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 Dec 10, 2019 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.ntxq 176Number of NIC tx queues used for a port. 177The default is 16 or the number 178of CPU cores in the system, whichever is less. 179.It Va hw.cxgbe.nrxq 180Number of NIC rx queues used for a port. 181The default is 8 or the number 182of CPU cores in the system, whichever is less. 183.It Va hw.cxgbe.nofldtxq 184Number of TOE tx queues used for a port. 185The default is 8 or the 186number of CPU cores in the system, whichever is less. 187.It Va hw.cxgbe.nofldrxq 188Number of TOE rx queues used for a port. 189The default is 2 or the 190number of CPU cores in the system, whichever is less. 191.It Va hw.cxgbe.num_vis 192Number of virtual interfaces (VIs) created for each port. 193Each virtual interface creates a separate network interface. 194The first virtual interface on each port is required and represents 195the primary network interface on the port. 196Additional virtual interfaces on a port are named using the Virtual Interface 197name from the table above. 198Additional virtual interfaces use a single pair of queues 199for rx and tx as well an additional pair of queues for TOE rx and tx. 200The default is 1. 201.It Va hw.cxgbe.holdoff_timer_idx 202.It Va hw.cxgbe.holdoff_timer_idx_ofld 203Timer index value used to delay interrupts. 204The holdoff timer list has the values 1, 5, 10, 50, 100, and 200 205by default (all values are in microseconds) and the index selects a 206value from this list. 207holdoff_timer_idx_ofld applies to queues used for TOE rx. 208The default value is 1 which means the timer value is 5us. 209Different interfaces can be assigned different values at any time via the 210dev.<port>.X.holdoff_tmr_idx and dev.<port>.X.holdoff_tmr_idx_ofld sysctls. 211.It Va hw.cxgbe.holdoff_pktc_idx 212.It Va hw.cxgbe.holdoff_pktc_idx_ofld 213Packet-count index value used to delay interrupts. 214The packet-count list has the values 1, 8, 16, and 32 by default, 215and the index selects a value from this list. 216holdoff_pktc_idx_ofld applies to queues used for TOE rx. 217The default value is -1 which means packet counting is disabled and interrupts 218are generated based solely on the holdoff timer value. 219Different interfaces can be assigned different values via the 220dev.<port>.X.holdoff_pktc_idx and dev.<port>.X.holdoff_pktc_idx_ofld sysctls. 221These sysctls work only when the interface has never been marked up (as done by 222ifconfig up). 223.It Va hw.cxgbe.qsize_txq 224Number of entries in a transmit queue's descriptor ring. 225A buf_ring of the same size is also allocated for additional 226software queuing. 227See 228.Xr ifnet 9 . 229The default value is 1024. 230Different interfaces can be assigned different values via the 231dev.<port>.X.qsize_txq sysctl. 232This sysctl works only when the interface has never been marked up (as done by 233ifconfig up). 234.It Va hw.cxgbe.qsize_rxq 235Number of entries in a receive queue's descriptor ring. 236The default value is 1024. 237Different interfaces can be assigned different values via the 238dev.<port>.X.qsize_rxq sysctl. 239This sysctl works only when the interface has never been marked up (as done by 240ifconfig up). 241.It Va hw.cxgbe.interrupt_types 242Permitted interrupt types. 243Bit 0 represents INTx (line interrupts), bit 1 MSI, and bit 2 MSI-X. 244The default is 7 (all allowed). 245The driver selects the best possible type out of the allowed types. 246.It Va hw.cxgbe.pcie_relaxed_ordering 247PCIe Relaxed Ordering. 248-1 indicates the driver should determine whether to enable or disable PCIe RO. 2490 disables PCIe RO. 2501 enables PCIe RO. 2512 indicates the driver should not modify the PCIe RO setting. 252The default is -1. 253.It Va hw.cxgbe.fw_install 2540 prohibits the driver from installing a firmware on the card. 2551 allows the driver to install a new firmware if internal driver 256heuristics indicate that the new firmware is preferable to the one 257already on the card. 2582 instructs the driver to always install the new firmware on the card as 259long as it is compatible with the driver and is a different version than 260the one already on the card. 261The default is 1. 262.It Va hw.cxgbe.fl_pktshift 263Number of padding bytes inserted before the beginning of an Ethernet 264frame in the receive buffer. 265The default value is 0. 266A value of of 2 would ensure that the Ethernet payload (usually the IP header) 267is at a 4 byte aligned address. 2680-7 are all valid values. 269.It Va hw.cxgbe.fl_pad 270A non-zero value ensures that writes from the hardware to a receive buffer are 271padded up to the specified boundary. 272The default is -1 which lets the driver pick a pad boundary. 2730 disables trailer padding completely. 274.It Va hw.cxgbe.cong_drop 275Controls the hardware response to congestion. 276-1 disables congestion feedback and is not recommended. 2770 instructs the hardware to backpressure its pipeline on congestion. 278This usually results in the port emitting PAUSE frames. 2791 instructs the hardware to drop frames destined for congested queues. 280.It Va hw.cxgbe.pause_settings 281PAUSE frame settings. 282Bit 0 is rx_pause, bit 1 is tx_pause, bit 2 is pause_autoneg. 283rx_pause = 1 instructs the hardware to heed incoming PAUSE frames, 0 instructs 284it to ignore them. 285tx_pause = 1 allows the hardware to emit PAUSE frames when its receive FIFO 286reaches a high threshold, 0 prohibits the hardware from emitting PAUSE frames. 287pause_autoneg = 1 overrides the rx_pause and tx_pause bits and instructs the 288hardware to negotiate PAUSE settings with the link peer. 289The default is 7 (all three = 1). 290This tunable establishes the default PAUSE settings for all ports. 291Settings can be displayed and controlled on a per-port basis via the 292dev.<port>.X.pause_settings sysctl. 293.It Va hw.cxgbe.fec 294Forward Error Correction settings. 295-1 (default) means driver should automatically pick a value. 2960 disables FEC. 297Finer grained control can be achieved by setting individual bits. 298Bit 0 enables RS FEC, bit 1 enables BASE-R FEC (aka Firecode FEC), bit 2992 enables NO FEC, and bit 6 enables the FEC that is recommended by the 300transceiver/cable that is plugged in. 301These bits can be set together in any combination. 302This tunable establishes the default FEC settings for all ports. 303Settings can be displayed and controlled on a per-port basis via the 304dev.<port>.X.fec sysctl. 305.It Va hw.cxgbe.autoneg 306Link autonegotiation settings. 307This tunable establishes the default autonegotiation settings for all ports. 308Settings can be displayed and controlled on a per-port basis via the 309dev.<port>.X.autoneg sysctl. 3100 disables autonegotiation. 3111 enables autonegotiation. 312The default is -1 which lets the driver pick a value. 313dev.<port>.X.autoneg is -1 for port and module combinations that do not support 314autonegotiation. 315.It Va hw.cxgbe.buffer_packing 316Allow the hardware to deliver multiple frames in the same receive buffer 317opportunistically. 318The default is -1 which lets the driver decide. 3190 or 1 explicitly disable or enable this feature. 320.It Va hw.cxgbe.allow_mbufs_in_cluster 3211 allows the driver to lay down one or more mbufs within the receive buffer 322opportunistically. 323This is the default. 3240 prohibits the driver from doing so. 325.It Va hw.cxgbe.largest_rx_cluster 326.It Va hw.cxgbe.safest_rx_cluster 327Sizes of rx clusters. 328Each of these must be set to one of the sizes available 329(usually 2048, 4096, 9216, and 16384) and largest_rx_cluster must be greater 330than or equal to safest_rx_cluster. 331The defaults are 16384 and 4096 respectively. 332The driver never attempts to allocate a receive buffer larger than 333largest_rx_cluster and falls back to allocating buffers of 334safest_rx_cluster size if an allocation larger than safest_rx_cluster fails. 335Note that largest_rx_cluster merely establishes a ceiling -- the driver is 336allowed to allocate buffers of smaller sizes. 337.It Va hw.cxgbe.config_file 338Select a pre-packaged device configuration file. 339A configuration file contains a recipe for partitioning and configuring the 340hardware resources on the card. 341This tunable is for specialized applications only and should not be used in 342normal operation. 343The configuration profile currently in use is available in the dev.<nexus>.X.cf 344and dev.<nexus>.X.cfcsum sysctls. 345.It Va hw.cxgbe.linkcaps_allowed 346.It Va hw.cxgbe.niccaps_allowed 347.It Va hw.cxgbe.toecaps_allowed 348.It Va hw.cxgbe.rdmacaps_allowed 349.It Va hw.cxgbe.iscsicaps_allowed 350.It Va hw.cxgbe.fcoecaps_allowed 351Disallowing capabilities provides a hint to the driver and firmware to not 352reserve hardware resources for that feature. 353Each of these is a bit field with a bit for each sub-capability within the 354capability. 355This tunable is for specialized applications only and should not be used in 356normal operation. 357The capabilities for which hardware resources have been reserved are listed in 358dev.<nexus>.X.*caps sysctls. 359.El 360.Sh SUPPORT 361For general information and support, 362go to the Chelsio support website at: 363.Pa http://www.chelsio.com/ . 364.Pp 365If an issue is identified with this driver with a supported adapter, 366email all the specific information related to the issue to 367.Aq Mt support@chelsio.com . 368.Sh SEE ALSO 369.Xr altq 4 , 370.Xr arp 4 , 371.Xr ccr 4 , 372.Xr cxgb 4 , 373.Xr cxgbev 4 , 374.Xr netintro 4 , 375.Xr ng_ether 4 , 376.Xr ifconfig 8 377.Sh HISTORY 378The 379.Nm 380device driver first appeared in 381.Fx 9.0 . 382Support for T5 cards first appeared in 383.Fx 9.2 384and 385.Fx 10.0 . 386Support for T6 cards first appeared in 387.Fx 11.1 388and 389.Fx 12.0 . 390.Sh AUTHORS 391.An -nosplit 392The 393.Nm 394driver was written by 395.An Navdeep Parhar Aq Mt np@FreeBSD.org . 396