1.\" Copyright (c) 2001-2003, Intel Corporation 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 Intel Corporation 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: src/share/man/man4/igb.4,v 1.2 2010/05/14 20:11:30 marius Exp $ 33.\" 34.Dd June 21, 2013 35.Dt IGB 4 36.Os 37.Sh NAME 38.Nm igb 39.Nd "Intel(R) PRO/1000 PCI Express Gigabit Ethernet adapter driver" 40.Sh SYNOPSIS 41To compile this driver into the kernel, 42place the following line in your 43kernel configuration file: 44.Bd -ragged -offset indent 45.Cd "device ig_hal" 46.Cd "device igb" 47.Ed 48.Pp 49Alternatively, to load the driver as a 50module at boot time, place the following line in 51.Xr loader.conf 5 : 52.Bd -literal -offset indent 53ig_hal_load="YES" 54if_igb_load="YES" 55.Ed 56.Sh DESCRIPTION 57The 58.Nm 59driver provides support for PCI Express Gigabit Ethernet adapters based on 60the Intel 6182575, 6282576, 6382580, 64I210, 65I211, 66and I350 67Ethernet controller chips. 68The 69.Nm 70driver supports: 71.Pp 72.Bl -item -offset indent -compact 73.It 74Transmit/Receive checksum offload for IP/UDP/TCP. 75.\"Jumbo Frames. 76.It 77Interrupt moderation 78.It 79TCP segmentation offload (TSO) 80.It 81Receive side scaling (RSS) 82.It 83Multiple tranmission queues 84.It 85Multiple vector MSI-X 86.It 87VLAN tag stripping and inserting 88.El 89.Pp 90If 91.Xr polling 4 92or MSI-X is used, 93by default, 94the 95.Nm 96driver will try enabling as many reception queues and transmission queues 97as are allowed by the number of CPUs in the system. 98.Pp 99If multiple transmission queues are used, 100the round-robin arbitration is performed among the transmission queues. 101It should be noted that 102if both TSO and multiple transmission queues are used, 103for 82575, the round-robin arbitration between transmission queues is done 104at the TSO packet boundary; 105for the reset of the hardwares, the round-robin arbitration 106between transmission queues is done at the TCP segment boundary after 107the hardware segmentation is performed. 108.Pp 10982575 supports 4 reception queues and 4 transmission queues. 110MSI-X is not enabled due to hardware errata. 111Under MSI or legacy interrupt mode, 1122 reception queues are enabled for hardware RSS hash 113and only 1 transmission queue is enable. 114.Pp 11582576 supports 16 reception queues and 16 transmission queues. 116MSI-X is enable by default. 117However, 118due to the number of MSI-X vectors is 10, 119at most 8 reception queues and 8 transmission queues will be enabled 120under MSI-X mode. 121When 122.Xr polling 4 123is enabled on the devices, 124at most 16 reception queues and 16 transmission queues will be enabled. 125.Pp 12682580 and I350 support 8 reception queues and 8 transmission queues. 127MSI-X is enabled by default. 128.Pp 129I210 supports 4 reception queues and 4 transmission queues. 130MSI-X is enabled by default. 131.Pp 132I211 supports 2 reception queues and 2 transmission queues. 133MSI-X is enabled by default. 134.Pp 135.\"For questions related to hardware requirements, refer to the 136.\"documentation supplied with your Intel PRO/1000 adapter. 137.\"All hardware requirements listed apply to use with 138.\".Dx . 139.\".Pp 140.\"Support for Jumbo Frames is provided via the interface MTU setting. 141.\"Selecting an MTU larger than 1500 bytes with the 142.\".Xr ifconfig 8 143.\"utility configures the adapter to receive and transmit Jumbo Frames. 144.\"The maximum MTU size for Jumbo Frames is 9216. 145.\".Pp 146The 147.Nm 148driver supports the following media types: 149.Bl -tag -width ".Cm 10baseT/UTP" 150.It Cm autoselect 151Enables auto-negotiation for speed and duplex. 152.It Cm 10baseT/UTP 153Sets 10Mbps operation. 154Use the 155.Cm mediaopt 156option to select 157.Cm full-duplex 158mode. 159.It Cm 100baseTX 160Sets 100Mbps operation. 161Use the 162.Cm mediaopt 163option to select 164.Cm full-duplex 165mode. 166.It Cm 1000baseSX 167Sets 1000Mbps operation. 168Only 169.Cm full-duplex 170mode is supported at this speed. 171.It Cm 1000baseT 172Sets 1000Mbps operation. 173Only 174.Cm full-duplex 175mode is supported at this speed. 176.El 177.Pp 178The 179.Nm 180driver supports the following media options: 181.Bl -tag -width ".Cm full-duplex" 182.It Cm full-duplex 183Forces full-duplex operation 184.It Cm half-duplex 185Forces half-duplex operation. 186.El 187.Pp 188Only use 189.Cm mediaopt 190to set the driver to 191.Cm full-duplex . 192If 193.Cm mediaopt 194is not specified, the driver defaults to 195.Cm half-duplex . 196.Pp 197For more information on configuring this device, see 198.Xr ifconfig 8 . 199The 200.Nm 201driver supports 202.Xr polling 4 . 203.Sh HARDWARE 204The 205.Nm 206driver supports Gigabit Ethernet adapters based on the Intel 20782575, 20882576, 20982580, 210I210, 211I211, 212and I350 213controller chips: 214.Pp 215.Bl -bullet -compact 216.It 217Intel 82575EB Gigabit Ethernet Controller 218.It 219Intel 82576 Gigabit Ethernet Controller 220.It 221Intel 82580EB Gigabit Ethernet Controller 222.It 223Intel Ethernet Controller I210 Series 224.It 225Intel Ethernet Controller I211 Series 226.It 227Intel Ethernet Controller I350 228.It 229Intel Ethernet Server Adapter I210-T1 230.It 231Intel Ethernet Server Adapter I340-F4 232.It 233Intel Ethernet Server Adapter I340-T4 234.It 235Intel Ethernet Server Adapter I350-F2 236.It 237Intel Ethernet Server Adapter I350-F4 238.It 239Intel Ethernet Server Adapter I350-T2 240.It 241Intel Ethernet Server Adapter I350-T4 242.It 243Intel Gigabit EF Dual Port Server Adapter 244.It 245Intel Gigabit ET Dual Port Server Adapter 246.It 247Intel Gigabit ET Quad Port Server Adapter 248.It 249Intel Gigabit ET2 Quad Port Server Adapter 250.It 251Intel Gigabit VT Quad Port Server Adapter 252.El 253.Sh TUNABLES 254Tunables can be set at the 255.Xr loader 8 256prompt before booting the kernel or stored in 257.Xr loader.conf 5 . 258.Em X 259is the device unit number. 260.Bl -tag -width ".Va hw.igbX.irq.unshared" 261.It Va hw.igb.rxd Va hw.igbX.rxd 262Number of receive descriptors allocated by the driver. 263The default value is 512. 264The minimum is 256, 265and the maximum is 4096. 266.It Va hw.igb.txd Va hw.igbX.txd 267Number of transmit descriptors allocated by the driver. 268The default value is 1024. 269The minimum is 256, 270and the maximum is 4096. 271.It Va hw.igb.rxr Va hw.igbX.rxr 272This tunable specifies the number of reception queues could be enabled. 273Maximum allowed value for these tunables is device specific 274and it must be power of 2 aligned. 275Setting these tunables to 0 allows the driver to make 276as many reception queues ready-for-use as allowed by the number of CPUs. 277.It Va hw.igb.txr Va hw.igbX.txr 278This tunable specifies the number of transmission queues could be enabled. 279Maximum allowed value for these tunables is device specific 280and it must be power of 2 aligned. 281Setting these tunables to 0 allows the driver to make 282as many transmission queues ready-for-use as allowed by the number of CPUs. 283.It Va hw.igb.msix.enable Va hw.igbX.msix.enable 284By default, 285the driver will use MSI-X if it is supported. 286This behaviour can be turned off by setting this tunable to 0. 287.It Va hw.igbX.msix.off 288If MSI-X is used 289and the number of MSI-X vectors is not enough to 290put transmission queue processing and reception queue processing 291onto independent MSI-X vector, 292this tunable specifies the leading target CPU for 293transmission and reception queues processing. 294The value specificed must be aligned to the maximum of 295the number of reception queues 296and the number of transmission queues enabled, 297and must be less than the power of 2 number of CPUs. 298.It Va hw.igbX.msix.rxoff 299If MSI-X is used 300and the number of MSI-X vectors is enough to 301put transmission queue processing and reception queue processing 302onto independent MSI-X vector, 303this tunable specifies the leading target CPU for reception queues processing. 304The value specificed must be aligned to the number of reception queues enabled 305and must be less than the power of 2 number of CPUs. 306.It Va hw.igbX.msix.txoff 307If MSI-X is used 308and the number of MSI-X vectors is enough to 309put transmission queue processing and reception queue processing 310onto independent MSI-X vector, 311this tunable specifies the leading target CPU 312for transmission queues processing. 313The value specificed must be aligned to 314the number of transmission queues enabled 315and must be less than the power of 2 number of CPUs. 316.It Va hw.igb.msi.enable Va hw.igbX.msi.enable 317If MSI-X is disabled and MSI is supported, 318the driver will use MSI. 319This behavior can be turned off by setting this tunable to 0. 320.It Va hw.igbX.msi.cpu 321If MSI is used, 322it specifies the MSI's target CPU. 323.It Va hw.igbX.irq.unshared 324If legacy interrupt is used, 325by default, 326the driver assumes the interrupt could be shared. 327Setting this tunable to 1 allows the driver to perform certain 328optimization based on the knowledge that the interrupt is not shared. 329.It Va hw.igbX.npoll.txoff 330This tunable specifies the leading target CPU for 331transmission queue 332.Xr polling 4 333processing. 334The value specificed must be aligned to the number of transmission queues 335enabled and must be less than the power of 2 number of CPUs. 336.It Va hw.igbX.npoll.rxoff 337This tunable specifies the leading target CPU for 338reception queue 339.Xr polling 4 340processing. 341The value specificed must be aligned to the number of reception queues 342enabled and must be less than the power of 2 number of CPUs. 343.El 344.Sh MIB Variables 345A number of per-interface variables are implemented in the 346.Va hw.igb Ns Em X 347branch of the 348.Xr sysctl 3 349MIB. 350.Bl -tag -width "tx_intr_nsegs" 351.It Va rxr 352Number of reception queues could be enabled (read-only). 353Use the tunable 354.Va hw.igb.rxr 355or 356.Va hw.igbX.rxr 357to configure it. 358.It Va rxr_inuse 359Number of reception queues being used (read-only). 360.It Va txr 361Number of transmission queues could be enabled (read-only). 362Use the tunable 363.Va hw.igb.txr 364or 365.Va hw.igbX.txr 366to configure it. 367.It Va txr_inuse 368Number of transmission queues being used (read-only). 369.It Va rxd 370Number of descriptors per reception queue (read-only). 371Use the tunable 372.Va hw.igb.rxd 373or 374.Va hw.igbX.rxd 375to configure it. 376.It Va txd 377Number of descriptors per transmission queue (read-only). 378Use the tunable 379.Va hw.igb.txd 380or 381.Va hw.igbX.txd 382to configure it. 383.It Va intr_rate 384If MSI or legacy interrupt is used, 385this sysctl controls the highest possible frequency 386that interrupt could be generated by the device. 387It is 6000 by default (~150us). 388.It Va msixY_rate 389If MSI-X is used, 390this sysctl controls the highest possible frequency 391that interrupt could be generated by the 392.Em Y 393MSI-X vector. 394For reception processing only MSI-X vector, 395the default value is 6000 (~150us). 396For transmission processing only MSI-X vector, 397the default value is 4000 (250us). 398For MSI-X vector which handles both reception and transmission, 399the default value is 6000 (~150us). 400.It Va tx_intr_nsegs 401Transmission interrupt is asked to be generated upon every 402.Va tx_intr_nsegs 403transmission descritors having been setup. 404The default value is 1/16 of the number of transmission descriptors per queue. 405.It Va tx_wreg_nsegs 406The number of transmission descriptors should be setup 407before the hardware register is written. 408Setting this value too high will have negative effect 409on transmission timeliness. 410Setting this value too low will hurt overall transmission performance 411due to the frequent hardware register writing. 412The default value is 8. 413.It Va rx_wreg_nsegs 414The number of reception descriptors should be setup 415before the hardware register is written. 416Setting this value too high will make device drop incoming packets. 417Setting this value too low will hurt overall reception performance 418due to the frequent hardware register writing. 419The default value is 32. 420.It Va npoll_rxoff 421See the tunable 422.Va hw.igbX.npoll.rxoff . 423The set value will take effect the next time 424.Xr polling 4 425is enabled on the device. 426.It Va npoll_txoff 427See the tunable 428.Va hw.igbX.npoll.txoff . 429The set value will take effect the next time 430.Xr polling 4 431is enabled on the device. 432.El 433.\".Sh SUPPORT 434.\"For general information and support, 435.\"go to the Intel support website at: 436.\".Pa http://support.intel.com . 437.\".Pp 438.\"If an issue is identified with the released source code on the supported kernel 439.\"with a supported adapter, email the specific information related to the 440.\"issue to 441.\".Aq Mt freebsdnic@mailbox.intel.com . 442.Sh SEE ALSO 443.Xr altq 4 , 444.Xr arp 4 , 445.Xr ifmedia 4 , 446.Xr netintro 4 , 447.Xr ng_ether 4 , 448.Xr polling 4 , 449.Xr vlan 4 , 450.Xr ifconfig 8 451.Sh HISTORY 452The 453.Nm 454device driver first appeared in 455.Fx 7.1 . 456.Sh AUTHORS 457The 458.Nm 459driver was written by 460.An Intel Corporation Aq Mt freebsdnic@mailbox.intel.com . 461