xref: /freebsd/share/man/man4/cxgbev.4 (revision 7f9dff23d3092aa33ad45b2b63e52469b3c13a6e)
1.\" Copyright (c) 2011-2016, Chelsio Inc
2.\" All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions are met:
6.\"
7.\" 1. Redistributions of source code must retain the above copyright notice,
8.\"    this list of conditions and the following disclaimer.
9.\"
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.
13.\"
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.
17.\"
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.
29.\"
30.\" * Other names and brands may be claimed as the property of others.
31.\"
32.\" $FreeBSD$
33.\"
34.Dd December 5, 2016
35.Dt CXGBEV 4
36.Os
37.Sh NAME
38.Nm cxgbev
39.Nd "Chelsio T4-, T5-, and T6-based 100Gb, 40Gb, 25Gb, 10Gb, and 1Gb Ethernet VF driver"
40.Sh SYNOPSIS
41To compile this driver into the kernel,
42place the following lines in your
43kernel configuration file:
44.Bd -ragged -offset indent
45.Cd "device cxgbe"
46.Cd "device cxgbev"
47.Ed
48.Pp
49To load the driver as a
50module at boot time, place the following line in
51.Xr loader.conf 5 :
52.Bd -literal -offset indent
53if_cxgbev_load="YES"
54.Ed
55.Sh DESCRIPTION
56The
57.Nm
58driver provides support for Virtual Functions on PCI Express Ethernet adapters
59based on the Chelsio Terminator 4, Terminator 5, and Terminator 6 ASICs
60(T4, T5, and T6).
61The driver supports Jumbo Frames, Transmit/Receive checksum offload,
62TCP segmentation offload (TSO), Large Receive Offload (LRO), VLAN
63tag insertion/extraction, VLAN checksum offload, VLAN TSO, and
64Receive Side Steering (RSS).
65For further hardware information and questions related to hardware
66requirements, see
67.Pa http://www.chelsio.com/ .
68.Pp
69The
70.Nm
71driver 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
76.It T6 Ta ccv Ta t6vf
77.El
78.Pp
79Loader tunables with the hw.cxgbe prefix apply to VFs from all cards.
80The Physical Function driver for Chelsio Terminator adapters shares these
81tunables.
82The driver provides sysctl MIBs for both ports and parent devices using
83the names above.
84For example, a T5 VF will provide port MIBs under dev.cxlv and
85parent device MIBs under dev.t5vf.
86References to sysctl MIBs in the remainder of this page will use
87dev.<port> for port MIBs and dev.<nexus> for parent device MIBs.
88.Pp
89For more information on configuring this device, see
90.Xr ifconfig 8 .
91.Sh HARDWARE
92The
93.Nm
94driver supports Virtual Functions on 100Gb and 25Gb Ethernet adapters
95based on the T6 ASIC:
96.Pp
97.Bl -bullet -compact
98.It
99Chelsio T6225-CR
100.It
101Chelsio T6225-SO-CR
102.It
103Chelsio T62100-LP-CR
104.It
105Chelsio T62100-SO-CR
106.It
107Chelsio T62100-CR
108.El
109.Pp
110The
111.Nm
112driver supports Virtual Functions on 40Gb, 10Gb and 1Gb Ethernet adapters
113based on the T5 ASIC:
114.Pp
115.Bl -bullet -compact
116.It
117Chelsio T580-CR
118.It
119Chelsio T580-LP-CR
120.It
121Chelsio T580-LP-SO-CR
122.It
123Chelsio T560-CR
124.It
125Chelsio T540-CR
126.It
127Chelsio T540-LP-CR
128.It
129Chelsio T522-CR
130.It
131Chelsio T520-LL-CR
132.It
133Chelsio T520-CR
134.It
135Chelsio T520-SO
136.It
137Chelsio T520-BT
138.It
139Chelsio T504-BT
140.El
141.Pp
142The
143.Nm
144driver supports Virtual Functions on 10Gb and 1Gb Ethernet adapters based
145on the T4 ASIC:
146.Pp
147.Bl -bullet -compact
148.It
149Chelsio T420-CR
150.It
151Chelsio T422-CR
152.It
153Chelsio T440-CR
154.It
155Chelsio T420-BCH
156.It
157Chelsio T440-BCH
158.It
159Chelsio T440-CH
160.It
161Chelsio T420-SO
162.It
163Chelsio T420-CX
164.It
165Chelsio T420-BT
166.It
167Chelsio T404-BT
168.El
169.Sh LOADER TUNABLES
170Tunables can be set at the
171.Xr loader 8
172prompt before booting the kernel or stored in
173.Xr loader.conf 5 .
174.Bl -tag -width indent
175.It Va hw.cxgbe.ntxq10g
176The number of tx queues to use for a 10Gb or higher-speed port.
177The default is 16 or the number
178of CPU cores in the system, whichever is less.
179.It Va hw.cxgbe.nrxq10g
180The number of rx queues to use for a 10Gb or higher-speed port.
181The default is 8 or the number
182of CPU cores in the system, whichever is less.
183.It Va hw.cxgbe.ntxq1g
184The number of tx queues to use for a 1Gb port.
185The default is 4 or the number
186of CPU cores in the system, whichever is less.
187.It Va hw.cxgbe.nrxq1g
188The number of rx queues to use for a 1Gb port.
189The default is 2 or the number
190of CPU cores in the system, whichever is less.
191.It Va hw.cxgbe.holdoff_timer_idx_10G
192.It Va hw.cxgbe.holdoff_timer_idx_1G
193The timer index value to use to delay interrupts.
194The holdoff timer list has the values 1, 5, 10, 50, 100, and 200
195by default (all values are in microseconds) and the index selects a
196value from this list.
197The default value is 1 which means the timer value is 5us.
198Different interfaces can be assigned different values at any time via the
199dev.<port>.X.holdoff_tmr_idx sysctl.
200.It Va hw.cxgbe.holdoff_pktc_idx_10G
201.It Va hw.cxgbe.holdoff_pktc_idx_1G
202The packet-count index value to use to delay interrupts.
203The packet-count list has the values 1, 8, 16, and 32 by default
204and the index selects a value from this list.
205The default value is -1 which means packet counting is disabled and interrupts
206are generated based solely on the holdoff timer value.
207Different interfaces can be assigned different values via the
208dev.<port>.X.holdoff_pktc_idx sysctl.
209This sysctl works only when the interface has never been marked up (as done by
210ifconfig up).
211.It Va hw.cxgbe.qsize_txq
212The size, in number of entries, of the descriptor ring used for a tx
213queue.
214A buf_ring of the same size is also allocated for additional
215software queuing.
216See
217.Xr ifnet 9 .
218The default value is 1024.
219Different interfaces can be assigned different values via the
220dev.<port>.X.qsize_txq sysctl.
221This sysctl works only when the interface has never been marked up (as done by
222ifconfig up).
223.It Va hw.cxgbe.qsize_rxq
224The size, in number of entries, of the descriptor ring used for an
225rx queue.
226The default value is 1024.
227Different interfaces can be assigned different values via the
228dev.<port>.X.qsize_rxq sysctl.
229This sysctl works only when the interface has never been marked up (as done by
230ifconfig up).
231.It Va hw.cxgbe.interrupt_types
232The interrupt types that the driver is allowed to use.
233Bit 0 represents INTx (line interrupts), bit 1 MSI, bit 2 MSI-X.
234The default is 7 (all allowed).
235The driver will select the best possible type out of the allowed types by
236itself.
237.It Va hw.cxgbe.fl_pktshift
238The number of bytes of padding inserted before the beginning of an Ethernet
239frame in the receive buffer.
240The default value of 2 ensures that the Ethernet payload (usually the IP header)
241is at a 4 byte aligned address.
2420-7 are all valid values.
243.It Va hw.cxgbe.fl_pad
244A non-zero value ensures that writes from the hardware to a receive buffer are
245padded up to the specified boundary.
246The default is -1 which lets the driver pick a pad boundary.
2470 disables trailer padding completely.
248.It Va hw.cxgbe.buffer_packing
249Allow the hardware to deliver multiple frames in the same receive buffer
250opportunistically.
251The default is -1 which lets the driver decide.
2520 or 1 explicitly disable or enable this feature.
253.It Va hw.cxgbe.allow_mbufs_in_cluster
2541 allows the driver to lay down one or more mbufs within the receive buffer
255opportunistically.
256This is the default.
2570 prohibits the driver from doing so.
258.It Va hw.cxgbe.largest_rx_cluster
259.It Va hw.cxgbe.safest_rx_cluster
260Sizes of rx clusters.
261Each of these must be set to one of the sizes available
262(usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater
263than or equal to safest_rx_cluster.
264The defaults are 16384 and 4096 respectively.
265The driver will never attempt to allocate a receive buffer larger than
266largest_rx_cluster and will fall back to allocating buffers of
267safest_rx_cluster size if an allocation larger than safest_rx_cluster fails.
268Note that largest_rx_cluster merely establishes a ceiling -- the driver is
269allowed to allocate buffers of smaller sizes.
270.El
271.Pp
272Certain settings and resources for Virtual Functions are dictated
273by the parent Physical Function driver.
274For example, the Physical Function driver limits the number of queues a
275Virtual Function is permitted to use.
276Some of these limits can be adjusted in the firmware configuration file
277used with the Physical Function driver.
278.Pp
279The PAUSE settings on the port of a Virtual Function are inherited from
280the settings of the same port on the Physical Function.
281Virtual Functions cannot modify the setting and track changes made to
282the associated port's setting by the Physical Function driver.
283.Pp
284Receive queues on a Virtual Function always drop packets in response to
285congestion
286.Po
287equivalent to setting
288.Va hw.cxgbe.cong_drop
289to 1
290.Pc .
291.Pp
292The VF driver currently depends on the PF driver.
293As a result, loading the VF driver will also load the PF driver as a
294dependency.
295.Sh SUPPORT
296For general information and support,
297go to the Chelsio support website at:
298.Pa http://www.chelsio.com/ .
299.Pp
300If an issue is identified with this driver with a supported adapter,
301email all the specific information related to the issue to
302.Aq Mt support@chelsio.com .
303.Sh SEE ALSO
304.Xr altq 4 ,
305.Xr arp 4 ,
306.Xr cxgbe 4 ,
307.Xr netintro 4 ,
308.Xr ng_ether 4 ,
309.Xr ifconfig 8
310.Sh HISTORY
311The
312.Nm
313device driver first appeared in
314.Fx 11.1
315and
316.Fx 12.0 .
317.Sh AUTHORS
318.An -nosplit
319The
320.Nm
321driver was written by
322.An Navdeep Parhar Aq Mt np@FreeBSD.org
323and
324.An John Baldwin Aq Mt jhb@FreeBSD.org .
325