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"
165 * A structure used to retrieve statistics for an Ethernet port.
166 * Not all statistics fields in struct rte_eth_stats are supported
167 * by any type of network interface card (NIC). If any statistics
168 * field is not supported, its value is 0.
170 struct rte_eth_stats {
171 uint64_t ipackets; /**< Total number of successfully received packets. */
172 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
173 uint64_t ibytes; /**< Total number of successfully received bytes. */
174 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
176 /**< Total of RX packets dropped by the HW,
177 * because there are no available buffer (i.e. RX queues are full).
179 uint64_t ierrors; /**< Total number of erroneous received packets. */
180 uint64_t oerrors; /**< Total number of failed transmitted packets. */
181 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
182 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
183 /**< Total number of queue RX packets. */
184 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
185 /**< Total number of queue TX packets. */
186 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
187 /**< Total number of successfully received queue bytes. */
188 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
189 /**< Total number of successfully transmitted queue bytes. */
190 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
191 /**< Total number of queue packets received that are dropped. */
195 * Device supported speeds bitmap flags
197 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
198 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
199 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
200 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
201 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
202 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
203 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
204 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
205 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
206 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
207 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
208 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
209 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
210 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
211 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
212 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
215 * Ethernet numeric link speeds in Mbps
217 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
218 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
219 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
220 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
221 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
222 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
223 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
224 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
225 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
226 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
227 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
228 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
229 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
232 * A structure used to retrieve link-level information of an Ethernet port.
235 struct rte_eth_link {
236 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
237 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
238 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
239 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
240 } __attribute__((aligned(8))); /**< aligned for atomic64 read/write */
242 /* Utility constants */
243 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
244 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
245 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
246 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
247 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
248 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
251 * A structure used to configure the ring threshold registers of an RX/TX
252 * queue for an Ethernet port.
254 struct rte_eth_thresh {
255 uint8_t pthresh; /**< Ring prefetch threshold. */
256 uint8_t hthresh; /**< Ring host threshold. */
257 uint8_t wthresh; /**< Ring writeback threshold. */
261 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
263 #define ETH_MQ_RX_RSS_FLAG 0x1
264 #define ETH_MQ_RX_DCB_FLAG 0x2
265 #define ETH_MQ_RX_VMDQ_FLAG 0x4
268 * A set of values to identify what method is to be used to route
269 * packets to multiple queues.
271 enum rte_eth_rx_mq_mode {
272 /** None of DCB,RSS or VMDQ mode */
275 /** For RX side, only RSS is on */
276 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
277 /** For RX side,only DCB is on. */
278 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
279 /** Both DCB and RSS enable */
280 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
282 /** Only VMDQ, no RSS nor DCB */
283 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
284 /** RSS mode with VMDQ */
285 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
286 /** Use VMDQ+DCB to route traffic to queues */
287 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
288 /** Enable both VMDQ and DCB in VMDq */
289 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
294 * for rx mq mode backward compatible
296 #define ETH_RSS ETH_MQ_RX_RSS
297 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
298 #define ETH_DCB_RX ETH_MQ_RX_DCB
301 * A set of values to identify what method is to be used to transmit
302 * packets using multi-TCs.
304 enum rte_eth_tx_mq_mode {
305 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
306 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
307 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
308 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
312 * for tx mq mode backward compatible
314 #define ETH_DCB_NONE ETH_MQ_TX_NONE
315 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
316 #define ETH_DCB_TX ETH_MQ_TX_DCB
319 * A structure used to configure the RX features of an Ethernet port.
321 struct rte_eth_rxmode {
322 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
323 enum rte_eth_rx_mq_mode mq_mode;
324 uint32_t max_rx_pkt_len; /**< Only used if jumbo_frame enabled. */
325 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
327 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
328 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
329 * structure are allowed to be set.
334 * Below bitfield API is obsolete. Application should
335 * enable per-port offloads using the offload field
338 uint16_t header_split : 1, /**< Header Split enable. */
339 hw_ip_checksum : 1, /**< IP/UDP/TCP checksum offload enable. */
340 hw_vlan_filter : 1, /**< VLAN filter enable. */
341 hw_vlan_strip : 1, /**< VLAN strip enable. */
342 hw_vlan_extend : 1, /**< Extended VLAN enable. */
343 jumbo_frame : 1, /**< Jumbo Frame Receipt enable. */
344 hw_strip_crc : 1, /**< Enable CRC stripping by hardware. */
345 enable_scatter : 1, /**< Enable scatter packets rx handler */
346 enable_lro : 1, /**< Enable LRO */
347 hw_timestamp : 1, /**< Enable HW timestamp */
348 security : 1, /**< Enable rte_security offloads */
350 * When set the offload bitfield should be ignored.
351 * Instead per-port Rx offloads should be set on offloads
353 * Per-queue offloads shuold be set on rte_eth_rxq_conf
355 * This bit is temporary till rxmode bitfield offloads API will
358 ignore_offload_bitfield : 1;
362 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
363 * Note that single VLAN is treated the same as inner VLAN.
366 ETH_VLAN_TYPE_UNKNOWN = 0,
367 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
368 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
373 * A structure used to describe a vlan filter.
374 * If the bit corresponding to a VID is set, such VID is on.
376 struct rte_vlan_filter_conf {
381 * A structure used to configure the Receive Side Scaling (RSS) feature
382 * of an Ethernet port.
383 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
384 * to an array holding the RSS key to use for hashing specific header
385 * fields of received packets. The length of this array should be indicated
386 * by *rss_key_len* below. Otherwise, a default random hash key is used by
389 * The *rss_key_len* field of the *rss_conf* structure indicates the length
390 * in bytes of the array pointed by *rss_key*. To be compatible, this length
391 * will be checked in i40e only. Others assume 40 bytes to be used as before.
393 * The *rss_hf* field of the *rss_conf* structure indicates the different
394 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
395 * Supplying an *rss_hf* equal to zero disables the RSS feature.
397 struct rte_eth_rss_conf {
398 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
399 uint8_t rss_key_len; /**< hash key length in bytes. */
400 uint64_t rss_hf; /**< Hash functions to apply - see below. */
404 * The RSS offload types are defined based on flow types which are defined
405 * in rte_eth_ctrl.h. Different NIC hardwares may support different RSS offload
406 * types. The supported flow types or RSS offload types can be queried by
407 * rte_eth_dev_info_get().
409 #define ETH_RSS_IPV4 (1ULL << RTE_ETH_FLOW_IPV4)
410 #define ETH_RSS_FRAG_IPV4 (1ULL << RTE_ETH_FLOW_FRAG_IPV4)
411 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_TCP)
412 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_UDP)
413 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_SCTP)
414 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_OTHER)
415 #define ETH_RSS_IPV6 (1ULL << RTE_ETH_FLOW_IPV6)
416 #define ETH_RSS_FRAG_IPV6 (1ULL << RTE_ETH_FLOW_FRAG_IPV6)
417 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_TCP)
418 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_UDP)
419 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_SCTP)
420 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_OTHER)
421 #define ETH_RSS_L2_PAYLOAD (1ULL << RTE_ETH_FLOW_L2_PAYLOAD)
422 #define ETH_RSS_IPV6_EX (1ULL << RTE_ETH_FLOW_IPV6_EX)
423 #define ETH_RSS_IPV6_TCP_EX (1ULL << RTE_ETH_FLOW_IPV6_TCP_EX)
424 #define ETH_RSS_IPV6_UDP_EX (1ULL << RTE_ETH_FLOW_IPV6_UDP_EX)
425 #define ETH_RSS_PORT (1ULL << RTE_ETH_FLOW_PORT)
426 #define ETH_RSS_VXLAN (1ULL << RTE_ETH_FLOW_VXLAN)
427 #define ETH_RSS_GENEVE (1ULL << RTE_ETH_FLOW_GENEVE)
428 #define ETH_RSS_NVGRE (1ULL << RTE_ETH_FLOW_NVGRE)
430 #define ETH_RSS_IP ( \
432 ETH_RSS_FRAG_IPV4 | \
433 ETH_RSS_NONFRAG_IPV4_OTHER | \
435 ETH_RSS_FRAG_IPV6 | \
436 ETH_RSS_NONFRAG_IPV6_OTHER | \
439 #define ETH_RSS_UDP ( \
440 ETH_RSS_NONFRAG_IPV4_UDP | \
441 ETH_RSS_NONFRAG_IPV6_UDP | \
444 #define ETH_RSS_TCP ( \
445 ETH_RSS_NONFRAG_IPV4_TCP | \
446 ETH_RSS_NONFRAG_IPV6_TCP | \
449 #define ETH_RSS_SCTP ( \
450 ETH_RSS_NONFRAG_IPV4_SCTP | \
451 ETH_RSS_NONFRAG_IPV6_SCTP)
453 #define ETH_RSS_TUNNEL ( \
458 /**< Mask of valid RSS hash protocols */
459 #define ETH_RSS_PROTO_MASK ( \
461 ETH_RSS_FRAG_IPV4 | \
462 ETH_RSS_NONFRAG_IPV4_TCP | \
463 ETH_RSS_NONFRAG_IPV4_UDP | \
464 ETH_RSS_NONFRAG_IPV4_SCTP | \
465 ETH_RSS_NONFRAG_IPV4_OTHER | \
467 ETH_RSS_FRAG_IPV6 | \
468 ETH_RSS_NONFRAG_IPV6_TCP | \
469 ETH_RSS_NONFRAG_IPV6_UDP | \
470 ETH_RSS_NONFRAG_IPV6_SCTP | \
471 ETH_RSS_NONFRAG_IPV6_OTHER | \
472 ETH_RSS_L2_PAYLOAD | \
474 ETH_RSS_IPV6_TCP_EX | \
475 ETH_RSS_IPV6_UDP_EX | \
482 * Definitions used for redirection table entry size.
483 * Some RSS RETA sizes may not be supported by some drivers, check the
484 * documentation or the description of relevant functions for more details.
486 #define ETH_RSS_RETA_SIZE_64 64
487 #define ETH_RSS_RETA_SIZE_128 128
488 #define ETH_RSS_RETA_SIZE_256 256
489 #define ETH_RSS_RETA_SIZE_512 512
490 #define RTE_RETA_GROUP_SIZE 64
492 /* Definitions used for VMDQ and DCB functionality */
493 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
494 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
495 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
496 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
498 /* DCB capability defines */
499 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
500 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
502 /* Definitions used for VLAN Offload functionality */
503 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
504 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
505 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
507 /* Definitions used for mask VLAN setting */
508 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
509 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
510 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
511 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
513 /* Definitions used for receive MAC address */
514 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
516 /* Definitions used for unicast hash */
517 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
519 /* Definitions used for VMDQ pool rx mode setting */
520 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
521 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
522 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
523 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
524 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
526 /** Maximum nb. of vlan per mirror rule */
527 #define ETH_MIRROR_MAX_VLANS 64
529 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
530 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
531 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
532 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
533 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
536 * A structure used to configure VLAN traffic mirror of an Ethernet port.
538 struct rte_eth_vlan_mirror {
539 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
540 /** VLAN ID list for vlan mirroring. */
541 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
545 * A structure used to configure traffic mirror of an Ethernet port.
547 struct rte_eth_mirror_conf {
548 uint8_t rule_type; /**< Mirroring rule type */
549 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
550 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
551 /** VLAN ID setting for VLAN mirroring. */
552 struct rte_eth_vlan_mirror vlan;
556 * A structure used to configure 64 entries of Redirection Table of the
557 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
558 * more than 64 entries supported by hardware, an array of this structure
561 struct rte_eth_rss_reta_entry64 {
563 /**< Mask bits indicate which entries need to be updated/queried. */
564 uint16_t reta[RTE_RETA_GROUP_SIZE];
565 /**< Group of 64 redirection table entries. */
569 * This enum indicates the possible number of traffic classes
570 * in DCB configurations
572 enum rte_eth_nb_tcs {
573 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
574 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
578 * This enum indicates the possible number of queue pools
579 * in VMDQ configurations.
581 enum rte_eth_nb_pools {
582 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
583 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
584 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
585 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
588 /* This structure may be extended in future. */
589 struct rte_eth_dcb_rx_conf {
590 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
591 /** Traffic class each UP mapped to. */
592 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
595 struct rte_eth_vmdq_dcb_tx_conf {
596 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
597 /** Traffic class each UP mapped to. */
598 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
601 struct rte_eth_dcb_tx_conf {
602 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
603 /** Traffic class each UP mapped to. */
604 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
607 struct rte_eth_vmdq_tx_conf {
608 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
612 * A structure used to configure the VMDQ+DCB feature
613 * of an Ethernet port.
615 * Using this feature, packets are routed to a pool of queues, based
616 * on the vlan id in the vlan tag, and then to a specific queue within
617 * that pool, using the user priority vlan tag field.
619 * A default pool may be used, if desired, to route all traffic which
620 * does not match the vlan filter rules.
622 struct rte_eth_vmdq_dcb_conf {
623 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
624 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
625 uint8_t default_pool; /**< The default pool, if applicable */
626 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
628 uint16_t vlan_id; /**< The vlan id of the received frame */
629 uint64_t pools; /**< Bitmask of pools for packet rx */
630 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
631 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
632 /**< Selects a queue in a pool */
636 * A structure used to configure the VMDQ feature of an Ethernet port when
637 * not combined with the DCB feature.
639 * Using this feature, packets are routed to a pool of queues. By default,
640 * the pool selection is based on the MAC address, the vlan id in the
641 * vlan tag as specified in the pool_map array.
642 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
643 * selection using only the MAC address. MAC address to pool mapping is done
644 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
645 * corresponding to the pool id.
647 * Queue selection within the selected pool will be done using RSS when
648 * it is enabled or revert to the first queue of the pool if not.
650 * A default pool may be used, if desired, to route all traffic which
651 * does not match the vlan filter rules or any pool MAC address.
653 struct rte_eth_vmdq_rx_conf {
654 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
655 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
656 uint8_t default_pool; /**< The default pool, if applicable */
657 uint8_t enable_loop_back; /**< Enable VT loop back */
658 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
659 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
661 uint16_t vlan_id; /**< The vlan id of the received frame */
662 uint64_t pools; /**< Bitmask of pools for packet rx */
663 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
667 * A structure used to configure the TX features of an Ethernet port.
669 struct rte_eth_txmode {
670 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
672 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
673 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
674 * structure are allowed to be set.
678 /* For i40e specifically */
681 uint8_t hw_vlan_reject_tagged : 1,
682 /**< If set, reject sending out tagged pkts */
683 hw_vlan_reject_untagged : 1,
684 /**< If set, reject sending out untagged pkts */
685 hw_vlan_insert_pvid : 1;
686 /**< If set, enable port based VLAN insertion */
690 * A structure used to configure an RX ring of an Ethernet port.
692 struct rte_eth_rxconf {
693 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
694 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
695 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
696 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
698 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
699 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
700 * fields on rte_eth_dev_info structure are allowed to be set.
705 #define ETH_TXQ_FLAGS_NOMULTSEGS 0x0001 /**< nb_segs=1 for all mbufs */
706 #define ETH_TXQ_FLAGS_NOREFCOUNT 0x0002 /**< refcnt can be ignored */
707 #define ETH_TXQ_FLAGS_NOMULTMEMP 0x0004 /**< all bufs come from same mempool */
708 #define ETH_TXQ_FLAGS_NOVLANOFFL 0x0100 /**< disable VLAN offload */
709 #define ETH_TXQ_FLAGS_NOXSUMSCTP 0x0200 /**< disable SCTP checksum offload */
710 #define ETH_TXQ_FLAGS_NOXSUMUDP 0x0400 /**< disable UDP checksum offload */
711 #define ETH_TXQ_FLAGS_NOXSUMTCP 0x0800 /**< disable TCP checksum offload */
712 #define ETH_TXQ_FLAGS_NOOFFLOADS \
713 (ETH_TXQ_FLAGS_NOVLANOFFL | ETH_TXQ_FLAGS_NOXSUMSCTP | \
714 ETH_TXQ_FLAGS_NOXSUMUDP | ETH_TXQ_FLAGS_NOXSUMTCP)
715 #define ETH_TXQ_FLAGS_NOXSUMS \
716 (ETH_TXQ_FLAGS_NOXSUMSCTP | ETH_TXQ_FLAGS_NOXSUMUDP | \
717 ETH_TXQ_FLAGS_NOXSUMTCP)
719 * When set the txq_flags should be ignored,
720 * instead per-queue Tx offloads will be set on offloads field
721 * located on rte_eth_txq_conf struct.
722 * This flag is temporary till the rte_eth_txq_conf.txq_flags
723 * API will be deprecated.
725 #define ETH_TXQ_FLAGS_IGNORE 0x8000
728 * A structure used to configure a TX ring of an Ethernet port.
730 struct rte_eth_txconf {
731 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
732 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
733 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
734 less free descriptors than this value. */
736 uint32_t txq_flags; /**< Set flags for the Tx queue */
737 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
739 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
740 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
741 * fields on rte_eth_dev_info structure are allowed to be set.
747 * A structure contains information about HW descriptor ring limitations.
749 struct rte_eth_desc_lim {
750 uint16_t nb_max; /**< Max allowed number of descriptors. */
751 uint16_t nb_min; /**< Min allowed number of descriptors. */
752 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
755 * Max allowed number of segments per whole packet.
757 * - For TSO packet this is the total number of data descriptors allowed
760 * @see nb_mtu_seg_max
765 * Max number of segments per one MTU.
767 * - For non-TSO packet, this is the maximum allowed number of segments
768 * in a single transmit packet.
770 * - For TSO packet each segment within the TSO may span up to this
775 uint16_t nb_mtu_seg_max;
779 * This enum indicates the flow control mode
781 enum rte_eth_fc_mode {
782 RTE_FC_NONE = 0, /**< Disable flow control. */
783 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
784 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
785 RTE_FC_FULL /**< Enable flow control on both side. */
789 * A structure used to configure Ethernet flow control parameter.
790 * These parameters will be configured into the register of the NIC.
791 * Please refer to the corresponding data sheet for proper value.
793 struct rte_eth_fc_conf {
794 uint32_t high_water; /**< High threshold value to trigger XOFF */
795 uint32_t low_water; /**< Low threshold value to trigger XON */
796 uint16_t pause_time; /**< Pause quota in the Pause frame */
797 uint16_t send_xon; /**< Is XON frame need be sent */
798 enum rte_eth_fc_mode mode; /**< Link flow control mode */
799 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
800 uint8_t autoneg; /**< Use Pause autoneg */
804 * A structure used to configure Ethernet priority flow control parameter.
805 * These parameters will be configured into the register of the NIC.
806 * Please refer to the corresponding data sheet for proper value.
808 struct rte_eth_pfc_conf {
809 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
810 uint8_t priority; /**< VLAN User Priority. */
814 * Memory space that can be configured to store Flow Director filters
815 * in the board memory.
817 enum rte_fdir_pballoc_type {
818 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
819 RTE_FDIR_PBALLOC_128K, /**< 128k. */
820 RTE_FDIR_PBALLOC_256K, /**< 256k. */
824 * Select report mode of FDIR hash information in RX descriptors.
826 enum rte_fdir_status_mode {
827 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
828 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
829 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
833 * A structure used to configure the Flow Director (FDIR) feature
834 * of an Ethernet port.
836 * If mode is RTE_FDIR_DISABLE, the pballoc value is ignored.
838 struct rte_fdir_conf {
839 enum rte_fdir_mode mode; /**< Flow Director mode. */
840 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
841 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
842 /** RX queue of packets matching a "drop" filter in perfect mode. */
844 struct rte_eth_fdir_masks mask;
845 struct rte_eth_fdir_flex_conf flex_conf;
846 /**< Flex payload configuration. */
850 * UDP tunneling configuration.
851 * Used to config the UDP port for a type of tunnel.
852 * NICs need the UDP port to identify the tunnel type.
853 * Normally a type of tunnel has a default UDP port, this structure can be used
854 * in case if the users want to change or support more UDP port.
856 struct rte_eth_udp_tunnel {
857 uint16_t udp_port; /**< UDP port used for the tunnel. */
858 uint8_t prot_type; /**< Tunnel type. Defined in rte_eth_tunnel_type. */
862 * A structure used to enable/disable specific device interrupts.
864 struct rte_intr_conf {
865 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
867 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
869 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
874 * A structure used to configure an Ethernet port.
875 * Depending upon the RX multi-queue mode, extra advanced
876 * configuration settings may be needed.
878 struct rte_eth_conf {
879 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
880 used. ETH_LINK_SPEED_FIXED disables link
881 autonegotiation, and a unique speed shall be
882 set. Otherwise, the bitmap defines the set of
883 speeds to be advertised. If the special value
884 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
885 supported are advertised. */
886 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
887 struct rte_eth_txmode txmode; /**< Port TX configuration. */
888 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
889 is 0, meaning the loopback mode is disabled.
890 Read the datasheet of given ethernet controller
891 for details. The possible values of this field
892 are defined in implementation of each driver. */
894 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
895 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
896 /**< Port vmdq+dcb configuration. */
897 struct rte_eth_dcb_rx_conf dcb_rx_conf;
898 /**< Port dcb RX configuration. */
899 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
900 /**< Port vmdq RX configuration. */
901 } rx_adv_conf; /**< Port RX filtering configuration. */
903 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
904 /**< Port vmdq+dcb TX configuration. */
905 struct rte_eth_dcb_tx_conf dcb_tx_conf;
906 /**< Port dcb TX configuration. */
907 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
908 /**< Port vmdq TX configuration. */
909 } tx_adv_conf; /**< Port TX DCB configuration (union). */
910 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
911 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
912 uint32_t dcb_capability_en;
913 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. */
914 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
918 * A structure used to retrieve the contextual information of
919 * an Ethernet device, such as the controlling driver of the device,
920 * its PCI context, etc...
924 * RX offload capabilities of a device.
926 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
927 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
928 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
929 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
930 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
931 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
932 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
933 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
934 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
935 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
936 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
937 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
938 #define DEV_RX_OFFLOAD_CRC_STRIP 0x00001000
939 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
940 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
941 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
942 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
943 DEV_RX_OFFLOAD_UDP_CKSUM | \
944 DEV_RX_OFFLOAD_TCP_CKSUM)
945 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
946 DEV_RX_OFFLOAD_VLAN_FILTER | \
947 DEV_RX_OFFLOAD_VLAN_EXTEND)
950 * If new Rx offload capabilities are defined, they also must be
951 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
955 * TX offload capabilities of a device.
957 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
958 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
959 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
960 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
961 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
962 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
963 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
964 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
965 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
966 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
967 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
968 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
969 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
970 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
971 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
972 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
973 * tx queue without SW lock.
975 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
976 /**< Device supports multi segment send. */
977 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
978 /**< Device supports optimization for fast release of mbufs.
979 * When set application must guarantee that per-queue all mbufs comes from
980 * the same mempool and has refcnt = 1.
982 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
984 * Device supports generic UDP tunneled packet TSO.
985 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
988 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
990 * Device supports generic IP tunneled packet TSO.
991 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
994 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
997 * If new Tx offload capabilities are defined, they also must be
998 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
1001 struct rte_pci_device;
1004 * Fallback default preferred Rx/Tx port parameters.
1005 * These are used if an application requests default parameters
1006 * but the PMD does not provide preferred values.
1008 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
1009 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
1010 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
1011 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
1014 * Preferred Rx/Tx port parameters.
1015 * There are separate instances of this structure for transmission
1016 * and reception respectively.
1018 struct rte_eth_dev_portconf {
1019 uint16_t burst_size; /**< Device-preferred burst size */
1020 uint16_t ring_size; /**< Device-preferred size of queue rings */
1021 uint16_t nb_queues; /**< Device-preferred number of queues */
1025 * Ethernet device information
1027 struct rte_eth_dev_info {
1028 struct rte_device *device; /** Generic device information */
1029 const char *driver_name; /**< Device Driver name. */
1030 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1031 Use if_indextoname() to translate into an interface name. */
1032 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1033 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1034 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1035 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1036 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1037 uint32_t max_hash_mac_addrs;
1038 /** Maximum number of hash MAC addresses for MTA and UTA. */
1039 uint16_t max_vfs; /**< Maximum number of VFs. */
1040 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1041 uint64_t rx_offload_capa;
1042 /**< Device per port RX offload capabilities. */
1043 uint64_t tx_offload_capa;
1044 /**< Device per port TX offload capabilities. */
1045 uint64_t rx_queue_offload_capa;
1046 /**< Device per queue RX offload capabilities. */
1047 uint64_t tx_queue_offload_capa;
1048 /**< Device per queue TX offload capabilities. */
1050 /**< Device redirection table size, the total number of entries. */
1051 uint8_t hash_key_size; /**< Hash key size in bytes */
1052 /** Bit mask of RSS offloads, the bit offset also means flow type */
1053 uint64_t flow_type_rss_offloads;
1054 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1055 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1056 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1057 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1058 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1059 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1060 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1061 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1062 /** Configured number of rx/tx queues */
1063 uint16_t nb_rx_queues; /**< Number of RX queues. */
1064 uint16_t nb_tx_queues; /**< Number of TX queues. */
1065 /** Rx parameter recommendations */
1066 struct rte_eth_dev_portconf default_rxportconf;
1067 /** Tx parameter recommendations */
1068 struct rte_eth_dev_portconf default_txportconf;
1072 * Ethernet device RX queue information structure.
1073 * Used to retieve information about configured queue.
1075 struct rte_eth_rxq_info {
1076 struct rte_mempool *mp; /**< mempool used by that queue. */
1077 struct rte_eth_rxconf conf; /**< queue config parameters. */
1078 uint8_t scattered_rx; /**< scattered packets RX supported. */
1079 uint16_t nb_desc; /**< configured number of RXDs. */
1080 } __rte_cache_min_aligned;
1083 * Ethernet device TX queue information structure.
1084 * Used to retrieve information about configured queue.
1086 struct rte_eth_txq_info {
1087 struct rte_eth_txconf conf; /**< queue config parameters. */
1088 uint16_t nb_desc; /**< configured number of TXDs. */
1089 } __rte_cache_min_aligned;
1091 /** Maximum name length for extended statistics counters */
1092 #define RTE_ETH_XSTATS_NAME_SIZE 64
1095 * An Ethernet device extended statistic structure
1097 * This structure is used by rte_eth_xstats_get() to provide
1098 * statistics that are not provided in the generic *rte_eth_stats*
1100 * It maps a name id, corresponding to an index in the array returned
1101 * by rte_eth_xstats_get_names(), to a statistic value.
1103 struct rte_eth_xstat {
1104 uint64_t id; /**< The index in xstats name array. */
1105 uint64_t value; /**< The statistic counter value. */
1109 * A name element for extended statistics.
1111 * An array of this structure is returned by rte_eth_xstats_get_names().
1112 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1113 * structure references these names by their array index.
1115 struct rte_eth_xstat_name {
1116 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1119 #define ETH_DCB_NUM_TCS 8
1120 #define ETH_MAX_VMDQ_POOL 64
1123 * A structure used to get the information of queue and
1124 * TC mapping on both TX and RX paths.
1126 struct rte_eth_dcb_tc_queue_mapping {
1127 /** rx queues assigned to tc per Pool */
1131 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1132 /** rx queues assigned to tc per Pool */
1136 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1140 * A structure used to get the information of DCB.
1141 * It includes TC UP mapping and queue TC mapping.
1143 struct rte_eth_dcb_info {
1144 uint8_t nb_tcs; /**< number of TCs */
1145 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1146 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1147 /** rx queues assigned to tc */
1148 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1152 * RX/TX queue states
1154 #define RTE_ETH_QUEUE_STATE_STOPPED 0
1155 #define RTE_ETH_QUEUE_STATE_STARTED 1
1157 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1159 /* Macros to check for valid port */
1160 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1161 if (!rte_eth_dev_is_valid_port(port_id)) { \
1162 RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
1167 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1168 if (!rte_eth_dev_is_valid_port(port_id)) { \
1169 RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
1175 * l2 tunnel configuration.
1178 /**< l2 tunnel enable mask */
1179 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1180 /**< l2 tunnel insertion mask */
1181 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1182 /**< l2 tunnel stripping mask */
1183 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1184 /**< l2 tunnel forwarding mask */
1185 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1188 * Function type used for RX packet processing packet callbacks.
1190 * The callback function is called on RX with a burst of packets that have
1191 * been received on the given port and queue.
1194 * The Ethernet port on which RX is being performed.
1196 * The queue on the Ethernet port which is being used to receive the packets.
1198 * The burst of packets that have just been received.
1200 * The number of packets in the burst pointed to by "pkts".
1202 * The max number of packets that can be stored in the "pkts" array.
1204 * The arbitrary user parameter passed in by the application when the callback
1205 * was originally configured.
1207 * The number of packets returned to the user.
1209 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1210 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1214 * Function type used for TX packet processing packet callbacks.
1216 * The callback function is called on TX with a burst of packets immediately
1217 * before the packets are put onto the hardware queue for transmission.
1220 * The Ethernet port on which TX is being performed.
1222 * The queue on the Ethernet port which is being used to transmit the packets.
1224 * The burst of packets that are about to be transmitted.
1226 * The number of packets in the burst pointed to by "pkts".
1228 * The arbitrary user parameter passed in by the application when the callback
1229 * was originally configured.
1231 * The number of packets to be written to the NIC.
1233 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1234 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1237 * A set of values to describe the possible states of an eth device.
1239 enum rte_eth_dev_state {
1240 RTE_ETH_DEV_UNUSED = 0,
1241 RTE_ETH_DEV_ATTACHED,
1242 RTE_ETH_DEV_DEFERRED,
1243 RTE_ETH_DEV_REMOVED,
1246 struct rte_eth_dev_sriov {
1247 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1248 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1249 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1250 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1252 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1254 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1256 #define RTE_ETH_DEV_NO_OWNER 0
1258 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1260 struct rte_eth_dev_owner {
1261 uint64_t id; /**< The owner unique identifier. */
1262 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1265 /** Device supports link state interrupt */
1266 #define RTE_ETH_DEV_INTR_LSC 0x0002
1267 /** Device is a bonded slave */
1268 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1269 /** Device supports device removal interrupt */
1270 #define RTE_ETH_DEV_INTR_RMV 0x0008
1274 * @b EXPERIMENTAL: this API may change without prior notice.
1276 * Iterates over valid ethdev ports owned by a specific owner.
1279 * The id of the next possible valid owned port.
1281 * The owner identifier.
1282 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1284 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1286 uint64_t __rte_experimental rte_eth_find_next_owned_by(uint16_t port_id,
1287 const uint64_t owner_id);
1290 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1292 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1293 for (p = rte_eth_find_next_owned_by(0, o); \
1294 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1295 p = rte_eth_find_next_owned_by(p + 1, o))
1298 * Iterates over valid ethdev ports.
1301 * The id of the next possible valid port.
1303 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1305 uint16_t rte_eth_find_next(uint16_t port_id);
1308 * Macro to iterate over all enabled and ownerless ethdev ports.
1310 #define RTE_ETH_FOREACH_DEV(p) \
1311 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1316 * @b EXPERIMENTAL: this API may change without prior notice.
1318 * Get a new unique owner identifier.
1319 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1320 * to avoid multiple management of device by different entities.
1323 * Owner identifier pointer.
1325 * Negative errno value on error, 0 on success.
1327 int __rte_experimental rte_eth_dev_owner_new(uint64_t *owner_id);
1331 * @b EXPERIMENTAL: this API may change without prior notice.
1333 * Set an Ethernet device owner.
1336 * The identifier of the port to own.
1338 * The owner pointer.
1340 * Negative errno value on error, 0 on success.
1342 int __rte_experimental rte_eth_dev_owner_set(const uint16_t port_id,
1343 const struct rte_eth_dev_owner *owner);
1347 * @b EXPERIMENTAL: this API may change without prior notice.
1349 * Unset Ethernet device owner to make the device ownerless.
1352 * The identifier of port to make ownerless.
1354 * The owner identifier.
1356 * 0 on success, negative errno value on error.
1358 int __rte_experimental rte_eth_dev_owner_unset(const uint16_t port_id,
1359 const uint64_t owner_id);
1363 * @b EXPERIMENTAL: this API may change without prior notice.
1365 * Remove owner from all Ethernet devices owned by a specific owner.
1368 * The owner identifier.
1370 void __rte_experimental rte_eth_dev_owner_delete(const uint64_t owner_id);
1374 * @b EXPERIMENTAL: this API may change without prior notice.
1376 * Get the owner of an Ethernet device.
1379 * The port identifier.
1381 * The owner structure pointer to fill.
1383 * 0 on success, negative errno value on error..
1385 int __rte_experimental rte_eth_dev_owner_get(const uint16_t port_id,
1386 struct rte_eth_dev_owner *owner);
1389 * Get the total number of Ethernet devices that have been successfully
1390 * initialized by the matching Ethernet driver during the PCI probing phase
1391 * and that are available for applications to use. These devices must be
1392 * accessed by using the ``RTE_ETH_FOREACH_DEV()`` macro to deal with
1393 * non-contiguous ranges of devices.
1394 * These non-contiguous ranges can be created by calls to hotplug functions or
1398 * - The total number of usable Ethernet devices.
1401 uint16_t rte_eth_dev_count(void);
1404 * Get the number of ports which are usable for the application.
1406 * These devices must be iterated by using the macro
1407 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
1408 * to deal with non-contiguous ranges of devices.
1411 * The count of available Ethernet devices.
1413 uint16_t __rte_experimental rte_eth_dev_count_avail(void);
1416 * Get the total number of ports which are allocated.
1418 * Some devices may not be available for the application.
1421 * The total count of Ethernet devices.
1423 uint16_t __rte_experimental rte_eth_dev_count_total(void);
1426 * Attach a new Ethernet device specified by arguments.
1429 * A pointer to a strings array describing the new device
1430 * to be attached. The strings should be a pci address like
1431 * '0000:01:00.0' or virtual device name like 'net_pcap0'.
1433 * A pointer to a port identifier actually attached.
1435 * 0 on success and port_id is filled, negative on error
1437 int rte_eth_dev_attach(const char *devargs, uint16_t *port_id);
1440 * Detach a Ethernet device specified by port identifier.
1441 * This function must be called when the device is in the
1445 * The port identifier of the device to detach.
1447 * A pointer to a buffer that will be filled with the device name.
1448 * This buffer must be at least RTE_DEV_NAME_MAX_LEN long.
1450 * 0 on success and devname is filled, negative on error
1452 int rte_eth_dev_detach(uint16_t port_id, char *devname);
1455 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
1456 * the bitmap link_speeds of the struct rte_eth_conf
1459 * Numerical speed value in Mbps
1461 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
1463 * 0 if the speed cannot be mapped
1465 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
1469 * @b EXPERIMENTAL: this API may change without prior notice
1471 * Get DEV_RX_OFFLOAD_* flag name.
1476 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1478 const char * __rte_experimental rte_eth_dev_rx_offload_name(uint64_t offload);
1482 * @b EXPERIMENTAL: this API may change without prior notice
1484 * Get DEV_TX_OFFLOAD_* flag name.
1489 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1491 const char * __rte_experimental rte_eth_dev_tx_offload_name(uint64_t offload);
1494 * Configure an Ethernet device.
1495 * This function must be invoked first before any other function in the
1496 * Ethernet API. This function can also be re-invoked when a device is in the
1500 * The port identifier of the Ethernet device to configure.
1501 * @param nb_rx_queue
1502 * The number of receive queues to set up for the Ethernet device.
1503 * @param nb_tx_queue
1504 * The number of transmit queues to set up for the Ethernet device.
1506 * The pointer to the configuration data to be used for the Ethernet device.
1507 * The *rte_eth_conf* structure includes:
1508 * - the hardware offload features to activate, with dedicated fields for
1509 * each statically configurable offload hardware feature provided by
1510 * Ethernet devices, such as IP checksum or VLAN tag stripping for
1512 * The Rx offload bitfield API is obsolete and will be deprecated.
1513 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
1514 * structure and use offloads field to set per-port offloads instead.
1515 * - the Receive Side Scaling (RSS) configuration when using multiple RX
1518 * Embedding all configuration information in a single data structure
1519 * is the more flexible method that allows the addition of new features
1520 * without changing the syntax of the API.
1522 * - 0: Success, device configured.
1523 * - <0: Error code returned by the driver configuration function.
1525 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
1526 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
1530 * @b EXPERIMENTAL: this API may change without prior notice.
1532 * Check if an Ethernet device was physically removed.
1535 * The port identifier of the Ethernet device.
1537 * 1 when the Ethernet device is removed, otherwise 0.
1539 int __rte_experimental
1540 rte_eth_dev_is_removed(uint16_t port_id);
1543 * Allocate and set up a receive queue for an Ethernet device.
1545 * The function allocates a contiguous block of memory for *nb_rx_desc*
1546 * receive descriptors from a memory zone associated with *socket_id*
1547 * and initializes each receive descriptor with a network buffer allocated
1548 * from the memory pool *mb_pool*.
1551 * The port identifier of the Ethernet device.
1552 * @param rx_queue_id
1553 * The index of the receive queue to set up.
1554 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1555 * to rte_eth_dev_configure().
1557 * The number of receive descriptors to allocate for the receive ring.
1559 * The *socket_id* argument is the socket identifier in case of NUMA.
1560 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1561 * the DMA memory allocated for the receive descriptors of the ring.
1563 * The pointer to the configuration data to be used for the receive queue.
1564 * NULL value is allowed, in which case default RX configuration
1566 * The *rx_conf* structure contains an *rx_thresh* structure with the values
1567 * of the Prefetch, Host, and Write-Back threshold registers of the receive
1569 * In addition it contains the hardware offloads features to activate using
1570 * the DEV_RX_OFFLOAD_* flags.
1572 * The pointer to the memory pool from which to allocate *rte_mbuf* network
1573 * memory buffers to populate each descriptor of the receive ring.
1575 * - 0: Success, receive queue correctly set up.
1576 * - -EIO: if device is removed.
1577 * - -EINVAL: The size of network buffers which can be allocated from the
1578 * memory pool does not fit the various buffer sizes allowed by the
1579 * device controller.
1580 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
1581 * allocate network memory buffers from the memory pool when
1582 * initializing receive descriptors.
1584 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
1585 uint16_t nb_rx_desc, unsigned int socket_id,
1586 const struct rte_eth_rxconf *rx_conf,
1587 struct rte_mempool *mb_pool);
1590 * Allocate and set up a transmit queue for an Ethernet device.
1593 * The port identifier of the Ethernet device.
1594 * @param tx_queue_id
1595 * The index of the transmit queue to set up.
1596 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1597 * to rte_eth_dev_configure().
1599 * The number of transmit descriptors to allocate for the transmit ring.
1601 * The *socket_id* argument is the socket identifier in case of NUMA.
1602 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1603 * the DMA memory allocated for the transmit descriptors of the ring.
1605 * The pointer to the configuration data to be used for the transmit queue.
1606 * NULL value is allowed, in which case default TX configuration
1608 * The *tx_conf* structure contains the following data:
1609 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
1610 * Write-Back threshold registers of the transmit ring.
1611 * When setting Write-Back threshold to the value greater then zero,
1612 * *tx_rs_thresh* value should be explicitly set to one.
1613 * - The *tx_free_thresh* value indicates the [minimum] number of network
1614 * buffers that must be pending in the transmit ring to trigger their
1615 * [implicit] freeing by the driver transmit function.
1616 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
1617 * descriptors that must be pending in the transmit ring before setting the
1618 * RS bit on a descriptor by the driver transmit function.
1619 * The *tx_rs_thresh* value should be less or equal then
1620 * *tx_free_thresh* value, and both of them should be less then
1622 * - The *txq_flags* member contains flags to pass to the TX queue setup
1623 * function to configure the behavior of the TX queue. This should be set
1624 * to 0 if no special configuration is required.
1625 * This API is obsolete and will be deprecated. Applications
1626 * should set it to ETH_TXQ_FLAGS_IGNORE and use
1627 * the offloads field below.
1628 * - The *offloads* member contains Tx offloads to be enabled.
1629 * Offloads which are not set cannot be used on the datapath.
1631 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
1632 * the transmit function to use default values.
1634 * - 0: Success, the transmit queue is correctly set up.
1635 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
1637 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
1638 uint16_t nb_tx_desc, unsigned int socket_id,
1639 const struct rte_eth_txconf *tx_conf);
1642 * Return the NUMA socket to which an Ethernet device is connected
1645 * The port identifier of the Ethernet device
1647 * The NUMA socket id to which the Ethernet device is connected or
1648 * a default of zero if the socket could not be determined.
1649 * -1 is returned is the port_id value is out of range.
1651 int rte_eth_dev_socket_id(uint16_t port_id);
1654 * Check if port_id of device is attached
1657 * The port identifier of the Ethernet device
1659 * - 0 if port is out of range or not attached
1660 * - 1 if device is attached
1662 int rte_eth_dev_is_valid_port(uint16_t port_id);
1665 * Start specified RX queue of a port. It is used when rx_deferred_start
1666 * flag of the specified queue is true.
1669 * The port identifier of the Ethernet device
1670 * @param rx_queue_id
1671 * The index of the rx queue to update the ring.
1672 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1673 * to rte_eth_dev_configure().
1675 * - 0: Success, the receive queue is started.
1676 * - -EINVAL: The port_id or the queue_id out of range.
1677 * - -EIO: if device is removed.
1678 * - -ENOTSUP: The function not supported in PMD driver.
1680 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
1683 * Stop specified RX queue of a port
1686 * The port identifier of the Ethernet device
1687 * @param rx_queue_id
1688 * The index of the rx queue to update the ring.
1689 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1690 * to rte_eth_dev_configure().
1692 * - 0: Success, the receive queue is stopped.
1693 * - -EINVAL: The port_id or the queue_id out of range.
1694 * - -EIO: if device is removed.
1695 * - -ENOTSUP: The function not supported in PMD driver.
1697 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
1700 * Start TX for specified queue of a port. It is used when tx_deferred_start
1701 * flag of the specified queue is true.
1704 * The port identifier of the Ethernet device
1705 * @param tx_queue_id
1706 * The index of the tx queue to update the ring.
1707 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1708 * to rte_eth_dev_configure().
1710 * - 0: Success, the transmit queue is started.
1711 * - -EINVAL: The port_id or the queue_id out of range.
1712 * - -EIO: if device is removed.
1713 * - -ENOTSUP: The function not supported in PMD driver.
1715 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
1718 * Stop specified TX queue of a port
1721 * The port identifier of the Ethernet device
1722 * @param tx_queue_id
1723 * The index of the tx queue to update the ring.
1724 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1725 * to rte_eth_dev_configure().
1727 * - 0: Success, the transmit queue is stopped.
1728 * - -EINVAL: The port_id or the queue_id out of range.
1729 * - -EIO: if device is removed.
1730 * - -ENOTSUP: The function not supported in PMD driver.
1732 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
1735 * Start an Ethernet device.
1737 * The device start step is the last one and consists of setting the configured
1738 * offload features and in starting the transmit and the receive units of the
1740 * On success, all basic functions exported by the Ethernet API (link status,
1741 * receive/transmit, and so on) can be invoked.
1744 * The port identifier of the Ethernet device.
1746 * - 0: Success, Ethernet device started.
1747 * - <0: Error code of the driver device start function.
1749 int rte_eth_dev_start(uint16_t port_id);
1752 * Stop an Ethernet device. The device can be restarted with a call to
1753 * rte_eth_dev_start()
1756 * The port identifier of the Ethernet device.
1758 void rte_eth_dev_stop(uint16_t port_id);
1761 * Link up an Ethernet device.
1763 * Set device link up will re-enable the device rx/tx
1764 * functionality after it is previously set device linked down.
1767 * The port identifier of the Ethernet device.
1769 * - 0: Success, Ethernet device linked up.
1770 * - <0: Error code of the driver device link up function.
1772 int rte_eth_dev_set_link_up(uint16_t port_id);
1775 * Link down an Ethernet device.
1776 * The device rx/tx functionality will be disabled if success,
1777 * and it can be re-enabled with a call to
1778 * rte_eth_dev_set_link_up()
1781 * The port identifier of the Ethernet device.
1783 int rte_eth_dev_set_link_down(uint16_t port_id);
1786 * Close a stopped Ethernet device. The device cannot be restarted!
1787 * The function frees all resources except for needed by the
1788 * closed state. To free these resources, call rte_eth_dev_detach().
1791 * The port identifier of the Ethernet device.
1793 void rte_eth_dev_close(uint16_t port_id);
1796 * Reset a Ethernet device and keep its port id.
1798 * When a port has to be reset passively, the DPDK application can invoke
1799 * this function. For example when a PF is reset, all its VFs should also
1800 * be reset. Normally a DPDK application can invoke this function when
1801 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
1802 * a port reset in other circumstances.
1804 * When this function is called, it first stops the port and then calls the
1805 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
1806 * state, in which no Tx and Rx queues are setup, as if the port has been
1807 * reset and not started. The port keeps the port id it had before the
1810 * After calling rte_eth_dev_reset( ), the application should use
1811 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
1812 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
1813 * to reconfigure the device as appropriate.
1815 * Note: To avoid unexpected behavior, the application should stop calling
1816 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
1817 * safety, all these controlling functions should be called from the same
1821 * The port identifier of the Ethernet device.
1824 * - (0) if successful.
1825 * - (-EINVAL) if port identifier is invalid.
1826 * - (-ENOTSUP) if hardware doesn't support this function.
1827 * - (-EPERM) if not ran from the primary process.
1828 * - (-EIO) if re-initialisation failed or device is removed.
1829 * - (-ENOMEM) if the reset failed due to OOM.
1830 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
1832 int rte_eth_dev_reset(uint16_t port_id);
1835 * Enable receipt in promiscuous mode for an Ethernet device.
1838 * The port identifier of the Ethernet device.
1840 void rte_eth_promiscuous_enable(uint16_t port_id);
1843 * Disable receipt in promiscuous mode for an Ethernet device.
1846 * The port identifier of the Ethernet device.
1848 void rte_eth_promiscuous_disable(uint16_t port_id);
1851 * Return the value of promiscuous mode for an Ethernet device.
1854 * The port identifier of the Ethernet device.
1856 * - (1) if promiscuous is enabled
1857 * - (0) if promiscuous is disabled.
1860 int rte_eth_promiscuous_get(uint16_t port_id);
1863 * Enable the receipt of any multicast frame by an Ethernet device.
1866 * The port identifier of the Ethernet device.
1868 void rte_eth_allmulticast_enable(uint16_t port_id);
1871 * Disable the receipt of all multicast frames by an Ethernet device.
1874 * The port identifier of the Ethernet device.
1876 void rte_eth_allmulticast_disable(uint16_t port_id);
1879 * Return the value of allmulticast mode for an Ethernet device.
1882 * The port identifier of the Ethernet device.
1884 * - (1) if allmulticast is enabled
1885 * - (0) if allmulticast is disabled.
1888 int rte_eth_allmulticast_get(uint16_t port_id);
1891 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1892 * or FULL-DUPLEX) of the physical link of an Ethernet device. It might need
1893 * to wait up to 9 seconds in it.
1896 * The port identifier of the Ethernet device.
1898 * A pointer to an *rte_eth_link* structure to be filled with
1899 * the status, the speed and the mode of the Ethernet device link.
1901 void rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
1904 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1905 * or FULL-DUPLEX) of the physical link of an Ethernet device. It is a no-wait
1906 * version of rte_eth_link_get().
1909 * The port identifier of the Ethernet device.
1911 * A pointer to an *rte_eth_link* structure to be filled with
1912 * the status, the speed and the mode of the Ethernet device link.
1914 void rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
1917 * Retrieve the general I/O statistics of an Ethernet device.
1920 * The port identifier of the Ethernet device.
1922 * A pointer to a structure of type *rte_eth_stats* to be filled with
1923 * the values of device counters for the following set of statistics:
1924 * - *ipackets* with the total of successfully received packets.
1925 * - *opackets* with the total of successfully transmitted packets.
1926 * - *ibytes* with the total of successfully received bytes.
1927 * - *obytes* with the total of successfully transmitted bytes.
1928 * - *ierrors* with the total of erroneous received packets.
1929 * - *oerrors* with the total of failed transmitted packets.
1931 * Zero if successful. Non-zero otherwise.
1933 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
1936 * Reset the general I/O statistics of an Ethernet device.
1939 * The port identifier of the Ethernet device.
1941 * - (0) if device notified to reset stats.
1942 * - (-ENOTSUP) if hardware doesn't support.
1943 * - (-ENODEV) if *port_id* invalid.
1945 int rte_eth_stats_reset(uint16_t port_id);
1948 * Retrieve names of extended statistics of an Ethernet device.
1951 * The port identifier of the Ethernet device.
1952 * @param xstats_names
1953 * An rte_eth_xstat_name array of at least *size* elements to
1954 * be filled. If set to NULL, the function returns the required number
1957 * The size of the xstats_names array (number of elements).
1959 * - A positive value lower or equal to size: success. The return value
1960 * is the number of entries filled in the stats table.
1961 * - A positive value higher than size: error, the given statistics table
1962 * is too small. The return value corresponds to the size that should
1963 * be given to succeed. The entries in the table are not valid and
1964 * shall not be used by the caller.
1965 * - A negative value on error (invalid port id).
1967 int rte_eth_xstats_get_names(uint16_t port_id,
1968 struct rte_eth_xstat_name *xstats_names,
1972 * Retrieve extended statistics of an Ethernet device.
1975 * The port identifier of the Ethernet device.
1977 * A pointer to a table of structure of type *rte_eth_xstat*
1978 * to be filled with device statistics ids and values: id is the
1979 * index of the name string in xstats_names (see rte_eth_xstats_get_names()),
1980 * and value is the statistic counter.
1981 * This parameter can be set to NULL if n is 0.
1983 * The size of the xstats array (number of elements).
1985 * - A positive value lower or equal to n: success. The return value
1986 * is the number of entries filled in the stats table.
1987 * - A positive value higher than n: error, the given statistics table
1988 * is too small. The return value corresponds to the size that should
1989 * be given to succeed. The entries in the table are not valid and
1990 * shall not be used by the caller.
1991 * - A negative value on error (invalid port id).
1993 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
1997 * Retrieve names of extended statistics of an Ethernet device.
2000 * The port identifier of the Ethernet device.
2001 * @param xstats_names
2002 * An rte_eth_xstat_name array of at least *size* elements to
2003 * be filled. If set to NULL, the function returns the required number
2006 * IDs array given by app to retrieve specific statistics
2008 * The size of the xstats_names array (number of elements).
2010 * - A positive value lower or equal to size: success. The return value
2011 * is the number of entries filled in the stats table.
2012 * - A positive value higher than size: error, the given statistics table
2013 * is too small. The return value corresponds to the size that should
2014 * be given to succeed. The entries in the table are not valid and
2015 * shall not be used by the caller.
2016 * - A negative value on error (invalid port id).
2019 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2020 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2024 * Retrieve extended statistics of an Ethernet device.
2027 * The port identifier of the Ethernet device.
2029 * A pointer to an ids array passed by application. This tells which
2030 * statistics values function should retrieve. This parameter
2031 * can be set to NULL if size is 0. In this case function will retrieve
2032 * all avalible statistics.
2034 * A pointer to a table to be filled with device statistics values.
2036 * The size of the ids 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_by_id(uint16_t port_id, const uint64_t *ids,
2047 uint64_t *values, unsigned int size);
2050 * Gets the ID of a statistic from its name.
2052 * This function searches for the statistics using string compares, and
2053 * as such should not be used on the fast-path. For fast-path retrieval of
2054 * specific statistics, store the ID as provided in *id* from this function,
2055 * and pass the ID to rte_eth_xstats_get()
2057 * @param port_id The port to look up statistics from
2058 * @param xstat_name The name of the statistic to return
2059 * @param[out] id A pointer to an app-supplied uint64_t which should be
2060 * set to the ID of the stat if the stat exists.
2063 * -ENODEV for invalid port_id,
2064 * -EIO if device is removed,
2065 * -EINVAL if the xstat_name doesn't exist in port_id
2067 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2071 * Reset extended statistics of an Ethernet device.
2074 * The port identifier of the Ethernet device.
2076 void rte_eth_xstats_reset(uint16_t port_id);
2079 * Set a mapping for the specified transmit queue to the specified per-queue
2080 * statistics counter.
2083 * The port identifier of the Ethernet device.
2084 * @param tx_queue_id
2085 * The index of the transmit queue for which a queue stats mapping is required.
2086 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2087 * to rte_eth_dev_configure().
2089 * The per-queue packet statistics functionality number that the transmit
2090 * queue is to be assigned.
2091 * The value must be in the range [0, RTE_MAX_ETHPORT_QUEUE_STATS_MAPS - 1].
2093 * Zero if successful. Non-zero otherwise.
2095 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2096 uint16_t tx_queue_id, uint8_t stat_idx);
2099 * Set a mapping for the specified receive queue to the specified per-queue
2100 * statistics counter.
2103 * The port identifier of the Ethernet device.
2104 * @param rx_queue_id
2105 * The index of the receive queue for which a queue stats mapping is required.
2106 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2107 * to rte_eth_dev_configure().
2109 * The per-queue packet statistics functionality number that the receive
2110 * queue is to be assigned.
2111 * The value must be in the range [0, RTE_MAX_ETHPORT_QUEUE_STATS_MAPS - 1].
2113 * Zero if successful. Non-zero otherwise.
2115 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2116 uint16_t rx_queue_id,
2120 * Retrieve the Ethernet address of an Ethernet device.
2123 * The port identifier of the Ethernet device.
2125 * A pointer to a structure of type *ether_addr* to be filled with
2126 * the Ethernet address of the Ethernet device.
2128 void rte_eth_macaddr_get(uint16_t port_id, struct ether_addr *mac_addr);
2131 * Retrieve the contextual information of an Ethernet device.
2134 * The port identifier of the Ethernet device.
2136 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
2137 * the contextual information of the Ethernet device.
2139 void rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
2142 * Retrieve the firmware version of a device.
2145 * The port identifier of the device.
2147 * A pointer to a string array storing the firmware version of a device,
2148 * the string includes terminating null. This pointer is allocated by caller.
2150 * The size of the string array pointed by fw_version, which should be
2151 * large enough to store firmware version of the device.
2153 * - (0) if successful.
2154 * - (-ENOTSUP) if operation is not supported.
2155 * - (-ENODEV) if *port_id* invalid.
2156 * - (-EIO) if device is removed.
2157 * - (>0) if *fw_size* is not enough to store firmware version, return
2158 * the size of the non truncated string.
2160 int rte_eth_dev_fw_version_get(uint16_t port_id,
2161 char *fw_version, size_t fw_size);
2164 * Retrieve the supported packet types of an Ethernet device.
2166 * When a packet type is announced as supported, it *must* be recognized by
2167 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
2168 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
2169 * packet types for these packets:
2170 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
2171 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
2172 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
2173 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
2175 * When a packet is received by a PMD, the most precise type must be
2176 * returned among the ones supported. However a PMD is allowed to set
2177 * packet type that is not in the supported list, at the condition that it
2178 * is more precise. Therefore, a PMD announcing no supported packet types
2179 * can still set a matching packet type in a received packet.
2182 * Better to invoke this API after the device is already started or rx burst
2183 * function is decided, to obtain correct supported ptypes.
2185 * if a given PMD does not report what ptypes it supports, then the supported
2186 * ptype count is reported as 0.
2188 * The port identifier of the Ethernet device.
2190 * A hint of what kind of packet type which the caller is interested in.
2192 * An array pointer to store adequate packet types, allocated by caller.
2194 * Size of the array pointed by param ptypes.
2196 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
2197 * only num entries will be filled into the ptypes array, but the full
2198 * count of supported ptypes will be returned.
2199 * - (-ENODEV) if *port_id* invalid.
2201 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
2202 uint32_t *ptypes, int num);
2205 * Retrieve the MTU of an Ethernet device.
2208 * The port identifier of the Ethernet device.
2210 * A pointer to a uint16_t where the retrieved MTU is to be stored.
2212 * - (0) if successful.
2213 * - (-ENODEV) if *port_id* invalid.
2215 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
2218 * Change the MTU of an Ethernet device.
2221 * The port identifier of the Ethernet device.
2223 * A uint16_t for the MTU to be applied.
2225 * - (0) if successful.
2226 * - (-ENOTSUP) if operation is not supported.
2227 * - (-ENODEV) if *port_id* invalid.
2228 * - (-EIO) if device is removed.
2229 * - (-EINVAL) if *mtu* invalid.
2230 * - (-EBUSY) if operation is not allowed when the port is running
2232 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
2235 * Enable/Disable hardware filtering by an Ethernet device of received
2236 * VLAN packets tagged with a given VLAN Tag Identifier.
2239 * The port identifier of the Ethernet device.
2241 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
2243 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
2244 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
2246 * - (0) if successful.
2247 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2248 * - (-ENODEV) if *port_id* invalid.
2249 * - (-EIO) if device is removed.
2250 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
2251 * - (-EINVAL) if *vlan_id* > 4095.
2253 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
2256 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
2257 * 82599/X540/X550 can support VLAN stripping at the rx queue level
2260 * The port identifier of the Ethernet device.
2261 * @param rx_queue_id
2262 * The index of the receive queue for which a queue stats mapping is required.
2263 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2264 * to rte_eth_dev_configure().
2266 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
2267 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
2269 * - (0) if successful.
2270 * - (-ENOSUP) if hardware-assisted VLAN stripping not configured.
2271 * - (-ENODEV) if *port_id* invalid.
2272 * - (-EINVAL) if *rx_queue_id* invalid.
2274 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
2278 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
2279 * the VLAN Header. This is a register setup available on some Intel NIC, not
2280 * but all, please check the data sheet for availability.
2283 * The port identifier of the Ethernet device.
2287 * The Tag Protocol ID
2289 * - (0) if successful.
2290 * - (-ENOSUP) if hardware-assisted VLAN TPID setup is not supported.
2291 * - (-ENODEV) if *port_id* invalid.
2292 * - (-EIO) if device is removed.
2294 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
2295 enum rte_vlan_type vlan_type,
2299 * Set VLAN offload configuration on an Ethernet device
2300 * Enable/Disable Extended VLAN by an Ethernet device, This is a register setup
2301 * available on some Intel NIC, not but all, please check the data sheet for
2303 * Enable/Disable VLAN Strip can be done on rx queue for certain NIC, but here
2304 * the configuration is applied on the port level.
2307 * The port identifier of the Ethernet device.
2308 * @param offload_mask
2309 * The VLAN Offload bit mask can be mixed use with "OR"
2310 * ETH_VLAN_STRIP_OFFLOAD
2311 * ETH_VLAN_FILTER_OFFLOAD
2312 * ETH_VLAN_EXTEND_OFFLOAD
2314 * - (0) if successful.
2315 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2316 * - (-ENODEV) if *port_id* invalid.
2317 * - (-EIO) if device is removed.
2319 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
2322 * Read VLAN Offload configuration from an Ethernet device
2325 * The port identifier of the Ethernet device.
2327 * - (>0) if successful. Bit mask to indicate
2328 * ETH_VLAN_STRIP_OFFLOAD
2329 * ETH_VLAN_FILTER_OFFLOAD
2330 * ETH_VLAN_EXTEND_OFFLOAD
2331 * - (-ENODEV) if *port_id* invalid.
2333 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
2336 * Set port based TX VLAN insertion on or off.
2339 * The port identifier of the Ethernet device.
2341 * Port based TX VLAN identifier together with user priority.
2343 * Turn on or off the port based TX VLAN insertion.
2346 * - (0) if successful.
2347 * - negative if failed.
2349 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
2351 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
2355 * Structure used to buffer packets for future TX
2356 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
2358 struct rte_eth_dev_tx_buffer {
2359 buffer_tx_error_fn error_callback;
2360 void *error_userdata;
2361 uint16_t size; /**< Size of buffer for buffered tx */
2362 uint16_t length; /**< Number of packets in the array */
2363 struct rte_mbuf *pkts[];
2364 /**< Pending packets to be sent on explicit flush or when full */
2368 * Calculate the size of the tx buffer.
2371 * Number of stored packets.
2373 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
2374 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
2377 * Initialize default values for buffered transmitting
2380 * Tx buffer to be initialized.
2387 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
2390 * Configure a callback for buffered packets which cannot be sent
2392 * Register a specific callback to be called when an attempt is made to send
2393 * all packets buffered on an ethernet port, but not all packets can
2394 * successfully be sent. The callback registered here will be called only
2395 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
2396 * The default callback configured for each queue by default just frees the
2397 * packets back to the calling mempool. If additional behaviour is required,
2398 * for example, to count dropped packets, or to retry transmission of packets
2399 * which cannot be sent, this function should be used to register a suitable
2400 * callback function to implement the desired behaviour.
2401 * The example callback "rte_eth_count_unsent_packet_callback()" is also
2402 * provided as reference.
2405 * The port identifier of the Ethernet device.
2407 * The function to be used as the callback.
2409 * Arbitrary parameter to be passed to the callback function
2411 * 0 on success, or -1 on error with rte_errno set appropriately
2414 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
2415 buffer_tx_error_fn callback, void *userdata);
2418 * Callback function for silently dropping unsent buffered packets.
2420 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2421 * adjust the default behavior when buffered packets cannot be sent. This
2422 * function drops any unsent packets silently and is used by tx buffered
2423 * operations as default behavior.
2425 * NOTE: this function should not be called directly, instead it should be used
2426 * as a callback for packet buffering.
2428 * NOTE: when configuring this function as a callback with
2429 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2430 * should point to an uint64_t value.
2433 * The previously buffered packets which could not be sent
2435 * The number of unsent packets in the pkts array
2440 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
2444 * Callback function for tracking unsent buffered packets.
2446 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2447 * adjust the default behavior when buffered packets cannot be sent. This
2448 * function drops any unsent packets, but also updates a user-supplied counter
2449 * to track the overall number of packets dropped. The counter should be an
2450 * uint64_t variable.
2452 * NOTE: this function should not be called directly, instead it should be used
2453 * as a callback for packet buffering.
2455 * NOTE: when configuring this function as a callback with
2456 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2457 * should point to an uint64_t value.
2460 * The previously buffered packets which could not be sent
2462 * The number of unsent packets in the pkts array
2464 * Pointer to an uint64_t value, which will be incremented by unsent
2467 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
2471 * Request the driver to free mbufs currently cached by the driver. The
2472 * driver will only free the mbuf if it is no longer in use. It is the
2473 * application's responsibity to ensure rte_eth_tx_buffer_flush(..) is
2477 * The port identifier of the Ethernet device.
2479 * The index of the transmit queue through which output packets must be
2481 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2482 * to rte_eth_dev_configure().
2484 * Maximum number of packets to free. Use 0 to indicate all possible packets
2485 * should be freed. Note that a packet may be using multiple mbufs.
2488 * -ENODEV: Invalid interface
2489 * -EIO: device is removed
2490 * -ENOTSUP: Driver does not support function
2492 * 0-n: Number of packets freed. More packets may still remain in ring that
2496 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
2499 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
2502 enum rte_eth_event_ipsec_subtype {
2503 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
2504 /**< Unknown event type */
2505 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
2506 /**< Sequence number overflow */
2507 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
2508 /**< Soft time expiry of SA */
2509 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
2510 /**< Soft byte expiry of SA */
2511 RTE_ETH_EVENT_IPSEC_MAX
2512 /**< Max value of this enum */
2516 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
2517 * information of the IPsec offload event.
2519 struct rte_eth_event_ipsec_desc {
2520 enum rte_eth_event_ipsec_subtype subtype;
2521 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
2523 /**< Event specific metadata
2525 * For the following events, *userdata* registered
2526 * with the *rte_security_session* would be returned
2529 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
2530 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
2531 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
2533 * @see struct rte_security_session_conf
2539 * The eth device event type for interrupt, and maybe others in the future.
2541 enum rte_eth_event_type {
2542 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
2543 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
2544 RTE_ETH_EVENT_QUEUE_STATE,
2545 /**< queue state event (enabled/disabled) */
2546 RTE_ETH_EVENT_INTR_RESET,
2547 /**< reset interrupt event, sent to VF on PF reset */
2548 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
2549 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
2550 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
2551 RTE_ETH_EVENT_NEW, /**< port is probed */
2552 RTE_ETH_EVENT_DESTROY, /**< port is released */
2553 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
2554 RTE_ETH_EVENT_MAX /**< max value of this enum */
2557 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
2558 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
2559 /**< user application callback to be registered for interrupts */
2562 * Register a callback function for port event.
2566 * RTE_ETH_ALL means register the event for all port ids.
2570 * User supplied callback function to be called.
2572 * Pointer to the parameters for the registered callback.
2575 * - On success, zero.
2576 * - On failure, a negative value.
2578 int rte_eth_dev_callback_register(uint16_t port_id,
2579 enum rte_eth_event_type event,
2580 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2583 * Unregister a callback function for port event.
2587 * RTE_ETH_ALL means unregister the event for all port ids.
2591 * User supplied callback function to be called.
2593 * Pointer to the parameters for the registered callback. -1 means to
2594 * remove all for the same callback address and same event.
2597 * - On success, zero.
2598 * - On failure, a negative value.
2600 int rte_eth_dev_callback_unregister(uint16_t port_id,
2601 enum rte_eth_event_type event,
2602 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2605 * When there is no rx packet coming in Rx Queue for a long time, we can
2606 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
2607 * to be triggered when Rx packet arrives.
2609 * The rte_eth_dev_rx_intr_enable() function enables rx queue
2610 * interrupt on specific rx queue of a port.
2613 * The port identifier of the Ethernet device.
2615 * The index of the receive queue from which to retrieve input packets.
2616 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2617 * to rte_eth_dev_configure().
2619 * - (0) if successful.
2620 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2622 * - (-ENODEV) if *port_id* invalid.
2623 * - (-EIO) if device is removed.
2625 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
2628 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
2629 * interrupt and returns to polling mode.
2631 * The rte_eth_dev_rx_intr_disable() function disables rx queue
2632 * interrupt on specific rx queue of a port.
2635 * The port identifier of the Ethernet device.
2637 * The index of the receive queue from which to retrieve input packets.
2638 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2639 * to rte_eth_dev_configure().
2641 * - (0) if successful.
2642 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2644 * - (-ENODEV) if *port_id* invalid.
2645 * - (-EIO) if device is removed.
2647 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
2650 * RX Interrupt control per port.
2653 * The port identifier of the Ethernet device.
2655 * Epoll instance fd which the intr vector associated to.
2656 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2658 * The operation be performed for the vector.
2659 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2663 * - On success, zero.
2664 * - On failure, a negative value.
2666 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
2669 * RX Interrupt control per queue.
2672 * The port identifier of the Ethernet device.
2674 * The index of the receive queue from which to retrieve input packets.
2675 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2676 * to rte_eth_dev_configure().
2678 * Epoll instance fd which the intr vector associated to.
2679 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2681 * The operation be performed for the vector.
2682 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2686 * - On success, zero.
2687 * - On failure, a negative value.
2689 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
2690 int epfd, int op, void *data);
2693 * Turn on the LED on the Ethernet device.
2694 * This function turns on the LED on the Ethernet device.
2697 * The port identifier of the Ethernet device.
2699 * - (0) if successful.
2700 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2702 * - (-ENODEV) if *port_id* invalid.
2703 * - (-EIO) if device is removed.
2705 int rte_eth_led_on(uint16_t port_id);
2708 * Turn off the LED on the Ethernet device.
2709 * This function turns off the LED on the Ethernet device.
2712 * The port identifier of the Ethernet device.
2714 * - (0) if successful.
2715 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2717 * - (-ENODEV) if *port_id* invalid.
2718 * - (-EIO) if device is removed.
2720 int rte_eth_led_off(uint16_t port_id);
2723 * Get current status of the Ethernet link flow control for Ethernet device
2726 * The port identifier of the Ethernet device.
2728 * The pointer to the structure where to store the flow control parameters.
2730 * - (0) if successful.
2731 * - (-ENOTSUP) if hardware doesn't support flow control.
2732 * - (-ENODEV) if *port_id* invalid.
2733 * - (-EIO) if device is removed.
2735 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
2736 struct rte_eth_fc_conf *fc_conf);
2739 * Configure the Ethernet link flow control for Ethernet device
2742 * The port identifier of the Ethernet device.
2744 * The pointer to the structure of the flow control parameters.
2746 * - (0) if successful.
2747 * - (-ENOTSUP) if hardware doesn't support flow control mode.
2748 * - (-ENODEV) if *port_id* invalid.
2749 * - (-EINVAL) if bad parameter
2750 * - (-EIO) if flow control setup failure or device is removed.
2752 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
2753 struct rte_eth_fc_conf *fc_conf);
2756 * Configure the Ethernet priority flow control under DCB environment
2757 * for Ethernet device.
2760 * The port identifier of the Ethernet device.
2762 * The pointer to the structure of the priority flow control parameters.
2764 * - (0) if successful.
2765 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
2766 * - (-ENODEV) if *port_id* invalid.
2767 * - (-EINVAL) if bad parameter
2768 * - (-EIO) if flow control setup failure or device is removed.
2770 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
2771 struct rte_eth_pfc_conf *pfc_conf);
2774 * Add a MAC address to an internal array of addresses used to enable whitelist
2775 * filtering to accept packets only if the destination MAC address matches.
2778 * The port identifier of the Ethernet device.
2780 * The MAC address to add.
2782 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
2783 * not enabled, this should be set to 0.
2785 * - (0) if successfully added or *mac_addr* was already added.
2786 * - (-ENOTSUP) if hardware doesn't support this feature.
2787 * - (-ENODEV) if *port* is invalid.
2788 * - (-EIO) if device is removed.
2789 * - (-ENOSPC) if no more MAC addresses can be added.
2790 * - (-EINVAL) if MAC address is invalid.
2792 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct ether_addr *mac_addr,
2796 * Remove a MAC address from the internal array of addresses.
2799 * The port identifier of the Ethernet device.
2801 * MAC address to remove.
2803 * - (0) if successful, or *mac_addr* didn't exist.
2804 * - (-ENOTSUP) if hardware doesn't support.
2805 * - (-ENODEV) if *port* invalid.
2806 * - (-EADDRINUSE) if attempting to remove the default MAC address
2808 int rte_eth_dev_mac_addr_remove(uint16_t port_id, struct ether_addr *mac_addr);
2811 * Set the default MAC address.
2814 * The port identifier of the Ethernet device.
2816 * New default MAC address.
2818 * - (0) if successful, or *mac_addr* didn't exist.
2819 * - (-ENOTSUP) if hardware doesn't support.
2820 * - (-ENODEV) if *port* invalid.
2821 * - (-EINVAL) if MAC address is invalid.
2823 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
2824 struct ether_addr *mac_addr);
2827 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2830 * The port identifier of the Ethernet device.
2834 * Redirection table size. The table size can be queried by
2835 * rte_eth_dev_info_get().
2837 * - (0) if successful.
2838 * - (-ENOTSUP) if hardware doesn't support.
2839 * - (-EINVAL) if bad parameter.
2840 * - (-EIO) if device is removed.
2842 int rte_eth_dev_rss_reta_update(uint16_t port_id,
2843 struct rte_eth_rss_reta_entry64 *reta_conf,
2844 uint16_t reta_size);
2847 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2850 * The port identifier of the Ethernet device.
2854 * Redirection table size. The table size can be queried by
2855 * rte_eth_dev_info_get().
2857 * - (0) if successful.
2858 * - (-ENOTSUP) if hardware doesn't support.
2859 * - (-EINVAL) if bad parameter.
2860 * - (-EIO) if device is removed.
2862 int rte_eth_dev_rss_reta_query(uint16_t port_id,
2863 struct rte_eth_rss_reta_entry64 *reta_conf,
2864 uint16_t reta_size);
2867 * Updates unicast hash table for receiving packet with the given destination
2868 * MAC address, and the packet is routed to all VFs for which the RX mode is
2869 * accept packets that match the unicast hash table.
2872 * The port identifier of the Ethernet device.
2874 * Unicast MAC address.
2876 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
2877 * 0 - Clear an unicast hash bit.
2879 * - (0) if successful.
2880 * - (-ENOTSUP) if hardware doesn't support.
2881 * - (-ENODEV) if *port_id* invalid.
2882 * - (-EIO) if device is removed.
2883 * - (-EINVAL) if bad parameter.
2885 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct ether_addr *addr,
2889 * Updates all unicast hash bitmaps for receiving packet with any Unicast
2890 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
2891 * mode is accept packets that match the unicast hash table.
2894 * The port identifier of the Ethernet device.
2896 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
2898 * 0 - Clear all unicast hash bitmaps
2900 * - (0) if successful.
2901 * - (-ENOTSUP) if hardware doesn't support.
2902 * - (-ENODEV) if *port_id* invalid.
2903 * - (-EIO) if device is removed.
2904 * - (-EINVAL) if bad parameter.
2906 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
2909 * Set a traffic mirroring rule on an Ethernet device
2912 * The port identifier of the Ethernet device.
2913 * @param mirror_conf
2914 * The pointer to the traffic mirroring structure describing the mirroring rule.
2915 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
2916 * destination pool and the value of rule if enable vlan or pool mirroring.
2919 * The index of traffic mirroring rule, we support four separated rules.
2921 * 1 - Enable a mirroring rule.
2922 * 0 - Disable a mirroring rule.
2924 * - (0) if successful.
2925 * - (-ENOTSUP) if hardware doesn't support this feature.
2926 * - (-ENODEV) if *port_id* invalid.
2927 * - (-EIO) if device is removed.
2928 * - (-EINVAL) if the mr_conf information is not correct.
2930 int rte_eth_mirror_rule_set(uint16_t port_id,
2931 struct rte_eth_mirror_conf *mirror_conf,
2936 * Reset a traffic mirroring rule on an Ethernet device.
2939 * The port identifier of the Ethernet device.
2941 * The index of traffic mirroring rule, we support four separated rules.
2943 * - (0) if successful.
2944 * - (-ENOTSUP) if hardware doesn't support this feature.
2945 * - (-ENODEV) if *port_id* invalid.
2946 * - (-EIO) if device is removed.
2947 * - (-EINVAL) if bad parameter.
2949 int rte_eth_mirror_rule_reset(uint16_t port_id,
2953 * Set the rate limitation for a queue on an Ethernet device.
2956 * The port identifier of the Ethernet device.
2960 * The tx rate in Mbps. Allocated from the total port link speed.
2962 * - (0) if successful.
2963 * - (-ENOTSUP) if hardware doesn't support this feature.
2964 * - (-ENODEV) if *port_id* invalid.
2965 * - (-EIO) if device is removed.
2966 * - (-EINVAL) if bad parameter.
2968 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
2972 * Configuration of Receive Side Scaling hash computation of Ethernet device.
2975 * The port identifier of the Ethernet device.
2977 * The new configuration to use for RSS hash computation on the port.
2979 * - (0) if successful.
2980 * - (-ENODEV) if port identifier is invalid.
2981 * - (-EIO) if device is removed.
2982 * - (-ENOTSUP) if hardware doesn't support.
2983 * - (-EINVAL) if bad parameter.
2985 int rte_eth_dev_rss_hash_update(uint16_t port_id,
2986 struct rte_eth_rss_conf *rss_conf);
2989 * Retrieve current configuration of Receive Side Scaling hash computation
2990 * of Ethernet device.
2993 * The port identifier of the Ethernet device.
2995 * Where to store the current RSS hash configuration of the Ethernet device.
2997 * - (0) if successful.
2998 * - (-ENODEV) if port identifier is invalid.
2999 * - (-EIO) if device is removed.
3000 * - (-ENOTSUP) if hardware doesn't support RSS.
3003 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
3004 struct rte_eth_rss_conf *rss_conf);
3007 * Add UDP tunneling port for a specific type of tunnel.
3008 * The packets with this UDP port will be identified as this type of tunnel.
3009 * Before enabling any offloading function for a tunnel, users can call this API
3010 * to change or add more UDP port for the tunnel. So the offloading function
3011 * can take effect on the packets with the specific UDP port.
3014 * The port identifier of the Ethernet device.
3016 * UDP tunneling configuration.
3019 * - (0) if successful.
3020 * - (-ENODEV) if port identifier is invalid.
3021 * - (-EIO) if device is removed.
3022 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3025 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
3026 struct rte_eth_udp_tunnel *tunnel_udp);
3029 * Delete UDP tunneling port a specific type of tunnel.
3030 * The packets with this UDP port will not be identified as this type of tunnel
3032 * Before enabling any offloading function for a tunnel, users can call this API
3033 * to delete a UDP port for the tunnel. So the offloading function will not take
3034 * effect on the packets with the specific UDP port.
3037 * The port identifier of the Ethernet device.
3039 * UDP tunneling configuration.
3042 * - (0) if successful.
3043 * - (-ENODEV) if port identifier is invalid.
3044 * - (-EIO) if device is removed.
3045 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3048 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
3049 struct rte_eth_udp_tunnel *tunnel_udp);
3052 * Check whether the filter type is supported on an Ethernet device.
3053 * All the supported filter types are defined in 'rte_eth_ctrl.h'.
3056 * The port identifier of the Ethernet device.
3057 * @param filter_type
3060 * - (0) if successful.
3061 * - (-ENOTSUP) if hardware doesn't support this filter type.
3062 * - (-ENODEV) if *port_id* invalid.
3063 * - (-EIO) if device is removed.
3065 int rte_eth_dev_filter_supported(uint16_t port_id,
3066 enum rte_filter_type filter_type);
3069 * Take operations to assigned filter type on an Ethernet device.
3070 * All the supported operations and filter types are defined in 'rte_eth_ctrl.h'.
3073 * The port identifier of the Ethernet device.
3074 * @param filter_type
3077 * Type of operation.
3079 * A pointer to arguments defined specifically for the operation.
3081 * - (0) if successful.
3082 * - (-ENOTSUP) if hardware doesn't support.
3083 * - (-ENODEV) if *port_id* invalid.
3084 * - (-EIO) if device is removed.
3085 * - others depends on the specific operations implementation.
3087 int rte_eth_dev_filter_ctrl(uint16_t port_id, enum rte_filter_type filter_type,
3088 enum rte_filter_op filter_op, void *arg);
3091 * Get DCB information on an Ethernet device.
3094 * The port identifier of the Ethernet device.
3098 * - (0) if successful.
3099 * - (-ENODEV) if port identifier is invalid.
3100 * - (-EIO) if device is removed.
3101 * - (-ENOTSUP) if hardware doesn't support.
3103 int rte_eth_dev_get_dcb_info(uint16_t port_id,
3104 struct rte_eth_dcb_info *dcb_info);
3106 struct rte_eth_rxtx_callback;
3109 * Add a callback to be called on packet RX on a given port and queue.
3111 * This API configures a function to be called for each burst of
3112 * packets received on a given NIC port queue. The return value is a pointer
3113 * that can be used to later remove the callback using
3114 * rte_eth_remove_rx_callback().
3116 * Multiple functions are called in the order that they are added.
3119 * The port identifier of the Ethernet device.
3121 * The queue on the Ethernet device on which the callback is to be added.
3123 * The callback function
3125 * A generic pointer parameter which will be passed to each invocation of the
3126 * callback function on this port and queue.
3130 * On success, a pointer value which can later be used to remove the callback.
3132 const struct rte_eth_rxtx_callback *
3133 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
3134 rte_rx_callback_fn fn, void *user_param);
3137 * Add a callback that must be called first on packet RX on a given port
3140 * This API configures a first function to be called for each burst of
3141 * packets received on a given NIC port queue. The return value is a pointer
3142 * that can be used to later remove the callback using
3143 * rte_eth_remove_rx_callback().
3145 * Multiple functions are called in the order that they are added.
3148 * The port identifier of the Ethernet device.
3150 * The queue on the Ethernet device on which the callback is to be added.
3152 * The callback function
3154 * A generic pointer parameter which will be passed to each invocation of the
3155 * callback function on this port and queue.
3159 * On success, a pointer value which can later be used to remove the callback.
3161 const struct rte_eth_rxtx_callback *
3162 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
3163 rte_rx_callback_fn fn, void *user_param);
3166 * Add a callback to be called on packet TX on a given port and queue.
3168 * This API configures a function to be called for each burst of
3169 * packets sent on a given NIC port queue. The return value is a pointer
3170 * that can be used to later remove the callback using
3171 * rte_eth_remove_tx_callback().
3173 * Multiple functions are called in the order that they are added.
3176 * The port identifier of the Ethernet device.
3178 * The queue on the Ethernet device on which the callback is to be added.
3180 * The callback function
3182 * A generic pointer parameter which will be passed to each invocation of the
3183 * callback function on this port and queue.
3187 * On success, a pointer value which can later be used to remove the callback.
3189 const struct rte_eth_rxtx_callback *
3190 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
3191 rte_tx_callback_fn fn, void *user_param);
3194 * Remove an RX packet callback from a given port and queue.
3196 * This function is used to removed callbacks that were added to a NIC port
3197 * queue using rte_eth_add_rx_callback().
3199 * Note: the callback is removed from the callback list but it isn't freed
3200 * since the it may still be in use. The memory for the callback can be
3201 * subsequently freed back by the application by calling rte_free():
3203 * - Immediately - if the port is stopped, or the user knows that no
3204 * callbacks are in flight e.g. if called from the thread doing RX/TX
3207 * - After a short delay - where the delay is sufficient to allow any
3208 * in-flight callbacks to complete.
3211 * The port identifier of the Ethernet device.
3213 * The queue on the Ethernet device from which the callback is to be removed.
3215 * User supplied callback created via rte_eth_add_rx_callback().
3218 * - 0: Success. Callback was removed.
3219 * - -ENOTSUP: Callback support is not available.
3220 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3221 * is NULL or not found for the port/queue.
3223 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
3224 const struct rte_eth_rxtx_callback *user_cb);
3227 * Remove a TX packet callback from a given port and queue.
3229 * This function is used to removed callbacks that were added to a NIC port
3230 * queue using rte_eth_add_tx_callback().
3232 * Note: the callback is removed from the callback list but it isn't freed
3233 * since the it may still be in use. The memory for the callback can be
3234 * subsequently freed back by the application by calling rte_free():
3236 * - Immediately - if the port is stopped, or the user knows that no
3237 * callbacks are in flight e.g. if called from the thread doing RX/TX
3240 * - After a short delay - where the delay is sufficient to allow any
3241 * in-flight callbacks to complete.
3244 * The port identifier of the Ethernet device.
3246 * The queue on the Ethernet device from which the callback is to be removed.
3248 * User supplied callback created via rte_eth_add_tx_callback().
3251 * - 0: Success. Callback was removed.
3252 * - -ENOTSUP: Callback support is not available.
3253 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3254 * is NULL or not found for the port/queue.
3256 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
3257 const struct rte_eth_rxtx_callback *user_cb);
3260 * Retrieve information about given port's RX queue.
3263 * The port identifier of the Ethernet device.
3265 * The RX queue on the Ethernet device for which information
3266 * will be retrieved.
3268 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
3269 * the information of the Ethernet device.
3273 * - -ENOTSUP: routine is not supported by the device PMD.
3274 * - -EINVAL: The port_id or the queue_id is out of range.
3276 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3277 struct rte_eth_rxq_info *qinfo);
3280 * Retrieve information about given port's TX queue.
3283 * The port identifier of the Ethernet device.
3285 * The TX queue on the Ethernet device for which information
3286 * will be retrieved.
3288 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
3289 * the information of the Ethernet device.
3293 * - -ENOTSUP: routine is not supported by the device PMD.
3294 * - -EINVAL: The port_id or the queue_id is out of range.
3296 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3297 struct rte_eth_txq_info *qinfo);
3300 * Retrieve device registers and register attributes (number of registers and
3304 * The port identifier of the Ethernet device.
3306 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
3307 * NULL the function fills in the width and length fields. If non-NULL
3308 * the registers are put into the buffer pointed at by the data field.
3310 * - (0) if successful.
3311 * - (-ENOTSUP) if hardware doesn't support.
3312 * - (-ENODEV) if *port_id* invalid.
3313 * - (-EIO) if device is removed.
3314 * - others depends on the specific operations implementation.
3316 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
3319 * Retrieve size of device EEPROM
3322 * The port identifier of the Ethernet device.
3324 * - (>=0) EEPROM size if successful.
3325 * - (-ENOTSUP) if hardware doesn't support.
3326 * - (-ENODEV) if *port_id* invalid.
3327 * - (-EIO) if device is removed.
3328 * - others depends on the specific operations implementation.
3330 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
3333 * Retrieve EEPROM and EEPROM attribute
3336 * The port identifier of the Ethernet device.
3338 * The template includes buffer for return EEPROM data and
3339 * EEPROM attributes to be filled.
3341 * - (0) if successful.
3342 * - (-ENOTSUP) if hardware doesn't support.
3343 * - (-ENODEV) if *port_id* invalid.
3344 * - (-EIO) if device is removed.
3345 * - others depends on the specific operations implementation.
3347 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3350 * Program EEPROM with provided data
3353 * The port identifier of the Ethernet device.
3355 * The template includes EEPROM data for programming and
3356 * EEPROM attributes to be filled
3358 * - (0) if successful.
3359 * - (-ENOTSUP) if hardware doesn't support.
3360 * - (-ENODEV) if *port_id* invalid.
3361 * - (-EIO) if device is removed.
3362 * - others depends on the specific operations implementation.
3364 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3367 * Set the list of multicast addresses to filter on an Ethernet device.
3370 * The port identifier of the Ethernet device.
3371 * @param mc_addr_set
3372 * The array of multicast addresses to set. Equal to NULL when the function
3373 * is invoked to flush the set of filtered addresses.
3375 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
3376 * when the function is invoked to flush the set of filtered addresses.
3378 * - (0) if successful.
3379 * - (-ENODEV) if *port_id* invalid.
3380 * - (-EIO) if device is removed.
3381 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
3382 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
3384 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
3385 struct ether_addr *mc_addr_set,
3386 uint32_t nb_mc_addr);
3389 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
3392 * The port identifier of the Ethernet device.
3396 * - -ENODEV: The port ID is invalid.
3397 * - -EIO: if device is removed.
3398 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3400 int rte_eth_timesync_enable(uint16_t port_id);
3403 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
3406 * The port identifier of the Ethernet device.
3410 * - -ENODEV: The port ID is invalid.
3411 * - -EIO: if device is removed.
3412 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3414 int rte_eth_timesync_disable(uint16_t port_id);
3417 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
3420 * The port identifier of the Ethernet device.
3422 * Pointer to the timestamp struct.
3424 * Device specific flags. Used to pass the RX timesync register index to
3425 * i40e. Unused in igb/ixgbe, pass 0 instead.
3429 * - -EINVAL: No timestamp is available.
3430 * - -ENODEV: The port ID is invalid.
3431 * - -EIO: if device is removed.
3432 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3434 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
3435 struct timespec *timestamp, uint32_t flags);
3438 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
3441 * The port identifier of the Ethernet device.
3443 * Pointer to the timestamp struct.
3447 * - -EINVAL: No timestamp is available.
3448 * - -ENODEV: The port ID is invalid.
3449 * - -EIO: if device is removed.
3450 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3452 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
3453 struct timespec *timestamp);
3456 * Adjust the timesync clock on an Ethernet device.
3458 * This is usually used in conjunction with other Ethdev timesync functions to
3459 * synchronize the device time using the IEEE1588/802.1AS protocol.
3462 * The port identifier of the Ethernet device.
3464 * The adjustment in nanoseconds.
3468 * - -ENODEV: The port ID is invalid.
3469 * - -EIO: if device is removed.
3470 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3472 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
3475 * Read the time from the timesync clock on an Ethernet device.
3477 * This is usually used in conjunction with other Ethdev timesync functions to
3478 * synchronize the device time using the IEEE1588/802.1AS protocol.
3481 * The port identifier of the Ethernet device.
3483 * Pointer to the timespec struct that holds the time.
3488 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
3491 * Set the time of the timesync clock on an Ethernet device.
3493 * This is usually used in conjunction with other Ethdev timesync functions to
3494 * synchronize the device time using the IEEE1588/802.1AS protocol.
3497 * The port identifier of the Ethernet device.
3499 * Pointer to the timespec struct that holds the time.
3503 * - -EINVAL: No timestamp is available.
3504 * - -ENODEV: The port ID is invalid.
3505 * - -EIO: if device is removed.
3506 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3508 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
3511 * Config l2 tunnel ether type of an Ethernet device for filtering specific
3512 * tunnel packets by ether type.
3515 * The port identifier of the Ethernet device.
3517 * l2 tunnel configuration.
3520 * - (0) if successful.
3521 * - (-ENODEV) if port identifier is invalid.
3522 * - (-EIO) if device is removed.
3523 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3526 rte_eth_dev_l2_tunnel_eth_type_conf(uint16_t port_id,
3527 struct rte_eth_l2_tunnel_conf *l2_tunnel);
3530 * Enable/disable l2 tunnel offload functions. Include,
3531 * 1, The ability of parsing a type of l2 tunnel of an Ethernet device.
3532 * Filtering, forwarding and offloading this type of tunnel packets depend on
3534 * 2, Stripping the l2 tunnel tag.
3535 * 3, Insertion of the l2 tunnel tag.
3536 * 4, Forwarding the packets based on the l2 tunnel tag.
3539 * The port identifier of the Ethernet device.
3541 * l2 tunnel parameters.
3543 * Indicate the offload function.
3545 * Enable or disable this function.
3548 * - (0) if successful.
3549 * - (-ENODEV) if port identifier is invalid.
3550 * - (-EIO) if device is removed.
3551 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3554 rte_eth_dev_l2_tunnel_offload_set(uint16_t port_id,
3555 struct rte_eth_l2_tunnel_conf *l2_tunnel,
3560 * Get the port id from pci address or device name
3561 * Ex: 0000:2:00.0 or vdev name net_pcap0
3564 * pci address or name of the device
3566 * pointer to port identifier of the device
3568 * - (0) if successful and port_id is filled.
3569 * - (-ENODEV or -EINVAL) on failure.
3572 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
3575 * Get the device name from port id
3578 * pointer to port identifier of the device
3580 * pci address or name of the device
3582 * - (0) if successful.
3583 * - (-EINVAL) on failure.
3586 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
3589 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
3590 * the ethernet device information, otherwise adjust them to boundaries.
3593 * The port identifier of the Ethernet device.
3595 * A pointer to a uint16_t where the number of receive
3596 * descriptors stored.
3598 * A pointer to a uint16_t where the number of transmit
3599 * descriptors stored.
3601 * - (0) if successful.
3602 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
3604 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
3605 uint16_t *nb_rx_desc,
3606 uint16_t *nb_tx_desc);
3609 * Test if a port supports specific mempool ops.
3612 * Port identifier of the Ethernet device.
3614 * The name of the pool operations to test.
3616 * - 0: best mempool ops choice for this port.
3617 * - 1: mempool ops are supported for this port.
3618 * - -ENOTSUP: mempool ops not supported for this port.
3619 * - -ENODEV: Invalid port Identifier.
3620 * - -EINVAL: Pool param is null.
3623 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
3626 * Get the security context for the Ethernet device.
3629 * Port identifier of the Ethernet device
3632 * - pointer to security context on success.
3635 rte_eth_dev_get_sec_ctx(uint16_t port_id);
3638 #include <rte_ethdev_core.h>
3642 * Retrieve a burst of input packets from a receive queue of an Ethernet
3643 * device. The retrieved packets are stored in *rte_mbuf* structures whose
3644 * pointers are supplied in the *rx_pkts* array.
3646 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
3647 * receive queue, up to *nb_pkts* packets, and for each completed RX
3648 * descriptor in the ring, it performs the following operations:
3650 * - Initialize the *rte_mbuf* data structure associated with the
3651 * RX descriptor according to the information provided by the NIC into
3652 * that RX descriptor.
3654 * - Store the *rte_mbuf* data structure into the next entry of the
3657 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
3658 * allocated from the memory pool associated with the receive queue at
3659 * initialization time.
3661 * When retrieving an input packet that was scattered by the controller
3662 * into multiple receive descriptors, the rte_eth_rx_burst() function
3663 * appends the associated *rte_mbuf* buffers to the first buffer of the
3666 * The rte_eth_rx_burst() function returns the number of packets
3667 * actually retrieved, which is the number of *rte_mbuf* data structures
3668 * effectively supplied into the *rx_pkts* array.
3669 * A return value equal to *nb_pkts* indicates that the RX queue contained
3670 * at least *rx_pkts* packets, and this is likely to signify that other
3671 * received packets remain in the input queue. Applications implementing
3672 * a "retrieve as much received packets as possible" policy can check this
3673 * specific case and keep invoking the rte_eth_rx_burst() function until
3674 * a value less than *nb_pkts* is returned.
3676 * This receive method has the following advantages:
3678 * - It allows a run-to-completion network stack engine to retrieve and
3679 * to immediately process received packets in a fast burst-oriented
3680 * approach, avoiding the overhead of unnecessary intermediate packet
3681 * queue/dequeue operations.
3683 * - Conversely, it also allows an asynchronous-oriented processing
3684 * method to retrieve bursts of received packets and to immediately
3685 * queue them for further parallel processing by another logical core,
3686 * for instance. However, instead of having received packets being
3687 * individually queued by the driver, this approach allows the caller
3688 * of the rte_eth_rx_burst() function to queue a burst of retrieved
3689 * packets at a time and therefore dramatically reduce the cost of
3690 * enqueue/dequeue operations per packet.
3692 * - It allows the rte_eth_rx_burst() function of the driver to take
3693 * advantage of burst-oriented hardware features (CPU cache,
3694 * prefetch instructions, and so on) to minimize the number of CPU
3695 * cycles per packet.
3697 * To summarize, the proposed receive API enables many
3698 * burst-oriented optimizations in both synchronous and asynchronous
3699 * packet processing environments with no overhead in both cases.
3701 * The rte_eth_rx_burst() function does not provide any error
3702 * notification to avoid the corresponding overhead. As a hint, the
3703 * upper-level application might check the status of the device link once
3704 * being systematically returned a 0 value for a given number of tries.
3707 * The port identifier of the Ethernet device.
3709 * The index of the receive queue from which to retrieve input packets.
3710 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3711 * to rte_eth_dev_configure().
3713 * The address of an array of pointers to *rte_mbuf* structures that
3714 * must be large enough to store *nb_pkts* pointers in it.
3716 * The maximum number of packets to retrieve.
3718 * The number of packets actually retrieved, which is the number
3719 * of pointers to *rte_mbuf* structures effectively supplied to the
3722 static inline uint16_t
3723 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
3724 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
3726 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3728 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3729 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
3730 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
3732 if (queue_id >= dev->data->nb_rx_queues) {
3733 RTE_PMD_DEBUG_TRACE("Invalid RX queue_id=%d\n", queue_id);
3737 int16_t nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
3740 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
3741 struct rte_eth_rxtx_callback *cb = dev->post_rx_burst_cbs[queue_id];
3743 if (unlikely(cb != NULL)) {
3745 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
3746 nb_pkts, cb->param);
3748 } while (cb != NULL);
3756 * Get the number of used descriptors of a rx queue
3759 * The port identifier of the Ethernet device.
3761 * The queue id on the specific port.
3763 * The number of used descriptors in the specific queue, or:
3764 * (-EINVAL) if *port_id* or *queue_id* is invalid
3765 * (-ENOTSUP) if the device does not support this function
3768 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
3770 struct rte_eth_dev *dev;
3772 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -EINVAL);
3773 dev = &rte_eth_devices[port_id];
3774 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_count, -ENOTSUP);
3775 if (queue_id >= dev->data->nb_rx_queues)
3778 return (*dev->dev_ops->rx_queue_count)(dev, queue_id);
3782 * Check if the DD bit of the specific RX descriptor in the queue has been set
3785 * The port identifier of the Ethernet device.
3787 * The queue id on the specific port.
3789 * The offset of the descriptor ID from tail.
3791 * - (1) if the specific DD bit is set.
3792 * - (0) if the specific DD bit is not set.
3793 * - (-ENODEV) if *port_id* invalid.
3794 * - (-ENOTSUP) if the device does not support this function
3797 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
3799 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3800 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3801 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_done, -ENOTSUP);
3802 return (*dev->dev_ops->rx_descriptor_done)( \
3803 dev->data->rx_queues[queue_id], offset);
3806 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
3807 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
3808 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
3811 * Check the status of a Rx descriptor in the queue
3813 * It should be called in a similar context than the Rx function:
3814 * - on a dataplane core
3815 * - not concurrently on the same queue
3817 * Since it's a dataplane function, no check is performed on port_id and
3818 * queue_id. The caller must therefore ensure that the port is enabled
3819 * and the queue is configured and running.
3821 * Note: accessing to a random descriptor in the ring may trigger cache
3822 * misses and have a performance impact.
3825 * A valid port identifier of the Ethernet device which.
3827 * A valid Rx queue identifier on this port.
3829 * The offset of the descriptor starting from tail (0 is the next
3830 * packet to be received by the driver).
3833 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
3835 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
3836 * not yet processed by the driver (i.e. in the receive queue).
3837 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
3838 * the driver and not yet returned to hw, or reserved by the hw.
3839 * - (-EINVAL) bad descriptor offset.
3840 * - (-ENOTSUP) if the device does not support this function.
3841 * - (-ENODEV) bad port or queue (only if compiled with debug).
3844 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
3847 struct rte_eth_dev *dev;
3850 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3851 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3853 dev = &rte_eth_devices[port_id];
3854 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3855 if (queue_id >= dev->data->nb_rx_queues)
3858 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_status, -ENOTSUP);
3859 rxq = dev->data->rx_queues[queue_id];
3861 return (*dev->dev_ops->rx_descriptor_status)(rxq, offset);
3864 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
3865 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
3866 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
3869 * Check the status of a Tx descriptor in the queue.
3871 * It should be called in a similar context than the Tx function:
3872 * - on a dataplane core
3873 * - not concurrently on the same queue
3875 * Since it's a dataplane function, no check is performed on port_id and
3876 * queue_id. The caller must therefore ensure that the port is enabled
3877 * and the queue is configured and running.
3879 * Note: accessing to a random descriptor in the ring may trigger cache
3880 * misses and have a performance impact.
3883 * A valid port identifier of the Ethernet device which.
3885 * A valid Tx queue identifier on this port.
3887 * The offset of the descriptor starting from tail (0 is the place where
3888 * the next packet will be send).
3891 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
3892 * in the transmit queue.
3893 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
3894 * be reused by the driver.
3895 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
3896 * driver or the hardware.
3897 * - (-EINVAL) bad descriptor offset.
3898 * - (-ENOTSUP) if the device does not support this function.
3899 * - (-ENODEV) bad port or queue (only if compiled with debug).
3901 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
3902 uint16_t queue_id, uint16_t offset)
3904 struct rte_eth_dev *dev;
3907 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3908 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3910 dev = &rte_eth_devices[port_id];
3911 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3912 if (queue_id >= dev->data->nb_tx_queues)
3915 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_descriptor_status, -ENOTSUP);
3916 txq = dev->data->tx_queues[queue_id];
3918 return (*dev->dev_ops->tx_descriptor_status)(txq, offset);
3922 * Send a burst of output packets on a transmit queue of an Ethernet device.
3924 * The rte_eth_tx_burst() function is invoked to transmit output packets
3925 * on the output queue *queue_id* of the Ethernet device designated by its
3927 * The *nb_pkts* parameter is the number of packets to send which are
3928 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
3929 * allocated from a pool created with rte_pktmbuf_pool_create().
3930 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
3931 * up to the number of transmit descriptors available in the TX ring of the
3933 * For each packet to send, the rte_eth_tx_burst() function performs
3934 * the following operations:
3936 * - Pick up the next available descriptor in the transmit ring.
3938 * - Free the network buffer previously sent with that descriptor, if any.
3940 * - Initialize the transmit descriptor with the information provided
3941 * in the *rte_mbuf data structure.
3943 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
3944 * the rte_eth_tx_burst() function uses several transmit descriptors
3947 * The rte_eth_tx_burst() function returns the number of packets it
3948 * actually sent. A return value equal to *nb_pkts* means that all packets
3949 * have been sent, and this is likely to signify that other output packets
3950 * could be immediately transmitted again. Applications that implement a
3951 * "send as many packets to transmit as possible" policy can check this
3952 * specific case and keep invoking the rte_eth_tx_burst() function until
3953 * a value less than *nb_pkts* is returned.
3955 * It is the responsibility of the rte_eth_tx_burst() function to
3956 * transparently free the memory buffers of packets previously sent.
3957 * This feature is driven by the *tx_free_thresh* value supplied to the
3958 * rte_eth_dev_configure() function at device configuration time.
3959 * When the number of free TX descriptors drops below this threshold, the
3960 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
3961 * of those packets whose transmission was effectively completed.
3963 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
3964 * invoke this function concurrently on the same tx queue without SW lock.
3965 * @see rte_eth_dev_info_get, struct rte_eth_txconf::txq_flags
3967 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
3971 * The port identifier of the Ethernet device.
3973 * The index of the transmit queue through which output packets must be
3975 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
3976 * to rte_eth_dev_configure().
3978 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
3979 * which contain the output packets.
3981 * The maximum number of packets to transmit.
3983 * The number of output packets actually stored in transmit descriptors of
3984 * the transmit ring. The return value can be less than the value of the
3985 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
3987 static inline uint16_t
3988 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
3989 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
3991 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3993 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3994 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
3995 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
3997 if (queue_id >= dev->data->nb_tx_queues) {
3998 RTE_PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", queue_id);
4003 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4004 struct rte_eth_rxtx_callback *cb = dev->pre_tx_burst_cbs[queue_id];
4006 if (unlikely(cb != NULL)) {
4008 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
4011 } while (cb != NULL);
4015 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
4020 * @b EXPERIMENTAL: this API may change without prior notice
4022 * Process a burst of output packets on a transmit queue of an Ethernet device.
4024 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
4025 * transmitted on the output queue *queue_id* of the Ethernet device designated
4027 * The *nb_pkts* parameter is the number of packets to be prepared which are
4028 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
4029 * allocated from a pool created with rte_pktmbuf_pool_create().
4030 * For each packet to send, the rte_eth_tx_prepare() function performs
4031 * the following operations:
4033 * - Check if packet meets devices requirements for tx offloads.
4035 * - Check limitations about number of segments.
4037 * - Check additional requirements when debug is enabled.
4039 * - Update and/or reset required checksums when tx offload is set for packet.
4041 * Since this function can modify packet data, provided mbufs must be safely
4042 * writable (e.g. modified data cannot be in shared segment).
4044 * The rte_eth_tx_prepare() function returns the number of packets ready to be
4045 * sent. A return value equal to *nb_pkts* means that all packets are valid and
4046 * ready to be sent, otherwise stops processing on the first invalid packet and
4047 * leaves the rest packets untouched.
4049 * When this functionality is not implemented in the driver, all packets are
4050 * are returned untouched.
4053 * The port identifier of the Ethernet device.
4054 * The value must be a valid port id.
4056 * The index of the transmit queue through which output packets must be
4058 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4059 * to rte_eth_dev_configure().
4061 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
4062 * which contain the output packets.
4064 * The maximum number of packets to process.
4066 * The number of packets correct and ready to be sent. The return value can be
4067 * less than the value of the *tx_pkts* parameter when some packet doesn't
4068 * meet devices requirements with rte_errno set appropriately:
4069 * - -EINVAL: offload flags are not correctly set
4070 * - -ENOTSUP: the offload feature is not supported by the hardware
4074 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
4076 static inline uint16_t
4077 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
4078 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4080 struct rte_eth_dev *dev;
4082 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4083 if (!rte_eth_dev_is_valid_port(port_id)) {
4084 RTE_PMD_DEBUG_TRACE("Invalid TX port_id=%d\n", port_id);
4085 rte_errno = -EINVAL;
4090 dev = &rte_eth_devices[port_id];
4092 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4093 if (queue_id >= dev->data->nb_tx_queues) {
4094 RTE_PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", queue_id);
4095 rte_errno = -EINVAL;
4100 if (!dev->tx_pkt_prepare)
4103 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
4110 * Native NOOP operation for compilation targets which doesn't require any
4111 * preparations steps, and functional NOOP may introduce unnecessary performance
4114 * Generally this is not a good idea to turn it on globally and didn't should
4115 * be used if behavior of tx_preparation can change.
4118 static inline uint16_t
4119 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
4120 __rte_unused uint16_t queue_id,
4121 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4129 * Send any packets queued up for transmission on a port and HW queue
4131 * This causes an explicit flush of packets previously buffered via the
4132 * rte_eth_tx_buffer() function. It returns the number of packets successfully
4133 * sent to the NIC, and calls the error callback for any unsent packets. Unless
4134 * explicitly set up otherwise, the default callback simply frees the unsent
4135 * packets back to the owning mempool.
4138 * The port identifier of the Ethernet device.
4140 * The index of the transmit queue through which output packets must be
4142 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4143 * to rte_eth_dev_configure().
4145 * Buffer of packets to be transmit.
4147 * The number of packets successfully sent to the Ethernet device. The error
4148 * callback is called for any packets which could not be sent.
4150 static inline uint16_t
4151 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
4152 struct rte_eth_dev_tx_buffer *buffer)
4155 uint16_t to_send = buffer->length;
4160 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
4164 /* All packets sent, or to be dealt with by callback below */
4165 if (unlikely(sent != to_send))
4166 buffer->error_callback(&buffer->pkts[sent], to_send - sent,
4167 buffer->error_userdata);
4173 * Buffer a single packet for future transmission on a port and queue
4175 * This function takes a single mbuf/packet and buffers it for later
4176 * transmission on the particular port and queue specified. Once the buffer is
4177 * full of packets, an attempt will be made to transmit all the buffered
4178 * packets. In case of error, where not all packets can be transmitted, a
4179 * callback is called with the unsent packets as a parameter. If no callback
4180 * is explicitly set up, the unsent packets are just freed back to the owning
4181 * mempool. The function returns the number of packets actually sent i.e.
4182 * 0 if no buffer flush occurred, otherwise the number of packets successfully
4186 * The port identifier of the Ethernet device.
4188 * The index of the transmit queue through which output packets must be
4190 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4191 * to rte_eth_dev_configure().
4193 * Buffer used to collect packets to be sent.
4195 * Pointer to the packet mbuf to be sent.
4197 * 0 = packet has been buffered for later transmission
4198 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
4199 * causing N packets to be sent, and the error callback to be called for
4202 static __rte_always_inline uint16_t
4203 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
4204 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
4206 buffer->pkts[buffer->length++] = tx_pkt;
4207 if (buffer->length < buffer->size)
4210 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
4217 #endif /* _RTE_ETHDEV_H_ */