1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2010-2017 Intel Corporation
11 * RTE Ethernet Device API
13 * The Ethernet Device API is composed of two parts:
15 * - The application-oriented Ethernet API that includes functions to setup
16 * an Ethernet device (configure it, setup its RX and TX queues and start it),
17 * to get its MAC address, the speed and the status of its physical link,
18 * to receive and to transmit packets, and so on.
20 * - The driver-oriented Ethernet API that exports functions allowing
21 * an Ethernet Poll Mode Driver (PMD) to allocate an Ethernet device instance,
22 * create memzone for HW rings and process registered callbacks, and so on.
23 * PMDs should include rte_ethdev_driver.h instead of this header.
25 * By default, all the functions of the Ethernet Device API exported by a PMD
26 * are lock-free functions which assume to not be invoked in parallel on
27 * different logical cores to work on the same target object. For instance,
28 * the receive function of a PMD cannot be invoked in parallel on two logical
29 * cores to poll the same RX queue [of the same port]. Of course, this function
30 * can be invoked in parallel by different logical cores on different RX queues.
31 * It is the responsibility of the upper level application to enforce this rule.
33 * If needed, parallel accesses by multiple logical cores to shared queues
34 * shall be explicitly protected by dedicated inline lock-aware functions
35 * built on top of their corresponding lock-free functions of the PMD API.
37 * In all functions of the Ethernet API, the Ethernet device is
38 * designated by an integer >= 0 named the device port identifier.
40 * At the Ethernet driver level, Ethernet devices are represented by a generic
41 * data structure of type *rte_eth_dev*.
43 * Ethernet devices are dynamically registered during the PCI probing phase
44 * performed at EAL initialization time.
45 * When an Ethernet device is being probed, an *rte_eth_dev* structure and
46 * a new port identifier are allocated for that device. Then, the eth_dev_init()
47 * function supplied by the Ethernet driver matching the probed PCI
48 * device is invoked to properly initialize the device.
50 * The role of the device init function consists of resetting the hardware,
51 * checking access to Non-volatile Memory (NVM), reading the MAC address
54 * If the device init operation is successful, the correspondence between
55 * the port identifier assigned to the new device and its associated
56 * *rte_eth_dev* structure is effectively registered.
57 * Otherwise, both the *rte_eth_dev* structure and the port identifier are
60 * The functions exported by the application Ethernet API to setup a device
61 * designated by its port identifier must be invoked in the following order:
62 * - rte_eth_dev_configure()
63 * - rte_eth_tx_queue_setup()
64 * - rte_eth_rx_queue_setup()
65 * - rte_eth_dev_start()
67 * Then, the network application can invoke, in any order, the functions
68 * exported by the Ethernet API to get the MAC address of a given device, to
69 * get the speed and the status of a device physical link, to receive/transmit
70 * [burst of] packets, and so on.
72 * If the application wants to change the configuration (i.e. call
73 * rte_eth_dev_configure(), rte_eth_tx_queue_setup(), or
74 * rte_eth_rx_queue_setup()), it must call rte_eth_dev_stop() first to stop the
75 * device and then do the reconfiguration before calling rte_eth_dev_start()
76 * again. The transmit and receive functions should not be invoked when the
79 * Please note that some configuration is not stored between calls to
80 * rte_eth_dev_stop()/rte_eth_dev_start(). The following configuration will
83 * - flow control settings
84 * - receive mode configuration (promiscuous mode, hardware checksum mode,
85 * RSS/VMDQ settings etc.)
86 * - VLAN filtering configuration
87 * - MAC addresses supplied to MAC address array
88 * - flow director filtering mode (but not filtering rules)
89 * - NIC queue statistics mappings
91 * Any other configuration will not be stored and will need to be re-entered
92 * before a call to rte_eth_dev_start().
94 * Finally, a network application can close an Ethernet device by invoking the
95 * rte_eth_dev_close() function.
97 * Each function of the application Ethernet API invokes a specific function
98 * of the PMD that controls the target device designated by its port
100 * For this purpose, all device-specific functions of an Ethernet driver are
101 * supplied through a set of pointers contained in a generic structure of type
103 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev*
104 * structure by the device init function of the Ethernet driver, which is
105 * invoked during the PCI probing phase, as explained earlier.
107 * In other words, each function of the Ethernet API simply retrieves the
108 * *rte_eth_dev* structure associated with the device port identifier and
109 * performs an indirect invocation of the corresponding driver function
110 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure.
112 * For performance reasons, the address of the burst-oriented RX and TX
113 * functions of the Ethernet driver are not contained in the *eth_dev_ops*
114 * structure. Instead, they are directly stored at the beginning of the
115 * *rte_eth_dev* structure to avoid an extra indirect memory access during
118 * RTE ethernet device drivers do not use interrupts for transmitting or
119 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit
120 * functions to applications.
121 * Both receive and transmit functions are packet-burst oriented to minimize
122 * their cost per packet through the following optimizations:
124 * - Sharing among multiple packets the incompressible cost of the
125 * invocation of receive/transmit functions.
127 * - Enabling receive/transmit functions to take advantage of burst-oriented
128 * hardware features (L1 cache, prefetch instructions, NIC head/tail
129 * registers) to minimize the number of CPU cycles per packet, for instance,
130 * by avoiding useless read memory accesses to ring descriptors, or by
131 * systematically using arrays of pointers that exactly fit L1 cache line
132 * boundaries and sizes.
134 * The burst-oriented receive function does not provide any error notification,
135 * to avoid the corresponding overhead. As a hint, the upper-level application
136 * might check the status of the device link once being systematically returned
137 * a 0 value by the receive function of the driver for a given number of tries.
146 /* Use this macro to check if LRO API is supported */
147 #define RTE_ETHDEV_HAS_LRO_SUPPORT
149 #include <rte_compat.h>
151 #include <rte_interrupts.h>
153 #include <rte_devargs.h>
154 #include <rte_errno.h>
155 #include <rte_common.h>
156 #include <rte_config.h>
158 #include "rte_ether.h"
159 #include "rte_eth_ctrl.h"
160 #include "rte_dev_info.h"
162 extern int rte_eth_dev_logtype;
164 #define RTE_ETHDEV_LOG(level, ...) \
165 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
171 * @b EXPERIMENTAL: this API may change without prior notice.
173 * Initializes a device iterator.
175 * This iterator allows accessing a list of devices matching some devargs.
178 * Device iterator handle initialized by the function.
179 * The fields bus_str and cls_str might be dynamically allocated,
180 * and could be freed by calling rte_eth_iterator_cleanup().
183 * Device description string.
186 * 0 on successful initialization, negative otherwise.
189 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs);
193 * @b EXPERIMENTAL: this API may change without prior notice.
195 * Iterates on devices with devargs filter.
196 * The ownership is not checked.
198 * The next port id is returned, and the iterator is updated.
201 * Device iterator handle initialized by rte_eth_iterator_init().
202 * Some fields bus_str and cls_str might be freed when no more port is found,
203 * by calling rte_eth_iterator_cleanup().
206 * A port id if found, RTE_MAX_ETHPORTS otherwise.
209 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter);
213 * @b EXPERIMENTAL: this API may change without prior notice.
215 * Free some allocated fields of the iterator.
217 * This function is automatically called by rte_eth_iterator_next()
218 * on the last iteration (i.e. when no more matching port is found).
220 * It is safe to call this function twice; it will do nothing more.
223 * Device iterator handle initialized by rte_eth_iterator_init().
224 * The fields bus_str and cls_str are freed if needed.
227 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter);
230 * Macro to iterate over all ethdev ports matching some devargs.
232 * If a break is done before the end of the loop,
233 * the function rte_eth_iterator_cleanup() must be called.
236 * Iterated port id of type uint16_t.
238 * Device parameters input as string of type char*.
240 * Iterator handle of type struct rte_dev_iterator, used internally.
242 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \
243 for (rte_eth_iterator_init(iter, devargs), \
244 id = rte_eth_iterator_next(iter); \
245 id != RTE_MAX_ETHPORTS; \
246 id = rte_eth_iterator_next(iter))
249 * A structure used to retrieve statistics for an Ethernet port.
250 * Not all statistics fields in struct rte_eth_stats are supported
251 * by any type of network interface card (NIC). If any statistics
252 * field is not supported, its value is 0.
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 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
267 /**< Total number of queue RX packets. */
268 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
269 /**< Total number of queue TX packets. */
270 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
271 /**< Total number of successfully received queue bytes. */
272 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
273 /**< Total number of successfully transmitted queue bytes. */
274 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
275 /**< Total number of queue packets received that are dropped. */
279 * Device supported speeds bitmap flags
281 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
282 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
283 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
284 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
285 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
286 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
287 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
288 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
289 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
290 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
291 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
292 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
293 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
294 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
295 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
296 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
299 * Ethernet numeric link speeds in Mbps
301 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
302 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
303 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
304 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
305 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
306 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
307 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
308 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
309 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
310 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
311 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
312 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
313 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
316 * A structure used to retrieve link-level information of an Ethernet port.
319 struct rte_eth_link {
320 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
321 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
322 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
323 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
324 } __attribute__((aligned(8))); /**< aligned for atomic64 read/write */
326 /* Utility constants */
327 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
328 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
329 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
330 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
331 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
332 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
335 * A structure used to configure the ring threshold registers of an RX/TX
336 * queue for an Ethernet port.
338 struct rte_eth_thresh {
339 uint8_t pthresh; /**< Ring prefetch threshold. */
340 uint8_t hthresh; /**< Ring host threshold. */
341 uint8_t wthresh; /**< Ring writeback threshold. */
345 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
347 #define ETH_MQ_RX_RSS_FLAG 0x1
348 #define ETH_MQ_RX_DCB_FLAG 0x2
349 #define ETH_MQ_RX_VMDQ_FLAG 0x4
352 * A set of values to identify what method is to be used to route
353 * packets to multiple queues.
355 enum rte_eth_rx_mq_mode {
356 /** None of DCB,RSS or VMDQ mode */
359 /** For RX side, only RSS is on */
360 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
361 /** For RX side,only DCB is on. */
362 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
363 /** Both DCB and RSS enable */
364 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
366 /** Only VMDQ, no RSS nor DCB */
367 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
368 /** RSS mode with VMDQ */
369 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
370 /** Use VMDQ+DCB to route traffic to queues */
371 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
372 /** Enable both VMDQ and DCB in VMDq */
373 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
378 * for rx mq mode backward compatible
380 #define ETH_RSS ETH_MQ_RX_RSS
381 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
382 #define ETH_DCB_RX ETH_MQ_RX_DCB
385 * A set of values to identify what method is to be used to transmit
386 * packets using multi-TCs.
388 enum rte_eth_tx_mq_mode {
389 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
390 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
391 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
392 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
396 * for tx mq mode backward compatible
398 #define ETH_DCB_NONE ETH_MQ_TX_NONE
399 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
400 #define ETH_DCB_TX ETH_MQ_TX_DCB
403 * A structure used to configure the RX features of an Ethernet port.
405 struct rte_eth_rxmode {
406 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
407 enum rte_eth_rx_mq_mode mq_mode;
408 uint32_t max_rx_pkt_len; /**< Only used if JUMBO_FRAME enabled. */
409 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
411 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
412 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
413 * structure are allowed to be set.
419 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
420 * Note that single VLAN is treated the same as inner VLAN.
423 ETH_VLAN_TYPE_UNKNOWN = 0,
424 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
425 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
430 * A structure used to describe a vlan filter.
431 * If the bit corresponding to a VID is set, such VID is on.
433 struct rte_vlan_filter_conf {
438 * A structure used to configure the Receive Side Scaling (RSS) feature
439 * of an Ethernet port.
440 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
441 * to an array holding the RSS key to use for hashing specific header
442 * fields of received packets. The length of this array should be indicated
443 * by *rss_key_len* below. Otherwise, a default random hash key is used by
446 * The *rss_key_len* field of the *rss_conf* structure indicates the length
447 * in bytes of the array pointed by *rss_key*. To be compatible, this length
448 * will be checked in i40e only. Others assume 40 bytes to be used as before.
450 * The *rss_hf* field of the *rss_conf* structure indicates the different
451 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
452 * Supplying an *rss_hf* equal to zero disables the RSS feature.
454 struct rte_eth_rss_conf {
455 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
456 uint8_t rss_key_len; /**< hash key length in bytes. */
457 uint64_t rss_hf; /**< Hash functions to apply - see below. */
461 * The RSS offload types are defined based on flow types which are defined
462 * in rte_eth_ctrl.h. Different NIC hardwares may support different RSS offload
463 * types. The supported flow types or RSS offload types can be queried by
464 * rte_eth_dev_info_get().
466 #define ETH_RSS_IPV4 (1ULL << RTE_ETH_FLOW_IPV4)
467 #define ETH_RSS_FRAG_IPV4 (1ULL << RTE_ETH_FLOW_FRAG_IPV4)
468 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_TCP)
469 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_UDP)
470 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_SCTP)
471 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_OTHER)
472 #define ETH_RSS_IPV6 (1ULL << RTE_ETH_FLOW_IPV6)
473 #define ETH_RSS_FRAG_IPV6 (1ULL << RTE_ETH_FLOW_FRAG_IPV6)
474 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_TCP)
475 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_UDP)
476 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_SCTP)
477 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_OTHER)
478 #define ETH_RSS_L2_PAYLOAD (1ULL << RTE_ETH_FLOW_L2_PAYLOAD)
479 #define ETH_RSS_IPV6_EX (1ULL << RTE_ETH_FLOW_IPV6_EX)
480 #define ETH_RSS_IPV6_TCP_EX (1ULL << RTE_ETH_FLOW_IPV6_TCP_EX)
481 #define ETH_RSS_IPV6_UDP_EX (1ULL << RTE_ETH_FLOW_IPV6_UDP_EX)
482 #define ETH_RSS_PORT (1ULL << RTE_ETH_FLOW_PORT)
483 #define ETH_RSS_VXLAN (1ULL << RTE_ETH_FLOW_VXLAN)
484 #define ETH_RSS_GENEVE (1ULL << RTE_ETH_FLOW_GENEVE)
485 #define ETH_RSS_NVGRE (1ULL << RTE_ETH_FLOW_NVGRE)
487 #define ETH_RSS_IP ( \
489 ETH_RSS_FRAG_IPV4 | \
490 ETH_RSS_NONFRAG_IPV4_OTHER | \
492 ETH_RSS_FRAG_IPV6 | \
493 ETH_RSS_NONFRAG_IPV6_OTHER | \
496 #define ETH_RSS_UDP ( \
497 ETH_RSS_NONFRAG_IPV4_UDP | \
498 ETH_RSS_NONFRAG_IPV6_UDP | \
501 #define ETH_RSS_TCP ( \
502 ETH_RSS_NONFRAG_IPV4_TCP | \
503 ETH_RSS_NONFRAG_IPV6_TCP | \
506 #define ETH_RSS_SCTP ( \
507 ETH_RSS_NONFRAG_IPV4_SCTP | \
508 ETH_RSS_NONFRAG_IPV6_SCTP)
510 #define ETH_RSS_TUNNEL ( \
515 /**< Mask of valid RSS hash protocols */
516 #define ETH_RSS_PROTO_MASK ( \
518 ETH_RSS_FRAG_IPV4 | \
519 ETH_RSS_NONFRAG_IPV4_TCP | \
520 ETH_RSS_NONFRAG_IPV4_UDP | \
521 ETH_RSS_NONFRAG_IPV4_SCTP | \
522 ETH_RSS_NONFRAG_IPV4_OTHER | \
524 ETH_RSS_FRAG_IPV6 | \
525 ETH_RSS_NONFRAG_IPV6_TCP | \
526 ETH_RSS_NONFRAG_IPV6_UDP | \
527 ETH_RSS_NONFRAG_IPV6_SCTP | \
528 ETH_RSS_NONFRAG_IPV6_OTHER | \
529 ETH_RSS_L2_PAYLOAD | \
531 ETH_RSS_IPV6_TCP_EX | \
532 ETH_RSS_IPV6_UDP_EX | \
539 * Definitions used for redirection table entry size.
540 * Some RSS RETA sizes may not be supported by some drivers, check the
541 * documentation or the description of relevant functions for more details.
543 #define ETH_RSS_RETA_SIZE_64 64
544 #define ETH_RSS_RETA_SIZE_128 128
545 #define ETH_RSS_RETA_SIZE_256 256
546 #define ETH_RSS_RETA_SIZE_512 512
547 #define RTE_RETA_GROUP_SIZE 64
549 /* Definitions used for VMDQ and DCB functionality */
550 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
551 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
552 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
553 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
555 /* DCB capability defines */
556 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
557 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
559 /* Definitions used for VLAN Offload functionality */
560 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
561 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
562 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
564 /* Definitions used for mask VLAN setting */
565 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
566 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
567 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
568 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
570 /* Definitions used for receive MAC address */
571 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
573 /* Definitions used for unicast hash */
574 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
576 /* Definitions used for VMDQ pool rx mode setting */
577 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
578 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
579 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
580 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
581 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
583 /** Maximum nb. of vlan per mirror rule */
584 #define ETH_MIRROR_MAX_VLANS 64
586 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
587 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
588 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
589 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
590 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
593 * A structure used to configure VLAN traffic mirror of an Ethernet port.
595 struct rte_eth_vlan_mirror {
596 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
597 /** VLAN ID list for vlan mirroring. */
598 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
602 * A structure used to configure traffic mirror of an Ethernet port.
604 struct rte_eth_mirror_conf {
605 uint8_t rule_type; /**< Mirroring rule type */
606 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
607 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
608 /** VLAN ID setting for VLAN mirroring. */
609 struct rte_eth_vlan_mirror vlan;
613 * A structure used to configure 64 entries of Redirection Table of the
614 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
615 * more than 64 entries supported by hardware, an array of this structure
618 struct rte_eth_rss_reta_entry64 {
620 /**< Mask bits indicate which entries need to be updated/queried. */
621 uint16_t reta[RTE_RETA_GROUP_SIZE];
622 /**< Group of 64 redirection table entries. */
626 * This enum indicates the possible number of traffic classes
627 * in DCB configurations
629 enum rte_eth_nb_tcs {
630 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
631 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
635 * This enum indicates the possible number of queue pools
636 * in VMDQ configurations.
638 enum rte_eth_nb_pools {
639 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
640 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
641 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
642 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
645 /* This structure may be extended in future. */
646 struct rte_eth_dcb_rx_conf {
647 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
648 /** Traffic class each UP mapped to. */
649 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
652 struct rte_eth_vmdq_dcb_tx_conf {
653 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
654 /** Traffic class each UP mapped to. */
655 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
658 struct rte_eth_dcb_tx_conf {
659 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
660 /** Traffic class each UP mapped to. */
661 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
664 struct rte_eth_vmdq_tx_conf {
665 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
669 * A structure used to configure the VMDQ+DCB feature
670 * of an Ethernet port.
672 * Using this feature, packets are routed to a pool of queues, based
673 * on the vlan id in the vlan tag, and then to a specific queue within
674 * that pool, using the user priority vlan tag field.
676 * A default pool may be used, if desired, to route all traffic which
677 * does not match the vlan filter rules.
679 struct rte_eth_vmdq_dcb_conf {
680 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
681 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
682 uint8_t default_pool; /**< The default pool, if applicable */
683 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
685 uint16_t vlan_id; /**< The vlan id of the received frame */
686 uint64_t pools; /**< Bitmask of pools for packet rx */
687 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
688 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
689 /**< Selects a queue in a pool */
693 * A structure used to configure the VMDQ feature of an Ethernet port when
694 * not combined with the DCB feature.
696 * Using this feature, packets are routed to a pool of queues. By default,
697 * the pool selection is based on the MAC address, the vlan id in the
698 * vlan tag as specified in the pool_map array.
699 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
700 * selection using only the MAC address. MAC address to pool mapping is done
701 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
702 * corresponding to the pool id.
704 * Queue selection within the selected pool will be done using RSS when
705 * it is enabled or revert to the first queue of the pool if not.
707 * A default pool may be used, if desired, to route all traffic which
708 * does not match the vlan filter rules or any pool MAC address.
710 struct rte_eth_vmdq_rx_conf {
711 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
712 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
713 uint8_t default_pool; /**< The default pool, if applicable */
714 uint8_t enable_loop_back; /**< Enable VT loop back */
715 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
716 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
718 uint16_t vlan_id; /**< The vlan id of the received frame */
719 uint64_t pools; /**< Bitmask of pools for packet rx */
720 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
724 * A structure used to configure the TX features of an Ethernet port.
726 struct rte_eth_txmode {
727 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
729 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
730 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
731 * structure are allowed to be set.
735 /* For i40e specifically */
738 uint8_t hw_vlan_reject_tagged : 1,
739 /**< If set, reject sending out tagged pkts */
740 hw_vlan_reject_untagged : 1,
741 /**< If set, reject sending out untagged pkts */
742 hw_vlan_insert_pvid : 1;
743 /**< If set, enable port based VLAN insertion */
747 * A structure used to configure an RX ring of an Ethernet port.
749 struct rte_eth_rxconf {
750 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
751 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
752 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
753 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
755 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
756 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
757 * fields on rte_eth_dev_info structure are allowed to be set.
763 * A structure used to configure a TX ring of an Ethernet port.
765 struct rte_eth_txconf {
766 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
767 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
768 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
769 less free descriptors than this value. */
771 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
773 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
774 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
775 * fields on rte_eth_dev_info structure are allowed to be set.
781 * A structure contains information about HW descriptor ring limitations.
783 struct rte_eth_desc_lim {
784 uint16_t nb_max; /**< Max allowed number of descriptors. */
785 uint16_t nb_min; /**< Min allowed number of descriptors. */
786 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
789 * Max allowed number of segments per whole packet.
791 * - For TSO packet this is the total number of data descriptors allowed
794 * @see nb_mtu_seg_max
799 * Max number of segments per one MTU.
801 * - For non-TSO packet, this is the maximum allowed number of segments
802 * in a single transmit packet.
804 * - For TSO packet each segment within the TSO may span up to this
809 uint16_t nb_mtu_seg_max;
813 * This enum indicates the flow control mode
815 enum rte_eth_fc_mode {
816 RTE_FC_NONE = 0, /**< Disable flow control. */
817 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
818 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
819 RTE_FC_FULL /**< Enable flow control on both side. */
823 * A structure used to configure Ethernet flow control parameter.
824 * These parameters will be configured into the register of the NIC.
825 * Please refer to the corresponding data sheet for proper value.
827 struct rte_eth_fc_conf {
828 uint32_t high_water; /**< High threshold value to trigger XOFF */
829 uint32_t low_water; /**< Low threshold value to trigger XON */
830 uint16_t pause_time; /**< Pause quota in the Pause frame */
831 uint16_t send_xon; /**< Is XON frame need be sent */
832 enum rte_eth_fc_mode mode; /**< Link flow control mode */
833 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
834 uint8_t autoneg; /**< Use Pause autoneg */
838 * A structure used to configure Ethernet priority flow control parameter.
839 * These parameters will be configured into the register of the NIC.
840 * Please refer to the corresponding data sheet for proper value.
842 struct rte_eth_pfc_conf {
843 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
844 uint8_t priority; /**< VLAN User Priority. */
848 * Memory space that can be configured to store Flow Director filters
849 * in the board memory.
851 enum rte_fdir_pballoc_type {
852 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
853 RTE_FDIR_PBALLOC_128K, /**< 128k. */
854 RTE_FDIR_PBALLOC_256K, /**< 256k. */
858 * Select report mode of FDIR hash information in RX descriptors.
860 enum rte_fdir_status_mode {
861 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
862 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
863 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
867 * A structure used to configure the Flow Director (FDIR) feature
868 * of an Ethernet port.
870 * If mode is RTE_FDIR_DISABLE, the pballoc value is ignored.
872 struct rte_fdir_conf {
873 enum rte_fdir_mode mode; /**< Flow Director mode. */
874 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
875 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
876 /** RX queue of packets matching a "drop" filter in perfect mode. */
878 struct rte_eth_fdir_masks mask;
879 struct rte_eth_fdir_flex_conf flex_conf;
880 /**< Flex payload configuration. */
884 * UDP tunneling configuration.
885 * Used to config the UDP port for a type of tunnel.
886 * NICs need the UDP port to identify the tunnel type.
887 * Normally a type of tunnel has a default UDP port, this structure can be used
888 * in case if the users want to change or support more UDP port.
890 struct rte_eth_udp_tunnel {
891 uint16_t udp_port; /**< UDP port used for the tunnel. */
892 uint8_t prot_type; /**< Tunnel type. Defined in rte_eth_tunnel_type. */
896 * A structure used to enable/disable specific device interrupts.
898 struct rte_intr_conf {
899 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
901 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
903 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
908 * A structure used to configure an Ethernet port.
909 * Depending upon the RX multi-queue mode, extra advanced
910 * configuration settings may be needed.
912 struct rte_eth_conf {
913 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
914 used. ETH_LINK_SPEED_FIXED disables link
915 autonegotiation, and a unique speed shall be
916 set. Otherwise, the bitmap defines the set of
917 speeds to be advertised. If the special value
918 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
919 supported are advertised. */
920 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
921 struct rte_eth_txmode txmode; /**< Port TX configuration. */
922 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
923 is 0, meaning the loopback mode is disabled.
924 Read the datasheet of given ethernet controller
925 for details. The possible values of this field
926 are defined in implementation of each driver. */
928 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
929 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
930 /**< Port vmdq+dcb configuration. */
931 struct rte_eth_dcb_rx_conf dcb_rx_conf;
932 /**< Port dcb RX configuration. */
933 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
934 /**< Port vmdq RX configuration. */
935 } rx_adv_conf; /**< Port RX filtering configuration. */
937 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
938 /**< Port vmdq+dcb TX configuration. */
939 struct rte_eth_dcb_tx_conf dcb_tx_conf;
940 /**< Port dcb TX configuration. */
941 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
942 /**< Port vmdq TX configuration. */
943 } tx_adv_conf; /**< Port TX DCB configuration (union). */
944 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
945 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
946 uint32_t dcb_capability_en;
947 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. */
948 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
952 * RX offload capabilities of a device.
954 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
955 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
956 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
957 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
958 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
959 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
960 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
961 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
962 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
963 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
964 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
965 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
966 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
967 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
968 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
969 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
970 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
971 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
973 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
974 DEV_RX_OFFLOAD_UDP_CKSUM | \
975 DEV_RX_OFFLOAD_TCP_CKSUM)
976 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
977 DEV_RX_OFFLOAD_VLAN_FILTER | \
978 DEV_RX_OFFLOAD_VLAN_EXTEND)
981 * If new Rx offload capabilities are defined, they also must be
982 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
986 * TX offload capabilities of a device.
988 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
989 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
990 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
991 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
992 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
993 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
994 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
995 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
996 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
997 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
998 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
999 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
1000 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
1001 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
1002 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
1003 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
1004 * tx queue without SW lock.
1006 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
1007 /**< Device supports multi segment send. */
1008 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
1009 /**< Device supports optimization for fast release of mbufs.
1010 * When set application must guarantee that per-queue all mbufs comes from
1011 * the same mempool and has refcnt = 1.
1013 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
1015 * Device supports generic UDP tunneled packet TSO.
1016 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
1019 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
1021 * Device supports generic IP tunneled packet TSO.
1022 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
1025 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
1026 /** Device supports outer UDP checksum */
1027 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
1029 * Device supports match on metadata Tx offload..
1030 * Application must set PKT_TX_METADATA and mbuf metadata field.
1032 #define DEV_TX_OFFLOAD_MATCH_METADATA 0x00200000
1034 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
1035 /**< Device supports Rx queue setup after device started*/
1036 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
1037 /**< Device supports Tx queue setup after device started*/
1040 * If new Tx offload capabilities are defined, they also must be
1041 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1045 * Fallback default preferred Rx/Tx port parameters.
1046 * These are used if an application requests default parameters
1047 * but the PMD does not provide preferred values.
1049 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1050 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1051 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1052 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1055 * Preferred Rx/Tx port parameters.
1056 * There are separate instances of this structure for transmission
1057 * and reception respectively.
1059 struct rte_eth_dev_portconf {
1060 uint16_t burst_size; /**< Device-preferred burst size */
1061 uint16_t ring_size; /**< Device-preferred size of queue rings */
1062 uint16_t nb_queues; /**< Device-preferred number of queues */
1066 * Default values for switch domain id when ethdev does not support switch
1067 * domain definitions.
1069 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (0)
1072 * Ethernet device associated switch information
1074 struct rte_eth_switch_info {
1075 const char *name; /**< switch name */
1076 uint16_t domain_id; /**< switch domain id */
1079 * mapping to the devices physical switch port as enumerated from the
1080 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1081 * device this may correspond to the VF_ID of each virtual function,
1082 * but each driver should explicitly define the mapping of switch
1083 * port identifier to that physical interconnect/switch
1088 * Ethernet device information
1092 * A structure used to retrieve the contextual information of
1093 * an Ethernet device, such as the controlling driver of the
1096 struct rte_eth_dev_info {
1097 struct rte_device *device; /** Generic device information */
1098 const char *driver_name; /**< Device Driver name. */
1099 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1100 Use if_indextoname() to translate into an interface name. */
1101 const uint32_t *dev_flags; /**< Device flags */
1102 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1103 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1104 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1105 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1106 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1107 uint32_t max_hash_mac_addrs;
1108 /** Maximum number of hash MAC addresses for MTA and UTA. */
1109 uint16_t max_vfs; /**< Maximum number of VFs. */
1110 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1111 uint64_t rx_offload_capa;
1112 /**< All RX offload capabilities including all per-queue ones */
1113 uint64_t tx_offload_capa;
1114 /**< All TX offload capabilities including all per-queue ones */
1115 uint64_t rx_queue_offload_capa;
1116 /**< Device per-queue RX offload capabilities. */
1117 uint64_t tx_queue_offload_capa;
1118 /**< Device per-queue TX offload capabilities. */
1120 /**< Device redirection table size, the total number of entries. */
1121 uint8_t hash_key_size; /**< Hash key size in bytes */
1122 /** Bit mask of RSS offloads, the bit offset also means flow type */
1123 uint64_t flow_type_rss_offloads;
1124 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1125 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1126 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1127 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1128 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1129 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1130 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1131 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1132 /** Configured number of rx/tx queues */
1133 uint16_t nb_rx_queues; /**< Number of RX queues. */
1134 uint16_t nb_tx_queues; /**< Number of TX queues. */
1135 /** Rx parameter recommendations */
1136 struct rte_eth_dev_portconf default_rxportconf;
1137 /** Tx parameter recommendations */
1138 struct rte_eth_dev_portconf default_txportconf;
1139 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1142 * Switching information for ports on a device with a
1143 * embedded managed interconnect/switch.
1145 struct rte_eth_switch_info switch_info;
1149 * Ethernet device RX queue information structure.
1150 * Used to retieve information about configured queue.
1152 struct rte_eth_rxq_info {
1153 struct rte_mempool *mp; /**< mempool used by that queue. */
1154 struct rte_eth_rxconf conf; /**< queue config parameters. */
1155 uint8_t scattered_rx; /**< scattered packets RX supported. */
1156 uint16_t nb_desc; /**< configured number of RXDs. */
1157 } __rte_cache_min_aligned;
1160 * Ethernet device TX queue information structure.
1161 * Used to retrieve information about configured queue.
1163 struct rte_eth_txq_info {
1164 struct rte_eth_txconf conf; /**< queue config parameters. */
1165 uint16_t nb_desc; /**< configured number of TXDs. */
1166 } __rte_cache_min_aligned;
1168 /** Maximum name length for extended statistics counters */
1169 #define RTE_ETH_XSTATS_NAME_SIZE 64
1172 * An Ethernet device extended statistic structure
1174 * This structure is used by rte_eth_xstats_get() to provide
1175 * statistics that are not provided in the generic *rte_eth_stats*
1177 * It maps a name id, corresponding to an index in the array returned
1178 * by rte_eth_xstats_get_names(), to a statistic value.
1180 struct rte_eth_xstat {
1181 uint64_t id; /**< The index in xstats name array. */
1182 uint64_t value; /**< The statistic counter value. */
1186 * A name element for extended statistics.
1188 * An array of this structure is returned by rte_eth_xstats_get_names().
1189 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1190 * structure references these names by their array index.
1192 struct rte_eth_xstat_name {
1193 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1196 #define ETH_DCB_NUM_TCS 8
1197 #define ETH_MAX_VMDQ_POOL 64
1200 * A structure used to get the information of queue and
1201 * TC mapping on both TX and RX paths.
1203 struct rte_eth_dcb_tc_queue_mapping {
1204 /** rx queues assigned to tc per Pool */
1208 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1209 /** rx queues assigned to tc per Pool */
1213 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1217 * A structure used to get the information of DCB.
1218 * It includes TC UP mapping and queue TC mapping.
1220 struct rte_eth_dcb_info {
1221 uint8_t nb_tcs; /**< number of TCs */
1222 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1223 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1224 /** rx queues assigned to tc */
1225 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1229 * RX/TX queue states
1231 #define RTE_ETH_QUEUE_STATE_STOPPED 0
1232 #define RTE_ETH_QUEUE_STATE_STARTED 1
1234 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1236 /* Macros to check for valid port */
1237 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1238 if (!rte_eth_dev_is_valid_port(port_id)) { \
1239 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1244 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1245 if (!rte_eth_dev_is_valid_port(port_id)) { \
1246 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1252 * l2 tunnel configuration.
1255 /**< l2 tunnel enable mask */
1256 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1257 /**< l2 tunnel insertion mask */
1258 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1259 /**< l2 tunnel stripping mask */
1260 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1261 /**< l2 tunnel forwarding mask */
1262 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1265 * Function type used for RX packet processing packet callbacks.
1267 * The callback function is called on RX with a burst of packets that have
1268 * been received on the given port and queue.
1271 * The Ethernet port on which RX is being performed.
1273 * The queue on the Ethernet port which is being used to receive the packets.
1275 * The burst of packets that have just been received.
1277 * The number of packets in the burst pointed to by "pkts".
1279 * The max number of packets that can be stored in the "pkts" array.
1281 * The arbitrary user parameter passed in by the application when the callback
1282 * was originally configured.
1284 * The number of packets returned to the user.
1286 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1287 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1291 * Function type used for TX packet processing packet callbacks.
1293 * The callback function is called on TX with a burst of packets immediately
1294 * before the packets are put onto the hardware queue for transmission.
1297 * The Ethernet port on which TX is being performed.
1299 * The queue on the Ethernet port which is being used to transmit the packets.
1301 * The burst of packets that are about to be transmitted.
1303 * The number of packets in the burst pointed to by "pkts".
1305 * The arbitrary user parameter passed in by the application when the callback
1306 * was originally configured.
1308 * The number of packets to be written to the NIC.
1310 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1311 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1314 * Possible states of an ethdev port.
1316 enum rte_eth_dev_state {
1317 /** Device is unused before being probed. */
1318 RTE_ETH_DEV_UNUSED = 0,
1319 /** Device is attached when allocated in probing. */
1320 RTE_ETH_DEV_ATTACHED,
1321 /** The deferred state is useless and replaced by ownership. */
1322 RTE_ETH_DEV_DEFERRED,
1323 /** Device is in removed state when plug-out is detected. */
1324 RTE_ETH_DEV_REMOVED,
1327 struct rte_eth_dev_sriov {
1328 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1329 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1330 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1331 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1333 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1335 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1337 #define RTE_ETH_DEV_NO_OWNER 0
1339 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1341 struct rte_eth_dev_owner {
1342 uint64_t id; /**< The owner unique identifier. */
1343 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1347 * Port is released (i.e. totally freed and data erased) on close.
1348 * Temporary flag for PMD migration to new rte_eth_dev_close() behaviour.
1350 #define RTE_ETH_DEV_CLOSE_REMOVE 0x0001
1351 /** Device supports link state interrupt */
1352 #define RTE_ETH_DEV_INTR_LSC 0x0002
1353 /** Device is a bonded slave */
1354 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1355 /** Device supports device removal interrupt */
1356 #define RTE_ETH_DEV_INTR_RMV 0x0008
1357 /** Device is port representor */
1358 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1359 /** Device does not support MAC change after started */
1360 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1363 * Iterates over valid ethdev ports owned by a specific owner.
1366 * The id of the next possible valid owned port.
1368 * The owner identifier.
1369 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1371 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1373 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1374 const uint64_t owner_id);
1377 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1379 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1380 for (p = rte_eth_find_next_owned_by(0, o); \
1381 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1382 p = rte_eth_find_next_owned_by(p + 1, o))
1385 * Iterates over valid ethdev ports.
1388 * The id of the next possible valid port.
1390 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1392 uint16_t rte_eth_find_next(uint16_t port_id);
1395 * Macro to iterate over all enabled and ownerless ethdev ports.
1397 #define RTE_ETH_FOREACH_DEV(p) \
1398 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1403 * @b EXPERIMENTAL: this API may change without prior notice.
1405 * Get a new unique owner identifier.
1406 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1407 * to avoid multiple management of device by different entities.
1410 * Owner identifier pointer.
1412 * Negative errno value on error, 0 on success.
1414 int __rte_experimental rte_eth_dev_owner_new(uint64_t *owner_id);
1418 * @b EXPERIMENTAL: this API may change without prior notice.
1420 * Set an Ethernet device owner.
1423 * The identifier of the port to own.
1425 * The owner pointer.
1427 * Negative errno value on error, 0 on success.
1429 int __rte_experimental rte_eth_dev_owner_set(const uint16_t port_id,
1430 const struct rte_eth_dev_owner *owner);
1434 * @b EXPERIMENTAL: this API may change without prior notice.
1436 * Unset Ethernet device owner to make the device ownerless.
1439 * The identifier of port to make ownerless.
1441 * The owner identifier.
1443 * 0 on success, negative errno value on error.
1445 int __rte_experimental rte_eth_dev_owner_unset(const uint16_t port_id,
1446 const uint64_t owner_id);
1450 * @b EXPERIMENTAL: this API may change without prior notice.
1452 * Remove owner from all Ethernet devices owned by a specific owner.
1455 * The owner identifier.
1457 void __rte_experimental rte_eth_dev_owner_delete(const uint64_t owner_id);
1461 * @b EXPERIMENTAL: this API may change without prior notice.
1463 * Get the owner of an Ethernet device.
1466 * The port identifier.
1468 * The owner structure pointer to fill.
1470 * 0 on success, negative errno value on error..
1472 int __rte_experimental rte_eth_dev_owner_get(const uint16_t port_id,
1473 struct rte_eth_dev_owner *owner);
1476 * Get the total number of Ethernet devices that have been successfully
1477 * initialized by the matching Ethernet driver during the PCI probing phase
1478 * and that are available for applications to use. These devices must be
1479 * accessed by using the ``RTE_ETH_FOREACH_DEV()`` macro to deal with
1480 * non-contiguous ranges of devices.
1481 * These non-contiguous ranges can be created by calls to hotplug functions or
1485 * - The total number of usable Ethernet devices.
1488 uint16_t rte_eth_dev_count(void);
1491 * Get the number of ports which are usable for the application.
1493 * These devices must be iterated by using the macro
1494 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
1495 * to deal with non-contiguous ranges of devices.
1498 * The count of available Ethernet devices.
1500 uint16_t rte_eth_dev_count_avail(void);
1503 * Get the total number of ports which are allocated.
1505 * Some devices may not be available for the application.
1508 * The total count of Ethernet devices.
1510 uint16_t __rte_experimental rte_eth_dev_count_total(void);
1513 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
1514 * the bitmap link_speeds of the struct rte_eth_conf
1517 * Numerical speed value in Mbps
1519 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
1521 * 0 if the speed cannot be mapped
1523 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
1526 * Get DEV_RX_OFFLOAD_* flag name.
1531 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1533 const char *rte_eth_dev_rx_offload_name(uint64_t offload);
1536 * Get DEV_TX_OFFLOAD_* flag name.
1541 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1543 const char *rte_eth_dev_tx_offload_name(uint64_t offload);
1546 * Configure an Ethernet device.
1547 * This function must be invoked first before any other function in the
1548 * Ethernet API. This function can also be re-invoked when a device is in the
1552 * The port identifier of the Ethernet device to configure.
1553 * @param nb_rx_queue
1554 * The number of receive queues to set up for the Ethernet device.
1555 * @param nb_tx_queue
1556 * The number of transmit queues to set up for the Ethernet device.
1558 * The pointer to the configuration data to be used for the Ethernet device.
1559 * The *rte_eth_conf* structure includes:
1560 * - the hardware offload features to activate, with dedicated fields for
1561 * each statically configurable offload hardware feature provided by
1562 * Ethernet devices, such as IP checksum or VLAN tag stripping for
1564 * The Rx offload bitfield API is obsolete and will be deprecated.
1565 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
1566 * structure and use offloads field to set per-port offloads instead.
1567 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
1568 * the [rt]x_offload_capa returned from rte_eth_dev_infos_get().
1569 * Any type of device supported offloading set in the input argument
1570 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
1571 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
1572 * - the Receive Side Scaling (RSS) configuration when using multiple RX
1573 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
1574 * must be within the flow_type_rss_offloads provided by drivers via
1575 * rte_eth_dev_infos_get() API.
1577 * Embedding all configuration information in a single data structure
1578 * is the more flexible method that allows the addition of new features
1579 * without changing the syntax of the API.
1581 * - 0: Success, device configured.
1582 * - <0: Error code returned by the driver configuration function.
1584 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
1585 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
1589 * @b EXPERIMENTAL: this API may change without prior notice.
1591 * Check if an Ethernet device was physically removed.
1594 * The port identifier of the Ethernet device.
1596 * 1 when the Ethernet device is removed, otherwise 0.
1598 int __rte_experimental
1599 rte_eth_dev_is_removed(uint16_t port_id);
1602 * Allocate and set up a receive queue for an Ethernet device.
1604 * The function allocates a contiguous block of memory for *nb_rx_desc*
1605 * receive descriptors from a memory zone associated with *socket_id*
1606 * and initializes each receive descriptor with a network buffer allocated
1607 * from the memory pool *mb_pool*.
1610 * The port identifier of the Ethernet device.
1611 * @param rx_queue_id
1612 * The index of the receive queue to set up.
1613 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1614 * to rte_eth_dev_configure().
1616 * The number of receive descriptors to allocate for the receive ring.
1618 * The *socket_id* argument is the socket identifier in case of NUMA.
1619 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1620 * the DMA memory allocated for the receive descriptors of the ring.
1622 * The pointer to the configuration data to be used for the receive queue.
1623 * NULL value is allowed, in which case default RX configuration
1625 * The *rx_conf* structure contains an *rx_thresh* structure with the values
1626 * of the Prefetch, Host, and Write-Back threshold registers of the receive
1628 * In addition it contains the hardware offloads features to activate using
1629 * the DEV_RX_OFFLOAD_* flags.
1630 * If an offloading set in rx_conf->offloads
1631 * hasn't been set in the input argument eth_conf->rxmode.offloads
1632 * to rte_eth_dev_configure(), it is a new added offloading, it must be
1633 * per-queue type and it is enabled for the queue.
1634 * No need to repeat any bit in rx_conf->offloads which has already been
1635 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
1636 * at port level can't be disabled at queue level.
1638 * The pointer to the memory pool from which to allocate *rte_mbuf* network
1639 * memory buffers to populate each descriptor of the receive ring.
1641 * - 0: Success, receive queue correctly set up.
1642 * - -EIO: if device is removed.
1643 * - -EINVAL: The size of network buffers which can be allocated from the
1644 * memory pool does not fit the various buffer sizes allowed by the
1645 * device controller.
1646 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
1647 * allocate network memory buffers from the memory pool when
1648 * initializing receive descriptors.
1650 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
1651 uint16_t nb_rx_desc, unsigned int socket_id,
1652 const struct rte_eth_rxconf *rx_conf,
1653 struct rte_mempool *mb_pool);
1656 * Allocate and set up a transmit queue for an Ethernet device.
1659 * The port identifier of the Ethernet device.
1660 * @param tx_queue_id
1661 * The index of the transmit queue to set up.
1662 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1663 * to rte_eth_dev_configure().
1665 * The number of transmit descriptors to allocate for the transmit ring.
1667 * The *socket_id* argument is the socket identifier in case of NUMA.
1668 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1669 * the DMA memory allocated for the transmit descriptors of the ring.
1671 * The pointer to the configuration data to be used for the transmit queue.
1672 * NULL value is allowed, in which case default TX configuration
1674 * The *tx_conf* structure contains the following data:
1675 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
1676 * Write-Back threshold registers of the transmit ring.
1677 * When setting Write-Back threshold to the value greater then zero,
1678 * *tx_rs_thresh* value should be explicitly set to one.
1679 * - The *tx_free_thresh* value indicates the [minimum] number of network
1680 * buffers that must be pending in the transmit ring to trigger their
1681 * [implicit] freeing by the driver transmit function.
1682 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
1683 * descriptors that must be pending in the transmit ring before setting the
1684 * RS bit on a descriptor by the driver transmit function.
1685 * The *tx_rs_thresh* value should be less or equal then
1686 * *tx_free_thresh* value, and both of them should be less then
1688 * - The *offloads* member contains Tx offloads to be enabled.
1689 * If an offloading set in tx_conf->offloads
1690 * hasn't been set in the input argument eth_conf->txmode.offloads
1691 * to rte_eth_dev_configure(), it is a new added offloading, it must be
1692 * per-queue type and it is enabled for the queue.
1693 * No need to repeat any bit in tx_conf->offloads which has already been
1694 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
1695 * at port level can't be disabled at queue level.
1697 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
1698 * the transmit function to use default values.
1700 * - 0: Success, the transmit queue is correctly set up.
1701 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
1703 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
1704 uint16_t nb_tx_desc, unsigned int socket_id,
1705 const struct rte_eth_txconf *tx_conf);
1708 * Return the NUMA socket to which an Ethernet device is connected
1711 * The port identifier of the Ethernet device
1713 * The NUMA socket id to which the Ethernet device is connected or
1714 * a default of zero if the socket could not be determined.
1715 * -1 is returned is the port_id value is out of range.
1717 int rte_eth_dev_socket_id(uint16_t port_id);
1720 * Check if port_id of device is attached
1723 * The port identifier of the Ethernet device
1725 * - 0 if port is out of range or not attached
1726 * - 1 if device is attached
1728 int rte_eth_dev_is_valid_port(uint16_t port_id);
1731 * Start specified RX queue of a port. It is used when rx_deferred_start
1732 * flag of the specified queue is true.
1735 * The port identifier of the Ethernet device
1736 * @param rx_queue_id
1737 * The index of the rx queue to update the ring.
1738 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1739 * to rte_eth_dev_configure().
1741 * - 0: Success, the receive queue is started.
1742 * - -EINVAL: The port_id or the queue_id out of range.
1743 * - -EIO: if device is removed.
1744 * - -ENOTSUP: The function not supported in PMD driver.
1746 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
1749 * Stop specified RX queue of a port
1752 * The port identifier of the Ethernet device
1753 * @param rx_queue_id
1754 * The index of the rx queue to update the ring.
1755 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1756 * to rte_eth_dev_configure().
1758 * - 0: Success, the receive queue is stopped.
1759 * - -EINVAL: The port_id or the queue_id out of range.
1760 * - -EIO: if device is removed.
1761 * - -ENOTSUP: The function not supported in PMD driver.
1763 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
1766 * Start TX for specified queue of a port. It is used when tx_deferred_start
1767 * flag of the specified queue is true.
1770 * The port identifier of the Ethernet device
1771 * @param tx_queue_id
1772 * The index of the tx queue to update the ring.
1773 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1774 * to rte_eth_dev_configure().
1776 * - 0: Success, the transmit queue is started.
1777 * - -EINVAL: The port_id or the queue_id out of range.
1778 * - -EIO: if device is removed.
1779 * - -ENOTSUP: The function not supported in PMD driver.
1781 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
1784 * Stop specified TX queue of a port
1787 * The port identifier of the Ethernet device
1788 * @param tx_queue_id
1789 * The index of the tx queue to update the ring.
1790 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1791 * to rte_eth_dev_configure().
1793 * - 0: Success, the transmit queue is stopped.
1794 * - -EINVAL: The port_id or the queue_id out of range.
1795 * - -EIO: if device is removed.
1796 * - -ENOTSUP: The function not supported in PMD driver.
1798 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
1801 * Start an Ethernet device.
1803 * The device start step is the last one and consists of setting the configured
1804 * offload features and in starting the transmit and the receive units of the
1807 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
1808 * PMD port start callback function is invoked.
1810 * On success, all basic functions exported by the Ethernet API (link status,
1811 * receive/transmit, and so on) can be invoked.
1814 * The port identifier of the Ethernet device.
1816 * - 0: Success, Ethernet device started.
1817 * - <0: Error code of the driver device start function.
1819 int rte_eth_dev_start(uint16_t port_id);
1822 * Stop an Ethernet device. The device can be restarted with a call to
1823 * rte_eth_dev_start()
1826 * The port identifier of the Ethernet device.
1828 void rte_eth_dev_stop(uint16_t port_id);
1831 * Link up an Ethernet device.
1833 * Set device link up will re-enable the device rx/tx
1834 * functionality after it is previously set device linked down.
1837 * The port identifier of the Ethernet device.
1839 * - 0: Success, Ethernet device linked up.
1840 * - <0: Error code of the driver device link up function.
1842 int rte_eth_dev_set_link_up(uint16_t port_id);
1845 * Link down an Ethernet device.
1846 * The device rx/tx functionality will be disabled if success,
1847 * and it can be re-enabled with a call to
1848 * rte_eth_dev_set_link_up()
1851 * The port identifier of the Ethernet device.
1853 int rte_eth_dev_set_link_down(uint16_t port_id);
1856 * Close a stopped Ethernet device. The device cannot be restarted!
1857 * The function frees all port resources if the driver supports
1858 * the flag RTE_ETH_DEV_CLOSE_REMOVE.
1861 * The port identifier of the Ethernet device.
1863 void rte_eth_dev_close(uint16_t port_id);
1866 * Reset a Ethernet device and keep its port id.
1868 * When a port has to be reset passively, the DPDK application can invoke
1869 * this function. For example when a PF is reset, all its VFs should also
1870 * be reset. Normally a DPDK application can invoke this function when
1871 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
1872 * a port reset in other circumstances.
1874 * When this function is called, it first stops the port and then calls the
1875 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
1876 * state, in which no Tx and Rx queues are setup, as if the port has been
1877 * reset and not started. The port keeps the port id it had before the
1880 * After calling rte_eth_dev_reset( ), the application should use
1881 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
1882 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
1883 * to reconfigure the device as appropriate.
1885 * Note: To avoid unexpected behavior, the application should stop calling
1886 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
1887 * safety, all these controlling functions should be called from the same
1891 * The port identifier of the Ethernet device.
1894 * - (0) if successful.
1895 * - (-EINVAL) if port identifier is invalid.
1896 * - (-ENOTSUP) if hardware doesn't support this function.
1897 * - (-EPERM) if not ran from the primary process.
1898 * - (-EIO) if re-initialisation failed or device is removed.
1899 * - (-ENOMEM) if the reset failed due to OOM.
1900 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
1902 int rte_eth_dev_reset(uint16_t port_id);
1905 * Enable receipt in promiscuous mode for an Ethernet device.
1908 * The port identifier of the Ethernet device.
1910 void rte_eth_promiscuous_enable(uint16_t port_id);
1913 * Disable receipt in promiscuous mode for an Ethernet device.
1916 * The port identifier of the Ethernet device.
1918 void rte_eth_promiscuous_disable(uint16_t port_id);
1921 * Return the value of promiscuous mode for an Ethernet device.
1924 * The port identifier of the Ethernet device.
1926 * - (1) if promiscuous is enabled
1927 * - (0) if promiscuous is disabled.
1930 int rte_eth_promiscuous_get(uint16_t port_id);
1933 * Enable the receipt of any multicast frame by an Ethernet device.
1936 * The port identifier of the Ethernet device.
1938 void rte_eth_allmulticast_enable(uint16_t port_id);
1941 * Disable the receipt of all multicast frames by an Ethernet device.
1944 * The port identifier of the Ethernet device.
1946 void rte_eth_allmulticast_disable(uint16_t port_id);
1949 * Return the value of allmulticast mode for an Ethernet device.
1952 * The port identifier of the Ethernet device.
1954 * - (1) if allmulticast is enabled
1955 * - (0) if allmulticast is disabled.
1958 int rte_eth_allmulticast_get(uint16_t port_id);
1961 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1962 * or FULL-DUPLEX) of the physical link of an Ethernet device. It might need
1963 * to wait up to 9 seconds in it.
1966 * The port identifier of the Ethernet device.
1968 * A pointer to an *rte_eth_link* structure to be filled with
1969 * the status, the speed and the mode of the Ethernet device link.
1971 void rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
1974 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1975 * or FULL-DUPLEX) of the physical link of an Ethernet device. It is a no-wait
1976 * version of rte_eth_link_get().
1979 * The port identifier of the Ethernet device.
1981 * A pointer to an *rte_eth_link* structure to be filled with
1982 * the status, the speed and the mode of the Ethernet device link.
1984 void rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
1987 * Retrieve the general I/O statistics of an Ethernet device.
1990 * The port identifier of the Ethernet device.
1992 * A pointer to a structure of type *rte_eth_stats* to be filled with
1993 * the values of device counters for the following set of statistics:
1994 * - *ipackets* with the total of successfully received packets.
1995 * - *opackets* with the total of successfully transmitted packets.
1996 * - *ibytes* with the total of successfully received bytes.
1997 * - *obytes* with the total of successfully transmitted bytes.
1998 * - *ierrors* with the total of erroneous received packets.
1999 * - *oerrors* with the total of failed transmitted packets.
2001 * Zero if successful. Non-zero otherwise.
2003 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
2006 * Reset the general I/O statistics of an Ethernet device.
2009 * The port identifier of the Ethernet device.
2011 * - (0) if device notified to reset stats.
2012 * - (-ENOTSUP) if hardware doesn't support.
2013 * - (-ENODEV) if *port_id* invalid.
2015 int rte_eth_stats_reset(uint16_t port_id);
2018 * Retrieve names of extended statistics of an Ethernet device.
2020 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2022 * xstats_names[i].name => xstats[i].value
2024 * And the array index is same with id field of 'struct rte_eth_xstat':
2027 * This assumption makes key-value pair matching less flexible but simpler.
2030 * The port identifier of the Ethernet device.
2031 * @param xstats_names
2032 * An rte_eth_xstat_name array of at least *size* elements to
2033 * be filled. If set to NULL, the function returns the required number
2036 * The size of the xstats_names array (number of elements).
2038 * - A positive value lower or equal to size: success. The return value
2039 * is the number of entries filled in the stats table.
2040 * - A positive value higher than size: error, the given statistics table
2041 * is too small. The return value corresponds to the size that should
2042 * be given to succeed. The entries in the table are not valid and
2043 * shall not be used by the caller.
2044 * - A negative value on error (invalid port id).
2046 int rte_eth_xstats_get_names(uint16_t port_id,
2047 struct rte_eth_xstat_name *xstats_names,
2051 * Retrieve extended statistics of an Ethernet device.
2053 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2055 * xstats_names[i].name => xstats[i].value
2057 * And the array index is same with id field of 'struct rte_eth_xstat':
2060 * This assumption makes key-value pair matching less flexible but simpler.
2063 * The port identifier of the Ethernet device.
2065 * A pointer to a table of structure of type *rte_eth_xstat*
2066 * to be filled with device statistics ids and values.
2067 * This parameter can be set to NULL if n is 0.
2069 * The size of the xstats array (number of elements).
2071 * - A positive value lower or equal to n: success. The return value
2072 * is the number of entries filled in the stats table.
2073 * - A positive value higher than n: error, the given statistics table
2074 * is too small. The return value corresponds to the size that should
2075 * be given to succeed. The entries in the table are not valid and
2076 * shall not be used by the caller.
2077 * - A negative value on error (invalid port id).
2079 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2083 * Retrieve names of extended statistics of an Ethernet device.
2086 * The port identifier of the Ethernet device.
2087 * @param xstats_names
2088 * An rte_eth_xstat_name array of at least *size* elements to
2089 * be filled. If set to NULL, the function returns the required number
2092 * IDs array given by app to retrieve specific statistics
2094 * The size of the xstats_names array (number of elements).
2096 * - A positive value lower or equal to size: success. The return value
2097 * is the number of entries filled in the stats table.
2098 * - A positive value higher than size: error, the given statistics table
2099 * is too small. The return value corresponds to the size that should
2100 * be given to succeed. The entries in the table are not valid and
2101 * shall not be used by the caller.
2102 * - A negative value on error (invalid port id).
2105 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2106 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2110 * Retrieve extended statistics of an Ethernet device.
2113 * The port identifier of the Ethernet device.
2115 * A pointer to an ids array passed by application. This tells which
2116 * statistics values function should retrieve. This parameter
2117 * can be set to NULL if size is 0. In this case function will retrieve
2118 * all avalible statistics.
2120 * A pointer to a table to be filled with device statistics values.
2122 * The size of the ids array (number of elements).
2124 * - A positive value lower or equal to size: success. The return value
2125 * is the number of entries filled in the stats table.
2126 * - A positive value higher than size: error, the given statistics table
2127 * is too small. The return value corresponds to the size that should
2128 * be given to succeed. The entries in the table are not valid and
2129 * shall not be used by the caller.
2130 * - A negative value on error (invalid port id).
2132 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2133 uint64_t *values, unsigned int size);
2136 * Gets the ID of a statistic from its name.
2138 * This function searches for the statistics using string compares, and
2139 * as such should not be used on the fast-path. For fast-path retrieval of
2140 * specific statistics, store the ID as provided in *id* from this function,
2141 * and pass the ID to rte_eth_xstats_get()
2143 * @param port_id The port to look up statistics from
2144 * @param xstat_name The name of the statistic to return
2145 * @param[out] id A pointer to an app-supplied uint64_t which should be
2146 * set to the ID of the stat if the stat exists.
2149 * -ENODEV for invalid port_id,
2150 * -EIO if device is removed,
2151 * -EINVAL if the xstat_name doesn't exist in port_id
2153 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2157 * Reset extended statistics of an Ethernet device.
2160 * The port identifier of the Ethernet device.
2162 void rte_eth_xstats_reset(uint16_t port_id);
2165 * Set a mapping for the specified transmit queue to the specified per-queue
2166 * statistics counter.
2169 * The port identifier of the Ethernet device.
2170 * @param tx_queue_id
2171 * The index of the transmit queue for which a queue stats mapping is required.
2172 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2173 * to rte_eth_dev_configure().
2175 * The per-queue packet statistics functionality number that the transmit
2176 * queue is to be assigned.
2177 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2179 * Zero if successful. Non-zero otherwise.
2181 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2182 uint16_t tx_queue_id, uint8_t stat_idx);
2185 * Set a mapping for the specified receive queue to the specified per-queue
2186 * statistics counter.
2189 * The port identifier of the Ethernet device.
2190 * @param rx_queue_id
2191 * The index of the receive queue for which a queue stats mapping is required.
2192 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2193 * to rte_eth_dev_configure().
2195 * The per-queue packet statistics functionality number that the receive
2196 * queue is to be assigned.
2197 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2199 * Zero if successful. Non-zero otherwise.
2201 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2202 uint16_t rx_queue_id,
2206 * Retrieve the Ethernet address of an Ethernet device.
2209 * The port identifier of the Ethernet device.
2211 * A pointer to a structure of type *ether_addr* to be filled with
2212 * the Ethernet address of the Ethernet device.
2214 void rte_eth_macaddr_get(uint16_t port_id, struct ether_addr *mac_addr);
2217 * Retrieve the contextual information of an Ethernet device.
2220 * The port identifier of the Ethernet device.
2222 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
2223 * the contextual information of the Ethernet device.
2225 void rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
2228 * Retrieve the firmware version of a device.
2231 * The port identifier of the device.
2233 * A pointer to a string array storing the firmware version of a device,
2234 * the string includes terminating null. This pointer is allocated by caller.
2236 * The size of the string array pointed by fw_version, which should be
2237 * large enough to store firmware version of the device.
2239 * - (0) if successful.
2240 * - (-ENOTSUP) if operation is not supported.
2241 * - (-ENODEV) if *port_id* invalid.
2242 * - (-EIO) if device is removed.
2243 * - (>0) if *fw_size* is not enough to store firmware version, return
2244 * the size of the non truncated string.
2246 int rte_eth_dev_fw_version_get(uint16_t port_id,
2247 char *fw_version, size_t fw_size);
2250 * Retrieve the supported packet types of an Ethernet device.
2252 * When a packet type is announced as supported, it *must* be recognized by
2253 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
2254 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
2255 * packet types for these packets:
2256 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
2257 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
2258 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
2259 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
2261 * When a packet is received by a PMD, the most precise type must be
2262 * returned among the ones supported. However a PMD is allowed to set
2263 * packet type that is not in the supported list, at the condition that it
2264 * is more precise. Therefore, a PMD announcing no supported packet types
2265 * can still set a matching packet type in a received packet.
2268 * Better to invoke this API after the device is already started or rx burst
2269 * function is decided, to obtain correct supported ptypes.
2271 * if a given PMD does not report what ptypes it supports, then the supported
2272 * ptype count is reported as 0.
2274 * The port identifier of the Ethernet device.
2276 * A hint of what kind of packet type which the caller is interested in.
2278 * An array pointer to store adequate packet types, allocated by caller.
2280 * Size of the array pointed by param ptypes.
2282 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
2283 * only num entries will be filled into the ptypes array, but the full
2284 * count of supported ptypes will be returned.
2285 * - (-ENODEV) if *port_id* invalid.
2287 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
2288 uint32_t *ptypes, int num);
2291 * Retrieve the MTU of an Ethernet device.
2294 * The port identifier of the Ethernet device.
2296 * A pointer to a uint16_t where the retrieved MTU is to be stored.
2298 * - (0) if successful.
2299 * - (-ENODEV) if *port_id* invalid.
2301 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
2304 * Change the MTU of an Ethernet device.
2307 * The port identifier of the Ethernet device.
2309 * A uint16_t for the MTU to be applied.
2311 * - (0) if successful.
2312 * - (-ENOTSUP) if operation is not supported.
2313 * - (-ENODEV) if *port_id* invalid.
2314 * - (-EIO) if device is removed.
2315 * - (-EINVAL) if *mtu* invalid.
2316 * - (-EBUSY) if operation is not allowed when the port is running
2318 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
2321 * Enable/Disable hardware filtering by an Ethernet device of received
2322 * VLAN packets tagged with a given VLAN Tag Identifier.
2325 * The port identifier of the Ethernet device.
2327 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
2329 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
2330 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
2332 * - (0) if successful.
2333 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2334 * - (-ENODEV) if *port_id* invalid.
2335 * - (-EIO) if device is removed.
2336 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
2337 * - (-EINVAL) if *vlan_id* > 4095.
2339 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
2342 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
2343 * 82599/X540/X550 can support VLAN stripping at the rx queue level
2346 * The port identifier of the Ethernet device.
2347 * @param rx_queue_id
2348 * The index of the receive queue for which a queue stats mapping is required.
2349 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2350 * to rte_eth_dev_configure().
2352 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
2353 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
2355 * - (0) if successful.
2356 * - (-ENOSUP) if hardware-assisted VLAN stripping not configured.
2357 * - (-ENODEV) if *port_id* invalid.
2358 * - (-EINVAL) if *rx_queue_id* invalid.
2360 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
2364 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
2365 * the VLAN Header. This is a register setup available on some Intel NIC, not
2366 * but all, please check the data sheet for availability.
2369 * The port identifier of the Ethernet device.
2373 * The Tag Protocol ID
2375 * - (0) if successful.
2376 * - (-ENOSUP) if hardware-assisted VLAN TPID setup is not supported.
2377 * - (-ENODEV) if *port_id* invalid.
2378 * - (-EIO) if device is removed.
2380 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
2381 enum rte_vlan_type vlan_type,
2385 * Set VLAN offload configuration on an Ethernet device
2386 * Enable/Disable Extended VLAN by an Ethernet device, This is a register setup
2387 * available on some Intel NIC, not but all, please check the data sheet for
2389 * Enable/Disable VLAN Strip can be done on rx queue for certain NIC, but here
2390 * the configuration is applied on the port level.
2393 * The port identifier of the Ethernet device.
2394 * @param offload_mask
2395 * The VLAN Offload bit mask can be mixed use with "OR"
2396 * ETH_VLAN_STRIP_OFFLOAD
2397 * ETH_VLAN_FILTER_OFFLOAD
2398 * ETH_VLAN_EXTEND_OFFLOAD
2400 * - (0) if successful.
2401 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2402 * - (-ENODEV) if *port_id* invalid.
2403 * - (-EIO) if device is removed.
2405 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
2408 * Read VLAN Offload configuration from an Ethernet device
2411 * The port identifier of the Ethernet device.
2413 * - (>0) if successful. Bit mask to indicate
2414 * ETH_VLAN_STRIP_OFFLOAD
2415 * ETH_VLAN_FILTER_OFFLOAD
2416 * ETH_VLAN_EXTEND_OFFLOAD
2417 * - (-ENODEV) if *port_id* invalid.
2419 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
2422 * Set port based TX VLAN insertion on or off.
2425 * The port identifier of the Ethernet device.
2427 * Port based TX VLAN identifier together with user priority.
2429 * Turn on or off the port based TX VLAN insertion.
2432 * - (0) if successful.
2433 * - negative if failed.
2435 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
2437 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
2441 * Structure used to buffer packets for future TX
2442 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
2444 struct rte_eth_dev_tx_buffer {
2445 buffer_tx_error_fn error_callback;
2446 void *error_userdata;
2447 uint16_t size; /**< Size of buffer for buffered tx */
2448 uint16_t length; /**< Number of packets in the array */
2449 struct rte_mbuf *pkts[];
2450 /**< Pending packets to be sent on explicit flush or when full */
2454 * Calculate the size of the tx buffer.
2457 * Number of stored packets.
2459 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
2460 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
2463 * Initialize default values for buffered transmitting
2466 * Tx buffer to be initialized.
2473 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
2476 * Configure a callback for buffered packets which cannot be sent
2478 * Register a specific callback to be called when an attempt is made to send
2479 * all packets buffered on an ethernet port, but not all packets can
2480 * successfully be sent. The callback registered here will be called only
2481 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
2482 * The default callback configured for each queue by default just frees the
2483 * packets back to the calling mempool. If additional behaviour is required,
2484 * for example, to count dropped packets, or to retry transmission of packets
2485 * which cannot be sent, this function should be used to register a suitable
2486 * callback function to implement the desired behaviour.
2487 * The example callback "rte_eth_count_unsent_packet_callback()" is also
2488 * provided as reference.
2491 * The port identifier of the Ethernet device.
2493 * The function to be used as the callback.
2495 * Arbitrary parameter to be passed to the callback function
2497 * 0 on success, or -1 on error with rte_errno set appropriately
2500 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
2501 buffer_tx_error_fn callback, void *userdata);
2504 * Callback function for silently dropping unsent buffered packets.
2506 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2507 * adjust the default behavior when buffered packets cannot be sent. This
2508 * function drops any unsent packets silently and is used by tx buffered
2509 * operations as default behavior.
2511 * NOTE: this function should not be called directly, instead it should be used
2512 * as a callback for packet buffering.
2514 * NOTE: when configuring this function as a callback with
2515 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2516 * should point to an uint64_t value.
2519 * The previously buffered packets which could not be sent
2521 * The number of unsent packets in the pkts array
2526 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
2530 * Callback function for tracking unsent buffered packets.
2532 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2533 * adjust the default behavior when buffered packets cannot be sent. This
2534 * function drops any unsent packets, but also updates a user-supplied counter
2535 * to track the overall number of packets dropped. The counter should be an
2536 * uint64_t variable.
2538 * NOTE: this function should not be called directly, instead it should be used
2539 * as a callback for packet buffering.
2541 * NOTE: when configuring this function as a callback with
2542 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2543 * should point to an uint64_t value.
2546 * The previously buffered packets which could not be sent
2548 * The number of unsent packets in the pkts array
2550 * Pointer to an uint64_t value, which will be incremented by unsent
2553 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
2557 * Request the driver to free mbufs currently cached by the driver. The
2558 * driver will only free the mbuf if it is no longer in use. It is the
2559 * application's responsibity to ensure rte_eth_tx_buffer_flush(..) is
2563 * The port identifier of the Ethernet device.
2565 * The index of the transmit queue through which output packets must be
2567 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2568 * to rte_eth_dev_configure().
2570 * Maximum number of packets to free. Use 0 to indicate all possible packets
2571 * should be freed. Note that a packet may be using multiple mbufs.
2574 * -ENODEV: Invalid interface
2575 * -EIO: device is removed
2576 * -ENOTSUP: Driver does not support function
2578 * 0-n: Number of packets freed. More packets may still remain in ring that
2582 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
2585 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
2588 enum rte_eth_event_ipsec_subtype {
2589 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
2590 /**< Unknown event type */
2591 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
2592 /**< Sequence number overflow */
2593 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
2594 /**< Soft time expiry of SA */
2595 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
2596 /**< Soft byte expiry of SA */
2597 RTE_ETH_EVENT_IPSEC_MAX
2598 /**< Max value of this enum */
2602 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
2603 * information of the IPsec offload event.
2605 struct rte_eth_event_ipsec_desc {
2606 enum rte_eth_event_ipsec_subtype subtype;
2607 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
2609 /**< Event specific metadata
2611 * For the following events, *userdata* registered
2612 * with the *rte_security_session* would be returned
2615 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
2616 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
2617 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
2619 * @see struct rte_security_session_conf
2625 * The eth device event type for interrupt, and maybe others in the future.
2627 enum rte_eth_event_type {
2628 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
2629 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
2630 RTE_ETH_EVENT_QUEUE_STATE,
2631 /**< queue state event (enabled/disabled) */
2632 RTE_ETH_EVENT_INTR_RESET,
2633 /**< reset interrupt event, sent to VF on PF reset */
2634 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
2635 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
2636 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
2637 RTE_ETH_EVENT_NEW, /**< port is probed */
2638 RTE_ETH_EVENT_DESTROY, /**< port is released */
2639 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
2640 RTE_ETH_EVENT_MAX /**< max value of this enum */
2643 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
2644 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
2645 /**< user application callback to be registered for interrupts */
2648 * Register a callback function for port event.
2652 * RTE_ETH_ALL means register the event for all port ids.
2656 * User supplied callback function to be called.
2658 * Pointer to the parameters for the registered callback.
2661 * - On success, zero.
2662 * - On failure, a negative value.
2664 int rte_eth_dev_callback_register(uint16_t port_id,
2665 enum rte_eth_event_type event,
2666 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2669 * Unregister a callback function for port event.
2673 * RTE_ETH_ALL means unregister the event for all port ids.
2677 * User supplied callback function to be called.
2679 * Pointer to the parameters for the registered callback. -1 means to
2680 * remove all for the same callback address and same event.
2683 * - On success, zero.
2684 * - On failure, a negative value.
2686 int rte_eth_dev_callback_unregister(uint16_t port_id,
2687 enum rte_eth_event_type event,
2688 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2691 * When there is no rx packet coming in Rx Queue for a long time, we can
2692 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
2693 * to be triggered when Rx packet arrives.
2695 * The rte_eth_dev_rx_intr_enable() function enables rx queue
2696 * interrupt on specific rx queue of a port.
2699 * The port identifier of the Ethernet device.
2701 * The index of the receive queue from which to retrieve input packets.
2702 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2703 * to rte_eth_dev_configure().
2705 * - (0) if successful.
2706 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2708 * - (-ENODEV) if *port_id* invalid.
2709 * - (-EIO) if device is removed.
2711 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
2714 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
2715 * interrupt and returns to polling mode.
2717 * The rte_eth_dev_rx_intr_disable() function disables rx queue
2718 * interrupt on specific rx queue of a port.
2721 * The port identifier of the Ethernet device.
2723 * The index of the receive queue from which to retrieve input packets.
2724 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2725 * to rte_eth_dev_configure().
2727 * - (0) if successful.
2728 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2730 * - (-ENODEV) if *port_id* invalid.
2731 * - (-EIO) if device is removed.
2733 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
2736 * RX Interrupt control per port.
2739 * The port identifier of the Ethernet device.
2741 * Epoll instance fd which the intr vector associated to.
2742 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2744 * The operation be performed for the vector.
2745 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2749 * - On success, zero.
2750 * - On failure, a negative value.
2752 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
2755 * RX Interrupt control per queue.
2758 * The port identifier of the Ethernet device.
2760 * The index of the receive queue from which to retrieve input packets.
2761 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2762 * to rte_eth_dev_configure().
2764 * Epoll instance fd which the intr vector associated to.
2765 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2767 * The operation be performed for the vector.
2768 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2772 * - On success, zero.
2773 * - On failure, a negative value.
2775 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
2776 int epfd, int op, void *data);
2780 * @b EXPERIMENTAL: this API may change without prior notice.
2782 * Get interrupt fd per Rx queue.
2785 * The port identifier of the Ethernet device.
2787 * The index of the receive queue from which to retrieve input packets.
2788 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2789 * to rte_eth_dev_configure().
2791 * - (>=0) the interrupt fd associated to the requested Rx queue if
2795 int __rte_experimental
2796 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
2799 * Turn on the LED on the Ethernet device.
2800 * This function turns on the LED on the Ethernet device.
2803 * The port identifier of the Ethernet device.
2805 * - (0) if successful.
2806 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2808 * - (-ENODEV) if *port_id* invalid.
2809 * - (-EIO) if device is removed.
2811 int rte_eth_led_on(uint16_t port_id);
2814 * Turn off the LED on the Ethernet device.
2815 * This function turns off the LED on the Ethernet device.
2818 * The port identifier of the Ethernet device.
2820 * - (0) if successful.
2821 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2823 * - (-ENODEV) if *port_id* invalid.
2824 * - (-EIO) if device is removed.
2826 int rte_eth_led_off(uint16_t port_id);
2829 * Get current status of the Ethernet link flow control for Ethernet device
2832 * The port identifier of the Ethernet device.
2834 * The pointer to the structure where to store the flow control parameters.
2836 * - (0) if successful.
2837 * - (-ENOTSUP) if hardware doesn't support flow control.
2838 * - (-ENODEV) if *port_id* invalid.
2839 * - (-EIO) if device is removed.
2841 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
2842 struct rte_eth_fc_conf *fc_conf);
2845 * Configure the Ethernet link flow control for Ethernet device
2848 * The port identifier of the Ethernet device.
2850 * The pointer to the structure of the flow control parameters.
2852 * - (0) if successful.
2853 * - (-ENOTSUP) if hardware doesn't support flow control mode.
2854 * - (-ENODEV) if *port_id* invalid.
2855 * - (-EINVAL) if bad parameter
2856 * - (-EIO) if flow control setup failure or device is removed.
2858 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
2859 struct rte_eth_fc_conf *fc_conf);
2862 * Configure the Ethernet priority flow control under DCB environment
2863 * for Ethernet device.
2866 * The port identifier of the Ethernet device.
2868 * The pointer to the structure of the priority flow control parameters.
2870 * - (0) if successful.
2871 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
2872 * - (-ENODEV) if *port_id* invalid.
2873 * - (-EINVAL) if bad parameter
2874 * - (-EIO) if flow control setup failure or device is removed.
2876 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
2877 struct rte_eth_pfc_conf *pfc_conf);
2880 * Add a MAC address to an internal array of addresses used to enable whitelist
2881 * filtering to accept packets only if the destination MAC address matches.
2884 * The port identifier of the Ethernet device.
2886 * The MAC address to add.
2888 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
2889 * not enabled, this should be set to 0.
2891 * - (0) if successfully added or *mac_addr* was already added.
2892 * - (-ENOTSUP) if hardware doesn't support this feature.
2893 * - (-ENODEV) if *port* is invalid.
2894 * - (-EIO) if device is removed.
2895 * - (-ENOSPC) if no more MAC addresses can be added.
2896 * - (-EINVAL) if MAC address is invalid.
2898 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct ether_addr *mac_addr,
2902 * Remove a MAC address from the internal array of addresses.
2905 * The port identifier of the Ethernet device.
2907 * MAC address to remove.
2909 * - (0) if successful, or *mac_addr* didn't exist.
2910 * - (-ENOTSUP) if hardware doesn't support.
2911 * - (-ENODEV) if *port* invalid.
2912 * - (-EADDRINUSE) if attempting to remove the default MAC address
2914 int rte_eth_dev_mac_addr_remove(uint16_t port_id, struct ether_addr *mac_addr);
2917 * Set the default MAC address.
2920 * The port identifier of the Ethernet device.
2922 * New default MAC address.
2924 * - (0) if successful, or *mac_addr* didn't exist.
2925 * - (-ENOTSUP) if hardware doesn't support.
2926 * - (-ENODEV) if *port* invalid.
2927 * - (-EINVAL) if MAC address is invalid.
2929 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
2930 struct ether_addr *mac_addr);
2933 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2936 * The port identifier of the Ethernet device.
2940 * Redirection table size. The table size can be queried by
2941 * rte_eth_dev_info_get().
2943 * - (0) if successful.
2944 * - (-ENOTSUP) if hardware doesn't support.
2945 * - (-EINVAL) if bad parameter.
2946 * - (-EIO) if device is removed.
2948 int rte_eth_dev_rss_reta_update(uint16_t port_id,
2949 struct rte_eth_rss_reta_entry64 *reta_conf,
2950 uint16_t reta_size);
2953 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2956 * The port identifier of the Ethernet device.
2960 * Redirection table size. The table size can be queried by
2961 * rte_eth_dev_info_get().
2963 * - (0) if successful.
2964 * - (-ENOTSUP) if hardware doesn't support.
2965 * - (-EINVAL) if bad parameter.
2966 * - (-EIO) if device is removed.
2968 int rte_eth_dev_rss_reta_query(uint16_t port_id,
2969 struct rte_eth_rss_reta_entry64 *reta_conf,
2970 uint16_t reta_size);
2973 * Updates unicast hash table for receiving packet with the given destination
2974 * MAC address, and the packet is routed to all VFs for which the RX mode is
2975 * accept packets that match the unicast hash table.
2978 * The port identifier of the Ethernet device.
2980 * Unicast MAC address.
2982 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
2983 * 0 - Clear an unicast hash bit.
2985 * - (0) if successful.
2986 * - (-ENOTSUP) if hardware doesn't support.
2987 * - (-ENODEV) if *port_id* invalid.
2988 * - (-EIO) if device is removed.
2989 * - (-EINVAL) if bad parameter.
2991 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct ether_addr *addr,
2995 * Updates all unicast hash bitmaps for receiving packet with any Unicast
2996 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
2997 * mode is accept packets that match the unicast hash table.
3000 * The port identifier of the Ethernet device.
3002 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
3004 * 0 - Clear all unicast hash bitmaps
3006 * - (0) if successful.
3007 * - (-ENOTSUP) if hardware doesn't support.
3008 * - (-ENODEV) if *port_id* invalid.
3009 * - (-EIO) if device is removed.
3010 * - (-EINVAL) if bad parameter.
3012 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
3015 * Set a traffic mirroring rule on an Ethernet device
3018 * The port identifier of the Ethernet device.
3019 * @param mirror_conf
3020 * The pointer to the traffic mirroring structure describing the mirroring rule.
3021 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
3022 * destination pool and the value of rule if enable vlan or pool mirroring.
3025 * The index of traffic mirroring rule, we support four separated rules.
3027 * 1 - Enable a mirroring rule.
3028 * 0 - Disable a mirroring rule.
3030 * - (0) if successful.
3031 * - (-ENOTSUP) if hardware doesn't support this feature.
3032 * - (-ENODEV) if *port_id* invalid.
3033 * - (-EIO) if device is removed.
3034 * - (-EINVAL) if the mr_conf information is not correct.
3036 int rte_eth_mirror_rule_set(uint16_t port_id,
3037 struct rte_eth_mirror_conf *mirror_conf,
3042 * Reset a traffic mirroring rule on an Ethernet device.
3045 * The port identifier of the Ethernet device.
3047 * The index of traffic mirroring rule, we support four separated rules.
3049 * - (0) if successful.
3050 * - (-ENOTSUP) if hardware doesn't support this feature.
3051 * - (-ENODEV) if *port_id* invalid.
3052 * - (-EIO) if device is removed.
3053 * - (-EINVAL) if bad parameter.
3055 int rte_eth_mirror_rule_reset(uint16_t port_id,
3059 * Set the rate limitation for a queue on an Ethernet device.
3062 * The port identifier of the Ethernet device.
3066 * The tx rate in Mbps. Allocated from the total port link speed.
3068 * - (0) if successful.
3069 * - (-ENOTSUP) if hardware doesn't support this feature.
3070 * - (-ENODEV) if *port_id* invalid.
3071 * - (-EIO) if device is removed.
3072 * - (-EINVAL) if bad parameter.
3074 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
3078 * Configuration of Receive Side Scaling hash computation of Ethernet device.
3081 * The port identifier of the Ethernet device.
3083 * The new configuration to use for RSS hash computation on the port.
3085 * - (0) if successful.
3086 * - (-ENODEV) if port identifier is invalid.
3087 * - (-EIO) if device is removed.
3088 * - (-ENOTSUP) if hardware doesn't support.
3089 * - (-EINVAL) if bad parameter.
3091 int rte_eth_dev_rss_hash_update(uint16_t port_id,
3092 struct rte_eth_rss_conf *rss_conf);
3095 * Retrieve current configuration of Receive Side Scaling hash computation
3096 * of Ethernet device.
3099 * The port identifier of the Ethernet device.
3101 * Where to store the current RSS hash configuration of the Ethernet device.
3103 * - (0) if successful.
3104 * - (-ENODEV) if port identifier is invalid.
3105 * - (-EIO) if device is removed.
3106 * - (-ENOTSUP) if hardware doesn't support RSS.
3109 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
3110 struct rte_eth_rss_conf *rss_conf);
3113 * Add UDP tunneling port for a specific type of tunnel.
3114 * The packets with this UDP port will be identified as this type of tunnel.
3115 * Before enabling any offloading function for a tunnel, users can call this API
3116 * to change or add more UDP port for the tunnel. So the offloading function
3117 * can take effect on the packets with the specific UDP port.
3120 * The port identifier of the Ethernet device.
3122 * UDP tunneling configuration.
3125 * - (0) if successful.
3126 * - (-ENODEV) if port identifier is invalid.
3127 * - (-EIO) if device is removed.
3128 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3131 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
3132 struct rte_eth_udp_tunnel *tunnel_udp);
3135 * Delete UDP tunneling port a specific type of tunnel.
3136 * The packets with this UDP port will not be identified as this type of tunnel
3138 * Before enabling any offloading function for a tunnel, users can call this API
3139 * to delete a UDP port for the tunnel. So the offloading function will not take
3140 * effect on the packets with the specific UDP port.
3143 * The port identifier of the Ethernet device.
3145 * UDP tunneling configuration.
3148 * - (0) if successful.
3149 * - (-ENODEV) if port identifier is invalid.
3150 * - (-EIO) if device is removed.
3151 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3154 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
3155 struct rte_eth_udp_tunnel *tunnel_udp);
3158 * Check whether the filter type is supported on an Ethernet device.
3159 * All the supported filter types are defined in 'rte_eth_ctrl.h'.
3162 * The port identifier of the Ethernet device.
3163 * @param filter_type
3166 * - (0) if successful.
3167 * - (-ENOTSUP) if hardware doesn't support this filter type.
3168 * - (-ENODEV) if *port_id* invalid.
3169 * - (-EIO) if device is removed.
3171 int rte_eth_dev_filter_supported(uint16_t port_id,
3172 enum rte_filter_type filter_type);
3175 * Take operations to assigned filter type on an Ethernet device.
3176 * All the supported operations and filter types are defined in 'rte_eth_ctrl.h'.
3179 * The port identifier of the Ethernet device.
3180 * @param filter_type
3183 * Type of operation.
3185 * A pointer to arguments defined specifically for the operation.
3187 * - (0) if successful.
3188 * - (-ENOTSUP) if hardware doesn't support.
3189 * - (-ENODEV) if *port_id* invalid.
3190 * - (-EIO) if device is removed.
3191 * - others depends on the specific operations implementation.
3193 int rte_eth_dev_filter_ctrl(uint16_t port_id, enum rte_filter_type filter_type,
3194 enum rte_filter_op filter_op, void *arg);
3197 * Get DCB information on an Ethernet device.
3200 * The port identifier of the Ethernet device.
3204 * - (0) if successful.
3205 * - (-ENODEV) if port identifier is invalid.
3206 * - (-EIO) if device is removed.
3207 * - (-ENOTSUP) if hardware doesn't support.
3209 int rte_eth_dev_get_dcb_info(uint16_t port_id,
3210 struct rte_eth_dcb_info *dcb_info);
3212 struct rte_eth_rxtx_callback;
3215 * Add a callback to be called on packet RX on a given port and queue.
3217 * This API configures a function to be called for each burst of
3218 * packets received on a given NIC port queue. The return value is a pointer
3219 * that can be used to later remove the callback using
3220 * rte_eth_remove_rx_callback().
3222 * Multiple functions are called in the order that they are added.
3225 * The port identifier of the Ethernet device.
3227 * The queue on the Ethernet device on which the callback is to be added.
3229 * The callback function
3231 * A generic pointer parameter which will be passed to each invocation of the
3232 * callback function on this port and queue.
3236 * On success, a pointer value which can later be used to remove the callback.
3238 const struct rte_eth_rxtx_callback *
3239 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
3240 rte_rx_callback_fn fn, void *user_param);
3243 * Add a callback that must be called first on packet RX on a given port
3246 * This API configures a first function to be called for each burst of
3247 * packets received on a given NIC port queue. The return value is a pointer
3248 * that can be used to later remove the callback using
3249 * rte_eth_remove_rx_callback().
3251 * Multiple functions are called in the order that they are added.
3254 * The port identifier of the Ethernet device.
3256 * The queue on the Ethernet device on which the callback is to be added.
3258 * The callback function
3260 * A generic pointer parameter which will be passed to each invocation of the
3261 * callback function on this port and queue.
3265 * On success, a pointer value which can later be used to remove the callback.
3267 const struct rte_eth_rxtx_callback *
3268 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
3269 rte_rx_callback_fn fn, void *user_param);
3272 * Add a callback to be called on packet TX on a given port and queue.
3274 * This API configures a function to be called for each burst of
3275 * packets sent on a given NIC port queue. The return value is a pointer
3276 * that can be used to later remove the callback using
3277 * rte_eth_remove_tx_callback().
3279 * Multiple functions are called in the order that they are added.
3282 * The port identifier of the Ethernet device.
3284 * The queue on the Ethernet device on which the callback is to be added.
3286 * The callback function
3288 * A generic pointer parameter which will be passed to each invocation of the
3289 * callback function on this port and queue.
3293 * On success, a pointer value which can later be used to remove the callback.
3295 const struct rte_eth_rxtx_callback *
3296 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
3297 rte_tx_callback_fn fn, void *user_param);
3300 * Remove an RX packet callback from a given port and queue.
3302 * This function is used to removed callbacks that were added to a NIC port
3303 * queue using rte_eth_add_rx_callback().
3305 * Note: the callback is removed from the callback list but it isn't freed
3306 * since the it may still be in use. The memory for the callback can be
3307 * subsequently freed back by the application by calling rte_free():
3309 * - Immediately - if the port is stopped, or the user knows that no
3310 * callbacks are in flight e.g. if called from the thread doing RX/TX
3313 * - After a short delay - where the delay is sufficient to allow any
3314 * in-flight callbacks to complete.
3317 * The port identifier of the Ethernet device.
3319 * The queue on the Ethernet device from which the callback is to be removed.
3321 * User supplied callback created via rte_eth_add_rx_callback().
3324 * - 0: Success. Callback was removed.
3325 * - -ENOTSUP: Callback support is not available.
3326 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3327 * is NULL or not found for the port/queue.
3329 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
3330 const struct rte_eth_rxtx_callback *user_cb);
3333 * Remove a TX packet callback from a given port and queue.
3335 * This function is used to removed callbacks that were added to a NIC port
3336 * queue using rte_eth_add_tx_callback().
3338 * Note: the callback is removed from the callback list but it isn't freed
3339 * since the it may still be in use. The memory for the callback can be
3340 * subsequently freed back by the application by calling rte_free():
3342 * - Immediately - if the port is stopped, or the user knows that no
3343 * callbacks are in flight e.g. if called from the thread doing RX/TX
3346 * - After a short delay - where the delay is sufficient to allow any
3347 * in-flight callbacks to complete.
3350 * The port identifier of the Ethernet device.
3352 * The queue on the Ethernet device from which the callback is to be removed.
3354 * User supplied callback created via rte_eth_add_tx_callback().
3357 * - 0: Success. Callback was removed.
3358 * - -ENOTSUP: Callback support is not available.
3359 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3360 * is NULL or not found for the port/queue.
3362 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
3363 const struct rte_eth_rxtx_callback *user_cb);
3366 * Retrieve information about given port's RX queue.
3369 * The port identifier of the Ethernet device.
3371 * The RX queue on the Ethernet device for which information
3372 * will be retrieved.
3374 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
3375 * the information of the Ethernet device.
3379 * - -ENOTSUP: routine is not supported by the device PMD.
3380 * - -EINVAL: The port_id or the queue_id is out of range.
3382 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3383 struct rte_eth_rxq_info *qinfo);
3386 * Retrieve information about given port's TX queue.
3389 * The port identifier of the Ethernet device.
3391 * The TX queue on the Ethernet device for which information
3392 * will be retrieved.
3394 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
3395 * the information of the Ethernet device.
3399 * - -ENOTSUP: routine is not supported by the device PMD.
3400 * - -EINVAL: The port_id or the queue_id is out of range.
3402 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3403 struct rte_eth_txq_info *qinfo);
3406 * Retrieve device registers and register attributes (number of registers and
3410 * The port identifier of the Ethernet device.
3412 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
3413 * NULL the function fills in the width and length fields. If non-NULL
3414 * the registers are put into the buffer pointed at by the data field.
3416 * - (0) if successful.
3417 * - (-ENOTSUP) if hardware doesn't support.
3418 * - (-ENODEV) if *port_id* invalid.
3419 * - (-EIO) if device is removed.
3420 * - others depends on the specific operations implementation.
3422 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
3425 * Retrieve size of device EEPROM
3428 * The port identifier of the Ethernet device.
3430 * - (>=0) EEPROM size if successful.
3431 * - (-ENOTSUP) if hardware doesn't support.
3432 * - (-ENODEV) if *port_id* invalid.
3433 * - (-EIO) if device is removed.
3434 * - others depends on the specific operations implementation.
3436 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
3439 * Retrieve EEPROM and EEPROM attribute
3442 * The port identifier of the Ethernet device.
3444 * The template includes buffer for return EEPROM data and
3445 * EEPROM attributes to be filled.
3447 * - (0) if successful.
3448 * - (-ENOTSUP) if hardware doesn't support.
3449 * - (-ENODEV) if *port_id* invalid.
3450 * - (-EIO) if device is removed.
3451 * - others depends on the specific operations implementation.
3453 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3456 * Program EEPROM with provided data
3459 * The port identifier of the Ethernet device.
3461 * The template includes EEPROM data for programming and
3462 * EEPROM attributes to be filled
3464 * - (0) if successful.
3465 * - (-ENOTSUP) if hardware doesn't support.
3466 * - (-ENODEV) if *port_id* invalid.
3467 * - (-EIO) if device is removed.
3468 * - others depends on the specific operations implementation.
3470 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3474 * @b EXPERIMENTAL: this API may change without prior notice.
3476 * Retrieve the type and size of plugin module EEPROM
3479 * The port identifier of the Ethernet device.
3481 * The type and size of plugin module EEPROM.
3483 * - (0) if successful.
3484 * - (-ENOTSUP) if hardware doesn't support.
3485 * - (-ENODEV) if *port_id* invalid.
3486 * - (-EIO) if device is removed.
3487 * - others depends on the specific operations implementation.
3489 int __rte_experimental
3490 rte_eth_dev_get_module_info(uint16_t port_id,
3491 struct rte_eth_dev_module_info *modinfo);
3495 * @b EXPERIMENTAL: this API may change without prior notice.
3497 * Retrieve the data of plugin module EEPROM
3500 * The port identifier of the Ethernet device.
3502 * The template includes the plugin module EEPROM attributes, and the
3503 * buffer for return plugin module EEPROM data.
3505 * - (0) if successful.
3506 * - (-ENOTSUP) if hardware doesn't support.
3507 * - (-ENODEV) if *port_id* invalid.
3508 * - (-EIO) if device is removed.
3509 * - others depends on the specific operations implementation.
3511 int __rte_experimental
3512 rte_eth_dev_get_module_eeprom(uint16_t port_id,
3513 struct rte_dev_eeprom_info *info);
3516 * Set the list of multicast addresses to filter on an Ethernet device.
3519 * The port identifier of the Ethernet device.
3520 * @param mc_addr_set
3521 * The array of multicast addresses to set. Equal to NULL when the function
3522 * is invoked to flush the set of filtered addresses.
3524 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
3525 * when the function is invoked to flush the set of filtered addresses.
3527 * - (0) if successful.
3528 * - (-ENODEV) if *port_id* invalid.
3529 * - (-EIO) if device is removed.
3530 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
3531 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
3533 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
3534 struct ether_addr *mc_addr_set,
3535 uint32_t nb_mc_addr);
3538 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
3541 * The port identifier of the Ethernet device.
3545 * - -ENODEV: The port ID is invalid.
3546 * - -EIO: if device is removed.
3547 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3549 int rte_eth_timesync_enable(uint16_t port_id);
3552 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
3555 * The port identifier of the Ethernet device.
3559 * - -ENODEV: The port ID is invalid.
3560 * - -EIO: if device is removed.
3561 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3563 int rte_eth_timesync_disable(uint16_t port_id);
3566 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
3569 * The port identifier of the Ethernet device.
3571 * Pointer to the timestamp struct.
3573 * Device specific flags. Used to pass the RX timesync register index to
3574 * i40e. Unused in igb/ixgbe, pass 0 instead.
3578 * - -EINVAL: No timestamp is available.
3579 * - -ENODEV: The port ID is invalid.
3580 * - -EIO: if device is removed.
3581 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3583 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
3584 struct timespec *timestamp, uint32_t flags);
3587 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
3590 * The port identifier of the Ethernet device.
3592 * Pointer to the timestamp struct.
3596 * - -EINVAL: No timestamp is available.
3597 * - -ENODEV: The port ID is invalid.
3598 * - -EIO: if device is removed.
3599 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3601 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
3602 struct timespec *timestamp);
3605 * Adjust the timesync clock on an Ethernet device.
3607 * This is usually used in conjunction with other Ethdev timesync functions to
3608 * synchronize the device time using the IEEE1588/802.1AS protocol.
3611 * The port identifier of the Ethernet device.
3613 * The adjustment in nanoseconds.
3617 * - -ENODEV: The port ID is invalid.
3618 * - -EIO: if device is removed.
3619 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3621 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
3624 * Read the time from the timesync clock on an Ethernet device.
3626 * This is usually used in conjunction with other Ethdev timesync functions to
3627 * synchronize the device time using the IEEE1588/802.1AS protocol.
3630 * The port identifier of the Ethernet device.
3632 * Pointer to the timespec struct that holds the time.
3637 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
3640 * Set the time of the timesync clock on an Ethernet device.
3642 * This is usually used in conjunction with other Ethdev timesync functions to
3643 * synchronize the device time using the IEEE1588/802.1AS protocol.
3646 * The port identifier of the Ethernet device.
3648 * Pointer to the timespec struct that holds the time.
3652 * - -EINVAL: No timestamp is available.
3653 * - -ENODEV: The port ID is invalid.
3654 * - -EIO: if device is removed.
3655 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3657 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
3660 * Config l2 tunnel ether type of an Ethernet device for filtering specific
3661 * tunnel packets by ether type.
3664 * The port identifier of the Ethernet device.
3666 * l2 tunnel configuration.
3669 * - (0) if successful.
3670 * - (-ENODEV) if port identifier is invalid.
3671 * - (-EIO) if device is removed.
3672 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3675 rte_eth_dev_l2_tunnel_eth_type_conf(uint16_t port_id,
3676 struct rte_eth_l2_tunnel_conf *l2_tunnel);
3679 * Enable/disable l2 tunnel offload functions. Include,
3680 * 1, The ability of parsing a type of l2 tunnel of an Ethernet device.
3681 * Filtering, forwarding and offloading this type of tunnel packets depend on
3683 * 2, Stripping the l2 tunnel tag.
3684 * 3, Insertion of the l2 tunnel tag.
3685 * 4, Forwarding the packets based on the l2 tunnel tag.
3688 * The port identifier of the Ethernet device.
3690 * l2 tunnel parameters.
3692 * Indicate the offload function.
3694 * Enable or disable this function.
3697 * - (0) if successful.
3698 * - (-ENODEV) if port identifier is invalid.
3699 * - (-EIO) if device is removed.
3700 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3703 rte_eth_dev_l2_tunnel_offload_set(uint16_t port_id,
3704 struct rte_eth_l2_tunnel_conf *l2_tunnel,
3709 * Get the port id from device name. The device name should be specified
3711 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
3712 * - SoC device name, for example- fsl-gmac0
3713 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
3716 * pci address or name of the device
3718 * pointer to port identifier of the device
3720 * - (0) if successful and port_id is filled.
3721 * - (-ENODEV or -EINVAL) on failure.
3724 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
3727 * Get the device name from port id. The device name is specified as below:
3728 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
3729 * - SoC device name, for example- fsl-gmac0
3730 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
3733 * Port identifier of the device.
3735 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
3737 * - (0) if successful.
3738 * - (-EINVAL) on failure.
3741 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
3744 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
3745 * the ethernet device information, otherwise adjust them to boundaries.
3748 * The port identifier of the Ethernet device.
3750 * A pointer to a uint16_t where the number of receive
3751 * descriptors stored.
3753 * A pointer to a uint16_t where the number of transmit
3754 * descriptors stored.
3756 * - (0) if successful.
3757 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
3759 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
3760 uint16_t *nb_rx_desc,
3761 uint16_t *nb_tx_desc);
3764 * Test if a port supports specific mempool ops.
3767 * Port identifier of the Ethernet device.
3769 * The name of the pool operations to test.
3771 * - 0: best mempool ops choice for this port.
3772 * - 1: mempool ops are supported for this port.
3773 * - -ENOTSUP: mempool ops not supported for this port.
3774 * - -ENODEV: Invalid port Identifier.
3775 * - -EINVAL: Pool param is null.
3778 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
3781 * Get the security context for the Ethernet device.
3784 * Port identifier of the Ethernet device
3787 * - pointer to security context on success.
3790 rte_eth_dev_get_sec_ctx(uint16_t port_id);
3793 #include <rte_ethdev_core.h>
3797 * Retrieve a burst of input packets from a receive queue of an Ethernet
3798 * device. The retrieved packets are stored in *rte_mbuf* structures whose
3799 * pointers are supplied in the *rx_pkts* array.
3801 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
3802 * receive queue, up to *nb_pkts* packets, and for each completed RX
3803 * descriptor in the ring, it performs the following operations:
3805 * - Initialize the *rte_mbuf* data structure associated with the
3806 * RX descriptor according to the information provided by the NIC into
3807 * that RX descriptor.
3809 * - Store the *rte_mbuf* data structure into the next entry of the
3812 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
3813 * allocated from the memory pool associated with the receive queue at
3814 * initialization time.
3816 * When retrieving an input packet that was scattered by the controller
3817 * into multiple receive descriptors, the rte_eth_rx_burst() function
3818 * appends the associated *rte_mbuf* buffers to the first buffer of the
3821 * The rte_eth_rx_burst() function returns the number of packets
3822 * actually retrieved, which is the number of *rte_mbuf* data structures
3823 * effectively supplied into the *rx_pkts* array.
3824 * A return value equal to *nb_pkts* indicates that the RX queue contained
3825 * at least *rx_pkts* packets, and this is likely to signify that other
3826 * received packets remain in the input queue. Applications implementing
3827 * a "retrieve as much received packets as possible" policy can check this
3828 * specific case and keep invoking the rte_eth_rx_burst() function until
3829 * a value less than *nb_pkts* is returned.
3831 * This receive method has the following advantages:
3833 * - It allows a run-to-completion network stack engine to retrieve and
3834 * to immediately process received packets in a fast burst-oriented
3835 * approach, avoiding the overhead of unnecessary intermediate packet
3836 * queue/dequeue operations.
3838 * - Conversely, it also allows an asynchronous-oriented processing
3839 * method to retrieve bursts of received packets and to immediately
3840 * queue them for further parallel processing by another logical core,
3841 * for instance. However, instead of having received packets being
3842 * individually queued by the driver, this approach allows the caller
3843 * of the rte_eth_rx_burst() function to queue a burst of retrieved
3844 * packets at a time and therefore dramatically reduce the cost of
3845 * enqueue/dequeue operations per packet.
3847 * - It allows the rte_eth_rx_burst() function of the driver to take
3848 * advantage of burst-oriented hardware features (CPU cache,
3849 * prefetch instructions, and so on) to minimize the number of CPU
3850 * cycles per packet.
3852 * To summarize, the proposed receive API enables many
3853 * burst-oriented optimizations in both synchronous and asynchronous
3854 * packet processing environments with no overhead in both cases.
3856 * The rte_eth_rx_burst() function does not provide any error
3857 * notification to avoid the corresponding overhead. As a hint, the
3858 * upper-level application might check the status of the device link once
3859 * being systematically returned a 0 value for a given number of tries.
3862 * The port identifier of the Ethernet device.
3864 * The index of the receive queue from which to retrieve input packets.
3865 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3866 * to rte_eth_dev_configure().
3868 * The address of an array of pointers to *rte_mbuf* structures that
3869 * must be large enough to store *nb_pkts* pointers in it.
3871 * The maximum number of packets to retrieve.
3873 * The number of packets actually retrieved, which is the number
3874 * of pointers to *rte_mbuf* structures effectively supplied to the
3877 static inline uint16_t
3878 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
3879 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
3881 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3884 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3885 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
3886 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
3888 if (queue_id >= dev->data->nb_rx_queues) {
3889 RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n", queue_id);
3893 nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
3896 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
3897 if (unlikely(dev->post_rx_burst_cbs[queue_id] != NULL)) {
3898 struct rte_eth_rxtx_callback *cb =
3899 dev->post_rx_burst_cbs[queue_id];
3902 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
3903 nb_pkts, cb->param);
3905 } while (cb != NULL);
3913 * Get the number of used descriptors of a rx queue
3916 * The port identifier of the Ethernet device.
3918 * The queue id on the specific port.
3920 * The number of used descriptors in the specific queue, or:
3921 * (-EINVAL) if *port_id* or *queue_id* is invalid
3922 * (-ENOTSUP) if the device does not support this function
3925 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
3927 struct rte_eth_dev *dev;
3929 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -EINVAL);
3930 dev = &rte_eth_devices[port_id];
3931 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_count, -ENOTSUP);
3932 if (queue_id >= dev->data->nb_rx_queues)
3935 return (int)(*dev->dev_ops->rx_queue_count)(dev, queue_id);
3939 * Check if the DD bit of the specific RX descriptor in the queue has been set
3942 * The port identifier of the Ethernet device.
3944 * The queue id on the specific port.
3946 * The offset of the descriptor ID from tail.
3948 * - (1) if the specific DD bit is set.
3949 * - (0) if the specific DD bit is not set.
3950 * - (-ENODEV) if *port_id* invalid.
3951 * - (-ENOTSUP) if the device does not support this function
3954 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
3956 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3957 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3958 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_done, -ENOTSUP);
3959 return (*dev->dev_ops->rx_descriptor_done)( \
3960 dev->data->rx_queues[queue_id], offset);
3963 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
3964 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
3965 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
3968 * Check the status of a Rx descriptor in the queue
3970 * It should be called in a similar context than the Rx function:
3971 * - on a dataplane core
3972 * - not concurrently on the same queue
3974 * Since it's a dataplane function, no check is performed on port_id and
3975 * queue_id. The caller must therefore ensure that the port is enabled
3976 * and the queue is configured and running.
3978 * Note: accessing to a random descriptor in the ring may trigger cache
3979 * misses and have a performance impact.
3982 * A valid port identifier of the Ethernet device which.
3984 * A valid Rx queue identifier on this port.
3986 * The offset of the descriptor starting from tail (0 is the next
3987 * packet to be received by the driver).
3990 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
3992 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
3993 * not yet processed by the driver (i.e. in the receive queue).
3994 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
3995 * the driver and not yet returned to hw, or reserved by the hw.
3996 * - (-EINVAL) bad descriptor offset.
3997 * - (-ENOTSUP) if the device does not support this function.
3998 * - (-ENODEV) bad port or queue (only if compiled with debug).
4001 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
4004 struct rte_eth_dev *dev;
4007 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4008 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4010 dev = &rte_eth_devices[port_id];
4011 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4012 if (queue_id >= dev->data->nb_rx_queues)
4015 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_status, -ENOTSUP);
4016 rxq = dev->data->rx_queues[queue_id];
4018 return (*dev->dev_ops->rx_descriptor_status)(rxq, offset);
4021 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
4022 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
4023 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
4026 * Check the status of a Tx descriptor in the queue.
4028 * It should be called in a similar context than the Tx function:
4029 * - on a dataplane core
4030 * - not concurrently on the same queue
4032 * Since it's a dataplane function, no check is performed on port_id and
4033 * queue_id. The caller must therefore ensure that the port is enabled
4034 * and the queue is configured and running.
4036 * Note: accessing to a random descriptor in the ring may trigger cache
4037 * misses and have a performance impact.
4040 * A valid port identifier of the Ethernet device which.
4042 * A valid Tx queue identifier on this port.
4044 * The offset of the descriptor starting from tail (0 is the place where
4045 * the next packet will be send).
4048 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
4049 * in the transmit queue.
4050 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
4051 * be reused by the driver.
4052 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
4053 * driver or the hardware.
4054 * - (-EINVAL) bad descriptor offset.
4055 * - (-ENOTSUP) if the device does not support this function.
4056 * - (-ENODEV) bad port or queue (only if compiled with debug).
4058 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
4059 uint16_t queue_id, uint16_t offset)
4061 struct rte_eth_dev *dev;
4064 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4065 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4067 dev = &rte_eth_devices[port_id];
4068 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4069 if (queue_id >= dev->data->nb_tx_queues)
4072 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_descriptor_status, -ENOTSUP);
4073 txq = dev->data->tx_queues[queue_id];
4075 return (*dev->dev_ops->tx_descriptor_status)(txq, offset);
4079 * Send a burst of output packets on a transmit queue of an Ethernet device.
4081 * The rte_eth_tx_burst() function is invoked to transmit output packets
4082 * on the output queue *queue_id* of the Ethernet device designated by its
4084 * The *nb_pkts* parameter is the number of packets to send which are
4085 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
4086 * allocated from a pool created with rte_pktmbuf_pool_create().
4087 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
4088 * up to the number of transmit descriptors available in the TX ring of the
4090 * For each packet to send, the rte_eth_tx_burst() function performs
4091 * the following operations:
4093 * - Pick up the next available descriptor in the transmit ring.
4095 * - Free the network buffer previously sent with that descriptor, if any.
4097 * - Initialize the transmit descriptor with the information provided
4098 * in the *rte_mbuf data structure.
4100 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
4101 * the rte_eth_tx_burst() function uses several transmit descriptors
4104 * The rte_eth_tx_burst() function returns the number of packets it
4105 * actually sent. A return value equal to *nb_pkts* means that all packets
4106 * have been sent, and this is likely to signify that other output packets
4107 * could be immediately transmitted again. Applications that implement a
4108 * "send as many packets to transmit as possible" policy can check this
4109 * specific case and keep invoking the rte_eth_tx_burst() function until
4110 * a value less than *nb_pkts* is returned.
4112 * It is the responsibility of the rte_eth_tx_burst() function to
4113 * transparently free the memory buffers of packets previously sent.
4114 * This feature is driven by the *tx_free_thresh* value supplied to the
4115 * rte_eth_dev_configure() function at device configuration time.
4116 * When the number of free TX descriptors drops below this threshold, the
4117 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
4118 * of those packets whose transmission was effectively completed.
4120 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
4121 * invoke this function concurrently on the same tx queue without SW lock.
4122 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
4124 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
4128 * The port identifier of the Ethernet device.
4130 * The index of the transmit queue through which output packets must be
4132 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4133 * to rte_eth_dev_configure().
4135 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
4136 * which contain the output packets.
4138 * The maximum number of packets to transmit.
4140 * The number of output packets actually stored in transmit descriptors of
4141 * the transmit ring. The return value can be less than the value of the
4142 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
4144 static inline uint16_t
4145 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
4146 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4148 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4150 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4151 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
4152 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
4154 if (queue_id >= dev->data->nb_tx_queues) {
4155 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
4160 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4161 struct rte_eth_rxtx_callback *cb = dev->pre_tx_burst_cbs[queue_id];
4163 if (unlikely(cb != NULL)) {
4165 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
4168 } while (cb != NULL);
4172 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
4177 * @b EXPERIMENTAL: this API may change without prior notice
4179 * Process a burst of output packets on a transmit queue of an Ethernet device.
4181 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
4182 * transmitted on the output queue *queue_id* of the Ethernet device designated
4184 * The *nb_pkts* parameter is the number of packets to be prepared which are
4185 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
4186 * allocated from a pool created with rte_pktmbuf_pool_create().
4187 * For each packet to send, the rte_eth_tx_prepare() function performs
4188 * the following operations:
4190 * - Check if packet meets devices requirements for tx offloads.
4192 * - Check limitations about number of segments.
4194 * - Check additional requirements when debug is enabled.
4196 * - Update and/or reset required checksums when tx offload is set for packet.
4198 * Since this function can modify packet data, provided mbufs must be safely
4199 * writable (e.g. modified data cannot be in shared segment).
4201 * The rte_eth_tx_prepare() function returns the number of packets ready to be
4202 * sent. A return value equal to *nb_pkts* means that all packets are valid and
4203 * ready to be sent, otherwise stops processing on the first invalid packet and
4204 * leaves the rest packets untouched.
4206 * When this functionality is not implemented in the driver, all packets are
4207 * are returned untouched.
4210 * The port identifier of the Ethernet device.
4211 * The value must be a valid port id.
4213 * The index of the transmit queue through which output packets must be
4215 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4216 * to rte_eth_dev_configure().
4218 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
4219 * which contain the output packets.
4221 * The maximum number of packets to process.
4223 * The number of packets correct and ready to be sent. The return value can be
4224 * less than the value of the *tx_pkts* parameter when some packet doesn't
4225 * meet devices requirements with rte_errno set appropriately:
4226 * - -EINVAL: offload flags are not correctly set
4227 * - -ENOTSUP: the offload feature is not supported by the hardware
4231 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
4233 static inline uint16_t
4234 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
4235 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4237 struct rte_eth_dev *dev;
4239 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4240 if (!rte_eth_dev_is_valid_port(port_id)) {
4241 RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
4242 rte_errno = -EINVAL;
4247 dev = &rte_eth_devices[port_id];
4249 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4250 if (queue_id >= dev->data->nb_tx_queues) {
4251 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
4252 rte_errno = -EINVAL;
4257 if (!dev->tx_pkt_prepare)
4260 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
4267 * Native NOOP operation for compilation targets which doesn't require any
4268 * preparations steps, and functional NOOP may introduce unnecessary performance
4271 * Generally this is not a good idea to turn it on globally and didn't should
4272 * be used if behavior of tx_preparation can change.
4275 static inline uint16_t
4276 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
4277 __rte_unused uint16_t queue_id,
4278 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4286 * Send any packets queued up for transmission on a port and HW queue
4288 * This causes an explicit flush of packets previously buffered via the
4289 * rte_eth_tx_buffer() function. It returns the number of packets successfully
4290 * sent to the NIC, and calls the error callback for any unsent packets. Unless
4291 * explicitly set up otherwise, the default callback simply frees the unsent
4292 * packets back to the owning mempool.
4295 * The port identifier of the Ethernet device.
4297 * The index of the transmit queue through which output packets must be
4299 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4300 * to rte_eth_dev_configure().
4302 * Buffer of packets to be transmit.
4304 * The number of packets successfully sent to the Ethernet device. The error
4305 * callback is called for any packets which could not be sent.
4307 static inline uint16_t
4308 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
4309 struct rte_eth_dev_tx_buffer *buffer)
4312 uint16_t to_send = buffer->length;
4317 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
4321 /* All packets sent, or to be dealt with by callback below */
4322 if (unlikely(sent != to_send))
4323 buffer->error_callback(&buffer->pkts[sent],
4324 (uint16_t)(to_send - sent),
4325 buffer->error_userdata);
4331 * Buffer a single packet for future transmission on a port and queue
4333 * This function takes a single mbuf/packet and buffers it for later
4334 * transmission on the particular port and queue specified. Once the buffer is
4335 * full of packets, an attempt will be made to transmit all the buffered
4336 * packets. In case of error, where not all packets can be transmitted, a
4337 * callback is called with the unsent packets as a parameter. If no callback
4338 * is explicitly set up, the unsent packets are just freed back to the owning
4339 * mempool. The function returns the number of packets actually sent i.e.
4340 * 0 if no buffer flush occurred, otherwise the number of packets successfully
4344 * The port identifier of the Ethernet device.
4346 * The index of the transmit queue through which output packets must be
4348 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4349 * to rte_eth_dev_configure().
4351 * Buffer used to collect packets to be sent.
4353 * Pointer to the packet mbuf to be sent.
4355 * 0 = packet has been buffered for later transmission
4356 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
4357 * causing N packets to be sent, and the error callback to be called for
4360 static __rte_always_inline uint16_t
4361 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
4362 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
4364 buffer->pkts[buffer->length++] = tx_pkt;
4365 if (buffer->length < buffer->size)
4368 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
4375 #endif /* _RTE_ETHDEV_H_ */