1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2010-2017 Intel Corporation
11 * RTE Ethernet Device API
13 * The Ethernet Device API is composed of two parts:
15 * - The application-oriented Ethernet API that includes functions to setup
16 * an Ethernet device (configure it, setup its RX and TX queues and start it),
17 * to get its MAC address, the speed and the status of its physical link,
18 * to receive and to transmit packets, and so on.
20 * - The driver-oriented Ethernet API that exports functions allowing
21 * an Ethernet Poll Mode Driver (PMD) to allocate an Ethernet device instance,
22 * create memzone for HW rings and process registered callbacks, and so on.
23 * PMDs should include rte_ethdev_driver.h instead of this header.
25 * By default, all the functions of the Ethernet Device API exported by a PMD
26 * are lock-free functions which assume to not be invoked in parallel on
27 * different logical cores to work on the same target object. For instance,
28 * the receive function of a PMD cannot be invoked in parallel on two logical
29 * cores to poll the same RX queue [of the same port]. Of course, this function
30 * can be invoked in parallel by different logical cores on different RX queues.
31 * It is the responsibility of the upper level application to enforce this rule.
33 * If needed, parallel accesses by multiple logical cores to shared queues
34 * shall be explicitly protected by dedicated inline lock-aware functions
35 * built on top of their corresponding lock-free functions of the PMD API.
37 * In all functions of the Ethernet API, the Ethernet device is
38 * designated by an integer >= 0 named the device port identifier.
40 * At the Ethernet driver level, Ethernet devices are represented by a generic
41 * data structure of type *rte_eth_dev*.
43 * Ethernet devices are dynamically registered during the PCI probing phase
44 * performed at EAL initialization time.
45 * When an Ethernet device is being probed, an *rte_eth_dev* structure and
46 * a new port identifier are allocated for that device. Then, the eth_dev_init()
47 * function supplied by the Ethernet driver matching the probed PCI
48 * device is invoked to properly initialize the device.
50 * The role of the device init function consists of resetting the hardware,
51 * checking access to Non-volatile Memory (NVM), reading the MAC address
54 * If the device init operation is successful, the correspondence between
55 * the port identifier assigned to the new device and its associated
56 * *rte_eth_dev* structure is effectively registered.
57 * Otherwise, both the *rte_eth_dev* structure and the port identifier are
60 * The functions exported by the application Ethernet API to setup a device
61 * designated by its port identifier must be invoked in the following order:
62 * - rte_eth_dev_configure()
63 * - rte_eth_tx_queue_setup()
64 * - rte_eth_rx_queue_setup()
65 * - rte_eth_dev_start()
67 * Then, the network application can invoke, in any order, the functions
68 * exported by the Ethernet API to get the MAC address of a given device, to
69 * get the speed and the status of a device physical link, to receive/transmit
70 * [burst of] packets, and so on.
72 * If the application wants to change the configuration (i.e. call
73 * rte_eth_dev_configure(), rte_eth_tx_queue_setup(), or
74 * rte_eth_rx_queue_setup()), it must call rte_eth_dev_stop() first to stop the
75 * device and then do the reconfiguration before calling rte_eth_dev_start()
76 * again. The transmit and receive functions should not be invoked when the
79 * Please note that some configuration is not stored between calls to
80 * rte_eth_dev_stop()/rte_eth_dev_start(). The following configuration will
83 * - flow control settings
84 * - receive mode configuration (promiscuous mode, hardware checksum mode,
85 * RSS/VMDQ settings etc.)
86 * - VLAN filtering configuration
87 * - MAC addresses supplied to MAC address array
88 * - flow director filtering mode (but not filtering rules)
89 * - NIC queue statistics mappings
91 * Any other configuration will not be stored and will need to be re-entered
92 * before a call to rte_eth_dev_start().
94 * Finally, a network application can close an Ethernet device by invoking the
95 * rte_eth_dev_close() function.
97 * Each function of the application Ethernet API invokes a specific function
98 * of the PMD that controls the target device designated by its port
100 * For this purpose, all device-specific functions of an Ethernet driver are
101 * supplied through a set of pointers contained in a generic structure of type
103 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev*
104 * structure by the device init function of the Ethernet driver, which is
105 * invoked during the PCI probing phase, as explained earlier.
107 * In other words, each function of the Ethernet API simply retrieves the
108 * *rte_eth_dev* structure associated with the device port identifier and
109 * performs an indirect invocation of the corresponding driver function
110 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure.
112 * For performance reasons, the address of the burst-oriented RX and TX
113 * functions of the Ethernet driver are not contained in the *eth_dev_ops*
114 * structure. Instead, they are directly stored at the beginning of the
115 * *rte_eth_dev* structure to avoid an extra indirect memory access during
118 * RTE ethernet device drivers do not use interrupts for transmitting or
119 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit
120 * functions to applications.
121 * Both receive and transmit functions are packet-burst oriented to minimize
122 * their cost per packet through the following optimizations:
124 * - Sharing among multiple packets the incompressible cost of the
125 * invocation of receive/transmit functions.
127 * - Enabling receive/transmit functions to take advantage of burst-oriented
128 * hardware features (L1 cache, prefetch instructions, NIC head/tail
129 * registers) to minimize the number of CPU cycles per packet, for instance,
130 * by avoiding useless read memory accesses to ring descriptors, or by
131 * systematically using arrays of pointers that exactly fit L1 cache line
132 * boundaries and sizes.
134 * The burst-oriented receive function does not provide any error notification,
135 * to avoid the corresponding overhead. As a hint, the upper-level application
136 * might check the status of the device link once being systematically returned
137 * a 0 value by the receive function of the driver for a given number of tries.
146 /* Use this macro to check if LRO API is supported */
147 #define RTE_ETHDEV_HAS_LRO_SUPPORT
149 #include <rte_compat.h>
151 #include <rte_interrupts.h>
153 #include <rte_devargs.h>
154 #include <rte_errno.h>
155 #include <rte_common.h>
156 #include <rte_config.h>
158 #include "rte_ether.h"
159 #include "rte_eth_ctrl.h"
160 #include "rte_dev_info.h"
162 extern int rte_eth_dev_logtype;
164 #define RTE_ETHDEV_LOG(level, ...) \
165 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
170 * A structure used to retrieve statistics for an Ethernet port.
171 * Not all statistics fields in struct rte_eth_stats are supported
172 * by any type of network interface card (NIC). If any statistics
173 * field is not supported, its value is 0.
175 struct rte_eth_stats {
176 uint64_t ipackets; /**< Total number of successfully received packets. */
177 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
178 uint64_t ibytes; /**< Total number of successfully received bytes. */
179 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
181 /**< Total of RX packets dropped by the HW,
182 * because there are no available buffer (i.e. RX queues are full).
184 uint64_t ierrors; /**< Total number of erroneous received packets. */
185 uint64_t oerrors; /**< Total number of failed transmitted packets. */
186 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
187 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
188 /**< Total number of queue RX packets. */
189 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
190 /**< Total number of queue TX packets. */
191 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
192 /**< Total number of successfully received queue bytes. */
193 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
194 /**< Total number of successfully transmitted queue bytes. */
195 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
196 /**< Total number of queue packets received that are dropped. */
200 * Device supported speeds bitmap flags
202 #define ETH_LINK_SPEED_AUTONEG (0 << 0) /**< Autonegotiate (all speeds) */
203 #define ETH_LINK_SPEED_FIXED (1 << 0) /**< Disable autoneg (fixed speed) */
204 #define ETH_LINK_SPEED_10M_HD (1 << 1) /**< 10 Mbps half-duplex */
205 #define ETH_LINK_SPEED_10M (1 << 2) /**< 10 Mbps full-duplex */
206 #define ETH_LINK_SPEED_100M_HD (1 << 3) /**< 100 Mbps half-duplex */
207 #define ETH_LINK_SPEED_100M (1 << 4) /**< 100 Mbps full-duplex */
208 #define ETH_LINK_SPEED_1G (1 << 5) /**< 1 Gbps */
209 #define ETH_LINK_SPEED_2_5G (1 << 6) /**< 2.5 Gbps */
210 #define ETH_LINK_SPEED_5G (1 << 7) /**< 5 Gbps */
211 #define ETH_LINK_SPEED_10G (1 << 8) /**< 10 Gbps */
212 #define ETH_LINK_SPEED_20G (1 << 9) /**< 20 Gbps */
213 #define ETH_LINK_SPEED_25G (1 << 10) /**< 25 Gbps */
214 #define ETH_LINK_SPEED_40G (1 << 11) /**< 40 Gbps */
215 #define ETH_LINK_SPEED_50G (1 << 12) /**< 50 Gbps */
216 #define ETH_LINK_SPEED_56G (1 << 13) /**< 56 Gbps */
217 #define ETH_LINK_SPEED_100G (1 << 14) /**< 100 Gbps */
220 * Ethernet numeric link speeds in Mbps
222 #define ETH_SPEED_NUM_NONE 0 /**< Not defined */
223 #define ETH_SPEED_NUM_10M 10 /**< 10 Mbps */
224 #define ETH_SPEED_NUM_100M 100 /**< 100 Mbps */
225 #define ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */
226 #define ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */
227 #define ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */
228 #define ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */
229 #define ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */
230 #define ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */
231 #define ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */
232 #define ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */
233 #define ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */
234 #define ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */
237 * A structure used to retrieve link-level information of an Ethernet port.
240 struct rte_eth_link {
241 uint32_t link_speed; /**< ETH_SPEED_NUM_ */
242 uint16_t link_duplex : 1; /**< ETH_LINK_[HALF/FULL]_DUPLEX */
243 uint16_t link_autoneg : 1; /**< ETH_LINK_[AUTONEG/FIXED] */
244 uint16_t link_status : 1; /**< ETH_LINK_[DOWN/UP] */
245 } __attribute__((aligned(8))); /**< aligned for atomic64 read/write */
247 /* Utility constants */
248 #define ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */
249 #define ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */
250 #define ETH_LINK_DOWN 0 /**< Link is down (see link_status). */
251 #define ETH_LINK_UP 1 /**< Link is up (see link_status). */
252 #define ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */
253 #define ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */
256 * A structure used to configure the ring threshold registers of an RX/TX
257 * queue for an Ethernet port.
259 struct rte_eth_thresh {
260 uint8_t pthresh; /**< Ring prefetch threshold. */
261 uint8_t hthresh; /**< Ring host threshold. */
262 uint8_t wthresh; /**< Ring writeback threshold. */
266 * Simple flags are used for rte_eth_conf.rxmode.mq_mode.
268 #define ETH_MQ_RX_RSS_FLAG 0x1
269 #define ETH_MQ_RX_DCB_FLAG 0x2
270 #define ETH_MQ_RX_VMDQ_FLAG 0x4
273 * A set of values to identify what method is to be used to route
274 * packets to multiple queues.
276 enum rte_eth_rx_mq_mode {
277 /** None of DCB,RSS or VMDQ mode */
280 /** For RX side, only RSS is on */
281 ETH_MQ_RX_RSS = ETH_MQ_RX_RSS_FLAG,
282 /** For RX side,only DCB is on. */
283 ETH_MQ_RX_DCB = ETH_MQ_RX_DCB_FLAG,
284 /** Both DCB and RSS enable */
285 ETH_MQ_RX_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG,
287 /** Only VMDQ, no RSS nor DCB */
288 ETH_MQ_RX_VMDQ_ONLY = ETH_MQ_RX_VMDQ_FLAG,
289 /** RSS mode with VMDQ */
290 ETH_MQ_RX_VMDQ_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_VMDQ_FLAG,
291 /** Use VMDQ+DCB to route traffic to queues */
292 ETH_MQ_RX_VMDQ_DCB = ETH_MQ_RX_VMDQ_FLAG | ETH_MQ_RX_DCB_FLAG,
293 /** Enable both VMDQ and DCB in VMDq */
294 ETH_MQ_RX_VMDQ_DCB_RSS = ETH_MQ_RX_RSS_FLAG | ETH_MQ_RX_DCB_FLAG |
299 * for rx mq mode backward compatible
301 #define ETH_RSS ETH_MQ_RX_RSS
302 #define VMDQ_DCB ETH_MQ_RX_VMDQ_DCB
303 #define ETH_DCB_RX ETH_MQ_RX_DCB
306 * A set of values to identify what method is to be used to transmit
307 * packets using multi-TCs.
309 enum rte_eth_tx_mq_mode {
310 ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */
311 ETH_MQ_TX_DCB, /**< For TX side,only DCB is on. */
312 ETH_MQ_TX_VMDQ_DCB, /**< For TX side,both DCB and VT is on. */
313 ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */
317 * for tx mq mode backward compatible
319 #define ETH_DCB_NONE ETH_MQ_TX_NONE
320 #define ETH_VMDQ_DCB_TX ETH_MQ_TX_VMDQ_DCB
321 #define ETH_DCB_TX ETH_MQ_TX_DCB
324 * A structure used to configure the RX features of an Ethernet port.
326 struct rte_eth_rxmode {
327 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
328 enum rte_eth_rx_mq_mode mq_mode;
329 uint32_t max_rx_pkt_len; /**< Only used if JUMBO_FRAME enabled. */
330 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
332 * Per-port Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
333 * Only offloads set on rx_offload_capa field on rte_eth_dev_info
334 * structure are allowed to be set.
340 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN.
341 * Note that single VLAN is treated the same as inner VLAN.
344 ETH_VLAN_TYPE_UNKNOWN = 0,
345 ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */
346 ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */
351 * A structure used to describe a vlan filter.
352 * If the bit corresponding to a VID is set, such VID is on.
354 struct rte_vlan_filter_conf {
359 * A structure used to configure the Receive Side Scaling (RSS) feature
360 * of an Ethernet port.
361 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
362 * to an array holding the RSS key to use for hashing specific header
363 * fields of received packets. The length of this array should be indicated
364 * by *rss_key_len* below. Otherwise, a default random hash key is used by
367 * The *rss_key_len* field of the *rss_conf* structure indicates the length
368 * in bytes of the array pointed by *rss_key*. To be compatible, this length
369 * will be checked in i40e only. Others assume 40 bytes to be used as before.
371 * The *rss_hf* field of the *rss_conf* structure indicates the different
372 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
373 * Supplying an *rss_hf* equal to zero disables the RSS feature.
375 struct rte_eth_rss_conf {
376 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
377 uint8_t rss_key_len; /**< hash key length in bytes. */
378 uint64_t rss_hf; /**< Hash functions to apply - see below. */
382 * The RSS offload types are defined based on flow types which are defined
383 * in rte_eth_ctrl.h. Different NIC hardwares may support different RSS offload
384 * types. The supported flow types or RSS offload types can be queried by
385 * rte_eth_dev_info_get().
387 #define ETH_RSS_IPV4 (1ULL << RTE_ETH_FLOW_IPV4)
388 #define ETH_RSS_FRAG_IPV4 (1ULL << RTE_ETH_FLOW_FRAG_IPV4)
389 #define ETH_RSS_NONFRAG_IPV4_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_TCP)
390 #define ETH_RSS_NONFRAG_IPV4_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_UDP)
391 #define ETH_RSS_NONFRAG_IPV4_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_SCTP)
392 #define ETH_RSS_NONFRAG_IPV4_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV4_OTHER)
393 #define ETH_RSS_IPV6 (1ULL << RTE_ETH_FLOW_IPV6)
394 #define ETH_RSS_FRAG_IPV6 (1ULL << RTE_ETH_FLOW_FRAG_IPV6)
395 #define ETH_RSS_NONFRAG_IPV6_TCP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_TCP)
396 #define ETH_RSS_NONFRAG_IPV6_UDP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_UDP)
397 #define ETH_RSS_NONFRAG_IPV6_SCTP (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_SCTP)
398 #define ETH_RSS_NONFRAG_IPV6_OTHER (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_OTHER)
399 #define ETH_RSS_L2_PAYLOAD (1ULL << RTE_ETH_FLOW_L2_PAYLOAD)
400 #define ETH_RSS_IPV6_EX (1ULL << RTE_ETH_FLOW_IPV6_EX)
401 #define ETH_RSS_IPV6_TCP_EX (1ULL << RTE_ETH_FLOW_IPV6_TCP_EX)
402 #define ETH_RSS_IPV6_UDP_EX (1ULL << RTE_ETH_FLOW_IPV6_UDP_EX)
403 #define ETH_RSS_PORT (1ULL << RTE_ETH_FLOW_PORT)
404 #define ETH_RSS_VXLAN (1ULL << RTE_ETH_FLOW_VXLAN)
405 #define ETH_RSS_GENEVE (1ULL << RTE_ETH_FLOW_GENEVE)
406 #define ETH_RSS_NVGRE (1ULL << RTE_ETH_FLOW_NVGRE)
408 #define ETH_RSS_IP ( \
410 ETH_RSS_FRAG_IPV4 | \
411 ETH_RSS_NONFRAG_IPV4_OTHER | \
413 ETH_RSS_FRAG_IPV6 | \
414 ETH_RSS_NONFRAG_IPV6_OTHER | \
417 #define ETH_RSS_UDP ( \
418 ETH_RSS_NONFRAG_IPV4_UDP | \
419 ETH_RSS_NONFRAG_IPV6_UDP | \
422 #define ETH_RSS_TCP ( \
423 ETH_RSS_NONFRAG_IPV4_TCP | \
424 ETH_RSS_NONFRAG_IPV6_TCP | \
427 #define ETH_RSS_SCTP ( \
428 ETH_RSS_NONFRAG_IPV4_SCTP | \
429 ETH_RSS_NONFRAG_IPV6_SCTP)
431 #define ETH_RSS_TUNNEL ( \
436 /**< Mask of valid RSS hash protocols */
437 #define ETH_RSS_PROTO_MASK ( \
439 ETH_RSS_FRAG_IPV4 | \
440 ETH_RSS_NONFRAG_IPV4_TCP | \
441 ETH_RSS_NONFRAG_IPV4_UDP | \
442 ETH_RSS_NONFRAG_IPV4_SCTP | \
443 ETH_RSS_NONFRAG_IPV4_OTHER | \
445 ETH_RSS_FRAG_IPV6 | \
446 ETH_RSS_NONFRAG_IPV6_TCP | \
447 ETH_RSS_NONFRAG_IPV6_UDP | \
448 ETH_RSS_NONFRAG_IPV6_SCTP | \
449 ETH_RSS_NONFRAG_IPV6_OTHER | \
450 ETH_RSS_L2_PAYLOAD | \
452 ETH_RSS_IPV6_TCP_EX | \
453 ETH_RSS_IPV6_UDP_EX | \
460 * Definitions used for redirection table entry size.
461 * Some RSS RETA sizes may not be supported by some drivers, check the
462 * documentation or the description of relevant functions for more details.
464 #define ETH_RSS_RETA_SIZE_64 64
465 #define ETH_RSS_RETA_SIZE_128 128
466 #define ETH_RSS_RETA_SIZE_256 256
467 #define ETH_RSS_RETA_SIZE_512 512
468 #define RTE_RETA_GROUP_SIZE 64
470 /* Definitions used for VMDQ and DCB functionality */
471 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
472 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
473 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
474 #define ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */
476 /* DCB capability defines */
477 #define ETH_DCB_PG_SUPPORT 0x00000001 /**< Priority Group(ETS) support. */
478 #define ETH_DCB_PFC_SUPPORT 0x00000002 /**< Priority Flow Control support. */
480 /* Definitions used for VLAN Offload functionality */
481 #define ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */
482 #define ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */
483 #define ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */
485 /* Definitions used for mask VLAN setting */
486 #define ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */
487 #define ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/
488 #define ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/
489 #define ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/
491 /* Definitions used for receive MAC address */
492 #define ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */
494 /* Definitions used for unicast hash */
495 #define ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */
497 /* Definitions used for VMDQ pool rx mode setting */
498 #define ETH_VMDQ_ACCEPT_UNTAG 0x0001 /**< accept untagged packets. */
499 #define ETH_VMDQ_ACCEPT_HASH_MC 0x0002 /**< accept packets in multicast table . */
500 #define ETH_VMDQ_ACCEPT_HASH_UC 0x0004 /**< accept packets in unicast table. */
501 #define ETH_VMDQ_ACCEPT_BROADCAST 0x0008 /**< accept broadcast packets. */
502 #define ETH_VMDQ_ACCEPT_MULTICAST 0x0010 /**< multicast promiscuous. */
504 /** Maximum nb. of vlan per mirror rule */
505 #define ETH_MIRROR_MAX_VLANS 64
507 #define ETH_MIRROR_VIRTUAL_POOL_UP 0x01 /**< Virtual Pool uplink Mirroring. */
508 #define ETH_MIRROR_UPLINK_PORT 0x02 /**< Uplink Port Mirroring. */
509 #define ETH_MIRROR_DOWNLINK_PORT 0x04 /**< Downlink Port Mirroring. */
510 #define ETH_MIRROR_VLAN 0x08 /**< VLAN Mirroring. */
511 #define ETH_MIRROR_VIRTUAL_POOL_DOWN 0x10 /**< Virtual Pool downlink Mirroring. */
514 * A structure used to configure VLAN traffic mirror of an Ethernet port.
516 struct rte_eth_vlan_mirror {
517 uint64_t vlan_mask; /**< mask for valid VLAN ID. */
518 /** VLAN ID list for vlan mirroring. */
519 uint16_t vlan_id[ETH_MIRROR_MAX_VLANS];
523 * A structure used to configure traffic mirror of an Ethernet port.
525 struct rte_eth_mirror_conf {
526 uint8_t rule_type; /**< Mirroring rule type */
527 uint8_t dst_pool; /**< Destination pool for this mirror rule. */
528 uint64_t pool_mask; /**< Bitmap of pool for pool mirroring */
529 /** VLAN ID setting for VLAN mirroring. */
530 struct rte_eth_vlan_mirror vlan;
534 * A structure used to configure 64 entries of Redirection Table of the
535 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure
536 * more than 64 entries supported by hardware, an array of this structure
539 struct rte_eth_rss_reta_entry64 {
541 /**< Mask bits indicate which entries need to be updated/queried. */
542 uint16_t reta[RTE_RETA_GROUP_SIZE];
543 /**< Group of 64 redirection table entries. */
547 * This enum indicates the possible number of traffic classes
548 * in DCB configurations
550 enum rte_eth_nb_tcs {
551 ETH_4_TCS = 4, /**< 4 TCs with DCB. */
552 ETH_8_TCS = 8 /**< 8 TCs with DCB. */
556 * This enum indicates the possible number of queue pools
557 * in VMDQ configurations.
559 enum rte_eth_nb_pools {
560 ETH_8_POOLS = 8, /**< 8 VMDq pools. */
561 ETH_16_POOLS = 16, /**< 16 VMDq pools. */
562 ETH_32_POOLS = 32, /**< 32 VMDq pools. */
563 ETH_64_POOLS = 64 /**< 64 VMDq pools. */
566 /* This structure may be extended in future. */
567 struct rte_eth_dcb_rx_conf {
568 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */
569 /** Traffic class each UP mapped to. */
570 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
573 struct rte_eth_vmdq_dcb_tx_conf {
574 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */
575 /** Traffic class each UP mapped to. */
576 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
579 struct rte_eth_dcb_tx_conf {
580 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */
581 /** Traffic class each UP mapped to. */
582 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
585 struct rte_eth_vmdq_tx_conf {
586 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */
590 * A structure used to configure the VMDQ+DCB feature
591 * of an Ethernet port.
593 * Using this feature, packets are routed to a pool of queues, based
594 * on the vlan id in the vlan tag, and then to a specific queue within
595 * that pool, using the user priority vlan tag field.
597 * A default pool may be used, if desired, to route all traffic which
598 * does not match the vlan filter rules.
600 struct rte_eth_vmdq_dcb_conf {
601 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
602 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
603 uint8_t default_pool; /**< The default pool, if applicable */
604 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
606 uint16_t vlan_id; /**< The vlan id of the received frame */
607 uint64_t pools; /**< Bitmask of pools for packet rx */
608 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
609 uint8_t dcb_tc[ETH_DCB_NUM_USER_PRIORITIES];
610 /**< Selects a queue in a pool */
614 * A structure used to configure the VMDQ feature of an Ethernet port when
615 * not combined with the DCB feature.
617 * Using this feature, packets are routed to a pool of queues. By default,
618 * the pool selection is based on the MAC address, the vlan id in the
619 * vlan tag as specified in the pool_map array.
620 * Passing the ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool
621 * selection using only the MAC address. MAC address to pool mapping is done
622 * using the rte_eth_dev_mac_addr_add function, with the pool parameter
623 * corresponding to the pool id.
625 * Queue selection within the selected pool will be done using RSS when
626 * it is enabled or revert to the first queue of the pool if not.
628 * A default pool may be used, if desired, to route all traffic which
629 * does not match the vlan filter rules or any pool MAC address.
631 struct rte_eth_vmdq_rx_conf {
632 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */
633 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
634 uint8_t default_pool; /**< The default pool, if applicable */
635 uint8_t enable_loop_back; /**< Enable VT loop back */
636 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
637 uint32_t rx_mode; /**< Flags from ETH_VMDQ_ACCEPT_* */
639 uint16_t vlan_id; /**< The vlan id of the received frame */
640 uint64_t pools; /**< Bitmask of pools for packet rx */
641 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
645 * A structure used to configure the TX features of an Ethernet port.
647 struct rte_eth_txmode {
648 enum rte_eth_tx_mq_mode mq_mode; /**< TX multi-queues mode. */
650 * Per-port Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
651 * Only offloads set on tx_offload_capa field on rte_eth_dev_info
652 * structure are allowed to be set.
656 /* For i40e specifically */
659 uint8_t hw_vlan_reject_tagged : 1,
660 /**< If set, reject sending out tagged pkts */
661 hw_vlan_reject_untagged : 1,
662 /**< If set, reject sending out untagged pkts */
663 hw_vlan_insert_pvid : 1;
664 /**< If set, enable port based VLAN insertion */
668 * A structure used to configure an RX ring of an Ethernet port.
670 struct rte_eth_rxconf {
671 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
672 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
673 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */
674 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
676 * Per-queue Rx offloads to be set using DEV_RX_OFFLOAD_* flags.
677 * Only offloads set on rx_queue_offload_capa or rx_offload_capa
678 * fields on rte_eth_dev_info structure are allowed to be set.
684 * A structure used to configure a TX ring of an Ethernet port.
686 struct rte_eth_txconf {
687 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
688 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
689 uint16_t tx_free_thresh; /**< Start freeing TX buffers if there are
690 less free descriptors than this value. */
692 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */
694 * Per-queue Tx offloads to be set using DEV_TX_OFFLOAD_* flags.
695 * Only offloads set on tx_queue_offload_capa or tx_offload_capa
696 * fields on rte_eth_dev_info structure are allowed to be set.
702 * A structure contains information about HW descriptor ring limitations.
704 struct rte_eth_desc_lim {
705 uint16_t nb_max; /**< Max allowed number of descriptors. */
706 uint16_t nb_min; /**< Min allowed number of descriptors. */
707 uint16_t nb_align; /**< Number of descriptors should be aligned to. */
710 * Max allowed number of segments per whole packet.
712 * - For TSO packet this is the total number of data descriptors allowed
715 * @see nb_mtu_seg_max
720 * Max number of segments per one MTU.
722 * - For non-TSO packet, this is the maximum allowed number of segments
723 * in a single transmit packet.
725 * - For TSO packet each segment within the TSO may span up to this
730 uint16_t nb_mtu_seg_max;
734 * This enum indicates the flow control mode
736 enum rte_eth_fc_mode {
737 RTE_FC_NONE = 0, /**< Disable flow control. */
738 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
739 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
740 RTE_FC_FULL /**< Enable flow control on both side. */
744 * A structure used to configure Ethernet flow control parameter.
745 * These parameters will be configured into the register of the NIC.
746 * Please refer to the corresponding data sheet for proper value.
748 struct rte_eth_fc_conf {
749 uint32_t high_water; /**< High threshold value to trigger XOFF */
750 uint32_t low_water; /**< Low threshold value to trigger XON */
751 uint16_t pause_time; /**< Pause quota in the Pause frame */
752 uint16_t send_xon; /**< Is XON frame need be sent */
753 enum rte_eth_fc_mode mode; /**< Link flow control mode */
754 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */
755 uint8_t autoneg; /**< Use Pause autoneg */
759 * A structure used to configure Ethernet priority flow control parameter.
760 * These parameters will be configured into the register of the NIC.
761 * Please refer to the corresponding data sheet for proper value.
763 struct rte_eth_pfc_conf {
764 struct rte_eth_fc_conf fc; /**< General flow control parameter. */
765 uint8_t priority; /**< VLAN User Priority. */
769 * Memory space that can be configured to store Flow Director filters
770 * in the board memory.
772 enum rte_fdir_pballoc_type {
773 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
774 RTE_FDIR_PBALLOC_128K, /**< 128k. */
775 RTE_FDIR_PBALLOC_256K, /**< 256k. */
779 * Select report mode of FDIR hash information in RX descriptors.
781 enum rte_fdir_status_mode {
782 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
783 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
784 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
788 * A structure used to configure the Flow Director (FDIR) feature
789 * of an Ethernet port.
791 * If mode is RTE_FDIR_DISABLE, the pballoc value is ignored.
793 struct rte_fdir_conf {
794 enum rte_fdir_mode mode; /**< Flow Director mode. */
795 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
796 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
797 /** RX queue of packets matching a "drop" filter in perfect mode. */
799 struct rte_eth_fdir_masks mask;
800 struct rte_eth_fdir_flex_conf flex_conf;
801 /**< Flex payload configuration. */
805 * UDP tunneling configuration.
806 * Used to config the UDP port for a type of tunnel.
807 * NICs need the UDP port to identify the tunnel type.
808 * Normally a type of tunnel has a default UDP port, this structure can be used
809 * in case if the users want to change or support more UDP port.
811 struct rte_eth_udp_tunnel {
812 uint16_t udp_port; /**< UDP port used for the tunnel. */
813 uint8_t prot_type; /**< Tunnel type. Defined in rte_eth_tunnel_type. */
817 * A structure used to enable/disable specific device interrupts.
819 struct rte_intr_conf {
820 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
822 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */
824 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */
829 * A structure used to configure an Ethernet port.
830 * Depending upon the RX multi-queue mode, extra advanced
831 * configuration settings may be needed.
833 struct rte_eth_conf {
834 uint32_t link_speeds; /**< bitmap of ETH_LINK_SPEED_XXX of speeds to be
835 used. ETH_LINK_SPEED_FIXED disables link
836 autonegotiation, and a unique speed shall be
837 set. Otherwise, the bitmap defines the set of
838 speeds to be advertised. If the special value
839 ETH_LINK_SPEED_AUTONEG (0) is used, all speeds
840 supported are advertised. */
841 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
842 struct rte_eth_txmode txmode; /**< Port TX configuration. */
843 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value
844 is 0, meaning the loopback mode is disabled.
845 Read the datasheet of given ethernet controller
846 for details. The possible values of this field
847 are defined in implementation of each driver. */
849 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
850 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
851 /**< Port vmdq+dcb configuration. */
852 struct rte_eth_dcb_rx_conf dcb_rx_conf;
853 /**< Port dcb RX configuration. */
854 struct rte_eth_vmdq_rx_conf vmdq_rx_conf;
855 /**< Port vmdq RX configuration. */
856 } rx_adv_conf; /**< Port RX filtering configuration. */
858 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf;
859 /**< Port vmdq+dcb TX configuration. */
860 struct rte_eth_dcb_tx_conf dcb_tx_conf;
861 /**< Port dcb TX configuration. */
862 struct rte_eth_vmdq_tx_conf vmdq_tx_conf;
863 /**< Port vmdq TX configuration. */
864 } tx_adv_conf; /**< Port TX DCB configuration (union). */
865 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC
866 is needed,and the variable must be set ETH_DCB_PFC_SUPPORT. */
867 uint32_t dcb_capability_en;
868 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. */
869 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
873 * RX offload capabilities of a device.
875 #define DEV_RX_OFFLOAD_VLAN_STRIP 0x00000001
876 #define DEV_RX_OFFLOAD_IPV4_CKSUM 0x00000002
877 #define DEV_RX_OFFLOAD_UDP_CKSUM 0x00000004
878 #define DEV_RX_OFFLOAD_TCP_CKSUM 0x00000008
879 #define DEV_RX_OFFLOAD_TCP_LRO 0x00000010
880 #define DEV_RX_OFFLOAD_QINQ_STRIP 0x00000020
881 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000040
882 #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x00000080
883 #define DEV_RX_OFFLOAD_HEADER_SPLIT 0x00000100
884 #define DEV_RX_OFFLOAD_VLAN_FILTER 0x00000200
885 #define DEV_RX_OFFLOAD_VLAN_EXTEND 0x00000400
886 #define DEV_RX_OFFLOAD_JUMBO_FRAME 0x00000800
887 #define DEV_RX_OFFLOAD_SCATTER 0x00002000
888 #define DEV_RX_OFFLOAD_TIMESTAMP 0x00004000
889 #define DEV_RX_OFFLOAD_SECURITY 0x00008000
890 #define DEV_RX_OFFLOAD_KEEP_CRC 0x00010000
891 #define DEV_RX_OFFLOAD_SCTP_CKSUM 0x00020000
892 #define DEV_RX_OFFLOAD_OUTER_UDP_CKSUM 0x00040000
894 #define DEV_RX_OFFLOAD_CHECKSUM (DEV_RX_OFFLOAD_IPV4_CKSUM | \
895 DEV_RX_OFFLOAD_UDP_CKSUM | \
896 DEV_RX_OFFLOAD_TCP_CKSUM)
897 #define DEV_RX_OFFLOAD_VLAN (DEV_RX_OFFLOAD_VLAN_STRIP | \
898 DEV_RX_OFFLOAD_VLAN_FILTER | \
899 DEV_RX_OFFLOAD_VLAN_EXTEND)
902 * If new Rx offload capabilities are defined, they also must be
903 * mentioned in rte_rx_offload_names in rte_ethdev.c file.
907 * TX offload capabilities of a device.
909 #define DEV_TX_OFFLOAD_VLAN_INSERT 0x00000001
910 #define DEV_TX_OFFLOAD_IPV4_CKSUM 0x00000002
911 #define DEV_TX_OFFLOAD_UDP_CKSUM 0x00000004
912 #define DEV_TX_OFFLOAD_TCP_CKSUM 0x00000008
913 #define DEV_TX_OFFLOAD_SCTP_CKSUM 0x00000010
914 #define DEV_TX_OFFLOAD_TCP_TSO 0x00000020
915 #define DEV_TX_OFFLOAD_UDP_TSO 0x00000040
916 #define DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM 0x00000080 /**< Used for tunneling packet. */
917 #define DEV_TX_OFFLOAD_QINQ_INSERT 0x00000100
918 #define DEV_TX_OFFLOAD_VXLAN_TNL_TSO 0x00000200 /**< Used for tunneling packet. */
919 #define DEV_TX_OFFLOAD_GRE_TNL_TSO 0x00000400 /**< Used for tunneling packet. */
920 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x00000800 /**< Used for tunneling packet. */
921 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO 0x00001000 /**< Used for tunneling packet. */
922 #define DEV_TX_OFFLOAD_MACSEC_INSERT 0x00002000
923 #define DEV_TX_OFFLOAD_MT_LOCKFREE 0x00004000
924 /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
925 * tx queue without SW lock.
927 #define DEV_TX_OFFLOAD_MULTI_SEGS 0x00008000
928 /**< Device supports multi segment send. */
929 #define DEV_TX_OFFLOAD_MBUF_FAST_FREE 0x00010000
930 /**< Device supports optimization for fast release of mbufs.
931 * When set application must guarantee that per-queue all mbufs comes from
932 * the same mempool and has refcnt = 1.
934 #define DEV_TX_OFFLOAD_SECURITY 0x00020000
936 * Device supports generic UDP tunneled packet TSO.
937 * Application must set PKT_TX_TUNNEL_UDP and other mbuf fields required
940 #define DEV_TX_OFFLOAD_UDP_TNL_TSO 0x00040000
942 * Device supports generic IP tunneled packet TSO.
943 * Application must set PKT_TX_TUNNEL_IP and other mbuf fields required
946 #define DEV_TX_OFFLOAD_IP_TNL_TSO 0x00080000
947 /** Device supports outer UDP checksum */
948 #define DEV_TX_OFFLOAD_OUTER_UDP_CKSUM 0x00100000
950 * Device supports match on metadata Tx offload..
951 * Application must set PKT_TX_METADATA and mbuf metadata field.
953 #define DEV_TX_OFFLOAD_MATCH_METADATA 0x00200000
955 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
956 /**< Device supports Rx queue setup after device started*/
957 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
958 /**< Device supports Tx queue setup after device started*/
961 * If new Tx offload capabilities are defined, they also must be
962 * mentioned in rte_tx_offload_names in rte_ethdev.c file.
966 * Fallback default preferred Rx/Tx port parameters.
967 * These are used if an application requests default parameters
968 * but the PMD does not provide preferred values.
970 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512
971 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512
972 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1
973 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1
976 * Preferred Rx/Tx port parameters.
977 * There are separate instances of this structure for transmission
978 * and reception respectively.
980 struct rte_eth_dev_portconf {
981 uint16_t burst_size; /**< Device-preferred burst size */
982 uint16_t ring_size; /**< Device-preferred size of queue rings */
983 uint16_t nb_queues; /**< Device-preferred number of queues */
987 * Default values for switch domain id when ethdev does not support switch
988 * domain definitions.
990 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (0)
993 * Ethernet device associated switch information
995 struct rte_eth_switch_info {
996 const char *name; /**< switch name */
997 uint16_t domain_id; /**< switch domain id */
1000 * mapping to the devices physical switch port as enumerated from the
1001 * perspective of the embedded interconnect/switch. For SR-IOV enabled
1002 * device this may correspond to the VF_ID of each virtual function,
1003 * but each driver should explicitly define the mapping of switch
1004 * port identifier to that physical interconnect/switch
1009 * Ethernet device information
1013 * A structure used to retrieve the contextual information of
1014 * an Ethernet device, such as the controlling driver of the
1017 struct rte_eth_dev_info {
1018 struct rte_device *device; /** Generic device information */
1019 const char *driver_name; /**< Device Driver name. */
1020 unsigned int if_index; /**< Index to bound host interface, or 0 if none.
1021 Use if_indextoname() to translate into an interface name. */
1022 const uint32_t *dev_flags; /**< Device flags */
1023 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
1024 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
1025 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
1026 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
1027 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
1028 uint32_t max_hash_mac_addrs;
1029 /** Maximum number of hash MAC addresses for MTA and UTA. */
1030 uint16_t max_vfs; /**< Maximum number of VFs. */
1031 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */
1032 uint64_t rx_offload_capa;
1033 /**< All RX offload capabilities including all per-queue ones */
1034 uint64_t tx_offload_capa;
1035 /**< All TX offload capabilities including all per-queue ones */
1036 uint64_t rx_queue_offload_capa;
1037 /**< Device per-queue RX offload capabilities. */
1038 uint64_t tx_queue_offload_capa;
1039 /**< Device per-queue TX offload capabilities. */
1041 /**< Device redirection table size, the total number of entries. */
1042 uint8_t hash_key_size; /**< Hash key size in bytes */
1043 /** Bit mask of RSS offloads, the bit offset also means flow type */
1044 uint64_t flow_type_rss_offloads;
1045 struct rte_eth_rxconf default_rxconf; /**< Default RX configuration */
1046 struct rte_eth_txconf default_txconf; /**< Default TX configuration */
1047 uint16_t vmdq_queue_base; /**< First queue ID for VMDQ pools. */
1048 uint16_t vmdq_queue_num; /**< Queue number for VMDQ pools. */
1049 uint16_t vmdq_pool_base; /**< First ID of VMDQ pools. */
1050 struct rte_eth_desc_lim rx_desc_lim; /**< RX descriptors limits */
1051 struct rte_eth_desc_lim tx_desc_lim; /**< TX descriptors limits */
1052 uint32_t speed_capa; /**< Supported speeds bitmap (ETH_LINK_SPEED_). */
1053 /** Configured number of rx/tx queues */
1054 uint16_t nb_rx_queues; /**< Number of RX queues. */
1055 uint16_t nb_tx_queues; /**< Number of TX queues. */
1056 /** Rx parameter recommendations */
1057 struct rte_eth_dev_portconf default_rxportconf;
1058 /** Tx parameter recommendations */
1059 struct rte_eth_dev_portconf default_txportconf;
1060 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */
1063 * Switching information for ports on a device with a
1064 * embedded managed interconnect/switch.
1066 struct rte_eth_switch_info switch_info;
1070 * Ethernet device RX queue information structure.
1071 * Used to retieve information about configured queue.
1073 struct rte_eth_rxq_info {
1074 struct rte_mempool *mp; /**< mempool used by that queue. */
1075 struct rte_eth_rxconf conf; /**< queue config parameters. */
1076 uint8_t scattered_rx; /**< scattered packets RX supported. */
1077 uint16_t nb_desc; /**< configured number of RXDs. */
1078 } __rte_cache_min_aligned;
1081 * Ethernet device TX queue information structure.
1082 * Used to retrieve information about configured queue.
1084 struct rte_eth_txq_info {
1085 struct rte_eth_txconf conf; /**< queue config parameters. */
1086 uint16_t nb_desc; /**< configured number of TXDs. */
1087 } __rte_cache_min_aligned;
1089 /** Maximum name length for extended statistics counters */
1090 #define RTE_ETH_XSTATS_NAME_SIZE 64
1093 * An Ethernet device extended statistic structure
1095 * This structure is used by rte_eth_xstats_get() to provide
1096 * statistics that are not provided in the generic *rte_eth_stats*
1098 * It maps a name id, corresponding to an index in the array returned
1099 * by rte_eth_xstats_get_names(), to a statistic value.
1101 struct rte_eth_xstat {
1102 uint64_t id; /**< The index in xstats name array. */
1103 uint64_t value; /**< The statistic counter value. */
1107 * A name element for extended statistics.
1109 * An array of this structure is returned by rte_eth_xstats_get_names().
1110 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat*
1111 * structure references these names by their array index.
1113 struct rte_eth_xstat_name {
1114 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */
1117 #define ETH_DCB_NUM_TCS 8
1118 #define ETH_MAX_VMDQ_POOL 64
1121 * A structure used to get the information of queue and
1122 * TC mapping on both TX and RX paths.
1124 struct rte_eth_dcb_tc_queue_mapping {
1125 /** rx queues assigned to tc per Pool */
1129 } tc_rxq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1130 /** rx queues assigned to tc per Pool */
1134 } tc_txq[ETH_MAX_VMDQ_POOL][ETH_DCB_NUM_TCS];
1138 * A structure used to get the information of DCB.
1139 * It includes TC UP mapping and queue TC mapping.
1141 struct rte_eth_dcb_info {
1142 uint8_t nb_tcs; /**< number of TCs */
1143 uint8_t prio_tc[ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */
1144 uint8_t tc_bws[ETH_DCB_NUM_TCS]; /**< TX BW percentage for each TC */
1145 /** rx queues assigned to tc */
1146 struct rte_eth_dcb_tc_queue_mapping tc_queue;
1150 * RX/TX queue states
1152 #define RTE_ETH_QUEUE_STATE_STOPPED 0
1153 #define RTE_ETH_QUEUE_STATE_STARTED 1
1155 #define RTE_ETH_ALL RTE_MAX_ETHPORTS
1157 /* Macros to check for valid port */
1158 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
1159 if (!rte_eth_dev_is_valid_port(port_id)) { \
1160 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1165 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
1166 if (!rte_eth_dev_is_valid_port(port_id)) { \
1167 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \
1173 * l2 tunnel configuration.
1176 /**< l2 tunnel enable mask */
1177 #define ETH_L2_TUNNEL_ENABLE_MASK 0x00000001
1178 /**< l2 tunnel insertion mask */
1179 #define ETH_L2_TUNNEL_INSERTION_MASK 0x00000002
1180 /**< l2 tunnel stripping mask */
1181 #define ETH_L2_TUNNEL_STRIPPING_MASK 0x00000004
1182 /**< l2 tunnel forwarding mask */
1183 #define ETH_L2_TUNNEL_FORWARDING_MASK 0x00000008
1186 * Function type used for RX packet processing packet callbacks.
1188 * The callback function is called on RX with a burst of packets that have
1189 * been received on the given port and queue.
1192 * The Ethernet port on which RX is being performed.
1194 * The queue on the Ethernet port which is being used to receive the packets.
1196 * The burst of packets that have just been received.
1198 * The number of packets in the burst pointed to by "pkts".
1200 * The max number of packets that can be stored in the "pkts" array.
1202 * The arbitrary user parameter passed in by the application when the callback
1203 * was originally configured.
1205 * The number of packets returned to the user.
1207 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue,
1208 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts,
1212 * Function type used for TX packet processing packet callbacks.
1214 * The callback function is called on TX with a burst of packets immediately
1215 * before the packets are put onto the hardware queue for transmission.
1218 * The Ethernet port on which TX is being performed.
1220 * The queue on the Ethernet port which is being used to transmit the packets.
1222 * The burst of packets that are about to be transmitted.
1224 * The number of packets in the burst pointed to by "pkts".
1226 * The arbitrary user parameter passed in by the application when the callback
1227 * was originally configured.
1229 * The number of packets to be written to the NIC.
1231 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue,
1232 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param);
1235 * Possible states of an ethdev port.
1237 enum rte_eth_dev_state {
1238 /** Device is unused before being probed. */
1239 RTE_ETH_DEV_UNUSED = 0,
1240 /** Device is attached when allocated in probing. */
1241 RTE_ETH_DEV_ATTACHED,
1242 /** The deferred state is useless and replaced by ownership. */
1243 RTE_ETH_DEV_DEFERRED,
1244 /** Device is in removed state when plug-out is detected. */
1245 RTE_ETH_DEV_REMOVED,
1248 struct rte_eth_dev_sriov {
1249 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */
1250 uint8_t nb_q_per_pool; /**< rx queue number per pool */
1251 uint16_t def_vmdq_idx; /**< Default pool num used for PF */
1252 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */
1254 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov)
1256 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN
1258 #define RTE_ETH_DEV_NO_OWNER 0
1260 #define RTE_ETH_MAX_OWNER_NAME_LEN 64
1262 struct rte_eth_dev_owner {
1263 uint64_t id; /**< The owner unique identifier. */
1264 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */
1268 * Port is released (i.e. totally freed and data erased) on close.
1269 * Temporary flag for PMD migration to new rte_eth_dev_close() behaviour.
1271 #define RTE_ETH_DEV_CLOSE_REMOVE 0x0001
1272 /** Device supports link state interrupt */
1273 #define RTE_ETH_DEV_INTR_LSC 0x0002
1274 /** Device is a bonded slave */
1275 #define RTE_ETH_DEV_BONDED_SLAVE 0x0004
1276 /** Device supports device removal interrupt */
1277 #define RTE_ETH_DEV_INTR_RMV 0x0008
1278 /** Device is port representor */
1279 #define RTE_ETH_DEV_REPRESENTOR 0x0010
1280 /** Device does not support MAC change after started */
1281 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR 0x0020
1284 * Iterates over valid ethdev ports owned by a specific owner.
1287 * The id of the next possible valid owned port.
1289 * The owner identifier.
1290 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports.
1292 * Next valid port id owned by owner_id, RTE_MAX_ETHPORTS if there is none.
1294 uint64_t rte_eth_find_next_owned_by(uint16_t port_id,
1295 const uint64_t owner_id);
1298 * Macro to iterate over all enabled ethdev ports owned by a specific owner.
1300 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \
1301 for (p = rte_eth_find_next_owned_by(0, o); \
1302 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \
1303 p = rte_eth_find_next_owned_by(p + 1, o))
1306 * Iterates over valid ethdev ports.
1309 * The id of the next possible valid port.
1311 * Next valid port id, RTE_MAX_ETHPORTS if there is none.
1313 uint16_t rte_eth_find_next(uint16_t port_id);
1316 * Macro to iterate over all enabled and ownerless ethdev ports.
1318 #define RTE_ETH_FOREACH_DEV(p) \
1319 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER)
1324 * @b EXPERIMENTAL: this API may change without prior notice.
1326 * Get a new unique owner identifier.
1327 * An owner identifier is used to owns Ethernet devices by only one DPDK entity
1328 * to avoid multiple management of device by different entities.
1331 * Owner identifier pointer.
1333 * Negative errno value on error, 0 on success.
1335 int __rte_experimental rte_eth_dev_owner_new(uint64_t *owner_id);
1339 * @b EXPERIMENTAL: this API may change without prior notice.
1341 * Set an Ethernet device owner.
1344 * The identifier of the port to own.
1346 * The owner pointer.
1348 * Negative errno value on error, 0 on success.
1350 int __rte_experimental rte_eth_dev_owner_set(const uint16_t port_id,
1351 const struct rte_eth_dev_owner *owner);
1355 * @b EXPERIMENTAL: this API may change without prior notice.
1357 * Unset Ethernet device owner to make the device ownerless.
1360 * The identifier of port to make ownerless.
1362 * The owner identifier.
1364 * 0 on success, negative errno value on error.
1366 int __rte_experimental rte_eth_dev_owner_unset(const uint16_t port_id,
1367 const uint64_t owner_id);
1371 * @b EXPERIMENTAL: this API may change without prior notice.
1373 * Remove owner from all Ethernet devices owned by a specific owner.
1376 * The owner identifier.
1378 void __rte_experimental rte_eth_dev_owner_delete(const uint64_t owner_id);
1382 * @b EXPERIMENTAL: this API may change without prior notice.
1384 * Get the owner of an Ethernet device.
1387 * The port identifier.
1389 * The owner structure pointer to fill.
1391 * 0 on success, negative errno value on error..
1393 int __rte_experimental rte_eth_dev_owner_get(const uint16_t port_id,
1394 struct rte_eth_dev_owner *owner);
1397 * Get the total number of Ethernet devices that have been successfully
1398 * initialized by the matching Ethernet driver during the PCI probing phase
1399 * and that are available for applications to use. These devices must be
1400 * accessed by using the ``RTE_ETH_FOREACH_DEV()`` macro to deal with
1401 * non-contiguous ranges of devices.
1402 * These non-contiguous ranges can be created by calls to hotplug functions or
1406 * - The total number of usable Ethernet devices.
1409 uint16_t rte_eth_dev_count(void);
1412 * Get the number of ports which are usable for the application.
1414 * These devices must be iterated by using the macro
1415 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY``
1416 * to deal with non-contiguous ranges of devices.
1419 * The count of available Ethernet devices.
1421 uint16_t rte_eth_dev_count_avail(void);
1424 * Get the total number of ports which are allocated.
1426 * Some devices may not be available for the application.
1429 * The total count of Ethernet devices.
1431 uint16_t __rte_experimental rte_eth_dev_count_total(void);
1434 * Attach a new Ethernet device specified by arguments.
1437 * A pointer to a strings array describing the new device
1438 * to be attached. The strings should be a pci address like
1439 * '0000:01:00.0' or virtual device name like 'net_pcap0'.
1441 * A pointer to a port identifier actually attached.
1443 * 0 on success and port_id is filled, negative on error
1446 int rte_eth_dev_attach(const char *devargs, uint16_t *port_id);
1449 * Detach a Ethernet device specified by port identifier.
1450 * This function must be called when the device is in the
1454 * The port identifier of the device to detach.
1456 * A pointer to a buffer that will be filled with the device name.
1457 * This buffer must be at least RTE_DEV_NAME_MAX_LEN long.
1459 * 0 on success and devname is filled, negative on error
1462 int rte_eth_dev_detach(uint16_t port_id, char *devname);
1465 * Convert a numerical speed in Mbps to a bitmap flag that can be used in
1466 * the bitmap link_speeds of the struct rte_eth_conf
1469 * Numerical speed value in Mbps
1471 * ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds)
1473 * 0 if the speed cannot be mapped
1475 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex);
1479 * @b EXPERIMENTAL: this API may change without prior notice
1481 * Get DEV_RX_OFFLOAD_* flag name.
1486 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1488 const char * __rte_experimental rte_eth_dev_rx_offload_name(uint64_t offload);
1492 * @b EXPERIMENTAL: this API may change without prior notice
1494 * Get DEV_TX_OFFLOAD_* flag name.
1499 * Offload name or 'UNKNOWN' if the flag cannot be recognised.
1501 const char * __rte_experimental rte_eth_dev_tx_offload_name(uint64_t offload);
1504 * Configure an Ethernet device.
1505 * This function must be invoked first before any other function in the
1506 * Ethernet API. This function can also be re-invoked when a device is in the
1510 * The port identifier of the Ethernet device to configure.
1511 * @param nb_rx_queue
1512 * The number of receive queues to set up for the Ethernet device.
1513 * @param nb_tx_queue
1514 * The number of transmit queues to set up for the Ethernet device.
1516 * The pointer to the configuration data to be used for the Ethernet device.
1517 * The *rte_eth_conf* structure includes:
1518 * - the hardware offload features to activate, with dedicated fields for
1519 * each statically configurable offload hardware feature provided by
1520 * Ethernet devices, such as IP checksum or VLAN tag stripping for
1522 * The Rx offload bitfield API is obsolete and will be deprecated.
1523 * Applications should set the ignore_bitfield_offloads bit on *rxmode*
1524 * structure and use offloads field to set per-port offloads instead.
1525 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within
1526 * the [rt]x_offload_capa returned from rte_eth_dev_infos_get().
1527 * Any type of device supported offloading set in the input argument
1528 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled
1529 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup()
1530 * - the Receive Side Scaling (RSS) configuration when using multiple RX
1531 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf
1532 * must be within the flow_type_rss_offloads provided by drivers via
1533 * rte_eth_dev_infos_get() API.
1535 * Embedding all configuration information in a single data structure
1536 * is the more flexible method that allows the addition of new features
1537 * without changing the syntax of the API.
1539 * - 0: Success, device configured.
1540 * - <0: Error code returned by the driver configuration function.
1542 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue,
1543 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf);
1547 * @b EXPERIMENTAL: this API may change without prior notice.
1549 * Check if an Ethernet device was physically removed.
1552 * The port identifier of the Ethernet device.
1554 * 1 when the Ethernet device is removed, otherwise 0.
1556 int __rte_experimental
1557 rte_eth_dev_is_removed(uint16_t port_id);
1560 * Allocate and set up a receive queue for an Ethernet device.
1562 * The function allocates a contiguous block of memory for *nb_rx_desc*
1563 * receive descriptors from a memory zone associated with *socket_id*
1564 * and initializes each receive descriptor with a network buffer allocated
1565 * from the memory pool *mb_pool*.
1568 * The port identifier of the Ethernet device.
1569 * @param rx_queue_id
1570 * The index of the receive queue to set up.
1571 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1572 * to rte_eth_dev_configure().
1574 * The number of receive descriptors to allocate for the receive ring.
1576 * The *socket_id* argument is the socket identifier in case of NUMA.
1577 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1578 * the DMA memory allocated for the receive descriptors of the ring.
1580 * The pointer to the configuration data to be used for the receive queue.
1581 * NULL value is allowed, in which case default RX configuration
1583 * The *rx_conf* structure contains an *rx_thresh* structure with the values
1584 * of the Prefetch, Host, and Write-Back threshold registers of the receive
1586 * In addition it contains the hardware offloads features to activate using
1587 * the DEV_RX_OFFLOAD_* flags.
1588 * If an offloading set in rx_conf->offloads
1589 * hasn't been set in the input argument eth_conf->rxmode.offloads
1590 * to rte_eth_dev_configure(), it is a new added offloading, it must be
1591 * per-queue type and it is enabled for the queue.
1592 * No need to repeat any bit in rx_conf->offloads which has already been
1593 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
1594 * at port level can't be disabled at queue level.
1596 * The pointer to the memory pool from which to allocate *rte_mbuf* network
1597 * memory buffers to populate each descriptor of the receive ring.
1599 * - 0: Success, receive queue correctly set up.
1600 * - -EIO: if device is removed.
1601 * - -EINVAL: The size of network buffers which can be allocated from the
1602 * memory pool does not fit the various buffer sizes allowed by the
1603 * device controller.
1604 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
1605 * allocate network memory buffers from the memory pool when
1606 * initializing receive descriptors.
1608 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
1609 uint16_t nb_rx_desc, unsigned int socket_id,
1610 const struct rte_eth_rxconf *rx_conf,
1611 struct rte_mempool *mb_pool);
1614 * Allocate and set up a transmit queue for an Ethernet device.
1617 * The port identifier of the Ethernet device.
1618 * @param tx_queue_id
1619 * The index of the transmit queue to set up.
1620 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1621 * to rte_eth_dev_configure().
1623 * The number of transmit descriptors to allocate for the transmit ring.
1625 * The *socket_id* argument is the socket identifier in case of NUMA.
1626 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1627 * the DMA memory allocated for the transmit descriptors of the ring.
1629 * The pointer to the configuration data to be used for the transmit queue.
1630 * NULL value is allowed, in which case default TX configuration
1632 * The *tx_conf* structure contains the following data:
1633 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
1634 * Write-Back threshold registers of the transmit ring.
1635 * When setting Write-Back threshold to the value greater then zero,
1636 * *tx_rs_thresh* value should be explicitly set to one.
1637 * - The *tx_free_thresh* value indicates the [minimum] number of network
1638 * buffers that must be pending in the transmit ring to trigger their
1639 * [implicit] freeing by the driver transmit function.
1640 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
1641 * descriptors that must be pending in the transmit ring before setting the
1642 * RS bit on a descriptor by the driver transmit function.
1643 * The *tx_rs_thresh* value should be less or equal then
1644 * *tx_free_thresh* value, and both of them should be less then
1646 * - The *offloads* member contains Tx offloads to be enabled.
1647 * If an offloading set in tx_conf->offloads
1648 * hasn't been set in the input argument eth_conf->txmode.offloads
1649 * to rte_eth_dev_configure(), it is a new added offloading, it must be
1650 * per-queue type and it is enabled for the queue.
1651 * No need to repeat any bit in tx_conf->offloads which has already been
1652 * enabled in rte_eth_dev_configure() at port level. An offloading enabled
1653 * at port level can't be disabled at queue level.
1655 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
1656 * the transmit function to use default values.
1658 * - 0: Success, the transmit queue is correctly set up.
1659 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
1661 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,
1662 uint16_t nb_tx_desc, unsigned int socket_id,
1663 const struct rte_eth_txconf *tx_conf);
1666 * Return the NUMA socket to which an Ethernet device is connected
1669 * The port identifier of the Ethernet device
1671 * The NUMA socket id to which the Ethernet device is connected or
1672 * a default of zero if the socket could not be determined.
1673 * -1 is returned is the port_id value is out of range.
1675 int rte_eth_dev_socket_id(uint16_t port_id);
1678 * Check if port_id of device is attached
1681 * The port identifier of the Ethernet device
1683 * - 0 if port is out of range or not attached
1684 * - 1 if device is attached
1686 int rte_eth_dev_is_valid_port(uint16_t port_id);
1689 * Start specified RX queue of a port. It is used when rx_deferred_start
1690 * flag of the specified queue is true.
1693 * The port identifier of the Ethernet device
1694 * @param rx_queue_id
1695 * The index of the rx queue to update the ring.
1696 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1697 * to rte_eth_dev_configure().
1699 * - 0: Success, the receive queue is started.
1700 * - -EINVAL: The port_id or the queue_id out of range.
1701 * - -EIO: if device is removed.
1702 * - -ENOTSUP: The function not supported in PMD driver.
1704 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id);
1707 * Stop specified RX queue of a port
1710 * The port identifier of the Ethernet device
1711 * @param rx_queue_id
1712 * The index of the rx queue to update the ring.
1713 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1714 * to rte_eth_dev_configure().
1716 * - 0: Success, the receive queue is stopped.
1717 * - -EINVAL: The port_id or the queue_id out of range.
1718 * - -EIO: if device is removed.
1719 * - -ENOTSUP: The function not supported in PMD driver.
1721 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id);
1724 * Start TX for specified queue of a port. It is used when tx_deferred_start
1725 * flag of the specified queue is true.
1728 * The port identifier of the Ethernet device
1729 * @param tx_queue_id
1730 * The index of the tx queue to update the ring.
1731 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1732 * to rte_eth_dev_configure().
1734 * - 0: Success, the transmit queue is started.
1735 * - -EINVAL: The port_id or the queue_id out of range.
1736 * - -EIO: if device is removed.
1737 * - -ENOTSUP: The function not supported in PMD driver.
1739 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id);
1742 * Stop specified TX queue of a port
1745 * The port identifier of the Ethernet device
1746 * @param tx_queue_id
1747 * The index of the tx queue to update the ring.
1748 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1749 * to rte_eth_dev_configure().
1751 * - 0: Success, the transmit queue is stopped.
1752 * - -EINVAL: The port_id or the queue_id out of range.
1753 * - -EIO: if device is removed.
1754 * - -ENOTSUP: The function not supported in PMD driver.
1756 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id);
1759 * Start an Ethernet device.
1761 * The device start step is the last one and consists of setting the configured
1762 * offload features and in starting the transmit and the receive units of the
1765 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before
1766 * PMD port start callback function is invoked.
1768 * On success, all basic functions exported by the Ethernet API (link status,
1769 * receive/transmit, and so on) can be invoked.
1772 * The port identifier of the Ethernet device.
1774 * - 0: Success, Ethernet device started.
1775 * - <0: Error code of the driver device start function.
1777 int rte_eth_dev_start(uint16_t port_id);
1780 * Stop an Ethernet device. The device can be restarted with a call to
1781 * rte_eth_dev_start()
1784 * The port identifier of the Ethernet device.
1786 void rte_eth_dev_stop(uint16_t port_id);
1789 * Link up an Ethernet device.
1791 * Set device link up will re-enable the device rx/tx
1792 * functionality after it is previously set device linked down.
1795 * The port identifier of the Ethernet device.
1797 * - 0: Success, Ethernet device linked up.
1798 * - <0: Error code of the driver device link up function.
1800 int rte_eth_dev_set_link_up(uint16_t port_id);
1803 * Link down an Ethernet device.
1804 * The device rx/tx functionality will be disabled if success,
1805 * and it can be re-enabled with a call to
1806 * rte_eth_dev_set_link_up()
1809 * The port identifier of the Ethernet device.
1811 int rte_eth_dev_set_link_down(uint16_t port_id);
1814 * Close a stopped Ethernet device. The device cannot be restarted!
1815 * The function frees all port resources if the driver supports
1816 * the flag RTE_ETH_DEV_CLOSE_REMOVE.
1819 * The port identifier of the Ethernet device.
1821 void rte_eth_dev_close(uint16_t port_id);
1824 * Reset a Ethernet device and keep its port id.
1826 * When a port has to be reset passively, the DPDK application can invoke
1827 * this function. For example when a PF is reset, all its VFs should also
1828 * be reset. Normally a DPDK application can invoke this function when
1829 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start
1830 * a port reset in other circumstances.
1832 * When this function is called, it first stops the port and then calls the
1833 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial
1834 * state, in which no Tx and Rx queues are setup, as if the port has been
1835 * reset and not started. The port keeps the port id it had before the
1838 * After calling rte_eth_dev_reset( ), the application should use
1839 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ),
1840 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( )
1841 * to reconfigure the device as appropriate.
1843 * Note: To avoid unexpected behavior, the application should stop calling
1844 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread
1845 * safety, all these controlling functions should be called from the same
1849 * The port identifier of the Ethernet device.
1852 * - (0) if successful.
1853 * - (-EINVAL) if port identifier is invalid.
1854 * - (-ENOTSUP) if hardware doesn't support this function.
1855 * - (-EPERM) if not ran from the primary process.
1856 * - (-EIO) if re-initialisation failed or device is removed.
1857 * - (-ENOMEM) if the reset failed due to OOM.
1858 * - (-EAGAIN) if the reset temporarily failed and should be retried later.
1860 int rte_eth_dev_reset(uint16_t port_id);
1863 * Enable receipt in promiscuous mode for an Ethernet device.
1866 * The port identifier of the Ethernet device.
1868 void rte_eth_promiscuous_enable(uint16_t port_id);
1871 * Disable receipt in promiscuous mode for an Ethernet device.
1874 * The port identifier of the Ethernet device.
1876 void rte_eth_promiscuous_disable(uint16_t port_id);
1879 * Return the value of promiscuous mode for an Ethernet device.
1882 * The port identifier of the Ethernet device.
1884 * - (1) if promiscuous is enabled
1885 * - (0) if promiscuous is disabled.
1888 int rte_eth_promiscuous_get(uint16_t port_id);
1891 * Enable the receipt of any multicast frame by an Ethernet device.
1894 * The port identifier of the Ethernet device.
1896 void rte_eth_allmulticast_enable(uint16_t port_id);
1899 * Disable the receipt of all multicast frames by an Ethernet device.
1902 * The port identifier of the Ethernet device.
1904 void rte_eth_allmulticast_disable(uint16_t port_id);
1907 * Return the value of allmulticast mode for an Ethernet device.
1910 * The port identifier of the Ethernet device.
1912 * - (1) if allmulticast is enabled
1913 * - (0) if allmulticast is disabled.
1916 int rte_eth_allmulticast_get(uint16_t port_id);
1919 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1920 * or FULL-DUPLEX) of the physical link of an Ethernet device. It might need
1921 * to wait up to 9 seconds in it.
1924 * The port identifier of the Ethernet device.
1926 * A pointer to an *rte_eth_link* structure to be filled with
1927 * the status, the speed and the mode of the Ethernet device link.
1929 void rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link);
1932 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1933 * or FULL-DUPLEX) of the physical link of an Ethernet device. It is a no-wait
1934 * version of rte_eth_link_get().
1937 * The port identifier of the Ethernet device.
1939 * A pointer to an *rte_eth_link* structure to be filled with
1940 * the status, the speed and the mode of the Ethernet device link.
1942 void rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link);
1945 * Retrieve the general I/O statistics of an Ethernet device.
1948 * The port identifier of the Ethernet device.
1950 * A pointer to a structure of type *rte_eth_stats* to be filled with
1951 * the values of device counters for the following set of statistics:
1952 * - *ipackets* with the total of successfully received packets.
1953 * - *opackets* with the total of successfully transmitted packets.
1954 * - *ibytes* with the total of successfully received bytes.
1955 * - *obytes* with the total of successfully transmitted bytes.
1956 * - *ierrors* with the total of erroneous received packets.
1957 * - *oerrors* with the total of failed transmitted packets.
1959 * Zero if successful. Non-zero otherwise.
1961 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats);
1964 * Reset the general I/O statistics of an Ethernet device.
1967 * The port identifier of the Ethernet device.
1969 * - (0) if device notified to reset stats.
1970 * - (-ENOTSUP) if hardware doesn't support.
1971 * - (-ENODEV) if *port_id* invalid.
1973 int rte_eth_stats_reset(uint16_t port_id);
1976 * Retrieve names of extended statistics of an Ethernet device.
1978 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
1980 * xstats_names[i].name => xstats[i].value
1982 * And the array index is same with id field of 'struct rte_eth_xstat':
1985 * This assumption makes key-value pair matching less flexible but simpler.
1988 * The port identifier of the Ethernet device.
1989 * @param xstats_names
1990 * An rte_eth_xstat_name array of at least *size* elements to
1991 * be filled. If set to NULL, the function returns the required number
1994 * The size of the xstats_names array (number of elements).
1996 * - A positive value lower or equal to size: success. The return value
1997 * is the number of entries filled in the stats table.
1998 * - A positive value higher than size: error, the given statistics table
1999 * is too small. The return value corresponds to the size that should
2000 * be given to succeed. The entries in the table are not valid and
2001 * shall not be used by the caller.
2002 * - A negative value on error (invalid port id).
2004 int rte_eth_xstats_get_names(uint16_t port_id,
2005 struct rte_eth_xstat_name *xstats_names,
2009 * Retrieve extended statistics of an Ethernet device.
2011 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched
2013 * xstats_names[i].name => xstats[i].value
2015 * And the array index is same with id field of 'struct rte_eth_xstat':
2018 * This assumption makes key-value pair matching less flexible but simpler.
2021 * The port identifier of the Ethernet device.
2023 * A pointer to a table of structure of type *rte_eth_xstat*
2024 * to be filled with device statistics ids and values.
2025 * This parameter can be set to NULL if n is 0.
2027 * The size of the xstats array (number of elements).
2029 * - A positive value lower or equal to n: success. The return value
2030 * is the number of entries filled in the stats table.
2031 * - A positive value higher than n: error, the given statistics table
2032 * is too small. The return value corresponds to the size that should
2033 * be given to succeed. The entries in the table are not valid and
2034 * shall not be used by the caller.
2035 * - A negative value on error (invalid port id).
2037 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats,
2041 * Retrieve names of extended statistics of an Ethernet device.
2044 * The port identifier of the Ethernet device.
2045 * @param xstats_names
2046 * An rte_eth_xstat_name array of at least *size* elements to
2047 * be filled. If set to NULL, the function returns the required number
2050 * IDs array given by app to retrieve specific statistics
2052 * The size of the xstats_names array (number of elements).
2054 * - A positive value lower or equal to size: success. The return value
2055 * is the number of entries filled in the stats table.
2056 * - A positive value higher than size: error, the given statistics table
2057 * is too small. The return value corresponds to the size that should
2058 * be given to succeed. The entries in the table are not valid and
2059 * shall not be used by the caller.
2060 * - A negative value on error (invalid port id).
2063 rte_eth_xstats_get_names_by_id(uint16_t port_id,
2064 struct rte_eth_xstat_name *xstats_names, unsigned int size,
2068 * Retrieve extended statistics of an Ethernet device.
2071 * The port identifier of the Ethernet device.
2073 * A pointer to an ids array passed by application. This tells which
2074 * statistics values function should retrieve. This parameter
2075 * can be set to NULL if size is 0. In this case function will retrieve
2076 * all avalible statistics.
2078 * A pointer to a table to be filled with device statistics values.
2080 * The size of the ids array (number of elements).
2082 * - A positive value lower or equal to size: success. The return value
2083 * is the number of entries filled in the stats table.
2084 * - A positive value higher than size: error, the given statistics table
2085 * is too small. The return value corresponds to the size that should
2086 * be given to succeed. The entries in the table are not valid and
2087 * shall not be used by the caller.
2088 * - A negative value on error (invalid port id).
2090 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids,
2091 uint64_t *values, unsigned int size);
2094 * Gets the ID of a statistic from its name.
2096 * This function searches for the statistics using string compares, and
2097 * as such should not be used on the fast-path. For fast-path retrieval of
2098 * specific statistics, store the ID as provided in *id* from this function,
2099 * and pass the ID to rte_eth_xstats_get()
2101 * @param port_id The port to look up statistics from
2102 * @param xstat_name The name of the statistic to return
2103 * @param[out] id A pointer to an app-supplied uint64_t which should be
2104 * set to the ID of the stat if the stat exists.
2107 * -ENODEV for invalid port_id,
2108 * -EIO if device is removed,
2109 * -EINVAL if the xstat_name doesn't exist in port_id
2111 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name,
2115 * Reset extended statistics of an Ethernet device.
2118 * The port identifier of the Ethernet device.
2120 void rte_eth_xstats_reset(uint16_t port_id);
2123 * Set a mapping for the specified transmit queue to the specified per-queue
2124 * statistics counter.
2127 * The port identifier of the Ethernet device.
2128 * @param tx_queue_id
2129 * The index of the transmit queue for which a queue stats mapping is required.
2130 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2131 * to rte_eth_dev_configure().
2133 * The per-queue packet statistics functionality number that the transmit
2134 * queue is to be assigned.
2135 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2137 * Zero if successful. Non-zero otherwise.
2139 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id,
2140 uint16_t tx_queue_id, uint8_t stat_idx);
2143 * Set a mapping for the specified receive queue to the specified per-queue
2144 * statistics counter.
2147 * The port identifier of the Ethernet device.
2148 * @param rx_queue_id
2149 * The index of the receive queue for which a queue stats mapping is required.
2150 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2151 * to rte_eth_dev_configure().
2153 * The per-queue packet statistics functionality number that the receive
2154 * queue is to be assigned.
2155 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1].
2157 * Zero if successful. Non-zero otherwise.
2159 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id,
2160 uint16_t rx_queue_id,
2164 * Retrieve the Ethernet address of an Ethernet device.
2167 * The port identifier of the Ethernet device.
2169 * A pointer to a structure of type *ether_addr* to be filled with
2170 * the Ethernet address of the Ethernet device.
2172 void rte_eth_macaddr_get(uint16_t port_id, struct ether_addr *mac_addr);
2175 * Retrieve the contextual information of an Ethernet device.
2178 * The port identifier of the Ethernet device.
2180 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
2181 * the contextual information of the Ethernet device.
2183 void rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info);
2186 * Retrieve the firmware version of a device.
2189 * The port identifier of the device.
2191 * A pointer to a string array storing the firmware version of a device,
2192 * the string includes terminating null. This pointer is allocated by caller.
2194 * The size of the string array pointed by fw_version, which should be
2195 * large enough to store firmware version of the device.
2197 * - (0) if successful.
2198 * - (-ENOTSUP) if operation is not supported.
2199 * - (-ENODEV) if *port_id* invalid.
2200 * - (-EIO) if device is removed.
2201 * - (>0) if *fw_size* is not enough to store firmware version, return
2202 * the size of the non truncated string.
2204 int rte_eth_dev_fw_version_get(uint16_t port_id,
2205 char *fw_version, size_t fw_size);
2208 * Retrieve the supported packet types of an Ethernet device.
2210 * When a packet type is announced as supported, it *must* be recognized by
2211 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN
2212 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following
2213 * packet types for these packets:
2214 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4
2215 * - Ether/Vlan/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4
2216 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER
2217 * - Ether/Vlan/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN
2219 * When a packet is received by a PMD, the most precise type must be
2220 * returned among the ones supported. However a PMD is allowed to set
2221 * packet type that is not in the supported list, at the condition that it
2222 * is more precise. Therefore, a PMD announcing no supported packet types
2223 * can still set a matching packet type in a received packet.
2226 * Better to invoke this API after the device is already started or rx burst
2227 * function is decided, to obtain correct supported ptypes.
2229 * if a given PMD does not report what ptypes it supports, then the supported
2230 * ptype count is reported as 0.
2232 * The port identifier of the Ethernet device.
2234 * A hint of what kind of packet type which the caller is interested in.
2236 * An array pointer to store adequate packet types, allocated by caller.
2238 * Size of the array pointed by param ptypes.
2240 * - (>=0) Number of supported ptypes. If the number of types exceeds num,
2241 * only num entries will be filled into the ptypes array, but the full
2242 * count of supported ptypes will be returned.
2243 * - (-ENODEV) if *port_id* invalid.
2245 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
2246 uint32_t *ptypes, int num);
2249 * Retrieve the MTU of an Ethernet device.
2252 * The port identifier of the Ethernet device.
2254 * A pointer to a uint16_t where the retrieved MTU is to be stored.
2256 * - (0) if successful.
2257 * - (-ENODEV) if *port_id* invalid.
2259 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu);
2262 * Change the MTU of an Ethernet device.
2265 * The port identifier of the Ethernet device.
2267 * A uint16_t for the MTU to be applied.
2269 * - (0) if successful.
2270 * - (-ENOTSUP) if operation is not supported.
2271 * - (-ENODEV) if *port_id* invalid.
2272 * - (-EIO) if device is removed.
2273 * - (-EINVAL) if *mtu* invalid.
2274 * - (-EBUSY) if operation is not allowed when the port is running
2276 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu);
2279 * Enable/Disable hardware filtering by an Ethernet device of received
2280 * VLAN packets tagged with a given VLAN Tag Identifier.
2283 * The port identifier of the Ethernet device.
2285 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
2287 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
2288 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
2290 * - (0) if successful.
2291 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2292 * - (-ENODEV) if *port_id* invalid.
2293 * - (-EIO) if device is removed.
2294 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
2295 * - (-EINVAL) if *vlan_id* > 4095.
2297 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on);
2300 * Enable/Disable hardware VLAN Strip by a rx queue of an Ethernet device.
2301 * 82599/X540/X550 can support VLAN stripping at the rx queue level
2304 * The port identifier of the Ethernet device.
2305 * @param rx_queue_id
2306 * The index of the receive queue for which a queue stats mapping is required.
2307 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2308 * to rte_eth_dev_configure().
2310 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port.
2311 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port.
2313 * - (0) if successful.
2314 * - (-ENOSUP) if hardware-assisted VLAN stripping not configured.
2315 * - (-ENODEV) if *port_id* invalid.
2316 * - (-EINVAL) if *rx_queue_id* invalid.
2318 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id,
2322 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to
2323 * the VLAN Header. This is a register setup available on some Intel NIC, not
2324 * but all, please check the data sheet for availability.
2327 * The port identifier of the Ethernet device.
2331 * The Tag Protocol ID
2333 * - (0) if successful.
2334 * - (-ENOSUP) if hardware-assisted VLAN TPID setup is not supported.
2335 * - (-ENODEV) if *port_id* invalid.
2336 * - (-EIO) if device is removed.
2338 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id,
2339 enum rte_vlan_type vlan_type,
2343 * Set VLAN offload configuration on an Ethernet device
2344 * Enable/Disable Extended VLAN by an Ethernet device, This is a register setup
2345 * available on some Intel NIC, not but all, please check the data sheet for
2347 * Enable/Disable VLAN Strip can be done on rx queue for certain NIC, but here
2348 * the configuration is applied on the port level.
2351 * The port identifier of the Ethernet device.
2352 * @param offload_mask
2353 * The VLAN Offload bit mask can be mixed use with "OR"
2354 * ETH_VLAN_STRIP_OFFLOAD
2355 * ETH_VLAN_FILTER_OFFLOAD
2356 * ETH_VLAN_EXTEND_OFFLOAD
2358 * - (0) if successful.
2359 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
2360 * - (-ENODEV) if *port_id* invalid.
2361 * - (-EIO) if device is removed.
2363 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask);
2366 * Read VLAN Offload configuration from an Ethernet device
2369 * The port identifier of the Ethernet device.
2371 * - (>0) if successful. Bit mask to indicate
2372 * ETH_VLAN_STRIP_OFFLOAD
2373 * ETH_VLAN_FILTER_OFFLOAD
2374 * ETH_VLAN_EXTEND_OFFLOAD
2375 * - (-ENODEV) if *port_id* invalid.
2377 int rte_eth_dev_get_vlan_offload(uint16_t port_id);
2380 * Set port based TX VLAN insertion on or off.
2383 * The port identifier of the Ethernet device.
2385 * Port based TX VLAN identifier together with user priority.
2387 * Turn on or off the port based TX VLAN insertion.
2390 * - (0) if successful.
2391 * - negative if failed.
2393 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on);
2395 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count,
2399 * Structure used to buffer packets for future TX
2400 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush
2402 struct rte_eth_dev_tx_buffer {
2403 buffer_tx_error_fn error_callback;
2404 void *error_userdata;
2405 uint16_t size; /**< Size of buffer for buffered tx */
2406 uint16_t length; /**< Number of packets in the array */
2407 struct rte_mbuf *pkts[];
2408 /**< Pending packets to be sent on explicit flush or when full */
2412 * Calculate the size of the tx buffer.
2415 * Number of stored packets.
2417 #define RTE_ETH_TX_BUFFER_SIZE(sz) \
2418 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *))
2421 * Initialize default values for buffered transmitting
2424 * Tx buffer to be initialized.
2431 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size);
2434 * Configure a callback for buffered packets which cannot be sent
2436 * Register a specific callback to be called when an attempt is made to send
2437 * all packets buffered on an ethernet port, but not all packets can
2438 * successfully be sent. The callback registered here will be called only
2439 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs.
2440 * The default callback configured for each queue by default just frees the
2441 * packets back to the calling mempool. If additional behaviour is required,
2442 * for example, to count dropped packets, or to retry transmission of packets
2443 * which cannot be sent, this function should be used to register a suitable
2444 * callback function to implement the desired behaviour.
2445 * The example callback "rte_eth_count_unsent_packet_callback()" is also
2446 * provided as reference.
2449 * The port identifier of the Ethernet device.
2451 * The function to be used as the callback.
2453 * Arbitrary parameter to be passed to the callback function
2455 * 0 on success, or -1 on error with rte_errno set appropriately
2458 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer,
2459 buffer_tx_error_fn callback, void *userdata);
2462 * Callback function for silently dropping unsent buffered packets.
2464 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2465 * adjust the default behavior when buffered packets cannot be sent. This
2466 * function drops any unsent packets silently and is used by tx buffered
2467 * operations as default behavior.
2469 * NOTE: this function should not be called directly, instead it should be used
2470 * as a callback for packet buffering.
2472 * NOTE: when configuring this function as a callback with
2473 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2474 * should point to an uint64_t value.
2477 * The previously buffered packets which could not be sent
2479 * The number of unsent packets in the pkts array
2484 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent,
2488 * Callback function for tracking unsent buffered packets.
2490 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to
2491 * adjust the default behavior when buffered packets cannot be sent. This
2492 * function drops any unsent packets, but also updates a user-supplied counter
2493 * to track the overall number of packets dropped. The counter should be an
2494 * uint64_t variable.
2496 * NOTE: this function should not be called directly, instead it should be used
2497 * as a callback for packet buffering.
2499 * NOTE: when configuring this function as a callback with
2500 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter
2501 * should point to an uint64_t value.
2504 * The previously buffered packets which could not be sent
2506 * The number of unsent packets in the pkts array
2508 * Pointer to an uint64_t value, which will be incremented by unsent
2511 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent,
2515 * Request the driver to free mbufs currently cached by the driver. The
2516 * driver will only free the mbuf if it is no longer in use. It is the
2517 * application's responsibity to ensure rte_eth_tx_buffer_flush(..) is
2521 * The port identifier of the Ethernet device.
2523 * The index of the transmit queue through which output packets must be
2525 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
2526 * to rte_eth_dev_configure().
2528 * Maximum number of packets to free. Use 0 to indicate all possible packets
2529 * should be freed. Note that a packet may be using multiple mbufs.
2532 * -ENODEV: Invalid interface
2533 * -EIO: device is removed
2534 * -ENOTSUP: Driver does not support function
2536 * 0-n: Number of packets freed. More packets may still remain in ring that
2540 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt);
2543 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by
2546 enum rte_eth_event_ipsec_subtype {
2547 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0,
2548 /**< Unknown event type */
2549 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW,
2550 /**< Sequence number overflow */
2551 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY,
2552 /**< Soft time expiry of SA */
2553 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY,
2554 /**< Soft byte expiry of SA */
2555 RTE_ETH_EVENT_IPSEC_MAX
2556 /**< Max value of this enum */
2560 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra
2561 * information of the IPsec offload event.
2563 struct rte_eth_event_ipsec_desc {
2564 enum rte_eth_event_ipsec_subtype subtype;
2565 /**< Type of RTE_ETH_EVENT_IPSEC_* event */
2567 /**< Event specific metadata
2569 * For the following events, *userdata* registered
2570 * with the *rte_security_session* would be returned
2573 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW
2574 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY
2575 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY
2577 * @see struct rte_security_session_conf
2583 * The eth device event type for interrupt, and maybe others in the future.
2585 enum rte_eth_event_type {
2586 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
2587 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
2588 RTE_ETH_EVENT_QUEUE_STATE,
2589 /**< queue state event (enabled/disabled) */
2590 RTE_ETH_EVENT_INTR_RESET,
2591 /**< reset interrupt event, sent to VF on PF reset */
2592 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */
2593 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */
2594 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */
2595 RTE_ETH_EVENT_NEW, /**< port is probed */
2596 RTE_ETH_EVENT_DESTROY, /**< port is released */
2597 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */
2598 RTE_ETH_EVENT_MAX /**< max value of this enum */
2601 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id,
2602 enum rte_eth_event_type event, void *cb_arg, void *ret_param);
2603 /**< user application callback to be registered for interrupts */
2606 * Register a callback function for port event.
2610 * RTE_ETH_ALL means register the event for all port ids.
2614 * User supplied callback function to be called.
2616 * Pointer to the parameters for the registered callback.
2619 * - On success, zero.
2620 * - On failure, a negative value.
2622 int rte_eth_dev_callback_register(uint16_t port_id,
2623 enum rte_eth_event_type event,
2624 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2627 * Unregister a callback function for port event.
2631 * RTE_ETH_ALL means unregister the event for all port ids.
2635 * User supplied callback function to be called.
2637 * Pointer to the parameters for the registered callback. -1 means to
2638 * remove all for the same callback address and same event.
2641 * - On success, zero.
2642 * - On failure, a negative value.
2644 int rte_eth_dev_callback_unregister(uint16_t port_id,
2645 enum rte_eth_event_type event,
2646 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
2649 * When there is no rx packet coming in Rx Queue for a long time, we can
2650 * sleep lcore related to RX Queue for power saving, and enable rx interrupt
2651 * to be triggered when Rx packet arrives.
2653 * The rte_eth_dev_rx_intr_enable() function enables rx queue
2654 * interrupt on specific rx queue of a port.
2657 * The port identifier of the Ethernet device.
2659 * The index of the receive queue from which to retrieve input packets.
2660 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2661 * to rte_eth_dev_configure().
2663 * - (0) if successful.
2664 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2666 * - (-ENODEV) if *port_id* invalid.
2667 * - (-EIO) if device is removed.
2669 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id);
2672 * When lcore wakes up from rx interrupt indicating packet coming, disable rx
2673 * interrupt and returns to polling mode.
2675 * The rte_eth_dev_rx_intr_disable() function disables rx queue
2676 * interrupt on specific rx queue of a port.
2679 * The port identifier of the Ethernet device.
2681 * The index of the receive queue from which to retrieve input packets.
2682 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2683 * to rte_eth_dev_configure().
2685 * - (0) if successful.
2686 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2688 * - (-ENODEV) if *port_id* invalid.
2689 * - (-EIO) if device is removed.
2691 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id);
2694 * RX Interrupt control per port.
2697 * The port identifier of the Ethernet device.
2699 * Epoll instance fd which the intr vector associated to.
2700 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2702 * The operation be performed for the vector.
2703 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2707 * - On success, zero.
2708 * - On failure, a negative value.
2710 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data);
2713 * RX Interrupt control per queue.
2716 * The port identifier of the Ethernet device.
2718 * The index of the receive queue from which to retrieve input packets.
2719 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2720 * to rte_eth_dev_configure().
2722 * Epoll instance fd which the intr vector associated to.
2723 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance.
2725 * The operation be performed for the vector.
2726 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}.
2730 * - On success, zero.
2731 * - On failure, a negative value.
2733 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id,
2734 int epfd, int op, void *data);
2738 * @b EXPERIMENTAL: this API may change without prior notice.
2740 * Get interrupt fd per Rx queue.
2743 * The port identifier of the Ethernet device.
2745 * The index of the receive queue from which to retrieve input packets.
2746 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
2747 * to rte_eth_dev_configure().
2749 * - (>=0) the interrupt fd associated to the requested Rx queue if
2753 int __rte_experimental
2754 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id);
2757 * Turn on the LED on the Ethernet device.
2758 * This function turns on the LED on the Ethernet device.
2761 * The port identifier of the Ethernet device.
2763 * - (0) if successful.
2764 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2766 * - (-ENODEV) if *port_id* invalid.
2767 * - (-EIO) if device is removed.
2769 int rte_eth_led_on(uint16_t port_id);
2772 * Turn off the LED on the Ethernet device.
2773 * This function turns off the LED on the Ethernet device.
2776 * The port identifier of the Ethernet device.
2778 * - (0) if successful.
2779 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
2781 * - (-ENODEV) if *port_id* invalid.
2782 * - (-EIO) if device is removed.
2784 int rte_eth_led_off(uint16_t port_id);
2787 * Get current status of the Ethernet link flow control for Ethernet device
2790 * The port identifier of the Ethernet device.
2792 * The pointer to the structure where to store the flow control parameters.
2794 * - (0) if successful.
2795 * - (-ENOTSUP) if hardware doesn't support flow control.
2796 * - (-ENODEV) if *port_id* invalid.
2797 * - (-EIO) if device is removed.
2799 int rte_eth_dev_flow_ctrl_get(uint16_t port_id,
2800 struct rte_eth_fc_conf *fc_conf);
2803 * Configure the Ethernet link flow control for Ethernet device
2806 * The port identifier of the Ethernet device.
2808 * The pointer to the structure of the flow control parameters.
2810 * - (0) if successful.
2811 * - (-ENOTSUP) if hardware doesn't support flow control mode.
2812 * - (-ENODEV) if *port_id* invalid.
2813 * - (-EINVAL) if bad parameter
2814 * - (-EIO) if flow control setup failure or device is removed.
2816 int rte_eth_dev_flow_ctrl_set(uint16_t port_id,
2817 struct rte_eth_fc_conf *fc_conf);
2820 * Configure the Ethernet priority flow control under DCB environment
2821 * for Ethernet device.
2824 * The port identifier of the Ethernet device.
2826 * The pointer to the structure of the priority flow control parameters.
2828 * - (0) if successful.
2829 * - (-ENOTSUP) if hardware doesn't support priority flow control mode.
2830 * - (-ENODEV) if *port_id* invalid.
2831 * - (-EINVAL) if bad parameter
2832 * - (-EIO) if flow control setup failure or device is removed.
2834 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id,
2835 struct rte_eth_pfc_conf *pfc_conf);
2838 * Add a MAC address to an internal array of addresses used to enable whitelist
2839 * filtering to accept packets only if the destination MAC address matches.
2842 * The port identifier of the Ethernet device.
2844 * The MAC address to add.
2846 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
2847 * not enabled, this should be set to 0.
2849 * - (0) if successfully added or *mac_addr* was already added.
2850 * - (-ENOTSUP) if hardware doesn't support this feature.
2851 * - (-ENODEV) if *port* is invalid.
2852 * - (-EIO) if device is removed.
2853 * - (-ENOSPC) if no more MAC addresses can be added.
2854 * - (-EINVAL) if MAC address is invalid.
2856 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct ether_addr *mac_addr,
2860 * Remove a MAC address from the internal array of addresses.
2863 * The port identifier of the Ethernet device.
2865 * MAC address to remove.
2867 * - (0) if successful, or *mac_addr* didn't exist.
2868 * - (-ENOTSUP) if hardware doesn't support.
2869 * - (-ENODEV) if *port* invalid.
2870 * - (-EADDRINUSE) if attempting to remove the default MAC address
2872 int rte_eth_dev_mac_addr_remove(uint16_t port_id, struct ether_addr *mac_addr);
2875 * Set the default MAC address.
2878 * The port identifier of the Ethernet device.
2880 * New default MAC address.
2882 * - (0) if successful, or *mac_addr* didn't exist.
2883 * - (-ENOTSUP) if hardware doesn't support.
2884 * - (-ENODEV) if *port* invalid.
2885 * - (-EINVAL) if MAC address is invalid.
2887 int rte_eth_dev_default_mac_addr_set(uint16_t port_id,
2888 struct ether_addr *mac_addr);
2891 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2894 * The port identifier of the Ethernet device.
2898 * Redirection table size. The table size can be queried by
2899 * rte_eth_dev_info_get().
2901 * - (0) if successful.
2902 * - (-ENOTSUP) if hardware doesn't support.
2903 * - (-EINVAL) if bad parameter.
2904 * - (-EIO) if device is removed.
2906 int rte_eth_dev_rss_reta_update(uint16_t port_id,
2907 struct rte_eth_rss_reta_entry64 *reta_conf,
2908 uint16_t reta_size);
2911 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device.
2914 * The port identifier of the Ethernet device.
2918 * Redirection table size. The table size can be queried by
2919 * rte_eth_dev_info_get().
2921 * - (0) if successful.
2922 * - (-ENOTSUP) if hardware doesn't support.
2923 * - (-EINVAL) if bad parameter.
2924 * - (-EIO) if device is removed.
2926 int rte_eth_dev_rss_reta_query(uint16_t port_id,
2927 struct rte_eth_rss_reta_entry64 *reta_conf,
2928 uint16_t reta_size);
2931 * Updates unicast hash table for receiving packet with the given destination
2932 * MAC address, and the packet is routed to all VFs for which the RX mode is
2933 * accept packets that match the unicast hash table.
2936 * The port identifier of the Ethernet device.
2938 * Unicast MAC address.
2940 * 1 - Set an unicast hash bit for receiving packets with the MAC address.
2941 * 0 - Clear an unicast hash bit.
2943 * - (0) if successful.
2944 * - (-ENOTSUP) if hardware doesn't support.
2945 * - (-ENODEV) if *port_id* invalid.
2946 * - (-EIO) if device is removed.
2947 * - (-EINVAL) if bad parameter.
2949 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct ether_addr *addr,
2953 * Updates all unicast hash bitmaps for receiving packet with any Unicast
2954 * Ethernet MAC addresses,the packet is routed to all VFs for which the RX
2955 * mode is accept packets that match the unicast hash table.
2958 * The port identifier of the Ethernet device.
2960 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet
2962 * 0 - Clear all unicast hash bitmaps
2964 * - (0) if successful.
2965 * - (-ENOTSUP) if hardware doesn't support.
2966 * - (-ENODEV) if *port_id* invalid.
2967 * - (-EIO) if device is removed.
2968 * - (-EINVAL) if bad parameter.
2970 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on);
2973 * Set a traffic mirroring rule on an Ethernet device
2976 * The port identifier of the Ethernet device.
2977 * @param mirror_conf
2978 * The pointer to the traffic mirroring structure describing the mirroring rule.
2979 * The *rte_eth_vm_mirror_conf* structure includes the type of mirroring rule,
2980 * destination pool and the value of rule if enable vlan or pool mirroring.
2983 * The index of traffic mirroring rule, we support four separated rules.
2985 * 1 - Enable a mirroring rule.
2986 * 0 - Disable a mirroring rule.
2988 * - (0) if successful.
2989 * - (-ENOTSUP) if hardware doesn't support this feature.
2990 * - (-ENODEV) if *port_id* invalid.
2991 * - (-EIO) if device is removed.
2992 * - (-EINVAL) if the mr_conf information is not correct.
2994 int rte_eth_mirror_rule_set(uint16_t port_id,
2995 struct rte_eth_mirror_conf *mirror_conf,
3000 * Reset a traffic mirroring rule on an Ethernet device.
3003 * The port identifier of the Ethernet device.
3005 * The index of traffic mirroring rule, we support four separated rules.
3007 * - (0) if successful.
3008 * - (-ENOTSUP) if hardware doesn't support this feature.
3009 * - (-ENODEV) if *port_id* invalid.
3010 * - (-EIO) if device is removed.
3011 * - (-EINVAL) if bad parameter.
3013 int rte_eth_mirror_rule_reset(uint16_t port_id,
3017 * Set the rate limitation for a queue on an Ethernet device.
3020 * The port identifier of the Ethernet device.
3024 * The tx rate in Mbps. Allocated from the total port link speed.
3026 * - (0) if successful.
3027 * - (-ENOTSUP) if hardware doesn't support this feature.
3028 * - (-ENODEV) if *port_id* invalid.
3029 * - (-EIO) if device is removed.
3030 * - (-EINVAL) if bad parameter.
3032 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx,
3036 * Configuration of Receive Side Scaling hash computation of Ethernet device.
3039 * The port identifier of the Ethernet device.
3041 * The new configuration to use for RSS hash computation on the port.
3043 * - (0) if successful.
3044 * - (-ENODEV) if port identifier is invalid.
3045 * - (-EIO) if device is removed.
3046 * - (-ENOTSUP) if hardware doesn't support.
3047 * - (-EINVAL) if bad parameter.
3049 int rte_eth_dev_rss_hash_update(uint16_t port_id,
3050 struct rte_eth_rss_conf *rss_conf);
3053 * Retrieve current configuration of Receive Side Scaling hash computation
3054 * of Ethernet device.
3057 * The port identifier of the Ethernet device.
3059 * Where to store the current RSS hash configuration of the Ethernet device.
3061 * - (0) if successful.
3062 * - (-ENODEV) if port identifier is invalid.
3063 * - (-EIO) if device is removed.
3064 * - (-ENOTSUP) if hardware doesn't support RSS.
3067 rte_eth_dev_rss_hash_conf_get(uint16_t port_id,
3068 struct rte_eth_rss_conf *rss_conf);
3071 * Add UDP tunneling port for a specific type of tunnel.
3072 * The packets with this UDP port will be identified as this type of tunnel.
3073 * Before enabling any offloading function for a tunnel, users can call this API
3074 * to change or add more UDP port for the tunnel. So the offloading function
3075 * can take effect on the packets with the specific UDP port.
3078 * The port identifier of the Ethernet device.
3080 * UDP tunneling configuration.
3083 * - (0) if successful.
3084 * - (-ENODEV) if port identifier is invalid.
3085 * - (-EIO) if device is removed.
3086 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3089 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id,
3090 struct rte_eth_udp_tunnel *tunnel_udp);
3093 * Delete UDP tunneling port a specific type of tunnel.
3094 * The packets with this UDP port will not be identified as this type of tunnel
3096 * Before enabling any offloading function for a tunnel, users can call this API
3097 * to delete a UDP port for the tunnel. So the offloading function will not take
3098 * effect on the packets with the specific UDP port.
3101 * The port identifier of the Ethernet device.
3103 * UDP tunneling configuration.
3106 * - (0) if successful.
3107 * - (-ENODEV) if port identifier is invalid.
3108 * - (-EIO) if device is removed.
3109 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3112 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id,
3113 struct rte_eth_udp_tunnel *tunnel_udp);
3116 * Check whether the filter type is supported on an Ethernet device.
3117 * All the supported filter types are defined in 'rte_eth_ctrl.h'.
3120 * The port identifier of the Ethernet device.
3121 * @param filter_type
3124 * - (0) if successful.
3125 * - (-ENOTSUP) if hardware doesn't support this filter type.
3126 * - (-ENODEV) if *port_id* invalid.
3127 * - (-EIO) if device is removed.
3129 int rte_eth_dev_filter_supported(uint16_t port_id,
3130 enum rte_filter_type filter_type);
3133 * Take operations to assigned filter type on an Ethernet device.
3134 * All the supported operations and filter types are defined in 'rte_eth_ctrl.h'.
3137 * The port identifier of the Ethernet device.
3138 * @param filter_type
3141 * Type of operation.
3143 * A pointer to arguments defined specifically for the operation.
3145 * - (0) if successful.
3146 * - (-ENOTSUP) if hardware doesn't support.
3147 * - (-ENODEV) if *port_id* invalid.
3148 * - (-EIO) if device is removed.
3149 * - others depends on the specific operations implementation.
3151 int rte_eth_dev_filter_ctrl(uint16_t port_id, enum rte_filter_type filter_type,
3152 enum rte_filter_op filter_op, void *arg);
3155 * Get DCB information on an Ethernet device.
3158 * The port identifier of the Ethernet device.
3162 * - (0) if successful.
3163 * - (-ENODEV) if port identifier is invalid.
3164 * - (-EIO) if device is removed.
3165 * - (-ENOTSUP) if hardware doesn't support.
3167 int rte_eth_dev_get_dcb_info(uint16_t port_id,
3168 struct rte_eth_dcb_info *dcb_info);
3170 struct rte_eth_rxtx_callback;
3173 * Add a callback to be called on packet RX on a given port and queue.
3175 * This API configures a function to be called for each burst of
3176 * packets received on a given NIC port queue. The return value is a pointer
3177 * that can be used to later remove the callback using
3178 * rte_eth_remove_rx_callback().
3180 * Multiple functions are called in the order that they are added.
3183 * The port identifier of the Ethernet device.
3185 * The queue on the Ethernet device on which the callback is to be added.
3187 * The callback function
3189 * A generic pointer parameter which will be passed to each invocation of the
3190 * callback function on this port and queue.
3194 * On success, a pointer value which can later be used to remove the callback.
3196 const struct rte_eth_rxtx_callback *
3197 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id,
3198 rte_rx_callback_fn fn, void *user_param);
3201 * Add a callback that must be called first on packet RX on a given port
3204 * This API configures a first function to be called for each burst of
3205 * packets received on a given NIC port queue. The return value is a pointer
3206 * that can be used to later remove the callback using
3207 * rte_eth_remove_rx_callback().
3209 * Multiple functions are called in the order that they are added.
3212 * The port identifier of the Ethernet device.
3214 * The queue on the Ethernet device on which the callback is to be added.
3216 * The callback function
3218 * A generic pointer parameter which will be passed to each invocation of the
3219 * callback function on this port and queue.
3223 * On success, a pointer value which can later be used to remove the callback.
3225 const struct rte_eth_rxtx_callback *
3226 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id,
3227 rte_rx_callback_fn fn, void *user_param);
3230 * Add a callback to be called on packet TX on a given port and queue.
3232 * This API configures a function to be called for each burst of
3233 * packets sent on a given NIC port queue. The return value is a pointer
3234 * that can be used to later remove the callback using
3235 * rte_eth_remove_tx_callback().
3237 * Multiple functions are called in the order that they are added.
3240 * The port identifier of the Ethernet device.
3242 * The queue on the Ethernet device on which the callback is to be added.
3244 * The callback function
3246 * A generic pointer parameter which will be passed to each invocation of the
3247 * callback function on this port and queue.
3251 * On success, a pointer value which can later be used to remove the callback.
3253 const struct rte_eth_rxtx_callback *
3254 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id,
3255 rte_tx_callback_fn fn, void *user_param);
3258 * Remove an RX packet callback from a given port and queue.
3260 * This function is used to removed callbacks that were added to a NIC port
3261 * queue using rte_eth_add_rx_callback().
3263 * Note: the callback is removed from the callback list but it isn't freed
3264 * since the it may still be in use. The memory for the callback can be
3265 * subsequently freed back by the application by calling rte_free():
3267 * - Immediately - if the port is stopped, or the user knows that no
3268 * callbacks are in flight e.g. if called from the thread doing RX/TX
3271 * - After a short delay - where the delay is sufficient to allow any
3272 * in-flight callbacks to complete.
3275 * The port identifier of the Ethernet device.
3277 * The queue on the Ethernet device from which the callback is to be removed.
3279 * User supplied callback created via rte_eth_add_rx_callback().
3282 * - 0: Success. Callback was removed.
3283 * - -ENOTSUP: Callback support is not available.
3284 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3285 * is NULL or not found for the port/queue.
3287 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id,
3288 const struct rte_eth_rxtx_callback *user_cb);
3291 * Remove a TX packet callback from a given port and queue.
3293 * This function is used to removed callbacks that were added to a NIC port
3294 * queue using rte_eth_add_tx_callback().
3296 * Note: the callback is removed from the callback list but it isn't freed
3297 * since the it may still be in use. The memory for the callback can be
3298 * subsequently freed back by the application by calling rte_free():
3300 * - Immediately - if the port is stopped, or the user knows that no
3301 * callbacks are in flight e.g. if called from the thread doing RX/TX
3304 * - After a short delay - where the delay is sufficient to allow any
3305 * in-flight callbacks to complete.
3308 * The port identifier of the Ethernet device.
3310 * The queue on the Ethernet device from which the callback is to be removed.
3312 * User supplied callback created via rte_eth_add_tx_callback().
3315 * - 0: Success. Callback was removed.
3316 * - -ENOTSUP: Callback support is not available.
3317 * - -EINVAL: The port_id or the queue_id is out of range, or the callback
3318 * is NULL or not found for the port/queue.
3320 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id,
3321 const struct rte_eth_rxtx_callback *user_cb);
3324 * Retrieve information about given port's RX queue.
3327 * The port identifier of the Ethernet device.
3329 * The RX queue on the Ethernet device for which information
3330 * will be retrieved.
3332 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with
3333 * the information of the Ethernet device.
3337 * - -ENOTSUP: routine is not supported by the device PMD.
3338 * - -EINVAL: The port_id or the queue_id is out of range.
3340 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3341 struct rte_eth_rxq_info *qinfo);
3344 * Retrieve information about given port's TX queue.
3347 * The port identifier of the Ethernet device.
3349 * The TX queue on the Ethernet device for which information
3350 * will be retrieved.
3352 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with
3353 * the information of the Ethernet device.
3357 * - -ENOTSUP: routine is not supported by the device PMD.
3358 * - -EINVAL: The port_id or the queue_id is out of range.
3360 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id,
3361 struct rte_eth_txq_info *qinfo);
3364 * Retrieve device registers and register attributes (number of registers and
3368 * The port identifier of the Ethernet device.
3370 * Pointer to rte_dev_reg_info structure to fill in. If info->data is
3371 * NULL the function fills in the width and length fields. If non-NULL
3372 * the registers are put into the buffer pointed at by the data field.
3374 * - (0) if successful.
3375 * - (-ENOTSUP) if hardware doesn't support.
3376 * - (-ENODEV) if *port_id* invalid.
3377 * - (-EIO) if device is removed.
3378 * - others depends on the specific operations implementation.
3380 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info);
3383 * Retrieve size of device EEPROM
3386 * The port identifier of the Ethernet device.
3388 * - (>=0) EEPROM size if successful.
3389 * - (-ENOTSUP) if hardware doesn't support.
3390 * - (-ENODEV) if *port_id* invalid.
3391 * - (-EIO) if device is removed.
3392 * - others depends on the specific operations implementation.
3394 int rte_eth_dev_get_eeprom_length(uint16_t port_id);
3397 * Retrieve EEPROM and EEPROM attribute
3400 * The port identifier of the Ethernet device.
3402 * The template includes buffer for return EEPROM data and
3403 * EEPROM attributes to be filled.
3405 * - (0) if successful.
3406 * - (-ENOTSUP) if hardware doesn't support.
3407 * - (-ENODEV) if *port_id* invalid.
3408 * - (-EIO) if device is removed.
3409 * - others depends on the specific operations implementation.
3411 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3414 * Program EEPROM with provided data
3417 * The port identifier of the Ethernet device.
3419 * The template includes EEPROM data for programming and
3420 * EEPROM attributes to be filled
3422 * - (0) if successful.
3423 * - (-ENOTSUP) if hardware doesn't support.
3424 * - (-ENODEV) if *port_id* invalid.
3425 * - (-EIO) if device is removed.
3426 * - others depends on the specific operations implementation.
3428 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info);
3432 * @b EXPERIMENTAL: this API may change without prior notice.
3434 * Retrieve the type and size of plugin module EEPROM
3437 * The port identifier of the Ethernet device.
3439 * The type and size of plugin module EEPROM.
3441 * - (0) if successful.
3442 * - (-ENOTSUP) if hardware doesn't support.
3443 * - (-ENODEV) if *port_id* invalid.
3444 * - (-EIO) if device is removed.
3445 * - others depends on the specific operations implementation.
3447 int __rte_experimental
3448 rte_eth_dev_get_module_info(uint16_t port_id,
3449 struct rte_eth_dev_module_info *modinfo);
3453 * @b EXPERIMENTAL: this API may change without prior notice.
3455 * Retrieve the data of plugin module EEPROM
3458 * The port identifier of the Ethernet device.
3460 * The template includes the plugin module EEPROM attributes, and the
3461 * buffer for return plugin module EEPROM data.
3463 * - (0) if successful.
3464 * - (-ENOTSUP) if hardware doesn't support.
3465 * - (-ENODEV) if *port_id* invalid.
3466 * - (-EIO) if device is removed.
3467 * - others depends on the specific operations implementation.
3469 int __rte_experimental
3470 rte_eth_dev_get_module_eeprom(uint16_t port_id,
3471 struct rte_dev_eeprom_info *info);
3474 * Set the list of multicast addresses to filter on an Ethernet device.
3477 * The port identifier of the Ethernet device.
3478 * @param mc_addr_set
3479 * The array of multicast addresses to set. Equal to NULL when the function
3480 * is invoked to flush the set of filtered addresses.
3482 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0
3483 * when the function is invoked to flush the set of filtered addresses.
3485 * - (0) if successful.
3486 * - (-ENODEV) if *port_id* invalid.
3487 * - (-EIO) if device is removed.
3488 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering.
3489 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources.
3491 int rte_eth_dev_set_mc_addr_list(uint16_t port_id,
3492 struct ether_addr *mc_addr_set,
3493 uint32_t nb_mc_addr);
3496 * Enable IEEE1588/802.1AS timestamping for an Ethernet device.
3499 * The port identifier of the Ethernet device.
3503 * - -ENODEV: The port ID is invalid.
3504 * - -EIO: if device is removed.
3505 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3507 int rte_eth_timesync_enable(uint16_t port_id);
3510 * Disable IEEE1588/802.1AS timestamping for an Ethernet device.
3513 * The port identifier of the Ethernet device.
3517 * - -ENODEV: The port ID is invalid.
3518 * - -EIO: if device is removed.
3519 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3521 int rte_eth_timesync_disable(uint16_t port_id);
3524 * Read an IEEE1588/802.1AS RX timestamp from an Ethernet device.
3527 * The port identifier of the Ethernet device.
3529 * Pointer to the timestamp struct.
3531 * Device specific flags. Used to pass the RX timesync register index to
3532 * i40e. Unused in igb/ixgbe, pass 0 instead.
3536 * - -EINVAL: No timestamp is available.
3537 * - -ENODEV: The port ID is invalid.
3538 * - -EIO: if device is removed.
3539 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3541 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id,
3542 struct timespec *timestamp, uint32_t flags);
3545 * Read an IEEE1588/802.1AS TX timestamp from an Ethernet device.
3548 * The port identifier of the Ethernet device.
3550 * Pointer to the timestamp struct.
3554 * - -EINVAL: No timestamp is available.
3555 * - -ENODEV: The port ID is invalid.
3556 * - -EIO: if device is removed.
3557 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3559 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id,
3560 struct timespec *timestamp);
3563 * Adjust the timesync clock on an Ethernet device.
3565 * This is usually used in conjunction with other Ethdev timesync functions to
3566 * synchronize the device time using the IEEE1588/802.1AS protocol.
3569 * The port identifier of the Ethernet device.
3571 * The adjustment in nanoseconds.
3575 * - -ENODEV: The port ID is invalid.
3576 * - -EIO: if device is removed.
3577 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3579 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta);
3582 * Read the time from the timesync clock on an Ethernet device.
3584 * This is usually used in conjunction with other Ethdev timesync functions to
3585 * synchronize the device time using the IEEE1588/802.1AS protocol.
3588 * The port identifier of the Ethernet device.
3590 * Pointer to the timespec struct that holds the time.
3595 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time);
3598 * Set the time of the timesync clock on an Ethernet device.
3600 * This is usually used in conjunction with other Ethdev timesync functions to
3601 * synchronize the device time using the IEEE1588/802.1AS protocol.
3604 * The port identifier of the Ethernet device.
3606 * Pointer to the timespec struct that holds the time.
3610 * - -EINVAL: No timestamp is available.
3611 * - -ENODEV: The port ID is invalid.
3612 * - -EIO: if device is removed.
3613 * - -ENOTSUP: The function is not supported by the Ethernet driver.
3615 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time);
3618 * Config l2 tunnel ether type of an Ethernet device for filtering specific
3619 * tunnel packets by ether type.
3622 * The port identifier of the Ethernet device.
3624 * l2 tunnel configuration.
3627 * - (0) if successful.
3628 * - (-ENODEV) if port identifier is invalid.
3629 * - (-EIO) if device is removed.
3630 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3633 rte_eth_dev_l2_tunnel_eth_type_conf(uint16_t port_id,
3634 struct rte_eth_l2_tunnel_conf *l2_tunnel);
3637 * Enable/disable l2 tunnel offload functions. Include,
3638 * 1, The ability of parsing a type of l2 tunnel of an Ethernet device.
3639 * Filtering, forwarding and offloading this type of tunnel packets depend on
3641 * 2, Stripping the l2 tunnel tag.
3642 * 3, Insertion of the l2 tunnel tag.
3643 * 4, Forwarding the packets based on the l2 tunnel tag.
3646 * The port identifier of the Ethernet device.
3648 * l2 tunnel parameters.
3650 * Indicate the offload function.
3652 * Enable or disable this function.
3655 * - (0) if successful.
3656 * - (-ENODEV) if port identifier is invalid.
3657 * - (-EIO) if device is removed.
3658 * - (-ENOTSUP) if hardware doesn't support tunnel type.
3661 rte_eth_dev_l2_tunnel_offload_set(uint16_t port_id,
3662 struct rte_eth_l2_tunnel_conf *l2_tunnel,
3667 * Get the port id from device name. The device name should be specified
3669 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0
3670 * - SoC device name, for example- fsl-gmac0
3671 * - vdev dpdk name, for example- net_[pcap0|null0|tap0]
3674 * pci address or name of the device
3676 * pointer to port identifier of the device
3678 * - (0) if successful and port_id is filled.
3679 * - (-ENODEV or -EINVAL) on failure.
3682 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id);
3685 * Get the device name from port id. The device name is specified as below:
3686 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0
3687 * - SoC device name, for example- fsl-gmac0
3688 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0]
3691 * Port identifier of the device.
3693 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name.
3695 * - (0) if successful.
3696 * - (-EINVAL) on failure.
3699 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name);
3702 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from
3703 * the ethernet device information, otherwise adjust them to boundaries.
3706 * The port identifier of the Ethernet device.
3708 * A pointer to a uint16_t where the number of receive
3709 * descriptors stored.
3711 * A pointer to a uint16_t where the number of transmit
3712 * descriptors stored.
3714 * - (0) if successful.
3715 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure.
3717 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
3718 uint16_t *nb_rx_desc,
3719 uint16_t *nb_tx_desc);
3722 * Test if a port supports specific mempool ops.
3725 * Port identifier of the Ethernet device.
3727 * The name of the pool operations to test.
3729 * - 0: best mempool ops choice for this port.
3730 * - 1: mempool ops are supported for this port.
3731 * - -ENOTSUP: mempool ops not supported for this port.
3732 * - -ENODEV: Invalid port Identifier.
3733 * - -EINVAL: Pool param is null.
3736 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
3739 * Get the security context for the Ethernet device.
3742 * Port identifier of the Ethernet device
3745 * - pointer to security context on success.
3748 rte_eth_dev_get_sec_ctx(uint16_t port_id);
3751 #include <rte_ethdev_core.h>
3755 * Retrieve a burst of input packets from a receive queue of an Ethernet
3756 * device. The retrieved packets are stored in *rte_mbuf* structures whose
3757 * pointers are supplied in the *rx_pkts* array.
3759 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
3760 * receive queue, up to *nb_pkts* packets, and for each completed RX
3761 * descriptor in the ring, it performs the following operations:
3763 * - Initialize the *rte_mbuf* data structure associated with the
3764 * RX descriptor according to the information provided by the NIC into
3765 * that RX descriptor.
3767 * - Store the *rte_mbuf* data structure into the next entry of the
3770 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
3771 * allocated from the memory pool associated with the receive queue at
3772 * initialization time.
3774 * When retrieving an input packet that was scattered by the controller
3775 * into multiple receive descriptors, the rte_eth_rx_burst() function
3776 * appends the associated *rte_mbuf* buffers to the first buffer of the
3779 * The rte_eth_rx_burst() function returns the number of packets
3780 * actually retrieved, which is the number of *rte_mbuf* data structures
3781 * effectively supplied into the *rx_pkts* array.
3782 * A return value equal to *nb_pkts* indicates that the RX queue contained
3783 * at least *rx_pkts* packets, and this is likely to signify that other
3784 * received packets remain in the input queue. Applications implementing
3785 * a "retrieve as much received packets as possible" policy can check this
3786 * specific case and keep invoking the rte_eth_rx_burst() function until
3787 * a value less than *nb_pkts* is returned.
3789 * This receive method has the following advantages:
3791 * - It allows a run-to-completion network stack engine to retrieve and
3792 * to immediately process received packets in a fast burst-oriented
3793 * approach, avoiding the overhead of unnecessary intermediate packet
3794 * queue/dequeue operations.
3796 * - Conversely, it also allows an asynchronous-oriented processing
3797 * method to retrieve bursts of received packets and to immediately
3798 * queue them for further parallel processing by another logical core,
3799 * for instance. However, instead of having received packets being
3800 * individually queued by the driver, this approach allows the caller
3801 * of the rte_eth_rx_burst() function to queue a burst of retrieved
3802 * packets at a time and therefore dramatically reduce the cost of
3803 * enqueue/dequeue operations per packet.
3805 * - It allows the rte_eth_rx_burst() function of the driver to take
3806 * advantage of burst-oriented hardware features (CPU cache,
3807 * prefetch instructions, and so on) to minimize the number of CPU
3808 * cycles per packet.
3810 * To summarize, the proposed receive API enables many
3811 * burst-oriented optimizations in both synchronous and asynchronous
3812 * packet processing environments with no overhead in both cases.
3814 * The rte_eth_rx_burst() function does not provide any error
3815 * notification to avoid the corresponding overhead. As a hint, the
3816 * upper-level application might check the status of the device link once
3817 * being systematically returned a 0 value for a given number of tries.
3820 * The port identifier of the Ethernet device.
3822 * The index of the receive queue from which to retrieve input packets.
3823 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
3824 * to rte_eth_dev_configure().
3826 * The address of an array of pointers to *rte_mbuf* structures that
3827 * must be large enough to store *nb_pkts* pointers in it.
3829 * The maximum number of packets to retrieve.
3831 * The number of packets actually retrieved, which is the number
3832 * of pointers to *rte_mbuf* structures effectively supplied to the
3835 static inline uint16_t
3836 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id,
3837 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts)
3839 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3842 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3843 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
3844 RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
3846 if (queue_id >= dev->data->nb_rx_queues) {
3847 RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n", queue_id);
3851 nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
3854 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
3855 if (unlikely(dev->post_rx_burst_cbs[queue_id] != NULL)) {
3856 struct rte_eth_rxtx_callback *cb =
3857 dev->post_rx_burst_cbs[queue_id];
3860 nb_rx = cb->fn.rx(port_id, queue_id, rx_pkts, nb_rx,
3861 nb_pkts, cb->param);
3863 } while (cb != NULL);
3871 * Get the number of used descriptors of a rx queue
3874 * The port identifier of the Ethernet device.
3876 * The queue id on the specific port.
3878 * The number of used descriptors in the specific queue, or:
3879 * (-EINVAL) if *port_id* or *queue_id* is invalid
3880 * (-ENOTSUP) if the device does not support this function
3883 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id)
3885 struct rte_eth_dev *dev;
3887 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -EINVAL);
3888 dev = &rte_eth_devices[port_id];
3889 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_count, -ENOTSUP);
3890 if (queue_id >= dev->data->nb_rx_queues)
3893 return (int)(*dev->dev_ops->rx_queue_count)(dev, queue_id);
3897 * Check if the DD bit of the specific RX descriptor in the queue has been set
3900 * The port identifier of the Ethernet device.
3902 * The queue id on the specific port.
3904 * The offset of the descriptor ID from tail.
3906 * - (1) if the specific DD bit is set.
3907 * - (0) if the specific DD bit is not set.
3908 * - (-ENODEV) if *port_id* invalid.
3909 * - (-ENOTSUP) if the device does not support this function
3912 rte_eth_rx_descriptor_done(uint16_t port_id, uint16_t queue_id, uint16_t offset)
3914 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
3915 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3916 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_done, -ENOTSUP);
3917 return (*dev->dev_ops->rx_descriptor_done)( \
3918 dev->data->rx_queues[queue_id], offset);
3921 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */
3922 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */
3923 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
3926 * Check the status of a Rx descriptor in the queue
3928 * It should be called in a similar context than the Rx function:
3929 * - on a dataplane core
3930 * - not concurrently on the same queue
3932 * Since it's a dataplane function, no check is performed on port_id and
3933 * queue_id. The caller must therefore ensure that the port is enabled
3934 * and the queue is configured and running.
3936 * Note: accessing to a random descriptor in the ring may trigger cache
3937 * misses and have a performance impact.
3940 * A valid port identifier of the Ethernet device which.
3942 * A valid Rx queue identifier on this port.
3944 * The offset of the descriptor starting from tail (0 is the next
3945 * packet to be received by the driver).
3948 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to
3950 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but
3951 * not yet processed by the driver (i.e. in the receive queue).
3952 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by
3953 * the driver and not yet returned to hw, or reserved by the hw.
3954 * - (-EINVAL) bad descriptor offset.
3955 * - (-ENOTSUP) if the device does not support this function.
3956 * - (-ENODEV) bad port or queue (only if compiled with debug).
3959 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id,
3962 struct rte_eth_dev *dev;
3965 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3966 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
3968 dev = &rte_eth_devices[port_id];
3969 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
3970 if (queue_id >= dev->data->nb_rx_queues)
3973 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_status, -ENOTSUP);
3974 rxq = dev->data->rx_queues[queue_id];
3976 return (*dev->dev_ops->rx_descriptor_status)(rxq, offset);
3979 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */
3980 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */
3981 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */
3984 * Check the status of a Tx descriptor in the queue.
3986 * It should be called in a similar context than the Tx function:
3987 * - on a dataplane core
3988 * - not concurrently on the same queue
3990 * Since it's a dataplane function, no check is performed on port_id and
3991 * queue_id. The caller must therefore ensure that the port is enabled
3992 * and the queue is configured and running.
3994 * Note: accessing to a random descriptor in the ring may trigger cache
3995 * misses and have a performance impact.
3998 * A valid port identifier of the Ethernet device which.
4000 * A valid Tx queue identifier on this port.
4002 * The offset of the descriptor starting from tail (0 is the place where
4003 * the next packet will be send).
4006 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e.
4007 * in the transmit queue.
4008 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can
4009 * be reused by the driver.
4010 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the
4011 * driver or the hardware.
4012 * - (-EINVAL) bad descriptor offset.
4013 * - (-ENOTSUP) if the device does not support this function.
4014 * - (-ENODEV) bad port or queue (only if compiled with debug).
4016 static inline int rte_eth_tx_descriptor_status(uint16_t port_id,
4017 uint16_t queue_id, uint16_t offset)
4019 struct rte_eth_dev *dev;
4022 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4023 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
4025 dev = &rte_eth_devices[port_id];
4026 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4027 if (queue_id >= dev->data->nb_tx_queues)
4030 RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_descriptor_status, -ENOTSUP);
4031 txq = dev->data->tx_queues[queue_id];
4033 return (*dev->dev_ops->tx_descriptor_status)(txq, offset);
4037 * Send a burst of output packets on a transmit queue of an Ethernet device.
4039 * The rte_eth_tx_burst() function is invoked to transmit output packets
4040 * on the output queue *queue_id* of the Ethernet device designated by its
4042 * The *nb_pkts* parameter is the number of packets to send which are
4043 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
4044 * allocated from a pool created with rte_pktmbuf_pool_create().
4045 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
4046 * up to the number of transmit descriptors available in the TX ring of the
4048 * For each packet to send, the rte_eth_tx_burst() function performs
4049 * the following operations:
4051 * - Pick up the next available descriptor in the transmit ring.
4053 * - Free the network buffer previously sent with that descriptor, if any.
4055 * - Initialize the transmit descriptor with the information provided
4056 * in the *rte_mbuf data structure.
4058 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
4059 * the rte_eth_tx_burst() function uses several transmit descriptors
4062 * The rte_eth_tx_burst() function returns the number of packets it
4063 * actually sent. A return value equal to *nb_pkts* means that all packets
4064 * have been sent, and this is likely to signify that other output packets
4065 * could be immediately transmitted again. Applications that implement a
4066 * "send as many packets to transmit as possible" policy can check this
4067 * specific case and keep invoking the rte_eth_tx_burst() function until
4068 * a value less than *nb_pkts* is returned.
4070 * It is the responsibility of the rte_eth_tx_burst() function to
4071 * transparently free the memory buffers of packets previously sent.
4072 * This feature is driven by the *tx_free_thresh* value supplied to the
4073 * rte_eth_dev_configure() function at device configuration time.
4074 * When the number of free TX descriptors drops below this threshold, the
4075 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers
4076 * of those packets whose transmission was effectively completed.
4078 * If the PMD is DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can
4079 * invoke this function concurrently on the same tx queue without SW lock.
4080 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads
4082 * @see rte_eth_tx_prepare to perform some prior checks or adjustments
4086 * The port identifier of the Ethernet device.
4088 * The index of the transmit queue through which output packets must be
4090 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4091 * to rte_eth_dev_configure().
4093 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
4094 * which contain the output packets.
4096 * The maximum number of packets to transmit.
4098 * The number of output packets actually stored in transmit descriptors of
4099 * the transmit ring. The return value can be less than the value of the
4100 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
4102 static inline uint16_t
4103 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id,
4104 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4106 struct rte_eth_dev *dev = &rte_eth_devices[port_id];
4108 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4109 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
4110 RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
4112 if (queue_id >= dev->data->nb_tx_queues) {
4113 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
4118 #ifdef RTE_ETHDEV_RXTX_CALLBACKS
4119 struct rte_eth_rxtx_callback *cb = dev->pre_tx_burst_cbs[queue_id];
4121 if (unlikely(cb != NULL)) {
4123 nb_pkts = cb->fn.tx(port_id, queue_id, tx_pkts, nb_pkts,
4126 } while (cb != NULL);
4130 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
4135 * @b EXPERIMENTAL: this API may change without prior notice
4137 * Process a burst of output packets on a transmit queue of an Ethernet device.
4139 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be
4140 * transmitted on the output queue *queue_id* of the Ethernet device designated
4142 * The *nb_pkts* parameter is the number of packets to be prepared which are
4143 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them
4144 * allocated from a pool created with rte_pktmbuf_pool_create().
4145 * For each packet to send, the rte_eth_tx_prepare() function performs
4146 * the following operations:
4148 * - Check if packet meets devices requirements for tx offloads.
4150 * - Check limitations about number of segments.
4152 * - Check additional requirements when debug is enabled.
4154 * - Update and/or reset required checksums when tx offload is set for packet.
4156 * Since this function can modify packet data, provided mbufs must be safely
4157 * writable (e.g. modified data cannot be in shared segment).
4159 * The rte_eth_tx_prepare() function returns the number of packets ready to be
4160 * sent. A return value equal to *nb_pkts* means that all packets are valid and
4161 * ready to be sent, otherwise stops processing on the first invalid packet and
4162 * leaves the rest packets untouched.
4164 * When this functionality is not implemented in the driver, all packets are
4165 * are returned untouched.
4168 * The port identifier of the Ethernet device.
4169 * The value must be a valid port id.
4171 * The index of the transmit queue through which output packets must be
4173 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4174 * to rte_eth_dev_configure().
4176 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
4177 * which contain the output packets.
4179 * The maximum number of packets to process.
4181 * The number of packets correct and ready to be sent. The return value can be
4182 * less than the value of the *tx_pkts* parameter when some packet doesn't
4183 * meet devices requirements with rte_errno set appropriately:
4184 * - -EINVAL: offload flags are not correctly set
4185 * - -ENOTSUP: the offload feature is not supported by the hardware
4189 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP
4191 static inline uint16_t
4192 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id,
4193 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4195 struct rte_eth_dev *dev;
4197 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4198 if (!rte_eth_dev_is_valid_port(port_id)) {
4199 RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
4200 rte_errno = -EINVAL;
4205 dev = &rte_eth_devices[port_id];
4207 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
4208 if (queue_id >= dev->data->nb_tx_queues) {
4209 RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n", queue_id);
4210 rte_errno = -EINVAL;
4215 if (!dev->tx_pkt_prepare)
4218 return (*dev->tx_pkt_prepare)(dev->data->tx_queues[queue_id],
4225 * Native NOOP operation for compilation targets which doesn't require any
4226 * preparations steps, and functional NOOP may introduce unnecessary performance
4229 * Generally this is not a good idea to turn it on globally and didn't should
4230 * be used if behavior of tx_preparation can change.
4233 static inline uint16_t
4234 rte_eth_tx_prepare(__rte_unused uint16_t port_id,
4235 __rte_unused uint16_t queue_id,
4236 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
4244 * Send any packets queued up for transmission on a port and HW queue
4246 * This causes an explicit flush of packets previously buffered via the
4247 * rte_eth_tx_buffer() function. It returns the number of packets successfully
4248 * sent to the NIC, and calls the error callback for any unsent packets. Unless
4249 * explicitly set up otherwise, the default callback simply frees the unsent
4250 * packets back to the owning mempool.
4253 * The port identifier of the Ethernet device.
4255 * The index of the transmit queue through which output packets must be
4257 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4258 * to rte_eth_dev_configure().
4260 * Buffer of packets to be transmit.
4262 * The number of packets successfully sent to the Ethernet device. The error
4263 * callback is called for any packets which could not be sent.
4265 static inline uint16_t
4266 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id,
4267 struct rte_eth_dev_tx_buffer *buffer)
4270 uint16_t to_send = buffer->length;
4275 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send);
4279 /* All packets sent, or to be dealt with by callback below */
4280 if (unlikely(sent != to_send))
4281 buffer->error_callback(&buffer->pkts[sent],
4282 (uint16_t)(to_send - sent),
4283 buffer->error_userdata);
4289 * Buffer a single packet for future transmission on a port and queue
4291 * This function takes a single mbuf/packet and buffers it for later
4292 * transmission on the particular port and queue specified. Once the buffer is
4293 * full of packets, an attempt will be made to transmit all the buffered
4294 * packets. In case of error, where not all packets can be transmitted, a
4295 * callback is called with the unsent packets as a parameter. If no callback
4296 * is explicitly set up, the unsent packets are just freed back to the owning
4297 * mempool. The function returns the number of packets actually sent i.e.
4298 * 0 if no buffer flush occurred, otherwise the number of packets successfully
4302 * The port identifier of the Ethernet device.
4304 * The index of the transmit queue through which output packets must be
4306 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
4307 * to rte_eth_dev_configure().
4309 * Buffer used to collect packets to be sent.
4311 * Pointer to the packet mbuf to be sent.
4313 * 0 = packet has been buffered for later transmission
4314 * N > 0 = packet has been buffered, and the buffer was subsequently flushed,
4315 * causing N packets to be sent, and the error callback to be called for
4318 static __rte_always_inline uint16_t
4319 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id,
4320 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt)
4322 buffer->pkts[buffer->length++] = tx_pkt;
4323 if (buffer->length < buffer->size)
4326 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer);
4333 #endif /* _RTE_ETHDEV_H_ */