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