offloads
2295 * hasn't been set in the input argument eth_conf->txmode.offloads
2296 * to rte_eth_dev_configure(), it is a new added offloading, it must be
2297 * per-queue type and it is enabled for the queue.
2298 * No need to repeat any bit in tx_conf->offloads which has already been
2299 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
2300 * at port level can't be disabled at queue level.
2302 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
2303 * the transmit function to use default values.
2305 * - 0: Success, the transmit queue is correctly set up.
2306 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
2308 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
2309 uint16_t nb_tx_desc, unsigned int socket_id,
2310 const struct rte_eth_txconf *tx_conf);
2314 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2316 * Allocate and set up a transmit hairpin queue for an Ethernet device.
2319 * The port identifier of the Ethernet device.
2320 * @param tx_queue_id
2321 * The index of the transmit queue to set up.
2322 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2323 * to rte_eth_dev_configure().
2325 * The number of transmit descriptors to allocate for the transmit ring.
2326 * 0 to set default PMD value.
2328 * The hairpin configuration.
2331 * - (0) if successful.
2332 * - (-ENODEV) if *port_id* is invalid.
2333 * - (-ENOTSUP) if hardware doesn't support.
2334 * - (-EINVAL) if bad parameter.
2335 * - (-ENOMEM) if unable to allocate the resources.
2338 int rte_eth_tx_hairpin_queue_setup
2339 (uint16_t port_id, uint16_t tx_queue_id, uint16_t nb_tx_desc,
2340 const struct rte_eth_hairpin_conf *conf);
2344 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2346 * Get all the hairpin peer Rx / Tx ports of the current port.
2347 * The caller should ensure that the array is large enough to save the ports
2351 * The port identifier of the Ethernet device.
2353 * Pointer to the array to store the peer ports list.
2355 * Length of the array to store the port identifiers.
2357 * Current port to peer port direction
2358 * positive - current used as Tx to get all peer Rx ports.
2359 * zero - current used as Rx to get all peer Tx ports.
2362 * - (0 or positive) actual peer ports number.
2363 * - (-EINVAL) if bad parameter.
2364 * - (-ENODEV) if *port_id* invalid
2365 * - (-ENOTSUP) if hardware doesn't support.
2366 * - Others detailed errors from PMD drivers.
2369 int rte_eth_hairpin_get_peer_ports(uint16_t port_id, uint16_t *peer_ports,
2370 size_t len, uint32_t direction);
2374 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2376 * Bind all hairpin Tx queues of one port to the Rx queues of the peer port.
2377 * It is only allowed to call this function after all hairpin queues are
2378 * configured properly and the devices are in started state.
2381 * The identifier of the Tx port.
2383 * The identifier of peer Rx port.
2384 * RTE_MAX_ETHPORTS is allowed for the traversal of all devices.
2385 * Rx port ID could have the same value as Tx port ID.
2388 * - (0) if successful.
2389 * - (-ENODEV) if Tx port ID is invalid.
2390 * - (-EBUSY) if device is not in started state.
2391 * - (-ENOTSUP) if hardware doesn't support.
2392 * - Others detailed errors from PMD drivers.
2395 int rte_eth_hairpin_bind(uint16_t tx_port, uint16_t rx_port);
2399 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
2401 * Unbind all hairpin Tx queues of one port from the Rx queues of the peer port.
2402 * This should be called before closing the Tx or Rx devices, if the bind
2403 * function is called before.
2404 * After unbinding the hairpin ports pair, it is allowed to bind them again.
2405 * Changing queues configuration should be after stopping the device(s).
2408 * The identifier of the Tx port.
2410 * The identifier of peer Rx port.
2411 * RTE_MAX_ETHPORTS is allowed for traversal of all devices.
2412 * Rx port ID could have the same value as Tx port ID.
2415 * - (0) if successful.
2416 * - (-ENODEV) if Tx port ID is invalid.
2417 * - (-EBUSY) if device is in stopped state.
2418 * - (-ENOTSUP) if hardware doesn't support.
2419 * - Others detailed errors from PMD drivers.
2422 int rte_eth_hairpin_unbind(uint16_t tx_port, uint16_t rx_port);
2425 * Return the NUMA socket to which an Ethernet device is connected
2428 * The port identifier of the Ethernet device
2430 * The NUMA socket id to which the Ethernet device is connected or
2431 * a default of zero if the socket could not be determined.
2432 * -1 is returned is the port_id value is out of range.
2434 int rte_eth_dev_socket_id(uint16_t port_id);
2437 * Check if port_id of device is attached
2440 * The port identifier of the Ethernet device
2442 * - 0 if port is out of range or not attached
2443 * - 1 if device is attached
2445 int rte_eth_dev_is_valid_port(uint16_t port_id);
2448 * Start specified RX queue of a port. It is used when rx_deferred_start
2449 * flag of the specified queue is true.
2452 * The port identifier of the Ethernet device
2453 * @param rx_queue_id
2454 * The index of the rx queue to update the ring.
2455 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2456 * to rte_eth_dev_configure().
2458 * - 0: Success, the receive queue is started.
2459 * - -ENODEV: if *port_id* is invalid.
2460 * - -EINVAL: The queue_id out of range or belong to hairpin.
2461 * - -EIO: if device is removed.
2462 * - -ENOTSUP: The function not supported in PMD driver.
2464 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
2467 * Stop specified RX queue of a port
2470 * The port identifier of the Ethernet device
2471 * @param rx_queue_id
2472 * The index of the rx queue to update the ring.
2473 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2474 * to rte_eth_dev_configure().
2476 * - 0: Success, the receive queue is stopped.
2477 * - -ENODEV: if *port_id* is invalid.
2478 * - -EINVAL: The queue_id out of range or belong to hairpin.
2479 * - -EIO: if device is removed.
2480 * - -ENOTSUP: The function not supported in PMD driver.
2482 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
2485 * Start TX for specified queue of a port. It is used when tx_deferred_start
2486 * flag of the specified queue is true.
2489 * The port identifier of the Ethernet device
2490 * @param tx_queue_id
2491 * The index of the tx queue to update the ring.
2492 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2493 * to rte_eth_dev_configure().
2495 * - 0: Success, the transmit queue is started.
2496 * - -ENODEV: if *port_id* is invalid.
2497 * - -EINVAL: The queue_id out of range or belong to hairpin.
2498 * - -EIO: if device is removed.
2499 * - -ENOTSUP: The function not supported in PMD driver.
2501 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
2504 * Stop specified TX queue of a port
2507 * The port identifier of the Ethernet device
2508 * @param tx_queue_id
2509 * The index of the tx queue to update the ring.
2510 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2511 * to rte_eth_dev_configure().
2513 * - 0: Success, the transmit queue is stopped.
2514 * - -ENODEV: if *port_id* is invalid.
2515 * - -EINVAL: The queue_id out of range or belong to hairpin.
2516 * - -EIO: if device is removed.
2517 * - -ENOTSUP: The function not supported in PMD driver.
2519 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
2522 * Start an Ethernet device.
2524 * The device start step is the last one and consists of setting the configured
2525 * offload features and in starting the transmit and the receive units of the
2528 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
2529 * PMD port start callback function is invoked.
2531 * On success, all basic functions exported by the Ethernet API (link status,
2532 * receive/transmit, and so on) can be invoked.
2535 * The port identifier of the Ethernet device.
2537 * - 0: Success, Ethernet device started.
2538 * - <0: Error code of the driver device start function.
2540 int rte_eth_dev_start(uint16_t port_id);
2543 * Stop an Ethernet device. The device can be restarted with a call to
2544 * rte_eth_dev_start()
2547 * The port identifier of the Ethernet device.
2549 * - 0: Success, Ethernet device stopped.
2550 * - <0: Error code of the driver device stop function.
2552 int rte_eth_dev_stop(uint16_t port_id);
2555 * Link up an Ethernet device.
2557 * Set device link up will re-enable the device rx/tx
2558 * functionality after it is previously set device linked down.
2561 * The port identifier of the Ethernet device.
2563 * - 0: Success, Ethernet device linked up.
2564 * - <0: Error code of the driver device link up function.
2566 int rte_eth_dev_set_link_up(uint16_t port_id);
2569 * Link down an Ethernet device.
2570 * The device rx/tx functionality will be disabled if success,
2571 * and it can be re-enabled with a call to
2572 * rte_eth_dev_set_link_up()
2575 * The port identifier of the Ethernet device.
2577 int rte_eth_dev_set_link_down(uint16_t port_id);
2580 * Close a stopped Ethernet device. The device cannot be restarted!
2581 * The function frees all port resources.
2584 * The port identifier of the Ethernet device.
2586 * - Zero if the port is closed successfully.
2587 * - Negative if something went wrong.
2589 int rte_eth_dev_close(uint16_t port_id);
2592 * Reset a Ethernet device and keep its port id.
2594 * When a port has to be reset passively, the DPDK application can invoke
2595 * this function. For example when a PF is reset, all its VFs should also
2596 * be reset. Normally a DPDK application can invoke this function when
2597 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
2598 * a port reset in other circumstances.
2600 * When this function is called, it first stops the port and then calls the
2601 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
2602 * state, in which no Tx and Rx queues are setup, as if the port has been
2603 * reset and not started. The port keeps the port id it had before the
2606 * After calling rte_eth_dev_reset( ), the application should use
2607 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
2608 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
2609 * to reconfigure the device as appropriate.
2611 * Note: To avoid unexpected behavior, the application should stop calling
2612 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
2613 * safety, all these controlling functions should be called from the same
2617 * The port identifier of the Ethernet device.
2620 * - (0) if successful.
2621 * - (-ENODEV) if *port_id* is invalid.
2622 * - (-ENOTSUP) if hardware doesn't support this function.
2623 * - (-EPERM) if not ran from the primary process.
2624 * - (-EIO) if re-initialisation failed or device is removed.
2625 * - (-ENOMEM) if the reset failed due to OOM.
2626 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
2628 int rte_eth_dev_reset(uint16_t port_id);
2631 * Enable receipt in promiscuous mode for an Ethernet device.
2634 * The port identifier of the Ethernet device.
2636 * - (0) if successful.
2637 * - (-ENOTSUP) if support for promiscuous_enable() does not exist
2639 * - (-ENODEV) if *port_id* invalid.
2641 int rte_eth_promiscuous_enable(uint16_t port_id);
2644 * Disable receipt in promiscuous mode for an Ethernet device.
2647 * The port identifier of the Ethernet device.
2649 * - (0) if successful.
2650 * - (-ENOTSUP) if support for promiscuous_disable() does not exist
2652 * - (-ENODEV) if *port_id* invalid.
2654 int rte_eth_promiscuous_disable(uint16_t port_id);
2657 * Return the value of promiscuous mode for an Ethernet device.
2660 * The port identifier of the Ethernet device.
2662 * - (1) if promiscuous is enabled
2663 * - (0) if promiscuous is disabled.
2666 int rte_eth_promiscuous_get(uint16_t port_id);
2669 * Enable the receipt of any multicast frame by an Ethernet device.
2672 * The port identifier of the Ethernet device.
2674 * - (0) if successful.
2675 * - (-ENOTSUP) if support for allmulticast_enable() does not exist
2677 * - (-ENODEV) if *port_id* invalid.
2679 int rte_eth_allmulticast_enable(uint16_t port_id);
2682 * Disable the receipt of all multicast frames by an Ethernet device.
2685 * The port identifier of the Ethernet device.
2687 * - (0) if successful.
2688 * - (-ENOTSUP) if support for allmulticast_disable() does not exist
2690 * - (-ENODEV) if *port_id* invalid.
2692 int rte_eth_allmulticast_disable(uint16_t port_id);
2695 * Return the value of allmulticast mode for an Ethernet device.
2698 * The port identifier of the Ethernet device.
2700 * - (1) if allmulticast is enabled
2701 * - (0) if allmulticast is disabled.
2704 int rte_eth_allmulticast_get(uint16_t port_id);
2707 * Retrieve the link status (up/down), the duplex mode (half/full),
2708 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2710 * It might need to wait up to 9 seconds.
2711 * @see rte_eth_link_get_nowait.
2714 * The port identifier of the Ethernet device.
2716 * Link information written back.
2718 * - (0) if successful.
2719 * - (-ENOTSUP) if the function is not supported in PMD driver.
2720 * - (-ENODEV) if *port_id* invalid.
2721 * - (-EINVAL) if bad parameter.
2723 int rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
2726 * Retrieve the link status (up/down), the duplex mode (half/full),
2727 * the negotiation (auto/fixed), and if available, the speed (Mbps).
2730 * The port identifier of the Ethernet device.
2732 * Link information written back.
2734 * - (0) if successful.
2735 * - (-ENOTSUP) if the function is not supported in PMD driver.
2736 * - (-ENODEV) if *port_id* invalid.
2737 * - (-EINVAL) if bad parameter.
2739 int rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
2743 * @b EXPERIMENTAL: this API may change without prior notice.
2745 * The function converts a link_speed to a string. It handles all special
2746 * values like unknown or none speed.
2749 * link_speed of rte_eth_link struct
2751 * Link speed in textual format. It's pointer to immutable memory.
2752 * No free is required.
2755 const char *rte_eth_link_speed_to_str(uint32_t link_speed);
2759 * @b EXPERIMENTAL: this API may change without prior notice.
2761 * The function converts a rte_eth_link struct representing a link status to
2765 * A pointer to a string to be filled with textual representation of
2766 * device status. At least ETH_LINK_MAX_STR_LEN bytes should be allocated to
2767 * store default link status text.
2769 * Length of available memory at 'str' string.
2771 * Link status returned by rte_eth_link_get function
2773 * Number of bytes written to str array or -EINVAL if bad parameter.
2776 int rte_eth_link_to_str(char *str, size_t len,
2777 const struct rte_eth_link *eth_link);
2780 * Retrieve the general I/O statistics of an Ethernet device.
2783 * The port identifier of the Ethernet device.
2785 * A pointer to a structure of type *rte_eth_stats* to be filled with
2786 * the values of device counters for the following set of statistics:
2787 * - *ipackets* with the total of successfully received packets.
2788 * - *opackets* with the total of successfully transmitted packets.
2789 * - *ibytes* with the total of successfully received bytes.
2790 * - *obytes* with the total of successfully transmitted bytes.
2791 * - *ierrors* with the total of erroneous received packets.
2792 * - *oerrors* with the total of failed transmitted packets.
2794 * Zero if successful. Non-zero otherwise.
2796 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
2799 * Reset the general I/O statistics of an Ethernet device.
2802 * The port identifier of the Ethernet device.
2804 * - (0) if device notified to reset stats.
2805 * - (-ENOTSUP) if hardware doesn't support.
2806 * - (-ENODEV) if *port_id* invalid.
2807 * - (<0): Error code of the driver stats reset function.
2809 int rte_eth_stats_reset(uint16_t port_id);
2812 * Retrieve names of extended statistics of an Ethernet device.
2814 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2816 * xstats_names[i].name => xstats[i].value
2818 * And the array index is same with id field of 'struct rte_eth_xstat':
2821 * This assumption makes key-value pair matching less flexible but simpler.
2824 * The port identifier of the Ethernet device.
2825 * @param xstats_names
2826 * An rte_eth_xstat_name array of at least *size* elements to
2827 * be filled. If set to NULL, the function returns the required number
2830 * The size of the xstats_names array (number of elements).
2832 * - A positive value lower or equal to size: success. The return value
2833 * is the number of entries filled in the stats table.
2834 * - A positive value higher than size: error, the given statistics table
2835 * is too small. The return value corresponds to the size that should
2836 * be given to succeed. The entries in the table are not valid and
2837 * shall not be used by the caller.
2838 * - A negative value on error (invalid port id).
2840 int rte_eth_xstats_get_names(uint16_t port_id,
2841 struct rte_eth_xstat_name *xstats_names,
2845 * Retrieve extended statistics of an Ethernet device.
2847 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2849 * xstats_names[i].name => xstats[i].value
2851 * And the array index is same with id field of 'struct rte_eth_xstat':
2854 * This assumption makes key-value pair matching less flexible but simpler.
2857 * The port identifier of the Ethernet device.
2859 * A pointer to a table of structure of type *rte_eth_xstat*
2860 * to be filled with device statistics ids and values.
2861 * This parameter can be set to NULL if n is 0.
2863 * The size of the xstats array (number of elements).
2865 * - A positive value lower or equal to n: success. The return value
2866 * is the number of entries filled in the stats table.
2867 * - A positive value higher than n: error, the given statistics table
2868 * is too small. The return value corresponds to the size that should
2869 * be given to succeed. The entries in the table are not valid and
2870 * shall not be used by the caller.
2871 * - A negative value on error (invalid port id).
2873 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2877 * Retrieve names of extended statistics of an Ethernet device.
2880 * The port identifier of the Ethernet device.
2881 * @param xstats_names
2882 * An rte_eth_xstat_name array of at least *size* elements to
2883 * be filled. If set to NULL, the function returns the required number
2886 * IDs array given by app to retrieve specific statistics
2888 * The size of the xstats_names array (number of elements).
2890 * - A positive value lower or equal to size: success. The return value
2891 * is the number of entries filled in the stats table.
2892 * - A positive value higher than size: error, the given statistics table
2893 * is too small. The return value corresponds to the size that should
2894 * be given to succeed. The entries in the table are not valid and
2895 * shall not be used by the caller.
2896 * - A negative value on error (invalid port id).
2899 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2900 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2904 * Retrieve extended statistics of an Ethernet device.
2907 * The port identifier of the Ethernet device.
2909 * A pointer to an ids array passed by application. This tells which
2910 * statistics values function should retrieve. This parameter
2911 * can be set to NULL if size is 0. In this case function will retrieve
2912 * all available statistics.
2914 * A pointer to a table to be filled with device statistics values.
2916 * The size of the ids array (number of elements).
2918 * - A positive value lower or equal to size: success. The return value
2919 * is the number of entries filled in the stats table.
2920 * - A positive value higher than size: error, the given statistics table
2921 * is too small. The return value corresponds to the size that should
2922 * be given to succeed. The entries in the table are not valid and
2923 * shall not be used by the caller.
2924 * - A negative value on error (invalid port id).
2926 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2927 uint64_t *values, unsigned int size);
2930 * Gets the ID of a statistic from its name.
2932 * This function searches for the statistics using string compares, and
2933 * as such should not be used on the fast-path. For fast-path retrieval of
2934 * specific statistics, store the ID as provided in *id* from this function,
2935 * and pass the ID to rte_eth_xstats_get()
2937 * @param port_id The port to look up statistics from
2938 * @param xstat_name The name of the statistic to return
2939 * @param[out] id A pointer to an app-supplied uint64_t which should be
2940 * set to the ID of the stat if the stat exists.
2943 * -ENODEV for invalid port_id,
2944 * -EIO if device is removed,
2945 * -EINVAL if the xstat_name doesn't exist in port_id
2946 * -ENOMEM if bad parameter.
2948 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2952 * Reset extended statistics of an Ethernet device.
2955 * The port identifier of the Ethernet device.
2957 * - (0) if device notified to reset extended stats.
2958 * - (-ENOTSUP) if pmd doesn't support both
2959 * extended stats and basic stats reset.
2960 * - (-ENODEV) if *port_id* invalid.
2961 * - (<0): Error code of the driver xstats reset function.
2963 int rte_eth_xstats_reset(uint16_t port_id);
2966 * Set a mapping for the specified transmit queue to the specified per-queue
2967 * statistics counter.
2970 * The port identifier of the Ethernet device.
2971 * @param tx_queue_id
2972 * The index of the transmit queue for which a queue stats mapping is required.
2973 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2974 * to rte_eth_dev_configure().
2976 * The per-queue packet statistics functionality number that the transmit
2977 * queue is to be assigned.
2978 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2979 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
2981 * Zero if successful. Non-zero otherwise.
2983 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2984 uint16_t tx_queue_id, uint8_t stat_idx);
2987 * Set a mapping for the specified receive queue to the specified per-queue
2988 * statistics counter.
2991 * The port identifier of the Ethernet device.
2992 * @param rx_queue_id
2993 * The index of the receive queue for which a queue stats mapping is required.
2994 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2995 * to rte_eth_dev_configure().
2997 * The per-queue packet statistics functionality number that the receive
2998 * queue is to be assigned.
2999 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
3000 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256.
3002 * Zero if successful. Non-zero otherwise.
3004 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
3005 uint16_t rx_queue_id,
3009 * Retrieve the Ethernet address of an Ethernet device.
3012 * The port identifier of the Ethernet device.
3014 * A pointer to a structure of type *ether_addr* to be filled with
3015 * the Ethernet address of the Ethernet device.
3017 * - (0) if successful
3018 * - (-ENODEV) if *port_id* invalid.
3019 * - (-EINVAL) if bad parameter.
3021 int rte_eth_macaddr_get(uint16_t port_id, struct rte_ether_addr *mac_addr);
3024 * Retrieve the contextual information of an Ethernet device.
3026 * As part of this function, a number of of fields in dev_info will be
3027 * initialized as follows:
3032 * Where lim is defined within the rte_eth_dev_info_get as
3034 * const struct rte_eth_desc_lim lim = {
3035 * .nb_max = UINT16_MAX,
3038 * .nb_seg_max = UINT16_MAX,
3039 * .nb_mtu_seg_max = UINT16_MAX,
3042 * device = dev->device
3043 * min_mtu = RTE_ETHER_MIN_MTU
3044 * max_mtu = UINT16_MAX
3046 * The following fields will be populated if support for dev_infos_get()
3047 * exists for the device and the rte_eth_dev 'dev' has been populated
3048 * successfully with a call to it:
3050 * driver_name = dev->device->driver->name
3051 * nb_rx_queues = dev->data->nb_rx_queues
3052 * nb_tx_queues = dev->data->nb_tx_queues
3053 * dev_flags = &dev->data->dev_flags
3056 * The port identifier of the Ethernet device.
3058 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
3059 * the contextual information of the Ethernet device.
3061 * - (0) if successful.
3062 * - (-ENOTSUP) if support for dev_infos_get() does not exist for the device.
3063 * - (-ENODEV) if *port_id* invalid.
3064 * - (-EINVAL) if bad parameter.
3066 int rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
3069 * Retrieve the firmware version of a device.
3072 * The port identifier of the device.
3074 * A pointer to a string array storing the firmware version of a device,
3075 * the string includes terminating null. This pointer is allocated by caller.
3077 * The size of the string array pointed by fw_version, which should be
3078 * large enough to store firmware version of the device.
3080 * - (0) if successful.
3081 * - (-ENOTSUP) if operation is not supported.
3082 * - (-ENODEV) if *port_id* invalid.
3083 * - (-EIO) if device is removed.
3084 * - (-EINVAL) if bad parameter.
3085 * - (>0) if *fw_size* is not enough to store firmware version, return
3086 * the size of the non truncated string.
3088 int rte_eth_dev_fw_version_get(uint16_t port_id,
3089 char *fw_version, size_t fw_size);
3092 * Retrieve the supported packet types of an Ethernet device.
3094 * When a packet type is announced as supported, it *must* be recognized by
3095 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
3096 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
3097 * packet types for these packets:
3098 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
3099 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
3100 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
3101 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
3103 * When a packet is received by a PMD, the most precise type must be
3104 * returned among the ones supported. However a PMD is allowed to set
3105 * packet type that is not in the supported list, at the condition that it
3106 * is more precise. Therefore, a PMD announcing no supported packet types
3107 * can still set a matching packet type in a received packet.
3110 * Better to invoke this API after the device is already started or rx burst
3111 * function is decided, to obtain correct supported ptypes.
3113 * if a given PMD does not report what ptypes it supports, then the supported
3114 * ptype count is reported as 0.
3116 * The port identifier of the Ethernet device.
3118 * A hint of what kind of packet type which the caller is interested in.
3120 * An array pointer to store adequate packet types, allocated by caller.
3122 * Size of the array pointed by param ptypes.
3124 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
3125 * only num entries will be filled into the ptypes array, but the full
3126 * count of supported ptypes will be returned.
3127 * - (-ENODEV) if *port_id* invalid.
3128 * - (-EINVAL) if bad parameter.
3130 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
3131 uint32_t *ptypes, int num);
3133 * Inform Ethernet device about reduced range of packet types to handle.
3135 * Application can use this function to set only specific ptypes that it's
3136 * interested. This information can be used by the PMD to optimize Rx path.
3138 * The function accepts an array `set_ptypes` allocated by the caller to
3139 * store the packet types set by the driver, the last element of the array
3140 * is set to RTE_PTYPE_UNKNOWN. The size of the `set_ptype` array should be
3141 * `rte_eth_dev_get_supported_ptypes() + 1` else it might only be filled
3145 * The port identifier of the Ethernet device.
3147 * The ptype family that application is interested in should be bitwise OR of
3148 * RTE_PTYPE_*_MASK or 0.
3150 * An array pointer to store set packet types, allocated by caller. The
3151 * function marks the end of array with RTE_PTYPE_UNKNOWN.
3153 * Size of the array pointed by param ptypes.
3154 * Should be rte_eth_dev_get_supported_ptypes() + 1 to accommodate the
3158 * - (-ENODEV) if *port_id* invalid.
3159 * - (-EINVAL) if *ptype_mask* is invalid (or) set_ptypes is NULL and
3162 int rte_eth_dev_set_ptypes(uint16_t port_id, uint32_t ptype_mask,
3163 uint32_t *set_ptypes, unsigned int num);
3166 * Retrieve the MTU of an Ethernet device.
3169 * The port identifier of the Ethernet device.
3171 * A pointer to a uint16_t where the retrieved MTU is to be stored.
3173 * - (0) if successful.
3174 * - (-ENODEV) if *port_id* invalid.
3175 * - (-EINVAL) if bad parameter.
3177 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
3180 * Change the MTU of an Ethernet device.
3183 * The port identifier of the Ethernet device.
3185 * A uint16_t for the MTU to be applied.
3187 * - (0) if successful.
3188 * - (-ENOTSUP) if operation is not supported.
3189 * - (-ENODEV) if *port_id* invalid.
3190 * - (-EIO) if device is removed.
3191 * - (-EINVAL) if *mtu* invalid, validation of mtu can occur within
3192 * rte_eth_dev_set_mtu if dev_infos_get is supported by the device or
3193 * when the mtu is set using dev->dev_ops->mtu_set.
3194 * - (-EBUSY) if operation is not allowed when the port is running
3196 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
3199 * Enable/Disable hardware filtering by an Ethernet device of received
3200 * VLAN packets tagged with a given VLAN Tag Identifier.
3203 * The port identifier of the Ethernet device.
3205 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
3207 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
3208 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
3210 * - (0) if successful.
3211 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3212 * - (-ENODEV) if *port_id* invalid.
3213 * - (-EIO) if device is removed.
3214 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
3215 * - (-EINVAL) if *vlan_id* > 4095.
3217 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
3220 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
3223 * The port identifier of the Ethernet device.
3224 * @param rx_queue_id
3225 * The index of the receive queue for which a queue stats mapping is required.
3226 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3227 * to rte_eth_dev_configure().
3229 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
3230 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
3232 * - (0) if successful.
3233 * - (-ENOTSUP) if hardware-assisted VLAN stripping not configured.
3234 * - (-ENODEV) if *port_id* invalid.
3235 * - (-EINVAL) if *rx_queue_id* invalid.
3237 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
3241 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
3245 * The port identifier of the Ethernet device.
3249 * The Tag Protocol ID
3251 * - (0) if successful.
3252 * - (-ENOTSUP) if hardware-assisted VLAN TPID setup is not supported.
3253 * - (-ENODEV) if *port_id* invalid.
3254 * - (-EIO) if device is removed.
3256 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
3257 enum rte_vlan_type vlan_type,
3261 * Set VLAN offload configuration on an Ethernet device.
3264 * The port identifier of the Ethernet device.
3265 * @param offload_mask
3266 * The VLAN Offload bit mask can be mixed use with "OR"
3267 * ETH_VLAN_STRIP_OFFLOAD
3268 * ETH_VLAN_FILTER_OFFLOAD
3269 * ETH_VLAN_EXTEND_OFFLOAD
3270 * ETH_QINQ_STRIP_OFFLOAD
3272 * - (0) if successful.
3273 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured.
3274 * - (-ENODEV) if *port_id* invalid.
3275 * - (-EIO) if device is removed.
3277 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
3280 * Read VLAN Offload configuration from an Ethernet device
3283 * The port identifier of the Ethernet device.
3285 * - (>0) if successful. Bit mask to indicate
3286 * ETH_VLAN_STRIP_OFFLOAD
3287 * ETH_VLAN_FILTER_OFFLOAD
3288 * ETH_VLAN_EXTEND_OFFLOAD
3289 * ETH_QINQ_STRIP_OFFLOAD
3290 * - (-ENODEV) if *port_id* invalid.
3292 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
3295 * Set port based TX VLAN insertion on or off.
3298 * The port identifier of the Ethernet device.
3300 * Port based TX VLAN identifier together with user priority.
3302 * Turn on or off the port based TX VLAN insertion.
3305 * - (0) if successful.
3306 * - negative if failed.
3308 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
3310 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
3314 * Structure used to buffer packets for future TX
3315 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
3317 struct rte_eth_dev_tx_buffer {
3318 buffer_tx_error_fn error_callback;
3319 void *error_userdata;
3320 uint16_t size; /**< Size of buffer for buffered tx */
3321 uint16_t length; /**< Number of packets in the array */
3322 struct rte_mbuf *pkts[];
3323 /**< Pending packets to be sent on explicit flush or when full */
3327 * Calculate the size of the tx buffer.
3330 * Number of stored packets.
3332 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
3333 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
3336 * Initialize default values for buffered transmitting
3339 * Tx buffer to be initialized.
3346 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
3349 * Configure a callback for buffered packets which cannot be sent
3351 * Register a specific callback to be called when an attempt is made to send
3352 * all packets buffered on an ethernet port, but not all packets can
3353 * successfully be sent. The callback registered here will be called only
3354 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
3355 * The default callback configured for each queue by default just frees the
3356 * packets back to the calling mempool. If additional behaviour is required,
3357 * for example, to count dropped packets, or to retry transmission of packets
3358 * which cannot be sent, this function should be used to register a suitable
3359 * callback function to implement the desired behaviour.
3360 * The example callback "rte_eth_count_unsent_packet_callback()" is also
3361 * provided as reference.
3364 * The port identifier of the Ethernet device.
3366 * The function to be used as the callback.
3368 * Arbitrary parameter to be passed to the callback function
3370 * 0 on success, or -EINVAL if bad parameter
3373 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
3374 buffer_tx_error_fn callback, void *userdata);
3377 * Callback function for silently dropping unsent buffered packets.
3379 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3380 * adjust the default behavior when buffered packets cannot be sent. This
3381 * function drops any unsent packets silently and is used by tx buffered
3382 * operations as default behavior.
3384 * NOTE: this function should not be called directly, instead it should be used
3385 * as a callback for packet buffering.
3387 * NOTE: when configuring this function as a callback with
3388 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3389 * should point to an uint64_t value.
3392 * The previously buffered packets which could not be sent
3394 * The number of unsent packets in the pkts array
3399 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
3403 * Callback function for tracking unsent buffered packets.
3405 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
3406 * adjust the default behavior when buffered packets cannot be sent. This
3407 * function drops any unsent packets, but also updates a user-supplied counter
3408 * to track the overall number of packets dropped. The counter should be an
3409 * uint64_t variable.
3411 * NOTE: this function should not be called directly, instead it should be used
3412 * as a callback for packet buffering.
3414 * NOTE: when configuring this function as a callback with
3415 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
3416 * should point to an uint64_t value.
3419 * The previously buffered packets which could not be sent
3421 * The number of unsent packets in the pkts array
3423 * Pointer to an uint64_t value, which will be incremented by unsent
3426 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
3430 * Request the driver to free mbufs currently cached by the driver. The
3431 * driver will only free the mbuf if it is no longer in use. It is the
3432 * application's responsibility to ensure rte_eth_tx_buffer_flush(..) is
3436 * The port identifier of the Ethernet device.
3438 * The index of the transmit queue through which output packets must be
3440 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
3441 * to rte_eth_dev_configure().
3443 * Maximum number of packets to free. Use 0 to indicate all possible packets
3444 * should be freed. Note that a packet may be using multiple mbufs.
3447 * -ENODEV: Invalid interface
3448 * -EIO: device is removed
3449 * -ENOTSUP: Driver does not support function
3451 * 0-n: Number of packets freed. More packets may still remain in ring that
3455 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
3458 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
3461 enum rte_eth_event_ipsec_subtype {
3462 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
3463 /**< Unknown event type */
3464 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
3465 /**< Sequence number overflow */
3466 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
3467 /**< Soft time expiry of SA */
3468 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
3469 /**< Soft byte expiry of SA */
3470 RTE_ETH_EVENT_IPSEC_MAX
3471 /**< Max value of this enum */
3475 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
3476 * information of the IPsec offload event.
3478 struct rte_eth_event_ipsec_desc {
3479 enum rte_eth_event_ipsec_subtype subtype;
3480 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
3482 /**< Event specific metadata
3484 * For the following events, *userdata* registered
3485 * with the *rte_security_session* would be returned
3488 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
3489 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
3490 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
3492 * @see struct rte_security_session_conf
3498 * The eth device event type for interrupt, and maybe others in the future.
3500 enum rte_eth_event_type {
3501 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
3502 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
3503 RTE_ETH_EVENT_QUEUE_STATE,
3504 /**< queue state event (enabled/disabled) */
3505 RTE_ETH_EVENT_INTR_RESET,
3506 /**< reset interrupt event, sent to VF on PF reset */
3507 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
3508 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
3509 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
3510 RTE_ETH_EVENT_NEW, /**< port is probed */
3511 RTE_ETH_EVENT_DESTROY, /**< port is released */
3512 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
3513 RTE_ETH_EVENT_FLOW_AGED,/**< New aged-out flows is detected */
3514 RTE_ETH_EVENT_MAX /**< max value of this enum */
3517 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
3518 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
3519 /**< user application callback to be registered for interrupts */
3522 * Register a callback function for port event.
3526 * RTE_ETH_ALL means register the event for all port ids.
3530 * User supplied callback function to be called.
3532 * Pointer to the parameters for the registered callback.
3535 * - On success, zero.
3536 * - On failure, a negative value.
3538 int rte_eth_dev_callback_register(uint16_t port_id,
3539 enum rte_eth_event_type event,
3540 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3543 * Unregister a callback function for port event.
3547 * RTE_ETH_ALL means unregister the event for all port ids.
3551 * User supplied callback function to be called.
3553 * Pointer to the parameters for the registered callback. -1 means to
3554 * remove all for the same callback address and same event.
3557 * - On success, zero.
3558 * - On failure, a negative value.
3560 int rte_eth_dev_callback_unregister(uint16_t port_id,
3561 enum rte_eth_event_type event,
3562 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
3565 * When there is no rx packet coming in Rx Queue for a long time, we can
3566 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
3567 * to be triggered when Rx packet arrives.
3569 * The rte_eth_dev_rx_intr_enable() function enables rx queue
3570 * interrupt on specific rx queue of a port.
3573 * The port identifier of the Ethernet device.
3575 * The index of the receive queue from which to retrieve input packets.
3576 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3577 * to rte_eth_dev_configure().
3579 * - (0) if successful.
3580 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3582 * - (-ENODEV) if *port_id* invalid.
3583 * - (-EIO) if device is removed.
3585 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
3588 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
3589 * interrupt and returns to polling mode.
3591 * The rte_eth_dev_rx_intr_disable() function disables rx queue
3592 * interrupt on specific rx queue of a port.
3595 * The port identifier of the Ethernet device.
3597 * The index of the receive queue from which to retrieve input packets.
3598 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3599 * to rte_eth_dev_configure().
3601 * - (0) if successful.
3602 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3604 * - (-ENODEV) if *port_id* invalid.
3605 * - (-EIO) if device is removed.
3607 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
3610 * RX Interrupt control per port.
3613 * The port identifier of the Ethernet device.
3615 * Epoll instance fd which the intr vector associated to.
3616 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3618 * The operation be performed for the vector.
3619 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3623 * - On success, zero.
3624 * - On failure, a negative value.
3626 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
3629 * RX Interrupt control per queue.
3632 * The port identifier of the Ethernet device.
3634 * The index of the receive queue from which to retrieve input packets.
3635 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3636 * to rte_eth_dev_configure().
3638 * Epoll instance fd which the intr vector associated to.
3639 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
3641 * The operation be performed for the vector.
3642 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
3646 * - On success, zero.
3647 * - On failure, a negative value.
3649 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
3650 int epfd, int op, void *data);
3653 * Get interrupt fd per Rx queue.
3656 * The port identifier of the Ethernet device.
3658 * The index of the receive queue from which to retrieve input packets.
3659 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3660 * to rte_eth_dev_configure().
3662 * - (>=0) the interrupt fd associated to the requested Rx queue if
3667 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
3670 * Turn on the LED on the Ethernet device.
3671 * This function turns on the LED on the Ethernet device.
3674 * The port identifier of the Ethernet device.
3676 * - (0) if successful.
3677 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3679 * - (-ENODEV) if *port_id* invalid.
3680 * - (-EIO) if device is removed.
3682 int rte_eth_led_on(uint16_t port_id);
3685 * Turn off the LED on the Ethernet device.
3686 * This function turns off the LED on the Ethernet device.
3689 * The port identifier of the Ethernet device.
3691 * - (0) if successful.
3692 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
3694 * - (-ENODEV) if *port_id* invalid.
3695 * - (-EIO) if device is removed.
3697 int rte_eth_led_off(uint16_t port_id);
3701 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3703 * Get Forward Error Correction(FEC) capability.
3706 * The port identifier of the Ethernet device.
3707 * @param speed_fec_capa
3708 * speed_fec_capa is out only with per-speed capabilities.
3709 * If set to NULL, the function returns the required number
3710 * of required array entries.
3712 * a number of elements in an speed_fec_capa array.
3715 * - A non-negative value lower or equal to num: success. The return value
3716 * is the number of entries filled in the fec capa array.
3717 * - A non-negative value higher than num: error, the given fec capa array
3718 * is too small. The return value corresponds to the num that should
3719 * be given to succeed. The entries in fec capa array are not valid and
3720 * shall not be used by the caller.
3721 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3723 * - (-EIO) if device is removed.
3724 * - (-ENODEV) if *port_id* invalid.
3725 * - (-EINVAL) if *num* or *speed_fec_capa* invalid
3728 int rte_eth_fec_get_capability(uint16_t port_id,
3729 struct rte_eth_fec_capa *speed_fec_capa,
3734 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3736 * Get current Forward Error Correction(FEC) mode.
3737 * If link is down and AUTO is enabled, AUTO is returned, otherwise,
3738 * configured FEC mode is returned.
3739 * If link is up, current FEC mode is returned.
3742 * The port identifier of the Ethernet device.
3744 * A bitmask of enabled FEC modes. If AUTO bit is set, other
3745 * bits specify FEC modes which may be negotiated. If AUTO
3746 * bit is clear, specify FEC modes to be used (only one valid
3747 * mode per speed may be set).
3749 * - (0) if successful.
3750 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3752 * - (-EIO) if device is removed.
3753 * - (-ENODEV) if *port_id* invalid.
3756 int rte_eth_fec_get(uint16_t port_id, uint32_t *fec_capa);
3760 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
3762 * Set Forward Error Correction(FEC) mode.
3765 * The port identifier of the Ethernet device.
3767 * A bitmask of allowed FEC modes. If AUTO bit is set, other
3768 * bits specify FEC modes which may be negotiated. If AUTO
3769 * bit is clear, specify FEC modes to be used (only one valid
3770 * mode per speed may be set).
3772 * - (0) if successful.
3773 * - (-EINVAL) if the FEC mode is not valid.
3774 * - (-ENOTSUP) if underlying hardware OR driver doesn't support.
3775 * - (-EIO) if device is removed.
3776 * - (-ENODEV) if *port_id* invalid.
3779 int rte_eth_fec_set(uint16_t port_id, uint32_t fec_capa);
3782 * Get current status of the Ethernet link flow control for Ethernet device
3785 * The port identifier of the Ethernet device.
3787 * The pointer to the structure where to store the flow control parameters.
3789 * - (0) if successful.
3790 * - (-ENOTSUP) if hardware doesn't support flow control.
3791 * - (-ENODEV) if *port_id* invalid.
3792 * - (-EIO) if device is removed.
3793 * - (-EINVAL) if bad parameter.
3795 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
3796 struct rte_eth_fc_conf *fc_conf);
3799 * Configure the Ethernet link flow control for Ethernet device
3802 * The port identifier of the Ethernet device.
3804 * The pointer to the structure of the flow control parameters.
3806 * - (0) if successful.
3807 * - (-ENOTSUP) if hardware doesn't support flow control mode.
3808 * - (-ENODEV) if *port_id* invalid.
3809 * - (-EINVAL) if bad parameter
3810 * - (-EIO) if flow control setup failure or device is removed.
3812 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
3813 struct rte_eth_fc_conf *fc_conf);
3816 * Configure the Ethernet priority flow control under DCB environment
3817 * for Ethernet device.
3820 * The port identifier of the Ethernet device.
3822 * The pointer to the structure of the priority flow control parameters.
3824 * - (0) if successful.
3825 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
3826 * - (-ENODEV) if *port_id* invalid.
3827 * - (-EINVAL) if bad parameter
3828 * - (-EIO) if flow control setup failure or device is removed.
3830 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
3831 struct rte_eth_pfc_conf *pfc_conf);
3834 * Add a MAC address to the set used for filtering incoming packets.
3837 * The port identifier of the Ethernet device.
3839 * The MAC address to add.
3841 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
3842 * not enabled, this should be set to 0.
3844 * - (0) if successfully added or *mac_addr* was already added.
3845 * - (-ENOTSUP) if hardware doesn't support this feature.
3846 * - (-ENODEV) if *port* is invalid.
3847 * - (-EIO) if device is removed.
3848 * - (-ENOSPC) if no more MAC addresses can be added.
3849 * - (-EINVAL) if MAC address is invalid.
3851 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct rte_ether_addr *mac_addr,
3855 * Remove a MAC address from the internal array of addresses.
3858 * The port identifier of the Ethernet device.
3860 * MAC address to remove.
3862 * - (0) if successful, or *mac_addr* didn't exist.
3863 * - (-ENOTSUP) if hardware doesn't support.
3864 * - (-ENODEV) if *port* invalid.
3865 * - (-EADDRINUSE) if attempting to remove the default MAC address.
3866 * - (-EINVAL) if MAC address is invalid.
3868 int rte_eth_dev_mac_addr_remove(uint16_t port_id,
3869 struct rte_ether_addr *mac_addr);
3872 * Set the default MAC address.
3875 * The port identifier of the Ethernet device.
3877 * New default MAC address.
3879 * - (0) if successful, or *mac_addr* didn't exist.
3880 * - (-ENOTSUP) if hardware doesn't support.
3881 * - (-ENODEV) if *port* invalid.
3882 * - (-EINVAL) if MAC address is invalid.
3884 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
3885 struct rte_ether_addr *mac_addr);
3888 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3891 * The port identifier of the Ethernet device.
3895 * Redirection table size. The table size can be queried by
3896 * rte_eth_dev_info_get().
3898 * - (0) if successful.
3899 * - (-ENODEV) if *port_id* is invalid.
3900 * - (-ENOTSUP) if hardware doesn't support.
3901 * - (-EINVAL) if bad parameter.
3902 * - (-EIO) if device is removed.
3904 int rte_eth_dev_rss_reta_update(uint16_t port_id,
3905 struct rte_eth_rss_reta_entry64 *reta_conf,
3906 uint16_t reta_size);
3909 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
3912 * The port identifier of the Ethernet device.
3914 * RETA to query. For each requested reta entry, corresponding bit
3915 * in mask must be set.
3917 * Redirection table size. The table size can be queried by
3918 * rte_eth_dev_info_get().
3920 * - (0) if successful.
3921 * - (-ENODEV) if *port_id* is invalid.
3922 * - (-ENOTSUP) if hardware doesn't support.
3923 * - (-EINVAL) if bad parameter.
3924 * - (-EIO) if device is removed.
3926 int rte_eth_dev_rss_reta_query(uint16_t port_id,
3927 struct rte_eth_rss_reta_entry64 *reta_conf,
3928 uint16_t reta_size);
3931 * Updates unicast hash table for receiving packet with the given destination
3932 * MAC address, and the packet is routed to all VFs for which the RX mode is
3933 * accept packets that match the unicast hash table.
3936 * The port identifier of the Ethernet device.
3938 * Unicast MAC address.
3940 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
3941 * 0 - Clear an unicast hash bit.
3943 * - (0) if successful.
3944 * - (-ENOTSUP) if hardware doesn't support.
3945 * - (-ENODEV) if *port_id* invalid.
3946 * - (-EIO) if device is removed.
3947 * - (-EINVAL) if bad parameter.
3949 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct rte_ether_addr *addr,
3953 * Updates all unicast hash bitmaps for receiving packet with any Unicast
3954 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
3955 * mode is accept packets that match the unicast hash table.
3958 * The port identifier of the Ethernet device.
3960 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
3962 * 0 - Clear all unicast hash bitmaps
3964 * - (0) if successful.
3965 * - (-ENOTSUP) if hardware doesn't support.
3966 * - (-ENODEV) if *port_id* invalid.
3967 * - (-EIO) if device is removed.
3968 * - (-EINVAL) if bad parameter.
3970 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
3973 * Set a traffic mirroring rule on an Ethernet device
3976 * The port identifier of the Ethernet device.
3977 * @param mirror_conf
3978 * The pointer to the traffic mirroring structure describing the mirroring rule.
3979 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
3980 * destination pool and the value of rule if enable vlan or pool mirroring.
3983 * The index of traffic mirroring rule, we support four separated rules.
3985 * 1 - Enable a mirroring rule.
3986 * 0 - Disable a mirroring rule.
3988 * - (0) if successful.
3989 * - (-ENOTSUP) if hardware doesn't support this feature.
3990 * - (-ENODEV) if *port_id* invalid.
3991 * - (-EIO) if device is removed.
3992 * - (-EINVAL) if the mr_conf information is not correct.
3994 int rte_eth_mirror_rule_set(uint16_t port_id,
3995 struct rte_eth_mirror_conf *mirror_conf,
4000 * Reset a traffic mirroring rule on an Ethernet device.
4003 * The port identifier of the Ethernet device.
4005 * The index of traffic mirroring rule, we support four separated rules.
4007 * - (0) if successful.
4008 * - (-ENOTSUP) if hardware doesn't support this feature.
4009 * - (-ENODEV) if *port_id* invalid.
4010 * - (-EIO) if device is removed.
4011 * - (-EINVAL) if bad parameter.
4013 int rte_eth_mirror_rule_reset(uint16_t port_id,
4017 * Set the rate limitation for a queue on an Ethernet device.
4020 * The port identifier of the Ethernet device.
4024 * The tx rate in Mbps. Allocated from the total port link speed.
4026 * - (0) if successful.
4027 * - (-ENOTSUP) if hardware doesn't support this feature.
4028 * - (-ENODEV) if *port_id* invalid.
4029 * - (-EIO) if device is removed.
4030 * - (-EINVAL) if bad parameter.
4032 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
4036 * Configuration of Receive Side Scaling hash computation of Ethernet device.
4039 * The port identifier of the Ethernet device.
4041 * The new configuration to use for RSS hash computation on the port.
4043 * - (0) if successful.
4044 * - (-ENODEV) if port identifier is invalid.
4045 * - (-EIO) if device is removed.
4046 * - (-ENOTSUP) if hardware doesn't support.
4047 * - (-EINVAL) if bad parameter.
4049 int rte_eth_dev_rss_hash_update(uint16_t port_id,
4050 struct rte_eth_rss_conf *rss_conf);
4053 * Retrieve current configuration of Receive Side Scaling hash computation
4054 * of Ethernet device.
4057 * The port identifier of the Ethernet device.
4059 * Where to store the current RSS hash configuration of the Ethernet device.
4061 * - (0) if successful.
4062 * - (-ENODEV) if port identifier is invalid.
4063 * - (-EIO) if device is removed.
4064 * - (-ENOTSUP) if hardware doesn't support RSS.
4065 * - (-EINVAL) if bad parameter.
4068 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
4069 struct rte_eth_rss_conf *rss_conf);
4072 * Add UDP tunneling port for a type of tunnel.
4074 * Some NICs may require such configuration to properly parse a tunnel
4075 * with any standard or custom UDP port.
4076 * The packets with this UDP port will be parsed for this type of tunnel.
4077 * The device parser will also check the rest of the tunnel headers
4078 * before classifying the packet.
4080 * With some devices, this API will affect packet classification, i.e.:
4081 * - mbuf.packet_type reported on Rx
4082 * - rte_flow rules with tunnel items
4085 * The port identifier of the Ethernet device.
4087 * UDP tunneling configuration.
4090 * - (0) if successful.
4091 * - (-ENODEV) if port identifier is invalid.
4092 * - (-EIO) if device is removed.
4093 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4096 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
4097 struct rte_eth_udp_tunnel *tunnel_udp);
4100 * Delete UDP tunneling port for a type of tunnel.
4102 * The packets with this UDP port will not be classified as this type of tunnel
4103 * anymore if the device use such mapping for tunnel packet classification.
4105 * @see rte_eth_dev_udp_tunnel_port_add
4108 * The port identifier of the Ethernet device.
4110 * UDP tunneling configuration.
4113 * - (0) if successful.
4114 * - (-ENODEV) if port identifier is invalid.
4115 * - (-EIO) if device is removed.
4116 * - (-ENOTSUP) if hardware doesn't support tunnel type.
4119 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
4120 struct rte_eth_udp_tunnel *tunnel_udp);
4123 * Get DCB information on an Ethernet device.
4126 * The port identifier of the Ethernet device.
4130 * - (0) if successful.
4131 * - (-ENODEV) if port identifier is invalid.
4132 * - (-EIO) if device is removed.
4133 * - (-ENOTSUP) if hardware doesn't support.
4134 * - (-EINVAL) if bad parameter.
4136 int rte_eth_dev_get_dcb_info(uint16_t port_id,
4137 struct rte_eth_dcb_info *dcb_info);
4139 struct rte_eth_rxtx_callback;
4142 * Add a callback to be called on packet RX on a given port and queue.
4144 * This API configures a function to be called for each burst of
4145 * packets received on a given NIC port queue. The return value is a pointer
4146 * that can be used to later remove the callback using
4147 * rte_eth_remove_rx_callback().
4149 * Multiple functions are called in the order that they are added.
4152 * The port identifier of the Ethernet device.
4154 * The queue on the Ethernet device on which the callback is to be added.
4156 * The callback function
4158 * A generic pointer parameter which will be passed to each invocation of the
4159 * callback function on this port and queue. Inter-thread synchronization
4160 * of any user data changes is the responsibility of the user.
4164 * On success, a pointer value which can later be used to remove the callback.
4166 const struct rte_eth_rxtx_callback *
4167 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
4168 rte_rx_callback_fn fn, void *user_param);
4171 * Add a callback that must be called first on packet RX on a given port
4174 * This API configures a first function to be called for each burst of
4175 * packets received on a given NIC port queue. The return value is a pointer
4176 * that can be used to later remove the callback using
4177 * rte_eth_remove_rx_callback().
4179 * Multiple functions are called in the order that they are added.
4182 * The port identifier of the Ethernet device.
4184 * The queue on the Ethernet device on which the callback is to be added.
4186 * The callback function
4188 * A generic pointer parameter which will be passed to each invocation of the
4189 * callback function on this port and queue. Inter-thread synchronization
4190 * of any user data changes is the responsibility of the user.
4194 * On success, a pointer value which can later be used to remove the callback.
4196 const struct rte_eth_rxtx_callback *
4197 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
4198 rte_rx_callback_fn fn, void *user_param);
4201 * Add a callback to be called on packet TX on a given port and queue.
4203 * This API configures a function to be called for each burst of
4204 * packets sent on a given NIC port queue. The return value is a pointer
4205 * that can be used to later remove the callback using
4206 * rte_eth_remove_tx_callback().
4208 * Multiple functions are called in the order that they are added.
4211 * The port identifier of the Ethernet device.
4213 * The queue on the Ethernet device on which the callback is to be added.
4215 * The callback function
4217 * A generic pointer parameter which will be passed to each invocation of the
4218 * callback function on this port and queue. Inter-thread synchronization
4219 * of any user data changes is the responsibility of the user.
4223 * On success, a pointer value which can later be used to remove the callback.
4225 const struct rte_eth_rxtx_callback *
4226 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
4227 rte_tx_callback_fn fn, void *user_param);
4230 * Remove an RX packet callback from a given port and queue.
4232 * This function is used to removed callbacks that were added to a NIC port
4233 * queue using rte_eth_add_rx_callback().
4235 * Note: the callback is removed from the callback list but it isn't freed
4236 * since the it may still be in use. The memory for the callback can be
4237 * subsequently freed back by the application by calling rte_free():
4239 * - Immediately - if the port is stopped, or the user knows that no
4240 * callbacks are in flight e.g. if called from the thread doing RX/TX
4243 * - After a short delay - where the delay is sufficient to allow any
4244 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4245 * used to detect when data plane threads have ceased referencing the
4249 * The port identifier of the Ethernet device.
4251 * The queue on the Ethernet device from which the callback is to be removed.
4253 * User supplied callback created via rte_eth_add_rx_callback().
4256 * - 0: Success. Callback was removed.
4257 * - -ENODEV: If *port_id* is invalid.
4258 * - -ENOTSUP: Callback support is not available.
4259 * - -EINVAL: The queue_id is out of range, or the callback
4260 * is NULL or not found for the port/queue.
4262 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
4263 const struct rte_eth_rxtx_callback *user_cb);
4266 * Remove a TX packet callback from a given port and queue.
4268 * This function is used to removed callbacks that were added to a NIC port
4269 * queue using rte_eth_add_tx_callback().
4271 * Note: the callback is removed from the callback list but it isn't freed
4272 * since the it may still be in use. The memory for the callback can be
4273 * subsequently freed back by the application by calling rte_free():
4275 * - Immediately - if the port is stopped, or the user knows that no
4276 * callbacks are in flight e.g. if called from the thread doing RX/TX
4279 * - After a short delay - where the delay is sufficient to allow any
4280 * in-flight callbacks to complete. Alternately, the RCU mechanism can be
4281 * used to detect when data plane threads have ceased referencing the
4285 * The port identifier of the Ethernet device.
4287 * The queue on the Ethernet device from which the callback is to be removed.
4289 * User supplied callback created via rte_eth_add_tx_callback().
4292 * - 0: Success. Callback was removed.
4293 * - -ENODEV: If *port_id* is invalid.
4294 * - -ENOTSUP: Callback support is not available.
4295 * - -EINVAL: The queue_id is out of range, or the callback
4296 * is NULL or not found for the port/queue.
4298 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
4299 const struct rte_eth_rxtx_callback *user_cb);
4302 * Retrieve information about given port's RX queue.
4305 * The port identifier of the Ethernet device.
4307 * The RX queue on the Ethernet device for which information
4308 * will be retrieved.
4310 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
4311 * the information of the Ethernet device.
4315 * - -ENODEV: If *port_id* is invalid.
4316 * - -ENOTSUP: routine is not supported by the device PMD.
4317 * - -EINVAL: The queue_id is out of range, or the queue
4320 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4321 struct rte_eth_rxq_info *qinfo);
4324 * Retrieve information about given port's TX queue.
4327 * The port identifier of the Ethernet device.
4329 * The TX queue on the Ethernet device for which information
4330 * will be retrieved.
4332 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
4333 * the information of the Ethernet device.
4337 * - -ENODEV: If *port_id* is invalid.
4338 * - -ENOTSUP: routine is not supported by the device PMD.
4339 * - -EINVAL: The queue_id is out of range, or the queue
4342 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
4343 struct rte_eth_txq_info *qinfo);
4346 * Retrieve information about the Rx packet burst mode.
4349 * The port identifier of the Ethernet device.
4351 * The Rx queue on the Ethernet device for which information
4352 * will be retrieved.
4354 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4355 * with the information of the packet burst mode.
4359 * - -ENODEV: If *port_id* is invalid.
4360 * - -ENOTSUP: routine is not supported by the device PMD.
4361 * - -EINVAL: The queue_id is out of range.
4363 int rte_eth_rx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4364 struct rte_eth_burst_mode *mode);
4367 * Retrieve information about the Tx packet burst mode.
4370 * The port identifier of the Ethernet device.
4372 * The Tx queue on the Ethernet device for which information
4373 * will be retrieved.
4375 * A pointer to a structure of type *rte_eth_burst_mode* to be filled
4376 * with the information of the packet burst mode.
4380 * - -ENODEV: If *port_id* is invalid.
4381 * - -ENOTSUP: routine is not supported by the device PMD.
4382 * - -EINVAL: The queue_id is out of range.
4384 int rte_eth_tx_burst_mode_get(uint16_t port_id, uint16_t queue_id,
4385 struct rte_eth_burst_mode *mode);
4389 * @b EXPERIMENTAL: this API may change without prior notice.
4391 * Retrieve the monitor condition for a given receive queue.
4394 * The port identifier of the Ethernet device.
4396 * The Rx queue on the Ethernet device for which information
4397 * will be retrieved.
4399 * The pointer to power-optimized monitoring condition structure.
4403 * -ENOTSUP: Operation not supported.
4404 * -EINVAL: Invalid parameters.
4405 * -ENODEV: Invalid port ID.
4408 int rte_eth_get_monitor_addr(uint16_t port_id, uint16_t queue_id,
4409 struct rte_power_monitor_cond *pmc);
4412 * Retrieve device registers and register attributes (number of registers and
4416 * The port identifier of the Ethernet device.
4418 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
4419 * NULL the function fills in the width and length fields. If non-NULL
4420 * the registers are put into the buffer pointed at by the data field.
4422 * - (0) if successful.
4423 * - (-ENOTSUP) if hardware doesn't support.
4424 * - (-EINVAL) if bad parameter.
4425 * - (-ENODEV) if *port_id* invalid.
4426 * - (-EIO) if device is removed.
4427 * - others depends on the specific operations implementation.
4429 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
4432 * Retrieve size of device EEPROM
4435 * The port identifier of the Ethernet device.
4437 * - (>=0) EEPROM size if successful.
4438 * - (-ENOTSUP) if hardware doesn't support.
4439 * - (-ENODEV) if *port_id* invalid.
4440 * - (-EIO) if device is removed.
4441 * - others depends on the specific operations implementation.
4443 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
4446 * Retrieve EEPROM and EEPROM attribute
4449 * The port identifier of the Ethernet device.
4451 * The template includes buffer for return EEPROM data and
4452 * EEPROM attributes to be filled.
4454 * - (0) if successful.
4455 * - (-ENOTSUP) if hardware doesn't support.
4456 * - (-EINVAL) if bad parameter.
4457 * - (-ENODEV) if *port_id* invalid.
4458 * - (-EIO) if device is removed.
4459 * - others depends on the specific operations implementation.
4461 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4464 * Program EEPROM with provided data
4467 * The port identifier of the Ethernet device.
4469 * The template includes EEPROM data for programming and
4470 * EEPROM attributes to be filled
4472 * - (0) if successful.
4473 * - (-ENOTSUP) if hardware doesn't support.
4474 * - (-ENODEV) if *port_id* invalid.
4475 * - (-EINVAL) if bad parameter.
4476 * - (-EIO) if device is removed.
4477 * - others depends on the specific operations implementation.
4479 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
4483 * @b EXPERIMENTAL: this API may change without prior notice.
4485 * Retrieve the type and size of plugin module EEPROM
4488 * The port identifier of the Ethernet device.
4490 * The type and size of plugin module EEPROM.
4492 * - (0) if successful.
4493 * - (-ENOTSUP) if hardware doesn't support.
4494 * - (-ENODEV) if *port_id* invalid.
4495 * - (-EINVAL) if bad parameter.
4496 * - (-EIO) if device is removed.
4497 * - others depends on the specific operations implementation.
4501 rte_eth_dev_get_module_info(uint16_t port_id,
4502 struct rte_eth_dev_module_info *modinfo);
4506 * @b EXPERIMENTAL: this API may change without prior notice.
4508 * Retrieve the data of plugin module EEPROM
4511 * The port identifier of the Ethernet device.
4513 * The template includes the plugin module EEPROM attributes, and the
4514 * buffer for return plugin module EEPROM data.
4516 * - (0) if successful.
4517 * - (-ENOTSUP) if hardware doesn't support.
4518 * - (-EINVAL) if bad parameter.
4519 * - (-ENODEV) if *port_id* invalid.
4520 * - (-EIO) if device is removed.
4521 * - others depends on the specific operations implementation.
4525 rte_eth_dev_get_module_eeprom(uint16_t port_id,
4526 struct rte_dev_eeprom_info *info);
4529 * Set the list of multicast addresses to filter on an Ethernet device.
4532 * The port identifier of the Ethernet device.
4533 * @param mc_addr_set
4534 * The array of multicast addresses to set. Equal to NULL when the function
4535 * is invoked to flush the set of filtered addresses.
4537 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
4538 * when the function is invoked to flush the set of filtered addresses.
4540 * - (0) if successful.
4541 * - (-ENODEV) if *port_id* invalid.
4542 * - (-EIO) if device is removed.
4543 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
4544 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
4545 * - (-EINVAL) if bad parameter.
4547 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
4548 struct rte_ether_addr *mc_addr_set,
4549 uint32_t nb_mc_addr);
4552 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
4555 * The port identifier of the Ethernet device.
4559 * - -ENODEV: The port ID is invalid.
4560 * - -EIO: if device is removed.
4561 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4563 int rte_eth_timesync_enable(uint16_t port_id);
4566 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
4569 * The port identifier of the Ethernet device.
4573 * - -ENODEV: The port ID is invalid.
4574 * - -EIO: if device is removed.
4575 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4577 int rte_eth_timesync_disable(uint16_t port_id);
4580 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
4583 * The port identifier of the Ethernet device.
4585 * Pointer to the timestamp struct.
4587 * Device specific flags. Used to pass the RX timesync register index to
4588 * i40e. Unused in igb/ixgbe, pass 0 instead.
4592 * - -EINVAL: No timestamp is available.
4593 * - -ENODEV: The port ID is invalid.
4594 * - -EIO: if device is removed.
4595 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4597 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
4598 struct timespec *timestamp, uint32_t flags);
4601 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
4604 * The port identifier of the Ethernet device.
4606 * Pointer to the timestamp struct.
4610 * - -EINVAL: No timestamp is available.
4611 * - -ENODEV: The port ID is invalid.
4612 * - -EIO: if device is removed.
4613 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4615 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
4616 struct timespec *timestamp);
4619 * Adjust the timesync clock on an Ethernet device.
4621 * This is usually used in conjunction with other Ethdev timesync functions to
4622 * synchronize the device time using the IEEE1588/802.1AS protocol.
4625 * The port identifier of the Ethernet device.
4627 * The adjustment in nanoseconds.
4631 * - -ENODEV: The port ID is invalid.
4632 * - -EIO: if device is removed.
4633 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4635 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
4638 * Read the time from the timesync clock on an Ethernet device.
4640 * This is usually used in conjunction with other Ethdev timesync functions to
4641 * synchronize the device time using the IEEE1588/802.1AS protocol.
4644 * The port identifier of the Ethernet device.
4646 * Pointer to the timespec struct that holds the time.
4650 * - -EINVAL: Bad parameter.
4652 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
4655 * Set the time of the timesync clock on an Ethernet device.
4657 * This is usually used in conjunction with other Ethdev timesync functions to
4658 * synchronize the device time using the IEEE1588/802.1AS protocol.
4661 * The port identifier of the Ethernet device.
4663 * Pointer to the timespec struct that holds the time.
4667 * - -EINVAL: No timestamp is available.
4668 * - -ENODEV: The port ID is invalid.
4669 * - -EIO: if device is removed.
4670 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4672 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
4676 * @b EXPERIMENTAL: this API may change without prior notice.
4678 * Read the current clock counter of an Ethernet device
4680 * This returns the current raw clock value of an Ethernet device. It is
4681 * a raw amount of ticks, with no given time reference.
4682 * The value returned here is from the same clock than the one
4683 * filling timestamp field of Rx packets when using hardware timestamp
4684 * offload. Therefore it can be used to compute a precise conversion of
4685 * the device clock to the real time.
4687 * E.g, a simple heuristic to derivate the frequency would be:
4688 * uint64_t start, end;
4689 * rte_eth_read_clock(port, start);
4690 * rte_delay_ms(100);
4691 * rte_eth_read_clock(port, end);
4692 * double freq = (end - start) * 10;
4694 * Compute a common reference with:
4695 * uint64_t base_time_sec = current_time();
4696 * uint64_t base_clock;
4697 * rte_eth_read_clock(port, base_clock);
4699 * Then, convert the raw mbuf timestamp with:
4700 * base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq;
4702 * This simple example will not provide a very good accuracy. One must
4703 * at least measure multiple times the frequency and do a regression.
4704 * To avoid deviation from the system time, the common reference can
4705 * be repeated from time to time. The integer division can also be
4706 * converted by a multiplication and a shift for better performance.
4709 * The port identifier of the Ethernet device.
4711 * Pointer to the uint64_t that holds the raw clock value.
4715 * - -ENODEV: The port ID is invalid.
4716 * - -ENOTSUP: The function is not supported by the Ethernet driver.
4717 * - -EINVAL: if bad parameter.
4721 rte_eth_read_clock(uint16_t port_id, uint64_t *clock);
4724 * Get the port id from device name. The device name should be specified
4726 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
4727 * - SoC device name, for example- fsl-gmac0
4728 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
4731 * pci address or name of the device
4733 * pointer to port identifier of the device
4735 * - (0) if successful and port_id is filled.
4736 * - (-ENODEV or -EINVAL) on failure.
4739 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
4742 * Get the device name from port id. The device name is specified as below:
4743 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
4744 * - SoC device name, for example- fsl-gmac0
4745 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
4748 * Port identifier of the device.
4750 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
4752 * - (0) if successful.
4753 * - (-ENODEV) if *port_id* is invalid.
4754 * - (-EINVAL) on failure.
4757 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
4760 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
4761 * the ethernet device information, otherwise adjust them to boundaries.
4764 * The port identifier of the Ethernet device.
4766 * A pointer to a uint16_t where the number of receive
4767 * descriptors stored.
4769 * A pointer to a uint16_t where the number of transmit
4770 * descriptors stored.
4772 * - (0) if successful.
4773 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
4775 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
4776 uint16_t *nb_rx_desc,
4777 uint16_t *nb_tx_desc);
4780 * Test if a port supports specific mempool ops.
4783 * Port identifier of the Ethernet device.
4785 * The name of the pool operations to test.
4787 * - 0: best mempool ops choice for this port.
4788 * - 1: mempool ops are supported for this port.
4789 * - -ENOTSUP: mempool ops not supported for this port.
4790 * - -ENODEV: Invalid port Identifier.
4791 * - -EINVAL: Pool param is null.
4794 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
4797 * Get the security context for the Ethernet device.
4800 * Port identifier of the Ethernet device
4803 * - pointer to security context on success.
4806 rte_eth_dev_get_sec_ctx(uint16_t port_id);
4810 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
4812 * Query the device hairpin capabilities.
4815 * The port identifier of the Ethernet device.
4817 * Pointer to a structure that will hold the hairpin capabilities.
4819 * - (0) if successful.
4820 * - (-ENOTSUP) if hardware doesn't support.
4821 * - (-EINVAL) if bad parameter.
4824 int rte_eth_dev_hairpin_capability_get(uint16_t port_id,
4825 struct rte_eth_hairpin_cap *cap);
4829 * @b EXPERIMENTAL: this structure may change without prior notice.
4831 * ethernet device representor ID range entry
4833 struct rte_eth_representor_range {
4834 enum rte_eth_representor_type type; /**< Representor type */
4835 int controller; /**< Controller index */
4836 int pf; /**< Physical function index */
4839 int vf; /**< VF start index */
4840 int sf; /**< SF start index */
4842 uint32_t id_base; /**< Representor ID start index */
4843 uint32_t id_end; /**< Representor ID end index */
4844 char name[RTE_DEV_NAME_MAX_LEN]; /**< Representor name */
4849 * @b EXPERIMENTAL: this structure may change without prior notice.
4851 * Ethernet device representor information
4853 struct rte_eth_representor_info {
4854 uint16_t controller; /**< Controller ID of caller device. */
4855 uint16_t pf; /**< Physical function ID of caller device. */
4856 uint32_t nb_ranges_alloc; /**< Size of the ranges array. */
4857 uint32_t nb_ranges; /**< Number of initialized ranges. */
4858 struct rte_eth_representor_range ranges[];/**< Representor ID range. */
4862 * Retrieve the representor info of the device.
4864 * Get device representor info to be able to calculate a unique
4865 * representor ID. @see rte_eth_representor_id_get helper.
4868 * The port identifier of the device.
4870 * A pointer to a representor info structure.
4871 * NULL to return number of range entries and allocate memory
4872 * for next call to store detail.
4873 * The number of ranges that were written into this structure
4874 * will be placed into its nb_ranges field. This number cannot be
4875 * larger than the nb_ranges_alloc that by the user before calling
4876 * this function. It can be smaller than the value returned by the
4877 * function, however.
4879 * - (-ENOTSUP) if operation is not supported.
4880 * - (-ENODEV) if *port_id* invalid.
4881 * - (-EIO) if device is removed.
4882 * - (>=0) number of available representor range entries.
4885 int rte_eth_representor_info_get(uint16_t port_id,
4886 struct rte_eth_representor_info *info);
4888 #include <rte_ethdev_core.h>
4892 * Retrieve a burst of input packets from a receive queue of an Ethernet
4893 * device. The retrieved packets are stored in *rte_mbuf* structures whose
4894 * pointers are supplied in the *rx_pkts* array.
4896 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
4897 * receive queue, up to *nb_pkts* packets, and for each completed RX
4898 * descriptor in the ring, it performs the following operations:
4900 * - Initialize the *rte_mbuf* data structure associated with the
4901 * RX descriptor according to the information provided by the NIC into
4902 * that RX descriptor.
4904 * - Store the *rte_mbuf* data structure into the next entry of the
4907 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
4908 * allocated from the memory pool associated with the receive queue at
4909 * initialization time.
4911 * When retrieving an input packet that was scattered by the controller
4912 * into multiple receive descriptors, the rte_eth_rx_burst() function
4913 * appends the associated *rte_mbuf* buffers to the first buffer of the
4916 * The rte_eth_rx_burst() function returns the number of packets
4917 * actually retrieved, which is the number of *rte_mbuf* data structures
4918 * effectively supplied into the *rx_pkts* array.
4919 * A return value equal to *nb_pkts* indicates that the RX queue contained
4920 * at least *rx_pkts* packets, and this is likely to signify that other
4921 * received packets remain in the input queue. Applications implementing
4922 * a "retrieve as much received packets as possible" policy can check this
4923 * specific case and keep invoking the rte_eth_rx_burst() function until
4924 * a value less than *nb_pkts* is returned.
4926 * This receive method has the following advantages:
4928 * - It allows a run-to-completion network stack engine to retrieve and
4929 * to immediately process received packets in a fast burst-oriented
4930 * approach, avoiding the overhead of unnecessary intermediate packet
4931 * queue/dequeue operations.
4933 * - Conversely, it also allows an asynchronous-oriented processing
4934 * method to retrieve bursts of received packets and to immediately
4935 * queue them for further parallel processing by another logical core,
4936 * for instance. However, instead of having received packets being
4937 * individually queued by the driver, this approach allows the caller
4938 * of the rte_eth_rx_burst() function to queue a burst of retrieved
4939 * packets at a time and therefore dramatically reduce the cost of
4940 * enqueue/dequeue operations per packet.
4942 * - It allows the rte_eth_rx_burst() function of the driver to take
4943 * advantage of burst-oriented hardware features (CPU cache,
4944 * prefetch instructions, and so on) to minimize the number of CPU
4945 * cycles per packet.
4947 * To summarize, the proposed receive API enables many
4948 * burst-oriented optimizations in both synchronous and asynchronous
4949 * packet processing environments with no overhead in both cases.
4952 * Some drivers using vector instructions require that *nb_pkts* is
4953 * divisible by 4 or 8, depending on the driver implementation.
4955 * The rte_eth_rx_burst() function does not provide any error
4956 * notification to avoid the corresponding overhead. As a hint, the
4957 * upper-level application might check the status of the device link once
4958 * being systematically returned a 0 value for a given number of tries.
4961 * The port identifier of the Ethernet device.
4963 * The index of the receive queue from which to retrieve input packets.
4964 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
4965 * to rte_eth_dev_configure().
4967 * The address of an array of pointers to *rte_mbuf* structures that
4968 * must be large enough to store *nb_pkts* pointers in it.
4970 * The maximum number of packets to retrieve.
4971 * The value must be divisible by 8 in order to work with any driver.
4973 * The number of packets actually retrieved, which is the number
4974 * of pointers to *rte_mbuf* structures effectively supplied to the
4977 static inline uint16_t
4978 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
4979 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
4981 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4984 #ifdef RTE_ETHDEV_DEBUG_RX
4985 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
4986 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
4988 if (queue_id >= dev->data->nb_rx_queues) {
4989 RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n", queue_id);
4993 nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
4996 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4997 struct rte_eth_rxtx_callback *cb;
4999 /* __ATOMIC_RELEASE memory order was used when the
5000 * call back was inserted into the list.
5001 * Since there is a clear dependency between loading
5002 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5005 cb = __atomic_load_n(&dev->post_rx_burst_cbs[queue_id],
5008 if (unlikely(cb != NULL)) {
5010 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
5011 nb_pkts, cb->param);
5013 } while (cb != NULL);
5017 rte_ethdev_trace_rx_burst(port_id, queue_id, (void **)rx_pkts, nb_rx);
5022 * Get the number of used descriptors of a rx queue
5025 * The port identifier of the Ethernet device.
5027 * The queue id on the specific port.
5029 * The number of used descriptors in the specific queue, or:
5030 * - (-ENODEV) if *port_id* is invalid.
5031 * (-EINVAL) if *queue_id* is invalid
5032 * (-ENOTSUP) if the device does not support this function
5035 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
5037 struct rte_eth_dev *dev;
5039 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5040 dev = &rte_eth_devices[port_id];
5041 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_queue_count, -ENOTSUP);
5042 if (queue_id >= dev->data->nb_rx_queues ||
5043 dev->data->rx_queues[queue_id] == NULL)
5046 return (int)(*dev->rx_queue_count)(dev, queue_id);
5050 * Check if the DD bit of the specific RX descriptor in the queue has been set
5053 * The port identifier of the Ethernet device.
5055 * The queue id on the specific port.
5057 * The offset of the descriptor ID from tail.
5059 * - (1) if the specific DD bit is set.
5060 * - (0) if the specific DD bit is not set.
5061 * - (-ENODEV) if *port_id* invalid.
5062 * - (-ENOTSUP) if the device does not support this function
5066 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
5068 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
5069 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5070 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_done, -ENOTSUP);
5071 return (*dev->rx_descriptor_done)(dev->data->rx_queues[queue_id], offset);
5074 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
5075 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
5076 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5079 * Check the status of a Rx descriptor in the queue
5081 * It should be called in a similar context than the Rx function:
5082 * - on a dataplane core
5083 * - not concurrently on the same queue
5085 * Since it's a dataplane function, no check is performed on port_id and
5086 * queue_id. The caller must therefore ensure that the port is enabled
5087 * and the queue is configured and running.
5089 * Note: accessing to a random descriptor in the ring may trigger cache
5090 * misses and have a performance impact.
5093 * A valid port identifier of the Ethernet device which.
5095 * A valid Rx queue identifier on this port.
5097 * The offset of the descriptor starting from tail (0 is the next
5098 * packet to be received by the driver).
5101 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
5103 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
5104 * not yet processed by the driver (i.e. in the receive queue).
5105 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
5106 * the driver and not yet returned to hw, or reserved by the hw.
5107 * - (-EINVAL) bad descriptor offset.
5108 * - (-ENOTSUP) if the device does not support this function.
5109 * - (-ENODEV) bad port or queue (only if compiled with debug).
5112 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
5115 struct rte_eth_dev *dev;
5118 #ifdef RTE_ETHDEV_DEBUG_RX
5119 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5121 dev = &rte_eth_devices[port_id];
5122 #ifdef RTE_ETHDEV_DEBUG_RX
5123 if (queue_id >= dev->data->nb_rx_queues)
5126 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_descriptor_status, -ENOTSUP);
5127 rxq = dev->data->rx_queues[queue_id];
5129 return (*dev->rx_descriptor_status)(rxq, offset);
5132 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
5133 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
5134 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
5137 * Check the status of a Tx descriptor in the queue.
5139 * It should be called in a similar context than the Tx function:
5140 * - on a dataplane core
5141 * - not concurrently on the same queue
5143 * Since it's a dataplane function, no check is performed on port_id and
5144 * queue_id. The caller must therefore ensure that the port is enabled
5145 * and the queue is configured and running.
5147 * Note: accessing to a random descriptor in the ring may trigger cache
5148 * misses and have a performance impact.
5151 * A valid port identifier of the Ethernet device which.
5153 * A valid Tx queue identifier on this port.
5155 * The offset of the descriptor starting from tail (0 is the place where
5156 * the next packet will be send).
5159 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
5160 * in the transmit queue.
5161 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
5162 * be reused by the driver.
5163 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
5164 * driver or the hardware.
5165 * - (-EINVAL) bad descriptor offset.
5166 * - (-ENOTSUP) if the device does not support this function.
5167 * - (-ENODEV) bad port or queue (only if compiled with debug).
5169 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
5170 uint16_t queue_id, uint16_t offset)
5172 struct rte_eth_dev *dev;
5175 #ifdef RTE_ETHDEV_DEBUG_TX
5176 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
5178 dev = &rte_eth_devices[port_id];
5179 #ifdef RTE_ETHDEV_DEBUG_TX
5180 if (queue_id >= dev->data->nb_tx_queues)
5183 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_descriptor_status, -ENOTSUP);
5184 txq = dev->data->tx_queues[queue_id];
5186 return (*dev->tx_descriptor_status)(txq, offset);
5190 * Send a burst of output packets on a transmit queue of an Ethernet device.
5192 * The rte_eth_tx_burst() function is invoked to transmit output packets
5193 * on the output queue *queue_id* of the Ethernet device designated by its
5195 * The *nb_pkts* parameter is the number of packets to send which are
5196 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5197 * allocated from a pool created with rte_pktmbuf_pool_create().
5198 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
5199 * up to the number of transmit descriptors available in the TX ring of the
5201 * For each packet to send, the rte_eth_tx_burst() function performs
5202 * the following operations:
5204 * - Pick up the next available descriptor in the transmit ring.
5206 * - Free the network buffer previously sent with that descriptor, if any.
5208 * - Initialize the transmit descriptor with the information provided
5209 * in the *rte_mbuf data structure.
5211 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
5212 * the rte_eth_tx_burst() function uses several transmit descriptors
5215 * The rte_eth_tx_burst() function returns the number of packets it
5216 * actually sent. A return value equal to *nb_pkts* means that all packets
5217 * have been sent, and this is likely to signify that other output packets
5218 * could be immediately transmitted again. Applications that implement a
5219 * "send as many packets to transmit as possible" policy can check this
5220 * specific case and keep invoking the rte_eth_tx_burst() function until
5221 * a value less than *nb_pkts* is returned.
5223 * It is the responsibility of the rte_eth_tx_burst() function to
5224 * transparently free the memory buffers of packets previously sent.
5225 * This feature is driven by the *tx_free_thresh* value supplied to the
5226 * rte_eth_dev_configure() function at device configuration time.
5227 * When the number of free TX descriptors drops below this threshold, the
5228 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
5229 * of those packets whose transmission was effectively completed.
5231 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
5232 * invoke this function concurrently on the same tx queue without SW lock.
5233 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
5235 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
5239 * The port identifier of the Ethernet device.
5241 * The index of the transmit queue through which output packets must be
5243 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5244 * to rte_eth_dev_configure().
5246 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5247 * which contain the output packets.
5249 * The maximum number of packets to transmit.
5251 * The number of output packets actually stored in transmit descriptors of
5252 * the transmit ring. The return value can be less than the value of the
5253 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
5255 static inline uint16_t
5256 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
5257 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5259 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
5261 #ifdef RTE_ETHDEV_DEBUG_TX
5262 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
5263 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
5265 if (queue_id >= dev->data->nb_tx_queues) {
5266 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5271 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
5272 struct rte_eth_rxtx_callback *cb;
5274 /* __ATOMIC_RELEASE memory order was used when the
5275 * call back was inserted into the list.
5276 * Since there is a clear dependency between loading
5277 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is
5280 cb = __atomic_load_n(&dev->pre_tx_burst_cbs[queue_id],
5283 if (unlikely(cb != NULL)) {
5285 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
5288 } while (cb != NULL);
5292 rte_ethdev_trace_tx_burst(port_id, queue_id, (void **)tx_pkts,
5294 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
5298 * Process a burst of output packets on a transmit queue of an Ethernet device.
5300 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
5301 * transmitted on the output queue *queue_id* of the Ethernet device designated
5303 * The *nb_pkts* parameter is the number of packets to be prepared which are
5304 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
5305 * allocated from a pool created with rte_pktmbuf_pool_create().
5306 * For each packet to send, the rte_eth_tx_prepare() function performs
5307 * the following operations:
5309 * - Check if packet meets devices requirements for tx offloads.
5311 * - Check limitations about number of segments.
5313 * - Check additional requirements when debug is enabled.
5315 * - Update and/or reset required checksums when tx offload is set for packet.
5317 * Since this function can modify packet data, provided mbufs must be safely
5318 * writable (e.g. modified data cannot be in shared segment).
5320 * The rte_eth_tx_prepare() function returns the number of packets ready to be
5321 * sent. A return value equal to *nb_pkts* means that all packets are valid and
5322 * ready to be sent, otherwise stops processing on the first invalid packet and
5323 * leaves the rest packets untouched.
5325 * When this functionality is not implemented in the driver, all packets are
5326 * are returned untouched.
5329 * The port identifier of the Ethernet device.
5330 * The value must be a valid port id.
5332 * The index of the transmit queue through which output packets must be
5334 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5335 * to rte_eth_dev_configure().
5337 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
5338 * which contain the output packets.
5340 * The maximum number of packets to process.
5342 * The number of packets correct and ready to be sent. The return value can be
5343 * less than the value of the *tx_pkts* parameter when some packet doesn't
5344 * meet devices requirements with rte_errno set appropriately:
5345 * - EINVAL: offload flags are not correctly set
5346 * - ENOTSUP: the offload feature is not supported by the hardware
5347 * - ENODEV: if *port_id* is invalid (with debug enabled only)
5351 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
5353 static inline uint16_t
5354 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
5355 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5357 struct rte_eth_dev *dev;
5359 #ifdef RTE_ETHDEV_DEBUG_TX
5360 if (!rte_eth_dev_is_valid_port(port_id)) {
5361 RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
5367 dev = &rte_eth_devices[port_id];
5369 #ifdef RTE_ETHDEV_DEBUG_TX
5370 if (queue_id >= dev->data->nb_tx_queues) {
5371 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
5377 if (!dev->tx_pkt_prepare)
5380 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
5387 * Native NOOP operation for compilation targets which doesn't require any
5388 * preparations steps, and functional NOOP may introduce unnecessary performance
5391 * Generally this is not a good idea to turn it on globally and didn't should
5392 * be used if behavior of tx_preparation can change.
5395 static inline uint16_t
5396 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
5397 __rte_unused uint16_t queue_id,
5398 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
5406 * Send any packets queued up for transmission on a port and HW queue
5408 * This causes an explicit flush of packets previously buffered via the
5409 * rte_eth_tx_buffer() function. It returns the number of packets successfully
5410 * sent to the NIC, and calls the error callback for any unsent packets. Unless
5411 * explicitly set up otherwise, the default callback simply frees the unsent
5412 * packets back to the owning mempool.
5415 * The port identifier of the Ethernet device.
5417 * The index of the transmit queue through which output packets must be
5419 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5420 * to rte_eth_dev_configure().
5422 * Buffer of packets to be transmit.
5424 * The number of packets successfully sent to the Ethernet device. The error
5425 * callback is called for any packets which could not be sent.
5427 static inline uint16_t
5428 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
5429 struct rte_eth_dev_tx_buffer *buffer)
5432 uint16_t to_send = buffer->length;
5437 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
5441 /* All packets sent, or to be dealt with by callback below */
5442 if (unlikely(sent != to_send))
5443 buffer->error_callback(&buffer->pkts[sent],
5444 (uint16_t)(to_send - sent),
5445 buffer->error_userdata);
5451 * Buffer a single packet for future transmission on a port and queue
5453 * This function takes a single mbuf/packet and buffers it for later
5454 * transmission on the particular port and queue specified. Once the buffer is
5455 * full of packets, an attempt will be made to transmit all the buffered
5456 * packets. In case of error, where not all packets can be transmitted, a
5457 * callback is called with the unsent packets as a parameter. If no callback
5458 * is explicitly set up, the unsent packets are just freed back to the owning
5459 * mempool. The function returns the number of packets actually sent i.e.
5460 * 0 if no buffer flush occurred, otherwise the number of packets successfully
5464 * The port identifier of the Ethernet device.
5466 * The index of the transmit queue through which output packets must be
5468 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
5469 * to rte_eth_dev_configure().
5471 * Buffer used to collect packets to be sent.
5473 * Pointer to the packet mbuf to be sent.
5475 * 0 = packet has been buffered for later transmission
5476 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
5477 * causing N packets to be sent, and the error callback to be called for
5480 static __rte_always_inline uint16_t
5481 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
5482 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
5484 buffer->pkts[buffer->length++] = tx_pkt;
5485 if (buffer->length < buffer->size)
5488 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
5495 #endif /* _RTE_ETHDEV_H_ */