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 22, 2016 35.Dt CXGBE 4 36.Os 37.Sh NAME 38.Nm cxgbe 39.Nd "Chelsio T4-, T5-, and T6-based 100Gb, 40Gb, 25Gb, 10Gb, and 1Gb Ethernet adapter 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.Ed 47.Pp 48To load the driver as a 49module at boot time, place the following lines in 50.Xr loader.conf 5 : 51.Bd -literal -offset indent 52t4fw_cfg_load="YES" 53t5fw_cfg_load="YES" 54t6fw_cfg_load="YES" 55if_cxgbe_load="YES" 56.Ed 57.Sh DESCRIPTION 58The 59.Nm 60driver provides support for PCI Express Ethernet adapters based on 61the Chelsio Terminator 4, Terminator 5, and Terminator 6 ASICs (T4, T5, and T6). 62The driver supports Jumbo Frames, Transmit/Receive checksum offload, 63TCP segmentation offload (TSO), Large Receive Offload (LRO), VLAN 64tag insertion/extraction, VLAN checksum offload, VLAN TSO, and 65Receive Side Steering (RSS). 66For further hardware information and questions related to hardware 67requirements, see 68.Pa http://www.chelsio.com/ . 69.Pp 70The 71.Nm 72driver 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 78.El 79.Pp 80Loader tunables with the hw.cxgbe prefix apply to all cards. 81The driver provides sysctl MIBs for both ports and parent devices using 82the names above. 83For example, a T5 adapter provides port MIBs under dev.cxl and 84adapter-wide MIBs under dev.t5nex. 85References to sysctl MIBs in the remainder of this page use 86dev.<port> for port MIBs and dev.<nexus> for adapter-wide MIBs. 87.Pp 88For more information on configuring this device, see 89.Xr ifconfig 8 . 90.Sh HARDWARE 91The 92.Nm 93driver supports 100Gb and 25Gb Ethernet adapters based 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 40Gb, 10Gb and 1Gb Ethernet adapters based on the T5 ASIC: 111.Pp 112.Bl -bullet -compact 113.It 114Chelsio T580-CR 115.It 116Chelsio T580-LP-CR 117.It 118Chelsio T580-LP-SO-CR 119.It 120Chelsio T560-CR 121.It 122Chelsio T540-CR 123.It 124Chelsio T540-LP-CR 125.It 126Chelsio T522-CR 127.It 128Chelsio T520-LL-CR 129.It 130Chelsio T520-CR 131.It 132Chelsio T520-SO 133.It 134Chelsio T520-BT 135.It 136Chelsio T504-BT 137.El 138.Pp 139The 140.Nm 141driver supports 10Gb and 1Gb Ethernet adapters based on the T4 ASIC: 142.Pp 143.Bl -bullet -compact 144.It 145Chelsio T420-CR 146.It 147Chelsio T422-CR 148.It 149Chelsio T440-CR 150.It 151Chelsio T420-BCH 152.It 153Chelsio T440-BCH 154.It 155Chelsio T440-CH 156.It 157Chelsio T420-SO 158.It 159Chelsio T420-CX 160.It 161Chelsio T420-BT 162.It 163Chelsio T404-BT 164.El 165.Sh LOADER TUNABLES 166Tunables can be set at the 167.Xr loader 8 168prompt before booting the kernel or stored in 169.Xr loader.conf 5 . 170There are multiple tunables that control the number of queues of various 171types. 172A negative value for such a tunable instructs the driver to create 173up to that many queues if there are enough CPU cores available. 174.Bl -tag -width indent 175.It Va hw.cxgbe.ntxq10g 176Number of tx queues used 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 180Number of rx queues used 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 184Number of tx queues used 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 188Number of rx queues used 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.nofldtxq10g 192Number of TOE tx queues used for a 10Gb or higher-speed port. 193The default is 8 or the 194number of CPU cores in the system, whichever is less. 195.It Va hw.cxgbe.nofldrxq10g 196Number of TOE rx queues used for a 10Gb or higher-speed port. 197The default is 2 or the 198number of CPU cores in the system, whichever is less. 199.It Va hw.cxgbe.nofldtxq1g 200Number of TOE tx queues used for a 1Gb port. 201The default is 2 or the 202number of CPU cores in the system, whichever is less. 203.It Va hw.cxgbe.nofldrxq1g 204Number of TOE rx queues used for a 1Gb port. 205The default is 1. 206.It Va hw.cxgbe.num_vis 207Number of virtual interfaces (VIs) created for each port. 208Each virtual interface creates a separate network interface. 209The first virtual interface on each port is required and represents 210the primary network interface on the port. 211Additional virtual interfaces on a port are named using the Virtual Interface 212name from the table above. 213Additional virtual interfaces use a single pair of queues 214for rx and tx as well an additional pair of queues for TOE rx and tx. 215The default is 1. 216.It Va hw.cxgbe.holdoff_timer_idx_10G 217.It Va hw.cxgbe.holdoff_timer_idx_1G 218Timer index value used to delay interrupts. 219The holdoff timer list has the values 1, 5, 10, 50, 100, and 200 220by default (all values are in microseconds) and the index selects a 221value from this list. 222The default value is 1 which means the timer value is 5us. 223Different interfaces can be assigned different values at any time via the 224dev.<port>.X.holdoff_tmr_idx sysctl. 225.It Va hw.cxgbe.holdoff_pktc_idx_10G 226.It Va hw.cxgbe.holdoff_pktc_idx_1G 227Packet-count index value used to delay interrupts. 228The packet-count list has the values 1, 8, 16, and 32 by default, 229and the index selects a value from this list. 230The default value is -1 which means packet counting is disabled and interrupts 231are generated based solely on the holdoff timer value. 232Different interfaces can be assigned different values via the 233dev.<port>.X.holdoff_pktc_idx sysctl. 234This sysctl works only when the interface has never been marked up (as done by 235ifconfig up). 236.It Va hw.cxgbe.qsize_txq 237Number of entries in a transmit queue's descriptor ring. 238A buf_ring of the same size is also allocated for additional 239software queuing. 240See 241.Xr ifnet 9 . 242The default value is 1024. 243Different interfaces can be assigned different values via the 244dev.<port>.X.qsize_txq sysctl. 245This sysctl works only when the interface has never been marked up (as done by 246ifconfig up). 247.It Va hw.cxgbe.qsize_rxq 248Number of entries in a receive queue's descriptor ring. 249The default value is 1024. 250Different interfaces can be assigned different values via the 251dev.<port>.X.qsize_rxq sysctl. 252This sysctl works only when the interface has never been marked up (as done by 253ifconfig up). 254.It Va hw.cxgbe.interrupt_types 255Permitted interrupt types. 256Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X. 257The default is 7 (all allowed). 258The driver selects the best possible type out of the allowed types. 259.It Va hw.cxgbe.fw_install 2600 prohibits the driver from installing a firmware on the card. 2611 allows the driver to install a new firmware if internal driver 262heuristics indicate that the new firmware is preferable to the one 263already on the card. 2642 instructs the driver to always install the new firmware on the card as 265long as it is compatible with the driver and is a different version than 266the one already on the card. 267The default is 1. 268.It Va hw.cxgbe.fl_pktshift 269Number of padding bytes inserted before the beginning of an Ethernet 270frame in the receive buffer. 271The default value of 2 ensures that the Ethernet payload (usually the IP header) 272is at a 4 byte aligned address. 2730-7 are all valid values. 274.It Va hw.cxgbe.fl_pad 275A non-zero value ensures that writes from the hardware to a receive buffer are 276padded up to the specified boundary. 277The default is -1 which lets the driver pick a pad boundary. 2780 disables trailer padding completely. 279.It Va hw.cxgbe.cong_drop 280Controls the hardware response to congestion. 281-1 disables congestion feedback and is not recommended. 2820 instructs the hardware to backpressure its pipeline on congestion. 283This usually results in the port emitting PAUSE frames. 2841 instructs the hardware to drop frames destined for congested queues. 285.It Va hw.cxgbe.pause_settings 286PAUSE frame settings. 287Bit 0 is rx_pause, bit 1 is tx_pause. 288rx_pause = 1 instructs the hardware to heed incoming PAUSE frames, 0 instructs 289it to ignore them. 290tx_pause = 1 allows the hardware to emit PAUSE frames when its receive FIFO 291reaches a high threshold, 0 prohibits the hardware from emitting PAUSE frames. 292The default is 3 (both rx_pause and tx_pause = 1). 293This tunable establishes the default PAUSE settings for all ports. 294Settings can be displayed and controlled on a per-port basis via the 295dev.<port>.X.pause_settings sysctl. 296.It Va hw.cxgbe.fec 297FEC (Forward Error Correction) settings. 2980 diables FEC. 299Bit 0 enables RS FEC, bit 1 enables BASE-R RS, bit 3 is reserved. 300The default is -1 which lets the driver pick a value. 301This tunable establishes the default FEC settings for all ports. 302Settings can be displayed and controlled on a per-port basis via the 303dev.<port>.X.fec sysctl. 304.It Va hw.cxgbe.autoneg 305Link autonegotiation settings. 306This tunable establishes the default autonegotiation settings for all ports. 307Settings can be displayed and controlled on a per-port basis via the 308dev.<port>.X.autoneg sysctl. 3090 disables autonegotiation. 3101 enables autonegotiation. 311The default is -1 which lets the driver pick a value. 312dev.<port>.X.autoneg is -1 for port and module combinations that do not support 313autonegotiation. 314.It Va hw.cxgbe.buffer_packing 315Allow the hardware to deliver multiple frames in the same receive buffer 316opportunistically. 317The default is -1 which lets the driver decide. 3180 or 1 explicitly disable or enable this feature. 319.It Va hw.cxgbe.allow_mbufs_in_cluster 3201 allows the driver to lay down one or more mbufs within the receive buffer 321opportunistically. 322This is the default. 3230 prohibits the driver from doing so. 324.It Va hw.cxgbe.largest_rx_cluster 325.It Va hw.cxgbe.safest_rx_cluster 326Sizes of rx clusters. 327Each of these must be set to one of the sizes available 328(usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater 329than or equal to safest_rx_cluster. 330The defaults are 16384 and 4096 respectively. 331The driver never attempts to allocate a receive buffer larger than 332largest_rx_cluster and falls back to allocating buffers of 333safest_rx_cluster size if an allocation larger than safest_rx_cluster fails. 334Note that largest_rx_cluster merely establishes a ceiling -- the driver is 335allowed to allocate buffers of smaller sizes. 336.It Va hw.cxgbe.config_file 337Select a pre-packaged device configuration file. 338A configuration file contains a recipe for partitioning and configuring the 339hardware resources on the card. 340This tunable is for specialized applications only and should not be used in 341normal operation. 342The configuration profile currently in use is available in the dev.<nexus>.X.cf 343and dev.<nexus>.X.cfcsum sysctls. 344.It Va hw.cxgbe.linkcaps_allowed 345.It Va hw.cxgbe.niccaps_allowed 346.It Va hw.cxgbe.toecaps_allowed 347.It Va hw.cxgbe.rdmacaps_allowed 348.It Va hw.cxgbe.iscsicaps_allowed 349.It Va hw.cxgbe.fcoecaps_allowed 350Disallowing capabilities provides a hint to the driver and firmware to not 351reserve hardware resources for that feature. 352Each of these is a bit field with a bit for each sub-capability within the 353capability. 354This tunable is for specialized applications only and should not be used in 355normal operation. 356The capabilities for which hardware resources have been reserved are listed in 357dev.<nexus>.X.*caps sysctls. 358.El 359.Sh SUPPORT 360For general information and support, 361go to the Chelsio support website at: 362.Pa http://www.chelsio.com/ . 363.Pp 364If an issue is identified with this driver with a supported adapter, 365email all the specific information related to the issue to 366.Aq Mt support@chelsio.com . 367.Sh SEE ALSO 368.Xr altq 4 , 369.Xr arp 4 , 370.Xr cxgb 4 , 371.Xr cxgbev 4 , 372.Xr netintro 4 , 373.Xr ng_ether 4 , 374.Xr ifconfig 8 375.Sh HISTORY 376The 377.Nm 378device driver first appeared in 379.Fx 9.0 . 380Support for T5 cards first appeared in 381.Fx 9.2 382and 383.Fx 10.0 . 384Support for T6 cards first appeared in 385.Fx 11.1 386and 387.Fx 12.0 . 388.Sh AUTHORS 389.An -nosplit 390The 391.Nm 392driver was written by 393.An Navdeep Parhar Aq Mt np@FreeBSD.org . 394