4 * Copyright(c) 2010-2012 Intel Corporation. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of Intel Corporation nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
35 #ifndef _RTE_ETHDEV_H_
36 #define _RTE_ETHDEV_H_
41 * RTE Ethernet Device API
43 * The Ethernet Device API is composed of two parts:
45 * - The application-oriented Ethernet API that includes functions to setup
46 * an Ethernet device (configure it, setup its RX and TX queues and start it),
47 * to get its MAC address, the speed and the status of its physical link,
48 * to receive and to transmit packets, and so on.
50 * - The driver-oriented Ethernet API that exports a function allowing
51 * an Ethernet Poll Mode Driver (PMD) to simultaneously register itself as
52 * an Ethernet device driver and as a PCI driver for a set of matching PCI
53 * [Ethernet] devices classes.
55 * By default, all the functions of the Ethernet Device API exported by a PMD
56 * are lock-free functions which assume to not be invoked in parallel on
57 * different logical cores to work on the same target object. For instance,
58 * the receive function of a PMD cannot be invoked in parallel on two logical
59 * cores to poll the same RX queue [of the same port]. Of course, this function
60 * can be invoked in parallel by different logical cores on different RX queues.
61 * It is the responsibility of the upper level application to enforce this rule.
63 * If needed, parallel accesses by multiple logical cores to shared queues
64 * shall be explicitly protected by dedicated inline lock-aware functions
65 * built on top of their corresponding lock-free functions of the PMD API.
67 * In all functions of the Ethernet API, the Ethernet device is
68 * designated by an integer >= 0 named the device port identifier.
70 * At the Ethernet driver level, Ethernet devices are represented by a generic
71 * data structure of type *rte_eth_dev*.
73 * Ethernet devices are dynamically registered during the PCI probing phase
74 * performed at EAL initialization time.
75 * When an Ethernet device is being probed, an *rte_eth_dev* structure and
76 * a new port identifier are allocated for that device. Then, the eth_dev_init()
77 * function supplied by the Ethernet driver matching the probed PCI
78 * device is invoked to properly initialize the device.
80 * The role of the device init function consists of resetting the hardware,
81 * checking access to Non-volatile Memory (NVM), reading the MAC address
84 * If the device init operation is successful, the correspondence between
85 * the port identifier assigned to the new device and its associated
86 * *rte_eth_dev* structure is effectively registered.
87 * Otherwise, both the *rte_eth_dev* structure and the port identifier are
90 * The functions exported by the application Ethernet API to setup a device
91 * designated by its port identifier must be invoked in the following order:
92 * - rte_eth_dev_configure()
93 * - rte_eth_tx_queue_setup()
94 * - rte_eth_rx_queue_setup()
95 * - rte_eth_dev_start()
97 * Then, the network application can invoke, in any order, the functions
98 * exported by the Ethernet API to get the MAC address of a given device, to
99 * get the speed and the status of a device physical link, to receive/transmit
100 * [burst of] packets, and so on.
102 * If the application wants to change the configuration (i.e. call
103 * rte_eth_dev_configure(), rte_eth_tx_queue_setup(), or
104 * rte_eth_rx_queue_setup()), it must call rte_eth_dev_stop() first to stop the
105 * device and then do the reconfiguration before calling rte_eth_dev_start()
106 * again. The tramsit and receive functions should not be invoked when the
109 * Please note that some configuration is not stored between calls to
110 * rte_eth_dev_stop()/rte_eth_dev_start(). The following configuration will
113 * - flow control settings
114 * - receive mode configuration (promiscuous mode, hardware checksum mode,
115 * RSS/VMDQ settings etc.)
116 * - VLAN filtering configuration
117 * - MAC addresses supplied to MAC address array
118 * - flow director filtering mode (but not filtering rules)
119 * - NIC queue statistics mappings
121 * Any other configuration will not be stored and will need to be re-entered
122 * after a call to rte_eth_dev_start().
124 * Finally, a network application can close an Ethernet device by invoking the
125 * rte_eth_dev_close() function.
127 * Each function of the application Ethernet API invokes a specific function
128 * of the PMD that controls the target device designated by its port
130 * For this purpose, all device-specific functions of an Ethernet driver are
131 * supplied through a set of pointers contained in a generic structure of type
133 * The address of the *eth_dev_ops* structure is stored in the *rte_eth_dev*
134 * structure by the device init function of the Ethernet driver, which is
135 * invoked during the PCI probing phase, as explained earlier.
137 * In other words, each function of the Ethernet API simply retrieves the
138 * *rte_eth_dev* structure associated with the device port identifier and
139 * performs an indirect invocation of the corresponding driver function
140 * supplied in the *eth_dev_ops* structure of the *rte_eth_dev* structure.
142 * For performance reasons, the address of the burst-oriented RX and TX
143 * functions of the Ethernet driver are not contained in the *eth_dev_ops*
144 * structure. Instead, they are directly stored at the beginning of the
145 * *rte_eth_dev* structure to avoid an extra indirect memory access during
148 * RTE ethernet device drivers do not use interrupts for transmitting or
149 * receiving. Instead, Ethernet drivers export Poll-Mode receive and transmit
150 * functions to applications.
151 * Both receive and transmit functions are packet-burst oriented to minimize
152 * their cost per packet through the following optimizations:
154 * - Sharing among multiple packets the incompressible cost of the
155 * invocation of receive/transmit functions.
157 * - Enabling receive/transmit functions to take advantage of burst-oriented
158 * hardware features (L1 cache, prefetch instructions, NIC head/tail
159 * registers) to minimize the number of CPU cycles per packet, for instance,
160 * by avoiding useless read memory accesses to ring descriptors, or by
161 * systematically using arrays of pointers that exactly fit L1 cache line
162 * boundaries and sizes.
164 * The burst-oriented receive function does not provide any error notification,
165 * to avoid the corresponding overhead. As a hint, the upper-level application
166 * might check the status of the device link once being systematically returned
167 * a 0 value by the receive function of the driver for a given number of tries.
177 #include <rte_interrupts.h>
179 #include <rte_mbuf.h>
180 #include "rte_ether.h"
183 * A structure used to retrieve statistics for an Ethernet port.
185 struct rte_eth_stats {
186 uint64_t ipackets; /**< Total number of successfully received packets. */
187 uint64_t opackets; /**< Total number of successfully transmitted packets.*/
188 uint64_t ibytes; /**< Total number of successfully received bytes. */
189 uint64_t obytes; /**< Total number of successfully transmitted bytes. */
190 uint64_t ierrors; /**< Total number of erroneous received packets. */
191 uint64_t oerrors; /**< Total number of failed transmitted packets. */
192 uint64_t imcasts; /**< Total number of multicast received packets. */
193 uint64_t rx_nombuf; /**< Total number of RX mbuf allocation failures. */
194 uint64_t fdirmatch; /**< Total number of RX packets matching a filter. */
195 uint64_t fdirmiss; /**< Total number of RX packets not matching any filter. */
196 uint64_t q_ipackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
197 /**< Total number of queue RX packets. */
198 uint64_t q_opackets[RTE_ETHDEV_QUEUE_STAT_CNTRS];
199 /**< Total number of queue TX packets. */
200 uint64_t q_ibytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
201 /**< Total number of successfully received queue bytes. */
202 uint64_t q_obytes[RTE_ETHDEV_QUEUE_STAT_CNTRS];
203 /**< Total number of successfully transmitted queue bytes. */
204 uint64_t q_errors[RTE_ETHDEV_QUEUE_STAT_CNTRS];
205 /**< Total number of queue packets received that are dropped. */
209 * A structure used to retrieve link-level information of an Ethernet port.
211 struct rte_eth_link {
212 uint16_t link_speed; /**< ETH_LINK_SPEED_[10, 100, 1000, 10000] */
213 uint16_t link_duplex; /**< ETH_LINK_[HALF_DUPLEX, FULL_DUPLEX] */
214 uint8_t link_status : 1; /**< 1 -> link up, 0 -> link down */
215 }__attribute__((aligned(8))); /**< aligned for atomic64 read/write */
217 #define ETH_LINK_SPEED_AUTONEG 0 /**< Auto-negotiate link speed. */
218 #define ETH_LINK_SPEED_10 10 /**< 10 megabits/second. */
219 #define ETH_LINK_SPEED_100 100 /**< 100 megabits/second. */
220 #define ETH_LINK_SPEED_1000 1000 /**< 1 gigabits/second. */
221 #define ETH_LINK_SPEED_10000 10000 /**< 10 gigabits/second. */
223 #define ETH_LINK_AUTONEG_DUPLEX 0 /**< Auto-negotiate duplex. */
224 #define ETH_LINK_HALF_DUPLEX 1 /**< Half-duplex connection. */
225 #define ETH_LINK_FULL_DUPLEX 2 /**< Full-duplex connection. */
228 * A structure used to configure the ring threshold registers of an RX/TX
229 * queue for an Ethernet port.
231 struct rte_eth_thresh {
232 uint8_t pthresh; /**< Ring prefetch threshold. */
233 uint8_t hthresh; /**< Ring host threshold. */
234 uint8_t wthresh; /**< Ring writeback threshold. */
238 * A set of values to identify what method is to be used to route
239 * packets to multiple queues.
241 enum rte_eth_rx_mq_mode {
242 ETH_RSS = 0, /**< Default to RSS mode */
243 ETH_VMDQ_DCB /**< Use VMDQ+DCB to route traffic to queues */
247 * A structure used to configure the RX features of an Ethernet port.
249 struct rte_eth_rxmode {
250 /** The multi-queue packet distribution mode to be used, e.g. RSS. */
251 enum rte_eth_rx_mq_mode mq_mode;
252 uint32_t max_rx_pkt_len; /**< Only used if jumbo_frame enabled. */
253 uint16_t split_hdr_size; /**< hdr buf size (header_split enabled).*/
254 uint8_t header_split : 1, /**< Header Split enable. */
255 hw_ip_checksum : 1, /**< IP/UDP/TCP checksum offload enable. */
256 hw_vlan_filter : 1, /**< VLAN filter enable. */
257 jumbo_frame : 1, /**< Jumbo Frame Receipt enable. */
258 hw_strip_crc : 1; /**< Enable CRC stripping by hardware. */
262 * A structure used to configure the Receive Side Scaling (RSS) feature
263 * of an Ethernet port.
264 * If not NULL, the *rss_key* pointer of the *rss_conf* structure points
265 * to an array of 40 bytes holding the RSS key to use for hashing specific
266 * header fields of received packets.
267 * Otherwise, a default random hash key is used by the device driver.
269 * The *rss_hf* field of the *rss_conf* structure indicates the different
270 * types of IPv4/IPv6 packets to which the RSS hashing must be applied.
271 * Supplying an *rss_hf* equal to zero disables the RSS feature.
273 struct rte_eth_rss_conf {
274 uint8_t *rss_key; /**< If not NULL, 40-byte hash key. */
275 uint16_t rss_hf; /**< Hash functions to apply - see below. */
278 #define ETH_RSS_IPV4 0x0001 /**< IPv4 packet. */
279 #define ETH_RSS_IPV4_TCP 0x0002 /**< IPv4/TCP packet. */
280 #define ETH_RSS_IPV6 0x0004 /**< IPv6 packet. */
281 #define ETH_RSS_IPV6_EX 0x0008 /**< IPv6 packet with extension headers.*/
282 #define ETH_RSS_IPV6_TCP 0x0010 /**< IPv6/TCP packet. */
283 #define ETH_RSS_IPV6_TCP_EX 0x0020 /**< IPv6/TCP with extension headers. */
284 /* Intel RSS extensions to UDP packets */
285 #define ETH_RSS_IPV4_UDP 0x0040 /**< IPv4/UDP packet. */
286 #define ETH_RSS_IPV6_UDP 0x0080 /**< IPv6/UDP packet. */
287 #define ETH_RSS_IPV6_UDP_EX 0x0100 /**< IPv6/UDP with extension headers. */
289 /* Definitions used for VMDQ and DCB functionality */
290 #define ETH_VMDQ_MAX_VLAN_FILTERS 64 /**< Maximum nb. of VMDQ vlan filters. */
291 #define ETH_DCB_NUM_USER_PRIORITIES 8 /**< Maximum nb. of DCB priorities. */
292 #define ETH_VMDQ_DCB_NUM_QUEUES 128 /**< Maximum nb. of VMDQ DCB queues. */
295 * This enum indicates the possible number of queue pools
296 * in VMDQ+DCB configurations.
298 enum rte_eth_nb_pools {
299 ETH_16_POOLS = 16, /**< 16 pools with DCB. */
300 ETH_32_POOLS = 32 /**< 32 pools with DCB. */
304 * A structure used to configure the VMDQ+DCB feature
305 * of an Ethernet port.
307 * Using this feature, packets are routed to a pool of queues, based
308 * on the vlan id in the vlan tag, and then to a specific queue within
309 * that pool, using the user priority vlan tag field.
311 * A default pool may be used, if desired, to route all traffic which
312 * does not match the vlan filter rules.
314 struct rte_eth_vmdq_dcb_conf {
315 enum rte_eth_nb_pools nb_queue_pools; /**< With DCB, 16 or 32 pools */
316 uint8_t enable_default_pool; /**< If non-zero, use a default pool */
317 uint8_t default_pool; /**< The default pool, if applicable */
318 uint8_t nb_pool_maps; /**< We can have up to 64 filters/mappings */
320 uint16_t vlan_id; /**< The vlan id of the received frame */
321 uint64_t pools; /**< Bitmask of pools for packet rx */
322 } pool_map[ETH_VMDQ_MAX_VLAN_FILTERS]; /**< VMDq vlan pool maps. */
323 uint8_t dcb_queue[ETH_DCB_NUM_USER_PRIORITIES];
324 /**< Selects a queue in a pool */
328 * A structure used to configure the TX features of an Ethernet port.
329 * For future extensions.
331 struct rte_eth_txmode {
335 * A structure used to configure an RX ring of an Ethernet port.
337 struct rte_eth_rxconf {
338 struct rte_eth_thresh rx_thresh; /**< RX ring threshold registers. */
339 uint16_t rx_free_thresh; /**< Drives the freeing of RX descriptors. */
343 * A structure used to configure a TX ring of an Ethernet port.
345 struct rte_eth_txconf {
346 struct rte_eth_thresh tx_thresh; /**< TX ring threshold registers. */
347 uint16_t tx_rs_thresh; /**< Drives the setting of RS bit on TXDs. */
348 uint16_t tx_free_thresh; /**< Drives the freeing of TX buffers. */
352 * This enum indicates the flow control mode
354 enum rte_eth_fc_mode {
355 RTE_FC_NONE = 0, /**< Disable flow control. */
356 RTE_FC_RX_PAUSE, /**< RX pause frame, enable flowctrl on TX side. */
357 RTE_FC_TX_PAUSE, /**< TX pause frame, enable flowctrl on RX side. */
358 RTE_FC_FULL /**< Enable flow control on both side. */
362 * A structure used to configure Ethernet flow control parameter.
363 * These parameters will be configured into the register of the NIC.
364 * Please refer to the corresponding data sheet for proper value.
366 struct rte_eth_fc_conf {
367 uint32_t high_water; /**< High threshold value to trigger XOFF */
368 uint32_t low_water; /**< Low threshold value to trigger XON */
369 uint16_t pause_time; /**< Pause quota in the Pause frame */
370 uint16_t send_xon; /**< Is XON frame need be sent */
371 enum rte_eth_fc_mode mode; /**< Link flow control mode */
375 * Flow Director setting modes: none (default), signature or perfect.
378 RTE_FDIR_MODE_NONE = 0, /**< Disable FDIR support. */
379 RTE_FDIR_MODE_SIGNATURE, /**< Enable FDIR signature filter mode. */
380 RTE_FDIR_MODE_PERFECT, /**< Enable FDIR perfect filter mode. */
384 * Memory space that can be configured to store Flow Director filters
385 * in the board memory.
387 enum rte_fdir_pballoc_type {
388 RTE_FDIR_PBALLOC_64K = 0, /**< 64k. */
389 RTE_FDIR_PBALLOC_128K, /**< 128k. */
390 RTE_FDIR_PBALLOC_256K, /**< 256k. */
394 * Select report mode of FDIR hash information in RX descriptors.
396 enum rte_fdir_status_mode {
397 RTE_FDIR_NO_REPORT_STATUS = 0, /**< Never report FDIR hash. */
398 RTE_FDIR_REPORT_STATUS, /**< Only report FDIR hash for matching pkts. */
399 RTE_FDIR_REPORT_STATUS_ALWAYS, /**< Always report FDIR hash. */
403 * A structure used to configure the Flow Director (FDIR) feature
404 * of an Ethernet port.
406 * If mode is RTE_FDIR_DISABLE, the pballoc value is ignored.
408 struct rte_fdir_conf {
409 enum rte_fdir_mode mode; /**< Flow Director mode. */
410 enum rte_fdir_pballoc_type pballoc; /**< Space for FDIR filters. */
411 enum rte_fdir_status_mode status; /**< How to report FDIR hash. */
412 /* Offset of flexbytes field in RX packets (in 16-bit word units). */
413 uint8_t flexbytes_offset;
414 /* RX queue of packets matching a "drop" filter in perfect mode. */
419 * Possible l4type of FDIR filters.
422 RTE_FDIR_L4TYPE_NONE = 0, /**< None. */
423 RTE_FDIR_L4TYPE_UDP, /**< UDP. */
424 RTE_FDIR_L4TYPE_TCP, /**< TCP. */
425 RTE_FDIR_L4TYPE_SCTP, /**< SCTP. */
429 * Select IPv4 or IPv6 FDIR filters.
432 RTE_FDIR_IPTYPE_IPV4 = 0, /**< IPv4. */
433 RTE_FDIR_IPTYPE_IPV6 , /**< IPv6. */
437 * A structure used to define a FDIR packet filter.
439 struct rte_fdir_filter {
440 uint16_t flex_bytes; /**< Flex bytes value to match. */
441 uint16_t vlan_id; /**< VLAN ID value to match, 0 otherwise. */
442 uint16_t port_src; /**< Source port to match, 0 otherwise. */
443 uint16_t port_dst; /**< Destination port to match, 0 otherwise. */
445 uint32_t ipv4_addr; /**< IPv4 source address to match. */
446 uint32_t ipv6_addr[4]; /**< IPv6 source address to match. */
447 } ip_src; /**< IPv4/IPv6 source address to match (union of above). */
449 uint32_t ipv4_addr; /**< IPv4 destination address to match. */
450 uint32_t ipv6_addr[4]; /**< IPv6 destination address to match */
451 } ip_dst; /**< IPv4/IPv6 destination address to match (union of above). */
452 enum rte_l4type l4type; /**< l4type to match: NONE/UDP/TCP/SCTP. */
453 enum rte_iptype iptype; /**< IP packet type to match: IPv4 or IPv6. */
457 * A structure used to configure FDIR masks that are used by the device
458 * to match the various fields of RX packet headers.
459 * @note The only_ip_flow field has the opposite meaning compared to other
462 struct rte_fdir_masks {
463 /** When set to 1, packet l4type is \b NOT relevant in filters, and
464 source and destination port masks must be set to zero. */
465 uint8_t only_ip_flow;
466 /** If set to 1, vlan_id is relevant in filters. */
468 /** If set to 1, vlan_prio is relevant in filters. */
470 /** If set to 1, flexbytes is relevant in filters. */
472 /** Mask of Destination IPv4 Address. All bits set to 1 define the
473 relevant bits to use in the destination address of an IPv4 packet
474 when matching it against FDIR filters. */
475 uint32_t dst_ipv4_mask;
476 /** Mask of Source IPv4 Address. All bits set to 1 define
477 the relevant bits to use in the source address of an IPv4 packet
478 when matching it against FDIR filters. */
479 uint32_t src_ipv4_mask;
480 /** Mask of Source IPv6 Address. All bits set to 1 define the
481 relevant BYTES to use in the source address of an IPv6 packet
482 when matching it against FDIR filters. */
483 uint16_t src_ipv6_mask;
484 /** Mask of Source Port. All bits set to 1 define the relevant
485 bits to use in the source port of an IP packets when matching it
486 against FDIR filters. */
487 uint16_t src_port_mask;
488 /** Mask of Destination Port. All bits set to 1 define the relevant
489 bits to use in the destination port of an IP packet when matching it
490 against FDIR filters. */
491 uint16_t dst_port_mask;
495 * A structure used to report the status of the flow director filters in use.
497 struct rte_eth_fdir {
498 /** Number of filters with collision indication. */
500 /** Number of free (non programmed) filters. */
502 /** The Lookup hash value of the added filter that updated the value
503 of the MAXLEN field */
505 /** Longest linked list of filters in the table. */
507 /** Number of added filters. */
509 /** Number of removed filters. */
511 /** Number of failed added filters (no more space in device). */
513 /** Number of failed removed filters. */
518 * A structure used to enable/disable specific device interrupts.
520 struct rte_intr_conf {
521 /** enable/disable lsc interrupt. 0 (default) - disable, 1 enable */
526 * A structure used to configure an Ethernet port.
527 * Depending upon the RX multi-queue mode, extra advanced
528 * configuration settings may be needed.
530 struct rte_eth_conf {
532 /**< ETH_LINK_SPEED_10[0|00|000], or 0 for autonegotation */
533 uint16_t link_duplex;
534 /**< ETH_LINK_[HALF_DUPLEX|FULL_DUPLEX], or 0 for autonegotation */
535 struct rte_eth_rxmode rxmode; /**< Port RX configuration. */
536 struct rte_eth_txmode txmode; /**< Port TX configuration. */
538 struct rte_eth_rss_conf rss_conf; /**< Port RSS configuration */
539 struct rte_eth_vmdq_dcb_conf vmdq_dcb_conf;
540 /**< Port vmdq+dcb configuration. */
541 } rx_adv_conf; /**< Port RX filtering configuration (union). */
542 struct rte_fdir_conf fdir_conf; /**< FDIR configuration. */
543 struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
547 * A structure used to retrieve the contextual information of
548 * an Ethernet device, such as the controlling driver of the device,
549 * its PCI context, etc...
551 struct rte_eth_dev_info {
552 struct rte_pci_device *pci_dev; /**< Device PCI information. */
553 const char *driver_name; /**< Device Driver name. */
554 uint32_t min_rx_bufsize; /**< Minimum size of RX buffer. */
555 uint32_t max_rx_pktlen; /**< Maximum configurable length of RX pkt. */
556 uint16_t max_rx_queues; /**< Maximum number of RX queues. */
557 uint16_t max_tx_queues; /**< Maximum number of TX queues. */
558 uint32_t max_mac_addrs; /**< Maximum number of MAC addresses. */
565 struct rte_eth_dev_callback;
566 /** @internal Structure to keep track of registered callbacks */
567 TAILQ_HEAD(rte_eth_dev_cb_list, rte_eth_dev_callback);
570 * Definitions of all functions exported by an Ethernet driver through the
571 * the generic structure of type *eth_dev_ops* supplied in the *rte_eth_dev*
572 * structure associated with an Ethernet device.
575 typedef int (*eth_dev_configure_t)(struct rte_eth_dev *dev, uint16_t nb_rx_q,
577 /**< @internal Ethernet device configuration. */
579 typedef int (*eth_dev_start_t)(struct rte_eth_dev *dev);
580 /**< @internal Function used to start a configured Ethernet device. */
582 typedef void (*eth_dev_stop_t)(struct rte_eth_dev *dev);
583 /**< @internal Function used to stop a configured Ethernet device. */
585 typedef void (*eth_dev_close_t)(struct rte_eth_dev *dev);
586 /**< @internal Function used to close a configured Ethernet device. */
588 typedef void (*eth_promiscuous_enable_t)(struct rte_eth_dev *dev);
589 /**< @internal Function used to enable the RX promiscuous mode of an Ethernet device. */
591 typedef void (*eth_promiscuous_disable_t)(struct rte_eth_dev *dev);
592 /**< @internal Function used to disable the RX promiscuous mode of an Ethernet device. */
594 typedef void (*eth_allmulticast_enable_t)(struct rte_eth_dev *dev);
595 /**< @internal Enable the receipt of all multicast packets by an Ethernet device. */
597 typedef void (*eth_allmulticast_disable_t)(struct rte_eth_dev *dev);
598 /**< @internal Disable the receipt of all multicast packets by an Ethernet device. */
600 typedef int (*eth_link_update_t)(struct rte_eth_dev *dev,
601 int wait_to_complete);
602 /**< @internal Get link speed, duplex mode and state (up/down) of an Ethernet device. */
604 typedef void (*eth_stats_get_t)(struct rte_eth_dev *dev,
605 struct rte_eth_stats *igb_stats);
606 /**< @internal Get global I/O statistics of an Ethernet device. */
608 typedef void (*eth_stats_reset_t)(struct rte_eth_dev *dev);
609 /**< @internal Reset global I/O statistics of an Ethernet device to 0. */
611 typedef int (*eth_queue_stats_mapping_set_t)(struct rte_eth_dev *dev,
615 /**< @internal Set a queue statistics mapping for a tx/rx queue of an Ethernet device. */
617 typedef void (*eth_dev_infos_get_t)(struct rte_eth_dev *dev,
618 struct rte_eth_dev_info *dev_info);
619 /**< @internal Get specific informations of an Ethernet device. */
621 typedef int (*eth_rx_queue_setup_t)(struct rte_eth_dev *dev,
622 uint16_t rx_queue_id,
624 unsigned int socket_id,
625 const struct rte_eth_rxconf *rx_conf,
626 struct rte_mempool *mb_pool);
627 /**< @internal Set up a receive queue of an Ethernet device. */
629 typedef int (*eth_tx_queue_setup_t)(struct rte_eth_dev *dev,
630 uint16_t tx_queue_id,
632 unsigned int socket_id,
633 const struct rte_eth_txconf *tx_conf);
634 /**< @internal Setup a transmit queue of an Ethernet device. */
636 typedef void (*vlan_filter_set_t)(struct rte_eth_dev *dev,
639 /**< @internal filtering of a VLAN Tag Identifier by an Ethernet device. */
641 typedef uint16_t (*eth_rx_burst_t)(struct igb_rx_queue *rxq,
642 struct rte_mbuf **rx_pkts,
644 /**< @internal Retrieve input packets from a receive queue of an Ethernet device. */
646 typedef uint16_t (*eth_tx_burst_t)(struct igb_tx_queue *txq,
647 struct rte_mbuf **tx_pkts,
649 /**< @internal Send output packets on a transmit queue of an Ethernet device. */
651 typedef int (*fdir_add_signature_filter_t)(struct rte_eth_dev *dev,
652 struct rte_fdir_filter *fdir_ftr,
654 /**< @internal Setup a new signature filter rule on an Ethernet device */
656 typedef int (*fdir_update_signature_filter_t)(struct rte_eth_dev *dev,
657 struct rte_fdir_filter *fdir_ftr,
659 /**< @internal Update a signature filter rule on an Ethernet device */
661 typedef int (*fdir_remove_signature_filter_t)(struct rte_eth_dev *dev,
662 struct rte_fdir_filter *fdir_ftr);
663 /**< @internal Remove a signature filter rule on an Ethernet device */
665 typedef void (*fdir_infos_get_t)(struct rte_eth_dev *dev,
666 struct rte_eth_fdir *fdir);
667 /**< @internal Get information about fdir status */
669 typedef int (*fdir_add_perfect_filter_t)(struct rte_eth_dev *dev,
670 struct rte_fdir_filter *fdir_ftr,
671 uint16_t soft_id, uint8_t rx_queue,
673 /**< @internal Setup a new perfect filter rule on an Ethernet device */
675 typedef int (*fdir_update_perfect_filter_t)(struct rte_eth_dev *dev,
676 struct rte_fdir_filter *fdir_ftr,
677 uint16_t soft_id, uint8_t rx_queue,
679 /**< @internal Update a perfect filter rule on an Ethernet device */
681 typedef int (*fdir_remove_perfect_filter_t)(struct rte_eth_dev *dev,
682 struct rte_fdir_filter *fdir_ftr,
684 /**< @internal Remove a perfect filter rule on an Ethernet device */
686 typedef int (*fdir_set_masks_t)(struct rte_eth_dev *dev,
687 struct rte_fdir_masks *fdir_masks);
688 /**< @internal Setup flow director masks on an Ethernet device */
690 typedef int (*flow_ctrl_set_t)(struct rte_eth_dev *dev,
691 struct rte_eth_fc_conf *fc_conf);
692 /**< @internal Setup flow control parameter on an Ethernet device */
694 typedef int (*eth_dev_led_on_t)(struct rte_eth_dev *dev);
695 /**< @internal Turn on SW controllable LED on an Ethernet device */
697 typedef int (*eth_dev_led_off_t)(struct rte_eth_dev *dev);
698 /**< @internal Turn off SW controllable LED on an Ethernet device */
700 typedef void (*eth_mac_addr_remove_t)(struct rte_eth_dev *dev, uint32_t index);
701 /**< @internal Remove MAC address from receive address register */
703 typedef void (*eth_mac_addr_add_t)(struct rte_eth_dev *dev,
704 struct ether_addr *mac_addr,
707 /**< @internal Set a MAC address into Receive Address Address Register */
710 * @internal A structure containing the functions exported by an Ethernet driver.
713 eth_dev_configure_t dev_configure; /**< Configure device. */
714 eth_dev_start_t dev_start; /**< Start device. */
715 eth_dev_stop_t dev_stop; /**< Stop device. */
716 eth_dev_close_t dev_close; /**< Close device. */
717 eth_promiscuous_enable_t promiscuous_enable; /**< Promiscuous ON. */
718 eth_promiscuous_disable_t promiscuous_disable;/**< Promiscuous OFF. */
719 eth_allmulticast_enable_t allmulticast_enable;/**< RX multicast ON. */
720 eth_allmulticast_disable_t allmulticast_disable;/**< RX multicast OF. */
721 eth_link_update_t link_update; /**< Get device link state. */
722 eth_stats_get_t stats_get; /**< Get device statistics. */
723 eth_stats_reset_t stats_reset; /**< Reset device statistics. */
724 eth_queue_stats_mapping_set_t queue_stats_mapping_set;
725 /**< Configure per queue stat counter mapping. */
726 eth_dev_infos_get_t dev_infos_get; /**< Get device info. */
727 vlan_filter_set_t vlan_filter_set; /**< Filter VLAN Setup. */
728 eth_rx_queue_setup_t rx_queue_setup;/**< Set up device RX queue.*/
729 eth_tx_queue_setup_t tx_queue_setup;/**< Set up device TX queue.*/
730 eth_dev_led_on_t dev_led_on; /**< Turn on LED. */
731 eth_dev_led_off_t dev_led_off; /**< Turn off LED. */
732 flow_ctrl_set_t flow_ctrl_set; /**< Setup flow control. */
733 eth_mac_addr_remove_t mac_addr_remove; /**< Remove MAC address */
734 eth_mac_addr_add_t mac_addr_add; /**< Add a MAC address */
736 /** Add a signature filter. */
737 fdir_add_signature_filter_t fdir_add_signature_filter;
738 /** Update a signature filter. */
739 fdir_update_signature_filter_t fdir_update_signature_filter;
740 /** Remove a signature filter. */
741 fdir_remove_signature_filter_t fdir_remove_signature_filter;
742 /** Get information about FDIR status. */
743 fdir_infos_get_t fdir_infos_get;
744 /** Add a perfect filter. */
745 fdir_add_perfect_filter_t fdir_add_perfect_filter;
746 /** Update a perfect filter. */
747 fdir_update_perfect_filter_t fdir_update_perfect_filter;
748 /** Remove a perfect filter. */
749 fdir_remove_perfect_filter_t fdir_remove_perfect_filter;
750 /** Setup masks for FDIR filtering. */
751 fdir_set_masks_t fdir_set_masks;
756 * The generic data structure associated with each ethernet device.
758 * Pointers to burst-oriented packet receive and transmit functions are
759 * located at the beginning of the structure, along with the pointer to
760 * where all the data elements for the particular device are stored in shared
761 * memory. This split allows the function pointer and driver data to be per-
762 * process, while the actual configuration data for the device is shared.
765 eth_rx_burst_t rx_pkt_burst; /**< Pointer to PMD receive function. */
766 eth_tx_burst_t tx_pkt_burst; /**< Pointer to PMD transmit function. */
767 struct rte_eth_dev_data *data; /**< Pointer to device data */
768 const struct eth_driver *driver;/**< Driver for this device */
769 struct eth_dev_ops *dev_ops; /**< Functions exported by PMD */
770 struct rte_pci_device *pci_dev; /**< PCI info. supplied by probing */
771 struct rte_eth_dev_cb_list callbacks; /**< User application callbacks */
776 * The data part, with no function pointers, associated with each ethernet device.
778 * This structure is safe to place in shared memory to be common among different
779 * processes in a multi-process configuration.
781 struct rte_eth_dev_data {
782 struct igb_rx_queue **rx_queues; /**< Array of pointers to RX queues. */
783 struct igb_tx_queue **tx_queues; /**< Array of pointers to TX queues. */
784 uint16_t nb_rx_queues; /**< Number of RX queues. */
785 uint16_t nb_tx_queues; /**< Number of TX queues. */
787 void *dev_private; /**< PMD-specific private data */
789 struct rte_eth_link dev_link;
790 /**< Link-level information & status */
792 struct rte_eth_conf dev_conf; /**< Configuration applied to device. */
793 uint16_t max_frame_size; /**< Default is ETHER_MAX_LEN (1518). */
795 uint64_t rx_mbuf_alloc_failed; /**< RX ring mbuf allocation failures. */
796 struct ether_addr* mac_addrs;/**< Device Ethernet Link address. */
797 uint8_t port_id; /**< Device [external] port identifier. */
798 uint8_t promiscuous : 1, /**< RX promiscuous mode ON(1) / OFF(0). */
799 scattered_rx : 1, /**< RX of scattered packets is ON(1) / OFF(0) */
800 all_multicast : 1, /**< RX all multicast mode ON(1) / OFF(0). */
801 dev_started : 1; /**< Device state: STARTED(1) / STOPPED(0). */
806 * The pool of *rte_eth_dev* structures. The size of the pool
807 * is configured at compile-time in the <rte_ethdev.c> file.
809 extern struct rte_eth_dev rte_eth_devices[];
812 * Get the total number of Ethernet devices that have been successfully
813 * initialized by the [matching] Ethernet driver during the PCI probing phase.
814 * All devices whose port identifier is in the range
815 * [0, rte_eth_dev_count() - 1] can be operated on by network applications.
818 * - The total number of usable Ethernet devices.
820 extern uint8_t rte_eth_dev_count(void);
825 * Initialization function of an Ethernet driver invoked for each matching
826 * Ethernet PCI device detected during the PCI probing phase.
829 * The pointer to the [matching] Ethernet driver structure supplied by
830 * the PMD when it registered itself.
832 * The *eth_dev* pointer is the address of the *rte_eth_dev* structure
833 * associated with the matching device and which have been [automatically]
834 * allocated in the *rte_eth_devices* array.
835 * The *eth_dev* structure is supplied to the driver initialization function
836 * with the following fields already initialized:
838 * - *pci_dev*: Holds the pointers to the *rte_pci_device* structure which
839 * contains the generic PCI information of the matching device.
841 * - *dev_private*: Holds a pointer to the device private data structure.
843 * - *max_frame_size*: Contains the default Ethernet maximum frame length
846 * - *port_id*: Contains the port index of the device (actually the index
847 * of the *eth_dev* structure in the *rte_eth_devices* array).
850 * - 0: Success, the device is properly initialized by the driver.
851 * In particular, the driver MUST have set up the *dev_ops* pointer
852 * of the *eth_dev* structure.
853 * - <0: Error code of the device initialization failure.
855 typedef int (*eth_dev_init_t)(struct eth_driver *eth_drv,
856 struct rte_eth_dev *eth_dev);
860 * The structure associated with a PMD Ethernet driver.
862 * Each Ethernet driver acts as a PCI driver and is represented by a generic
863 * *eth_driver* structure that holds:
865 * - An *rte_pci_driver* structure (which must be the first field).
867 * - The *eth_dev_init* function invoked for each matching PCI device.
869 * - The size of the private data to allocate for each matching device.
872 struct rte_pci_driver pci_drv; /**< The PMD is also a PCI driver. */
873 eth_dev_init_t eth_dev_init; /**< Device init function. */
874 unsigned int dev_private_size; /**< Size of device private data. */
879 * A function invoked by the initialization function of an Ethernet driver
880 * to simultaneously register itself as a PCI driver and as an Ethernet
881 * Poll Mode Driver (PMD).
884 * The pointer to the *eth_driver* structure associated with
885 * the Ethernet driver.
887 extern void rte_eth_driver_register(struct eth_driver *eth_drv);
890 * The initialization function of the driver for
891 * Intel(r) IGB Gigabit Ethernet Controller devices.
892 * This function is invoked once at EAL start time.
896 extern int rte_igb_pmd_init(void);
899 * The initialization function of the driver for
900 * Intel(r) EM Gigabit Ethernet Controller devices.
901 * This function is invoked once at EAL start time.
905 extern int rte_em_pmd_init(void);
908 * The initialization function of the driver for 1Gbps Intel IGB_VF
910 * Invoked once at EAL start time.
914 extern int rte_igbvf_pmd_init(void);
917 * The initialization function of the driver for 10Gbps Intel IXGBE
919 * Invoked once at EAL start time.
923 extern int rte_ixgbe_pmd_init(void);
926 * The initialization function of the driver for 10Gbps Intel IXGBE_VF
928 * Invoked once at EAL start time.
932 extern int rte_ixgbevf_pmd_init(void);
935 * The initialization function of *all* supported and enabled drivers.
936 * Right now, the following PMDs are supported:
942 * This function is invoked once at EAL start time.
945 * Error code of the device initialization failure,
946 * -ENODEV if there are no drivers available
947 * (e.g. if all driver config options are = n).
950 int rte_pmd_init_all(void)
954 #ifdef RTE_LIBRTE_IGB_PMD
955 if ((ret = rte_igb_pmd_init()) != 0) {
956 RTE_LOG(ERR, PMD, "Cannot init igb PMD\n");
959 if ((ret = rte_igbvf_pmd_init()) != 0) {
960 RTE_LOG(ERR, PMD, "Cannot init igbvf PMD\n");
963 #endif /* RTE_LIBRTE_IGB_PMD */
965 #ifdef RTE_LIBRTE_EM_PMD
966 if ((ret = rte_em_pmd_init()) != 0) {
967 RTE_LOG(ERR, PMD, "Cannot init em PMD\n");
970 #endif /* RTE_LIBRTE_EM_PMD */
972 #ifdef RTE_LIBRTE_IXGBE_PMD
973 if ((ret = rte_ixgbe_pmd_init()) != 0) {
974 RTE_LOG(ERR, PMD, "Cannot init ixgbe PMD\n");
977 if ((ret = rte_ixgbevf_pmd_init()) != 0) {
978 RTE_LOG(ERR, PMD, "Cannot init ixgbevf PMD\n");
981 #endif /* RTE_LIBRTE_IXGBE_PMD */
984 RTE_LOG(ERR, PMD, "No PMD(s) are configured\n");
989 * Configure an Ethernet device.
990 * This function must be invoked first before any other function in the
991 * Ethernet API. This function can also be re-invoked when a device is in the
995 * The port identifier of the Ethernet device to configure.
997 * The number of receive queues to set up for the Ethernet device.
999 * The number of transmit queues to set up for the Ethernet device.
1001 * The pointer to the configuration data to be used for the Ethernet device.
1002 * The *rte_eth_conf* structure includes:
1003 * - the hardware offload features to activate, with dedicated fields for
1004 * each statically configurable offload hardware feature provided by
1005 * Ethernet devices, such as IP checksum or VLAN tag stripping for
1007 * - the Receive Side Scaling (RSS) configuration when using multiple RX
1010 * Embedding all configuration information in a single data structure
1011 * is the more flexible method that allows the addition of new features
1012 * without changing the syntax of the API.
1014 * - 0: Success, device configured.
1015 * - <0: Error code returned by the driver configuration function.
1017 extern int rte_eth_dev_configure(uint8_t port_id,
1018 uint16_t nb_rx_queue,
1019 uint16_t nb_tx_queue,
1020 const struct rte_eth_conf *eth_conf);
1023 * Allocate and set up a receive queue for an Ethernet device.
1025 * The function allocates a contiguous block of memory for *nb_rx_desc*
1026 * receive descriptors from a memory zone associated with *socket_id*
1027 * and initializes each receive descriptor with a network buffer allocated
1028 * from the memory pool *mb_pool*.
1031 * The port identifier of the Ethernet device.
1032 * @param rx_queue_id
1033 * The index of the receive queue to set up.
1034 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1035 * to rte_eth_dev_configure().
1037 * The number of receive descriptors to allocate for the receive ring.
1039 * The *socket_id* argument is the socket identifier in case of NUMA.
1040 * The value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1041 * the DMA memory allocated for the receive descriptors of the ring.
1043 * The pointer to the configuration data to be used for the receive queue.
1044 * The *rx_conf* structure contains an *rx_thresh* structure with the values
1045 * of the Prefetch, Host, and Write-Back threshold registers of the receive
1048 * The pointer to the memory pool from which to allocate *rte_mbuf* network
1049 * memory buffers to populate each descriptor of the receive ring.
1051 * - 0: Success, receive queue correctly set up.
1052 * - -EINVAL: The size of network buffers which can be allocated from the
1053 * memory pool does not fit the various buffer sizes allowed by the
1054 * device controller.
1055 * - -ENOMEM: Unable to allocate the receive ring descriptors or to
1056 * allocate network memory buffers from the memory pool when
1057 * initializing receive descriptors.
1059 extern int rte_eth_rx_queue_setup(uint8_t port_id, uint16_t rx_queue_id,
1060 uint16_t nb_rx_desc, unsigned int socket_id,
1061 const struct rte_eth_rxconf *rx_conf,
1062 struct rte_mempool *mb_pool);
1065 * Allocate and set up a transmit queue for an Ethernet device.
1068 * The port identifier of the Ethernet device.
1069 * @param tx_queue_id
1070 * The index of the transmit queue to set up.
1071 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1072 * to rte_eth_dev_configure().
1074 * The number of transmit descriptors to allocate for the transmit ring.
1076 * The *socket_id* argument is the socket identifier in case of NUMA.
1077 * Its value can be *SOCKET_ID_ANY* if there is no NUMA constraint for
1078 * the DMA memory allocated for the transmit descriptors of the ring.
1080 * The pointer to the configuration data to be used for the transmit queue.
1081 * The *tx_conf* structure contains the following data:
1082 * - The *tx_thresh* structure with the values of the Prefetch, Host, and
1083 * Write-Back threshold registers of the transmit ring.
1084 * When setting Write-Back threshold to the value greater then zero,
1085 * *tx_rs_thresh* value should be explicitly set to one.
1086 * - The *tx_free_thresh* value indicates the [minimum] number of network
1087 * buffers that must be pending in the transmit ring to trigger their
1088 * [implicit] freeing by the driver transmit function.
1089 * - The *tx_rs_thresh* value indicates the [minimum] number of transmit
1090 * descriptors that must be pending in the transmit ring before setting the
1091 * RS bit on a descriptor by the driver transmit function.
1092 * The *tx_rs_thresh* value should be less or equal then
1093 * *tx_free_thresh* value, and both of them should be less then
1096 * Note that setting *tx_free_thresh* or *tx_rs_thresh* value to 0 forces
1097 * the transmit function to use default values.
1099 * - 0: Success, the transmit queue is correctly set up.
1100 * - -ENOMEM: Unable to allocate the transmit ring descriptors.
1102 extern int rte_eth_tx_queue_setup(uint8_t port_id, uint16_t tx_queue_id,
1103 uint16_t nb_tx_desc, unsigned int socket_id,
1104 const struct rte_eth_txconf *tx_conf);
1107 * Start an Ethernet device.
1109 * The device start step is the last one and consists of setting the configured
1110 * offload features and in starting the transmit and the receive units of the
1112 * On success, all basic functions exported by the Ethernet API (link status,
1113 * receive/transmit, and so on) can be invoked.
1116 * The port identifier of the Ethernet device.
1118 * - 0: Success, Ethernet device started.
1119 * - <0: Error code of the driver device start function.
1121 extern int rte_eth_dev_start(uint8_t port_id);
1124 * Stop an Ethernet device. The device can be restarted with a call to
1125 * rte_eth_dev_start()
1128 * The port identifier of the Ethernet device.
1130 extern void rte_eth_dev_stop(uint8_t port_id);
1133 * Close an Ethernet device. The device cannot be restarted!
1136 * The port identifier of the Ethernet device.
1138 extern void rte_eth_dev_close(uint8_t port_id);
1141 * Enable receipt in promiscuous mode for an Ethernet device.
1144 * The port identifier of the Ethernet device.
1146 extern void rte_eth_promiscuous_enable(uint8_t port_id);
1149 * Disable receipt in promiscuous mode for an Ethernet device.
1152 * The port identifier of the Ethernet device.
1154 extern void rte_eth_promiscuous_disable(uint8_t port_id);
1157 * Return the value of promiscuous mode for an Ethernet device.
1160 * The port identifier of the Ethernet device.
1162 * - (1) if promiscuous is enabled
1163 * - (0) if promiscuous is disabled.
1166 extern int rte_eth_promiscuous_get(uint8_t port_id);
1169 * Enable the receipt of any multicast frame by an Ethernet device.
1172 * The port identifier of the Ethernet device.
1174 extern void rte_eth_allmulticast_enable(uint8_t port_id);
1177 * Disable the receipt of all multicast frames by an Ethernet device.
1180 * The port identifier of the Ethernet device.
1182 extern void rte_eth_allmulticast_disable(uint8_t port_id);
1185 * Return the value of allmulticast mode for an Ethernet device.
1188 * The port identifier of the Ethernet device.
1190 * - (1) if allmulticast is enabled
1191 * - (0) if allmulticast is disabled.
1194 extern int rte_eth_allmulticast_get(uint8_t port_id);
1197 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1198 * or FULL-DUPLEX) of the physical link of an Ethernet device. It might need
1199 * to wait up to 9 seconds in it.
1202 * The port identifier of the Ethernet device.
1204 * A pointer to an *rte_eth_link* structure to be filled with
1205 * the status, the speed and the mode of the Ethernet device link.
1207 extern void rte_eth_link_get(uint8_t port_id, struct rte_eth_link *link);
1210 * Retrieve the status (ON/OFF), the speed (in Mbps) and the mode (HALF-DUPLEX
1211 * or FULL-DUPLEX) of the physical link of an Ethernet device. It is a no-wait
1212 * version of rte_eth_link_get().
1215 * The port identifier of the Ethernet device.
1217 * A pointer to an *rte_eth_link* structure to be filled with
1218 * the status, the speed and the mode of the Ethernet device link.
1220 extern void rte_eth_link_get_nowait(uint8_t port_id,
1221 struct rte_eth_link *link);
1224 * Retrieve the general I/O statistics of an Ethernet device.
1227 * The port identifier of the Ethernet device.
1229 * A pointer to a structure of type *rte_eth_stats* to be filled with
1230 * the values of device counters for the following set of statistics:
1231 * - *ipackets* with the total of successfully received packets.
1232 * - *opackets* with the total of successfully transmitted packets.
1233 * - *ibytes* with the total of successfully received bytes.
1234 * - *obytes* with the total of successfully transmitted bytes.
1235 * - *ierrors* with the total of erroneous received packets.
1236 * - *oerrors* with the total of failed transmitted packets.
1238 extern void rte_eth_stats_get(uint8_t port_id, struct rte_eth_stats *stats);
1241 * Reset the general I/O statistics of an Ethernet device.
1244 * The port identifier of the Ethernet device.
1246 extern void rte_eth_stats_reset(uint8_t port_id);
1249 * Set a mapping for the specified transmit queue to the specified per-queue
1250 * statistics counter.
1253 * The port identifier of the Ethernet device.
1254 * @param tx_queue_id
1255 * The index of the transmit queue for which a queue stats mapping is required.
1256 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1257 * to rte_eth_dev_configure().
1259 * The per-queue packet statistics functionality number that the transmit
1260 * queue is to be assigned.
1261 * The value must be in the range [0, RTE_MAX_ETHPORT_QUEUE_STATS_MAPS - 1].
1263 * Zero if successful. Non-zero otherwise.
1265 extern int rte_eth_dev_set_tx_queue_stats_mapping(uint8_t port_id,
1266 uint16_t tx_queue_id,
1270 * Set a mapping for the specified receive queue to the specified per-queue
1271 * statistics counter.
1274 * The port identifier of the Ethernet device.
1275 * @param rx_queue_id
1276 * The index of the receive queue for which a queue stats mapping is required.
1277 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1278 * to rte_eth_dev_configure().
1280 * The per-queue packet statistics functionality number that the receive
1281 * queue is to be assigned.
1282 * The value must be in the range [0, RTE_MAX_ETHPORT_QUEUE_STATS_MAPS - 1].
1284 * Zero if successful. Non-zero otherwise.
1286 extern int rte_eth_dev_set_rx_queue_stats_mapping(uint8_t port_id,
1287 uint16_t rx_queue_id,
1291 * Retrieve the Ethernet address of an Ethernet device.
1294 * The port identifier of the Ethernet device.
1296 * A pointer to a structure of type *ether_addr* to be filled with
1297 * the Ethernet address of the Ethernet device.
1299 extern void rte_eth_macaddr_get(uint8_t port_id, struct ether_addr *mac_addr);
1302 * Retrieve the contextual information of an Ethernet device.
1305 * The port identifier of the Ethernet device.
1307 * A pointer to a structure of type *rte_eth_dev_info* to be filled with
1308 * the contextual information of the Ethernet device.
1310 extern void rte_eth_dev_info_get(uint8_t port_id,
1311 struct rte_eth_dev_info *dev_info);
1314 * Enable/Disable hardware filtering by an Ethernet device of received
1315 * VLAN packets tagged with a given VLAN Tag Identifier.
1318 * The port identifier of the Ethernet device.
1320 * The VLAN Tag Identifier whose filtering must be enabled or disabled.
1322 * If > 0, enable VLAN filtering of VLAN packets tagged with *vlan_id*.
1323 * Otherwise, disable VLAN filtering of VLAN packets tagged with *vlan_id*.
1325 * - (0) if successful.
1326 * - (-ENOSUP) if hardware-assisted VLAN filtering not configured.
1327 * - (-ENODEV) if *port_id* invalid.
1328 * - (-ENOSYS) if VLAN filtering on *port_id* disabled.
1329 * - (-EINVAL) if *vlan_id* > 4095.
1331 extern int rte_eth_dev_vlan_filter(uint8_t port_id, uint16_t vlan_id, int on);
1335 * Retrieve a burst of input packets from a receive queue of an Ethernet
1336 * device. The retrieved packets are stored in *rte_mbuf* structures whose
1337 * pointers are supplied in the *rx_pkts* array.
1339 * The rte_eth_rx_burst() function loops, parsing the RX ring of the
1340 * receive queue, up to *nb_pkts* packets, and for each completed RX
1341 * descriptor in the ring, it performs the following operations:
1343 * - Initialize the *rte_mbuf* data structure associated with the
1344 * RX descriptor according to the information provided by the NIC into
1345 * that RX descriptor.
1347 * - Store the *rte_mbuf* data structure into the next entry of the
1350 * - Replenish the RX descriptor with a new *rte_mbuf* buffer
1351 * allocated from the memory pool associated with the receive queue at
1352 * initialization time.
1354 * When retrieving an input packet that was scattered by the controller
1355 * into multiple receive descriptors, the rte_eth_rx_burst() function
1356 * appends the associated *rte_mbuf* buffers to the first buffer of the
1359 * The rte_eth_rx_burst() function returns the number of packets
1360 * actually retrieved, which is the number of *rte_mbuf* data structures
1361 * effectively supplied into the *rx_pkts* array.
1362 * A return value equal to *nb_pkts* indicates that the RX queue contained
1363 * at least *rx_pkts* packets, and this is likely to signify that other
1364 * received packets remain in the input queue. Applications implementing
1365 * a "retrieve as much received packets as possible" policy can check this
1366 * specific case and keep invoking the rte_eth_rx_burst() function until
1367 * a value less than *nb_pkts* is returned.
1369 * This receive method has the following advantages:
1371 * - It allows a run-to-completion network stack engine to retrieve and
1372 * to immediately process received packets in a fast burst-oriented
1373 * approach, avoiding the overhead of unnecessary intermediate packet
1374 * queue/dequeue operations.
1376 * - Conversely, it also allows an asynchronous-oriented processing
1377 * method to retrieve bursts of received packets and to immediately
1378 * queue them for further parallel processing by another logical core,
1379 * for instance. However, instead of having received packets being
1380 * individually queued by the driver, this approach allows the invoker
1381 * of the rte_eth_rx_burst() function to queue a burst of retrieved
1382 * packets at a time and therefore dramatically reduce the cost of
1383 * enqueue/dequeue operations per packet.
1385 * - It allows the rte_eth_rx_burst() function of the driver to take
1386 * advantage of burst-oriented hardware features (CPU cache,
1387 * prefetch instructions, and so on) to minimize the number of CPU
1388 * cycles per packet.
1390 * To summarize, the proposed receive API enables many
1391 * burst-oriented optimizations in both synchronous and asynchronous
1392 * packet processing environments with no overhead in both cases.
1394 * The rte_eth_rx_burst() function does not provide any error
1395 * notification to avoid the corresponding overhead. As a hint, the
1396 * upper-level application might check the status of the device link once
1397 * being systematically returned a 0 value for a given number of tries.
1400 * The port identifier of the Ethernet device.
1402 * The index of the receive queue from which to retrieve input packets.
1403 * The value must be in the range [0, nb_rx_queue - 1] previously supplied
1404 * to rte_eth_dev_configure().
1406 * The address of an array of pointers to *rte_mbuf* structures that
1407 * must be large enough to store *nb_pkts* pointers in it.
1409 * The maximum number of packets to retrieve.
1411 * The number of packets actually retrieved, which is the number
1412 * of pointers to *rte_mbuf* structures effectively supplied to the
1415 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
1416 extern uint16_t rte_eth_rx_burst(uint8_t port_id, uint16_t queue_id,
1417 struct rte_mbuf **rx_pkts, uint16_t nb_pkts);
1419 static inline uint16_t
1420 rte_eth_rx_burst(uint8_t port_id, uint16_t queue_id,
1421 struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
1423 struct rte_eth_dev *dev;
1425 dev = &rte_eth_devices[port_id];
1426 return (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id], rx_pkts, nb_pkts);
1431 * Send a burst of output packets on a transmit queue of an Ethernet device.
1433 * The rte_eth_tx_burst() function is invoked to transmit output packets
1434 * on the output queue *queue_id* of the Ethernet device designated by its
1436 * The *nb_pkts* parameter is the number of packets to send which are
1437 * supplied in the *tx_pkts* array of *rte_mbuf* structures.
1438 * The rte_eth_tx_burst() function loops, sending *nb_pkts* packets,
1439 * up to the number of transmit descriptors available in the TX ring of the
1441 * For each packet to send, the rte_eth_tx_burst() function performs
1442 * the following operations:
1444 * - Pick up the next available descriptor in the transmit ring.
1446 * - Free the network buffer previously sent with that descriptor, if any.
1448 * - Initialize the transmit descriptor with the information provided
1449 * in the *rte_mbuf data structure.
1451 * In the case of a segmented packet composed of a list of *rte_mbuf* buffers,
1452 * the rte_eth_tx_burst() function uses several transmit descriptors
1455 * The rte_eth_tx_burst() function returns the number of packets it
1456 * actually sent. A return value equal to *nb_pkts* means that all packets
1457 * have been sent, and this is likely to signify that other output packets
1458 * could be immediately transmitted again. Applications that implement a
1459 * "send as many packets to transmit as possible" policy can check this
1460 * specific case and keep invoking the rte_eth_tx_burst() function until
1461 * a value less than *nb_pkts* is returned.
1463 * It is the responsibility of the rte_eth_tx_burst() function to
1464 * transparently free the memory buffers of packets previously sent.
1465 * This feature is driven by the *tx_free_thresh* value supplied to the
1466 * rte_eth_dev_configure() function at device configuration time.
1467 * When the number of previously sent packets reached the "minimum transmit
1468 * packets to free" threshold, the rte_eth_tx_burst() function must
1469 * [attempt to] free the *rte_mbuf* buffers of those packets whose
1470 * transmission was effectively completed.
1473 * The port identifier of the Ethernet device.
1475 * The index of the transmit queue through which output packets must be
1477 * The value must be in the range [0, nb_tx_queue - 1] previously supplied
1478 * to rte_eth_dev_configure().
1480 * The address of an array of *nb_pkts* pointers to *rte_mbuf* structures
1481 * which contain the output packets.
1483 * The maximum number of packets to transmit.
1485 * The number of output packets actually stored in transmit descriptors of
1486 * the transmit ring. The return value can be less than the value of the
1487 * *tx_pkts* parameter when the transmit ring is full or has been filled up.
1489 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
1490 extern uint16_t rte_eth_tx_burst(uint8_t port_id, uint16_t queue_id,
1491 struct rte_mbuf **tx_pkts, uint16_t nb_pkts);
1493 static inline uint16_t
1494 rte_eth_tx_burst(uint8_t port_id, uint16_t queue_id,
1495 struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
1497 struct rte_eth_dev *dev;
1499 dev = &rte_eth_devices[port_id];
1500 return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id], tx_pkts, nb_pkts);
1505 * Setup a new signature filter rule on an Ethernet device
1508 * The port identifier of the Ethernet device.
1509 * @param fdir_filter
1510 * The pointer to the fdir filter structure describing the signature filter
1512 * The *rte_fdir_filter* structure includes the values of the different fields
1513 * to match: source and destination IP addresses, vlan id, flexbytes, source
1514 * and destination ports, and so on.
1516 * The index of the RX queue where to store RX packets matching the added
1517 * signature filter defined in fdir_filter.
1519 * - (0) if successful.
1520 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1521 * - (-ENODEV) if *port_id* invalid.
1522 * - (-ENOSYS) if the FDIR mode is not configured in signature mode
1524 * - (-EINVAL) if the fdir_filter information is not correct.
1526 int rte_eth_dev_fdir_add_signature_filter(uint8_t port_id,
1527 struct rte_fdir_filter *fdir_filter,
1531 * Update a signature filter rule on an Ethernet device.
1532 * If the rule doesn't exits, it is created.
1535 * The port identifier of the Ethernet device.
1537 * The pointer to the structure describing the signature filter rule.
1538 * The *rte_fdir_filter* structure includes the values of the different fields
1539 * to match: source and destination IP addresses, vlan id, flexbytes, source
1540 * and destination ports, and so on.
1542 * The index of the RX queue where to store RX packets matching the added
1543 * signature filter defined in fdir_ftr.
1545 * - (0) if successful.
1546 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1547 * - (-ENODEV) if *port_id* invalid.
1548 * - (-ENOSYS) if the flow director mode is not configured in signature mode
1550 * - (-EINVAL) if the fdir_filter information is not correct.
1552 int rte_eth_dev_fdir_update_signature_filter(uint8_t port_id,
1553 struct rte_fdir_filter *fdir_ftr,
1557 * Remove a signature filter rule on an Ethernet device.
1560 * The port identifier of the Ethernet device.
1562 * The pointer to the structure describing the signature filter rule.
1563 * The *rte_fdir_filter* structure includes the values of the different fields
1564 * to match: source and destination IP addresses, vlan id, flexbytes, source
1565 * and destination ports, and so on.
1567 * - (0) if successful.
1568 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1569 * - (-ENODEV) if *port_id* invalid.
1570 * - (-ENOSYS) if the flow director mode is not configured in signature mode
1572 * - (-EINVAL) if the fdir_filter information is not correct.
1574 int rte_eth_dev_fdir_remove_signature_filter(uint8_t port_id,
1575 struct rte_fdir_filter *fdir_ftr);
1578 * Retrieve the flow director information of an Ethernet device.
1581 * The port identifier of the Ethernet device.
1583 * A pointer to a structure of type *rte_eth_dev_fdir* to be filled with
1584 * the flow director information of the Ethernet device.
1586 * - (0) if successful.
1587 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1588 * - (-ENODEV) if *port_id* invalid.
1589 * - (-ENOSYS) if the flow director mode is not configured on *port_id*.
1591 int rte_eth_dev_fdir_get_infos(uint8_t port_id, struct rte_eth_fdir *fdir);
1594 * Add a new perfect filter rule on an Ethernet device.
1597 * The port identifier of the Ethernet device.
1598 * @param fdir_filter
1599 * The pointer to the structure describing the perfect filter rule.
1600 * The *rte_fdir_filter* structure includes the values of the different fields
1601 * to match: source and destination IP addresses, vlan id, flexbytes, source
1602 * and destination ports, and so on.
1603 * IPv6 are not supported.
1605 * The 16-bit value supplied in the field hash.fdir.id of mbuf for RX
1606 * packets matching the perfect filter.
1608 * The index of the RX queue where to store RX packets matching the added
1609 * perfect filter defined in fdir_filter.
1611 * If drop is set to 1, matching RX packets are stored into the RX drop
1612 * queue defined in the rte_fdir_conf.
1614 * - (0) if successful.
1615 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1616 * - (-ENODEV) if *port_id* invalid.
1617 * - (-ENOSYS) if the flow director mode is not configured in perfect mode
1619 * - (-EINVAL) if the fdir_filter information is not correct.
1621 int rte_eth_dev_fdir_add_perfect_filter(uint8_t port_id,
1622 struct rte_fdir_filter *fdir_filter,
1623 uint16_t soft_id, uint8_t rx_queue,
1627 * Update a perfect filter rule on an Ethernet device.
1628 * If the rule doesn't exits, it is created.
1631 * The port identifier of the Ethernet device.
1632 * @param fdir_filter
1633 * The pointer to the structure describing the perfect filter rule.
1634 * The *rte_fdir_filter* structure includes the values of the different fields
1635 * to match: source and destination IP addresses, vlan id, flexbytes, source
1636 * and destination ports, and so on.
1637 * IPv6 are not supported.
1639 * The 16-bit value supplied in the field hash.fdir.id of mbuf for RX
1640 * packets matching the perfect filter.
1642 * The index of the RX queue where to store RX packets matching the added
1643 * perfect filter defined in fdir_filter.
1645 * If drop is set to 1, matching RX packets are stored into the RX drop
1646 * queue defined in the rte_fdir_conf.
1648 * - (0) if successful.
1649 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1650 * - (-ENODEV) if *port_id* invalid.
1651 * - (-ENOSYS) if the flow director mode is not configured in perfect mode
1653 * - (-EINVAL) if the fdir_filter information is not correct.
1655 int rte_eth_dev_fdir_update_perfect_filter(uint8_t port_id,
1656 struct rte_fdir_filter *fdir_filter,
1657 uint16_t soft_id, uint8_t rx_queue,
1661 * Remove a perfect filter rule on an Ethernet device.
1664 * The port identifier of the Ethernet device.
1665 * @param fdir_filter
1666 * The pointer to the structure describing the perfect filter rule.
1667 * The *rte_fdir_filter* structure includes the values of the different fields
1668 * to match: source and destination IP addresses, vlan id, flexbytes, source
1669 * and destination ports, and so on.
1670 * IPv6 are not supported.
1672 * The soft_id value provided when adding/updating the removed filter.
1674 * - (0) if successful.
1675 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1676 * - (-ENODEV) if *port_id* invalid.
1677 * - (-ENOSYS) if the flow director mode is not configured in perfect mode
1679 * - (-EINVAL) if the fdir_filter information is not correct.
1681 int rte_eth_dev_fdir_remove_perfect_filter(uint8_t port_id,
1682 struct rte_fdir_filter *fdir_filter,
1685 * Configure globally the masks for flow director mode for an Ethernet device.
1686 * For example, the device can match packets with only the first 24 bits of
1687 * the IPv4 source address.
1689 * The following fields can be masked: IPv4 addresses and L4 port numbers.
1690 * The following fields can be either enabled or disabled completely for the
1691 * matching functionality: VLAN ID tag; VLAN Priority + CFI bit; Flexible 2-byte
1693 * IPv6 masks are not supported.
1695 * All filters must comply with the masks previously configured.
1696 * For example, with a mask equal to 255.255.255.0 for the source IPv4 address,
1697 * all IPv4 filters must be created with a source IPv4 address that fits the
1700 * This function flushes all filters that have been previously added in
1704 * The port identifier of the Ethernet device.
1706 * The pointer to the fdir mask structure describing relevant headers fields
1707 * and relevant bits to use when matching packets addresses and ports.
1708 * IPv6 masks are not supported.
1710 * - (0) if successful.
1711 * - (-ENOTSUP) if hardware doesn't support flow director mode.
1712 * - (-ENODEV) if *port_id* invalid.
1713 * - (-ENOSYS) if the flow director mode is not configured in perfect
1714 * mode on *port_id*.
1715 * - (-EINVAL) if the fdir_filter information is not correct
1717 int rte_eth_dev_fdir_set_masks(uint8_t port_id,
1718 struct rte_fdir_masks *fdir_mask);
1721 * The eth device event type for interrupt, and maybe others in the future.
1723 enum rte_eth_event_type {
1724 RTE_ETH_EVENT_UNKNOWN, /**< unknown event type */
1725 RTE_ETH_EVENT_INTR_LSC, /**< lsc interrupt event */
1726 RTE_ETH_EVENT_MAX /**< max value of this enum */
1729 typedef void (*rte_eth_dev_cb_fn)(uint8_t port_id, \
1730 enum rte_eth_event_type event, void *cb_arg);
1731 /**< user application callback to be registered for interrupts */
1736 * Register a callback function for specific port id.
1743 * User supplied callback function to be called.
1745 * Pointer to the parameters for the registered callback.
1748 * - On success, zero.
1749 * - On failure, a negative value.
1751 int rte_eth_dev_callback_register(uint8_t port_id,
1752 enum rte_eth_event_type event,
1753 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
1756 * Unregister a callback function for specific port id.
1763 * User supplied callback function to be called.
1765 * Pointer to the parameters for the registered callback. -1 means to
1766 * remove all for the same callback address and same event.
1769 * - On success, zero.
1770 * - On failure, a negative value.
1772 int rte_eth_dev_callback_unregister(uint8_t port_id,
1773 enum rte_eth_event_type event,
1774 rte_eth_dev_cb_fn cb_fn, void *cb_arg);
1777 * @internal Executes all the user application registered callbacks for
1778 * the specific device. It is for DPDK internal user only. User
1779 * application should not call it directly.
1782 * Pointer to struct rte_eth_dev.
1784 * Eth device interrupt event type.
1789 void _rte_eth_dev_callback_process(struct rte_eth_dev *dev,
1790 enum rte_eth_event_type event);
1793 * Turn on the LED on the Ethernet device.
1794 * This function turns on the LED on the Ethernet device.
1797 * The port identifier of the Ethernet device.
1799 * - (0) if successful.
1800 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
1802 * - (-ENODEV) if *port_id* invalid.
1804 int rte_eth_led_on(uint8_t port_id);
1807 * Turn off the LED on the Ethernet device.
1808 * This function turns off the LED on the Ethernet device.
1811 * The port identifier of the Ethernet device.
1813 * - (0) if successful.
1814 * - (-ENOTSUP) if underlying hardware OR driver doesn't support
1816 * - (-ENODEV) if *port_id* invalid.
1818 int rte_eth_led_off(uint8_t port_id);
1821 * Configure the Ethernet link flow control for Ethernet device
1824 * The port identifier of the Ethernet device.
1826 * The pointer to the structure of the flow control parameters.
1828 * - (0) if successful.
1829 * - (-ENOTSUP) if hardware doesn't support flow control mode.
1830 * - (-ENODEV) if *port_id* invalid.
1831 * - (-EINVAL) if bad parameter
1832 * - (-EIO) if flow control setup failure
1834 int rte_eth_dev_flow_ctrl_set(uint8_t port_id,
1835 struct rte_eth_fc_conf *fc_conf);
1838 * Add a MAC address to an internal array of addresses used to enable whitelist
1839 * filtering to accept packets only if the destination MAC address matches.
1842 * The port identifier of the Ethernet device.
1844 * The MAC address to add.
1846 * VMDq pool index to associate address with (if VMDq is enabled). If VMDq is
1847 * not enabled, this should be set to 0.
1849 * - (0) if successfully added or *mac_addr" was already added.
1850 * - (-ENOTSUP) if hardware doesn't support this feature.
1851 * - (-ENODEV) if *port* is invalid.
1852 * - (-ENOSPC) if no more MAC addresses can be added.
1853 * - (-EINVAL) if MAC address is invalid.
1855 int rte_eth_dev_mac_addr_add(uint8_t port, struct ether_addr *mac_addr,
1859 * Remove a MAC address from the internal array of addresses.
1862 * The port identifier of the Ethernet device.
1864 * MAC address to remove.
1866 * - (0) if successful, or *mac_addr* didn't exist.
1867 * - (-ENOTSUP) if hardware doesn't support.
1868 * - (-ENODEV) if *port* invalid.
1869 * - (-EADDRINUSE) if attempting to remove the default MAC address
1871 int rte_eth_dev_mac_addr_remove(uint8_t port, struct ether_addr *mac_addr);
1878 #endif /* _RTE_ETHDEV_H_ */