1 /* SPDX-License-Identifier: BSD-3-Clause 2 * Copyright(c) 2010-2017 Intel Corporation 3 */ 4 5 #ifndef _RTE_ETHDEV_H_ 6 #define _RTE_ETHDEV_H_ 7 8 /** 9 * @file 10 * 11 * RTE Ethernet Device API 12 * 13 * The Ethernet Device API is composed of two parts: 14 * 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. 19 * 20 * - The driver-oriented Ethernet API that exports functions allowing 21 * an Ethernet Poll Mode Driver (PMD) to allocate an Ethernet device instance, 22 * create memzone for HW rings and process registered callbacks, and so on. 23 * PMDs should include ethdev_driver.h instead of this header. 24 * 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. 32 * 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. 36 * 37 * In all functions of the Ethernet API, the Ethernet device is 38 * designated by an integer >= 0 named the device port identifier. 39 * 40 * At the Ethernet driver level, Ethernet devices are represented by a generic 41 * data structure of type *rte_eth_dev*. 42 * 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. 49 * 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 52 * from NVM etc. 53 * 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 58 * freed. 59 * 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() 66 * 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. 71 * 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 77 * device or the queue is stopped. 78 * 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 81 * be retained: 82 * 83 * - MTU 84 * - flow control settings 85 * - receive mode configuration (promiscuous mode, all-multicast mode, 86 * hardware checksum mode, RSS/VMDq settings etc.) 87 * - VLAN filtering configuration 88 * - default MAC address 89 * - MAC addresses supplied to MAC address array 90 * - flow director filtering mode (but not filtering rules) 91 * - NIC queue statistics mappings 92 * 93 * The following configuration may be retained or not 94 * depending on the device capabilities: 95 * 96 * - flow rules 97 * - flow-related shared objects, e.g. indirect actions 98 * 99 * Any other configuration will not be stored and will need to be re-entered 100 * before a call to rte_eth_dev_start(). 101 * 102 * Finally, a network application can close an Ethernet device by invoking the 103 * rte_eth_dev_close() function. 104 * 105 * Each function of the application Ethernet API invokes a specific function 106 * of the PMD that controls the target device designated by its port 107 * identifier. 108 * For this purpose, all device-specific functions of an Ethernet driver are 109 * supplied through a set of pointers contained in a generic structure of type 110 * *eth_dev_ops*. 111 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev* 112 * structure by the device init function of the Ethernet driver, which is 113 * invoked during the PCI probing phase, as explained earlier. 114 * 115 * In other words, each function of the Ethernet API simply retrieves the 116 * *rte_eth_dev* structure associated with the device port identifier and 117 * performs an indirect invocation of the corresponding driver function 118 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure. 119 * 120 * For performance reasons, the address of the burst-oriented Rx and Tx 121 * functions of the Ethernet driver are not contained in the *eth_dev_ops* 122 * structure. Instead, they are directly stored at the beginning of the 123 * *rte_eth_dev* structure to avoid an extra indirect memory access during 124 * their invocation. 125 * 126 * RTE Ethernet device drivers do not use interrupts for transmitting or 127 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit 128 * functions to applications. 129 * Both receive and transmit functions are packet-burst oriented to minimize 130 * their cost per packet through the following optimizations: 131 * 132 * - Sharing among multiple packets the incompressible cost of the 133 * invocation of receive/transmit functions. 134 * 135 * - Enabling receive/transmit functions to take advantage of burst-oriented 136 * hardware features (L1 cache, prefetch instructions, NIC head/tail 137 * registers) to minimize the number of CPU cycles per packet, for instance, 138 * by avoiding useless read memory accesses to ring descriptors, or by 139 * systematically using arrays of pointers that exactly fit L1 cache line 140 * boundaries and sizes. 141 * 142 * The burst-oriented receive function does not provide any error notification, 143 * to avoid the corresponding overhead. As a hint, the upper-level application 144 * might check the status of the device link once being systematically returned 145 * a 0 value by the receive function of the driver for a given number of tries. 146 */ 147 148 #ifdef __cplusplus 149 extern "C" { 150 #endif 151 152 #include <stdint.h> 153 154 /* Use this macro to check if LRO API is supported */ 155 #define RTE_ETHDEV_HAS_LRO_SUPPORT 156 157 /* Alias RTE_LIBRTE_ETHDEV_DEBUG for backward compatibility. */ 158 #ifdef RTE_LIBRTE_ETHDEV_DEBUG 159 #define RTE_ETHDEV_DEBUG_RX 160 #define RTE_ETHDEV_DEBUG_TX 161 #endif 162 163 #include <rte_cman.h> 164 #include <rte_compat.h> 165 #include <rte_log.h> 166 #include <rte_interrupts.h> 167 #include <rte_dev.h> 168 #include <rte_devargs.h> 169 #include <rte_bitops.h> 170 #include <rte_errno.h> 171 #include <rte_common.h> 172 #include <rte_config.h> 173 #include <rte_power_intrinsics.h> 174 175 #include "rte_ethdev_trace_fp.h" 176 #include "rte_dev_info.h" 177 178 extern int rte_eth_dev_logtype; 179 180 #define RTE_ETHDEV_LOG(level, ...) \ 181 rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__) 182 183 struct rte_mbuf; 184 185 /** 186 * Initializes a device iterator. 187 * 188 * This iterator allows accessing a list of devices matching some devargs. 189 * 190 * @param iter 191 * Device iterator handle initialized by the function. 192 * The fields bus_str and cls_str might be dynamically allocated, 193 * and could be freed by calling rte_eth_iterator_cleanup(). 194 * 195 * @param devargs 196 * Device description string. 197 * 198 * @return 199 * 0 on successful initialization, negative otherwise. 200 */ 201 int rte_eth_iterator_init(struct rte_dev_iterator *iter, const char *devargs); 202 203 /** 204 * Iterates on devices with devargs filter. 205 * The ownership is not checked. 206 * 207 * The next port ID is returned, and the iterator is updated. 208 * 209 * @param iter 210 * Device iterator handle initialized by rte_eth_iterator_init(). 211 * Some fields bus_str and cls_str might be freed when no more port is found, 212 * by calling rte_eth_iterator_cleanup(). 213 * 214 * @return 215 * A port ID if found, RTE_MAX_ETHPORTS otherwise. 216 */ 217 uint16_t rte_eth_iterator_next(struct rte_dev_iterator *iter); 218 219 /** 220 * Free some allocated fields of the iterator. 221 * 222 * This function is automatically called by rte_eth_iterator_next() 223 * on the last iteration (i.e. when no more matching port is found). 224 * 225 * It is safe to call this function twice; it will do nothing more. 226 * 227 * @param iter 228 * Device iterator handle initialized by rte_eth_iterator_init(). 229 * The fields bus_str and cls_str are freed if needed. 230 */ 231 void rte_eth_iterator_cleanup(struct rte_dev_iterator *iter); 232 233 /** 234 * Macro to iterate over all ethdev ports matching some devargs. 235 * 236 * If a break is done before the end of the loop, 237 * the function rte_eth_iterator_cleanup() must be called. 238 * 239 * @param id 240 * Iterated port ID of type uint16_t. 241 * @param devargs 242 * Device parameters input as string of type char*. 243 * @param iter 244 * Iterator handle of type struct rte_dev_iterator, used internally. 245 */ 246 #define RTE_ETH_FOREACH_MATCHING_DEV(id, devargs, iter) \ 247 for (rte_eth_iterator_init(iter, devargs), \ 248 id = rte_eth_iterator_next(iter); \ 249 id != RTE_MAX_ETHPORTS; \ 250 id = rte_eth_iterator_next(iter)) 251 252 /** 253 * A structure used to retrieve statistics for an Ethernet port. 254 * Not all statistics fields in struct rte_eth_stats are supported 255 * by any type of network interface card (NIC). If any statistics 256 * field is not supported, its value is 0. 257 * All byte-related statistics do not include Ethernet FCS regardless 258 * of whether these bytes have been delivered to the application 259 * (see RTE_ETH_RX_OFFLOAD_KEEP_CRC). 260 */ 261 struct rte_eth_stats { 262 uint64_t ipackets; /**< Total number of successfully received packets. */ 263 uint64_t opackets; /**< Total number of successfully transmitted packets.*/ 264 uint64_t ibytes; /**< Total number of successfully received bytes. */ 265 uint64_t obytes; /**< Total number of successfully transmitted bytes. */ 266 /** 267 * Total of Rx packets dropped by the HW, 268 * because there are no available buffer (i.e. Rx queues are full). 269 */ 270 uint64_t imissed; 271 uint64_t ierrors; /**< Total number of erroneous received packets. */ 272 uint64_t oerrors; /**< Total number of failed transmitted packets. */ 273 uint64_t rx_nombuf; /**< Total number of Rx mbuf allocation failures. */ 274 /* Queue stats are limited to max 256 queues */ 275 /** Total number of queue Rx packets. */ 276 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS]; 277 /** Total number of queue Tx packets. */ 278 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS]; 279 /** Total number of successfully received queue bytes. */ 280 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS]; 281 /** Total number of successfully transmitted queue bytes. */ 282 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS]; 283 /** Total number of queue packets received that are dropped. */ 284 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS]; 285 }; 286 287 /**@{@name Link speed capabilities 288 * Device supported speeds bitmap flags 289 */ 290 #define RTE_ETH_LINK_SPEED_AUTONEG 0 /**< Autonegotiate (all speeds) */ 291 #define RTE_ETH_LINK_SPEED_FIXED RTE_BIT32(0) /**< Disable autoneg (fixed speed) */ 292 #define RTE_ETH_LINK_SPEED_10M_HD RTE_BIT32(1) /**< 10 Mbps half-duplex */ 293 #define RTE_ETH_LINK_SPEED_10M RTE_BIT32(2) /**< 10 Mbps full-duplex */ 294 #define RTE_ETH_LINK_SPEED_100M_HD RTE_BIT32(3) /**< 100 Mbps half-duplex */ 295 #define RTE_ETH_LINK_SPEED_100M RTE_BIT32(4) /**< 100 Mbps full-duplex */ 296 #define RTE_ETH_LINK_SPEED_1G RTE_BIT32(5) /**< 1 Gbps */ 297 #define RTE_ETH_LINK_SPEED_2_5G RTE_BIT32(6) /**< 2.5 Gbps */ 298 #define RTE_ETH_LINK_SPEED_5G RTE_BIT32(7) /**< 5 Gbps */ 299 #define RTE_ETH_LINK_SPEED_10G RTE_BIT32(8) /**< 10 Gbps */ 300 #define RTE_ETH_LINK_SPEED_20G RTE_BIT32(9) /**< 20 Gbps */ 301 #define RTE_ETH_LINK_SPEED_25G RTE_BIT32(10) /**< 25 Gbps */ 302 #define RTE_ETH_LINK_SPEED_40G RTE_BIT32(11) /**< 40 Gbps */ 303 #define RTE_ETH_LINK_SPEED_50G RTE_BIT32(12) /**< 50 Gbps */ 304 #define RTE_ETH_LINK_SPEED_56G RTE_BIT32(13) /**< 56 Gbps */ 305 #define RTE_ETH_LINK_SPEED_100G RTE_BIT32(14) /**< 100 Gbps */ 306 #define RTE_ETH_LINK_SPEED_200G RTE_BIT32(15) /**< 200 Gbps */ 307 #define RTE_ETH_LINK_SPEED_400G RTE_BIT32(16) /**< 400 Gbps */ 308 /**@}*/ 309 310 /**@{@name Link speed 311 * Ethernet numeric link speeds in Mbps 312 */ 313 #define RTE_ETH_SPEED_NUM_NONE 0 /**< Not defined */ 314 #define RTE_ETH_SPEED_NUM_10M 10 /**< 10 Mbps */ 315 #define RTE_ETH_SPEED_NUM_100M 100 /**< 100 Mbps */ 316 #define RTE_ETH_SPEED_NUM_1G 1000 /**< 1 Gbps */ 317 #define RTE_ETH_SPEED_NUM_2_5G 2500 /**< 2.5 Gbps */ 318 #define RTE_ETH_SPEED_NUM_5G 5000 /**< 5 Gbps */ 319 #define RTE_ETH_SPEED_NUM_10G 10000 /**< 10 Gbps */ 320 #define RTE_ETH_SPEED_NUM_20G 20000 /**< 20 Gbps */ 321 #define RTE_ETH_SPEED_NUM_25G 25000 /**< 25 Gbps */ 322 #define RTE_ETH_SPEED_NUM_40G 40000 /**< 40 Gbps */ 323 #define RTE_ETH_SPEED_NUM_50G 50000 /**< 50 Gbps */ 324 #define RTE_ETH_SPEED_NUM_56G 56000 /**< 56 Gbps */ 325 #define RTE_ETH_SPEED_NUM_100G 100000 /**< 100 Gbps */ 326 #define RTE_ETH_SPEED_NUM_200G 200000 /**< 200 Gbps */ 327 #define RTE_ETH_SPEED_NUM_400G 400000 /**< 400 Gbps */ 328 #define RTE_ETH_SPEED_NUM_UNKNOWN UINT32_MAX /**< Unknown */ 329 /**@}*/ 330 331 /** 332 * A structure used to retrieve link-level information of an Ethernet port. 333 */ 334 __extension__ 335 struct rte_eth_link { 336 uint32_t link_speed; /**< RTE_ETH_SPEED_NUM_ */ 337 uint16_t link_duplex : 1; /**< RTE_ETH_LINK_[HALF/FULL]_DUPLEX */ 338 uint16_t link_autoneg : 1; /**< RTE_ETH_LINK_[AUTONEG/FIXED] */ 339 uint16_t link_status : 1; /**< RTE_ETH_LINK_[DOWN/UP] */ 340 } __rte_aligned(8); /**< aligned for atomic64 read/write */ 341 342 /**@{@name Link negotiation 343 * Constants used in link management. 344 */ 345 #define RTE_ETH_LINK_HALF_DUPLEX 0 /**< Half-duplex connection (see link_duplex). */ 346 #define RTE_ETH_LINK_FULL_DUPLEX 1 /**< Full-duplex connection (see link_duplex). */ 347 #define RTE_ETH_LINK_DOWN 0 /**< Link is down (see link_status). */ 348 #define RTE_ETH_LINK_UP 1 /**< Link is up (see link_status). */ 349 #define RTE_ETH_LINK_FIXED 0 /**< No autonegotiation (see link_autoneg). */ 350 #define RTE_ETH_LINK_AUTONEG 1 /**< Autonegotiated (see link_autoneg). */ 351 #define RTE_ETH_LINK_MAX_STR_LEN 40 /**< Max length of default link string. */ 352 /**@}*/ 353 354 /** 355 * A structure used to configure the ring threshold registers of an Rx/Tx 356 * queue for an Ethernet port. 357 */ 358 struct rte_eth_thresh { 359 uint8_t pthresh; /**< Ring prefetch threshold. */ 360 uint8_t hthresh; /**< Ring host threshold. */ 361 uint8_t wthresh; /**< Ring writeback threshold. */ 362 }; 363 364 /**@{@name Multi-queue mode 365 * @see rte_eth_conf.rxmode.mq_mode. 366 */ 367 #define RTE_ETH_MQ_RX_RSS_FLAG RTE_BIT32(0) /**< Enable RSS. @see rte_eth_rss_conf */ 368 #define RTE_ETH_MQ_RX_DCB_FLAG RTE_BIT32(1) /**< Enable DCB. */ 369 #define RTE_ETH_MQ_RX_VMDQ_FLAG RTE_BIT32(2) /**< Enable VMDq. */ 370 /**@}*/ 371 372 /** 373 * A set of values to identify what method is to be used to route 374 * packets to multiple queues. 375 */ 376 enum rte_eth_rx_mq_mode { 377 /** None of DCB, RSS or VMDq mode */ 378 RTE_ETH_MQ_RX_NONE = 0, 379 380 /** For Rx side, only RSS is on */ 381 RTE_ETH_MQ_RX_RSS = RTE_ETH_MQ_RX_RSS_FLAG, 382 /** For Rx side,only DCB is on. */ 383 RTE_ETH_MQ_RX_DCB = RTE_ETH_MQ_RX_DCB_FLAG, 384 /** Both DCB and RSS enable */ 385 RTE_ETH_MQ_RX_DCB_RSS = RTE_ETH_MQ_RX_RSS_FLAG | RTE_ETH_MQ_RX_DCB_FLAG, 386 387 /** Only VMDq, no RSS nor DCB */ 388 RTE_ETH_MQ_RX_VMDQ_ONLY = RTE_ETH_MQ_RX_VMDQ_FLAG, 389 /** RSS mode with VMDq */ 390 RTE_ETH_MQ_RX_VMDQ_RSS = RTE_ETH_MQ_RX_RSS_FLAG | RTE_ETH_MQ_RX_VMDQ_FLAG, 391 /** Use VMDq+DCB to route traffic to queues */ 392 RTE_ETH_MQ_RX_VMDQ_DCB = RTE_ETH_MQ_RX_VMDQ_FLAG | RTE_ETH_MQ_RX_DCB_FLAG, 393 /** Enable both VMDq and DCB in VMDq */ 394 RTE_ETH_MQ_RX_VMDQ_DCB_RSS = RTE_ETH_MQ_RX_RSS_FLAG | RTE_ETH_MQ_RX_DCB_FLAG | 395 RTE_ETH_MQ_RX_VMDQ_FLAG, 396 }; 397 398 /** 399 * A set of values to identify what method is to be used to transmit 400 * packets using multi-TCs. 401 */ 402 enum rte_eth_tx_mq_mode { 403 RTE_ETH_MQ_TX_NONE = 0, /**< It is in neither DCB nor VT mode. */ 404 RTE_ETH_MQ_TX_DCB, /**< For Tx side,only DCB is on. */ 405 RTE_ETH_MQ_TX_VMDQ_DCB, /**< For Tx side,both DCB and VT is on. */ 406 RTE_ETH_MQ_TX_VMDQ_ONLY, /**< Only VT on, no DCB */ 407 }; 408 409 /** 410 * A structure used to configure the Rx features of an Ethernet port. 411 */ 412 struct rte_eth_rxmode { 413 /** The multi-queue packet distribution mode to be used, e.g. RSS. */ 414 enum rte_eth_rx_mq_mode mq_mode; 415 uint32_t mtu; /**< Requested MTU. */ 416 /** Maximum allowed size of LRO aggregated packet. */ 417 uint32_t max_lro_pkt_size; 418 /** 419 * Per-port Rx offloads to be set using RTE_ETH_RX_OFFLOAD_* flags. 420 * Only offloads set on rx_offload_capa field on rte_eth_dev_info 421 * structure are allowed to be set. 422 */ 423 uint64_t offloads; 424 425 uint64_t reserved_64s[2]; /**< Reserved for future fields */ 426 void *reserved_ptrs[2]; /**< Reserved for future fields */ 427 }; 428 429 /** 430 * VLAN types to indicate if it is for single VLAN, inner VLAN or outer VLAN. 431 * Note that single VLAN is treated the same as inner VLAN. 432 */ 433 enum rte_vlan_type { 434 RTE_ETH_VLAN_TYPE_UNKNOWN = 0, 435 RTE_ETH_VLAN_TYPE_INNER, /**< Inner VLAN. */ 436 RTE_ETH_VLAN_TYPE_OUTER, /**< Single VLAN, or outer VLAN. */ 437 RTE_ETH_VLAN_TYPE_MAX, 438 }; 439 440 /** 441 * A structure used to describe a VLAN filter. 442 * If the bit corresponding to a VID is set, such VID is on. 443 */ 444 struct rte_vlan_filter_conf { 445 uint64_t ids[64]; 446 }; 447 448 /** 449 * A structure used to configure the Receive Side Scaling (RSS) feature 450 * of an Ethernet port. 451 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points 452 * to an array holding the RSS key to use for hashing specific header 453 * fields of received packets. The length of this array should be indicated 454 * by *rss_key_len* below. Otherwise, a default random hash key is used by 455 * the device driver. 456 * 457 * The *rss_key_len* field of the *rss_conf* structure indicates the length 458 * in bytes of the array pointed by *rss_key*. To be compatible, this length 459 * will be checked in i40e only. Others assume 40 bytes to be used as before. 460 * 461 * The *rss_hf* field of the *rss_conf* structure indicates the different 462 * types of IPv4/IPv6 packets to which the RSS hashing must be applied. 463 * Supplying an *rss_hf* equal to zero disables the RSS feature. 464 */ 465 struct rte_eth_rss_conf { 466 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */ 467 uint8_t rss_key_len; /**< hash key length in bytes. */ 468 uint64_t rss_hf; /**< Hash functions to apply - see below. */ 469 }; 470 471 /* 472 * A packet can be identified by hardware as different flow types. Different 473 * NIC hardware may support different flow types. 474 * Basically, the NIC hardware identifies the flow type as deep protocol as 475 * possible, and exclusively. For example, if a packet is identified as 476 * 'RTE_ETH_FLOW_NONFRAG_IPV4_TCP', it will not be any of other flow types, 477 * though it is an actual IPV4 packet. 478 */ 479 #define RTE_ETH_FLOW_UNKNOWN 0 480 #define RTE_ETH_FLOW_RAW 1 481 #define RTE_ETH_FLOW_IPV4 2 482 #define RTE_ETH_FLOW_FRAG_IPV4 3 483 #define RTE_ETH_FLOW_NONFRAG_IPV4_TCP 4 484 #define RTE_ETH_FLOW_NONFRAG_IPV4_UDP 5 485 #define RTE_ETH_FLOW_NONFRAG_IPV4_SCTP 6 486 #define RTE_ETH_FLOW_NONFRAG_IPV4_OTHER 7 487 #define RTE_ETH_FLOW_IPV6 8 488 #define RTE_ETH_FLOW_FRAG_IPV6 9 489 #define RTE_ETH_FLOW_NONFRAG_IPV6_TCP 10 490 #define RTE_ETH_FLOW_NONFRAG_IPV6_UDP 11 491 #define RTE_ETH_FLOW_NONFRAG_IPV6_SCTP 12 492 #define RTE_ETH_FLOW_NONFRAG_IPV6_OTHER 13 493 #define RTE_ETH_FLOW_L2_PAYLOAD 14 494 #define RTE_ETH_FLOW_IPV6_EX 15 495 #define RTE_ETH_FLOW_IPV6_TCP_EX 16 496 #define RTE_ETH_FLOW_IPV6_UDP_EX 17 497 /** Consider device port number as a flow differentiator */ 498 #define RTE_ETH_FLOW_PORT 18 499 #define RTE_ETH_FLOW_VXLAN 19 /**< VXLAN protocol based flow */ 500 #define RTE_ETH_FLOW_GENEVE 20 /**< GENEVE protocol based flow */ 501 #define RTE_ETH_FLOW_NVGRE 21 /**< NVGRE protocol based flow */ 502 #define RTE_ETH_FLOW_VXLAN_GPE 22 /**< VXLAN-GPE protocol based flow */ 503 #define RTE_ETH_FLOW_GTPU 23 /**< GTPU protocol based flow */ 504 #define RTE_ETH_FLOW_MAX 24 505 506 /* 507 * Below macros are defined for RSS offload types, they can be used to 508 * fill rte_eth_rss_conf.rss_hf or rte_flow_action_rss.types. 509 */ 510 #define RTE_ETH_RSS_IPV4 RTE_BIT64(2) 511 #define RTE_ETH_RSS_FRAG_IPV4 RTE_BIT64(3) 512 #define RTE_ETH_RSS_NONFRAG_IPV4_TCP RTE_BIT64(4) 513 #define RTE_ETH_RSS_NONFRAG_IPV4_UDP RTE_BIT64(5) 514 #define RTE_ETH_RSS_NONFRAG_IPV4_SCTP RTE_BIT64(6) 515 #define RTE_ETH_RSS_NONFRAG_IPV4_OTHER RTE_BIT64(7) 516 #define RTE_ETH_RSS_IPV6 RTE_BIT64(8) 517 #define RTE_ETH_RSS_FRAG_IPV6 RTE_BIT64(9) 518 #define RTE_ETH_RSS_NONFRAG_IPV6_TCP RTE_BIT64(10) 519 #define RTE_ETH_RSS_NONFRAG_IPV6_UDP RTE_BIT64(11) 520 #define RTE_ETH_RSS_NONFRAG_IPV6_SCTP RTE_BIT64(12) 521 #define RTE_ETH_RSS_NONFRAG_IPV6_OTHER RTE_BIT64(13) 522 #define RTE_ETH_RSS_L2_PAYLOAD RTE_BIT64(14) 523 #define RTE_ETH_RSS_IPV6_EX RTE_BIT64(15) 524 #define RTE_ETH_RSS_IPV6_TCP_EX RTE_BIT64(16) 525 #define RTE_ETH_RSS_IPV6_UDP_EX RTE_BIT64(17) 526 #define RTE_ETH_RSS_PORT RTE_BIT64(18) 527 #define RTE_ETH_RSS_VXLAN RTE_BIT64(19) 528 #define RTE_ETH_RSS_GENEVE RTE_BIT64(20) 529 #define RTE_ETH_RSS_NVGRE RTE_BIT64(21) 530 #define RTE_ETH_RSS_GTPU RTE_BIT64(23) 531 #define RTE_ETH_RSS_ETH RTE_BIT64(24) 532 #define RTE_ETH_RSS_S_VLAN RTE_BIT64(25) 533 #define RTE_ETH_RSS_C_VLAN RTE_BIT64(26) 534 #define RTE_ETH_RSS_ESP RTE_BIT64(27) 535 #define RTE_ETH_RSS_AH RTE_BIT64(28) 536 #define RTE_ETH_RSS_L2TPV3 RTE_BIT64(29) 537 #define RTE_ETH_RSS_PFCP RTE_BIT64(30) 538 #define RTE_ETH_RSS_PPPOE RTE_BIT64(31) 539 #define RTE_ETH_RSS_ECPRI RTE_BIT64(32) 540 #define RTE_ETH_RSS_MPLS RTE_BIT64(33) 541 #define RTE_ETH_RSS_IPV4_CHKSUM RTE_BIT64(34) 542 543 /** 544 * The RTE_ETH_RSS_L4_CHKSUM works on checksum field of any L4 header. 545 * It is similar to RTE_ETH_RSS_PORT that they don't specify the specific type of 546 * L4 header. This macro is defined to replace some specific L4 (TCP/UDP/SCTP) 547 * checksum type for constructing the use of RSS offload bits. 548 * 549 * Due to above reason, some old APIs (and configuration) don't support 550 * RTE_ETH_RSS_L4_CHKSUM. The rte_flow RSS API supports it. 551 * 552 * For the case that checksum is not used in an UDP header, 553 * it takes the reserved value 0 as input for the hash function. 554 */ 555 #define RTE_ETH_RSS_L4_CHKSUM RTE_BIT64(35) 556 557 #define RTE_ETH_RSS_L2TPV2 RTE_BIT64(36) 558 559 /* 560 * We use the following macros to combine with above RTE_ETH_RSS_* for 561 * more specific input set selection. These bits are defined starting 562 * from the high end of the 64 bits. 563 * Note: If we use above RTE_ETH_RSS_* without SRC/DST_ONLY, it represents 564 * both SRC and DST are taken into account. If SRC_ONLY and DST_ONLY of 565 * the same level are used simultaneously, it is the same case as none of 566 * them are added. 567 */ 568 #define RTE_ETH_RSS_L3_SRC_ONLY RTE_BIT64(63) 569 #define RTE_ETH_RSS_L3_DST_ONLY RTE_BIT64(62) 570 #define RTE_ETH_RSS_L4_SRC_ONLY RTE_BIT64(61) 571 #define RTE_ETH_RSS_L4_DST_ONLY RTE_BIT64(60) 572 #define RTE_ETH_RSS_L2_SRC_ONLY RTE_BIT64(59) 573 #define RTE_ETH_RSS_L2_DST_ONLY RTE_BIT64(58) 574 575 /* 576 * Only select IPV6 address prefix as RSS input set according to 577 * https://tools.ietf.org/html/rfc6052 578 * Must be combined with RTE_ETH_RSS_IPV6, RTE_ETH_RSS_NONFRAG_IPV6_UDP, 579 * RTE_ETH_RSS_NONFRAG_IPV6_TCP, RTE_ETH_RSS_NONFRAG_IPV6_SCTP. 580 */ 581 #define RTE_ETH_RSS_L3_PRE32 RTE_BIT64(57) 582 #define RTE_ETH_RSS_L3_PRE40 RTE_BIT64(56) 583 #define RTE_ETH_RSS_L3_PRE48 RTE_BIT64(55) 584 #define RTE_ETH_RSS_L3_PRE56 RTE_BIT64(54) 585 #define RTE_ETH_RSS_L3_PRE64 RTE_BIT64(53) 586 #define RTE_ETH_RSS_L3_PRE96 RTE_BIT64(52) 587 588 /* 589 * Use the following macros to combine with the above layers 590 * to choose inner and outer layers or both for RSS computation. 591 * Bits 50 and 51 are reserved for this. 592 */ 593 594 /** 595 * level 0, requests the default behavior. 596 * Depending on the packet type, it can mean outermost, innermost, 597 * anything in between or even no RSS. 598 * It basically stands for the innermost encapsulation level RSS 599 * can be performed on according to PMD and device capabilities. 600 */ 601 #define RTE_ETH_RSS_LEVEL_PMD_DEFAULT (UINT64_C(0) << 50) 602 603 /** 604 * level 1, requests RSS to be performed on the outermost packet 605 * encapsulation level. 606 */ 607 #define RTE_ETH_RSS_LEVEL_OUTERMOST (UINT64_C(1) << 50) 608 609 /** 610 * level 2, requests RSS to be performed on the specified inner packet 611 * encapsulation level, from outermost to innermost (lower to higher values). 612 */ 613 #define RTE_ETH_RSS_LEVEL_INNERMOST (UINT64_C(2) << 50) 614 #define RTE_ETH_RSS_LEVEL_MASK (UINT64_C(3) << 50) 615 616 #define RTE_ETH_RSS_LEVEL(rss_hf) ((rss_hf & RTE_ETH_RSS_LEVEL_MASK) >> 50) 617 618 /** 619 * For input set change of hash filter, if SRC_ONLY and DST_ONLY of 620 * the same level are used simultaneously, it is the same case as 621 * none of them are added. 622 * 623 * @param rss_hf 624 * RSS types with SRC/DST_ONLY. 625 * @return 626 * RSS types. 627 */ 628 static inline uint64_t 629 rte_eth_rss_hf_refine(uint64_t rss_hf) 630 { 631 if ((rss_hf & RTE_ETH_RSS_L3_SRC_ONLY) && (rss_hf & RTE_ETH_RSS_L3_DST_ONLY)) 632 rss_hf &= ~(RTE_ETH_RSS_L3_SRC_ONLY | RTE_ETH_RSS_L3_DST_ONLY); 633 634 if ((rss_hf & RTE_ETH_RSS_L4_SRC_ONLY) && (rss_hf & RTE_ETH_RSS_L4_DST_ONLY)) 635 rss_hf &= ~(RTE_ETH_RSS_L4_SRC_ONLY | RTE_ETH_RSS_L4_DST_ONLY); 636 637 return rss_hf; 638 } 639 640 #define RTE_ETH_RSS_IPV6_PRE32 ( \ 641 RTE_ETH_RSS_IPV6 | \ 642 RTE_ETH_RSS_L3_PRE32) 643 644 #define RTE_ETH_RSS_IPV6_PRE40 ( \ 645 RTE_ETH_RSS_IPV6 | \ 646 RTE_ETH_RSS_L3_PRE40) 647 648 #define RTE_ETH_RSS_IPV6_PRE48 ( \ 649 RTE_ETH_RSS_IPV6 | \ 650 RTE_ETH_RSS_L3_PRE48) 651 652 #define RTE_ETH_RSS_IPV6_PRE56 ( \ 653 RTE_ETH_RSS_IPV6 | \ 654 RTE_ETH_RSS_L3_PRE56) 655 656 #define RTE_ETH_RSS_IPV6_PRE64 ( \ 657 RTE_ETH_RSS_IPV6 | \ 658 RTE_ETH_RSS_L3_PRE64) 659 660 #define RTE_ETH_RSS_IPV6_PRE96 ( \ 661 RTE_ETH_RSS_IPV6 | \ 662 RTE_ETH_RSS_L3_PRE96) 663 664 #define RTE_ETH_RSS_IPV6_PRE32_UDP ( \ 665 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 666 RTE_ETH_RSS_L3_PRE32) 667 668 #define RTE_ETH_RSS_IPV6_PRE40_UDP ( \ 669 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 670 RTE_ETH_RSS_L3_PRE40) 671 672 #define RTE_ETH_RSS_IPV6_PRE48_UDP ( \ 673 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 674 RTE_ETH_RSS_L3_PRE48) 675 676 #define RTE_ETH_RSS_IPV6_PRE56_UDP ( \ 677 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 678 RTE_ETH_RSS_L3_PRE56) 679 680 #define RTE_ETH_RSS_IPV6_PRE64_UDP ( \ 681 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 682 RTE_ETH_RSS_L3_PRE64) 683 684 #define RTE_ETH_RSS_IPV6_PRE96_UDP ( \ 685 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 686 RTE_ETH_RSS_L3_PRE96) 687 688 #define RTE_ETH_RSS_IPV6_PRE32_TCP ( \ 689 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 690 RTE_ETH_RSS_L3_PRE32) 691 692 #define RTE_ETH_RSS_IPV6_PRE40_TCP ( \ 693 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 694 RTE_ETH_RSS_L3_PRE40) 695 696 #define RTE_ETH_RSS_IPV6_PRE48_TCP ( \ 697 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 698 RTE_ETH_RSS_L3_PRE48) 699 700 #define RTE_ETH_RSS_IPV6_PRE56_TCP ( \ 701 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 702 RTE_ETH_RSS_L3_PRE56) 703 704 #define RTE_ETH_RSS_IPV6_PRE64_TCP ( \ 705 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 706 RTE_ETH_RSS_L3_PRE64) 707 708 #define RTE_ETH_RSS_IPV6_PRE96_TCP ( \ 709 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 710 RTE_ETH_RSS_L3_PRE96) 711 712 #define RTE_ETH_RSS_IPV6_PRE32_SCTP ( \ 713 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 714 RTE_ETH_RSS_L3_PRE32) 715 716 #define RTE_ETH_RSS_IPV6_PRE40_SCTP ( \ 717 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 718 RTE_ETH_RSS_L3_PRE40) 719 720 #define RTE_ETH_RSS_IPV6_PRE48_SCTP ( \ 721 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 722 RTE_ETH_RSS_L3_PRE48) 723 724 #define RTE_ETH_RSS_IPV6_PRE56_SCTP ( \ 725 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 726 RTE_ETH_RSS_L3_PRE56) 727 728 #define RTE_ETH_RSS_IPV6_PRE64_SCTP ( \ 729 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 730 RTE_ETH_RSS_L3_PRE64) 731 732 #define RTE_ETH_RSS_IPV6_PRE96_SCTP ( \ 733 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 734 RTE_ETH_RSS_L3_PRE96) 735 736 #define RTE_ETH_RSS_IP ( \ 737 RTE_ETH_RSS_IPV4 | \ 738 RTE_ETH_RSS_FRAG_IPV4 | \ 739 RTE_ETH_RSS_NONFRAG_IPV4_OTHER | \ 740 RTE_ETH_RSS_IPV6 | \ 741 RTE_ETH_RSS_FRAG_IPV6 | \ 742 RTE_ETH_RSS_NONFRAG_IPV6_OTHER | \ 743 RTE_ETH_RSS_IPV6_EX) 744 745 #define RTE_ETH_RSS_UDP ( \ 746 RTE_ETH_RSS_NONFRAG_IPV4_UDP | \ 747 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 748 RTE_ETH_RSS_IPV6_UDP_EX) 749 750 #define RTE_ETH_RSS_TCP ( \ 751 RTE_ETH_RSS_NONFRAG_IPV4_TCP | \ 752 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 753 RTE_ETH_RSS_IPV6_TCP_EX) 754 755 #define RTE_ETH_RSS_SCTP ( \ 756 RTE_ETH_RSS_NONFRAG_IPV4_SCTP | \ 757 RTE_ETH_RSS_NONFRAG_IPV6_SCTP) 758 759 #define RTE_ETH_RSS_TUNNEL ( \ 760 RTE_ETH_RSS_VXLAN | \ 761 RTE_ETH_RSS_GENEVE | \ 762 RTE_ETH_RSS_NVGRE) 763 764 #define RTE_ETH_RSS_VLAN ( \ 765 RTE_ETH_RSS_S_VLAN | \ 766 RTE_ETH_RSS_C_VLAN) 767 768 /** Mask of valid RSS hash protocols */ 769 #define RTE_ETH_RSS_PROTO_MASK ( \ 770 RTE_ETH_RSS_IPV4 | \ 771 RTE_ETH_RSS_FRAG_IPV4 | \ 772 RTE_ETH_RSS_NONFRAG_IPV4_TCP | \ 773 RTE_ETH_RSS_NONFRAG_IPV4_UDP | \ 774 RTE_ETH_RSS_NONFRAG_IPV4_SCTP | \ 775 RTE_ETH_RSS_NONFRAG_IPV4_OTHER | \ 776 RTE_ETH_RSS_IPV6 | \ 777 RTE_ETH_RSS_FRAG_IPV6 | \ 778 RTE_ETH_RSS_NONFRAG_IPV6_TCP | \ 779 RTE_ETH_RSS_NONFRAG_IPV6_UDP | \ 780 RTE_ETH_RSS_NONFRAG_IPV6_SCTP | \ 781 RTE_ETH_RSS_NONFRAG_IPV6_OTHER | \ 782 RTE_ETH_RSS_L2_PAYLOAD | \ 783 RTE_ETH_RSS_IPV6_EX | \ 784 RTE_ETH_RSS_IPV6_TCP_EX | \ 785 RTE_ETH_RSS_IPV6_UDP_EX | \ 786 RTE_ETH_RSS_PORT | \ 787 RTE_ETH_RSS_VXLAN | \ 788 RTE_ETH_RSS_GENEVE | \ 789 RTE_ETH_RSS_NVGRE | \ 790 RTE_ETH_RSS_MPLS) 791 792 /* 793 * Definitions used for redirection table entry size. 794 * Some RSS RETA sizes may not be supported by some drivers, check the 795 * documentation or the description of relevant functions for more details. 796 */ 797 #define RTE_ETH_RSS_RETA_SIZE_64 64 798 #define RTE_ETH_RSS_RETA_SIZE_128 128 799 #define RTE_ETH_RSS_RETA_SIZE_256 256 800 #define RTE_ETH_RSS_RETA_SIZE_512 512 801 #define RTE_ETH_RETA_GROUP_SIZE 64 802 803 /**@{@name VMDq and DCB maximums */ 804 #define RTE_ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDq VLAN filters. */ 805 #define RTE_ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */ 806 #define RTE_ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDq DCB queues. */ 807 #define RTE_ETH_DCB_NUM_QUEUES 128 /**< Maximum nb. of DCB queues. */ 808 /**@}*/ 809 810 /**@{@name DCB capabilities */ 811 #define RTE_ETH_DCB_PG_SUPPORT RTE_BIT32(0) /**< Priority Group(ETS) support. */ 812 #define RTE_ETH_DCB_PFC_SUPPORT RTE_BIT32(1) /**< Priority Flow Control support. */ 813 /**@}*/ 814 815 /**@{@name VLAN offload bits */ 816 #define RTE_ETH_VLAN_STRIP_OFFLOAD 0x0001 /**< VLAN Strip On/Off */ 817 #define RTE_ETH_VLAN_FILTER_OFFLOAD 0x0002 /**< VLAN Filter On/Off */ 818 #define RTE_ETH_VLAN_EXTEND_OFFLOAD 0x0004 /**< VLAN Extend On/Off */ 819 #define RTE_ETH_QINQ_STRIP_OFFLOAD 0x0008 /**< QINQ Strip On/Off */ 820 821 #define RTE_ETH_VLAN_STRIP_MASK 0x0001 /**< VLAN Strip setting mask */ 822 #define RTE_ETH_VLAN_FILTER_MASK 0x0002 /**< VLAN Filter setting mask*/ 823 #define RTE_ETH_VLAN_EXTEND_MASK 0x0004 /**< VLAN Extend setting mask*/ 824 #define RTE_ETH_QINQ_STRIP_MASK 0x0008 /**< QINQ Strip setting mask */ 825 #define RTE_ETH_VLAN_ID_MAX 0x0FFF /**< VLAN ID is in lower 12 bits*/ 826 /**@}*/ 827 828 /* Definitions used for receive MAC address */ 829 #define RTE_ETH_NUM_RECEIVE_MAC_ADDR 128 /**< Maximum nb. of receive mac addr. */ 830 831 /* Definitions used for unicast hash */ 832 #define RTE_ETH_VMDQ_NUM_UC_HASH_ARRAY 128 /**< Maximum nb. of UC hash array. */ 833 834 /**@{@name VMDq Rx mode 835 * @see rte_eth_vmdq_rx_conf.rx_mode 836 */ 837 /** Accept untagged packets. */ 838 #define RTE_ETH_VMDQ_ACCEPT_UNTAG RTE_BIT32(0) 839 /** Accept packets in multicast table. */ 840 #define RTE_ETH_VMDQ_ACCEPT_HASH_MC RTE_BIT32(1) 841 /** Accept packets in unicast table. */ 842 #define RTE_ETH_VMDQ_ACCEPT_HASH_UC RTE_BIT32(2) 843 /** Accept broadcast packets. */ 844 #define RTE_ETH_VMDQ_ACCEPT_BROADCAST RTE_BIT32(3) 845 /** Multicast promiscuous. */ 846 #define RTE_ETH_VMDQ_ACCEPT_MULTICAST RTE_BIT32(4) 847 /**@}*/ 848 849 /** 850 * A structure used to configure 64 entries of Redirection Table of the 851 * Receive Side Scaling (RSS) feature of an Ethernet port. To configure 852 * more than 64 entries supported by hardware, an array of this structure 853 * is needed. 854 */ 855 struct rte_eth_rss_reta_entry64 { 856 /** Mask bits indicate which entries need to be updated/queried. */ 857 uint64_t mask; 858 /** Group of 64 redirection table entries. */ 859 uint16_t reta[RTE_ETH_RETA_GROUP_SIZE]; 860 }; 861 862 /** 863 * This enum indicates the possible number of traffic classes 864 * in DCB configurations 865 */ 866 enum rte_eth_nb_tcs { 867 RTE_ETH_4_TCS = 4, /**< 4 TCs with DCB. */ 868 RTE_ETH_8_TCS = 8 /**< 8 TCs with DCB. */ 869 }; 870 871 /** 872 * This enum indicates the possible number of queue pools 873 * in VMDq configurations. 874 */ 875 enum rte_eth_nb_pools { 876 RTE_ETH_8_POOLS = 8, /**< 8 VMDq pools. */ 877 RTE_ETH_16_POOLS = 16, /**< 16 VMDq pools. */ 878 RTE_ETH_32_POOLS = 32, /**< 32 VMDq pools. */ 879 RTE_ETH_64_POOLS = 64 /**< 64 VMDq pools. */ 880 }; 881 882 /* This structure may be extended in future. */ 883 struct rte_eth_dcb_rx_conf { 884 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs */ 885 /** Traffic class each UP mapped to. */ 886 uint8_t dcb_tc[RTE_ETH_DCB_NUM_USER_PRIORITIES]; 887 }; 888 889 struct rte_eth_vmdq_dcb_tx_conf { 890 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools. */ 891 /** Traffic class each UP mapped to. */ 892 uint8_t dcb_tc[RTE_ETH_DCB_NUM_USER_PRIORITIES]; 893 }; 894 895 struct rte_eth_dcb_tx_conf { 896 enum rte_eth_nb_tcs nb_tcs; /**< Possible DCB TCs, 4 or 8 TCs. */ 897 /** Traffic class each UP mapped to. */ 898 uint8_t dcb_tc[RTE_ETH_DCB_NUM_USER_PRIORITIES]; 899 }; 900 901 struct rte_eth_vmdq_tx_conf { 902 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq mode, 64 pools. */ 903 }; 904 905 /** 906 * A structure used to configure the VMDq+DCB feature 907 * of an Ethernet port. 908 * 909 * Using this feature, packets are routed to a pool of queues, based 910 * on the VLAN ID in the VLAN tag, and then to a specific queue within 911 * that pool, using the user priority VLAN tag field. 912 * 913 * A default pool may be used, if desired, to route all traffic which 914 * does not match the VLAN filter rules. 915 */ 916 struct rte_eth_vmdq_dcb_conf { 917 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */ 918 uint8_t enable_default_pool; /**< If non-zero, use a default pool */ 919 uint8_t default_pool; /**< The default pool, if applicable */ 920 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */ 921 struct { 922 uint16_t vlan_id; /**< The VLAN ID of the received frame */ 923 uint64_t pools; /**< Bitmask of pools for packet Rx */ 924 } pool_map[RTE_ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq VLAN pool maps. */ 925 /** Selects a queue in a pool */ 926 uint8_t dcb_tc[RTE_ETH_DCB_NUM_USER_PRIORITIES]; 927 }; 928 929 /** 930 * A structure used to configure the VMDq feature of an Ethernet port when 931 * not combined with the DCB feature. 932 * 933 * Using this feature, packets are routed to a pool of queues. By default, 934 * the pool selection is based on the MAC address, the VLAN ID in the 935 * VLAN tag as specified in the pool_map array. 936 * Passing the RTE_ETH_VMDQ_ACCEPT_UNTAG in the rx_mode field allows pool 937 * selection using only the MAC address. MAC address to pool mapping is done 938 * using the rte_eth_dev_mac_addr_add function, with the pool parameter 939 * corresponding to the pool ID. 940 * 941 * Queue selection within the selected pool will be done using RSS when 942 * it is enabled or revert to the first queue of the pool if not. 943 * 944 * A default pool may be used, if desired, to route all traffic which 945 * does not match the VLAN filter rules or any pool MAC address. 946 */ 947 struct rte_eth_vmdq_rx_conf { 948 enum rte_eth_nb_pools nb_queue_pools; /**< VMDq only mode, 8 or 64 pools */ 949 uint8_t enable_default_pool; /**< If non-zero, use a default pool */ 950 uint8_t default_pool; /**< The default pool, if applicable */ 951 uint8_t enable_loop_back; /**< Enable VT loop back */ 952 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */ 953 uint32_t rx_mode; /**< Flags from RTE_ETH_VMDQ_ACCEPT_* */ 954 struct { 955 uint16_t vlan_id; /**< The VLAN ID of the received frame */ 956 uint64_t pools; /**< Bitmask of pools for packet Rx */ 957 } pool_map[RTE_ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq VLAN pool maps. */ 958 }; 959 960 /** 961 * A structure used to configure the Tx features of an Ethernet port. 962 */ 963 struct rte_eth_txmode { 964 enum rte_eth_tx_mq_mode mq_mode; /**< Tx multi-queues mode. */ 965 /** 966 * Per-port Tx offloads to be set using RTE_ETH_TX_OFFLOAD_* flags. 967 * Only offloads set on tx_offload_capa field on rte_eth_dev_info 968 * structure are allowed to be set. 969 */ 970 uint64_t offloads; 971 972 uint16_t pvid; 973 __extension__ 974 uint8_t /** If set, reject sending out tagged pkts */ 975 hw_vlan_reject_tagged : 1, 976 /** If set, reject sending out untagged pkts */ 977 hw_vlan_reject_untagged : 1, 978 /** If set, enable port based VLAN insertion */ 979 hw_vlan_insert_pvid : 1; 980 981 uint64_t reserved_64s[2]; /**< Reserved for future fields */ 982 void *reserved_ptrs[2]; /**< Reserved for future fields */ 983 }; 984 985 /** 986 * @warning 987 * @b EXPERIMENTAL: this structure may change without prior notice. 988 * 989 * A structure used to configure an Rx packet segment to split. 990 * 991 * If RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT flag is set in offloads field, 992 * the PMD will split the received packets into multiple segments 993 * according to the specification in the description array: 994 * 995 * - The first network buffer will be allocated from the memory pool, 996 * specified in the first array element, the second buffer, from the 997 * pool in the second element, and so on. 998 * 999 * - The proto_hdrs in the elements define the split position of 1000 * received packets. 1001 * 1002 * - The offsets from the segment description elements specify 1003 * the data offset from the buffer beginning except the first mbuf. 1004 * The first segment offset is added with RTE_PKTMBUF_HEADROOM. 1005 * 1006 * - The lengths in the elements define the maximal data amount 1007 * being received to each segment. The receiving starts with filling 1008 * up the first mbuf data buffer up to specified length. If the 1009 * there are data remaining (packet is longer than buffer in the first 1010 * mbuf) the following data will be pushed to the next segment 1011 * up to its own length, and so on. 1012 * 1013 * - If the length in the segment description element is zero 1014 * the actual buffer size will be deduced from the appropriate 1015 * memory pool properties. 1016 * 1017 * - If there is not enough elements to describe the buffer for entire 1018 * packet of maximal length the following parameters will be used 1019 * for the all remaining segments: 1020 * - pool from the last valid element 1021 * - the buffer size from this pool 1022 * - zero offset 1023 * 1024 * - Length based buffer split: 1025 * - mp, length, offset should be configured. 1026 * - The proto_hdr field must be 0. 1027 * 1028 * - Protocol header based buffer split: 1029 * - mp, offset, proto_hdr should be configured. 1030 * - The length field must be 0. 1031 * - The proto_hdr field in the last segment should be 0. 1032 * 1033 * - When protocol header split is enabled, NIC may receive packets 1034 * which do not match all the protocol headers within the Rx segments. 1035 * At this point, NIC will have two possible split behaviors according to 1036 * matching results, one is exact match, another is longest match. 1037 * The split result of NIC must belong to one of them. 1038 * The exact match means NIC only do split when the packets exactly match all 1039 * the protocol headers in the segments. 1040 * Otherwise, the whole packet will be put into the last valid mempool. 1041 * The longest match means NIC will do split until packets mismatch 1042 * the protocol header in the segments. 1043 * The rest will be put into the last valid pool. 1044 */ 1045 struct rte_eth_rxseg_split { 1046 struct rte_mempool *mp; /**< Memory pool to allocate segment from. */ 1047 uint16_t length; /**< Segment data length, configures split point. */ 1048 uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */ 1049 /** 1050 * proto_hdr defines a bit mask of the protocol sequence as RTE_PTYPE_*. 1051 * The last RTE_PTYPE* in the mask indicates the split position. 1052 * 1053 * If one protocol header is defined to split packets into two segments, 1054 * for non-tunneling packets, the complete protocol sequence should be defined. 1055 * For tunneling packets, for simplicity, only the tunnel and inner part of 1056 * complete protocol sequence is required. 1057 * If several protocol headers are defined to split packets into multi-segments, 1058 * the repeated parts of adjacent segments should be omitted. 1059 */ 1060 uint32_t proto_hdr; 1061 }; 1062 1063 /** 1064 * @warning 1065 * @b EXPERIMENTAL: this structure may change without prior notice. 1066 * 1067 * A common structure used to describe Rx packet segment properties. 1068 */ 1069 union rte_eth_rxseg { 1070 /* The settings for buffer split offload. */ 1071 struct rte_eth_rxseg_split split; 1072 /* The other features settings should be added here. */ 1073 }; 1074 1075 /** 1076 * A structure used to configure an Rx ring of an Ethernet port. 1077 */ 1078 struct rte_eth_rxconf { 1079 struct rte_eth_thresh rx_thresh; /**< Rx ring threshold registers. */ 1080 uint16_t rx_free_thresh; /**< Drives the freeing of Rx descriptors. */ 1081 uint8_t rx_drop_en; /**< Drop packets if no descriptors are available. */ 1082 uint8_t rx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */ 1083 uint16_t rx_nseg; /**< Number of descriptions in rx_seg array. */ 1084 /** 1085 * Share group index in Rx domain and switch domain. 1086 * Non-zero value to enable Rx queue share, zero value disable share. 1087 * PMD is responsible for Rx queue consistency checks to avoid member 1088 * port's configuration contradict to each other. 1089 */ 1090 uint16_t share_group; 1091 uint16_t share_qid; /**< Shared Rx queue ID in group */ 1092 /** 1093 * Per-queue Rx offloads to be set using RTE_ETH_RX_OFFLOAD_* flags. 1094 * Only offloads set on rx_queue_offload_capa or rx_offload_capa 1095 * fields on rte_eth_dev_info structure are allowed to be set. 1096 */ 1097 uint64_t offloads; 1098 /** 1099 * Points to the array of segment descriptions for an entire packet. 1100 * Array elements are properties for consecutive Rx segments. 1101 * 1102 * The supported capabilities of receiving segmentation is reported 1103 * in rte_eth_dev_info.rx_seg_capa field. 1104 */ 1105 union rte_eth_rxseg *rx_seg; 1106 1107 /** 1108 * Array of mempools to allocate Rx buffers from. 1109 * 1110 * This provides support for multiple mbuf pools per Rx queue. 1111 * The capability is reported in device info via positive 1112 * max_rx_mempools. 1113 * 1114 * It could be useful for more efficient usage of memory when an 1115 * application creates different mempools to steer the specific 1116 * size of the packet. 1117 * 1118 * If many mempools are specified, packets received using Rx 1119 * burst may belong to any provided mempool. From ethdev user point 1120 * of view it is undefined how PMD/NIC chooses mempool for a packet. 1121 * 1122 * If Rx scatter is enabled, a packet may be delivered using a chain 1123 * of mbufs obtained from single mempool or multiple mempools based 1124 * on the NIC implementation. 1125 */ 1126 struct rte_mempool **rx_mempools; 1127 uint16_t rx_nmempool; /** < Number of Rx mempools */ 1128 1129 uint64_t reserved_64s[2]; /**< Reserved for future fields */ 1130 void *reserved_ptrs[2]; /**< Reserved for future fields */ 1131 }; 1132 1133 /** 1134 * A structure used to configure a Tx ring of an Ethernet port. 1135 */ 1136 struct rte_eth_txconf { 1137 struct rte_eth_thresh tx_thresh; /**< Tx ring threshold registers. */ 1138 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */ 1139 uint16_t tx_free_thresh; /**< Start freeing Tx buffers if there are 1140 less free descriptors than this value. */ 1141 1142 uint8_t tx_deferred_start; /**< Do not start queue with rte_eth_dev_start(). */ 1143 /** 1144 * Per-queue Tx offloads to be set using RTE_ETH_TX_OFFLOAD_* flags. 1145 * Only offloads set on tx_queue_offload_capa or tx_offload_capa 1146 * fields on rte_eth_dev_info structure are allowed to be set. 1147 */ 1148 uint64_t offloads; 1149 1150 uint64_t reserved_64s[2]; /**< Reserved for future fields */ 1151 void *reserved_ptrs[2]; /**< Reserved for future fields */ 1152 }; 1153 1154 /** 1155 * @warning 1156 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1157 * 1158 * A structure used to return the Tx or Rx hairpin queue capabilities. 1159 */ 1160 struct rte_eth_hairpin_queue_cap { 1161 /** 1162 * When set, PMD supports placing descriptors and/or data buffers 1163 * in dedicated device memory. 1164 */ 1165 uint32_t locked_device_memory:1; 1166 1167 /** 1168 * When set, PMD supports placing descriptors and/or data buffers 1169 * in host memory managed by DPDK. 1170 */ 1171 uint32_t rte_memory:1; 1172 1173 uint32_t reserved:30; /**< Reserved for future fields */ 1174 }; 1175 1176 /** 1177 * @warning 1178 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1179 * 1180 * A structure used to return the hairpin capabilities that are supported. 1181 */ 1182 struct rte_eth_hairpin_cap { 1183 /** The max number of hairpin queues (different bindings). */ 1184 uint16_t max_nb_queues; 1185 /** Max number of Rx queues to be connected to one Tx queue. */ 1186 uint16_t max_rx_2_tx; 1187 /** Max number of Tx queues to be connected to one Rx queue. */ 1188 uint16_t max_tx_2_rx; 1189 uint16_t max_nb_desc; /**< The max num of descriptors. */ 1190 struct rte_eth_hairpin_queue_cap rx_cap; /**< Rx hairpin queue capabilities. */ 1191 struct rte_eth_hairpin_queue_cap tx_cap; /**< Tx hairpin queue capabilities. */ 1192 }; 1193 1194 #define RTE_ETH_MAX_HAIRPIN_PEERS 32 1195 1196 /** 1197 * @warning 1198 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1199 * 1200 * A structure used to hold hairpin peer data. 1201 */ 1202 struct rte_eth_hairpin_peer { 1203 uint16_t port; /**< Peer port. */ 1204 uint16_t queue; /**< Peer queue. */ 1205 }; 1206 1207 /** 1208 * @warning 1209 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1210 * 1211 * A structure used to configure hairpin binding. 1212 */ 1213 struct rte_eth_hairpin_conf { 1214 uint32_t peer_count:16; /**< The number of peers. */ 1215 1216 /** 1217 * Explicit Tx flow rule mode. 1218 * One hairpin pair of queues should have the same attribute. 1219 * 1220 * - When set, the user should be responsible for inserting the hairpin 1221 * Tx part flows and removing them. 1222 * - When clear, the PMD will try to handle the Tx part of the flows, 1223 * e.g., by splitting one flow into two parts. 1224 */ 1225 uint32_t tx_explicit:1; 1226 1227 /** 1228 * Manually bind hairpin queues. 1229 * One hairpin pair of queues should have the same attribute. 1230 * 1231 * - When set, to enable hairpin, the user should call the hairpin bind 1232 * function after all the queues are set up properly and the ports are 1233 * started. Also, the hairpin unbind function should be called 1234 * accordingly before stopping a port that with hairpin configured. 1235 * - When cleared, the PMD will try to enable the hairpin with the queues 1236 * configured automatically during port start. 1237 */ 1238 uint32_t manual_bind:1; 1239 1240 /** 1241 * Use locked device memory as a backing storage. 1242 * 1243 * - When set, PMD will attempt place descriptors and/or data buffers 1244 * in dedicated device memory. 1245 * - When cleared, PMD will use default memory type as a backing storage. 1246 * Please refer to PMD documentation for details. 1247 * 1248 * API user should check if PMD supports this configuration flag using 1249 * @see rte_eth_dev_hairpin_capability_get. 1250 */ 1251 uint32_t use_locked_device_memory:1; 1252 1253 /** 1254 * Use DPDK memory as backing storage. 1255 * 1256 * - When set, PMD will attempt place descriptors and/or data buffers 1257 * in host memory managed by DPDK. 1258 * - When cleared, PMD will use default memory type as a backing storage. 1259 * Please refer to PMD documentation for details. 1260 * 1261 * API user should check if PMD supports this configuration flag using 1262 * @see rte_eth_dev_hairpin_capability_get. 1263 */ 1264 uint32_t use_rte_memory:1; 1265 1266 /** 1267 * Force usage of hairpin memory configuration. 1268 * 1269 * - When set, PMD will attempt to use specified memory settings. 1270 * If resource allocation fails, then hairpin queue allocation 1271 * will result in an error. 1272 * - When clear, PMD will attempt to use specified memory settings. 1273 * If resource allocation fails, then PMD will retry 1274 * allocation with default configuration. 1275 */ 1276 uint32_t force_memory:1; 1277 1278 uint32_t reserved:11; /**< Reserved bits. */ 1279 1280 struct rte_eth_hairpin_peer peers[RTE_ETH_MAX_HAIRPIN_PEERS]; 1281 }; 1282 1283 /** 1284 * A structure contains information about HW descriptor ring limitations. 1285 */ 1286 struct rte_eth_desc_lim { 1287 uint16_t nb_max; /**< Max allowed number of descriptors. */ 1288 uint16_t nb_min; /**< Min allowed number of descriptors. */ 1289 uint16_t nb_align; /**< Number of descriptors should be aligned to. */ 1290 1291 /** 1292 * Max allowed number of segments per whole packet. 1293 * 1294 * - For TSO packet this is the total number of data descriptors allowed 1295 * by device. 1296 * 1297 * @see nb_mtu_seg_max 1298 */ 1299 uint16_t nb_seg_max; 1300 1301 /** 1302 * Max number of segments per one MTU. 1303 * 1304 * - For non-TSO packet, this is the maximum allowed number of segments 1305 * in a single transmit packet. 1306 * 1307 * - For TSO packet each segment within the TSO may span up to this 1308 * value. 1309 * 1310 * @see nb_seg_max 1311 */ 1312 uint16_t nb_mtu_seg_max; 1313 }; 1314 1315 /** 1316 * This enum indicates the flow control mode 1317 */ 1318 enum rte_eth_fc_mode { 1319 RTE_ETH_FC_NONE = 0, /**< Disable flow control. */ 1320 RTE_ETH_FC_RX_PAUSE, /**< Rx pause frame, enable flowctrl on Tx side. */ 1321 RTE_ETH_FC_TX_PAUSE, /**< Tx pause frame, enable flowctrl on Rx side. */ 1322 RTE_ETH_FC_FULL /**< Enable flow control on both side. */ 1323 }; 1324 1325 /** 1326 * A structure used to configure Ethernet flow control parameter. 1327 * These parameters will be configured into the register of the NIC. 1328 * Please refer to the corresponding data sheet for proper value. 1329 */ 1330 struct rte_eth_fc_conf { 1331 uint32_t high_water; /**< High threshold value to trigger XOFF */ 1332 uint32_t low_water; /**< Low threshold value to trigger XON */ 1333 uint16_t pause_time; /**< Pause quota in the Pause frame */ 1334 uint16_t send_xon; /**< Is XON frame need be sent */ 1335 enum rte_eth_fc_mode mode; /**< Link flow control mode */ 1336 uint8_t mac_ctrl_frame_fwd; /**< Forward MAC control frames */ 1337 uint8_t autoneg; /**< Use Pause autoneg */ 1338 }; 1339 1340 /** 1341 * A structure used to configure Ethernet priority flow control parameter. 1342 * These parameters will be configured into the register of the NIC. 1343 * Please refer to the corresponding data sheet for proper value. 1344 */ 1345 struct rte_eth_pfc_conf { 1346 struct rte_eth_fc_conf fc; /**< General flow control parameter. */ 1347 uint8_t priority; /**< VLAN User Priority. */ 1348 }; 1349 1350 /** 1351 * @warning 1352 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1353 * 1354 * A structure used to retrieve information of queue based PFC. 1355 */ 1356 struct rte_eth_pfc_queue_info { 1357 /** 1358 * Maximum supported traffic class as per PFC (802.1Qbb) specification. 1359 */ 1360 uint8_t tc_max; 1361 /** PFC queue mode capabilities. */ 1362 enum rte_eth_fc_mode mode_capa; 1363 }; 1364 1365 /** 1366 * @warning 1367 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 1368 * 1369 * A structure used to configure Ethernet priority flow control parameters for 1370 * ethdev queues. 1371 * 1372 * rte_eth_pfc_queue_conf::rx_pause structure shall be used to configure given 1373 * tx_qid with corresponding tc. When ethdev device receives PFC frame with 1374 * rte_eth_pfc_queue_conf::rx_pause::tc, traffic will be paused on 1375 * rte_eth_pfc_queue_conf::rx_pause::tx_qid for that tc. 1376 * 1377 * rte_eth_pfc_queue_conf::tx_pause structure shall be used to configure given 1378 * rx_qid. When rx_qid is congested, PFC frames are generated with 1379 * rte_eth_pfc_queue_conf::rx_pause::tc and 1380 * rte_eth_pfc_queue_conf::rx_pause::pause_time to the peer. 1381 */ 1382 struct rte_eth_pfc_queue_conf { 1383 enum rte_eth_fc_mode mode; /**< Link flow control mode */ 1384 1385 struct { 1386 uint16_t tx_qid; /**< Tx queue ID */ 1387 /** Traffic class as per PFC (802.1Qbb) spec. The value must be 1388 * in the range [0, rte_eth_pfc_queue_info::tx_max - 1] 1389 */ 1390 uint8_t tc; 1391 } rx_pause; /* Valid when (mode == FC_RX_PAUSE || mode == FC_FULL) */ 1392 1393 struct { 1394 uint16_t pause_time; /**< Pause quota in the Pause frame */ 1395 uint16_t rx_qid; /**< Rx queue ID */ 1396 /** Traffic class as per PFC (802.1Qbb) spec. The value must be 1397 * in the range [0, rte_eth_pfc_queue_info::tx_max - 1] 1398 */ 1399 uint8_t tc; 1400 } tx_pause; /* Valid when (mode == FC_TX_PAUSE || mode == FC_FULL) */ 1401 }; 1402 1403 /** 1404 * Tunnel type for device-specific classifier configuration. 1405 * @see rte_eth_udp_tunnel 1406 */ 1407 enum rte_eth_tunnel_type { 1408 RTE_ETH_TUNNEL_TYPE_NONE = 0, 1409 RTE_ETH_TUNNEL_TYPE_VXLAN, 1410 RTE_ETH_TUNNEL_TYPE_GENEVE, 1411 RTE_ETH_TUNNEL_TYPE_TEREDO, 1412 RTE_ETH_TUNNEL_TYPE_NVGRE, 1413 RTE_ETH_TUNNEL_TYPE_IP_IN_GRE, 1414 RTE_ETH_L2_TUNNEL_TYPE_E_TAG, 1415 RTE_ETH_TUNNEL_TYPE_VXLAN_GPE, 1416 RTE_ETH_TUNNEL_TYPE_ECPRI, 1417 RTE_ETH_TUNNEL_TYPE_MAX, 1418 }; 1419 1420 /* Deprecated API file for rte_eth_dev_filter_* functions */ 1421 #include "rte_eth_ctrl.h" 1422 1423 /** 1424 * UDP tunneling configuration. 1425 * 1426 * Used to configure the classifier of a device, 1427 * associating an UDP port with a type of tunnel. 1428 * 1429 * Some NICs may need such configuration to properly parse a tunnel 1430 * with any standard or custom UDP port. 1431 */ 1432 struct rte_eth_udp_tunnel { 1433 uint16_t udp_port; /**< UDP port used for the tunnel. */ 1434 uint8_t prot_type; /**< Tunnel type. @see rte_eth_tunnel_type */ 1435 }; 1436 1437 /** 1438 * A structure used to enable/disable specific device interrupts. 1439 */ 1440 struct rte_eth_intr_conf { 1441 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */ 1442 uint32_t lsc:1; 1443 /** enable/disable rxq interrupt. 0 (default) - disable, 1 enable */ 1444 uint32_t rxq:1; 1445 /** enable/disable rmv interrupt. 0 (default) - disable, 1 enable */ 1446 uint32_t rmv:1; 1447 }; 1448 1449 #define rte_intr_conf rte_eth_intr_conf 1450 1451 /** 1452 * A structure used to configure an Ethernet port. 1453 * Depending upon the Rx multi-queue mode, extra advanced 1454 * configuration settings may be needed. 1455 */ 1456 struct rte_eth_conf { 1457 uint32_t link_speeds; /**< bitmap of RTE_ETH_LINK_SPEED_XXX of speeds to be 1458 used. RTE_ETH_LINK_SPEED_FIXED disables link 1459 autonegotiation, and a unique speed shall be 1460 set. Otherwise, the bitmap defines the set of 1461 speeds to be advertised. If the special value 1462 RTE_ETH_LINK_SPEED_AUTONEG (0) is used, all speeds 1463 supported are advertised. */ 1464 struct rte_eth_rxmode rxmode; /**< Port Rx configuration. */ 1465 struct rte_eth_txmode txmode; /**< Port Tx configuration. */ 1466 uint32_t lpbk_mode; /**< Loopback operation mode. By default the value 1467 is 0, meaning the loopback mode is disabled. 1468 Read the datasheet of given Ethernet controller 1469 for details. The possible values of this field 1470 are defined in implementation of each driver. */ 1471 struct { 1472 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */ 1473 /** Port VMDq+DCB configuration. */ 1474 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf; 1475 /** Port DCB Rx configuration. */ 1476 struct rte_eth_dcb_rx_conf dcb_rx_conf; 1477 /** Port VMDq Rx configuration. */ 1478 struct rte_eth_vmdq_rx_conf vmdq_rx_conf; 1479 } rx_adv_conf; /**< Port Rx filtering configuration. */ 1480 union { 1481 /** Port VMDq+DCB Tx configuration. */ 1482 struct rte_eth_vmdq_dcb_tx_conf vmdq_dcb_tx_conf; 1483 /** Port DCB Tx configuration. */ 1484 struct rte_eth_dcb_tx_conf dcb_tx_conf; 1485 /** Port VMDq Tx configuration. */ 1486 struct rte_eth_vmdq_tx_conf vmdq_tx_conf; 1487 } tx_adv_conf; /**< Port Tx DCB configuration (union). */ 1488 /** Currently,Priority Flow Control(PFC) are supported,if DCB with PFC 1489 is needed,and the variable must be set RTE_ETH_DCB_PFC_SUPPORT. */ 1490 uint32_t dcb_capability_en; 1491 struct rte_eth_intr_conf intr_conf; /**< Interrupt mode configuration. */ 1492 }; 1493 1494 /** 1495 * Rx offload capabilities of a device. 1496 */ 1497 #define RTE_ETH_RX_OFFLOAD_VLAN_STRIP RTE_BIT64(0) 1498 #define RTE_ETH_RX_OFFLOAD_IPV4_CKSUM RTE_BIT64(1) 1499 #define RTE_ETH_RX_OFFLOAD_UDP_CKSUM RTE_BIT64(2) 1500 #define RTE_ETH_RX_OFFLOAD_TCP_CKSUM RTE_BIT64(3) 1501 #define RTE_ETH_RX_OFFLOAD_TCP_LRO RTE_BIT64(4) 1502 #define RTE_ETH_RX_OFFLOAD_QINQ_STRIP RTE_BIT64(5) 1503 #define RTE_ETH_RX_OFFLOAD_OUTER_IPV4_CKSUM RTE_BIT64(6) 1504 #define RTE_ETH_RX_OFFLOAD_MACSEC_STRIP RTE_BIT64(7) 1505 #define RTE_ETH_RX_OFFLOAD_VLAN_FILTER RTE_BIT64(9) 1506 #define RTE_ETH_RX_OFFLOAD_VLAN_EXTEND RTE_BIT64(10) 1507 #define RTE_ETH_RX_OFFLOAD_SCATTER RTE_BIT64(13) 1508 /** 1509 * Timestamp is set by the driver in RTE_MBUF_DYNFIELD_TIMESTAMP_NAME 1510 * and RTE_MBUF_DYNFLAG_RX_TIMESTAMP_NAME is set in ol_flags. 1511 * The mbuf field and flag are registered when the offload is configured. 1512 */ 1513 #define RTE_ETH_RX_OFFLOAD_TIMESTAMP RTE_BIT64(14) 1514 #define RTE_ETH_RX_OFFLOAD_SECURITY RTE_BIT64(15) 1515 #define RTE_ETH_RX_OFFLOAD_KEEP_CRC RTE_BIT64(16) 1516 #define RTE_ETH_RX_OFFLOAD_SCTP_CKSUM RTE_BIT64(17) 1517 #define RTE_ETH_RX_OFFLOAD_OUTER_UDP_CKSUM RTE_BIT64(18) 1518 #define RTE_ETH_RX_OFFLOAD_RSS_HASH RTE_BIT64(19) 1519 #define RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT RTE_BIT64(20) 1520 1521 #define RTE_ETH_RX_OFFLOAD_CHECKSUM (RTE_ETH_RX_OFFLOAD_IPV4_CKSUM | \ 1522 RTE_ETH_RX_OFFLOAD_UDP_CKSUM | \ 1523 RTE_ETH_RX_OFFLOAD_TCP_CKSUM) 1524 #define RTE_ETH_RX_OFFLOAD_VLAN (RTE_ETH_RX_OFFLOAD_VLAN_STRIP | \ 1525 RTE_ETH_RX_OFFLOAD_VLAN_FILTER | \ 1526 RTE_ETH_RX_OFFLOAD_VLAN_EXTEND | \ 1527 RTE_ETH_RX_OFFLOAD_QINQ_STRIP) 1528 1529 /* 1530 * If new Rx offload capabilities are defined, they also must be 1531 * mentioned in rte_rx_offload_names in rte_ethdev.c file. 1532 */ 1533 1534 /** 1535 * Tx offload capabilities of a device. 1536 */ 1537 #define RTE_ETH_TX_OFFLOAD_VLAN_INSERT RTE_BIT64(0) 1538 #define RTE_ETH_TX_OFFLOAD_IPV4_CKSUM RTE_BIT64(1) 1539 #define RTE_ETH_TX_OFFLOAD_UDP_CKSUM RTE_BIT64(2) 1540 #define RTE_ETH_TX_OFFLOAD_TCP_CKSUM RTE_BIT64(3) 1541 #define RTE_ETH_TX_OFFLOAD_SCTP_CKSUM RTE_BIT64(4) 1542 #define RTE_ETH_TX_OFFLOAD_TCP_TSO RTE_BIT64(5) 1543 #define RTE_ETH_TX_OFFLOAD_UDP_TSO RTE_BIT64(6) 1544 #define RTE_ETH_TX_OFFLOAD_OUTER_IPV4_CKSUM RTE_BIT64(7) /**< Used for tunneling packet. */ 1545 #define RTE_ETH_TX_OFFLOAD_QINQ_INSERT RTE_BIT64(8) 1546 #define RTE_ETH_TX_OFFLOAD_VXLAN_TNL_TSO RTE_BIT64(9) /**< Used for tunneling packet. */ 1547 #define RTE_ETH_TX_OFFLOAD_GRE_TNL_TSO RTE_BIT64(10) /**< Used for tunneling packet. */ 1548 #define RTE_ETH_TX_OFFLOAD_IPIP_TNL_TSO RTE_BIT64(11) /**< Used for tunneling packet. */ 1549 #define RTE_ETH_TX_OFFLOAD_GENEVE_TNL_TSO RTE_BIT64(12) /**< Used for tunneling packet. */ 1550 #define RTE_ETH_TX_OFFLOAD_MACSEC_INSERT RTE_BIT64(13) 1551 /** 1552 * Multiple threads can invoke rte_eth_tx_burst() concurrently on the same 1553 * Tx queue without SW lock. 1554 */ 1555 #define RTE_ETH_TX_OFFLOAD_MT_LOCKFREE RTE_BIT64(14) 1556 /** Device supports multi segment send. */ 1557 #define RTE_ETH_TX_OFFLOAD_MULTI_SEGS RTE_BIT64(15) 1558 /** 1559 * Device supports optimization for fast release of mbufs. 1560 * When set application must guarantee that per-queue all mbufs comes from 1561 * the same mempool and has refcnt = 1. 1562 */ 1563 #define RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE RTE_BIT64(16) 1564 #define RTE_ETH_TX_OFFLOAD_SECURITY RTE_BIT64(17) 1565 /** 1566 * Device supports generic UDP tunneled packet TSO. 1567 * Application must set RTE_MBUF_F_TX_TUNNEL_UDP and other mbuf fields required 1568 * for tunnel TSO. 1569 */ 1570 #define RTE_ETH_TX_OFFLOAD_UDP_TNL_TSO RTE_BIT64(18) 1571 /** 1572 * Device supports generic IP tunneled packet TSO. 1573 * Application must set RTE_MBUF_F_TX_TUNNEL_IP and other mbuf fields required 1574 * for tunnel TSO. 1575 */ 1576 #define RTE_ETH_TX_OFFLOAD_IP_TNL_TSO RTE_BIT64(19) 1577 /** Device supports outer UDP checksum */ 1578 #define RTE_ETH_TX_OFFLOAD_OUTER_UDP_CKSUM RTE_BIT64(20) 1579 /** 1580 * Device sends on time read from RTE_MBUF_DYNFIELD_TIMESTAMP_NAME 1581 * if RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME is set in ol_flags. 1582 * The mbuf field and flag are registered when the offload is configured. 1583 */ 1584 #define RTE_ETH_TX_OFFLOAD_SEND_ON_TIMESTAMP RTE_BIT64(21) 1585 /* 1586 * If new Tx offload capabilities are defined, they also must be 1587 * mentioned in rte_tx_offload_names in rte_ethdev.c file. 1588 */ 1589 1590 /**@{@name Device capabilities 1591 * Non-offload capabilities reported in rte_eth_dev_info.dev_capa. 1592 */ 1593 /** Device supports Rx queue setup after device started. */ 1594 #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP RTE_BIT64(0) 1595 /** Device supports Tx queue setup after device started. */ 1596 #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP RTE_BIT64(1) 1597 /** 1598 * Device supports shared Rx queue among ports within Rx domain and 1599 * switch domain. Mbufs are consumed by shared Rx queue instead of 1600 * each queue. Multiple groups are supported by share_group of Rx 1601 * queue configuration. Shared Rx queue is identified by PMD using 1602 * share_qid of Rx queue configuration. Polling any port in the group 1603 * receive packets of all member ports, source port identified by 1604 * mbuf->port field. 1605 */ 1606 #define RTE_ETH_DEV_CAPA_RXQ_SHARE RTE_BIT64(2) 1607 /** Device supports keeping flow rules across restart. */ 1608 #define RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP RTE_BIT64(3) 1609 /** Device supports keeping shared flow objects across restart. */ 1610 #define RTE_ETH_DEV_CAPA_FLOW_SHARED_OBJECT_KEEP RTE_BIT64(4) 1611 /**@}*/ 1612 1613 /* 1614 * Fallback default preferred Rx/Tx port parameters. 1615 * These are used if an application requests default parameters 1616 * but the PMD does not provide preferred values. 1617 */ 1618 #define RTE_ETH_DEV_FALLBACK_RX_RINGSIZE 512 1619 #define RTE_ETH_DEV_FALLBACK_TX_RINGSIZE 512 1620 #define RTE_ETH_DEV_FALLBACK_RX_NBQUEUES 1 1621 #define RTE_ETH_DEV_FALLBACK_TX_NBQUEUES 1 1622 1623 /** 1624 * Preferred Rx/Tx port parameters. 1625 * There are separate instances of this structure for transmission 1626 * and reception respectively. 1627 */ 1628 struct rte_eth_dev_portconf { 1629 uint16_t burst_size; /**< Device-preferred burst size */ 1630 uint16_t ring_size; /**< Device-preferred size of queue rings */ 1631 uint16_t nb_queues; /**< Device-preferred number of queues */ 1632 }; 1633 1634 /** 1635 * Default values for switch domain ID when ethdev does not support switch 1636 * domain definitions. 1637 */ 1638 #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID (UINT16_MAX) 1639 1640 /** 1641 * Ethernet device associated switch information 1642 */ 1643 struct rte_eth_switch_info { 1644 const char *name; /**< switch name */ 1645 uint16_t domain_id; /**< switch domain ID */ 1646 /** 1647 * Mapping to the devices physical switch port as enumerated from the 1648 * perspective of the embedded interconnect/switch. For SR-IOV enabled 1649 * device this may correspond to the VF_ID of each virtual function, 1650 * but each driver should explicitly define the mapping of switch 1651 * port identifier to that physical interconnect/switch 1652 */ 1653 uint16_t port_id; 1654 /** 1655 * Shared Rx queue sub-domain boundary. Only ports in same Rx domain 1656 * and switch domain can share Rx queue. Valid only if device advertised 1657 * RTE_ETH_DEV_CAPA_RXQ_SHARE capability. 1658 */ 1659 uint16_t rx_domain; 1660 }; 1661 1662 /** 1663 * @warning 1664 * @b EXPERIMENTAL: this structure may change without prior notice. 1665 * 1666 * Ethernet device Rx buffer segmentation capabilities. 1667 */ 1668 struct rte_eth_rxseg_capa { 1669 __extension__ 1670 uint32_t multi_pools:1; /**< Supports receiving to multiple pools.*/ 1671 uint32_t offset_allowed:1; /**< Supports buffer offsets. */ 1672 uint32_t offset_align_log2:4; /**< Required offset alignment. */ 1673 uint16_t max_nseg; /**< Maximum amount of segments to split. */ 1674 uint16_t reserved; /**< Reserved field. */ 1675 }; 1676 1677 /** 1678 * Ethernet device information 1679 */ 1680 1681 /** 1682 * Ethernet device representor port type. 1683 */ 1684 enum rte_eth_representor_type { 1685 RTE_ETH_REPRESENTOR_NONE, /**< not a representor. */ 1686 RTE_ETH_REPRESENTOR_VF, /**< representor of Virtual Function. */ 1687 RTE_ETH_REPRESENTOR_SF, /**< representor of Sub Function. */ 1688 RTE_ETH_REPRESENTOR_PF, /**< representor of Physical Function. */ 1689 }; 1690 1691 /** 1692 * @warning 1693 * @b EXPERIMENTAL: this enumeration may change without prior notice. 1694 * 1695 * Ethernet device error handling mode. 1696 */ 1697 enum rte_eth_err_handle_mode { 1698 /** No error handling modes are supported. */ 1699 RTE_ETH_ERROR_HANDLE_MODE_NONE, 1700 /** Passive error handling, after the PMD detects that a reset is required, 1701 * the PMD reports @see RTE_ETH_EVENT_INTR_RESET event, 1702 * and the application invokes @see rte_eth_dev_reset to recover the port. 1703 */ 1704 RTE_ETH_ERROR_HANDLE_MODE_PASSIVE, 1705 /** Proactive error handling, after the PMD detects that a reset is required, 1706 * the PMD reports @see RTE_ETH_EVENT_ERR_RECOVERING event, 1707 * do recovery internally, and finally reports the recovery result event 1708 * (@see RTE_ETH_EVENT_RECOVERY_*). 1709 */ 1710 RTE_ETH_ERROR_HANDLE_MODE_PROACTIVE, 1711 }; 1712 1713 /** 1714 * A structure used to retrieve the contextual information of 1715 * an Ethernet device, such as the controlling driver of the 1716 * device, etc... 1717 */ 1718 struct rte_eth_dev_info { 1719 struct rte_device *device; /**< Generic device information */ 1720 const char *driver_name; /**< Device Driver name. */ 1721 unsigned int if_index; /**< Index to bound host interface, or 0 if none. 1722 Use if_indextoname() to translate into an interface name. */ 1723 uint16_t min_mtu; /**< Minimum MTU allowed */ 1724 uint16_t max_mtu; /**< Maximum MTU allowed */ 1725 const uint32_t *dev_flags; /**< Device flags */ 1726 uint32_t min_rx_bufsize; /**< Minimum size of Rx buffer. */ 1727 uint32_t max_rx_pktlen; /**< Maximum configurable length of Rx pkt. */ 1728 /** Maximum configurable size of LRO aggregated packet. */ 1729 uint32_t max_lro_pkt_size; 1730 uint16_t max_rx_queues; /**< Maximum number of Rx queues. */ 1731 uint16_t max_tx_queues; /**< Maximum number of Tx queues. */ 1732 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */ 1733 /** Maximum number of hash MAC addresses for MTA and UTA. */ 1734 uint32_t max_hash_mac_addrs; 1735 uint16_t max_vfs; /**< Maximum number of VFs. */ 1736 uint16_t max_vmdq_pools; /**< Maximum number of VMDq pools. */ 1737 struct rte_eth_rxseg_capa rx_seg_capa; /**< Segmentation capability.*/ 1738 /** All Rx offload capabilities including all per-queue ones */ 1739 uint64_t rx_offload_capa; 1740 /** All Tx offload capabilities including all per-queue ones */ 1741 uint64_t tx_offload_capa; 1742 /** Device per-queue Rx offload capabilities. */ 1743 uint64_t rx_queue_offload_capa; 1744 /** Device per-queue Tx offload capabilities. */ 1745 uint64_t tx_queue_offload_capa; 1746 /** Device redirection table size, the total number of entries. */ 1747 uint16_t reta_size; 1748 uint8_t hash_key_size; /**< Hash key size in bytes */ 1749 /** Bit mask of RSS offloads, the bit offset also means flow type */ 1750 uint64_t flow_type_rss_offloads; 1751 struct rte_eth_rxconf default_rxconf; /**< Default Rx configuration */ 1752 struct rte_eth_txconf default_txconf; /**< Default Tx configuration */ 1753 uint16_t vmdq_queue_base; /**< First queue ID for VMDq pools. */ 1754 uint16_t vmdq_queue_num; /**< Queue number for VMDq pools. */ 1755 uint16_t vmdq_pool_base; /**< First ID of VMDq pools. */ 1756 struct rte_eth_desc_lim rx_desc_lim; /**< Rx descriptors limits */ 1757 struct rte_eth_desc_lim tx_desc_lim; /**< Tx descriptors limits */ 1758 uint32_t speed_capa; /**< Supported speeds bitmap (RTE_ETH_LINK_SPEED_). */ 1759 /** Configured number of Rx/Tx queues */ 1760 uint16_t nb_rx_queues; /**< Number of Rx queues. */ 1761 uint16_t nb_tx_queues; /**< Number of Tx queues. */ 1762 /** 1763 * Maximum number of Rx mempools supported per Rx queue. 1764 * 1765 * Value greater than 0 means that the driver supports Rx queue 1766 * mempools specification via rx_conf->rx_mempools. 1767 */ 1768 uint16_t max_rx_mempools; 1769 /** Rx parameter recommendations */ 1770 struct rte_eth_dev_portconf default_rxportconf; 1771 /** Tx parameter recommendations */ 1772 struct rte_eth_dev_portconf default_txportconf; 1773 /** Generic device capabilities (RTE_ETH_DEV_CAPA_). */ 1774 uint64_t dev_capa; 1775 /** 1776 * Switching information for ports on a device with a 1777 * embedded managed interconnect/switch. 1778 */ 1779 struct rte_eth_switch_info switch_info; 1780 /** Supported error handling mode. */ 1781 enum rte_eth_err_handle_mode err_handle_mode; 1782 1783 uint64_t reserved_64s[2]; /**< Reserved for future fields */ 1784 void *reserved_ptrs[2]; /**< Reserved for future fields */ 1785 }; 1786 1787 /**@{@name Rx/Tx queue states */ 1788 #define RTE_ETH_QUEUE_STATE_STOPPED 0 /**< Queue stopped. */ 1789 #define RTE_ETH_QUEUE_STATE_STARTED 1 /**< Queue started. */ 1790 #define RTE_ETH_QUEUE_STATE_HAIRPIN 2 /**< Queue used for hairpin. */ 1791 /**@}*/ 1792 1793 /** 1794 * Ethernet device Rx queue information structure. 1795 * Used to retrieve information about configured queue. 1796 */ 1797 struct rte_eth_rxq_info { 1798 struct rte_mempool *mp; /**< mempool used by that queue. */ 1799 struct rte_eth_rxconf conf; /**< queue config parameters. */ 1800 uint8_t scattered_rx; /**< scattered packets Rx supported. */ 1801 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */ 1802 uint16_t nb_desc; /**< configured number of RXDs. */ 1803 uint16_t rx_buf_size; /**< hardware receive buffer size. */ 1804 /** 1805 * Available Rx descriptors threshold defined as percentage 1806 * of Rx queue size. If number of available descriptors is lower, 1807 * the event RTE_ETH_EVENT_RX_AVAIL_THESH is generated. 1808 * Value 0 means that the threshold monitoring is disabled. 1809 */ 1810 uint8_t avail_thresh; 1811 } __rte_cache_min_aligned; 1812 1813 /** 1814 * Ethernet device Tx queue information structure. 1815 * Used to retrieve information about configured queue. 1816 */ 1817 struct rte_eth_txq_info { 1818 struct rte_eth_txconf conf; /**< queue config parameters. */ 1819 uint16_t nb_desc; /**< configured number of TXDs. */ 1820 uint8_t queue_state; /**< one of RTE_ETH_QUEUE_STATE_*. */ 1821 } __rte_cache_min_aligned; 1822 1823 /* Generic Burst mode flag definition, values can be ORed. */ 1824 1825 /** 1826 * If the queues have different burst mode description, this bit will be set 1827 * by PMD, then the application can iterate to retrieve burst description for 1828 * all other queues. 1829 */ 1830 #define RTE_ETH_BURST_FLAG_PER_QUEUE RTE_BIT64(0) 1831 1832 /** 1833 * Ethernet device Rx/Tx queue packet burst mode information structure. 1834 * Used to retrieve information about packet burst mode setting. 1835 */ 1836 struct rte_eth_burst_mode { 1837 uint64_t flags; /**< The ORed values of RTE_ETH_BURST_FLAG_xxx */ 1838 1839 #define RTE_ETH_BURST_MODE_INFO_SIZE 1024 /**< Maximum size for information */ 1840 char info[RTE_ETH_BURST_MODE_INFO_SIZE]; /**< burst mode information */ 1841 }; 1842 1843 /** Maximum name length for extended statistics counters */ 1844 #define RTE_ETH_XSTATS_NAME_SIZE 64 1845 1846 /** 1847 * An Ethernet device extended statistic structure 1848 * 1849 * This structure is used by rte_eth_xstats_get() to provide 1850 * statistics that are not provided in the generic *rte_eth_stats* 1851 * structure. 1852 * It maps a name ID, corresponding to an index in the array returned 1853 * by rte_eth_xstats_get_names(), to a statistic value. 1854 */ 1855 struct rte_eth_xstat { 1856 uint64_t id; /**< The index in xstats name array. */ 1857 uint64_t value; /**< The statistic counter value. */ 1858 }; 1859 1860 /** 1861 * A name element for extended statistics. 1862 * 1863 * An array of this structure is returned by rte_eth_xstats_get_names(). 1864 * It lists the names of extended statistics for a PMD. The *rte_eth_xstat* 1865 * structure references these names by their array index. 1866 * 1867 * The xstats should follow a common naming scheme. 1868 * Some names are standardized in rte_stats_strings. 1869 * Examples: 1870 * - rx_missed_errors 1871 * - tx_q3_bytes 1872 * - tx_size_128_to_255_packets 1873 */ 1874 struct rte_eth_xstat_name { 1875 char name[RTE_ETH_XSTATS_NAME_SIZE]; /**< The statistic name. */ 1876 }; 1877 1878 #define RTE_ETH_DCB_NUM_TCS 8 1879 #define RTE_ETH_MAX_VMDQ_POOL 64 1880 1881 /** 1882 * A structure used to get the information of queue and 1883 * TC mapping on both Tx and Rx paths. 1884 */ 1885 struct rte_eth_dcb_tc_queue_mapping { 1886 /** Rx queues assigned to tc per Pool */ 1887 struct { 1888 uint16_t base; 1889 uint16_t nb_queue; 1890 } tc_rxq[RTE_ETH_MAX_VMDQ_POOL][RTE_ETH_DCB_NUM_TCS]; 1891 /** Rx queues assigned to tc per Pool */ 1892 struct { 1893 uint16_t base; 1894 uint16_t nb_queue; 1895 } tc_txq[RTE_ETH_MAX_VMDQ_POOL][RTE_ETH_DCB_NUM_TCS]; 1896 }; 1897 1898 /** 1899 * A structure used to get the information of DCB. 1900 * It includes TC UP mapping and queue TC mapping. 1901 */ 1902 struct rte_eth_dcb_info { 1903 uint8_t nb_tcs; /**< number of TCs */ 1904 uint8_t prio_tc[RTE_ETH_DCB_NUM_USER_PRIORITIES]; /**< Priority to tc */ 1905 uint8_t tc_bws[RTE_ETH_DCB_NUM_TCS]; /**< Tx BW percentage for each TC */ 1906 /** Rx queues assigned to tc */ 1907 struct rte_eth_dcb_tc_queue_mapping tc_queue; 1908 }; 1909 1910 /** 1911 * This enum indicates the possible Forward Error Correction (FEC) modes 1912 * of an ethdev port. 1913 */ 1914 enum rte_eth_fec_mode { 1915 RTE_ETH_FEC_NOFEC = 0, /**< FEC is off */ 1916 RTE_ETH_FEC_AUTO, /**< FEC autonegotiation modes */ 1917 RTE_ETH_FEC_BASER, /**< FEC using common algorithm */ 1918 RTE_ETH_FEC_RS, /**< FEC using RS algorithm */ 1919 }; 1920 1921 /* Translate from FEC mode to FEC capa */ 1922 #define RTE_ETH_FEC_MODE_TO_CAPA(x) RTE_BIT32(x) 1923 1924 /* This macro indicates FEC capa mask */ 1925 #define RTE_ETH_FEC_MODE_CAPA_MASK(x) RTE_BIT32(RTE_ETH_FEC_ ## x) 1926 1927 /* A structure used to get capabilities per link speed */ 1928 struct rte_eth_fec_capa { 1929 uint32_t speed; /**< Link speed (see RTE_ETH_SPEED_NUM_*) */ 1930 uint32_t capa; /**< FEC capabilities bitmask */ 1931 }; 1932 1933 #define RTE_ETH_ALL RTE_MAX_ETHPORTS 1934 1935 /* Macros to check for valid port */ 1936 #define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \ 1937 if (!rte_eth_dev_is_valid_port(port_id)) { \ 1938 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \ 1939 return retval; \ 1940 } \ 1941 } while (0) 1942 1943 #define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \ 1944 if (!rte_eth_dev_is_valid_port(port_id)) { \ 1945 RTE_ETHDEV_LOG(ERR, "Invalid port_id=%u\n", port_id); \ 1946 return; \ 1947 } \ 1948 } while (0) 1949 1950 /** 1951 * Function type used for Rx packet processing packet callbacks. 1952 * 1953 * The callback function is called on Rx with a burst of packets that have 1954 * been received on the given port and queue. 1955 * 1956 * @param port_id 1957 * The Ethernet port on which Rx is being performed. 1958 * @param queue 1959 * The queue on the Ethernet port which is being used to receive the packets. 1960 * @param pkts 1961 * The burst of packets that have just been received. 1962 * @param nb_pkts 1963 * The number of packets in the burst pointed to by "pkts". 1964 * @param max_pkts 1965 * The max number of packets that can be stored in the "pkts" array. 1966 * @param user_param 1967 * The arbitrary user parameter passed in by the application when the callback 1968 * was originally configured. 1969 * @return 1970 * The number of packets returned to the user. 1971 */ 1972 typedef uint16_t (*rte_rx_callback_fn)(uint16_t port_id, uint16_t queue, 1973 struct rte_mbuf *pkts[], uint16_t nb_pkts, uint16_t max_pkts, 1974 void *user_param); 1975 1976 /** 1977 * Function type used for Tx packet processing packet callbacks. 1978 * 1979 * The callback function is called on Tx with a burst of packets immediately 1980 * before the packets are put onto the hardware queue for transmission. 1981 * 1982 * @param port_id 1983 * The Ethernet port on which Tx is being performed. 1984 * @param queue 1985 * The queue on the Ethernet port which is being used to transmit the packets. 1986 * @param pkts 1987 * The burst of packets that are about to be transmitted. 1988 * @param nb_pkts 1989 * The number of packets in the burst pointed to by "pkts". 1990 * @param user_param 1991 * The arbitrary user parameter passed in by the application when the callback 1992 * was originally configured. 1993 * @return 1994 * The number of packets to be written to the NIC. 1995 */ 1996 typedef uint16_t (*rte_tx_callback_fn)(uint16_t port_id, uint16_t queue, 1997 struct rte_mbuf *pkts[], uint16_t nb_pkts, void *user_param); 1998 1999 /** 2000 * Possible states of an ethdev port. 2001 */ 2002 enum rte_eth_dev_state { 2003 /** Device is unused before being probed. */ 2004 RTE_ETH_DEV_UNUSED = 0, 2005 /** Device is attached when allocated in probing. */ 2006 RTE_ETH_DEV_ATTACHED, 2007 /** Device is in removed state when plug-out is detected. */ 2008 RTE_ETH_DEV_REMOVED, 2009 }; 2010 2011 struct rte_eth_dev_sriov { 2012 uint8_t active; /**< SRIOV is active with 16, 32 or 64 pools */ 2013 uint8_t nb_q_per_pool; /**< Rx queue number per pool */ 2014 uint16_t def_vmdq_idx; /**< Default pool num used for PF */ 2015 uint16_t def_pool_q_idx; /**< Default pool queue start reg index */ 2016 }; 2017 #define RTE_ETH_DEV_SRIOV(dev) ((dev)->data->sriov) 2018 2019 #define RTE_ETH_NAME_MAX_LEN RTE_DEV_NAME_MAX_LEN 2020 2021 #define RTE_ETH_DEV_NO_OWNER 0 2022 2023 #define RTE_ETH_MAX_OWNER_NAME_LEN 64 2024 2025 struct rte_eth_dev_owner { 2026 uint64_t id; /**< The owner unique identifier. */ 2027 char name[RTE_ETH_MAX_OWNER_NAME_LEN]; /**< The owner name. */ 2028 }; 2029 2030 /**@{@name Device flags 2031 * Flags internally saved in rte_eth_dev_data.dev_flags 2032 * and reported in rte_eth_dev_info.dev_flags. 2033 */ 2034 /** PMD supports thread-safe flow operations */ 2035 #define RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE RTE_BIT32(0) 2036 /** Device supports link state interrupt */ 2037 #define RTE_ETH_DEV_INTR_LSC RTE_BIT32(1) 2038 /** Device is a bonded slave */ 2039 #define RTE_ETH_DEV_BONDED_SLAVE RTE_BIT32(2) 2040 /** Device supports device removal interrupt */ 2041 #define RTE_ETH_DEV_INTR_RMV RTE_BIT32(3) 2042 /** Device is port representor */ 2043 #define RTE_ETH_DEV_REPRESENTOR RTE_BIT32(4) 2044 /** Device does not support MAC change after started */ 2045 #define RTE_ETH_DEV_NOLIVE_MAC_ADDR RTE_BIT32(5) 2046 /** 2047 * Queue xstats filled automatically by ethdev layer. 2048 * PMDs filling the queue xstats themselves should not set this flag 2049 */ 2050 #define RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS RTE_BIT32(6) 2051 /**@}*/ 2052 2053 /** 2054 * Iterates over valid ethdev ports owned by a specific owner. 2055 * 2056 * @param port_id 2057 * The ID of the next possible valid owned port. 2058 * @param owner_id 2059 * The owner identifier. 2060 * RTE_ETH_DEV_NO_OWNER means iterate over all valid ownerless ports. 2061 * @return 2062 * Next valid port ID owned by owner_id, RTE_MAX_ETHPORTS if there is none. 2063 */ 2064 uint64_t rte_eth_find_next_owned_by(uint16_t port_id, 2065 const uint64_t owner_id); 2066 2067 /** 2068 * Macro to iterate over all enabled ethdev ports owned by a specific owner. 2069 */ 2070 #define RTE_ETH_FOREACH_DEV_OWNED_BY(p, o) \ 2071 for (p = rte_eth_find_next_owned_by(0, o); \ 2072 (unsigned int)p < (unsigned int)RTE_MAX_ETHPORTS; \ 2073 p = rte_eth_find_next_owned_by(p + 1, o)) 2074 2075 /** 2076 * Iterates over valid ethdev ports. 2077 * 2078 * @param port_id 2079 * The ID of the next possible valid port. 2080 * @return 2081 * Next valid port ID, RTE_MAX_ETHPORTS if there is none. 2082 */ 2083 uint16_t rte_eth_find_next(uint16_t port_id); 2084 2085 /** 2086 * Macro to iterate over all enabled and ownerless ethdev ports. 2087 */ 2088 #define RTE_ETH_FOREACH_DEV(p) \ 2089 RTE_ETH_FOREACH_DEV_OWNED_BY(p, RTE_ETH_DEV_NO_OWNER) 2090 2091 /** 2092 * Iterates over ethdev ports of a specified device. 2093 * 2094 * @param port_id_start 2095 * The ID of the next possible valid port. 2096 * @param parent 2097 * The generic device behind the ports to iterate. 2098 * @return 2099 * Next port ID of the device, possibly port_id_start, 2100 * RTE_MAX_ETHPORTS if there is none. 2101 */ 2102 uint16_t 2103 rte_eth_find_next_of(uint16_t port_id_start, 2104 const struct rte_device *parent); 2105 2106 /** 2107 * Macro to iterate over all ethdev ports of a specified device. 2108 * 2109 * @param port_id 2110 * The ID of the matching port being iterated. 2111 * @param parent 2112 * The rte_device pointer matching the iterated ports. 2113 */ 2114 #define RTE_ETH_FOREACH_DEV_OF(port_id, parent) \ 2115 for (port_id = rte_eth_find_next_of(0, parent); \ 2116 port_id < RTE_MAX_ETHPORTS; \ 2117 port_id = rte_eth_find_next_of(port_id + 1, parent)) 2118 2119 /** 2120 * Iterates over sibling ethdev ports (i.e. sharing the same rte_device). 2121 * 2122 * @param port_id_start 2123 * The ID of the next possible valid sibling port. 2124 * @param ref_port_id 2125 * The ID of a reference port to compare rte_device with. 2126 * @return 2127 * Next sibling port ID, possibly port_id_start or ref_port_id itself, 2128 * RTE_MAX_ETHPORTS if there is none. 2129 */ 2130 uint16_t 2131 rte_eth_find_next_sibling(uint16_t port_id_start, uint16_t ref_port_id); 2132 2133 /** 2134 * Macro to iterate over all ethdev ports sharing the same rte_device 2135 * as the specified port. 2136 * Note: the specified reference port is part of the loop iterations. 2137 * 2138 * @param port_id 2139 * The ID of the matching port being iterated. 2140 * @param ref_port_id 2141 * The ID of the port being compared. 2142 */ 2143 #define RTE_ETH_FOREACH_DEV_SIBLING(port_id, ref_port_id) \ 2144 for (port_id = rte_eth_find_next_sibling(0, ref_port_id); \ 2145 port_id < RTE_MAX_ETHPORTS; \ 2146 port_id = rte_eth_find_next_sibling(port_id + 1, ref_port_id)) 2147 2148 /** 2149 * Get a new unique owner identifier. 2150 * An owner identifier is used to owns Ethernet devices by only one DPDK entity 2151 * to avoid multiple management of device by different entities. 2152 * 2153 * @param owner_id 2154 * Owner identifier pointer. 2155 * @return 2156 * Negative errno value on error, 0 on success. 2157 */ 2158 int rte_eth_dev_owner_new(uint64_t *owner_id); 2159 2160 /** 2161 * Set an Ethernet device owner. 2162 * 2163 * @param port_id 2164 * The identifier of the port to own. 2165 * @param owner 2166 * The owner pointer. 2167 * @return 2168 * Negative errno value on error, 0 on success. 2169 */ 2170 int rte_eth_dev_owner_set(const uint16_t port_id, 2171 const struct rte_eth_dev_owner *owner); 2172 2173 /** 2174 * Unset Ethernet device owner to make the device ownerless. 2175 * 2176 * @param port_id 2177 * The identifier of port to make ownerless. 2178 * @param owner_id 2179 * The owner identifier. 2180 * @return 2181 * 0 on success, negative errno value on error. 2182 */ 2183 int rte_eth_dev_owner_unset(const uint16_t port_id, 2184 const uint64_t owner_id); 2185 2186 /** 2187 * Remove owner from all Ethernet devices owned by a specific owner. 2188 * 2189 * @param owner_id 2190 * The owner identifier. 2191 * @return 2192 * 0 on success, negative errno value on error. 2193 */ 2194 int rte_eth_dev_owner_delete(const uint64_t owner_id); 2195 2196 /** 2197 * Get the owner of an Ethernet device. 2198 * 2199 * @param port_id 2200 * The port identifier. 2201 * @param owner 2202 * The owner structure pointer to fill. 2203 * @return 2204 * 0 on success, negative errno value on error.. 2205 */ 2206 int rte_eth_dev_owner_get(const uint16_t port_id, 2207 struct rte_eth_dev_owner *owner); 2208 2209 /** 2210 * Get the number of ports which are usable for the application. 2211 * 2212 * These devices must be iterated by using the macro 2213 * ``RTE_ETH_FOREACH_DEV`` or ``RTE_ETH_FOREACH_DEV_OWNED_BY`` 2214 * to deal with non-contiguous ranges of devices. 2215 * 2216 * @return 2217 * The count of available Ethernet devices. 2218 */ 2219 uint16_t rte_eth_dev_count_avail(void); 2220 2221 /** 2222 * Get the total number of ports which are allocated. 2223 * 2224 * Some devices may not be available for the application. 2225 * 2226 * @return 2227 * The total count of Ethernet devices. 2228 */ 2229 uint16_t rte_eth_dev_count_total(void); 2230 2231 /** 2232 * Convert a numerical speed in Mbps to a bitmap flag that can be used in 2233 * the bitmap link_speeds of the struct rte_eth_conf 2234 * 2235 * @param speed 2236 * Numerical speed value in Mbps 2237 * @param duplex 2238 * RTE_ETH_LINK_[HALF/FULL]_DUPLEX (only for 10/100M speeds) 2239 * @return 2240 * 0 if the speed cannot be mapped 2241 */ 2242 uint32_t rte_eth_speed_bitflag(uint32_t speed, int duplex); 2243 2244 /** 2245 * Get RTE_ETH_RX_OFFLOAD_* flag name. 2246 * 2247 * @param offload 2248 * Offload flag. 2249 * @return 2250 * Offload name or 'UNKNOWN' if the flag cannot be recognised. 2251 */ 2252 const char *rte_eth_dev_rx_offload_name(uint64_t offload); 2253 2254 /** 2255 * Get RTE_ETH_TX_OFFLOAD_* flag name. 2256 * 2257 * @param offload 2258 * Offload flag. 2259 * @return 2260 * Offload name or 'UNKNOWN' if the flag cannot be recognised. 2261 */ 2262 const char *rte_eth_dev_tx_offload_name(uint64_t offload); 2263 2264 /** 2265 * @warning 2266 * @b EXPERIMENTAL: this API may change without prior notice. 2267 * 2268 * Get RTE_ETH_DEV_CAPA_* flag name. 2269 * 2270 * @param capability 2271 * Capability flag. 2272 * @return 2273 * Capability name or 'UNKNOWN' if the flag cannot be recognized. 2274 */ 2275 __rte_experimental 2276 const char *rte_eth_dev_capability_name(uint64_t capability); 2277 2278 /** 2279 * Configure an Ethernet device. 2280 * This function must be invoked first before any other function in the 2281 * Ethernet API. This function can also be re-invoked when a device is in the 2282 * stopped state. 2283 * 2284 * @param port_id 2285 * The port identifier of the Ethernet device to configure. 2286 * @param nb_rx_queue 2287 * The number of receive queues to set up for the Ethernet device. 2288 * @param nb_tx_queue 2289 * The number of transmit queues to set up for the Ethernet device. 2290 * @param eth_conf 2291 * The pointer to the configuration data to be used for the Ethernet device. 2292 * The *rte_eth_conf* structure includes: 2293 * - the hardware offload features to activate, with dedicated fields for 2294 * each statically configurable offload hardware feature provided by 2295 * Ethernet devices, such as IP checksum or VLAN tag stripping for 2296 * example. 2297 * The Rx offload bitfield API is obsolete and will be deprecated. 2298 * Applications should set the ignore_bitfield_offloads bit on *rxmode* 2299 * structure and use offloads field to set per-port offloads instead. 2300 * - Any offloading set in eth_conf->[rt]xmode.offloads must be within 2301 * the [rt]x_offload_capa returned from rte_eth_dev_info_get(). 2302 * Any type of device supported offloading set in the input argument 2303 * eth_conf->[rt]xmode.offloads to rte_eth_dev_configure() is enabled 2304 * on all queues and it can't be disabled in rte_eth_[rt]x_queue_setup() 2305 * - the Receive Side Scaling (RSS) configuration when using multiple Rx 2306 * queues per port. Any RSS hash function set in eth_conf->rss_conf.rss_hf 2307 * must be within the flow_type_rss_offloads provided by drivers via 2308 * rte_eth_dev_info_get() API. 2309 * 2310 * Embedding all configuration information in a single data structure 2311 * is the more flexible method that allows the addition of new features 2312 * without changing the syntax of the API. 2313 * @return 2314 * - 0: Success, device configured. 2315 * - <0: Error code returned by the driver configuration function. 2316 */ 2317 int rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_queue, 2318 uint16_t nb_tx_queue, const struct rte_eth_conf *eth_conf); 2319 2320 /** 2321 * Check if an Ethernet device was physically removed. 2322 * 2323 * @param port_id 2324 * The port identifier of the Ethernet device. 2325 * @return 2326 * 1 when the Ethernet device is removed, otherwise 0. 2327 */ 2328 int 2329 rte_eth_dev_is_removed(uint16_t port_id); 2330 2331 /** 2332 * Allocate and set up a receive queue for an Ethernet device. 2333 * 2334 * The function allocates a contiguous block of memory for *nb_rx_desc* 2335 * receive descriptors from a memory zone associated with *socket_id* 2336 * and initializes each receive descriptor with a network buffer allocated 2337 * from the memory pool *mb_pool*. 2338 * 2339 * @param port_id 2340 * The port identifier of the Ethernet device. 2341 * @param rx_queue_id 2342 * The index of the receive queue to set up. 2343 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 2344 * to rte_eth_dev_configure(). 2345 * @param nb_rx_desc 2346 * The number of receive descriptors to allocate for the receive ring. 2347 * @param socket_id 2348 * The *socket_id* argument is the socket identifier in case of NUMA. 2349 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for 2350 * the DMA memory allocated for the receive descriptors of the ring. 2351 * @param rx_conf 2352 * The pointer to the configuration data to be used for the receive queue. 2353 * NULL value is allowed, in which case default Rx configuration 2354 * will be used. 2355 * The *rx_conf* structure contains an *rx_thresh* structure with the values 2356 * of the Prefetch, Host, and Write-Back threshold registers of the receive 2357 * ring. 2358 * In addition it contains the hardware offloads features to activate using 2359 * the RTE_ETH_RX_OFFLOAD_* flags. 2360 * If an offloading set in rx_conf->offloads 2361 * hasn't been set in the input argument eth_conf->rxmode.offloads 2362 * to rte_eth_dev_configure(), it is a new added offloading, it must be 2363 * per-queue type and it is enabled for the queue. 2364 * No need to repeat any bit in rx_conf->offloads which has already been 2365 * enabled in rte_eth_dev_configure() at port level. An offloading enabled 2366 * at port level can't be disabled at queue level. 2367 * The configuration structure also contains the pointer to the array 2368 * of the receiving buffer segment descriptions, see rx_seg and rx_nseg 2369 * fields, this extended configuration might be used by split offloads like 2370 * RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT. If mb_pool is not NULL, 2371 * the extended configuration fields must be set to NULL and zero. 2372 * @param mb_pool 2373 * The pointer to the memory pool from which to allocate *rte_mbuf* network 2374 * memory buffers to populate each descriptor of the receive ring. There are 2375 * two options to provide Rx buffer configuration: 2376 * - single pool: 2377 * mb_pool is not NULL, rx_conf.rx_nseg is 0. 2378 * - multiple segments description: 2379 * mb_pool is NULL, rx_conf.rx_seg is not NULL, rx_conf.rx_nseg is not 0. 2380 * Taken only if flag RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT is set in offloads. 2381 * 2382 * @return 2383 * - 0: Success, receive queue correctly set up. 2384 * - -EIO: if device is removed. 2385 * - -ENODEV: if *port_id* is invalid. 2386 * - -EINVAL: The memory pool pointer is null or the size of network buffers 2387 * which can be allocated from this memory pool does not fit the various 2388 * buffer sizes allowed by the device controller. 2389 * - -ENOMEM: Unable to allocate the receive ring descriptors or to 2390 * allocate network memory buffers from the memory pool when 2391 * initializing receive descriptors. 2392 */ 2393 int rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id, 2394 uint16_t nb_rx_desc, unsigned int socket_id, 2395 const struct rte_eth_rxconf *rx_conf, 2396 struct rte_mempool *mb_pool); 2397 2398 /** 2399 * @warning 2400 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 2401 * 2402 * Allocate and set up a hairpin receive queue for an Ethernet device. 2403 * 2404 * The function set up the selected queue to be used in hairpin. 2405 * 2406 * @param port_id 2407 * The port identifier of the Ethernet device. 2408 * @param rx_queue_id 2409 * The index of the receive queue to set up. 2410 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 2411 * to rte_eth_dev_configure(). 2412 * @param nb_rx_desc 2413 * The number of receive descriptors to allocate for the receive ring. 2414 * 0 means the PMD will use default value. 2415 * @param conf 2416 * The pointer to the hairpin configuration. 2417 * 2418 * @return 2419 * - (0) if successful. 2420 * - (-ENODEV) if *port_id* is invalid. 2421 * - (-ENOTSUP) if hardware doesn't support. 2422 * - (-EINVAL) if bad parameter. 2423 * - (-ENOMEM) if unable to allocate the resources. 2424 */ 2425 __rte_experimental 2426 int rte_eth_rx_hairpin_queue_setup 2427 (uint16_t port_id, uint16_t rx_queue_id, uint16_t nb_rx_desc, 2428 const struct rte_eth_hairpin_conf *conf); 2429 2430 /** 2431 * Allocate and set up a transmit queue for an Ethernet device. 2432 * 2433 * @param port_id 2434 * The port identifier of the Ethernet device. 2435 * @param tx_queue_id 2436 * The index of the transmit queue to set up. 2437 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 2438 * to rte_eth_dev_configure(). 2439 * @param nb_tx_desc 2440 * The number of transmit descriptors to allocate for the transmit ring. 2441 * @param socket_id 2442 * The *socket_id* argument is the socket identifier in case of NUMA. 2443 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for 2444 * the DMA memory allocated for the transmit descriptors of the ring. 2445 * @param tx_conf 2446 * The pointer to the configuration data to be used for the transmit queue. 2447 * NULL value is allowed, in which case default Tx configuration 2448 * will be used. 2449 * The *tx_conf* structure contains the following data: 2450 * - The *tx_thresh* structure with the values of the Prefetch, Host, and 2451 * Write-Back threshold registers of the transmit ring. 2452 * When setting Write-Back threshold to the value greater then zero, 2453 * *tx_rs_thresh* value should be explicitly set to one. 2454 * - The *tx_free_thresh* value indicates the [minimum] number of network 2455 * buffers that must be pending in the transmit ring to trigger their 2456 * [implicit] freeing by the driver transmit function. 2457 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit 2458 * descriptors that must be pending in the transmit ring before setting the 2459 * RS bit on a descriptor by the driver transmit function. 2460 * The *tx_rs_thresh* value should be less or equal then 2461 * *tx_free_thresh* value, and both of them should be less then 2462 * *nb_tx_desc* - 3. 2463 * - The *offloads* member contains Tx offloads to be enabled. 2464 * If an offloading set in tx_conf->offloads 2465 * hasn't been set in the input argument eth_conf->txmode.offloads 2466 * to rte_eth_dev_configure(), it is a new added offloading, it must be 2467 * per-queue type and it is enabled for the queue. 2468 * No need to repeat any bit in tx_conf->offloads which has already been 2469 * enabled in rte_eth_dev_configure() at port level. An offloading enabled 2470 * at port level can't be disabled at queue level. 2471 * 2472 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces 2473 * the transmit function to use default values. 2474 * @return 2475 * - 0: Success, the transmit queue is correctly set up. 2476 * - -ENOMEM: Unable to allocate the transmit ring descriptors. 2477 */ 2478 int rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id, 2479 uint16_t nb_tx_desc, unsigned int socket_id, 2480 const struct rte_eth_txconf *tx_conf); 2481 2482 /** 2483 * @warning 2484 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 2485 * 2486 * Allocate and set up a transmit hairpin queue for an Ethernet device. 2487 * 2488 * @param port_id 2489 * The port identifier of the Ethernet device. 2490 * @param tx_queue_id 2491 * The index of the transmit queue to set up. 2492 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 2493 * to rte_eth_dev_configure(). 2494 * @param nb_tx_desc 2495 * The number of transmit descriptors to allocate for the transmit ring. 2496 * 0 to set default PMD value. 2497 * @param conf 2498 * The hairpin configuration. 2499 * 2500 * @return 2501 * - (0) if successful. 2502 * - (-ENODEV) if *port_id* is invalid. 2503 * - (-ENOTSUP) if hardware doesn't support. 2504 * - (-EINVAL) if bad parameter. 2505 * - (-ENOMEM) if unable to allocate the resources. 2506 */ 2507 __rte_experimental 2508 int rte_eth_tx_hairpin_queue_setup 2509 (uint16_t port_id, uint16_t tx_queue_id, uint16_t nb_tx_desc, 2510 const struct rte_eth_hairpin_conf *conf); 2511 2512 /** 2513 * @warning 2514 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 2515 * 2516 * Get all the hairpin peer Rx / Tx ports of the current port. 2517 * The caller should ensure that the array is large enough to save the ports 2518 * list. 2519 * 2520 * @param port_id 2521 * The port identifier of the Ethernet device. 2522 * @param peer_ports 2523 * Pointer to the array to store the peer ports list. 2524 * @param len 2525 * Length of the array to store the port identifiers. 2526 * @param direction 2527 * Current port to peer port direction 2528 * positive - current used as Tx to get all peer Rx ports. 2529 * zero - current used as Rx to get all peer Tx ports. 2530 * 2531 * @return 2532 * - (0 or positive) actual peer ports number. 2533 * - (-EINVAL) if bad parameter. 2534 * - (-ENODEV) if *port_id* invalid 2535 * - (-ENOTSUP) if hardware doesn't support. 2536 * - Others detailed errors from PMDs. 2537 */ 2538 __rte_experimental 2539 int rte_eth_hairpin_get_peer_ports(uint16_t port_id, uint16_t *peer_ports, 2540 size_t len, uint32_t direction); 2541 2542 /** 2543 * @warning 2544 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 2545 * 2546 * Bind all hairpin Tx queues of one port to the Rx queues of the peer port. 2547 * It is only allowed to call this function after all hairpin queues are 2548 * configured properly and the devices are in started state. 2549 * 2550 * @param tx_port 2551 * The identifier of the Tx port. 2552 * @param rx_port 2553 * The identifier of peer Rx port. 2554 * RTE_MAX_ETHPORTS is allowed for the traversal of all devices. 2555 * Rx port ID could have the same value as Tx port ID. 2556 * 2557 * @return 2558 * - (0) if successful. 2559 * - (-ENODEV) if Tx port ID is invalid. 2560 * - (-EBUSY) if device is not in started state. 2561 * - (-ENOTSUP) if hardware doesn't support. 2562 * - Others detailed errors from PMDs. 2563 */ 2564 __rte_experimental 2565 int rte_eth_hairpin_bind(uint16_t tx_port, uint16_t rx_port); 2566 2567 /** 2568 * @warning 2569 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 2570 * 2571 * Unbind all hairpin Tx queues of one port from the Rx queues of the peer port. 2572 * This should be called before closing the Tx or Rx devices, if the bind 2573 * function is called before. 2574 * After unbinding the hairpin ports pair, it is allowed to bind them again. 2575 * Changing queues configuration should be after stopping the device(s). 2576 * 2577 * @param tx_port 2578 * The identifier of the Tx port. 2579 * @param rx_port 2580 * The identifier of peer Rx port. 2581 * RTE_MAX_ETHPORTS is allowed for traversal of all devices. 2582 * Rx port ID could have the same value as Tx port ID. 2583 * 2584 * @return 2585 * - (0) if successful. 2586 * - (-ENODEV) if Tx port ID is invalid. 2587 * - (-EBUSY) if device is in stopped state. 2588 * - (-ENOTSUP) if hardware doesn't support. 2589 * - Others detailed errors from PMDs. 2590 */ 2591 __rte_experimental 2592 int rte_eth_hairpin_unbind(uint16_t tx_port, uint16_t rx_port); 2593 2594 /** 2595 * @warning 2596 * @b EXPERIMENTAL: this API may change without prior notice. 2597 * 2598 * Get the number of aggregated ports of the DPDK port (specified with port_id). 2599 * It is used when multiple ports are aggregated into a single one. 2600 * 2601 * For the regular physical port doesn't have aggregated ports, 2602 * the number of aggregated ports is reported as 0. 2603 * 2604 * @param port_id 2605 * The port identifier of the Ethernet device. 2606 * @return 2607 * - (>=0) the number of aggregated port if success. 2608 */ 2609 __rte_experimental 2610 int rte_eth_dev_count_aggr_ports(uint16_t port_id); 2611 2612 /** 2613 * @warning 2614 * @b EXPERIMENTAL: this API may change without prior notice. 2615 * 2616 * Map a Tx queue with an aggregated port of the DPDK port (specified with port_id). 2617 * When multiple ports are aggregated into a single one, 2618 * it allows to choose which port to use for Tx via a queue. 2619 * 2620 * The application should use rte_eth_dev_map_aggr_tx_affinity() 2621 * after rte_eth_dev_configure(), rte_eth_tx_queue_setup(), and 2622 * before rte_eth_dev_start(). 2623 * 2624 * @param port_id 2625 * The identifier of the port used in rte_eth_tx_burst(). 2626 * @param tx_queue_id 2627 * The index of the transmit queue used in rte_eth_tx_burst(). 2628 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 2629 * to rte_eth_dev_configure(). 2630 * @param affinity 2631 * The number of the aggregated port. 2632 * Value 0 means no affinity and traffic could be routed to any aggregated port. 2633 * The first aggregated port is number 1 and so on. 2634 * The maximum number is given by rte_eth_dev_count_aggr_ports(). 2635 * 2636 * @return 2637 * Zero if successful. Non-zero otherwise. 2638 */ 2639 __rte_experimental 2640 int rte_eth_dev_map_aggr_tx_affinity(uint16_t port_id, uint16_t tx_queue_id, 2641 uint8_t affinity); 2642 2643 /** 2644 * Return the NUMA socket to which an Ethernet device is connected 2645 * 2646 * @param port_id 2647 * The port identifier of the Ethernet device 2648 * @return 2649 * - The NUMA socket ID which the Ethernet device is connected to. 2650 * - -1 (which translates to SOCKET_ID_ANY) if the socket could not be 2651 * determined. rte_errno is then set to: 2652 * - EINVAL is the port_id is invalid, 2653 * - 0 is the socket could not be determined, 2654 */ 2655 int rte_eth_dev_socket_id(uint16_t port_id); 2656 2657 /** 2658 * Check if port_id of device is attached 2659 * 2660 * @param port_id 2661 * The port identifier of the Ethernet device 2662 * @return 2663 * - 0 if port is out of range or not attached 2664 * - 1 if device is attached 2665 */ 2666 int rte_eth_dev_is_valid_port(uint16_t port_id); 2667 2668 /** 2669 * Start specified Rx queue of a port. It is used when rx_deferred_start 2670 * flag of the specified queue is true. 2671 * 2672 * @param port_id 2673 * The port identifier of the Ethernet device 2674 * @param rx_queue_id 2675 * The index of the Rx queue to update the ring. 2676 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 2677 * to rte_eth_dev_configure(). 2678 * @return 2679 * - 0: Success, the receive queue is started. 2680 * - -ENODEV: if *port_id* is invalid. 2681 * - -EINVAL: The queue_id out of range or belong to hairpin. 2682 * - -EIO: if device is removed. 2683 * - -ENOTSUP: The function not supported in PMD. 2684 */ 2685 int rte_eth_dev_rx_queue_start(uint16_t port_id, uint16_t rx_queue_id); 2686 2687 /** 2688 * Stop specified Rx queue of a port 2689 * 2690 * @param port_id 2691 * The port identifier of the Ethernet device 2692 * @param rx_queue_id 2693 * The index of the Rx queue to update the ring. 2694 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 2695 * to rte_eth_dev_configure(). 2696 * @return 2697 * - 0: Success, the receive queue is stopped. 2698 * - -ENODEV: if *port_id* is invalid. 2699 * - -EINVAL: The queue_id out of range or belong to hairpin. 2700 * - -EIO: if device is removed. 2701 * - -ENOTSUP: The function not supported in PMD. 2702 */ 2703 int rte_eth_dev_rx_queue_stop(uint16_t port_id, uint16_t rx_queue_id); 2704 2705 /** 2706 * Start Tx for specified queue of a port. It is used when tx_deferred_start 2707 * flag of the specified queue is true. 2708 * 2709 * @param port_id 2710 * The port identifier of the Ethernet device 2711 * @param tx_queue_id 2712 * The index of the Tx queue to update the ring. 2713 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 2714 * to rte_eth_dev_configure(). 2715 * @return 2716 * - 0: Success, the transmit queue is started. 2717 * - -ENODEV: if *port_id* is invalid. 2718 * - -EINVAL: The queue_id out of range or belong to hairpin. 2719 * - -EIO: if device is removed. 2720 * - -ENOTSUP: The function not supported in PMD. 2721 */ 2722 int rte_eth_dev_tx_queue_start(uint16_t port_id, uint16_t tx_queue_id); 2723 2724 /** 2725 * Stop specified Tx queue of a port 2726 * 2727 * @param port_id 2728 * The port identifier of the Ethernet device 2729 * @param tx_queue_id 2730 * The index of the Tx queue to update the ring. 2731 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 2732 * to rte_eth_dev_configure(). 2733 * @return 2734 * - 0: Success, the transmit queue is stopped. 2735 * - -ENODEV: if *port_id* is invalid. 2736 * - -EINVAL: The queue_id out of range or belong to hairpin. 2737 * - -EIO: if device is removed. 2738 * - -ENOTSUP: The function not supported in PMD. 2739 */ 2740 int rte_eth_dev_tx_queue_stop(uint16_t port_id, uint16_t tx_queue_id); 2741 2742 /** 2743 * Start an Ethernet device. 2744 * 2745 * The device start step is the last one and consists of setting the configured 2746 * offload features and in starting the transmit and the receive units of the 2747 * device. 2748 * 2749 * Device RTE_ETH_DEV_NOLIVE_MAC_ADDR flag causes MAC address to be set before 2750 * PMD port start callback function is invoked. 2751 * 2752 * On success, all basic functions exported by the Ethernet API (link status, 2753 * receive/transmit, and so on) can be invoked. 2754 * 2755 * @param port_id 2756 * The port identifier of the Ethernet device. 2757 * @return 2758 * - 0: Success, Ethernet device started. 2759 * - -EAGAIN: If start operation must be retried. 2760 * - <0: Error code of the driver device start function. 2761 */ 2762 int rte_eth_dev_start(uint16_t port_id); 2763 2764 /** 2765 * Stop an Ethernet device. The device can be restarted with a call to 2766 * rte_eth_dev_start() 2767 * 2768 * @param port_id 2769 * The port identifier of the Ethernet device. 2770 * @return 2771 * - 0: Success, Ethernet device stopped. 2772 * - -EBUSY: If stopping the port is not allowed in current state. 2773 * - <0: Error code of the driver device stop function. 2774 */ 2775 int rte_eth_dev_stop(uint16_t port_id); 2776 2777 /** 2778 * Link up an Ethernet device. 2779 * 2780 * Set device link up will re-enable the device Rx/Tx 2781 * functionality after it is previously set device linked down. 2782 * 2783 * @param port_id 2784 * The port identifier of the Ethernet device. 2785 * @return 2786 * - 0: Success, Ethernet device linked up. 2787 * - <0: Error code of the driver device link up function. 2788 */ 2789 int rte_eth_dev_set_link_up(uint16_t port_id); 2790 2791 /** 2792 * Link down an Ethernet device. 2793 * The device Rx/Tx functionality will be disabled if success, 2794 * and it can be re-enabled with a call to 2795 * rte_eth_dev_set_link_up() 2796 * 2797 * @param port_id 2798 * The port identifier of the Ethernet device. 2799 */ 2800 int rte_eth_dev_set_link_down(uint16_t port_id); 2801 2802 /** 2803 * Close a stopped Ethernet device. The device cannot be restarted! 2804 * The function frees all port resources. 2805 * 2806 * @param port_id 2807 * The port identifier of the Ethernet device. 2808 * @return 2809 * - Zero if the port is closed successfully. 2810 * - Negative if something went wrong. 2811 */ 2812 int rte_eth_dev_close(uint16_t port_id); 2813 2814 /** 2815 * Reset a Ethernet device and keep its port ID. 2816 * 2817 * When a port has to be reset passively, the DPDK application can invoke 2818 * this function. For example when a PF is reset, all its VFs should also 2819 * be reset. Normally a DPDK application can invoke this function when 2820 * RTE_ETH_EVENT_INTR_RESET event is detected, but can also use it to start 2821 * a port reset in other circumstances. 2822 * 2823 * When this function is called, it first stops the port and then calls the 2824 * PMD specific dev_uninit( ) and dev_init( ) to return the port to initial 2825 * state, in which no Tx and Rx queues are setup, as if the port has been 2826 * reset and not started. The port keeps the port ID it had before the 2827 * function call. 2828 * 2829 * After calling rte_eth_dev_reset( ), the application should use 2830 * rte_eth_dev_configure( ), rte_eth_rx_queue_setup( ), 2831 * rte_eth_tx_queue_setup( ), and rte_eth_dev_start( ) 2832 * to reconfigure the device as appropriate. 2833 * 2834 * Note: To avoid unexpected behavior, the application should stop calling 2835 * Tx and Rx functions before calling rte_eth_dev_reset( ). For thread 2836 * safety, all these controlling functions should be called from the same 2837 * thread. 2838 * 2839 * @param port_id 2840 * The port identifier of the Ethernet device. 2841 * 2842 * @return 2843 * - (0) if successful. 2844 * - (-ENODEV) if *port_id* is invalid. 2845 * - (-ENOTSUP) if hardware doesn't support this function. 2846 * - (-EPERM) if not ran from the primary process. 2847 * - (-EIO) if re-initialisation failed or device is removed. 2848 * - (-ENOMEM) if the reset failed due to OOM. 2849 * - (-EAGAIN) if the reset temporarily failed and should be retried later. 2850 */ 2851 int rte_eth_dev_reset(uint16_t port_id); 2852 2853 /** 2854 * Enable receipt in promiscuous mode for an Ethernet device. 2855 * 2856 * @param port_id 2857 * The port identifier of the Ethernet device. 2858 * @return 2859 * - (0) if successful. 2860 * - (-ENOTSUP) if support for promiscuous_enable() does not exist 2861 * for the device. 2862 * - (-ENODEV) if *port_id* invalid. 2863 */ 2864 int rte_eth_promiscuous_enable(uint16_t port_id); 2865 2866 /** 2867 * Disable receipt in promiscuous mode for an Ethernet device. 2868 * 2869 * @param port_id 2870 * The port identifier of the Ethernet device. 2871 * @return 2872 * - (0) if successful. 2873 * - (-ENOTSUP) if support for promiscuous_disable() does not exist 2874 * for the device. 2875 * - (-ENODEV) if *port_id* invalid. 2876 */ 2877 int rte_eth_promiscuous_disable(uint16_t port_id); 2878 2879 /** 2880 * Return the value of promiscuous mode for an Ethernet device. 2881 * 2882 * @param port_id 2883 * The port identifier of the Ethernet device. 2884 * @return 2885 * - (1) if promiscuous is enabled 2886 * - (0) if promiscuous is disabled. 2887 * - (-1) on error 2888 */ 2889 int rte_eth_promiscuous_get(uint16_t port_id); 2890 2891 /** 2892 * Enable the receipt of any multicast frame by an Ethernet device. 2893 * 2894 * @param port_id 2895 * The port identifier of the Ethernet device. 2896 * @return 2897 * - (0) if successful. 2898 * - (-ENOTSUP) if support for allmulticast_enable() does not exist 2899 * for the device. 2900 * - (-ENODEV) if *port_id* invalid. 2901 */ 2902 int rte_eth_allmulticast_enable(uint16_t port_id); 2903 2904 /** 2905 * Disable the receipt of all multicast frames by an Ethernet device. 2906 * 2907 * @param port_id 2908 * The port identifier of the Ethernet device. 2909 * @return 2910 * - (0) if successful. 2911 * - (-ENOTSUP) if support for allmulticast_disable() does not exist 2912 * for the device. 2913 * - (-ENODEV) if *port_id* invalid. 2914 */ 2915 int rte_eth_allmulticast_disable(uint16_t port_id); 2916 2917 /** 2918 * Return the value of allmulticast mode for an Ethernet device. 2919 * 2920 * @param port_id 2921 * The port identifier of the Ethernet device. 2922 * @return 2923 * - (1) if allmulticast is enabled 2924 * - (0) if allmulticast is disabled. 2925 * - (-1) on error 2926 */ 2927 int rte_eth_allmulticast_get(uint16_t port_id); 2928 2929 /** 2930 * Retrieve the link status (up/down), the duplex mode (half/full), 2931 * the negotiation (auto/fixed), and if available, the speed (Mbps). 2932 * 2933 * It might need to wait up to 9 seconds. 2934 * @see rte_eth_link_get_nowait. 2935 * 2936 * @param port_id 2937 * The port identifier of the Ethernet device. 2938 * @param link 2939 * Link information written back. 2940 * @return 2941 * - (0) if successful. 2942 * - (-ENOTSUP) if the function is not supported in PMD. 2943 * - (-ENODEV) if *port_id* invalid. 2944 * - (-EINVAL) if bad parameter. 2945 */ 2946 int rte_eth_link_get(uint16_t port_id, struct rte_eth_link *link); 2947 2948 /** 2949 * Retrieve the link status (up/down), the duplex mode (half/full), 2950 * the negotiation (auto/fixed), and if available, the speed (Mbps). 2951 * 2952 * @param port_id 2953 * The port identifier of the Ethernet device. 2954 * @param link 2955 * Link information written back. 2956 * @return 2957 * - (0) if successful. 2958 * - (-ENOTSUP) if the function is not supported in PMD. 2959 * - (-ENODEV) if *port_id* invalid. 2960 * - (-EINVAL) if bad parameter. 2961 */ 2962 int rte_eth_link_get_nowait(uint16_t port_id, struct rte_eth_link *link); 2963 2964 /** 2965 * @warning 2966 * @b EXPERIMENTAL: this API may change without prior notice. 2967 * 2968 * The function converts a link_speed to a string. It handles all special 2969 * values like unknown or none speed. 2970 * 2971 * @param link_speed 2972 * link_speed of rte_eth_link struct 2973 * @return 2974 * Link speed in textual format. It's pointer to immutable memory. 2975 * No free is required. 2976 */ 2977 __rte_experimental 2978 const char *rte_eth_link_speed_to_str(uint32_t link_speed); 2979 2980 /** 2981 * @warning 2982 * @b EXPERIMENTAL: this API may change without prior notice. 2983 * 2984 * The function converts a rte_eth_link struct representing a link status to 2985 * a string. 2986 * 2987 * @param str 2988 * A pointer to a string to be filled with textual representation of 2989 * device status. At least RTE_ETH_LINK_MAX_STR_LEN bytes should be allocated to 2990 * store default link status text. 2991 * @param len 2992 * Length of available memory at 'str' string. 2993 * @param eth_link 2994 * Link status returned by rte_eth_link_get function 2995 * @return 2996 * Number of bytes written to str array or -EINVAL if bad parameter. 2997 */ 2998 __rte_experimental 2999 int rte_eth_link_to_str(char *str, size_t len, 3000 const struct rte_eth_link *eth_link); 3001 3002 /** 3003 * Retrieve the general I/O statistics of an Ethernet device. 3004 * 3005 * @param port_id 3006 * The port identifier of the Ethernet device. 3007 * @param stats 3008 * A pointer to a structure of type *rte_eth_stats* to be filled with 3009 * the values of device counters for the following set of statistics: 3010 * - *ipackets* with the total of successfully received packets. 3011 * - *opackets* with the total of successfully transmitted packets. 3012 * - *ibytes* with the total of successfully received bytes. 3013 * - *obytes* with the total of successfully transmitted bytes. 3014 * - *ierrors* with the total of erroneous received packets. 3015 * - *oerrors* with the total of failed transmitted packets. 3016 * @return 3017 * Zero if successful. Non-zero otherwise. 3018 */ 3019 int rte_eth_stats_get(uint16_t port_id, struct rte_eth_stats *stats); 3020 3021 /** 3022 * Reset the general I/O statistics of an Ethernet device. 3023 * 3024 * @param port_id 3025 * The port identifier of the Ethernet device. 3026 * @return 3027 * - (0) if device notified to reset stats. 3028 * - (-ENOTSUP) if hardware doesn't support. 3029 * - (-ENODEV) if *port_id* invalid. 3030 * - (<0): Error code of the driver stats reset function. 3031 */ 3032 int rte_eth_stats_reset(uint16_t port_id); 3033 3034 /** 3035 * Retrieve names of extended statistics of an Ethernet device. 3036 * 3037 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched 3038 * by array index: 3039 * xstats_names[i].name => xstats[i].value 3040 * 3041 * And the array index is same with id field of 'struct rte_eth_xstat': 3042 * xstats[i].id == i 3043 * 3044 * This assumption makes key-value pair matching less flexible but simpler. 3045 * 3046 * @param port_id 3047 * The port identifier of the Ethernet device. 3048 * @param xstats_names 3049 * An rte_eth_xstat_name array of at least *size* elements to 3050 * be filled. If set to NULL, the function returns the required number 3051 * of elements. 3052 * @param size 3053 * The size of the xstats_names array (number of elements). 3054 * @return 3055 * - A positive value lower or equal to size: success. The return value 3056 * is the number of entries filled in the stats table. 3057 * - A positive value higher than size: error, the given statistics table 3058 * is too small. The return value corresponds to the size that should 3059 * be given to succeed. The entries in the table are not valid and 3060 * shall not be used by the caller. 3061 * - A negative value on error (invalid port ID). 3062 */ 3063 int rte_eth_xstats_get_names(uint16_t port_id, 3064 struct rte_eth_xstat_name *xstats_names, 3065 unsigned int size); 3066 3067 /** 3068 * Retrieve extended statistics of an Ethernet device. 3069 * 3070 * There is an assumption that 'xstat_names' and 'xstats' arrays are matched 3071 * by array index: 3072 * xstats_names[i].name => xstats[i].value 3073 * 3074 * And the array index is same with id field of 'struct rte_eth_xstat': 3075 * xstats[i].id == i 3076 * 3077 * This assumption makes key-value pair matching less flexible but simpler. 3078 * 3079 * @param port_id 3080 * The port identifier of the Ethernet device. 3081 * @param xstats 3082 * A pointer to a table of structure of type *rte_eth_xstat* 3083 * to be filled with device statistics ids and values. 3084 * This parameter can be set to NULL if and only if n is 0. 3085 * @param n 3086 * The size of the xstats array (number of elements). 3087 * If lower than the required number of elements, the function returns 3088 * the required number of elements. 3089 * If equal to zero, the xstats must be NULL, the function returns the 3090 * required number of elements. 3091 * @return 3092 * - A positive value lower or equal to n: success. The return value 3093 * is the number of entries filled in the stats table. 3094 * - A positive value higher than n: error, the given statistics table 3095 * is too small. The return value corresponds to the size that should 3096 * be given to succeed. The entries in the table are not valid and 3097 * shall not be used by the caller. 3098 * - A negative value on error (invalid port ID). 3099 */ 3100 int rte_eth_xstats_get(uint16_t port_id, struct rte_eth_xstat *xstats, 3101 unsigned int n); 3102 3103 /** 3104 * Retrieve names of extended statistics of an Ethernet device. 3105 * 3106 * @param port_id 3107 * The port identifier of the Ethernet device. 3108 * @param xstats_names 3109 * Array to be filled in with names of requested device statistics. 3110 * Must not be NULL if @p ids are specified (not NULL). 3111 * @param size 3112 * Number of elements in @p xstats_names array (if not NULL) and in 3113 * @p ids array (if not NULL). Must be 0 if both array pointers are NULL. 3114 * @param ids 3115 * IDs array given by app to retrieve specific statistics. May be NULL to 3116 * retrieve names of all available statistics or, if @p xstats_names is 3117 * NULL as well, just the number of available statistics. 3118 * @return 3119 * - A positive value lower or equal to size: success. The return value 3120 * is the number of entries filled in the stats table. 3121 * - A positive value higher than size: success. The given statistics table 3122 * is too small. The return value corresponds to the size that should 3123 * be given to succeed. The entries in the table are not valid and 3124 * shall not be used by the caller. 3125 * - A negative value on error. 3126 */ 3127 int 3128 rte_eth_xstats_get_names_by_id(uint16_t port_id, 3129 struct rte_eth_xstat_name *xstats_names, unsigned int size, 3130 uint64_t *ids); 3131 3132 /** 3133 * Retrieve extended statistics of an Ethernet device. 3134 * 3135 * @param port_id 3136 * The port identifier of the Ethernet device. 3137 * @param ids 3138 * IDs array given by app to retrieve specific statistics. May be NULL to 3139 * retrieve all available statistics or, if @p values is NULL as well, 3140 * just the number of available statistics. 3141 * @param values 3142 * Array to be filled in with requested device statistics. 3143 * Must not be NULL if ids are specified (not NULL). 3144 * @param size 3145 * Number of elements in @p values array (if not NULL) and in @p ids 3146 * array (if not NULL). Must be 0 if both array pointers are NULL. 3147 * @return 3148 * - A positive value lower or equal to size: success. The return value 3149 * is the number of entries filled in the stats table. 3150 * - A positive value higher than size: success: The given statistics table 3151 * is too small. The return value corresponds to the size that should 3152 * be given to succeed. The entries in the table are not valid and 3153 * shall not be used by the caller. 3154 * - A negative value on error. 3155 */ 3156 int rte_eth_xstats_get_by_id(uint16_t port_id, const uint64_t *ids, 3157 uint64_t *values, unsigned int size); 3158 3159 /** 3160 * Gets the ID of a statistic from its name. 3161 * 3162 * This function searches for the statistics using string compares, and 3163 * as such should not be used on the fast-path. For fast-path retrieval of 3164 * specific statistics, store the ID as provided in *id* from this function, 3165 * and pass the ID to rte_eth_xstats_get() 3166 * 3167 * @param port_id The port to look up statistics from 3168 * @param xstat_name The name of the statistic to return 3169 * @param[out] id A pointer to an app-supplied uint64_t which should be 3170 * set to the ID of the stat if the stat exists. 3171 * @return 3172 * 0 on success 3173 * -ENODEV for invalid port_id, 3174 * -EIO if device is removed, 3175 * -EINVAL if the xstat_name doesn't exist in port_id 3176 * -ENOMEM if bad parameter. 3177 */ 3178 int rte_eth_xstats_get_id_by_name(uint16_t port_id, const char *xstat_name, 3179 uint64_t *id); 3180 3181 /** 3182 * Reset extended statistics of an Ethernet device. 3183 * 3184 * @param port_id 3185 * The port identifier of the Ethernet device. 3186 * @return 3187 * - (0) if device notified to reset extended stats. 3188 * - (-ENOTSUP) if pmd doesn't support both 3189 * extended stats and basic stats reset. 3190 * - (-ENODEV) if *port_id* invalid. 3191 * - (<0): Error code of the driver xstats reset function. 3192 */ 3193 int rte_eth_xstats_reset(uint16_t port_id); 3194 3195 /** 3196 * Set a mapping for the specified transmit queue to the specified per-queue 3197 * statistics counter. 3198 * 3199 * @param port_id 3200 * The port identifier of the Ethernet device. 3201 * @param tx_queue_id 3202 * The index of the transmit queue for which a queue stats mapping is required. 3203 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 3204 * to rte_eth_dev_configure(). 3205 * @param stat_idx 3206 * The per-queue packet statistics functionality number that the transmit 3207 * queue is to be assigned. 3208 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1]. 3209 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256. 3210 * @return 3211 * Zero if successful. Non-zero otherwise. 3212 */ 3213 int rte_eth_dev_set_tx_queue_stats_mapping(uint16_t port_id, 3214 uint16_t tx_queue_id, uint8_t stat_idx); 3215 3216 /** 3217 * Set a mapping for the specified receive queue to the specified per-queue 3218 * statistics counter. 3219 * 3220 * @param port_id 3221 * The port identifier of the Ethernet device. 3222 * @param rx_queue_id 3223 * The index of the receive queue for which a queue stats mapping is required. 3224 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 3225 * to rte_eth_dev_configure(). 3226 * @param stat_idx 3227 * The per-queue packet statistics functionality number that the receive 3228 * queue is to be assigned. 3229 * The value must be in the range [0, RTE_ETHDEV_QUEUE_STAT_CNTRS - 1]. 3230 * Max RTE_ETHDEV_QUEUE_STAT_CNTRS being 256. 3231 * @return 3232 * Zero if successful. Non-zero otherwise. 3233 */ 3234 int rte_eth_dev_set_rx_queue_stats_mapping(uint16_t port_id, 3235 uint16_t rx_queue_id, 3236 uint8_t stat_idx); 3237 3238 /** 3239 * Retrieve the Ethernet address of an Ethernet device. 3240 * 3241 * @param port_id 3242 * The port identifier of the Ethernet device. 3243 * @param mac_addr 3244 * A pointer to a structure of type *ether_addr* to be filled with 3245 * the Ethernet address of the Ethernet device. 3246 * @return 3247 * - (0) if successful 3248 * - (-ENODEV) if *port_id* invalid. 3249 * - (-EINVAL) if bad parameter. 3250 */ 3251 int rte_eth_macaddr_get(uint16_t port_id, struct rte_ether_addr *mac_addr); 3252 3253 /** 3254 * @warning 3255 * @b EXPERIMENTAL: this API may change without prior notice 3256 * 3257 * Retrieve the Ethernet addresses of an Ethernet device. 3258 * 3259 * @param port_id 3260 * The port identifier of the Ethernet device. 3261 * @param ma 3262 * A pointer to an array of structures of type *ether_addr* to be filled with 3263 * the Ethernet addresses of the Ethernet device. 3264 * @param num 3265 * Number of elements in the @p ma array. 3266 * Note that rte_eth_dev_info::max_mac_addrs can be used to retrieve 3267 * max number of Ethernet addresses for given port. 3268 * @return 3269 * - number of retrieved addresses if successful 3270 * - (-ENODEV) if *port_id* invalid. 3271 * - (-EINVAL) if bad parameter. 3272 */ 3273 __rte_experimental 3274 int rte_eth_macaddrs_get(uint16_t port_id, struct rte_ether_addr *ma, 3275 unsigned int num); 3276 3277 /** 3278 * Retrieve the contextual information of an Ethernet device. 3279 * 3280 * This function returns the Ethernet device information based 3281 * on the values stored internally in the device specific data. 3282 * For example: number of queues, descriptor limits, device 3283 * capabilities and offload flags. 3284 * 3285 * @param port_id 3286 * The port identifier of the Ethernet device. 3287 * @param dev_info 3288 * A pointer to a structure of type *rte_eth_dev_info* to be filled with 3289 * the contextual information of the Ethernet device. 3290 * @return 3291 * - (0) if successful. 3292 * - (-ENOTSUP) if support for dev_infos_get() does not exist for the device. 3293 * - (-ENODEV) if *port_id* invalid. 3294 * - (-EINVAL) if bad parameter. 3295 */ 3296 int rte_eth_dev_info_get(uint16_t port_id, struct rte_eth_dev_info *dev_info); 3297 3298 /** 3299 * @warning 3300 * @b EXPERIMENTAL: this API may change without prior notice. 3301 * 3302 * Retrieve the configuration of an Ethernet device. 3303 * 3304 * @param port_id 3305 * The port identifier of the Ethernet device. 3306 * @param dev_conf 3307 * Location for Ethernet device configuration to be filled in. 3308 * @return 3309 * - (0) if successful. 3310 * - (-ENODEV) if *port_id* invalid. 3311 * - (-EINVAL) if bad parameter. 3312 */ 3313 __rte_experimental 3314 int rte_eth_dev_conf_get(uint16_t port_id, struct rte_eth_conf *dev_conf); 3315 3316 /** 3317 * Retrieve the firmware version of a device. 3318 * 3319 * @param port_id 3320 * The port identifier of the device. 3321 * @param fw_version 3322 * A pointer to a string array storing the firmware version of a device, 3323 * the string includes terminating null. This pointer is allocated by caller. 3324 * @param fw_size 3325 * The size of the string array pointed by fw_version, which should be 3326 * large enough to store firmware version of the device. 3327 * @return 3328 * - (0) if successful. 3329 * - (-ENOTSUP) if operation is not supported. 3330 * - (-ENODEV) if *port_id* invalid. 3331 * - (-EIO) if device is removed. 3332 * - (-EINVAL) if bad parameter. 3333 * - (>0) if *fw_size* is not enough to store firmware version, return 3334 * the size of the non truncated string. 3335 */ 3336 int rte_eth_dev_fw_version_get(uint16_t port_id, 3337 char *fw_version, size_t fw_size); 3338 3339 /** 3340 * Retrieve the supported packet types of an Ethernet device. 3341 * 3342 * When a packet type is announced as supported, it *must* be recognized by 3343 * the PMD. For instance, if RTE_PTYPE_L2_ETHER, RTE_PTYPE_L2_ETHER_VLAN 3344 * and RTE_PTYPE_L3_IPV4 are announced, the PMD must return the following 3345 * packet types for these packets: 3346 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4 3347 * - Ether/VLAN/IPv4 -> RTE_PTYPE_L2_ETHER_VLAN | RTE_PTYPE_L3_IPV4 3348 * - Ether/[anything else] -> RTE_PTYPE_L2_ETHER 3349 * - Ether/VLAN/[anything else] -> RTE_PTYPE_L2_ETHER_VLAN 3350 * 3351 * When a packet is received by a PMD, the most precise type must be 3352 * returned among the ones supported. However a PMD is allowed to set 3353 * packet type that is not in the supported list, at the condition that it 3354 * is more precise. Therefore, a PMD announcing no supported packet types 3355 * can still set a matching packet type in a received packet. 3356 * 3357 * @note 3358 * Better to invoke this API after the device is already started or Rx burst 3359 * function is decided, to obtain correct supported ptypes. 3360 * @note 3361 * if a given PMD does not report what ptypes it supports, then the supported 3362 * ptype count is reported as 0. 3363 * @param port_id 3364 * The port identifier of the Ethernet device. 3365 * @param ptype_mask 3366 * A hint of what kind of packet type which the caller is interested in. 3367 * @param ptypes 3368 * An array pointer to store adequate packet types, allocated by caller. 3369 * @param num 3370 * Size of the array pointed by param ptypes. 3371 * @return 3372 * - (>=0) Number of supported ptypes. If the number of types exceeds num, 3373 * only num entries will be filled into the ptypes array, but the full 3374 * count of supported ptypes will be returned. 3375 * - (-ENODEV) if *port_id* invalid. 3376 * - (-EINVAL) if bad parameter. 3377 */ 3378 int rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask, 3379 uint32_t *ptypes, int num); 3380 /** 3381 * Inform Ethernet device about reduced range of packet types to handle. 3382 * 3383 * Application can use this function to set only specific ptypes that it's 3384 * interested. This information can be used by the PMD to optimize Rx path. 3385 * 3386 * The function accepts an array `set_ptypes` allocated by the caller to 3387 * store the packet types set by the driver, the last element of the array 3388 * is set to RTE_PTYPE_UNKNOWN. The size of the `set_ptype` array should be 3389 * `rte_eth_dev_get_supported_ptypes() + 1` else it might only be filled 3390 * partially. 3391 * 3392 * @param port_id 3393 * The port identifier of the Ethernet device. 3394 * @param ptype_mask 3395 * The ptype family that application is interested in should be bitwise OR of 3396 * RTE_PTYPE_*_MASK or 0. 3397 * @param set_ptypes 3398 * An array pointer to store set packet types, allocated by caller. The 3399 * function marks the end of array with RTE_PTYPE_UNKNOWN. 3400 * @param num 3401 * Size of the array pointed by param ptypes. 3402 * Should be rte_eth_dev_get_supported_ptypes() + 1 to accommodate the 3403 * set ptypes. 3404 * @return 3405 * - (0) if Success. 3406 * - (-ENODEV) if *port_id* invalid. 3407 * - (-EINVAL) if *ptype_mask* is invalid (or) set_ptypes is NULL and 3408 * num > 0. 3409 */ 3410 int rte_eth_dev_set_ptypes(uint16_t port_id, uint32_t ptype_mask, 3411 uint32_t *set_ptypes, unsigned int num); 3412 3413 /** 3414 * Retrieve the MTU of an Ethernet device. 3415 * 3416 * @param port_id 3417 * The port identifier of the Ethernet device. 3418 * @param mtu 3419 * A pointer to a uint16_t where the retrieved MTU is to be stored. 3420 * @return 3421 * - (0) if successful. 3422 * - (-ENODEV) if *port_id* invalid. 3423 * - (-EINVAL) if bad parameter. 3424 */ 3425 int rte_eth_dev_get_mtu(uint16_t port_id, uint16_t *mtu); 3426 3427 /** 3428 * Change the MTU of an Ethernet device. 3429 * 3430 * @param port_id 3431 * The port identifier of the Ethernet device. 3432 * @param mtu 3433 * A uint16_t for the MTU to be applied. 3434 * @return 3435 * - (0) if successful. 3436 * - (-ENOTSUP) if operation is not supported. 3437 * - (-ENODEV) if *port_id* invalid. 3438 * - (-EIO) if device is removed. 3439 * - (-EINVAL) if *mtu* invalid, validation of mtu can occur within 3440 * rte_eth_dev_set_mtu if dev_infos_get is supported by the device or 3441 * when the mtu is set using dev->dev_ops->mtu_set. 3442 * - (-EBUSY) if operation is not allowed when the port is running 3443 */ 3444 int rte_eth_dev_set_mtu(uint16_t port_id, uint16_t mtu); 3445 3446 /** 3447 * Enable/Disable hardware filtering by an Ethernet device of received 3448 * VLAN packets tagged with a given VLAN Tag Identifier. 3449 * 3450 * @param port_id 3451 * The port identifier of the Ethernet device. 3452 * @param vlan_id 3453 * The VLAN Tag Identifier whose filtering must be enabled or disabled. 3454 * @param on 3455 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*. 3456 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*. 3457 * @return 3458 * - (0) if successful. 3459 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured. 3460 * - (-ENODEV) if *port_id* invalid. 3461 * - (-EIO) if device is removed. 3462 * - (-ENOSYS) if VLAN filtering on *port_id* disabled. 3463 * - (-EINVAL) if *vlan_id* > 4095. 3464 */ 3465 int rte_eth_dev_vlan_filter(uint16_t port_id, uint16_t vlan_id, int on); 3466 3467 /** 3468 * Enable/Disable hardware VLAN Strip by a Rx queue of an Ethernet device. 3469 * 3470 * @param port_id 3471 * The port identifier of the Ethernet device. 3472 * @param rx_queue_id 3473 * The index of the receive queue for which a queue stats mapping is required. 3474 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 3475 * to rte_eth_dev_configure(). 3476 * @param on 3477 * If 1, Enable VLAN Stripping of the receive queue of the Ethernet port. 3478 * If 0, Disable VLAN Stripping of the receive queue of the Ethernet port. 3479 * @return 3480 * - (0) if successful. 3481 * - (-ENOTSUP) if hardware-assisted VLAN stripping not configured. 3482 * - (-ENODEV) if *port_id* invalid. 3483 * - (-EINVAL) if *rx_queue_id* invalid. 3484 */ 3485 int rte_eth_dev_set_vlan_strip_on_queue(uint16_t port_id, uint16_t rx_queue_id, 3486 int on); 3487 3488 /** 3489 * Set the Outer VLAN Ether Type by an Ethernet device, it can be inserted to 3490 * the VLAN header. 3491 * 3492 * @param port_id 3493 * The port identifier of the Ethernet device. 3494 * @param vlan_type 3495 * The VLAN type. 3496 * @param tag_type 3497 * The Tag Protocol ID 3498 * @return 3499 * - (0) if successful. 3500 * - (-ENOTSUP) if hardware-assisted VLAN TPID setup is not supported. 3501 * - (-ENODEV) if *port_id* invalid. 3502 * - (-EIO) if device is removed. 3503 */ 3504 int rte_eth_dev_set_vlan_ether_type(uint16_t port_id, 3505 enum rte_vlan_type vlan_type, 3506 uint16_t tag_type); 3507 3508 /** 3509 * Set VLAN offload configuration on an Ethernet device. 3510 * 3511 * @param port_id 3512 * The port identifier of the Ethernet device. 3513 * @param offload_mask 3514 * The VLAN Offload bit mask can be mixed use with "OR" 3515 * RTE_ETH_VLAN_STRIP_OFFLOAD 3516 * RTE_ETH_VLAN_FILTER_OFFLOAD 3517 * RTE_ETH_VLAN_EXTEND_OFFLOAD 3518 * RTE_ETH_QINQ_STRIP_OFFLOAD 3519 * @return 3520 * - (0) if successful. 3521 * - (-ENOTSUP) if hardware-assisted VLAN filtering not configured. 3522 * - (-ENODEV) if *port_id* invalid. 3523 * - (-EIO) if device is removed. 3524 */ 3525 int rte_eth_dev_set_vlan_offload(uint16_t port_id, int offload_mask); 3526 3527 /** 3528 * Read VLAN Offload configuration from an Ethernet device 3529 * 3530 * @param port_id 3531 * The port identifier of the Ethernet device. 3532 * @return 3533 * - (>0) if successful. Bit mask to indicate 3534 * RTE_ETH_VLAN_STRIP_OFFLOAD 3535 * RTE_ETH_VLAN_FILTER_OFFLOAD 3536 * RTE_ETH_VLAN_EXTEND_OFFLOAD 3537 * RTE_ETH_QINQ_STRIP_OFFLOAD 3538 * - (-ENODEV) if *port_id* invalid. 3539 */ 3540 int rte_eth_dev_get_vlan_offload(uint16_t port_id); 3541 3542 /** 3543 * Set port based Tx VLAN insertion on or off. 3544 * 3545 * @param port_id 3546 * The port identifier of the Ethernet device. 3547 * @param pvid 3548 * Port based Tx VLAN identifier together with user priority. 3549 * @param on 3550 * Turn on or off the port based Tx VLAN insertion. 3551 * 3552 * @return 3553 * - (0) if successful. 3554 * - negative if failed. 3555 */ 3556 int rte_eth_dev_set_vlan_pvid(uint16_t port_id, uint16_t pvid, int on); 3557 3558 /** 3559 * @warning 3560 * @b EXPERIMENTAL: this API may change without prior notice. 3561 * 3562 * Set Rx queue available descriptors threshold. 3563 * 3564 * @param port_id 3565 * The port identifier of the Ethernet device. 3566 * @param queue_id 3567 * The index of the receive queue. 3568 * @param avail_thresh 3569 * The available descriptors threshold is percentage of Rx queue size 3570 * which describes the availability of Rx queue for hardware. 3571 * If the Rx queue availability is below it, 3572 * the event RTE_ETH_EVENT_RX_AVAIL_THRESH is triggered. 3573 * [1-99] to set a new available descriptors threshold. 3574 * 0 to disable threshold monitoring. 3575 * 3576 * @return 3577 * - 0 if successful. 3578 * - (-ENODEV) if @p port_id is invalid. 3579 * - (-EINVAL) if bad parameter. 3580 * - (-ENOTSUP) if available Rx descriptors threshold is not supported. 3581 * - (-EIO) if device is removed. 3582 */ 3583 __rte_experimental 3584 int rte_eth_rx_avail_thresh_set(uint16_t port_id, uint16_t queue_id, 3585 uint8_t avail_thresh); 3586 3587 /** 3588 * @warning 3589 * @b EXPERIMENTAL: this API may change without prior notice. 3590 * 3591 * Find Rx queue with RTE_ETH_EVENT_RX_AVAIL_THRESH event pending. 3592 * 3593 * @param port_id 3594 * The port identifier of the Ethernet device. 3595 * @param[inout] queue_id 3596 * On input starting Rx queue index to search from. 3597 * If the queue_id is bigger than maximum queue ID of the port, 3598 * search is started from 0. So that application can keep calling 3599 * this function to handle all pending events with a simple increment 3600 * of queue_id on the next call. 3601 * On output if return value is 1, Rx queue index with the event pending. 3602 * @param[out] avail_thresh 3603 * Location for available descriptors threshold of the found Rx queue. 3604 * 3605 * @return 3606 * - 1 if an Rx queue with pending event is found. 3607 * - 0 if no Rx queue with pending event is found. 3608 * - (-ENODEV) if @p port_id is invalid. 3609 * - (-EINVAL) if bad parameter (e.g. @p queue_id is NULL). 3610 * - (-ENOTSUP) if operation is not supported. 3611 * - (-EIO) if device is removed. 3612 */ 3613 __rte_experimental 3614 int rte_eth_rx_avail_thresh_query(uint16_t port_id, uint16_t *queue_id, 3615 uint8_t *avail_thresh); 3616 3617 typedef void (*buffer_tx_error_fn)(struct rte_mbuf **unsent, uint16_t count, 3618 void *userdata); 3619 3620 /** 3621 * Structure used to buffer packets for future Tx 3622 * Used by APIs rte_eth_tx_buffer and rte_eth_tx_buffer_flush 3623 */ 3624 struct rte_eth_dev_tx_buffer { 3625 buffer_tx_error_fn error_callback; 3626 void *error_userdata; 3627 uint16_t size; /**< Size of buffer for buffered Tx */ 3628 uint16_t length; /**< Number of packets in the array */ 3629 /** Pending packets to be sent on explicit flush or when full */ 3630 struct rte_mbuf *pkts[]; 3631 }; 3632 3633 /** 3634 * Calculate the size of the Tx buffer. 3635 * 3636 * @param sz 3637 * Number of stored packets. 3638 */ 3639 #define RTE_ETH_TX_BUFFER_SIZE(sz) \ 3640 (sizeof(struct rte_eth_dev_tx_buffer) + (sz) * sizeof(struct rte_mbuf *)) 3641 3642 /** 3643 * Initialize default values for buffered transmitting 3644 * 3645 * @param buffer 3646 * Tx buffer to be initialized. 3647 * @param size 3648 * Buffer size 3649 * @return 3650 * 0 if no error 3651 */ 3652 int 3653 rte_eth_tx_buffer_init(struct rte_eth_dev_tx_buffer *buffer, uint16_t size); 3654 3655 /** 3656 * Configure a callback for buffered packets which cannot be sent 3657 * 3658 * Register a specific callback to be called when an attempt is made to send 3659 * all packets buffered on an Ethernet port, but not all packets can 3660 * successfully be sent. The callback registered here will be called only 3661 * from calls to rte_eth_tx_buffer() and rte_eth_tx_buffer_flush() APIs. 3662 * The default callback configured for each queue by default just frees the 3663 * packets back to the calling mempool. If additional behaviour is required, 3664 * for example, to count dropped packets, or to retry transmission of packets 3665 * which cannot be sent, this function should be used to register a suitable 3666 * callback function to implement the desired behaviour. 3667 * The example callback "rte_eth_count_unsent_packet_callback()" is also 3668 * provided as reference. 3669 * 3670 * @param buffer 3671 * The port identifier of the Ethernet device. 3672 * @param callback 3673 * The function to be used as the callback. 3674 * @param userdata 3675 * Arbitrary parameter to be passed to the callback function 3676 * @return 3677 * 0 on success, or -EINVAL if bad parameter 3678 */ 3679 int 3680 rte_eth_tx_buffer_set_err_callback(struct rte_eth_dev_tx_buffer *buffer, 3681 buffer_tx_error_fn callback, void *userdata); 3682 3683 /** 3684 * Callback function for silently dropping unsent buffered packets. 3685 * 3686 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to 3687 * adjust the default behavior when buffered packets cannot be sent. This 3688 * function drops any unsent packets silently and is used by Tx buffered 3689 * operations as default behavior. 3690 * 3691 * NOTE: this function should not be called directly, instead it should be used 3692 * as a callback for packet buffering. 3693 * 3694 * NOTE: when configuring this function as a callback with 3695 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter 3696 * should point to an uint64_t value. 3697 * 3698 * @param pkts 3699 * The previously buffered packets which could not be sent 3700 * @param unsent 3701 * The number of unsent packets in the pkts array 3702 * @param userdata 3703 * Not used 3704 */ 3705 void 3706 rte_eth_tx_buffer_drop_callback(struct rte_mbuf **pkts, uint16_t unsent, 3707 void *userdata); 3708 3709 /** 3710 * Callback function for tracking unsent buffered packets. 3711 * 3712 * This function can be passed to rte_eth_tx_buffer_set_err_callback() to 3713 * adjust the default behavior when buffered packets cannot be sent. This 3714 * function drops any unsent packets, but also updates a user-supplied counter 3715 * to track the overall number of packets dropped. The counter should be an 3716 * uint64_t variable. 3717 * 3718 * NOTE: this function should not be called directly, instead it should be used 3719 * as a callback for packet buffering. 3720 * 3721 * NOTE: when configuring this function as a callback with 3722 * rte_eth_tx_buffer_set_err_callback(), the final, userdata parameter 3723 * should point to an uint64_t value. 3724 * 3725 * @param pkts 3726 * The previously buffered packets which could not be sent 3727 * @param unsent 3728 * The number of unsent packets in the pkts array 3729 * @param userdata 3730 * Pointer to an uint64_t value, which will be incremented by unsent 3731 */ 3732 void 3733 rte_eth_tx_buffer_count_callback(struct rte_mbuf **pkts, uint16_t unsent, 3734 void *userdata); 3735 3736 /** 3737 * Request the driver to free mbufs currently cached by the driver. The 3738 * driver will only free the mbuf if it is no longer in use. It is the 3739 * application's responsibility to ensure rte_eth_tx_buffer_flush(..) is 3740 * called if needed. 3741 * 3742 * @param port_id 3743 * The port identifier of the Ethernet device. 3744 * @param queue_id 3745 * The index of the transmit queue through which output packets must be 3746 * sent. 3747 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 3748 * to rte_eth_dev_configure(). 3749 * @param free_cnt 3750 * Maximum number of packets to free. Use 0 to indicate all possible packets 3751 * should be freed. Note that a packet may be using multiple mbufs. 3752 * @return 3753 * Failure: < 0 3754 * -ENODEV: Invalid interface 3755 * -EIO: device is removed 3756 * -ENOTSUP: Driver does not support function 3757 * Success: >= 0 3758 * 0-n: Number of packets freed. More packets may still remain in ring that 3759 * are in use. 3760 */ 3761 int 3762 rte_eth_tx_done_cleanup(uint16_t port_id, uint16_t queue_id, uint32_t free_cnt); 3763 3764 /** 3765 * Subtypes for MACsec offload event (@ref RTE_ETH_EVENT_MACSEC) 3766 * raised by Ethernet device. 3767 */ 3768 enum rte_eth_event_macsec_subtype { 3769 /** Notifies unknown MACsec subevent. */ 3770 RTE_ETH_SUBEVENT_MACSEC_UNKNOWN, 3771 /** 3772 * Subevent of RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR sectag validation events 3773 * Validation check: SecTag.TCI.V = 1 3774 */ 3775 RTE_ETH_SUBEVENT_MACSEC_RX_SECTAG_V_EQ1, 3776 /** 3777 * Subevent of RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR sectag validation events 3778 * Validation check: SecTag.TCI.E = 0 && SecTag.TCI.C = 1 3779 */ 3780 RTE_ETH_SUBEVENT_MACSEC_RX_SECTAG_E_EQ0_C_EQ1, 3781 /** 3782 * Subevent of RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR sectag validation events 3783 * Validation check: SecTag.SL >= 'd48 3784 */ 3785 RTE_ETH_SUBEVENT_MACSEC_RX_SECTAG_SL_GTE48, 3786 /** 3787 * Subevent of RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR sectag validation events 3788 * Validation check: SecTag.TCI.ES = 1 && SecTag.TCI.SC = 1 3789 */ 3790 RTE_ETH_SUBEVENT_MACSEC_RX_SECTAG_ES_EQ1_SC_EQ1, 3791 /** 3792 * Subevent of RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR sectag validation events 3793 * Validation check: SecTag.TCI.SC = 1 && SecTag.TCI.SCB = 1 3794 */ 3795 RTE_ETH_SUBEVENT_MACSEC_RX_SECTAG_SC_EQ1_SCB_EQ1, 3796 }; 3797 3798 /** 3799 * Event types for MACsec offload event (@ref RTE_ETH_EVENT_MACSEC) 3800 * raised by eth device. 3801 */ 3802 enum rte_eth_event_macsec_type { 3803 /** Notifies unknown MACsec event. */ 3804 RTE_ETH_EVENT_MACSEC_UNKNOWN, 3805 /** Notifies Sectag validation failure events. */ 3806 RTE_ETH_EVENT_MACSEC_SECTAG_VAL_ERR, 3807 /** Notifies Rx SA hard expiry events. */ 3808 RTE_ETH_EVENT_MACSEC_RX_SA_PN_HARD_EXP, 3809 /** Notifies Rx SA soft expiry events. */ 3810 RTE_ETH_EVENT_MACSEC_RX_SA_PN_SOFT_EXP, 3811 /** Notifies Tx SA hard expiry events. */ 3812 RTE_ETH_EVENT_MACSEC_TX_SA_PN_HARD_EXP, 3813 /** Notifies Tx SA soft events. */ 3814 RTE_ETH_EVENT_MACSEC_TX_SA_PN_SOFT_EXP, 3815 /** Notifies Invalid SA event. */ 3816 RTE_ETH_EVENT_MACSEC_SA_NOT_VALID, 3817 }; 3818 3819 /** 3820 * Descriptor for @ref RTE_ETH_EVENT_MACSEC event. 3821 * Used by ethdev to send extra information of the MACsec offload event. 3822 */ 3823 struct rte_eth_event_macsec_desc { 3824 /** Type of RTE_ETH_EVENT_MACSEC_* event. */ 3825 enum rte_eth_event_macsec_type type; 3826 /** Type of RTE_ETH_SUBEVENT_MACSEC_* subevent. */ 3827 enum rte_eth_event_macsec_subtype subtype; 3828 /** 3829 * Event specific metadata. 3830 * 3831 * For the following events, *userdata* registered 3832 * with the *rte_security_session* would be returned 3833 * as metadata. 3834 * 3835 * @see struct rte_security_session_conf 3836 */ 3837 uint64_t metadata; 3838 }; 3839 3840 /** 3841 * Subtypes for IPsec offload event(@ref RTE_ETH_EVENT_IPSEC) raised by 3842 * eth device. 3843 */ 3844 enum rte_eth_event_ipsec_subtype { 3845 /** Unknown event type */ 3846 RTE_ETH_EVENT_IPSEC_UNKNOWN = 0, 3847 /** Sequence number overflow */ 3848 RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW, 3849 /** Soft time expiry of SA */ 3850 RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY, 3851 /** 3852 * Soft byte expiry of SA determined by 3853 * @ref rte_security_ipsec_lifetime::bytes_soft_limit 3854 */ 3855 RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY, 3856 /** 3857 * Soft packet expiry of SA determined by 3858 * @ref rte_security_ipsec_lifetime::packets_soft_limit 3859 */ 3860 RTE_ETH_EVENT_IPSEC_SA_PKT_EXPIRY, 3861 /** 3862 * Hard byte expiry of SA determined by 3863 * @ref rte_security_ipsec_lifetime::bytes_hard_limit 3864 */ 3865 RTE_ETH_EVENT_IPSEC_SA_BYTE_HARD_EXPIRY, 3866 /** 3867 * Hard packet expiry of SA determined by 3868 * @ref rte_security_ipsec_lifetime::packets_hard_limit 3869 */ 3870 RTE_ETH_EVENT_IPSEC_SA_PKT_HARD_EXPIRY, 3871 /** Max value of this enum */ 3872 RTE_ETH_EVENT_IPSEC_MAX 3873 }; 3874 3875 /** 3876 * Descriptor for @ref RTE_ETH_EVENT_IPSEC event. Used by eth dev to send extra 3877 * information of the IPsec offload event. 3878 */ 3879 struct rte_eth_event_ipsec_desc { 3880 /** Type of RTE_ETH_EVENT_IPSEC_* event */ 3881 enum rte_eth_event_ipsec_subtype subtype; 3882 /** 3883 * Event specific metadata. 3884 * 3885 * For the following events, *userdata* registered 3886 * with the *rte_security_session* would be returned 3887 * as metadata, 3888 * 3889 * - @ref RTE_ETH_EVENT_IPSEC_ESN_OVERFLOW 3890 * - @ref RTE_ETH_EVENT_IPSEC_SA_TIME_EXPIRY 3891 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_EXPIRY 3892 * - @ref RTE_ETH_EVENT_IPSEC_SA_PKT_EXPIRY 3893 * - @ref RTE_ETH_EVENT_IPSEC_SA_BYTE_HARD_EXPIRY 3894 * - @ref RTE_ETH_EVENT_IPSEC_SA_PKT_HARD_EXPIRY 3895 * 3896 * @see struct rte_security_session_conf 3897 * 3898 */ 3899 uint64_t metadata; 3900 }; 3901 3902 /** 3903 * The eth device event type for interrupt, and maybe others in the future. 3904 */ 3905 enum rte_eth_event_type { 3906 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */ 3907 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */ 3908 /** queue state event (enabled/disabled) */ 3909 RTE_ETH_EVENT_QUEUE_STATE, 3910 /** reset interrupt event, sent to VF on PF reset */ 3911 RTE_ETH_EVENT_INTR_RESET, 3912 RTE_ETH_EVENT_VF_MBOX, /**< message from the VF received by PF */ 3913 RTE_ETH_EVENT_MACSEC, /**< MACsec offload related event */ 3914 RTE_ETH_EVENT_INTR_RMV, /**< device removal event */ 3915 RTE_ETH_EVENT_NEW, /**< port is probed */ 3916 RTE_ETH_EVENT_DESTROY, /**< port is released */ 3917 RTE_ETH_EVENT_IPSEC, /**< IPsec offload related event */ 3918 RTE_ETH_EVENT_FLOW_AGED,/**< New aged-out flows is detected */ 3919 /** 3920 * Number of available Rx descriptors is smaller than the threshold. 3921 * @see rte_eth_rx_avail_thresh_set() 3922 */ 3923 RTE_ETH_EVENT_RX_AVAIL_THRESH, 3924 /** Port recovering from a hardware or firmware error. 3925 * If PMD supports proactive error recovery, 3926 * it should trigger this event to notify application 3927 * that it detected an error and the recovery is being started. 3928 * Upon receiving the event, the application should not invoke any control path API 3929 * (such as rte_eth_dev_configure/rte_eth_dev_stop...) until receiving 3930 * RTE_ETH_EVENT_RECOVERY_SUCCESS or RTE_ETH_EVENT_RECOVERY_FAILED event. 3931 * The PMD will set the data path pointers to dummy functions, 3932 * and re-set the data path pointers to non-dummy functions 3933 * before reporting RTE_ETH_EVENT_RECOVERY_SUCCESS event. 3934 * It means that the application cannot send or receive any packets 3935 * during this period. 3936 * @note Before the PMD reports the recovery result, 3937 * the PMD may report the RTE_ETH_EVENT_ERR_RECOVERING event again, 3938 * because a larger error may occur during the recovery. 3939 */ 3940 RTE_ETH_EVENT_ERR_RECOVERING, 3941 /** Port recovers successfully from the error. 3942 * The PMD already re-configured the port, 3943 * and the effect is the same as a restart operation. 3944 * a) The following operation will be retained: (alphabetically) 3945 * - DCB configuration 3946 * - FEC configuration 3947 * - Flow control configuration 3948 * - LRO configuration 3949 * - LSC configuration 3950 * - MTU 3951 * - MAC address (default and those supplied by MAC address array) 3952 * - Promiscuous and allmulticast mode 3953 * - PTP configuration 3954 * - Queue (Rx/Tx) settings 3955 * - Queue statistics mappings 3956 * - RSS configuration by rte_eth_dev_rss_xxx() family 3957 * - Rx checksum configuration 3958 * - Rx interrupt settings 3959 * - Traffic management configuration 3960 * - VLAN configuration (including filtering, tpid, strip, pvid) 3961 * - VMDq configuration 3962 * b) The following configuration maybe retained 3963 * or not depending on the device capabilities: 3964 * - flow rules 3965 * (@see RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP) 3966 * - shared flow objects 3967 * (@see RTE_ETH_DEV_CAPA_FLOW_SHARED_OBJECT_KEEP) 3968 * c) Any other configuration will not be stored 3969 * and will need to be re-configured. 3970 */ 3971 RTE_ETH_EVENT_RECOVERY_SUCCESS, 3972 /** Port recovery failed. 3973 * It means that the port should not be usable anymore. 3974 * The application should close the port. 3975 */ 3976 RTE_ETH_EVENT_RECOVERY_FAILED, 3977 RTE_ETH_EVENT_MAX /**< max value of this enum */ 3978 }; 3979 3980 /** User application callback to be registered for interrupts. */ 3981 typedef int (*rte_eth_dev_cb_fn)(uint16_t port_id, 3982 enum rte_eth_event_type event, void *cb_arg, void *ret_param); 3983 3984 /** 3985 * Register a callback function for port event. 3986 * 3987 * @param port_id 3988 * Port ID. 3989 * RTE_ETH_ALL means register the event for all port ids. 3990 * @param event 3991 * Event interested. 3992 * @param cb_fn 3993 * User supplied callback function to be called. 3994 * @param cb_arg 3995 * Pointer to the parameters for the registered callback. 3996 * 3997 * @return 3998 * - On success, zero. 3999 * - On failure, a negative value. 4000 */ 4001 int rte_eth_dev_callback_register(uint16_t port_id, 4002 enum rte_eth_event_type event, 4003 rte_eth_dev_cb_fn cb_fn, void *cb_arg); 4004 4005 /** 4006 * Unregister a callback function for port event. 4007 * 4008 * @param port_id 4009 * Port ID. 4010 * RTE_ETH_ALL means unregister the event for all port ids. 4011 * @param event 4012 * Event interested. 4013 * @param cb_fn 4014 * User supplied callback function to be called. 4015 * @param cb_arg 4016 * Pointer to the parameters for the registered callback. -1 means to 4017 * remove all for the same callback address and same event. 4018 * 4019 * @return 4020 * - On success, zero. 4021 * - On failure, a negative value. 4022 */ 4023 int rte_eth_dev_callback_unregister(uint16_t port_id, 4024 enum rte_eth_event_type event, 4025 rte_eth_dev_cb_fn cb_fn, void *cb_arg); 4026 4027 /** 4028 * When there is no Rx packet coming in Rx Queue for a long time, we can 4029 * sleep lcore related to Rx Queue for power saving, and enable Rx interrupt 4030 * to be triggered when Rx packet arrives. 4031 * 4032 * The rte_eth_dev_rx_intr_enable() function enables Rx queue 4033 * interrupt on specific Rx queue of a port. 4034 * 4035 * @param port_id 4036 * The port identifier of the Ethernet device. 4037 * @param queue_id 4038 * The index of the receive queue from which to retrieve input packets. 4039 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 4040 * to rte_eth_dev_configure(). 4041 * @return 4042 * - (0) if successful. 4043 * - (-ENOTSUP) if underlying hardware OR driver doesn't support 4044 * that operation. 4045 * - (-ENODEV) if *port_id* invalid. 4046 * - (-EIO) if device is removed. 4047 */ 4048 int rte_eth_dev_rx_intr_enable(uint16_t port_id, uint16_t queue_id); 4049 4050 /** 4051 * When lcore wakes up from Rx interrupt indicating packet coming, disable Rx 4052 * interrupt and returns to polling mode. 4053 * 4054 * The rte_eth_dev_rx_intr_disable() function disables Rx queue 4055 * interrupt on specific Rx queue of a port. 4056 * 4057 * @param port_id 4058 * The port identifier of the Ethernet device. 4059 * @param queue_id 4060 * The index of the receive queue from which to retrieve input packets. 4061 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 4062 * to rte_eth_dev_configure(). 4063 * @return 4064 * - (0) if successful. 4065 * - (-ENOTSUP) if underlying hardware OR driver doesn't support 4066 * that operation. 4067 * - (-ENODEV) if *port_id* invalid. 4068 * - (-EIO) if device is removed. 4069 */ 4070 int rte_eth_dev_rx_intr_disable(uint16_t port_id, uint16_t queue_id); 4071 4072 /** 4073 * Rx Interrupt control per port. 4074 * 4075 * @param port_id 4076 * The port identifier of the Ethernet device. 4077 * @param epfd 4078 * Epoll instance fd which the intr vector associated to. 4079 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance. 4080 * @param op 4081 * The operation be performed for the vector. 4082 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}. 4083 * @param data 4084 * User raw data. 4085 * @return 4086 * - On success, zero. 4087 * - On failure, a negative value. 4088 */ 4089 int rte_eth_dev_rx_intr_ctl(uint16_t port_id, int epfd, int op, void *data); 4090 4091 /** 4092 * Rx Interrupt control per queue. 4093 * 4094 * @param port_id 4095 * The port identifier of the Ethernet device. 4096 * @param queue_id 4097 * The index of the receive queue from which to retrieve input packets. 4098 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 4099 * to rte_eth_dev_configure(). 4100 * @param epfd 4101 * Epoll instance fd which the intr vector associated to. 4102 * Using RTE_EPOLL_PER_THREAD allows to use per thread epoll instance. 4103 * @param op 4104 * The operation be performed for the vector. 4105 * Operation type of {RTE_INTR_EVENT_ADD, RTE_INTR_EVENT_DEL}. 4106 * @param data 4107 * User raw data. 4108 * @return 4109 * - On success, zero. 4110 * - On failure, a negative value. 4111 */ 4112 int rte_eth_dev_rx_intr_ctl_q(uint16_t port_id, uint16_t queue_id, 4113 int epfd, int op, void *data); 4114 4115 /** 4116 * Get interrupt fd per Rx queue. 4117 * 4118 * @param port_id 4119 * The port identifier of the Ethernet device. 4120 * @param queue_id 4121 * The index of the receive queue from which to retrieve input packets. 4122 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 4123 * to rte_eth_dev_configure(). 4124 * @return 4125 * - (>=0) the interrupt fd associated to the requested Rx queue if 4126 * successful. 4127 * - (-1) on error. 4128 */ 4129 int 4130 rte_eth_dev_rx_intr_ctl_q_get_fd(uint16_t port_id, uint16_t queue_id); 4131 4132 /** 4133 * Turn on the LED on the Ethernet device. 4134 * This function turns on the LED on the Ethernet device. 4135 * 4136 * @param port_id 4137 * The port identifier of the Ethernet device. 4138 * @return 4139 * - (0) if successful. 4140 * - (-ENOTSUP) if underlying hardware OR driver doesn't support 4141 * that operation. 4142 * - (-ENODEV) if *port_id* invalid. 4143 * - (-EIO) if device is removed. 4144 */ 4145 int rte_eth_led_on(uint16_t port_id); 4146 4147 /** 4148 * Turn off the LED on the Ethernet device. 4149 * This function turns off the LED on the Ethernet device. 4150 * 4151 * @param port_id 4152 * The port identifier of the Ethernet device. 4153 * @return 4154 * - (0) if successful. 4155 * - (-ENOTSUP) if underlying hardware OR driver doesn't support 4156 * that operation. 4157 * - (-ENODEV) if *port_id* invalid. 4158 * - (-EIO) if device is removed. 4159 */ 4160 int rte_eth_led_off(uint16_t port_id); 4161 4162 /** 4163 * @warning 4164 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 4165 * 4166 * Get Forward Error Correction(FEC) capability. 4167 * 4168 * @param port_id 4169 * The port identifier of the Ethernet device. 4170 * @param speed_fec_capa 4171 * speed_fec_capa is out only with per-speed capabilities. 4172 * If set to NULL, the function returns the required number 4173 * of required array entries. 4174 * @param num 4175 * a number of elements in an speed_fec_capa array. 4176 * 4177 * @return 4178 * - A non-negative value lower or equal to num: success. The return value 4179 * is the number of entries filled in the fec capa array. 4180 * - A non-negative value higher than num: error, the given fec capa array 4181 * is too small. The return value corresponds to the num that should 4182 * be given to succeed. The entries in fec capa array are not valid and 4183 * shall not be used by the caller. 4184 * - (-ENOTSUP) if underlying hardware OR driver doesn't support. 4185 * that operation. 4186 * - (-EIO) if device is removed. 4187 * - (-ENODEV) if *port_id* invalid. 4188 * - (-EINVAL) if *num* or *speed_fec_capa* invalid 4189 */ 4190 __rte_experimental 4191 int rte_eth_fec_get_capability(uint16_t port_id, 4192 struct rte_eth_fec_capa *speed_fec_capa, 4193 unsigned int num); 4194 4195 /** 4196 * @warning 4197 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 4198 * 4199 * Get current Forward Error Correction(FEC) mode. 4200 * If link is down and AUTO is enabled, AUTO is returned, otherwise, 4201 * configured FEC mode is returned. 4202 * If link is up, current FEC mode is returned. 4203 * 4204 * @param port_id 4205 * The port identifier of the Ethernet device. 4206 * @param fec_capa 4207 * A bitmask of enabled FEC modes. If AUTO bit is set, other 4208 * bits specify FEC modes which may be negotiated. If AUTO 4209 * bit is clear, specify FEC modes to be used (only one valid 4210 * mode per speed may be set). 4211 * @return 4212 * - (0) if successful. 4213 * - (-ENOTSUP) if underlying hardware OR driver doesn't support. 4214 * that operation. 4215 * - (-EIO) if device is removed. 4216 * - (-ENODEV) if *port_id* invalid. 4217 */ 4218 __rte_experimental 4219 int rte_eth_fec_get(uint16_t port_id, uint32_t *fec_capa); 4220 4221 /** 4222 * @warning 4223 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 4224 * 4225 * Set Forward Error Correction(FEC) mode. 4226 * 4227 * @param port_id 4228 * The port identifier of the Ethernet device. 4229 * @param fec_capa 4230 * A bitmask of allowed FEC modes. If AUTO bit is set, other 4231 * bits specify FEC modes which may be negotiated. If AUTO 4232 * bit is clear, specify FEC modes to be used (only one valid 4233 * mode per speed may be set). 4234 * @return 4235 * - (0) if successful. 4236 * - (-EINVAL) if the FEC mode is not valid. 4237 * - (-ENOTSUP) if underlying hardware OR driver doesn't support. 4238 * - (-EIO) if device is removed. 4239 * - (-ENODEV) if *port_id* invalid. 4240 */ 4241 __rte_experimental 4242 int rte_eth_fec_set(uint16_t port_id, uint32_t fec_capa); 4243 4244 /** 4245 * Get current status of the Ethernet link flow control for Ethernet device 4246 * 4247 * @param port_id 4248 * The port identifier of the Ethernet device. 4249 * @param fc_conf 4250 * The pointer to the structure where to store the flow control parameters. 4251 * @return 4252 * - (0) if successful. 4253 * - (-ENOTSUP) if hardware doesn't support flow control. 4254 * - (-ENODEV) if *port_id* invalid. 4255 * - (-EIO) if device is removed. 4256 * - (-EINVAL) if bad parameter. 4257 */ 4258 int rte_eth_dev_flow_ctrl_get(uint16_t port_id, 4259 struct rte_eth_fc_conf *fc_conf); 4260 4261 /** 4262 * Configure the Ethernet link flow control for Ethernet device 4263 * 4264 * @param port_id 4265 * The port identifier of the Ethernet device. 4266 * @param fc_conf 4267 * The pointer to the structure of the flow control parameters. 4268 * @return 4269 * - (0) if successful. 4270 * - (-ENOTSUP) if hardware doesn't support flow control mode. 4271 * - (-ENODEV) if *port_id* invalid. 4272 * - (-EINVAL) if bad parameter 4273 * - (-EIO) if flow control setup failure or device is removed. 4274 */ 4275 int rte_eth_dev_flow_ctrl_set(uint16_t port_id, 4276 struct rte_eth_fc_conf *fc_conf); 4277 4278 /** 4279 * Configure the Ethernet priority flow control under DCB environment 4280 * for Ethernet device. 4281 * 4282 * @param port_id 4283 * The port identifier of the Ethernet device. 4284 * @param pfc_conf 4285 * The pointer to the structure of the priority flow control parameters. 4286 * @return 4287 * - (0) if successful. 4288 * - (-ENOTSUP) if hardware doesn't support priority flow control mode. 4289 * - (-ENODEV) if *port_id* invalid. 4290 * - (-EINVAL) if bad parameter 4291 * - (-EIO) if flow control setup failure or device is removed. 4292 */ 4293 int rte_eth_dev_priority_flow_ctrl_set(uint16_t port_id, 4294 struct rte_eth_pfc_conf *pfc_conf); 4295 4296 /** 4297 * Add a MAC address to the set used for filtering incoming packets. 4298 * 4299 * @param port_id 4300 * The port identifier of the Ethernet device. 4301 * @param mac_addr 4302 * The MAC address to add. 4303 * @param pool 4304 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is 4305 * not enabled, this should be set to 0. 4306 * @return 4307 * - (0) if successfully added or *mac_addr* was already added. 4308 * - (-ENOTSUP) if hardware doesn't support this feature. 4309 * - (-ENODEV) if *port* is invalid. 4310 * - (-EIO) if device is removed. 4311 * - (-ENOSPC) if no more MAC addresses can be added. 4312 * - (-EINVAL) if MAC address is invalid. 4313 */ 4314 int rte_eth_dev_mac_addr_add(uint16_t port_id, struct rte_ether_addr *mac_addr, 4315 uint32_t pool); 4316 4317 /** 4318 * @warning 4319 * @b EXPERIMENTAL: this API may change without prior notice. 4320 * 4321 * Retrieve the information for queue based PFC. 4322 * 4323 * @param port_id 4324 * The port identifier of the Ethernet device. 4325 * @param pfc_queue_info 4326 * A pointer to a structure of type *rte_eth_pfc_queue_info* to be filled with 4327 * the information about queue based PFC. 4328 * @return 4329 * - (0) if successful. 4330 * - (-ENOTSUP) if support for priority_flow_ctrl_queue_info_get does not exist. 4331 * - (-ENODEV) if *port_id* invalid. 4332 * - (-EINVAL) if bad parameter. 4333 */ 4334 __rte_experimental 4335 int rte_eth_dev_priority_flow_ctrl_queue_info_get(uint16_t port_id, 4336 struct rte_eth_pfc_queue_info *pfc_queue_info); 4337 4338 /** 4339 * @warning 4340 * @b EXPERIMENTAL: this API may change without prior notice. 4341 * 4342 * Configure the queue based priority flow control for a given queue 4343 * for Ethernet device. 4344 * 4345 * @note When an ethdev port switches to queue based PFC mode, the 4346 * unconfigured queues shall be configured by the driver with 4347 * default values such as lower priority value for TC etc. 4348 * 4349 * @param port_id 4350 * The port identifier of the Ethernet device. 4351 * @param pfc_queue_conf 4352 * The pointer to the structure of the priority flow control parameters 4353 * for the queue. 4354 * @return 4355 * - (0) if successful. 4356 * - (-ENOTSUP) if hardware doesn't support queue based PFC mode. 4357 * - (-ENODEV) if *port_id* invalid. 4358 * - (-EINVAL) if bad parameter 4359 * - (-EIO) if flow control setup queue failure 4360 */ 4361 __rte_experimental 4362 int rte_eth_dev_priority_flow_ctrl_queue_configure(uint16_t port_id, 4363 struct rte_eth_pfc_queue_conf *pfc_queue_conf); 4364 4365 /** 4366 * Remove a MAC address from the internal array of addresses. 4367 * 4368 * @param port_id 4369 * The port identifier of the Ethernet device. 4370 * @param mac_addr 4371 * MAC address to remove. 4372 * @return 4373 * - (0) if successful, or *mac_addr* didn't exist. 4374 * - (-ENOTSUP) if hardware doesn't support. 4375 * - (-ENODEV) if *port* invalid. 4376 * - (-EADDRINUSE) if attempting to remove the default MAC address. 4377 * - (-EINVAL) if MAC address is invalid. 4378 */ 4379 int rte_eth_dev_mac_addr_remove(uint16_t port_id, 4380 struct rte_ether_addr *mac_addr); 4381 4382 /** 4383 * Set the default MAC address. 4384 * 4385 * @param port_id 4386 * The port identifier of the Ethernet device. 4387 * @param mac_addr 4388 * New default MAC address. 4389 * @return 4390 * - (0) if successful, or *mac_addr* didn't exist. 4391 * - (-ENOTSUP) if hardware doesn't support. 4392 * - (-ENODEV) if *port* invalid. 4393 * - (-EINVAL) if MAC address is invalid. 4394 */ 4395 int rte_eth_dev_default_mac_addr_set(uint16_t port_id, 4396 struct rte_ether_addr *mac_addr); 4397 4398 /** 4399 * Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device. 4400 * 4401 * @param port_id 4402 * The port identifier of the Ethernet device. 4403 * @param reta_conf 4404 * RETA to update. 4405 * @param reta_size 4406 * Redirection table size. The table size can be queried by 4407 * rte_eth_dev_info_get(). 4408 * @return 4409 * - (0) if successful. 4410 * - (-ENODEV) if *port_id* is invalid. 4411 * - (-ENOTSUP) if hardware doesn't support. 4412 * - (-EINVAL) if bad parameter. 4413 * - (-EIO) if device is removed. 4414 */ 4415 int rte_eth_dev_rss_reta_update(uint16_t port_id, 4416 struct rte_eth_rss_reta_entry64 *reta_conf, 4417 uint16_t reta_size); 4418 4419 /** 4420 * Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device. 4421 * 4422 * @param port_id 4423 * The port identifier of the Ethernet device. 4424 * @param reta_conf 4425 * RETA to query. For each requested reta entry, corresponding bit 4426 * in mask must be set. 4427 * @param reta_size 4428 * Redirection table size. The table size can be queried by 4429 * rte_eth_dev_info_get(). 4430 * @return 4431 * - (0) if successful. 4432 * - (-ENODEV) if *port_id* is invalid. 4433 * - (-ENOTSUP) if hardware doesn't support. 4434 * - (-EINVAL) if bad parameter. 4435 * - (-EIO) if device is removed. 4436 */ 4437 int rte_eth_dev_rss_reta_query(uint16_t port_id, 4438 struct rte_eth_rss_reta_entry64 *reta_conf, 4439 uint16_t reta_size); 4440 4441 /** 4442 * Updates unicast hash table for receiving packet with the given destination 4443 * MAC address, and the packet is routed to all VFs for which the Rx mode is 4444 * accept packets that match the unicast hash table. 4445 * 4446 * @param port_id 4447 * The port identifier of the Ethernet device. 4448 * @param addr 4449 * Unicast MAC address. 4450 * @param on 4451 * 1 - Set an unicast hash bit for receiving packets with the MAC address. 4452 * 0 - Clear an unicast hash bit. 4453 * @return 4454 * - (0) if successful. 4455 * - (-ENOTSUP) if hardware doesn't support. 4456 * - (-ENODEV) if *port_id* invalid. 4457 * - (-EIO) if device is removed. 4458 * - (-EINVAL) if bad parameter. 4459 */ 4460 int rte_eth_dev_uc_hash_table_set(uint16_t port_id, struct rte_ether_addr *addr, 4461 uint8_t on); 4462 4463 /** 4464 * Updates all unicast hash bitmaps for receiving packet with any Unicast 4465 * Ethernet MAC addresses,the packet is routed to all VFs for which the Rx 4466 * mode is accept packets that match the unicast hash table. 4467 * 4468 * @param port_id 4469 * The port identifier of the Ethernet device. 4470 * @param on 4471 * 1 - Set all unicast hash bitmaps for receiving all the Ethernet 4472 * MAC addresses 4473 * 0 - Clear all unicast hash bitmaps 4474 * @return 4475 * - (0) if successful. 4476 * - (-ENOTSUP) if hardware doesn't support. 4477 * - (-ENODEV) if *port_id* invalid. 4478 * - (-EIO) if device is removed. 4479 * - (-EINVAL) if bad parameter. 4480 */ 4481 int rte_eth_dev_uc_all_hash_table_set(uint16_t port_id, uint8_t on); 4482 4483 /** 4484 * Set the rate limitation for a queue on an Ethernet device. 4485 * 4486 * @param port_id 4487 * The port identifier of the Ethernet device. 4488 * @param queue_idx 4489 * The queue ID. 4490 * @param tx_rate 4491 * The Tx rate in Mbps. Allocated from the total port link speed. 4492 * @return 4493 * - (0) if successful. 4494 * - (-ENOTSUP) if hardware doesn't support this feature. 4495 * - (-ENODEV) if *port_id* invalid. 4496 * - (-EIO) if device is removed. 4497 * - (-EINVAL) if bad parameter. 4498 */ 4499 int rte_eth_set_queue_rate_limit(uint16_t port_id, uint16_t queue_idx, 4500 uint32_t tx_rate); 4501 4502 /** 4503 * Configuration of Receive Side Scaling hash computation of Ethernet device. 4504 * 4505 * @param port_id 4506 * The port identifier of the Ethernet device. 4507 * @param rss_conf 4508 * The new configuration to use for RSS hash computation on the port. 4509 * @return 4510 * - (0) if successful. 4511 * - (-ENODEV) if port identifier is invalid. 4512 * - (-EIO) if device is removed. 4513 * - (-ENOTSUP) if hardware doesn't support. 4514 * - (-EINVAL) if bad parameter. 4515 */ 4516 int rte_eth_dev_rss_hash_update(uint16_t port_id, 4517 struct rte_eth_rss_conf *rss_conf); 4518 4519 /** 4520 * Retrieve current configuration of Receive Side Scaling hash computation 4521 * of Ethernet device. 4522 * 4523 * @param port_id 4524 * The port identifier of the Ethernet device. 4525 * @param rss_conf 4526 * Where to store the current RSS hash configuration of the Ethernet device. 4527 * @return 4528 * - (0) if successful. 4529 * - (-ENODEV) if port identifier is invalid. 4530 * - (-EIO) if device is removed. 4531 * - (-ENOTSUP) if hardware doesn't support RSS. 4532 * - (-EINVAL) if bad parameter. 4533 */ 4534 int 4535 rte_eth_dev_rss_hash_conf_get(uint16_t port_id, 4536 struct rte_eth_rss_conf *rss_conf); 4537 4538 /** 4539 * Add UDP tunneling port for a type of tunnel. 4540 * 4541 * Some NICs may require such configuration to properly parse a tunnel 4542 * with any standard or custom UDP port. 4543 * The packets with this UDP port will be parsed for this type of tunnel. 4544 * The device parser will also check the rest of the tunnel headers 4545 * before classifying the packet. 4546 * 4547 * With some devices, this API will affect packet classification, i.e.: 4548 * - mbuf.packet_type reported on Rx 4549 * - rte_flow rules with tunnel items 4550 * 4551 * @param port_id 4552 * The port identifier of the Ethernet device. 4553 * @param tunnel_udp 4554 * UDP tunneling configuration. 4555 * 4556 * @return 4557 * - (0) if successful. 4558 * - (-ENODEV) if port identifier is invalid. 4559 * - (-EIO) if device is removed. 4560 * - (-ENOTSUP) if hardware doesn't support tunnel type. 4561 */ 4562 int 4563 rte_eth_dev_udp_tunnel_port_add(uint16_t port_id, 4564 struct rte_eth_udp_tunnel *tunnel_udp); 4565 4566 /** 4567 * Delete UDP tunneling port for a type of tunnel. 4568 * 4569 * The packets with this UDP port will not be classified as this type of tunnel 4570 * anymore if the device use such mapping for tunnel packet classification. 4571 * 4572 * @see rte_eth_dev_udp_tunnel_port_add 4573 * 4574 * @param port_id 4575 * The port identifier of the Ethernet device. 4576 * @param tunnel_udp 4577 * UDP tunneling configuration. 4578 * 4579 * @return 4580 * - (0) if successful. 4581 * - (-ENODEV) if port identifier is invalid. 4582 * - (-EIO) if device is removed. 4583 * - (-ENOTSUP) if hardware doesn't support tunnel type. 4584 */ 4585 int 4586 rte_eth_dev_udp_tunnel_port_delete(uint16_t port_id, 4587 struct rte_eth_udp_tunnel *tunnel_udp); 4588 4589 /** 4590 * Get DCB information on an Ethernet device. 4591 * 4592 * @param port_id 4593 * The port identifier of the Ethernet device. 4594 * @param dcb_info 4595 * DCB information. 4596 * @return 4597 * - (0) if successful. 4598 * - (-ENODEV) if port identifier is invalid. 4599 * - (-EIO) if device is removed. 4600 * - (-ENOTSUP) if hardware doesn't support. 4601 * - (-EINVAL) if bad parameter. 4602 */ 4603 int rte_eth_dev_get_dcb_info(uint16_t port_id, 4604 struct rte_eth_dcb_info *dcb_info); 4605 4606 struct rte_eth_rxtx_callback; 4607 4608 /** 4609 * Add a callback to be called on packet Rx on a given port and queue. 4610 * 4611 * This API configures a function to be called for each burst of 4612 * packets received on a given NIC port queue. The return value is a pointer 4613 * that can be used to later remove the callback using 4614 * rte_eth_remove_rx_callback(). 4615 * 4616 * Multiple functions are called in the order that they are added. 4617 * 4618 * @param port_id 4619 * The port identifier of the Ethernet device. 4620 * @param queue_id 4621 * The queue on the Ethernet device on which the callback is to be added. 4622 * @param fn 4623 * The callback function 4624 * @param user_param 4625 * A generic pointer parameter which will be passed to each invocation of the 4626 * callback function on this port and queue. Inter-thread synchronization 4627 * of any user data changes is the responsibility of the user. 4628 * 4629 * @return 4630 * NULL on error. 4631 * On success, a pointer value which can later be used to remove the callback. 4632 */ 4633 const struct rte_eth_rxtx_callback * 4634 rte_eth_add_rx_callback(uint16_t port_id, uint16_t queue_id, 4635 rte_rx_callback_fn fn, void *user_param); 4636 4637 /** 4638 * Add a callback that must be called first on packet Rx on a given port 4639 * and queue. 4640 * 4641 * This API configures a first function to be called for each burst of 4642 * packets received on a given NIC port queue. The return value is a pointer 4643 * that can be used to later remove the callback using 4644 * rte_eth_remove_rx_callback(). 4645 * 4646 * Multiple functions are called in the order that they are added. 4647 * 4648 * @param port_id 4649 * The port identifier of the Ethernet device. 4650 * @param queue_id 4651 * The queue on the Ethernet device on which the callback is to be added. 4652 * @param fn 4653 * The callback function 4654 * @param user_param 4655 * A generic pointer parameter which will be passed to each invocation of the 4656 * callback function on this port and queue. Inter-thread synchronization 4657 * of any user data changes is the responsibility of the user. 4658 * 4659 * @return 4660 * NULL on error. 4661 * On success, a pointer value which can later be used to remove the callback. 4662 */ 4663 const struct rte_eth_rxtx_callback * 4664 rte_eth_add_first_rx_callback(uint16_t port_id, uint16_t queue_id, 4665 rte_rx_callback_fn fn, void *user_param); 4666 4667 /** 4668 * Add a callback to be called on packet Tx on a given port and queue. 4669 * 4670 * This API configures a function to be called for each burst of 4671 * packets sent on a given NIC port queue. The return value is a pointer 4672 * that can be used to later remove the callback using 4673 * rte_eth_remove_tx_callback(). 4674 * 4675 * Multiple functions are called in the order that they are added. 4676 * 4677 * @param port_id 4678 * The port identifier of the Ethernet device. 4679 * @param queue_id 4680 * The queue on the Ethernet device on which the callback is to be added. 4681 * @param fn 4682 * The callback function 4683 * @param user_param 4684 * A generic pointer parameter which will be passed to each invocation of the 4685 * callback function on this port and queue. Inter-thread synchronization 4686 * of any user data changes is the responsibility of the user. 4687 * 4688 * @return 4689 * NULL on error. 4690 * On success, a pointer value which can later be used to remove the callback. 4691 */ 4692 const struct rte_eth_rxtx_callback * 4693 rte_eth_add_tx_callback(uint16_t port_id, uint16_t queue_id, 4694 rte_tx_callback_fn fn, void *user_param); 4695 4696 /** 4697 * Remove an Rx packet callback from a given port and queue. 4698 * 4699 * This function is used to removed callbacks that were added to a NIC port 4700 * queue using rte_eth_add_rx_callback(). 4701 * 4702 * Note: the callback is removed from the callback list but it isn't freed 4703 * since the it may still be in use. The memory for the callback can be 4704 * subsequently freed back by the application by calling rte_free(): 4705 * 4706 * - Immediately - if the port is stopped, or the user knows that no 4707 * callbacks are in flight e.g. if called from the thread doing Rx/Tx 4708 * on that queue. 4709 * 4710 * - After a short delay - where the delay is sufficient to allow any 4711 * in-flight callbacks to complete. Alternately, the RCU mechanism can be 4712 * used to detect when data plane threads have ceased referencing the 4713 * callback memory. 4714 * 4715 * @param port_id 4716 * The port identifier of the Ethernet device. 4717 * @param queue_id 4718 * The queue on the Ethernet device from which the callback is to be removed. 4719 * @param user_cb 4720 * User supplied callback created via rte_eth_add_rx_callback(). 4721 * 4722 * @return 4723 * - 0: Success. Callback was removed. 4724 * - -ENODEV: If *port_id* is invalid. 4725 * - -ENOTSUP: Callback support is not available. 4726 * - -EINVAL: The queue_id is out of range, or the callback 4727 * is NULL or not found for the port/queue. 4728 */ 4729 int rte_eth_remove_rx_callback(uint16_t port_id, uint16_t queue_id, 4730 const struct rte_eth_rxtx_callback *user_cb); 4731 4732 /** 4733 * Remove a Tx packet callback from a given port and queue. 4734 * 4735 * This function is used to removed callbacks that were added to a NIC port 4736 * queue using rte_eth_add_tx_callback(). 4737 * 4738 * Note: the callback is removed from the callback list but it isn't freed 4739 * since the it may still be in use. The memory for the callback can be 4740 * subsequently freed back by the application by calling rte_free(): 4741 * 4742 * - Immediately - if the port is stopped, or the user knows that no 4743 * callbacks are in flight e.g. if called from the thread doing Rx/Tx 4744 * on that queue. 4745 * 4746 * - After a short delay - where the delay is sufficient to allow any 4747 * in-flight callbacks to complete. Alternately, the RCU mechanism can be 4748 * used to detect when data plane threads have ceased referencing the 4749 * callback memory. 4750 * 4751 * @param port_id 4752 * The port identifier of the Ethernet device. 4753 * @param queue_id 4754 * The queue on the Ethernet device from which the callback is to be removed. 4755 * @param user_cb 4756 * User supplied callback created via rte_eth_add_tx_callback(). 4757 * 4758 * @return 4759 * - 0: Success. Callback was removed. 4760 * - -ENODEV: If *port_id* is invalid. 4761 * - -ENOTSUP: Callback support is not available. 4762 * - -EINVAL: The queue_id is out of range, or the callback 4763 * is NULL or not found for the port/queue. 4764 */ 4765 int rte_eth_remove_tx_callback(uint16_t port_id, uint16_t queue_id, 4766 const struct rte_eth_rxtx_callback *user_cb); 4767 4768 /** 4769 * Retrieve information about given port's Rx queue. 4770 * 4771 * @param port_id 4772 * The port identifier of the Ethernet device. 4773 * @param queue_id 4774 * The Rx queue on the Ethernet device for which information 4775 * will be retrieved. 4776 * @param qinfo 4777 * A pointer to a structure of type *rte_eth_rxq_info_info* to be filled with 4778 * the information of the Ethernet device. 4779 * 4780 * @return 4781 * - 0: Success 4782 * - -ENODEV: If *port_id* is invalid. 4783 * - -ENOTSUP: routine is not supported by the device PMD. 4784 * - -EINVAL: The queue_id is out of range, or the queue 4785 * is hairpin queue. 4786 */ 4787 int rte_eth_rx_queue_info_get(uint16_t port_id, uint16_t queue_id, 4788 struct rte_eth_rxq_info *qinfo); 4789 4790 /** 4791 * Retrieve information about given port's Tx queue. 4792 * 4793 * @param port_id 4794 * The port identifier of the Ethernet device. 4795 * @param queue_id 4796 * The Tx queue on the Ethernet device for which information 4797 * will be retrieved. 4798 * @param qinfo 4799 * A pointer to a structure of type *rte_eth_txq_info_info* to be filled with 4800 * the information of the Ethernet device. 4801 * 4802 * @return 4803 * - 0: Success 4804 * - -ENODEV: If *port_id* is invalid. 4805 * - -ENOTSUP: routine is not supported by the device PMD. 4806 * - -EINVAL: The queue_id is out of range, or the queue 4807 * is hairpin queue. 4808 */ 4809 int rte_eth_tx_queue_info_get(uint16_t port_id, uint16_t queue_id, 4810 struct rte_eth_txq_info *qinfo); 4811 4812 /** 4813 * Retrieve information about the Rx packet burst mode. 4814 * 4815 * @param port_id 4816 * The port identifier of the Ethernet device. 4817 * @param queue_id 4818 * The Rx queue on the Ethernet device for which information 4819 * will be retrieved. 4820 * @param mode 4821 * A pointer to a structure of type *rte_eth_burst_mode* to be filled 4822 * with the information of the packet burst mode. 4823 * 4824 * @return 4825 * - 0: Success 4826 * - -ENODEV: If *port_id* is invalid. 4827 * - -ENOTSUP: routine is not supported by the device PMD. 4828 * - -EINVAL: The queue_id is out of range. 4829 */ 4830 int rte_eth_rx_burst_mode_get(uint16_t port_id, uint16_t queue_id, 4831 struct rte_eth_burst_mode *mode); 4832 4833 /** 4834 * Retrieve information about the Tx packet burst mode. 4835 * 4836 * @param port_id 4837 * The port identifier of the Ethernet device. 4838 * @param queue_id 4839 * The Tx queue on the Ethernet device for which information 4840 * will be retrieved. 4841 * @param mode 4842 * A pointer to a structure of type *rte_eth_burst_mode* to be filled 4843 * with the information of the packet burst mode. 4844 * 4845 * @return 4846 * - 0: Success 4847 * - -ENODEV: If *port_id* is invalid. 4848 * - -ENOTSUP: routine is not supported by the device PMD. 4849 * - -EINVAL: The queue_id is out of range. 4850 */ 4851 int rte_eth_tx_burst_mode_get(uint16_t port_id, uint16_t queue_id, 4852 struct rte_eth_burst_mode *mode); 4853 4854 /** 4855 * @warning 4856 * @b EXPERIMENTAL: this API may change without prior notice. 4857 * 4858 * Retrieve the monitor condition for a given receive queue. 4859 * 4860 * @param port_id 4861 * The port identifier of the Ethernet device. 4862 * @param queue_id 4863 * The Rx queue on the Ethernet device for which information 4864 * will be retrieved. 4865 * @param pmc 4866 * The pointer to power-optimized monitoring condition structure. 4867 * 4868 * @return 4869 * - 0: Success. 4870 * -ENOTSUP: Operation not supported. 4871 * -EINVAL: Invalid parameters. 4872 * -ENODEV: Invalid port ID. 4873 */ 4874 __rte_experimental 4875 int rte_eth_get_monitor_addr(uint16_t port_id, uint16_t queue_id, 4876 struct rte_power_monitor_cond *pmc); 4877 4878 /** 4879 * Retrieve device registers and register attributes (number of registers and 4880 * register size) 4881 * 4882 * @param port_id 4883 * The port identifier of the Ethernet device. 4884 * @param info 4885 * Pointer to rte_dev_reg_info structure to fill in. If info->data is 4886 * NULL the function fills in the width and length fields. If non-NULL 4887 * the registers are put into the buffer pointed at by the data field. 4888 * @return 4889 * - (0) if successful. 4890 * - (-ENOTSUP) if hardware doesn't support. 4891 * - (-EINVAL) if bad parameter. 4892 * - (-ENODEV) if *port_id* invalid. 4893 * - (-EIO) if device is removed. 4894 * - others depends on the specific operations implementation. 4895 */ 4896 int rte_eth_dev_get_reg_info(uint16_t port_id, struct rte_dev_reg_info *info); 4897 4898 /** 4899 * Retrieve size of device EEPROM 4900 * 4901 * @param port_id 4902 * The port identifier of the Ethernet device. 4903 * @return 4904 * - (>=0) EEPROM size if successful. 4905 * - (-ENOTSUP) if hardware doesn't support. 4906 * - (-ENODEV) if *port_id* invalid. 4907 * - (-EIO) if device is removed. 4908 * - others depends on the specific operations implementation. 4909 */ 4910 int rte_eth_dev_get_eeprom_length(uint16_t port_id); 4911 4912 /** 4913 * Retrieve EEPROM and EEPROM attribute 4914 * 4915 * @param port_id 4916 * The port identifier of the Ethernet device. 4917 * @param info 4918 * The template includes buffer for return EEPROM data and 4919 * EEPROM attributes to be filled. 4920 * @return 4921 * - (0) if successful. 4922 * - (-ENOTSUP) if hardware doesn't support. 4923 * - (-EINVAL) if bad parameter. 4924 * - (-ENODEV) if *port_id* invalid. 4925 * - (-EIO) if device is removed. 4926 * - others depends on the specific operations implementation. 4927 */ 4928 int rte_eth_dev_get_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info); 4929 4930 /** 4931 * Program EEPROM with provided data 4932 * 4933 * @param port_id 4934 * The port identifier of the Ethernet device. 4935 * @param info 4936 * The template includes EEPROM data for programming and 4937 * EEPROM attributes to be filled 4938 * @return 4939 * - (0) if successful. 4940 * - (-ENOTSUP) if hardware doesn't support. 4941 * - (-ENODEV) if *port_id* invalid. 4942 * - (-EINVAL) if bad parameter. 4943 * - (-EIO) if device is removed. 4944 * - others depends on the specific operations implementation. 4945 */ 4946 int rte_eth_dev_set_eeprom(uint16_t port_id, struct rte_dev_eeprom_info *info); 4947 4948 /** 4949 * @warning 4950 * @b EXPERIMENTAL: this API may change without prior notice. 4951 * 4952 * Retrieve the type and size of plugin module EEPROM 4953 * 4954 * @param port_id 4955 * The port identifier of the Ethernet device. 4956 * @param modinfo 4957 * The type and size of plugin module EEPROM. 4958 * @return 4959 * - (0) if successful. 4960 * - (-ENOTSUP) if hardware doesn't support. 4961 * - (-ENODEV) if *port_id* invalid. 4962 * - (-EINVAL) if bad parameter. 4963 * - (-EIO) if device is removed. 4964 * - others depends on the specific operations implementation. 4965 */ 4966 __rte_experimental 4967 int 4968 rte_eth_dev_get_module_info(uint16_t port_id, 4969 struct rte_eth_dev_module_info *modinfo); 4970 4971 /** 4972 * @warning 4973 * @b EXPERIMENTAL: this API may change without prior notice. 4974 * 4975 * Retrieve the data of plugin module EEPROM 4976 * 4977 * @param port_id 4978 * The port identifier of the Ethernet device. 4979 * @param info 4980 * The template includes the plugin module EEPROM attributes, and the 4981 * buffer for return plugin module EEPROM data. 4982 * @return 4983 * - (0) if successful. 4984 * - (-ENOTSUP) if hardware doesn't support. 4985 * - (-EINVAL) if bad parameter. 4986 * - (-ENODEV) if *port_id* invalid. 4987 * - (-EIO) if device is removed. 4988 * - others depends on the specific operations implementation. 4989 */ 4990 __rte_experimental 4991 int 4992 rte_eth_dev_get_module_eeprom(uint16_t port_id, 4993 struct rte_dev_eeprom_info *info); 4994 4995 /** 4996 * Set the list of multicast addresses to filter on an Ethernet device. 4997 * 4998 * @param port_id 4999 * The port identifier of the Ethernet device. 5000 * @param mc_addr_set 5001 * The array of multicast addresses to set. Equal to NULL when the function 5002 * is invoked to flush the set of filtered addresses. 5003 * @param nb_mc_addr 5004 * The number of multicast addresses in the *mc_addr_set* array. Equal to 0 5005 * when the function is invoked to flush the set of filtered addresses. 5006 * @return 5007 * - (0) if successful. 5008 * - (-ENODEV) if *port_id* invalid. 5009 * - (-EIO) if device is removed. 5010 * - (-ENOTSUP) if PMD of *port_id* doesn't support multicast filtering. 5011 * - (-ENOSPC) if *port_id* has not enough multicast filtering resources. 5012 * - (-EINVAL) if bad parameter. 5013 */ 5014 int rte_eth_dev_set_mc_addr_list(uint16_t port_id, 5015 struct rte_ether_addr *mc_addr_set, 5016 uint32_t nb_mc_addr); 5017 5018 /** 5019 * Enable IEEE1588/802.1AS timestamping for an Ethernet device. 5020 * 5021 * @param port_id 5022 * The port identifier of the Ethernet device. 5023 * 5024 * @return 5025 * - 0: Success. 5026 * - -ENODEV: The port ID is invalid. 5027 * - -EIO: if device is removed. 5028 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5029 */ 5030 int rte_eth_timesync_enable(uint16_t port_id); 5031 5032 /** 5033 * Disable IEEE1588/802.1AS timestamping for an Ethernet device. 5034 * 5035 * @param port_id 5036 * The port identifier of the Ethernet device. 5037 * 5038 * @return 5039 * - 0: Success. 5040 * - -ENODEV: The port ID is invalid. 5041 * - -EIO: if device is removed. 5042 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5043 */ 5044 int rte_eth_timesync_disable(uint16_t port_id); 5045 5046 /** 5047 * Read an IEEE1588/802.1AS Rx timestamp from an Ethernet device. 5048 * 5049 * @param port_id 5050 * The port identifier of the Ethernet device. 5051 * @param timestamp 5052 * Pointer to the timestamp struct. 5053 * @param flags 5054 * Device specific flags. Used to pass the Rx timesync register index to 5055 * i40e. Unused in igb/ixgbe, pass 0 instead. 5056 * 5057 * @return 5058 * - 0: Success. 5059 * - -EINVAL: No timestamp is available. 5060 * - -ENODEV: The port ID is invalid. 5061 * - -EIO: if device is removed. 5062 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5063 */ 5064 int rte_eth_timesync_read_rx_timestamp(uint16_t port_id, 5065 struct timespec *timestamp, uint32_t flags); 5066 5067 /** 5068 * Read an IEEE1588/802.1AS Tx timestamp from an Ethernet device. 5069 * 5070 * @param port_id 5071 * The port identifier of the Ethernet device. 5072 * @param timestamp 5073 * Pointer to the timestamp struct. 5074 * 5075 * @return 5076 * - 0: Success. 5077 * - -EINVAL: No timestamp is available. 5078 * - -ENODEV: The port ID is invalid. 5079 * - -EIO: if device is removed. 5080 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5081 */ 5082 int rte_eth_timesync_read_tx_timestamp(uint16_t port_id, 5083 struct timespec *timestamp); 5084 5085 /** 5086 * Adjust the timesync clock on an Ethernet device. 5087 * 5088 * This is usually used in conjunction with other Ethdev timesync functions to 5089 * synchronize the device time using the IEEE1588/802.1AS protocol. 5090 * 5091 * @param port_id 5092 * The port identifier of the Ethernet device. 5093 * @param delta 5094 * The adjustment in nanoseconds. 5095 * 5096 * @return 5097 * - 0: Success. 5098 * - -ENODEV: The port ID is invalid. 5099 * - -EIO: if device is removed. 5100 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5101 */ 5102 int rte_eth_timesync_adjust_time(uint16_t port_id, int64_t delta); 5103 5104 /** 5105 * Read the time from the timesync clock on an Ethernet device. 5106 * 5107 * This is usually used in conjunction with other Ethdev timesync functions to 5108 * synchronize the device time using the IEEE1588/802.1AS protocol. 5109 * 5110 * @param port_id 5111 * The port identifier of the Ethernet device. 5112 * @param time 5113 * Pointer to the timespec struct that holds the time. 5114 * 5115 * @return 5116 * - 0: Success. 5117 * - -EINVAL: Bad parameter. 5118 */ 5119 int rte_eth_timesync_read_time(uint16_t port_id, struct timespec *time); 5120 5121 /** 5122 * Set the time of the timesync clock on an Ethernet device. 5123 * 5124 * This is usually used in conjunction with other Ethdev timesync functions to 5125 * synchronize the device time using the IEEE1588/802.1AS protocol. 5126 * 5127 * @param port_id 5128 * The port identifier of the Ethernet device. 5129 * @param time 5130 * Pointer to the timespec struct that holds the time. 5131 * 5132 * @return 5133 * - 0: Success. 5134 * - -EINVAL: No timestamp is available. 5135 * - -ENODEV: The port ID is invalid. 5136 * - -EIO: if device is removed. 5137 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5138 */ 5139 int rte_eth_timesync_write_time(uint16_t port_id, const struct timespec *time); 5140 5141 /** 5142 * @warning 5143 * @b EXPERIMENTAL: this API may change without prior notice. 5144 * 5145 * Read the current clock counter of an Ethernet device 5146 * 5147 * This returns the current raw clock value of an Ethernet device. It is 5148 * a raw amount of ticks, with no given time reference. 5149 * The value returned here is from the same clock than the one 5150 * filling timestamp field of Rx packets when using hardware timestamp 5151 * offload. Therefore it can be used to compute a precise conversion of 5152 * the device clock to the real time. 5153 * 5154 * E.g, a simple heuristic to derivate the frequency would be: 5155 * uint64_t start, end; 5156 * rte_eth_read_clock(port, start); 5157 * rte_delay_ms(100); 5158 * rte_eth_read_clock(port, end); 5159 * double freq = (end - start) * 10; 5160 * 5161 * Compute a common reference with: 5162 * uint64_t base_time_sec = current_time(); 5163 * uint64_t base_clock; 5164 * rte_eth_read_clock(port, base_clock); 5165 * 5166 * Then, convert the raw mbuf timestamp with: 5167 * base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq; 5168 * 5169 * This simple example will not provide a very good accuracy. One must 5170 * at least measure multiple times the frequency and do a regression. 5171 * To avoid deviation from the system time, the common reference can 5172 * be repeated from time to time. The integer division can also be 5173 * converted by a multiplication and a shift for better performance. 5174 * 5175 * @param port_id 5176 * The port identifier of the Ethernet device. 5177 * @param clock 5178 * Pointer to the uint64_t that holds the raw clock value. 5179 * 5180 * @return 5181 * - 0: Success. 5182 * - -ENODEV: The port ID is invalid. 5183 * - -ENOTSUP: The function is not supported by the Ethernet driver. 5184 * - -EINVAL: if bad parameter. 5185 */ 5186 __rte_experimental 5187 int 5188 rte_eth_read_clock(uint16_t port_id, uint64_t *clock); 5189 5190 /** 5191 * Get the port ID from device name. The device name should be specified 5192 * as below: 5193 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:2:00.0 5194 * - SoC device name, for example- fsl-gmac0 5195 * - vdev dpdk name, for example- net_[pcap0|null0|tap0] 5196 * 5197 * @param name 5198 * pci address or name of the device 5199 * @param port_id 5200 * pointer to port identifier of the device 5201 * @return 5202 * - (0) if successful and port_id is filled. 5203 * - (-ENODEV or -EINVAL) on failure. 5204 */ 5205 int 5206 rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id); 5207 5208 /** 5209 * Get the device name from port ID. The device name is specified as below: 5210 * - PCIe address (Domain:Bus:Device.Function), for example- 0000:02:00.0 5211 * - SoC device name, for example- fsl-gmac0 5212 * - vdev dpdk name, for example- net_[pcap0|null0|tun0|tap0] 5213 * 5214 * @param port_id 5215 * Port identifier of the device. 5216 * @param name 5217 * Buffer of size RTE_ETH_NAME_MAX_LEN to store the name. 5218 * @return 5219 * - (0) if successful. 5220 * - (-ENODEV) if *port_id* is invalid. 5221 * - (-EINVAL) on failure. 5222 */ 5223 int 5224 rte_eth_dev_get_name_by_port(uint16_t port_id, char *name); 5225 5226 /** 5227 * Check that numbers of Rx and Tx descriptors satisfy descriptors limits from 5228 * the Ethernet device information, otherwise adjust them to boundaries. 5229 * 5230 * @param port_id 5231 * The port identifier of the Ethernet device. 5232 * @param nb_rx_desc 5233 * A pointer to a uint16_t where the number of receive 5234 * descriptors stored. 5235 * @param nb_tx_desc 5236 * A pointer to a uint16_t where the number of transmit 5237 * descriptors stored. 5238 * @return 5239 * - (0) if successful. 5240 * - (-ENOTSUP, -ENODEV or -EINVAL) on failure. 5241 */ 5242 int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id, 5243 uint16_t *nb_rx_desc, 5244 uint16_t *nb_tx_desc); 5245 5246 /** 5247 * Test if a port supports specific mempool ops. 5248 * 5249 * @param port_id 5250 * Port identifier of the Ethernet device. 5251 * @param [in] pool 5252 * The name of the pool operations to test. 5253 * @return 5254 * - 0: best mempool ops choice for this port. 5255 * - 1: mempool ops are supported for this port. 5256 * - -ENOTSUP: mempool ops not supported for this port. 5257 * - -ENODEV: Invalid port Identifier. 5258 * - -EINVAL: Pool param is null. 5259 */ 5260 int 5261 rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool); 5262 5263 /** 5264 * Get the security context for the Ethernet device. 5265 * 5266 * @param port_id 5267 * Port identifier of the Ethernet device 5268 * @return 5269 * - NULL on error. 5270 * - pointer to security context on success. 5271 */ 5272 void * 5273 rte_eth_dev_get_sec_ctx(uint16_t port_id); 5274 5275 /** 5276 * @warning 5277 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5278 * 5279 * Query the device hairpin capabilities. 5280 * 5281 * @param port_id 5282 * The port identifier of the Ethernet device. 5283 * @param cap 5284 * Pointer to a structure that will hold the hairpin capabilities. 5285 * @return 5286 * - (0) if successful. 5287 * - (-ENOTSUP) if hardware doesn't support. 5288 * - (-EINVAL) if bad parameter. 5289 */ 5290 __rte_experimental 5291 int rte_eth_dev_hairpin_capability_get(uint16_t port_id, 5292 struct rte_eth_hairpin_cap *cap); 5293 5294 /** 5295 * @warning 5296 * @b EXPERIMENTAL: this structure may change without prior notice. 5297 * 5298 * Ethernet device representor ID range entry 5299 */ 5300 struct rte_eth_representor_range { 5301 enum rte_eth_representor_type type; /**< Representor type */ 5302 int controller; /**< Controller index */ 5303 int pf; /**< Physical function index */ 5304 __extension__ 5305 union { 5306 int vf; /**< VF start index */ 5307 int sf; /**< SF start index */ 5308 }; 5309 uint32_t id_base; /**< Representor ID start index */ 5310 uint32_t id_end; /**< Representor ID end index */ 5311 char name[RTE_DEV_NAME_MAX_LEN]; /**< Representor name */ 5312 }; 5313 5314 /** 5315 * @warning 5316 * @b EXPERIMENTAL: this structure may change without prior notice. 5317 * 5318 * Ethernet device representor information 5319 */ 5320 struct rte_eth_representor_info { 5321 uint16_t controller; /**< Controller ID of caller device. */ 5322 uint16_t pf; /**< Physical function ID of caller device. */ 5323 uint32_t nb_ranges_alloc; /**< Size of the ranges array. */ 5324 uint32_t nb_ranges; /**< Number of initialized ranges. */ 5325 struct rte_eth_representor_range ranges[];/**< Representor ID range. */ 5326 }; 5327 5328 /** 5329 * Retrieve the representor info of the device. 5330 * 5331 * Get device representor info to be able to calculate a unique 5332 * representor ID. @see rte_eth_representor_id_get helper. 5333 * 5334 * @param port_id 5335 * The port identifier of the device. 5336 * @param info 5337 * A pointer to a representor info structure. 5338 * NULL to return number of range entries and allocate memory 5339 * for next call to store detail. 5340 * The number of ranges that were written into this structure 5341 * will be placed into its nb_ranges field. This number cannot be 5342 * larger than the nb_ranges_alloc that by the user before calling 5343 * this function. It can be smaller than the value returned by the 5344 * function, however. 5345 * @return 5346 * - (-ENOTSUP) if operation is not supported. 5347 * - (-ENODEV) if *port_id* invalid. 5348 * - (-EIO) if device is removed. 5349 * - (>=0) number of available representor range entries. 5350 */ 5351 __rte_experimental 5352 int rte_eth_representor_info_get(uint16_t port_id, 5353 struct rte_eth_representor_info *info); 5354 5355 /** The NIC is able to deliver flag (if set) with packets to the PMD. */ 5356 #define RTE_ETH_RX_METADATA_USER_FLAG RTE_BIT64(0) 5357 5358 /** The NIC is able to deliver mark ID with packets to the PMD. */ 5359 #define RTE_ETH_RX_METADATA_USER_MARK RTE_BIT64(1) 5360 5361 /** The NIC is able to deliver tunnel ID with packets to the PMD. */ 5362 #define RTE_ETH_RX_METADATA_TUNNEL_ID RTE_BIT64(2) 5363 5364 /** 5365 * Negotiate the NIC's ability to deliver specific kinds of metadata to the PMD. 5366 * 5367 * Invoke this API before the first rte_eth_dev_configure() invocation 5368 * to let the PMD make preparations that are inconvenient to do later. 5369 * 5370 * The negotiation process is as follows: 5371 * 5372 * - the application requests features intending to use at least some of them; 5373 * - the PMD responds with the guaranteed subset of the requested feature set; 5374 * - the application can retry negotiation with another set of features; 5375 * - the application can pass zero to clear the negotiation result; 5376 * - the last negotiated result takes effect upon 5377 * the ethdev configure and start. 5378 * 5379 * @note 5380 * The PMD is supposed to first consider enabling the requested feature set 5381 * in its entirety. Only if it fails to do so, does it have the right to 5382 * respond with a smaller set of the originally requested features. 5383 * 5384 * @note 5385 * Return code (-ENOTSUP) does not necessarily mean that the requested 5386 * features are unsupported. In this case, the application should just 5387 * assume that these features can be used without prior negotiations. 5388 * 5389 * @param port_id 5390 * Port (ethdev) identifier 5391 * 5392 * @param[inout] features 5393 * Feature selection buffer 5394 * 5395 * @return 5396 * - (-EBUSY) if the port can't handle this in its current state; 5397 * - (-ENOTSUP) if the method itself is not supported by the PMD; 5398 * - (-ENODEV) if *port_id* is invalid; 5399 * - (-EINVAL) if *features* is NULL; 5400 * - (-EIO) if the device is removed; 5401 * - (0) on success 5402 */ 5403 int rte_eth_rx_metadata_negotiate(uint16_t port_id, uint64_t *features); 5404 5405 /** Flag to offload IP reassembly for IPv4 packets. */ 5406 #define RTE_ETH_DEV_REASSEMBLY_F_IPV4 (RTE_BIT32(0)) 5407 /** Flag to offload IP reassembly for IPv6 packets. */ 5408 #define RTE_ETH_DEV_REASSEMBLY_F_IPV6 (RTE_BIT32(1)) 5409 5410 /** 5411 * A structure used to get/set IP reassembly configuration. It is also used 5412 * to get the maximum capability values that a PMD can support. 5413 * 5414 * If rte_eth_ip_reassembly_capability_get() returns 0, IP reassembly can be 5415 * enabled using rte_eth_ip_reassembly_conf_set() and params values lower than 5416 * capability params can be set in the PMD. 5417 */ 5418 struct rte_eth_ip_reassembly_params { 5419 /** Maximum time in ms which PMD can wait for other fragments. */ 5420 uint32_t timeout_ms; 5421 /** Maximum number of fragments that can be reassembled. */ 5422 uint16_t max_frags; 5423 /** 5424 * Flags to enable reassembly of packet types - 5425 * RTE_ETH_DEV_REASSEMBLY_F_xxx. 5426 */ 5427 uint16_t flags; 5428 }; 5429 5430 /** 5431 * @warning 5432 * @b EXPERIMENTAL: this API may change without prior notice 5433 * 5434 * Get IP reassembly capabilities supported by the PMD. This is the first API 5435 * to be called for enabling the IP reassembly offload feature. PMD will return 5436 * the maximum values of parameters that PMD can support and user can call 5437 * rte_eth_ip_reassembly_conf_set() with param values lower than capability. 5438 * 5439 * @param port_id 5440 * The port identifier of the device. 5441 * @param capa 5442 * A pointer to rte_eth_ip_reassembly_params structure. 5443 * @return 5444 * - (-ENOTSUP) if offload configuration is not supported by device. 5445 * - (-ENODEV) if *port_id* invalid. 5446 * - (-EIO) if device is removed. 5447 * - (-EINVAL) if device is not configured or *capa* passed is NULL. 5448 * - (0) on success. 5449 */ 5450 __rte_experimental 5451 int rte_eth_ip_reassembly_capability_get(uint16_t port_id, 5452 struct rte_eth_ip_reassembly_params *capa); 5453 5454 /** 5455 * @warning 5456 * @b EXPERIMENTAL: this API may change without prior notice 5457 * 5458 * Get IP reassembly configuration parameters currently set in PMD. 5459 * The API will return error if the configuration is not already 5460 * set using rte_eth_ip_reassembly_conf_set() before calling this API or if 5461 * the device is not configured. 5462 * 5463 * @param port_id 5464 * The port identifier of the device. 5465 * @param conf 5466 * A pointer to rte_eth_ip_reassembly_params structure. 5467 * @return 5468 * - (-ENOTSUP) if offload configuration is not supported by device. 5469 * - (-ENODEV) if *port_id* invalid. 5470 * - (-EIO) if device is removed. 5471 * - (-EINVAL) if device is not configured or if *conf* passed is NULL or if 5472 * configuration is not set using rte_eth_ip_reassembly_conf_set(). 5473 * - (0) on success. 5474 */ 5475 __rte_experimental 5476 int rte_eth_ip_reassembly_conf_get(uint16_t port_id, 5477 struct rte_eth_ip_reassembly_params *conf); 5478 5479 /** 5480 * @warning 5481 * @b EXPERIMENTAL: this API may change without prior notice 5482 * 5483 * Set IP reassembly configuration parameters if the PMD supports IP reassembly 5484 * offload. User should first call rte_eth_ip_reassembly_capability_get() to 5485 * check the maximum values supported by the PMD before setting the 5486 * configuration. The use of this API is mandatory to enable this feature and 5487 * should be called before rte_eth_dev_start(). 5488 * 5489 * In datapath, PMD cannot guarantee that IP reassembly is always successful. 5490 * Hence, PMD shall register mbuf dynamic field and dynamic flag using 5491 * rte_eth_ip_reassembly_dynfield_register() to denote incomplete IP reassembly. 5492 * If dynfield is not successfully registered, error will be returned and 5493 * IP reassembly offload cannot be used. 5494 * 5495 * @param port_id 5496 * The port identifier of the device. 5497 * @param conf 5498 * A pointer to rte_eth_ip_reassembly_params structure. 5499 * @return 5500 * - (-ENOTSUP) if offload configuration is not supported by device. 5501 * - (-ENODEV) if *port_id* invalid. 5502 * - (-EIO) if device is removed. 5503 * - (-EINVAL) if device is not configured or if device is already started or 5504 * if *conf* passed is NULL or if mbuf dynfield is not registered 5505 * successfully by the PMD. 5506 * - (0) on success. 5507 */ 5508 __rte_experimental 5509 int rte_eth_ip_reassembly_conf_set(uint16_t port_id, 5510 const struct rte_eth_ip_reassembly_params *conf); 5511 5512 /** 5513 * In case of IP reassembly offload failure, packet will be updated with 5514 * dynamic flag - RTE_MBUF_DYNFLAG_IP_REASSEMBLY_INCOMPLETE_NAME and packets 5515 * will be returned without alteration. 5516 * The application can retrieve the attached fragments using mbuf dynamic field 5517 * RTE_MBUF_DYNFIELD_IP_REASSEMBLY_NAME. 5518 */ 5519 typedef struct { 5520 /** 5521 * Next fragment packet. Application should fetch dynamic field of 5522 * each fragment until a NULL is received and nb_frags is 0. 5523 */ 5524 struct rte_mbuf *next_frag; 5525 /** Time spent(in ms) by HW in waiting for further fragments. */ 5526 uint16_t time_spent; 5527 /** Number of more fragments attached in mbuf dynamic fields. */ 5528 uint16_t nb_frags; 5529 } rte_eth_ip_reassembly_dynfield_t; 5530 5531 /** 5532 * @warning 5533 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5534 * 5535 * Dump private info from device to a file. Provided data and the order depends 5536 * on the PMD. 5537 * 5538 * @param port_id 5539 * The port identifier of the Ethernet device. 5540 * @param file 5541 * A pointer to a file for output. 5542 * @return 5543 * - (0) on success. 5544 * - (-ENODEV) if *port_id* is invalid. 5545 * - (-EINVAL) if null file. 5546 * - (-ENOTSUP) if the device does not support this function. 5547 * - (-EIO) if device is removed. 5548 */ 5549 __rte_experimental 5550 int rte_eth_dev_priv_dump(uint16_t port_id, FILE *file); 5551 5552 /** 5553 * @warning 5554 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5555 * 5556 * Dump ethdev Rx descriptor info to a file. 5557 * 5558 * This API is used for debugging, not a dataplane API. 5559 * 5560 * @param port_id 5561 * The port identifier of the Ethernet device. 5562 * @param queue_id 5563 * A Rx queue identifier on this port. 5564 * @param offset 5565 * The offset of the descriptor starting from tail. (0 is the next 5566 * packet to be received by the driver). 5567 * @param num 5568 * The number of the descriptors to dump. 5569 * @param file 5570 * A pointer to a file for output. 5571 * @return 5572 * - On success, zero. 5573 * - On failure, a negative value. 5574 */ 5575 __rte_experimental 5576 int rte_eth_rx_descriptor_dump(uint16_t port_id, uint16_t queue_id, 5577 uint16_t offset, uint16_t num, FILE *file); 5578 5579 /** 5580 * @warning 5581 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5582 * 5583 * Dump ethdev Tx descriptor info to a file. 5584 * 5585 * This API is used for debugging, not a dataplane API. 5586 * 5587 * @param port_id 5588 * The port identifier of the Ethernet device. 5589 * @param queue_id 5590 * A Tx queue identifier on this port. 5591 * @param offset 5592 * The offset of the descriptor starting from tail. (0 is the place where 5593 * the next packet will be send). 5594 * @param num 5595 * The number of the descriptors to dump. 5596 * @param file 5597 * A pointer to a file for output. 5598 * @return 5599 * - On success, zero. 5600 * - On failure, a negative value. 5601 */ 5602 __rte_experimental 5603 int rte_eth_tx_descriptor_dump(uint16_t port_id, uint16_t queue_id, 5604 uint16_t offset, uint16_t num, FILE *file); 5605 5606 5607 /* Congestion management */ 5608 5609 /** Enumerate list of ethdev congestion management objects */ 5610 enum rte_eth_cman_obj { 5611 /** Congestion management based on Rx queue depth */ 5612 RTE_ETH_CMAN_OBJ_RX_QUEUE = RTE_BIT32(0), 5613 /** 5614 * Congestion management based on mempool depth associated with Rx queue 5615 * @see rte_eth_rx_queue_setup() 5616 */ 5617 RTE_ETH_CMAN_OBJ_RX_QUEUE_MEMPOOL = RTE_BIT32(1), 5618 }; 5619 5620 /** 5621 * @warning 5622 * @b EXPERIMENTAL: this structure may change, or be removed, without prior notice 5623 * 5624 * A structure used to retrieve information of ethdev congestion management. 5625 */ 5626 struct rte_eth_cman_info { 5627 /** 5628 * Set of supported congestion management modes 5629 * @see enum rte_cman_mode 5630 */ 5631 uint64_t modes_supported; 5632 /** 5633 * Set of supported congestion management objects 5634 * @see enum rte_eth_cman_obj 5635 */ 5636 uint64_t objs_supported; 5637 /** 5638 * Reserved for future fields. Always returned as 0 when 5639 * rte_eth_cman_info_get() is invoked 5640 */ 5641 uint8_t rsvd[8]; 5642 }; 5643 5644 /** 5645 * @warning 5646 * @b EXPERIMENTAL: this structure may change, or be removed, without prior notice 5647 * 5648 * A structure used to configure the ethdev congestion management. 5649 */ 5650 struct rte_eth_cman_config { 5651 /** Congestion management object */ 5652 enum rte_eth_cman_obj obj; 5653 /** Congestion management mode */ 5654 enum rte_cman_mode mode; 5655 union { 5656 /** 5657 * Rx queue to configure congestion management. 5658 * 5659 * Valid when object is RTE_ETH_CMAN_OBJ_RX_QUEUE or 5660 * RTE_ETH_CMAN_OBJ_RX_QUEUE_MEMPOOL. 5661 */ 5662 uint16_t rx_queue; 5663 /** 5664 * Reserved for future fields. 5665 * It must be set to 0 when rte_eth_cman_config_set() is invoked 5666 * and will be returned as 0 when rte_eth_cman_config_get() is 5667 * invoked. 5668 */ 5669 uint8_t rsvd_obj_params[4]; 5670 } obj_param; 5671 union { 5672 /** 5673 * RED configuration parameters. 5674 * 5675 * Valid when mode is RTE_CMAN_RED. 5676 */ 5677 struct rte_cman_red_params red; 5678 /** 5679 * Reserved for future fields. 5680 * It must be set to 0 when rte_eth_cman_config_set() is invoked 5681 * and will be returned as 0 when rte_eth_cman_config_get() is 5682 * invoked. 5683 */ 5684 uint8_t rsvd_mode_params[4]; 5685 } mode_param; 5686 }; 5687 5688 /** 5689 * @warning 5690 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5691 * 5692 * Retrieve the information for ethdev congestion management 5693 * 5694 * @param port_id 5695 * The port identifier of the Ethernet device. 5696 * @param info 5697 * A pointer to a structure of type *rte_eth_cman_info* to be filled with 5698 * the information about congestion management. 5699 * @return 5700 * - (0) if successful. 5701 * - (-ENOTSUP) if support for cman_info_get does not exist. 5702 * - (-ENODEV) if *port_id* invalid. 5703 * - (-EINVAL) if bad parameter. 5704 */ 5705 __rte_experimental 5706 int rte_eth_cman_info_get(uint16_t port_id, struct rte_eth_cman_info *info); 5707 5708 /** 5709 * @warning 5710 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5711 * 5712 * Initialize the ethdev congestion management configuration structure with default values. 5713 * 5714 * @param port_id 5715 * The port identifier of the Ethernet device. 5716 * @param config 5717 * A pointer to a structure of type *rte_eth_cman_config* to be initialized 5718 * with default value. 5719 * @return 5720 * - (0) if successful. 5721 * - (-ENOTSUP) if support for cman_config_init does not exist. 5722 * - (-ENODEV) if *port_id* invalid. 5723 * - (-EINVAL) if bad parameter. 5724 */ 5725 __rte_experimental 5726 int rte_eth_cman_config_init(uint16_t port_id, struct rte_eth_cman_config *config); 5727 5728 /** 5729 * @warning 5730 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5731 * 5732 * Configure ethdev congestion management 5733 * 5734 * @param port_id 5735 * The port identifier of the Ethernet device. 5736 * @param config 5737 * A pointer to a structure of type *rte_eth_cman_config* to be configured. 5738 * @return 5739 * - (0) if successful. 5740 * - (-ENOTSUP) if support for cman_config_set does not exist. 5741 * - (-ENODEV) if *port_id* invalid. 5742 * - (-EINVAL) if bad parameter. 5743 */ 5744 __rte_experimental 5745 int rte_eth_cman_config_set(uint16_t port_id, const struct rte_eth_cman_config *config); 5746 5747 /** 5748 * @warning 5749 * @b EXPERIMENTAL: this API may change, or be removed, without prior notice 5750 * 5751 * Retrieve the applied ethdev congestion management parameters for the given port. 5752 * 5753 * @param port_id 5754 * The port identifier of the Ethernet device. 5755 * @param config 5756 * A pointer to a structure of type *rte_eth_cman_config* to retrieve 5757 * congestion management parameters for the given object. 5758 * Application must fill all parameters except mode_param parameter in 5759 * struct rte_eth_cman_config. 5760 * 5761 * @return 5762 * - (0) if successful. 5763 * - (-ENOTSUP) if support for cman_config_get does not exist. 5764 * - (-ENODEV) if *port_id* invalid. 5765 * - (-EINVAL) if bad parameter. 5766 */ 5767 __rte_experimental 5768 int rte_eth_cman_config_get(uint16_t port_id, struct rte_eth_cman_config *config); 5769 5770 #include <rte_ethdev_core.h> 5771 5772 /** 5773 * @internal 5774 * Helper routine for rte_eth_rx_burst(). 5775 * Should be called at exit from PMD's rte_eth_rx_bulk implementation. 5776 * Does necessary post-processing - invokes Rx callbacks if any, etc. 5777 * 5778 * @param port_id 5779 * The port identifier of the Ethernet device. 5780 * @param queue_id 5781 * The index of the receive queue from which to retrieve input packets. 5782 * @param rx_pkts 5783 * The address of an array of pointers to *rte_mbuf* structures that 5784 * have been retrieved from the device. 5785 * @param nb_rx 5786 * The number of packets that were retrieved from the device. 5787 * @param nb_pkts 5788 * The number of elements in @p rx_pkts array. 5789 * @param opaque 5790 * Opaque pointer of Rx queue callback related data. 5791 * 5792 * @return 5793 * The number of packets effectively supplied to the @p rx_pkts array. 5794 */ 5795 uint16_t rte_eth_call_rx_callbacks(uint16_t port_id, uint16_t queue_id, 5796 struct rte_mbuf **rx_pkts, uint16_t nb_rx, uint16_t nb_pkts, 5797 void *opaque); 5798 5799 /** 5800 * 5801 * Retrieve a burst of input packets from a receive queue of an Ethernet 5802 * device. The retrieved packets are stored in *rte_mbuf* structures whose 5803 * pointers are supplied in the *rx_pkts* array. 5804 * 5805 * The rte_eth_rx_burst() function loops, parsing the Rx ring of the 5806 * receive queue, up to *nb_pkts* packets, and for each completed Rx 5807 * descriptor in the ring, it performs the following operations: 5808 * 5809 * - Initialize the *rte_mbuf* data structure associated with the 5810 * Rx descriptor according to the information provided by the NIC into 5811 * that Rx descriptor. 5812 * 5813 * - Store the *rte_mbuf* data structure into the next entry of the 5814 * *rx_pkts* array. 5815 * 5816 * - Replenish the Rx descriptor with a new *rte_mbuf* buffer 5817 * allocated from the memory pool associated with the receive queue at 5818 * initialization time. 5819 * 5820 * When retrieving an input packet that was scattered by the controller 5821 * into multiple receive descriptors, the rte_eth_rx_burst() function 5822 * appends the associated *rte_mbuf* buffers to the first buffer of the 5823 * packet. 5824 * 5825 * The rte_eth_rx_burst() function returns the number of packets 5826 * actually retrieved, which is the number of *rte_mbuf* data structures 5827 * effectively supplied into the *rx_pkts* array. 5828 * A return value equal to *nb_pkts* indicates that the Rx queue contained 5829 * at least *rx_pkts* packets, and this is likely to signify that other 5830 * received packets remain in the input queue. Applications implementing 5831 * a "retrieve as much received packets as possible" policy can check this 5832 * specific case and keep invoking the rte_eth_rx_burst() function until 5833 * a value less than *nb_pkts* is returned. 5834 * 5835 * This receive method has the following advantages: 5836 * 5837 * - It allows a run-to-completion network stack engine to retrieve and 5838 * to immediately process received packets in a fast burst-oriented 5839 * approach, avoiding the overhead of unnecessary intermediate packet 5840 * queue/dequeue operations. 5841 * 5842 * - Conversely, it also allows an asynchronous-oriented processing 5843 * method to retrieve bursts of received packets and to immediately 5844 * queue them for further parallel processing by another logical core, 5845 * for instance. However, instead of having received packets being 5846 * individually queued by the driver, this approach allows the caller 5847 * of the rte_eth_rx_burst() function to queue a burst of retrieved 5848 * packets at a time and therefore dramatically reduce the cost of 5849 * enqueue/dequeue operations per packet. 5850 * 5851 * - It allows the rte_eth_rx_burst() function of the driver to take 5852 * advantage of burst-oriented hardware features (CPU cache, 5853 * prefetch instructions, and so on) to minimize the number of CPU 5854 * cycles per packet. 5855 * 5856 * To summarize, the proposed receive API enables many 5857 * burst-oriented optimizations in both synchronous and asynchronous 5858 * packet processing environments with no overhead in both cases. 5859 * 5860 * @note 5861 * Some drivers using vector instructions require that *nb_pkts* is 5862 * divisible by 4 or 8, depending on the driver implementation. 5863 * 5864 * The rte_eth_rx_burst() function does not provide any error 5865 * notification to avoid the corresponding overhead. As a hint, the 5866 * upper-level application might check the status of the device link once 5867 * being systematically returned a 0 value for a given number of tries. 5868 * 5869 * @param port_id 5870 * The port identifier of the Ethernet device. 5871 * @param queue_id 5872 * The index of the receive queue from which to retrieve input packets. 5873 * The value must be in the range [0, nb_rx_queue - 1] previously supplied 5874 * to rte_eth_dev_configure(). 5875 * @param rx_pkts 5876 * The address of an array of pointers to *rte_mbuf* structures that 5877 * must be large enough to store *nb_pkts* pointers in it. 5878 * @param nb_pkts 5879 * The maximum number of packets to retrieve. 5880 * The value must be divisible by 8 in order to work with any driver. 5881 * @return 5882 * The number of packets actually retrieved, which is the number 5883 * of pointers to *rte_mbuf* structures effectively supplied to the 5884 * *rx_pkts* array. 5885 */ 5886 static inline uint16_t 5887 rte_eth_rx_burst(uint16_t port_id, uint16_t queue_id, 5888 struct rte_mbuf **rx_pkts, const uint16_t nb_pkts) 5889 { 5890 uint16_t nb_rx; 5891 struct rte_eth_fp_ops *p; 5892 void *qd; 5893 5894 #ifdef RTE_ETHDEV_DEBUG_RX 5895 if (port_id >= RTE_MAX_ETHPORTS || 5896 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 5897 RTE_ETHDEV_LOG(ERR, 5898 "Invalid port_id=%u or queue_id=%u\n", 5899 port_id, queue_id); 5900 return 0; 5901 } 5902 #endif 5903 5904 /* fetch pointer to queue data */ 5905 p = &rte_eth_fp_ops[port_id]; 5906 qd = p->rxq.data[queue_id]; 5907 5908 #ifdef RTE_ETHDEV_DEBUG_RX 5909 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0); 5910 5911 if (qd == NULL) { 5912 RTE_ETHDEV_LOG(ERR, "Invalid Rx queue_id=%u for port_id=%u\n", 5913 queue_id, port_id); 5914 return 0; 5915 } 5916 #endif 5917 5918 nb_rx = p->rx_pkt_burst(qd, rx_pkts, nb_pkts); 5919 5920 #ifdef RTE_ETHDEV_RXTX_CALLBACKS 5921 { 5922 void *cb; 5923 5924 /* __ATOMIC_RELEASE memory order was used when the 5925 * call back was inserted into the list. 5926 * Since there is a clear dependency between loading 5927 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is 5928 * not required. 5929 */ 5930 cb = __atomic_load_n((void **)&p->rxq.clbk[queue_id], 5931 __ATOMIC_RELAXED); 5932 if (unlikely(cb != NULL)) 5933 nb_rx = rte_eth_call_rx_callbacks(port_id, queue_id, 5934 rx_pkts, nb_rx, nb_pkts, cb); 5935 } 5936 #endif 5937 5938 rte_ethdev_trace_rx_burst(port_id, queue_id, (void **)rx_pkts, nb_rx); 5939 return nb_rx; 5940 } 5941 5942 /** 5943 * Get the number of used descriptors of a Rx queue 5944 * 5945 * Since it's a dataplane function, no check is performed on port_id and 5946 * queue_id. The caller must therefore ensure that the port is enabled 5947 * and the queue is configured and running. 5948 * 5949 * @param port_id 5950 * The port identifier of the Ethernet device. 5951 * @param queue_id 5952 * The queue ID on the specific port. 5953 * @return 5954 * The number of used descriptors in the specific queue, or: 5955 * - (-ENODEV) if *port_id* is invalid. 5956 * - (-EINVAL) if *queue_id* is invalid 5957 * - (-ENOTSUP) if the device does not support this function 5958 */ 5959 static inline int 5960 rte_eth_rx_queue_count(uint16_t port_id, uint16_t queue_id) 5961 { 5962 struct rte_eth_fp_ops *p; 5963 void *qd; 5964 5965 #ifdef RTE_ETHDEV_DEBUG_RX 5966 if (port_id >= RTE_MAX_ETHPORTS || 5967 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 5968 RTE_ETHDEV_LOG(ERR, 5969 "Invalid port_id=%u or queue_id=%u\n", 5970 port_id, queue_id); 5971 return -EINVAL; 5972 } 5973 #endif 5974 5975 /* fetch pointer to queue data */ 5976 p = &rte_eth_fp_ops[port_id]; 5977 qd = p->rxq.data[queue_id]; 5978 5979 #ifdef RTE_ETHDEV_DEBUG_RX 5980 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV); 5981 if (qd == NULL) 5982 return -EINVAL; 5983 #endif 5984 5985 if (*p->rx_queue_count == NULL) 5986 return -ENOTSUP; 5987 return (int)(*p->rx_queue_count)(qd); 5988 } 5989 5990 /**@{@name Rx hardware descriptor states 5991 * @see rte_eth_rx_descriptor_status 5992 */ 5993 #define RTE_ETH_RX_DESC_AVAIL 0 /**< Desc available for hw. */ 5994 #define RTE_ETH_RX_DESC_DONE 1 /**< Desc done, filled by hw. */ 5995 #define RTE_ETH_RX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */ 5996 /**@}*/ 5997 5998 /** 5999 * Check the status of a Rx descriptor in the queue 6000 * 6001 * It should be called in a similar context than the Rx function: 6002 * - on a dataplane core 6003 * - not concurrently on the same queue 6004 * 6005 * Since it's a dataplane function, no check is performed on port_id and 6006 * queue_id. The caller must therefore ensure that the port is enabled 6007 * and the queue is configured and running. 6008 * 6009 * Note: accessing to a random descriptor in the ring may trigger cache 6010 * misses and have a performance impact. 6011 * 6012 * @param port_id 6013 * A valid port identifier of the Ethernet device which. 6014 * @param queue_id 6015 * A valid Rx queue identifier on this port. 6016 * @param offset 6017 * The offset of the descriptor starting from tail (0 is the next 6018 * packet to be received by the driver). 6019 * 6020 * @return 6021 * - (RTE_ETH_RX_DESC_AVAIL): Descriptor is available for the hardware to 6022 * receive a packet. 6023 * - (RTE_ETH_RX_DESC_DONE): Descriptor is done, it is filled by hw, but 6024 * not yet processed by the driver (i.e. in the receive queue). 6025 * - (RTE_ETH_RX_DESC_UNAVAIL): Descriptor is unavailable, either hold by 6026 * the driver and not yet returned to hw, or reserved by the hw. 6027 * - (-EINVAL) bad descriptor offset. 6028 * - (-ENOTSUP) if the device does not support this function. 6029 * - (-ENODEV) bad port or queue (only if compiled with debug). 6030 */ 6031 static inline int 6032 rte_eth_rx_descriptor_status(uint16_t port_id, uint16_t queue_id, 6033 uint16_t offset) 6034 { 6035 struct rte_eth_fp_ops *p; 6036 void *qd; 6037 6038 #ifdef RTE_ETHDEV_DEBUG_RX 6039 if (port_id >= RTE_MAX_ETHPORTS || 6040 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 6041 RTE_ETHDEV_LOG(ERR, 6042 "Invalid port_id=%u or queue_id=%u\n", 6043 port_id, queue_id); 6044 return -EINVAL; 6045 } 6046 #endif 6047 6048 /* fetch pointer to queue data */ 6049 p = &rte_eth_fp_ops[port_id]; 6050 qd = p->rxq.data[queue_id]; 6051 6052 #ifdef RTE_ETHDEV_DEBUG_RX 6053 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV); 6054 if (qd == NULL) 6055 return -ENODEV; 6056 #endif 6057 if (*p->rx_descriptor_status == NULL) 6058 return -ENOTSUP; 6059 return (*p->rx_descriptor_status)(qd, offset); 6060 } 6061 6062 /**@{@name Tx hardware descriptor states 6063 * @see rte_eth_tx_descriptor_status 6064 */ 6065 #define RTE_ETH_TX_DESC_FULL 0 /**< Desc filled for hw, waiting xmit. */ 6066 #define RTE_ETH_TX_DESC_DONE 1 /**< Desc done, packet is transmitted. */ 6067 #define RTE_ETH_TX_DESC_UNAVAIL 2 /**< Desc used by driver or hw. */ 6068 /**@}*/ 6069 6070 /** 6071 * Check the status of a Tx descriptor in the queue. 6072 * 6073 * It should be called in a similar context than the Tx function: 6074 * - on a dataplane core 6075 * - not concurrently on the same queue 6076 * 6077 * Since it's a dataplane function, no check is performed on port_id and 6078 * queue_id. The caller must therefore ensure that the port is enabled 6079 * and the queue is configured and running. 6080 * 6081 * Note: accessing to a random descriptor in the ring may trigger cache 6082 * misses and have a performance impact. 6083 * 6084 * @param port_id 6085 * A valid port identifier of the Ethernet device which. 6086 * @param queue_id 6087 * A valid Tx queue identifier on this port. 6088 * @param offset 6089 * The offset of the descriptor starting from tail (0 is the place where 6090 * the next packet will be send). 6091 * 6092 * @return 6093 * - (RTE_ETH_TX_DESC_FULL) Descriptor is being processed by the hw, i.e. 6094 * in the transmit queue. 6095 * - (RTE_ETH_TX_DESC_DONE) Hardware is done with this descriptor, it can 6096 * be reused by the driver. 6097 * - (RTE_ETH_TX_DESC_UNAVAIL): Descriptor is unavailable, reserved by the 6098 * driver or the hardware. 6099 * - (-EINVAL) bad descriptor offset. 6100 * - (-ENOTSUP) if the device does not support this function. 6101 * - (-ENODEV) bad port or queue (only if compiled with debug). 6102 */ 6103 static inline int rte_eth_tx_descriptor_status(uint16_t port_id, 6104 uint16_t queue_id, uint16_t offset) 6105 { 6106 struct rte_eth_fp_ops *p; 6107 void *qd; 6108 6109 #ifdef RTE_ETHDEV_DEBUG_TX 6110 if (port_id >= RTE_MAX_ETHPORTS || 6111 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 6112 RTE_ETHDEV_LOG(ERR, 6113 "Invalid port_id=%u or queue_id=%u\n", 6114 port_id, queue_id); 6115 return -EINVAL; 6116 } 6117 #endif 6118 6119 /* fetch pointer to queue data */ 6120 p = &rte_eth_fp_ops[port_id]; 6121 qd = p->txq.data[queue_id]; 6122 6123 #ifdef RTE_ETHDEV_DEBUG_TX 6124 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV); 6125 if (qd == NULL) 6126 return -ENODEV; 6127 #endif 6128 if (*p->tx_descriptor_status == NULL) 6129 return -ENOTSUP; 6130 return (*p->tx_descriptor_status)(qd, offset); 6131 } 6132 6133 /** 6134 * @internal 6135 * Helper routine for rte_eth_tx_burst(). 6136 * Should be called before entry PMD's rte_eth_tx_bulk implementation. 6137 * Does necessary pre-processing - invokes Tx callbacks if any, etc. 6138 * 6139 * @param port_id 6140 * The port identifier of the Ethernet device. 6141 * @param queue_id 6142 * The index of the transmit queue through which output packets must be 6143 * sent. 6144 * @param tx_pkts 6145 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures 6146 * which contain the output packets. 6147 * @param nb_pkts 6148 * The maximum number of packets to transmit. 6149 * @return 6150 * The number of output packets to transmit. 6151 */ 6152 uint16_t rte_eth_call_tx_callbacks(uint16_t port_id, uint16_t queue_id, 6153 struct rte_mbuf **tx_pkts, uint16_t nb_pkts, void *opaque); 6154 6155 /** 6156 * Send a burst of output packets on a transmit queue of an Ethernet device. 6157 * 6158 * The rte_eth_tx_burst() function is invoked to transmit output packets 6159 * on the output queue *queue_id* of the Ethernet device designated by its 6160 * *port_id*. 6161 * The *nb_pkts* parameter is the number of packets to send which are 6162 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them 6163 * allocated from a pool created with rte_pktmbuf_pool_create(). 6164 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets, 6165 * up to the number of transmit descriptors available in the Tx ring of the 6166 * transmit queue. 6167 * For each packet to send, the rte_eth_tx_burst() function performs 6168 * the following operations: 6169 * 6170 * - Pick up the next available descriptor in the transmit ring. 6171 * 6172 * - Free the network buffer previously sent with that descriptor, if any. 6173 * 6174 * - Initialize the transmit descriptor with the information provided 6175 * in the *rte_mbuf data structure. 6176 * 6177 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers, 6178 * the rte_eth_tx_burst() function uses several transmit descriptors 6179 * of the ring. 6180 * 6181 * The rte_eth_tx_burst() function returns the number of packets it 6182 * actually sent. A return value equal to *nb_pkts* means that all packets 6183 * have been sent, and this is likely to signify that other output packets 6184 * could be immediately transmitted again. Applications that implement a 6185 * "send as many packets to transmit as possible" policy can check this 6186 * specific case and keep invoking the rte_eth_tx_burst() function until 6187 * a value less than *nb_pkts* is returned. 6188 * 6189 * It is the responsibility of the rte_eth_tx_burst() function to 6190 * transparently free the memory buffers of packets previously sent. 6191 * This feature is driven by the *tx_free_thresh* value supplied to the 6192 * rte_eth_dev_configure() function at device configuration time. 6193 * When the number of free Tx descriptors drops below this threshold, the 6194 * rte_eth_tx_burst() function must [attempt to] free the *rte_mbuf* buffers 6195 * of those packets whose transmission was effectively completed. 6196 * 6197 * If the PMD is RTE_ETH_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads can 6198 * invoke this function concurrently on the same Tx queue without SW lock. 6199 * @see rte_eth_dev_info_get, struct rte_eth_txconf::offloads 6200 * 6201 * @see rte_eth_tx_prepare to perform some prior checks or adjustments 6202 * for offloads. 6203 * 6204 * @note This function must not modify mbufs (including packets data) 6205 * unless the refcnt is 1. 6206 * An exception is the bonding PMD, which does not have "Tx prepare" support, 6207 * in this case, mbufs may be modified. 6208 * 6209 * @param port_id 6210 * The port identifier of the Ethernet device. 6211 * @param queue_id 6212 * The index of the transmit queue through which output packets must be 6213 * sent. 6214 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 6215 * to rte_eth_dev_configure(). 6216 * @param tx_pkts 6217 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures 6218 * which contain the output packets. 6219 * @param nb_pkts 6220 * The maximum number of packets to transmit. 6221 * @return 6222 * The number of output packets actually stored in transmit descriptors of 6223 * the transmit ring. The return value can be less than the value of the 6224 * *tx_pkts* parameter when the transmit ring is full or has been filled up. 6225 */ 6226 static inline uint16_t 6227 rte_eth_tx_burst(uint16_t port_id, uint16_t queue_id, 6228 struct rte_mbuf **tx_pkts, uint16_t nb_pkts) 6229 { 6230 struct rte_eth_fp_ops *p; 6231 void *qd; 6232 6233 #ifdef RTE_ETHDEV_DEBUG_TX 6234 if (port_id >= RTE_MAX_ETHPORTS || 6235 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 6236 RTE_ETHDEV_LOG(ERR, 6237 "Invalid port_id=%u or queue_id=%u\n", 6238 port_id, queue_id); 6239 return 0; 6240 } 6241 #endif 6242 6243 /* fetch pointer to queue data */ 6244 p = &rte_eth_fp_ops[port_id]; 6245 qd = p->txq.data[queue_id]; 6246 6247 #ifdef RTE_ETHDEV_DEBUG_TX 6248 RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0); 6249 6250 if (qd == NULL) { 6251 RTE_ETHDEV_LOG(ERR, "Invalid Tx queue_id=%u for port_id=%u\n", 6252 queue_id, port_id); 6253 return 0; 6254 } 6255 #endif 6256 6257 #ifdef RTE_ETHDEV_RXTX_CALLBACKS 6258 { 6259 void *cb; 6260 6261 /* __ATOMIC_RELEASE memory order was used when the 6262 * call back was inserted into the list. 6263 * Since there is a clear dependency between loading 6264 * cb and cb->fn/cb->next, __ATOMIC_ACQUIRE memory order is 6265 * not required. 6266 */ 6267 cb = __atomic_load_n((void **)&p->txq.clbk[queue_id], 6268 __ATOMIC_RELAXED); 6269 if (unlikely(cb != NULL)) 6270 nb_pkts = rte_eth_call_tx_callbacks(port_id, queue_id, 6271 tx_pkts, nb_pkts, cb); 6272 } 6273 #endif 6274 6275 nb_pkts = p->tx_pkt_burst(qd, tx_pkts, nb_pkts); 6276 6277 rte_ethdev_trace_tx_burst(port_id, queue_id, (void **)tx_pkts, nb_pkts); 6278 return nb_pkts; 6279 } 6280 6281 /** 6282 * Process a burst of output packets on a transmit queue of an Ethernet device. 6283 * 6284 * The rte_eth_tx_prepare() function is invoked to prepare output packets to be 6285 * transmitted on the output queue *queue_id* of the Ethernet device designated 6286 * by its *port_id*. 6287 * The *nb_pkts* parameter is the number of packets to be prepared which are 6288 * supplied in the *tx_pkts* array of *rte_mbuf* structures, each of them 6289 * allocated from a pool created with rte_pktmbuf_pool_create(). 6290 * For each packet to send, the rte_eth_tx_prepare() function performs 6291 * the following operations: 6292 * 6293 * - Check if packet meets devices requirements for Tx offloads. 6294 * 6295 * - Check limitations about number of segments. 6296 * 6297 * - Check additional requirements when debug is enabled. 6298 * 6299 * - Update and/or reset required checksums when Tx offload is set for packet. 6300 * 6301 * Since this function can modify packet data, provided mbufs must be safely 6302 * writable (e.g. modified data cannot be in shared segment). 6303 * 6304 * The rte_eth_tx_prepare() function returns the number of packets ready to be 6305 * sent. A return value equal to *nb_pkts* means that all packets are valid and 6306 * ready to be sent, otherwise stops processing on the first invalid packet and 6307 * leaves the rest packets untouched. 6308 * 6309 * When this functionality is not implemented in the driver, all packets are 6310 * are returned untouched. 6311 * 6312 * @param port_id 6313 * The port identifier of the Ethernet device. 6314 * The value must be a valid port ID. 6315 * @param queue_id 6316 * The index of the transmit queue through which output packets must be 6317 * sent. 6318 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 6319 * to rte_eth_dev_configure(). 6320 * @param tx_pkts 6321 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures 6322 * which contain the output packets. 6323 * @param nb_pkts 6324 * The maximum number of packets to process. 6325 * @return 6326 * The number of packets correct and ready to be sent. The return value can be 6327 * less than the value of the *tx_pkts* parameter when some packet doesn't 6328 * meet devices requirements with rte_errno set appropriately: 6329 * - EINVAL: offload flags are not correctly set 6330 * - ENOTSUP: the offload feature is not supported by the hardware 6331 * - ENODEV: if *port_id* is invalid (with debug enabled only) 6332 * 6333 */ 6334 6335 #ifndef RTE_ETHDEV_TX_PREPARE_NOOP 6336 6337 static inline uint16_t 6338 rte_eth_tx_prepare(uint16_t port_id, uint16_t queue_id, 6339 struct rte_mbuf **tx_pkts, uint16_t nb_pkts) 6340 { 6341 struct rte_eth_fp_ops *p; 6342 void *qd; 6343 6344 #ifdef RTE_ETHDEV_DEBUG_TX 6345 if (port_id >= RTE_MAX_ETHPORTS || 6346 queue_id >= RTE_MAX_QUEUES_PER_PORT) { 6347 RTE_ETHDEV_LOG(ERR, 6348 "Invalid port_id=%u or queue_id=%u\n", 6349 port_id, queue_id); 6350 rte_errno = ENODEV; 6351 return 0; 6352 } 6353 #endif 6354 6355 /* fetch pointer to queue data */ 6356 p = &rte_eth_fp_ops[port_id]; 6357 qd = p->txq.data[queue_id]; 6358 6359 #ifdef RTE_ETHDEV_DEBUG_TX 6360 if (!rte_eth_dev_is_valid_port(port_id)) { 6361 RTE_ETHDEV_LOG(ERR, "Invalid Tx port_id=%u\n", port_id); 6362 rte_errno = ENODEV; 6363 return 0; 6364 } 6365 if (qd == NULL) { 6366 RTE_ETHDEV_LOG(ERR, "Invalid Tx queue_id=%u for port_id=%u\n", 6367 queue_id, port_id); 6368 rte_errno = EINVAL; 6369 return 0; 6370 } 6371 #endif 6372 6373 if (!p->tx_pkt_prepare) 6374 return nb_pkts; 6375 6376 return p->tx_pkt_prepare(qd, tx_pkts, nb_pkts); 6377 } 6378 6379 #else 6380 6381 /* 6382 * Native NOOP operation for compilation targets which doesn't require any 6383 * preparations steps, and functional NOOP may introduce unnecessary performance 6384 * drop. 6385 * 6386 * Generally this is not a good idea to turn it on globally and didn't should 6387 * be used if behavior of tx_preparation can change. 6388 */ 6389 6390 static inline uint16_t 6391 rte_eth_tx_prepare(__rte_unused uint16_t port_id, 6392 __rte_unused uint16_t queue_id, 6393 __rte_unused struct rte_mbuf **tx_pkts, uint16_t nb_pkts) 6394 { 6395 return nb_pkts; 6396 } 6397 6398 #endif 6399 6400 /** 6401 * Send any packets queued up for transmission on a port and HW queue 6402 * 6403 * This causes an explicit flush of packets previously buffered via the 6404 * rte_eth_tx_buffer() function. It returns the number of packets successfully 6405 * sent to the NIC, and calls the error callback for any unsent packets. Unless 6406 * explicitly set up otherwise, the default callback simply frees the unsent 6407 * packets back to the owning mempool. 6408 * 6409 * @param port_id 6410 * The port identifier of the Ethernet device. 6411 * @param queue_id 6412 * The index of the transmit queue through which output packets must be 6413 * sent. 6414 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 6415 * to rte_eth_dev_configure(). 6416 * @param buffer 6417 * Buffer of packets to be transmit. 6418 * @return 6419 * The number of packets successfully sent to the Ethernet device. The error 6420 * callback is called for any packets which could not be sent. 6421 */ 6422 static inline uint16_t 6423 rte_eth_tx_buffer_flush(uint16_t port_id, uint16_t queue_id, 6424 struct rte_eth_dev_tx_buffer *buffer) 6425 { 6426 uint16_t sent; 6427 uint16_t to_send = buffer->length; 6428 6429 if (to_send == 0) 6430 return 0; 6431 6432 sent = rte_eth_tx_burst(port_id, queue_id, buffer->pkts, to_send); 6433 6434 buffer->length = 0; 6435 6436 /* All packets sent, or to be dealt with by callback below */ 6437 if (unlikely(sent != to_send)) 6438 buffer->error_callback(&buffer->pkts[sent], 6439 (uint16_t)(to_send - sent), 6440 buffer->error_userdata); 6441 6442 return sent; 6443 } 6444 6445 /** 6446 * Buffer a single packet for future transmission on a port and queue 6447 * 6448 * This function takes a single mbuf/packet and buffers it for later 6449 * transmission on the particular port and queue specified. Once the buffer is 6450 * full of packets, an attempt will be made to transmit all the buffered 6451 * packets. In case of error, where not all packets can be transmitted, a 6452 * callback is called with the unsent packets as a parameter. If no callback 6453 * is explicitly set up, the unsent packets are just freed back to the owning 6454 * mempool. The function returns the number of packets actually sent i.e. 6455 * 0 if no buffer flush occurred, otherwise the number of packets successfully 6456 * flushed 6457 * 6458 * @param port_id 6459 * The port identifier of the Ethernet device. 6460 * @param queue_id 6461 * The index of the transmit queue through which output packets must be 6462 * sent. 6463 * The value must be in the range [0, nb_tx_queue - 1] previously supplied 6464 * to rte_eth_dev_configure(). 6465 * @param buffer 6466 * Buffer used to collect packets to be sent. 6467 * @param tx_pkt 6468 * Pointer to the packet mbuf to be sent. 6469 * @return 6470 * 0 = packet has been buffered for later transmission 6471 * N > 0 = packet has been buffered, and the buffer was subsequently flushed, 6472 * causing N packets to be sent, and the error callback to be called for 6473 * the rest. 6474 */ 6475 static __rte_always_inline uint16_t 6476 rte_eth_tx_buffer(uint16_t port_id, uint16_t queue_id, 6477 struct rte_eth_dev_tx_buffer *buffer, struct rte_mbuf *tx_pkt) 6478 { 6479 buffer->pkts[buffer->length++] = tx_pkt; 6480 if (buffer->length < buffer->size) 6481 return 0; 6482 6483 return rte_eth_tx_buffer_flush(port_id, queue_id, buffer); 6484 } 6485 6486 /** 6487 * @warning 6488 * @b EXPERIMENTAL: this API may change without prior notice 6489 * 6490 * Get supported header protocols to split on Rx. 6491 * 6492 * When a packet type is announced to be split, 6493 * it *must* be supported by the PMD. 6494 * For instance, if eth-ipv4, eth-ipv4-udp is announced, 6495 * the PMD must return the following packet types for these packets: 6496 * - Ether/IPv4 -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4 6497 * - Ether/IPv4/UDP -> RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L4_UDP 6498 * 6499 * @param port_id 6500 * The port identifier of the device. 6501 * @param[out] ptypes 6502 * An array pointer to store supported protocol headers, allocated by caller. 6503 * These ptypes are composed with RTE_PTYPE_*. 6504 * @param num 6505 * Size of the array pointed by param ptypes. 6506 * @return 6507 * - (>=0) Number of supported ptypes. If the number of types exceeds num, 6508 * only num entries will be filled into the ptypes array, 6509 * but the full count of supported ptypes will be returned. 6510 * - (-ENOTSUP) if header protocol is not supported by device. 6511 * - (-ENODEV) if *port_id* invalid. 6512 * - (-EINVAL) if bad parameter. 6513 */ 6514 __rte_experimental 6515 int rte_eth_buffer_split_get_supported_hdr_ptypes(uint16_t port_id, uint32_t *ptypes, int num); 6516 6517 #ifdef __cplusplus 6518 } 6519 #endif 6520 6521 #endif /* _RTE_ETHDEV_H_ */ 6522