1 .\" Copyright (c) 2011-2016, Chelsio Inc
2 .\" All rights reserved.
4 .\" Redistribution and use in source and binary forms, with or without
5 .\" modification, are permitted provided that the following conditions are met:
7 .\" 1. Redistributions of source code must retain the above copyright notice,
8 .\" this list of conditions and the following disclaimer.
10 .\" 2. Redistributions in binary form must reproduce the above copyright
11 .\" notice, this list of conditions and the following disclaimer in the
12 .\" documentation and/or other materials provided with the distribution.
14 .\" 3. Neither the name of the Chelsio Inc nor the names of its
15 .\" contributors may be used to endorse or promote products derived from
16 .\" this software without specific prior written permission.
18 .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
19 .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
22 .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28 .\" POSSIBILITY OF SUCH DAMAGE.
30 .\" * Other names and brands may be claimed as the property of others.
39 .Nd "Chelsio T4-, T5-, and T6-based 100Gb, 40Gb, 25Gb, 10Gb, and 1Gb Ethernet adapter driver"
41 To compile this driver into the kernel,
42 place the following lines in your
43 kernel configuration file:
44 .Bd -ragged -offset indent
48 To load the driver as a
49 module at boot time, place the following lines in
51 .Bd -literal -offset indent
60 driver provides support for PCI Express Ethernet adapters based on
61 the Chelsio Terminator 4, Terminator 5, and Terminator 6 ASICs (T4, T5, and T6).
62 The driver supports Jumbo Frames, Transmit/Receive checksum offload,
63 TCP segmentation offload (TSO), Large Receive Offload (LRO), VLAN
64 tag insertion/extraction, VLAN checksum offload, VLAN TSO, and
65 Receive Side Steering (RSS).
66 For further hardware information and questions related to hardware
68 .Pa http://www.chelsio.com/ .
72 driver uses different names for devices based on the associated ASIC:
73 .Bl -column -offset indent "ASIC" "Port Name" "Parent Device"
74 .It Sy ASIC Ta Sy Port Name Ta Sy Parent Device Ta Sy Virtual Interface
75 .It T4 Ta cxgbe Ta t4nex Ta vcxgbe
76 .It T5 Ta cxl Ta t5nex Ta vcxl
77 .It T6 Ta cc Ta t6nex Ta vcc
81 Loader tunables with the hw.cxgbe prefix apply to all cards.
82 The driver provides sysctl MIBs for both ports and parent devices using
84 For example, a T5 adapter provides port MIBs under dev.cxl and
85 adapter-wide MIBs under dev.t5nex.
86 References to sysctl MIBs in the remainder of this page use
87 dev.<port> for port MIBs and dev.<nexus> for adapter-wide MIBs.
89 For more information on configuring this device, see
94 driver supports 100Gb and 25Gb Ethernet adapters based on the T6 ASIC:
111 driver supports 40Gb, 10Gb and 1Gb Ethernet adapters based on the T5 ASIC:
119 Chelsio T580-LP-SO-CR
142 driver supports 10Gb and 1Gb Ethernet adapters based on the T4 ASIC:
167 Tunables can be set at the
169 prompt before booting the kernel or stored in
171 .Bl -tag -width indent
172 .It Va hw.cxgbe.ntxq10g
173 Number of tx queues used for a 10Gb or higher-speed port.
174 The default is 16 or the number
175 of CPU cores in the system, whichever is less.
176 .It Va hw.cxgbe.nrxq10g
177 Number of rx queues used for a 10Gb or higher-speed port.
178 The default is 8 or the number
179 of CPU cores in the system, whichever is less.
180 .It Va hw.cxgbe.ntxq1g
181 Number of tx queues used for a 1Gb port.
182 The default is 4 or the number
183 of CPU cores in the system, whichever is less.
184 .It Va hw.cxgbe.nrxq1g
185 Number of rx queues used for a 1Gb port.
186 The default is 2 or the number
187 of CPU cores in the system, whichever is less.
188 .It Va hw.cxgbe.nofldtxq10g
189 Number of TOE tx queues used for a 10Gb or higher-speed port.
190 The default is 8 or the
191 number of CPU cores in the system, whichever is less.
192 .It Va hw.cxgbe.nofldrxq10g
193 Number of TOE rx queues used for a 10Gb or higher-speed port.
194 The default is 2 or the
195 number of CPU cores in the system, whichever is less.
196 .It Va hw.cxgbe.nofldtxq1g
197 Number of TOE tx queues used for a 1Gb port.
198 The default is 2 or the
199 number of CPU cores in the system, whichever is less.
200 .It Va hw.cxgbe.nofldrxq1g
201 Number of TOE rx queues used for a 1Gb port.
203 .It Va hw.cxgbe.num_vis
204 Number of virtual interfaces (VIs) created for each port.
205 Each virtual interface creates a separate network interface.
206 The first virtual interface on each port is required and represents
207 the primary network interface on the port.
208 Additional virtual interfaces on a port are named using the Virtual Interface
209 name from the table above.
210 Additional virtual interfaces use a single pair of queues
211 for rx and tx as well an additional pair of queues for TOE rx and tx.
213 .It Va hw.cxgbe.holdoff_timer_idx_10G
214 .It Va hw.cxgbe.holdoff_timer_idx_1G
215 Timer index value used to delay interrupts.
216 The holdoff timer list has the values 1, 5, 10, 50, 100, and 200
217 by default (all values are in microseconds) and the index selects a
218 value from this list.
219 The default value is 1 which means the timer value is 5us.
220 Different interfaces can be assigned different values at any time via the
221 dev.<port>.X.holdoff_tmr_idx sysctl.
222 .It Va hw.cxgbe.holdoff_pktc_idx_10G
223 .It Va hw.cxgbe.holdoff_pktc_idx_1G
224 Packet-count index value used to delay interrupts.
225 The packet-count list has the values 1, 8, 16, and 32 by default,
226 and the index selects a value from this list.
227 The default value is -1 which means packet counting is disabled and interrupts
228 are generated based solely on the holdoff timer value.
229 Different interfaces can be assigned different values via the
230 dev.<port>.X.holdoff_pktc_idx sysctl.
231 This sysctl works only when the interface has never been marked up (as done by
233 .It Va hw.cxgbe.qsize_txq
234 Number of entries in a transmit queue's descriptor ring.
235 A buf_ring of the same size is also allocated for additional
239 The default value is 1024.
240 Different interfaces can be assigned different values via the
241 dev.<port>.X.qsize_txq sysctl.
242 This sysctl works only when the interface has never been marked up (as done by
244 .It Va hw.cxgbe.qsize_rxq
245 Number of entries in a receive queue's descriptor ring.
246 The default value is 1024.
247 Different interfaces can be assigned different values via the
248 dev.<port>.X.qsize_rxq sysctl.
249 This sysctl works only when the interface has never been marked up (as done by
251 .It Va hw.cxgbe.interrupt_types
252 Permitted interrupt types.
253 Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X.
254 The default is 7 (all allowed).
255 The driver selects the best possible type out of the allowed types.
256 .It Va hw.cxgbe.fw_install
257 0 prohibits the driver from installing a firmware on the card.
258 1 allows the driver to install a new firmware if internal driver
259 heuristics indicate that the new firmware is preferable to the one
261 2 instructs the driver to always install the new firmware on the card as
262 long as it is compatible with the driver and is a different version than
263 the one already on the card.
265 .It Va hw.cxgbe.fl_pktshift
266 Number of padding bytes inserted before the beginning of an Ethernet
267 frame in the receive buffer.
268 The default value of 2 ensures that the Ethernet payload (usually the IP header)
269 is at a 4 byte aligned address.
270 0-7 are all valid values.
271 .It Va hw.cxgbe.fl_pad
272 A non-zero value ensures that writes from the hardware to a receive buffer are
273 padded up to the specified boundary.
274 The default is -1 which lets the driver pick a pad boundary.
275 0 disables trailer padding completely.
276 .It Va hw.cxgbe.cong_drop
277 Controls the hardware response to congestion.
278 -1 disables congestion feedback and is not recommended.
279 0 instructs the hardware to backpressure its pipeline on congestion.
280 This usually results in the port emitting PAUSE frames.
281 1 instructs the hardware to drop frames destined for congested queues.
282 .It Va hw.cxgbe.pause_settings
283 PAUSE frame settings.
284 Bit 0 is rx_pause, bit 1 is tx_pause.
285 rx_pause = 1 instructs the hardware to heed incoming PAUSE frames, 0 instructs
287 tx_pause = 1 allows the hardware to emit PAUSE frames when its receive FIFO
288 reaches a high threshold, 0 prohibits the hardware from emitting PAUSE frames.
289 The default is 3 (both rx_pause and tx_pause = 1).
290 This tunable establishes the default PAUSE settings for all ports.
291 Settings can be displayed and controlled on a per-port basis via the
292 dev.<port>.X.pause_settings sysctl.
294 FEC (Forward Error Correction) settings.
296 Bit 0 enables RS FEC, bit 1 enables BASE-R RS, bit 3 is reserved.
297 The default is -1 which lets the driver pick a value.
298 This tunable establishes the default FEC settings for all ports.
299 Settings can be displayed and controlled on a per-port basis via the
300 dev.<port>.X.fec sysctl.
301 .It Va hw.cxgbe.autoneg
302 Link autonegotiation settings.
303 This tunable establishes the default autonegotiation settings for all ports.
304 Settings can be displayed and controlled on a per-port basis via the
305 dev.<port>.X.autoneg sysctl.
306 0 disables autonegotiation.
307 1 enables autonegotiation.
308 The default is -1 which lets the driver pick a value.
309 dev.<port>.X.autoneg is -1 for port and module combinations that do not support
311 .It Va hw.cxgbe.buffer_packing
312 Allow the hardware to deliver multiple frames in the same receive buffer
314 The default is -1 which lets the driver decide.
315 0 or 1 explicitly disable or enable this feature.
316 .It Va hw.cxgbe.allow_mbufs_in_cluster
317 1 allows the driver to lay down one or more mbufs within the receive buffer
318 opportunistically. This is the default.
319 0 prohibits the driver from doing so.
320 .It Va hw.cxgbe.largest_rx_cluster
321 .It Va hw.cxgbe.safest_rx_cluster
322 Sizes of rx clusters. Each of these must be set to one of the sizes available
323 (usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater
324 than or equal to safest_rx_cluster.
325 The defaults are 16384 and 4096 respectively.
326 The driver never attempts to allocate a receive buffer larger than
327 largest_rx_cluster and falls back to allocating buffers of
328 safest_rx_cluster size if an allocation larger than safest_rx_cluster fails.
329 Note that largest_rx_cluster merely establishes a ceiling -- the driver is
330 allowed to allocate buffers of smaller sizes.
331 .It Va hw.cxgbe.config_file
332 Select a pre-packaged device configuration file.
333 A configuration file contains a recipe for partitioning and configuring the
334 hardware resources on the card.
335 This tunable is for specialized applications only and should not be used in
337 The configuration profile currently in use is available in the dev.<nexus>.X.cf
338 and dev.<nexus>.X.cfcsum sysctls.
339 .It Va hw.cxgbe.linkcaps_allowed
340 .It Va hw.cxgbe.niccaps_allowed
341 .It Va hw.cxgbe.toecaps_allowed
342 .It Va hw.cxgbe.rdmacaps_allowed
343 .It Va hw.cxgbe.iscsicaps_allowed
344 .It Va hw.cxgbe.fcoecaps_allowed
345 Disallowing capabilities provides a hint to the driver and firmware to not
346 reserve hardware resources for that feature.
347 Each of these is a bit field with a bit for each sub-capability within the
349 This tunable is for specialized applications only and should not be used in
351 The capabilities for which hardware resources have been reserved are listed in
352 dev.<nexus>.X.*caps sysctls.
355 For general information and support,
356 go to the Chelsio support website at:
357 .Pa http://www.chelsio.com/ .
359 If an issue is identified with this driver with a supported adapter,
360 email all the specific information related to the issue to
361 .Aq support@chelsio.com .
373 device driver first appeared in
375 Support for T5 cards first appeared in
379 Support for T6 cards first appeared in
387 driver was written by
388 .An Navdeep Parhar Aq np@FreeBSD.org .