1 .\" SPDX-License-Identifier: BSD-2-Clause
3 .\" Copyright (c) 2015-2022 Amazon.com, Inc. or its affiliates.
4 .\" All rights reserved.
6 .\" Redistribution and use in source and binary forms, with or without
7 .\" modification, are permitted provided that the following conditions
10 .\" 1. Redistributions of source code must retain the above copyright
11 .\" notice, this list of conditions and the following disclaimer.
13 .\" 2. 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
18 .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 .\" "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 .\" LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 .\" A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 .\" OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 .\" SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 .\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 .\" DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 .\" THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 .\" (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 .\" OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
35 .Nd "FreeBSD kernel driver for Elastic Network Adapter (ENA) family"
37 To compile this driver into the kernel,
38 place the following line in the
39 kernel configuration file:
40 .Bd -ragged -offset indent
44 Alternatively, to load the driver as a
45 module at boot time, place the following line in
47 .Bd -literal -offset indent
51 The ENA is a networking interface designed to make good use of modern CPU
52 features and system architectures.
54 The ENA device exposes a lightweight management interface with a
55 minimal set of memory mapped registers and extendable command set
56 through an Admin Queue.
58 The driver supports a range of ENA devices, is link-speed independent
59 (i.e., the same driver is used for 10GbE, 25GbE, 40GbE, etc.), and has
60 a negotiated and extendable feature set.
62 Some ENA devices support SR-IOV.
63 This driver is used for both the SR-IOV Physical Function (PF) and Virtual
64 Function (VF) devices.
66 The ENA devices enable high speed and low overhead network traffic
67 processing by providing multiple Tx/Rx queue pairs (the maximum number
68 is advertised by the device via the Admin Queue), a dedicated MSI-X
69 interrupt vector per Tx/Rx queue pair, and CPU cacheline optimized
72 When RSS is enabled, each Tx/Rx queue pair is bound to a corresponding
73 CPU core and its NUMA domain.
74 The order of those bindings is based on the RSS bucket mapping.
75 For builds with RSS support disabled, the
76 CPU and NUMA management is left to the kernel.
77 Receive-side scaling (RSS) is supported for multi-core scaling.
81 driver and its corresponding devices implement health
82 monitoring mechanisms such as watchdog, enabling the device and driver
83 to recover in a manner transparent to the application, as well as
86 Some of the ENA devices support a working mode called Low-latency
87 Queue (LLQ), which saves several more microseconds.
91 framework is provided by the
94 Kernel must be built with the DEV_NETMAP option to be able to use this feature.
96 Supported PCI vendor ID/device IDs:
102 1d0f:1ec2 - ENA PF with LLQ support
106 1d0f:ec21 - ENA VF with LLQ support
111 driver's behavior can be changed using run-time or boot-time sysctl
113 The boot-time arguments can be set at the
115 prompt before booting the kernel, or stored in the
117 The run-time arguments can be set using the
122 .Bl -tag -width indent
123 .It Va hw.ena.enable_9k_mbufs
124 Use 9k mbufs for the Rx descriptors.
126 If the node value is set to 1, 9k mbufs will be used for the Rx buffers.
127 If set to 0, page size mbufs will be used instead.
129 Using 9k buffers for Rx can improve Rx throughput, but in low memory conditions
130 it might increase allocation time, as the system has to look for 3 contiguous
132 This can further lead to OS instability, together with ENA driver reset and NVMe
134 If network performance is critical and memory capacity is sufficient, the 9k
136 .It Va hw.ena.force_large_llq_headers
137 Force the driver to use large LLQ headers (224 bytes).
139 If the node value is set to 0, the regular size LLQ header will be used, which
141 In some cases, the packet header can be bigger than this (for example -
142 IPv6 with multiple extensions).
143 In such a situation, the large LLQ headers should be used by setting this node
145 This will take effect only if the device supports both LLQ and large LLQ
147 Otherwise, it will fallback to the no LLQ mode or regular header size.
149 Increasing LLQ header size reduces the size of the Tx queue by half, so it may
150 affect the number of dropped Tx packets.
154 .Bl -tag -width indent
155 .It Va hw.ena.log_level
156 Controls extra logging verbosity of the driver.
158 The higher the logging level, the more logs will be printed out. 0 means all
159 extra logs are disabled and only error logs will be printed out.
160 Default value (2) reports errors, warnings and is verbose about driver
163 The possible flags are:
167 0 - ENA_ERR - Enable driver error messages and ena_com error logs.
169 1 - ENA_WARN - Enable logs for non-critical errors.
171 2 - ENA_INFO - Make the driver more verbose about its actions.
173 3 - ENA_DBG - Enable debug logs.
176 NOTE: In order to enable logging on the Tx/Rx data path, driver must be compiled
177 with ENA_LOG_IO_ENABLE compilation flag.
180 To enable logs for errors and warnings, the following command should be used:
181 .Bd -literal -offset indent
182 sysctl hw.ena.log_level=1
184 .It Va dev.ena.X.io_queues_nb
185 Number of the currently allocated and used IO queues.
186 The default is max_num_io_queues.
187 Controls the number of IO queue pairs (Tx/Rx). As this call has to reallocate
188 the queues, it will reset the interface and restart all the queues - this means
189 that everything, which was currently held in the queue, will be lost, leading to
190 potential packet drops.
192 This call can fail if the system isn't able to provide the driver with enough
194 In that situation, the driver will try to revert the previous number of the IO
196 If this also fails, the device reset will be triggered.
199 To use only 2 Tx and Rx queues for the device ena1, the following command should
201 .Bd -literal -offset indent
202 sysctl dev.ena.1.io_queues_nb=2
204 .It Va dev.ena.X.rx_queue_size
205 Size of the Rx queue.
207 Controls the number of IO descriptors for each Rx queue.
208 The user may want to increase the Rx queue size if they observe a high number of
209 Rx drops in the driver's statistics.
210 For performance reasons, the Rx queue size must be a power of 2.
212 This call can fail if the system isn't able to provide the driver with enough
214 In that situation, the driver will try to revert to the previous number of the
216 If this also fails, the device reset will be triggered.
219 To increase Rx ring size to 8K descriptors for the device ena0, the following
220 command should be used:
221 .Bd -literal -offset indent
222 sysctl dev.ena.0.rx_queue_size=8192
224 .It Va dev.ena.X.buf_ring_size
225 Size of the Tx buffer ring (drbr).
227 Input must be a power of 2.
228 Controls the number of mbufs that can be held in the Tx buffer ring.
229 The drbr is used as a multiple-producer, single-consumer lockless ring for
230 buffering extra mbufs coming from the stack in case the Tx procedure is busy
231 sending the packets, or the Tx ring is full.
232 Increasing the size of the buffer ring may reduce the number of Tx packets being
233 dropped in case of a big Tx burst, which cannot be handled by the IO queue
235 Each Tx queue has its own drbr.
237 It is recommended to keep the drbr with at least the default value, but in case
238 the system lacks the resources, it can be reduced.
239 This call can fail if the system is not able to provide the driver with enough
241 In that situation, the driver will try to revert to the previous number of the
242 drbr and trigger the device reset.
245 To set drbr size for interface ena0 to 2048, the following command should
247 .Bd -literal -offset indent
248 sysctl dev.ena.0.buf_ring_size=2048
250 .It Va dev.ena.X.eni_metrics.sample_interval
251 Interval in seconds for updating ENI metrics.
253 Determines how often (if ever) the ENI metrics should be updated.
254 The ENI metrics are being updated asynchronously in a timer service in order to
255 avoid admin queue overload by sysctl node reading.
256 The value in this node controls the interval between issuing admin commands to
257 the device, which will update the ENI metrics values.
259 If some application is periodically monitoring the eni_metrics, then the ENI
260 metrics interval can be adjusted accordingly.
261 Value 0 turns off the update completely.
262 Value 1 is the minimum interval and is equal to 1 second.
263 The maximum allowed update interval is 1 hour.
266 To update ENI metrics for the device ena1 every 10 seconds, the following
267 command should be used:
268 .Bd -literal -offset indent
269 sysctl dev.ena.1.eni_metrics.sample_interval=10
271 .It Va dev.ena.X.rss.indir_table_size
272 RSS indirection table size.
274 Returns the number of entries in the RSS indirection table.
277 To read the RSS indirection table size, the following command should be used:
278 .Bd -literal -offset indent
279 sysctl dev.ena.0.rss.indir_table_size
281 .It Va dev.ena.X.rss.indir_table
282 RSS indirection table mapping.
283 The default is x:y key-pairs of indir_table_size length.
284 Updates selected indices of the RSS indirection table.
286 The entry string consists of one or more x:y keypairs, where x stands for
287 the table index and y for its new value.
288 Table indices that don't need to be
289 updated can be omitted from the string and will retain their existing values.
291 If an index is entered more than once, the last value is used.
294 To update two selected indices in the RSS indirection table, e.g. setting index
295 0 to queue 5 and then index 5 to queue 0, the following command should be used:
296 .Bd -literal -offset indent
297 sysctl dev.ena.0.rss.indir_table="0:5 5:0"
299 .It Va dev.ena.X.rss.key
301 The default is 40 bytes long randomly generated hash key.
302 Controls the RSS Toeplitz hash algorithm key value.
304 Only available when driver compiled without the kernel side RSS support.
307 To change the RSS hash key value to
309 0x6d, 0x5a, 0x56, 0xda, 0x25, 0x5b, 0x0e, 0xc2,
311 0x41, 0x67, 0x25, 0x3d, 0x43, 0xa3, 0x8f, 0xb0,
313 0xd0, 0xca, 0x2b, 0xcb, 0xae, 0x7b, 0x30, 0xb4,
315 0x77, 0xcb, 0x2d, 0xa3, 0x80, 0x30, 0xf2, 0x0c,
317 0x6a, 0x42, 0xb7, 0x3b, 0xbe, 0xac, 0x01, 0xfa
319 the following command should be used:
320 .Bd -literal -offset indent
321 sysctl dev.ena.0.rss.key=6d5a56da255b0ec24167253d43a38fb0d0ca2bcbae7b30b477cb2da38030f20c6a42b73bbeac01fa
325 .Ss Device initialization phase
327 .It ena%d: failed to init mmio read less
329 Error occurred during initialization of the mmio register read request.
330 .It ena%d: Can not reset device
332 Device could not be reset.
334 Device may not be responding or is already during reset.
335 .It ena%d: device version is too low
337 Version of the controller is too old and it is not supported by the driver.
338 .It ena%d: Invalid dma width value %d
340 The controller is unable to request dma transaction width.
342 Device stopped responding or it demanded invalid value.
343 .It ena%d: Can not initialize ena admin queue with device
345 Initialization of the Admin Queue failed.
347 Device may not be responding or there was a problem with initialization of
349 .It ena%d: Cannot get attribute for ena device rc: %d
351 Failed to get attributes of the device from the controller.
352 .It ena%d: Cannot configure aenq groups rc: %d
354 Errors occurred when trying to configure AENQ groups.
356 .Ss Driver initialization/shutdown phase
358 .It ena%d: PCI resource allocation failed!
359 .It ena%d: failed to pmap registers bar
360 .It ena%d: can not allocate ifnet structure
361 .It ena%d: Error with network interface setup
362 .It ena%d: Failed to enable and set the admin interrupts
363 .It ena%d: Error, MSI-X is already enabled
364 .It ena%d: Failed to enable MSIX, vectors %d rc %d
365 .It ena%d: Not enough number of MSI-X allocated: %d
366 .It ena%d: Error with MSI-X enablement
367 .It ena%d: could not allocate irq vector: %d
368 .It ena%d: unable to allocate bus resource: registers!
369 .It ena%d: unable to allocate bus resource: msix!
371 Resource allocation failed when initializing the device.
373 Driver will not be attached.
374 .It ena%d: ENA device init failed (err: %d)
375 .It ena%d: Cannot initialize device
377 Device initialization failed.
379 Driver will not be attached.
380 .It ena%d: failed to register interrupt handler for irq %ju: %d
382 Error occurred when trying to register Admin Queue interrupt handler.
383 .It ena%d: Cannot setup mgmnt queue intr
385 Error occurred during configuration of the Admin Queue interrupts.
386 .It ena%d: Enable MSI-X failed
388 Configuration of the MSI-X for Admin Queue failed.
390 There could be lack of resources or interrupts could not have been configured.
392 Driver will not be attached.
393 .It ena%d: VLAN is in use, detach first
395 VLANs are being used when trying to detach the driver.
397 VLANs must be detached first and then detach routine have to be called again.
398 .It ena%d: Unmapped RX DMA tag associations
399 .It ena%d: Unmapped TX DMA tag associations
401 Error occurred when trying to destroy RX/TX DMA tag.
402 .It ena%d: Cannot init indirect table
403 .It ena%d: Cannot fill indirect table
404 .It ena%d: Cannot fill hash function
405 .It ena%d: Cannot fill hash control
406 .It ena%d: WARNING: RSS was not properly initialized, it will affect bandwidth
408 Error occurred during initialization of one of RSS resources.
410 The device will work with reduced performance because all RX packets will be
411 passed to queue 0 and there will be no hash information.
412 .It ena%d: LLQ is not supported. Fallback to host mode policy.
413 .It ena%d: Failed to configure the device mode. Fallback to host mode policy.
414 .It ena%d: unable to allocate LLQ bar resource. Fallback to host mode policy.
416 Error occurred during Low-latency Queue mode setup.
418 The device will work, but without the LLQ performance gain.
419 .It ena%d: failed to enable write combining.
421 Error occurred while setting the Write Combining mode, required for the LLQ.
422 .It ena%d: failed to tear down irq: %d
423 .It ena%d: dev has no parent while releasing res for irq: %d
424 Release of the interrupts failed.
426 .Ss Additional diagnostic
428 .It ena%d: Invalid MTU setting. new_mtu: %d max_mtu: %d min mtu: %d
430 Requested MTU value is not supported and will not be set.
431 .It ena%d: Failed to set MTU to %d
433 This message appears when either MTU change feature is not supported, or device
434 communication error has occurred.
435 .It ena%d: Keep alive watchdog timeout.
437 Device stopped responding and will be reset.
438 .It ena%d: Found a Tx that wasn't completed on time, qid %d, index %d.
440 Packet was pushed to the NIC but not sent within given time limit.
442 It may be caused by hang of the IO queue.
443 .It ena%d: The number of lost tx completion is above the threshold (%d > %d). Reset the device
445 If too many Tx weren't completed on time the device is going to be reset.
447 It may be caused by hanged queue or device.
448 .It ena%d: Trigger reset is on
450 Device will be reset.
452 Reset is triggered either by watchdog or if too many TX packets were not
454 .It ena%d: device reset scheduled but trigger_reset is off
456 Reset task has been triggered, but the driver did not request it.
458 Device reset will not be performed.
459 .It ena%d: Device reset failed
461 Error occurred while trying to reset the device.
462 .It ena%d: Cannot initialize device
463 .It ena%d: Error, mac address are different
464 .It ena%d: Error, device max mtu is smaller than ifp MTU
465 .It ena%d: Validation of device parameters failed
466 .It ena%d: Enable MSI-X failed
467 .It ena%d: Failed to create I/O queues
468 .It ena%d: Reset attempt failed. Can not reset the device
470 Error occurred while trying to restore the device after reset.
471 .It ena%d: Device reset completed successfully, Driver info: %s
473 Device has been correctly restored after reset and is ready to use.
474 .It ena%d: Allocation for Tx Queue %u failed
475 .It ena%d: Allocation for Rx Queue %u failed
476 .It ena%d: Unable to create Rx DMA map for buffer %d
477 .It ena%d: Failed to create io TX queue #%d rc: %d
478 .It ena%d: Failed to get TX queue handlers. TX queue num %d rc: %d
479 .It ena%d: Failed to create io RX queue[%d] rc: %d
480 .It ena%d: Failed to get RX queue handlers. RX queue num %d rc: %d
481 .It ena%d: could not allocate irq vector: %d
482 .It ena%d: failed to register interrupt handler for irq %ju: %d
484 IO resources initialization failed.
486 Interface will not be brought up.
487 .It ena%d: LRO[%d] Initialization failed!
489 Initialization of the LRO for the RX ring failed.
490 .It ena%d: failed to alloc buffer for rx queue
491 .It ena%d: failed to add buffer for rx queue %d
492 .It ena%d: refilled rx qid %d with only %d mbufs (from %d)
494 Allocation of resources used on RX path failed.
496 If happened during initialization of the IO queue, the interface will not be
498 .It ena%d: NULL mbuf in rx_info
500 Error occurred while assembling mbuf from descriptors.
501 .It ena%d: tx_info doesn't have valid mbuf
502 .It ena%d: Invalid req_id: %hu
503 .It ena%d: failed to prepare tx bufs
505 Error occurred while preparing a packet for transmission.
506 .It ena%d: ioctl promisc/allmulti
508 IOCTL request for the device to work in promiscuous/allmulti mode.
515 If an issue is identified with the released source code with a supported
516 adapter, please email the specific information related to the issue to
517 .Aq Mt akiyano@amazon.com ,
518 .Aq Mt osamaabb@amazon.com
520 .Aq Mt darinzon@amazon.com .
528 driver first appeared in
533 driver was developed by Amazon and originally written by