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
80 Loader tunables with the hw.cxgbe prefix apply to all cards.
81 The driver provides sysctl MIBs for both ports and parent devices using
83 For example, a T5 adapter provides port MIBs under dev.cxl and
84 adapter-wide MIBs under dev.t5nex.
85 References to sysctl MIBs in the remainder of this page use
86 dev.<port> for port MIBs and dev.<nexus> for adapter-wide MIBs.
88 For more information on configuring this device, see
93 driver supports 100Gb and 25Gb Ethernet adapters based on the T6 ASIC:
110 driver supports 40Gb, 10Gb and 1Gb Ethernet adapters based on the T5 ASIC:
118 Chelsio T580-LP-SO-CR
141 driver supports 10Gb and 1Gb Ethernet adapters based on the T4 ASIC:
166 Tunables can be set at the
168 prompt before booting the kernel or stored in
170 There are multiple tunables that control the number of queues of various
172 A negative value for such a tunable instructs the driver to create
173 up to that many queues if there are enough CPU cores available.
174 .Bl -tag -width indent
176 Number of NIC tx queues used for a port.
177 The default is 16 or the number
178 of CPU cores in the system, whichever is less.
180 Number of NIC rx queues used for a port.
181 The default is 8 or the number
182 of CPU cores in the system, whichever is less.
183 .It Va hw.cxgbe.nofldtxq
184 Number of TOE tx queues used for a port.
185 The default is 8 or the
186 number of CPU cores in the system, whichever is less.
187 .It Va hw.cxgbe.nofldrxq
188 Number of TOE rx queues used for a port.
189 The default is 2 or the
190 number of CPU cores in the system, whichever is less.
191 .It Va hw.cxgbe.num_vis
192 Number of virtual interfaces (VIs) created for each port.
193 Each virtual interface creates a separate network interface.
194 The first virtual interface on each port is required and represents
195 the primary network interface on the port.
196 Additional virtual interfaces on a port are named using the Virtual Interface
197 name from the table above.
198 Additional virtual interfaces use a single pair of queues
199 for rx and tx as well an additional pair of queues for TOE rx and tx.
201 .It Va hw.cxgbe.holdoff_timer_idx
202 .It Va hw.cxgbe.holdoff_timer_idx_ofld
203 Timer index value used to delay interrupts.
204 The holdoff timer list has the values 1, 5, 10, 50, 100, and 200
205 by default (all values are in microseconds) and the index selects a
206 value from this list.
207 holdoff_timer_idx_ofld applies to queues used for TOE rx.
208 The default value is 1 which means the timer value is 5us.
209 Different interfaces can be assigned different values at any time via the
210 dev.<port>.X.holdoff_tmr_idx and dev.<port>.X.holdoff_tmr_idx_ofld sysctls.
211 .It Va hw.cxgbe.holdoff_pktc_idx
212 .It Va hw.cxgbe.holdoff_pktc_idx_ofld
213 Packet-count index value used to delay interrupts.
214 The packet-count list has the values 1, 8, 16, and 32 by default,
215 and the index selects a value from this list.
216 holdoff_pktc_idx_ofld applies to queues used for TOE rx.
217 The default value is -1 which means packet counting is disabled and interrupts
218 are generated based solely on the holdoff timer value.
219 Different interfaces can be assigned different values via the
220 dev.<port>.X.holdoff_pktc_idx and dev.<port>.X.holdoff_pktc_idx_ofld sysctls.
221 These sysctls work only when the interface has never been marked up (as done by
223 .It Va hw.cxgbe.qsize_txq
224 Number of entries in a transmit queue's descriptor ring.
225 A buf_ring of the same size is also allocated for additional
229 The default value is 1024.
230 Different interfaces can be assigned different values via the
231 dev.<port>.X.qsize_txq sysctl.
232 This sysctl works only when the interface has never been marked up (as done by
234 .It Va hw.cxgbe.qsize_rxq
235 Number of entries in a receive queue's descriptor ring.
236 The default value is 1024.
237 Different interfaces can be assigned different values via the
238 dev.<port>.X.qsize_rxq sysctl.
239 This sysctl works only when the interface has never been marked up (as done by
241 .It Va hw.cxgbe.interrupt_types
242 Permitted interrupt types.
243 Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X.
244 The default is 7 (all allowed).
245 The driver selects the best possible type out of the allowed types.
246 .It Va hw.cxgbe.fw_install
247 0 prohibits the driver from installing a firmware on the card.
248 1 allows the driver to install a new firmware if internal driver
249 heuristics indicate that the new firmware is preferable to the one
251 2 instructs the driver to always install the new firmware on the card as
252 long as it is compatible with the driver and is a different version than
253 the one already on the card.
255 .It Va hw.cxgbe.fl_pktshift
256 Number of padding bytes inserted before the beginning of an Ethernet
257 frame in the receive buffer.
258 The default value of 2 ensures that the Ethernet payload (usually the IP header)
259 is at a 4 byte aligned address.
260 0-7 are all valid values.
261 .It Va hw.cxgbe.fl_pad
262 A non-zero value ensures that writes from the hardware to a receive buffer are
263 padded up to the specified boundary.
264 The default is -1 which lets the driver pick a pad boundary.
265 0 disables trailer padding completely.
266 .It Va hw.cxgbe.cong_drop
267 Controls the hardware response to congestion.
268 -1 disables congestion feedback and is not recommended.
269 0 instructs the hardware to backpressure its pipeline on congestion.
270 This usually results in the port emitting PAUSE frames.
271 1 instructs the hardware to drop frames destined for congested queues.
272 .It Va hw.cxgbe.pause_settings
273 PAUSE frame settings.
274 Bit 0 is rx_pause, bit 1 is tx_pause.
275 rx_pause = 1 instructs the hardware to heed incoming PAUSE frames, 0 instructs
277 tx_pause = 1 allows the hardware to emit PAUSE frames when its receive FIFO
278 reaches a high threshold, 0 prohibits the hardware from emitting PAUSE frames.
279 The default is 3 (both rx_pause and tx_pause = 1).
280 This tunable establishes the default PAUSE settings for all ports.
281 Settings can be displayed and controlled on a per-port basis via the
282 dev.<port>.X.pause_settings sysctl.
284 FEC (Forward Error Correction) settings.
286 Bit 0 enables RS FEC, bit 1 enables BASE-R RS, bit 3 is reserved.
287 The default is -1 which lets the driver pick a value.
288 This tunable establishes the default FEC settings for all ports.
289 Settings can be displayed and controlled on a per-port basis via the
290 dev.<port>.X.fec sysctl.
291 .It Va hw.cxgbe.autoneg
292 Link autonegotiation settings.
293 This tunable establishes the default autonegotiation settings for all ports.
294 Settings can be displayed and controlled on a per-port basis via the
295 dev.<port>.X.autoneg sysctl.
296 0 disables autonegotiation.
297 1 enables autonegotiation.
298 The default is -1 which lets the driver pick a value.
299 dev.<port>.X.autoneg is -1 for port and module combinations that do not support
301 .It Va hw.cxgbe.buffer_packing
302 Allow the hardware to deliver multiple frames in the same receive buffer
304 The default is -1 which lets the driver decide.
305 0 or 1 explicitly disable or enable this feature.
306 .It Va hw.cxgbe.allow_mbufs_in_cluster
307 1 allows the driver to lay down one or more mbufs within the receive buffer
310 0 prohibits the driver from doing so.
311 .It Va hw.cxgbe.largest_rx_cluster
312 .It Va hw.cxgbe.safest_rx_cluster
313 Sizes of rx clusters.
314 Each of these must be set to one of the sizes available
315 (usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater
316 than or equal to safest_rx_cluster.
317 The defaults are 16384 and 4096 respectively.
318 The driver never attempts to allocate a receive buffer larger than
319 largest_rx_cluster and falls back to allocating buffers of
320 safest_rx_cluster size if an allocation larger than safest_rx_cluster fails.
321 Note that largest_rx_cluster merely establishes a ceiling -- the driver is
322 allowed to allocate buffers of smaller sizes.
323 .It Va hw.cxgbe.config_file
324 Select a pre-packaged device configuration file.
325 A configuration file contains a recipe for partitioning and configuring the
326 hardware resources on the card.
327 This tunable is for specialized applications only and should not be used in
329 The configuration profile currently in use is available in the dev.<nexus>.X.cf
330 and dev.<nexus>.X.cfcsum sysctls.
331 .It Va hw.cxgbe.linkcaps_allowed
332 .It Va hw.cxgbe.niccaps_allowed
333 .It Va hw.cxgbe.toecaps_allowed
334 .It Va hw.cxgbe.rdmacaps_allowed
335 .It Va hw.cxgbe.iscsicaps_allowed
336 .It Va hw.cxgbe.fcoecaps_allowed
337 Disallowing capabilities provides a hint to the driver and firmware to not
338 reserve hardware resources for that feature.
339 Each of these is a bit field with a bit for each sub-capability within the
341 This tunable is for specialized applications only and should not be used in
343 The capabilities for which hardware resources have been reserved are listed in
344 dev.<nexus>.X.*caps sysctls.
347 For general information and support,
348 go to the Chelsio support website at:
349 .Pa http://www.chelsio.com/ .
351 If an issue is identified with this driver with a supported adapter,
352 email all the specific information related to the issue to
353 .Aq Mt support@chelsio.com .
366 device driver first appeared in
368 Support for T5 cards first appeared in
372 Support for T6 cards first appeared in
380 driver was written by
381 .An Navdeep Parhar Aq Mt np@FreeBSD.org .