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 VF 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
49 To load the driver as a
50 module at boot time, place the following line in
52 .Bd -literal -offset indent
58 driver provides support for Virtual Functions on PCI Express Ethernet adapters
59 based on the Chelsio Terminator 4, Terminator 5, and Terminator 6 ASICs
61 The driver supports Jumbo Frames, Transmit/Receive checksum offload,
62 TCP segmentation offload (TSO), Large Receive Offload (LRO), VLAN
63 tag insertion/extraction, VLAN checksum offload, VLAN TSO, and
64 Receive Side Steering (RSS).
65 For further hardware information and questions related to hardware
67 .Pa http://www.chelsio.com/ .
71 driver uses different names for devices based on the associated ASIC:
72 .Bl -column -offset indent "ASIC" "Port Name"
73 .It Sy ASIC Ta Sy Port Name Ta Sy Parent Device
74 .It T4 Ta cxgbev Ta t4vf
75 .It T5 Ta cxlv Ta t5vf
79 Loader tunables with the hw.cxgbe prefix apply to VFs from all cards.
80 The Physical Function driver for Chelsio Terminator adapters shares these
82 The driver provides sysctl MIBs for both ports and parent devices using
84 For example, a T5 VF provides port MIBs under dev.cxlv and
85 parent device MIBs under dev.t5vf.
86 References to sysctl MIBs in the remainder of this page use
87 dev.<port> for port MIBs and dev.<nexus> for parent device MIBs.
89 For more information on configuring this device, see
94 driver supports Virtual Functions on 100Gb and 25Gb Ethernet adapters
112 driver supports Virtual Functions on 40Gb, 10Gb and 1Gb Ethernet adapters
113 based on the T5 ASIC:
121 Chelsio T580-LP-SO-CR
144 driver supports Virtual Functions on 10Gb and 1Gb Ethernet adapters based
170 Tunables can be set at the
172 prompt before booting the kernel or stored in
174 .Bl -tag -width indent
176 Number of 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 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.holdoff_timer_idx
184 Timer index value used to delay interrupts.
185 The holdoff timer list has the values 1, 5, 10, 50, 100, and 200
186 by default (all values are in microseconds) and the index selects a
187 value from this list.
188 The default value is 1 which means the timer value is 5us.
189 Different interfaces can be assigned different values at any time via the
190 dev.<port>.X.holdoff_tmr_idx sysctl.
191 .It Va hw.cxgbe.holdoff_pktc_idx
192 Packet-count index value used to delay interrupts.
193 The packet-count list has the values 1, 8, 16, and 32 by default,
194 and the index selects a value from this list.
195 The default value is -1 which means packet counting is disabled and interrupts
196 are generated based solely on the holdoff timer value.
197 Different interfaces can be assigned different values via the
198 dev.<port>.X.holdoff_pktc_idx sysctl.
199 This sysctl works only when the interface has never been marked up (as done by
201 .It Va hw.cxgbe.qsize_txq
202 Number of entries in a transmit queue's descriptor ring.
203 A buf_ring of the same size is also allocated for additional
207 The default value is 1024.
208 Different interfaces can be assigned different values via the
209 dev.<port>.X.qsize_txq sysctl.
210 This sysctl works only when the interface has never been marked up (as done by
212 .It Va hw.cxgbe.qsize_rxq
213 Number of entries in a receive queue's descriptor ring.
214 The default value is 1024.
215 Different interfaces can be assigned different values via the
216 dev.<port>.X.qsize_rxq sysctl.
217 This sysctl works only when the interface has never been marked up (as done by
219 .It Va hw.cxgbe.interrupt_types
220 Permitted interrupt types.
221 Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X.
222 The default is 7 (all allowed).
223 The driver selects the best possible type out of the allowed types.
224 Note that Virtual Functions do not support INTx interrupts and fail
225 to attach if neither MSI nor MSI-X are enabled.
226 .It Va hw.cxgbe.fl_pktshift
227 Number of padding bytes inserted before the beginning of an Ethernet
228 frame in the receive buffer.
229 The default value of 2 ensures that the Ethernet payload (usually the IP header)
230 is at a 4 byte aligned address.
231 0-7 are all valid values.
232 .It Va hw.cxgbe.fl_pad
233 A non-zero value ensures that writes from the hardware to a receive buffer are
234 padded up to the specified boundary.
235 The default is -1 which lets the driver pick a pad boundary.
236 0 disables trailer padding completely.
237 .It Va hw.cxgbe.buffer_packing
238 Allow the hardware to deliver multiple frames in the same receive buffer
240 The default is -1 which lets the driver decide.
241 0 or 1 explicitly disable or enable this feature.
242 .It Va hw.cxgbe.allow_mbufs_in_cluster
243 1 allows the driver to lay down one or more mbufs within the receive buffer
246 0 prohibits the driver from doing so.
247 .It Va hw.cxgbe.largest_rx_cluster
248 .It Va hw.cxgbe.safest_rx_cluster
249 Sizes of rx clusters.
250 Each of these must be set to one of the sizes available
251 (usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater
252 than or equal to safest_rx_cluster.
253 The defaults are 16384 and 4096 respectively.
254 The driver never attempts to allocate a receive buffer larger than
255 largest_rx_cluster and falls back to allocating buffers of
256 safest_rx_cluster size if an allocation larger than safest_rx_cluster fails.
257 Note that largest_rx_cluster merely establishes a ceiling -- the driver is
258 allowed to allocate buffers of smaller sizes.
261 Certain settings and resources for Virtual Functions are dictated
262 by the parent Physical Function driver.
263 For example, the Physical Function driver limits the number of queues
264 available to a Virtual Function.
265 Some of these limits can be adjusted in the firmware configuration file
266 used with the Physical Function driver.
268 The PAUSE settings on the port of a Virtual Function are inherited from
269 the settings of the same port on the Physical Function.
270 Virtual Functions cannot modify the setting and track changes made to
271 the associated port's setting by the Physical Function driver.
273 Receive queues on a Virtual Function always drop packets in response to
276 equivalent to setting
277 .Va hw.cxgbe.cong_drop
281 The VF driver currently depends on the PF driver.
282 As a result, loading the VF driver also loads the PF driver as a
285 For general information and support,
286 go to the Chelsio support website at:
287 .Pa http://www.chelsio.com/ .
289 If an issue is identified with this driver with a supported adapter,
290 email all the specific information related to the issue to
291 .Aq Mt support@chelsio.com .
301 device driver first appeared in
309 driver was written by
310 .An Navdeep Parhar Aq Mt np@FreeBSD.org
312 .An John Baldwin Aq Mt jhb@FreeBSD.org .