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.Dd November 10, 2022 33.Dt CXGBEV 4 34.Os 35.Sh NAME 36.Nm cxgbev 37.Nd "Chelsio T4-, T5-, and T6-based 100Gb, 40Gb, 25Gb, 10Gb, and 1Gb Ethernet VF driver" 38.Sh SYNOPSIS 39To compile this driver into the kernel, 40place the following lines in your 41kernel configuration file: 42.Bd -ragged -offset indent 43.Cd "device cxgbe" 44.Cd "device cxgbev" 45.Ed 46.Pp 47To load the driver as a 48module at boot time, place the following line in 49.Xr loader.conf 5 : 50.Bd -literal -offset indent 51if_cxgbev_load="YES" 52.Ed 53.Sh DESCRIPTION 54The 55.Nm 56driver provides support for Virtual Functions on PCI Express Ethernet adapters 57based on the Chelsio Terminator 4, Terminator 5, and Terminator 6 ASICs 58(T4, T5, and T6). 59The driver supports Jumbo Frames, Transmit/Receive checksum offload, 60TCP segmentation offload (TSO), Large Receive Offload (LRO), VLAN 61tag insertion/extraction, VLAN checksum offload, VLAN TSO, and 62Receive Side Steering (RSS). 63For further hardware information and questions related to hardware 64requirements, see 65.Pa http://www.chelsio.com/ . 66.Pp 67The 68.Nm 69driver uses different names for devices based on the associated ASIC: 70.Bl -column -offset indent "ASIC" "Port Name" 71.It Sy ASIC Ta Sy Port Name Ta Sy Parent Device 72.It T4 Ta cxgbev Ta t4vf 73.It T5 Ta cxlv Ta t5vf 74.It T6 Ta ccv Ta t6vf 75.El 76.Pp 77Loader tunables with the hw.cxgbe prefix apply to VFs from all cards. 78The Physical Function driver for Chelsio Terminator adapters shares these 79tunables. 80The driver provides sysctl MIBs for both ports and parent devices using 81the names above. 82For example, a T5 VF provides port MIBs under dev.cxlv and 83parent device MIBs under dev.t5vf. 84References to sysctl MIBs in the remainder of this page use 85dev.<port> for port MIBs and dev.<nexus> for parent device MIBs. 86.Pp 87For more information on configuring this device, see 88.Xr ifconfig 8 . 89.Sh HARDWARE 90The 91.Nm 92driver supports Virtual Functions on 100Gb and 25Gb Ethernet adapters 93based on the T6 ASIC: 94.Pp 95.Bl -bullet -compact 96.It 97Chelsio T6225-CR 98.It 99Chelsio T6225-SO-CR 100.It 101Chelsio T62100-LP-CR 102.It 103Chelsio T62100-SO-CR 104.It 105Chelsio T62100-CR 106.El 107.Pp 108The 109.Nm 110driver supports Virtual Functions on 40Gb, 10Gb and 1Gb Ethernet adapters 111based on the T5 ASIC: 112.Pp 113.Bl -bullet -compact 114.It 115Chelsio T580-CR 116.It 117Chelsio T580-LP-CR 118.It 119Chelsio T580-LP-SO-CR 120.It 121Chelsio T560-CR 122.It 123Chelsio T540-CR 124.It 125Chelsio T540-LP-CR 126.It 127Chelsio T522-CR 128.It 129Chelsio T520-LL-CR 130.It 131Chelsio T520-CR 132.It 133Chelsio T520-SO 134.It 135Chelsio T520-BT 136.It 137Chelsio T504-BT 138.El 139.Pp 140The 141.Nm 142driver supports Virtual Functions on 10Gb and 1Gb Ethernet adapters based 143on the T4 ASIC: 144.Pp 145.Bl -bullet -compact 146.It 147Chelsio T420-CR 148.It 149Chelsio T422-CR 150.It 151Chelsio T440-CR 152.It 153Chelsio T420-BCH 154.It 155Chelsio T440-BCH 156.It 157Chelsio T440-CH 158.It 159Chelsio T420-SO 160.It 161Chelsio T420-CX 162.It 163Chelsio T420-BT 164.It 165Chelsio T404-BT 166.El 167.Sh LOADER TUNABLES 168Tunables can be set at the 169.Xr loader 8 170prompt before booting the kernel or stored in 171.Xr loader.conf 5 . 172.Bl -tag -width indent 173.It Va hw.cxgbe.ntxq 174Number of tx queues used for a port. 175The default is 16 or the number 176of CPU cores in the system, whichever is less. 177.It Va hw.cxgbe.nrxq 178Number of rx queues used for a port. 179The default is 8 or the number 180of CPU cores in the system, whichever is less. 181.It Va hw.cxgbe.holdoff_timer_idx 182Timer index value used to delay interrupts. 183The holdoff timer list has the values 1, 5, 10, 50, 100, and 200 184by default (all values are in microseconds) and the index selects a 185value from this list. 186The default value is 1 which means the timer value is 5us. 187Different interfaces can be assigned different values at any time via the 188dev.<port>.X.holdoff_tmr_idx sysctl. 189.It Va hw.cxgbe.holdoff_pktc_idx 190Packet-count index value used to delay interrupts. 191The packet-count list has the values 1, 8, 16, and 32 by default, 192and the index selects a value from this list. 193The default value is -1 which means packet counting is disabled and interrupts 194are generated based solely on the holdoff timer value. 195Different interfaces can be assigned different values via the 196dev.<port>.X.holdoff_pktc_idx sysctl. 197This sysctl works only when the interface has never been marked up (as done by 198ifconfig up). 199.It Va hw.cxgbe.qsize_txq 200Number of entries in a transmit queue's descriptor ring. 201A buf_ring of the same size is also allocated for additional 202software queuing. 203See 204.Xr ifnet 9 . 205The default value is 1024. 206Different interfaces can be assigned different values via the 207dev.<port>.X.qsize_txq sysctl. 208This sysctl works only when the interface has never been marked up (as done by 209ifconfig up). 210.It Va hw.cxgbe.qsize_rxq 211Number of entries in a receive queue's descriptor ring. 212The default value is 1024. 213Different interfaces can be assigned different values via the 214dev.<port>.X.qsize_rxq sysctl. 215This sysctl works only when the interface has never been marked up (as done by 216ifconfig up). 217.It Va hw.cxgbe.interrupt_types 218Permitted interrupt types. 219Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X. 220The default is 7 (all allowed). 221The driver selects the best possible type out of the allowed types. 222Note that Virtual Functions do not support INTx interrupts and fail 223to attach if neither MSI nor MSI-X are enabled. 224.It Va hw.cxgbe.fl_pktshift 225Number of padding bytes inserted before the beginning of an Ethernet 226frame in the receive buffer. 227The default value of 2 ensures that the Ethernet payload (usually the IP header) 228is at a 4 byte aligned address. 2290-7 are all valid values. 230.It Va hw.cxgbe.fl_pad 231A non-zero value ensures that writes from the hardware to a receive buffer are 232padded up to the specified boundary. 233The default is -1 which lets the driver pick a pad boundary. 2340 disables trailer padding completely. 235.It Va hw.cxgbe.buffer_packing 236Allow the hardware to deliver multiple frames in the same receive buffer 237opportunistically. 238The default is -1 which lets the driver decide. 2390 or 1 explicitly disable or enable this feature. 240.It Va hw.cxgbe.allow_mbufs_in_cluster 2411 allows the driver to lay down one or more mbufs within the receive buffer 242opportunistically. 243This is the default. 2440 prohibits the driver from doing so. 245.It Va hw.cxgbe.largest_rx_cluster 246.It Va hw.cxgbe.safest_rx_cluster 247Sizes of rx clusters. 248Each of these must be set to one of the sizes available 249(usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater 250than or equal to safest_rx_cluster. 251The defaults are 16384 and 4096 respectively. 252The driver never attempts to allocate a receive buffer larger than 253largest_rx_cluster and falls back to allocating buffers of 254safest_rx_cluster size if an allocation larger than safest_rx_cluster fails. 255Note that largest_rx_cluster merely establishes a ceiling -- the driver is 256allowed to allocate buffers of smaller sizes. 257.El 258.Pp 259Certain settings and resources for Virtual Functions are dictated 260by the parent Physical Function driver. 261For example, the Physical Function driver limits the number of queues 262available to a Virtual Function. 263Some of these limits can be adjusted in the firmware configuration file 264used with the Physical Function driver. 265.Pp 266The PAUSE settings on the port of a Virtual Function are inherited from 267the settings of the same port on the Physical Function. 268Virtual Functions cannot modify the setting and track changes made to 269the associated port's setting by the Physical Function driver. 270.Pp 271Receive queues on a Virtual Function always drop packets in response to 272congestion 273.Po 274equivalent to setting 275.Va hw.cxgbe.cong_drop 276to 1 277.Pc . 278.Pp 279The VF driver currently depends on the PF driver. 280As a result, loading the VF driver also loads the PF driver as a 281dependency. 282.Sh SUPPORT 283For general information and support, 284go to the Chelsio support website at: 285.Pa http://www.chelsio.com/ . 286.Pp 287If an issue is identified with this driver with a supported adapter, 288email all the specific information related to the issue to 289.Aq Mt support@chelsio.com . 290.Sh SEE ALSO 291.Xr arp 4 , 292.Xr cxgbe 4 , 293.Xr netintro 4 , 294.Xr ng_ether 4 , 295.Xr ifconfig 8 296.Sh HISTORY 297The 298.Nm 299device driver first appeared in 300.Fx 11.1 301and 302.Fx 11.1 . 303.Sh AUTHORS 304.An -nosplit 305The 306.Nm 307driver was written by 308.An Navdeep Parhar Aq Mt np@FreeBSD.org 309and 310.An John Baldwin Aq Mt jhb@FreeBSD.org . 311