1Linux* Base Driver for the Intel(R) PRO/1000 Family of Adapters 2=============================================================== 3 4September 26, 2006 5 6 7Contents 8======== 9 10- In This Release 11- Identifying Your Adapter 12- Building and Installation 13- Command Line Parameters 14- Speed and Duplex Configuration 15- Additional Configurations 16- Known Issues 17- Support 18 19 20In This Release 21=============== 22 23This file describes the Linux* Base Driver for the Intel(R) PRO/1000 Family 24of Adapters. This driver includes support for Itanium(R)2-based systems. 25 26For questions related to hardware requirements, refer to the documentation 27supplied with your Intel PRO/1000 adapter. All hardware requirements listed 28apply to use with Linux. 29 30The following features are now available in supported kernels: 31 - Native VLANs 32 - Channel Bonding (teaming) 33 - SNMP 34 35Channel Bonding documentation can be found in the Linux kernel source: 36/Documentation/networking/bonding.txt 37 38The driver information previously displayed in the /proc filesystem is not 39supported in this release. Alternatively, you can use ethtool (version 1.6 40or later), lspci, and ifconfig to obtain the same information. 41 42Instructions on updating ethtool can be found in the section "Additional 43Configurations" later in this document. 44 45NOTE: The Intel(R) 82562v 10/100 Network Connection only provides 10/100 46support. 47 48 49Identifying Your Adapter 50======================== 51 52For more information on how to identify your adapter, go to the Adapter & 53Driver ID Guide at: 54 55 http://support.intel.com/support/network/adapter/pro100/21397.htm 56 57For the latest Intel network drivers for Linux, refer to the following 58website. In the search field, enter your adapter name or type, or use the 59networking link on the left to search for your adapter: 60 61 http://downloadfinder.intel.com/scripts-df/support_intel.asp 62 63 64Command Line Parameters 65======================= 66 67If the driver is built as a module, the following optional parameters 68are used by entering them on the command line with the modprobe command 69using this syntax: 70 71 modprobe e1000 [<option>=<VAL1>,<VAL2>,...] 72 73For example, with two PRO/1000 PCI adapters, entering: 74 75 modprobe e1000 TxDescriptors=80,128 76 77loads the e1000 driver with 80 TX descriptors for the first adapter and 78128 TX descriptors for the second adapter. 79 80The default value for each parameter is generally the recommended setting, 81unless otherwise noted. 82 83NOTES: For more information about the AutoNeg, Duplex, and Speed 84 parameters, see the "Speed and Duplex Configuration" section in 85 this document. 86 87 For more information about the InterruptThrottleRate, 88 RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay 89 parameters, see the application note at: 90 http://www.intel.com/design/network/applnots/ap450.htm 91 92 A descriptor describes a data buffer and attributes related to 93 the data buffer. This information is accessed by the hardware. 94 95 96AutoNeg 97------- 98(Supported only on adapters with copper connections) 99Valid Range: 0x01-0x0F, 0x20-0x2F 100Default Value: 0x2F 101 102This parameter is a bit-mask that specifies the speed and duplex settings 103advertised by the adapter. When this parameter is used, the Speed and 104Duplex parameters must not be specified. 105 106NOTE: Refer to the Speed and Duplex section of this readme for more 107 information on the AutoNeg parameter. 108 109 110Duplex 111------ 112(Supported only on adapters with copper connections) 113Valid Range: 0-2 (0=auto-negotiate, 1=half, 2=full) 114Default Value: 0 115 116This defines the direction in which data is allowed to flow. Can be 117either one or two-directional. If both Duplex and the link partner are 118set to auto-negotiate, the board auto-detects the correct duplex. If the 119link partner is forced (either full or half), Duplex defaults to half- 120duplex. 121 122 123FlowControl 124----------- 125Valid Range: 0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx) 126Default Value: Reads flow control settings from the EEPROM 127 128This parameter controls the automatic generation(Tx) and response(Rx) 129to Ethernet PAUSE frames. 130 131 132InterruptThrottleRate 133--------------------- 134(not supported on Intel(R) 82542, 82543 or 82544-based adapters) 135Valid Range: 0,1,3,100-100000 (0=off, 1=dynamic, 3=dynamic conservative) 136Default Value: 3 137 138The driver can limit the amount of interrupts per second that the adapter 139will generate for incoming packets. It does this by writing a value to the 140adapter that is based on the maximum amount of interrupts that the adapter 141will generate per second. 142 143Setting InterruptThrottleRate to a value greater or equal to 100 144will program the adapter to send out a maximum of that many interrupts 145per second, even if more packets have come in. This reduces interrupt 146load on the system and can lower CPU utilization under heavy load, 147but will increase latency as packets are not processed as quickly. 148 149The default behaviour of the driver previously assumed a static 150InterruptThrottleRate value of 8000, providing a good fallback value for 151all traffic types,but lacking in small packet performance and latency. 152The hardware can handle many more small packets per second however, and 153for this reason an adaptive interrupt moderation algorithm was implemented. 154 155Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in which 156it dynamically adjusts the InterruptThrottleRate value based on the traffic 157that it receives. After determining the type of incoming traffic in the last 158timeframe, it will adjust the InterruptThrottleRate to an appropriate value 159for that traffic. 160 161The algorithm classifies the incoming traffic every interval into 162classes. Once the class is determined, the InterruptThrottleRate value is 163adjusted to suit that traffic type the best. There are three classes defined: 164"Bulk traffic", for large amounts of packets of normal size; "Low latency", 165for small amounts of traffic and/or a significant percentage of small 166packets; and "Lowest latency", for almost completely small packets or 167minimal traffic. 168 169In dynamic conservative mode, the InterruptThrottleRate value is set to 4000 170for traffic that falls in class "Bulk traffic". If traffic falls in the "Low 171latency" or "Lowest latency" class, the InterruptThrottleRate is increased 172stepwise to 20000. This default mode is suitable for most applications. 173 174For situations where low latency is vital such as cluster or 175grid computing, the algorithm can reduce latency even more when 176InterruptThrottleRate is set to mode 1. In this mode, which operates 177the same as mode 3, the InterruptThrottleRate will be increased stepwise to 17870000 for traffic in class "Lowest latency". 179 180Setting InterruptThrottleRate to 0 turns off any interrupt moderation 181and may improve small packet latency, but is generally not suitable 182for bulk throughput traffic. 183 184NOTE: InterruptThrottleRate takes precedence over the TxAbsIntDelay and 185 RxAbsIntDelay parameters. In other words, minimizing the receive 186 and/or transmit absolute delays does not force the controller to 187 generate more interrupts than what the Interrupt Throttle Rate 188 allows. 189 190CAUTION: If you are using the Intel(R) PRO/1000 CT Network Connection 191 (controller 82547), setting InterruptThrottleRate to a value 192 greater than 75,000, may hang (stop transmitting) adapters 193 under certain network conditions. If this occurs a NETDEV 194 WATCHDOG message is logged in the system event log. In 195 addition, the controller is automatically reset, restoring 196 the network connection. To eliminate the potential for the 197 hang, ensure that InterruptThrottleRate is set no greater 198 than 75,000 and is not set to 0. 199 200NOTE: When e1000 is loaded with default settings and multiple adapters 201 are in use simultaneously, the CPU utilization may increase non- 202 linearly. In order to limit the CPU utilization without impacting 203 the overall throughput, we recommend that you load the driver as 204 follows: 205 206 modprobe e1000 InterruptThrottleRate=3000,3000,3000 207 208 This sets the InterruptThrottleRate to 3000 interrupts/sec for 209 the first, second, and third instances of the driver. The range 210 of 2000 to 3000 interrupts per second works on a majority of 211 systems and is a good starting point, but the optimal value will 212 be platform-specific. If CPU utilization is not a concern, use 213 RX_POLLING (NAPI) and default driver settings. 214 215 216 217RxDescriptors 218------------- 219Valid Range: 80-256 for 82542 and 82543-based adapters 220 80-4096 for all other supported adapters 221Default Value: 256 222 223This value specifies the number of receive buffer descriptors allocated 224by the driver. Increasing this value allows the driver to buffer more 225incoming packets, at the expense of increased system memory utilization. 226 227Each descriptor is 16 bytes. A receive buffer is also allocated for each 228descriptor and can be either 2048, 4096, 8192, or 16384 bytes, depending 229on the MTU setting. The maximum MTU size is 16110. 230 231NOTE: MTU designates the frame size. It only needs to be set for Jumbo 232 Frames. Depending on the available system resources, the request 233 for a higher number of receive descriptors may be denied. In this 234 case, use a lower number. 235 236 237RxIntDelay 238---------- 239Valid Range: 0-65535 (0=off) 240Default Value: 0 241 242This value delays the generation of receive interrupts in units of 1.024 243microseconds. Receive interrupt reduction can improve CPU efficiency if 244properly tuned for specific network traffic. Increasing this value adds 245extra latency to frame reception and can end up decreasing the throughput 246of TCP traffic. If the system is reporting dropped receives, this value 247may be set too high, causing the driver to run out of available receive 248descriptors. 249 250CAUTION: When setting RxIntDelay to a value other than 0, adapters may 251 hang (stop transmitting) under certain network conditions. If 252 this occurs a NETDEV WATCHDOG message is logged in the system 253 event log. In addition, the controller is automatically reset, 254 restoring the network connection. To eliminate the potential 255 for the hang ensure that RxIntDelay is set to 0. 256 257 258RxAbsIntDelay 259------------- 260(This parameter is supported only on 82540, 82545 and later adapters.) 261Valid Range: 0-65535 (0=off) 262Default Value: 128 263 264This value, in units of 1.024 microseconds, limits the delay in which a 265receive interrupt is generated. Useful only if RxIntDelay is non-zero, 266this value ensures that an interrupt is generated after the initial 267packet is received within the set amount of time. Proper tuning, 268along with RxIntDelay, may improve traffic throughput in specific network 269conditions. 270 271 272Speed 273----- 274(This parameter is supported only on adapters with copper connections.) 275Valid Settings: 0, 10, 100, 1000 276Default Value: 0 (auto-negotiate at all supported speeds) 277 278Speed forces the line speed to the specified value in megabits per second 279(Mbps). If this parameter is not specified or is set to 0 and the link 280partner is set to auto-negotiate, the board will auto-detect the correct 281speed. Duplex should also be set when Speed is set to either 10 or 100. 282 283 284TxDescriptors 285------------- 286Valid Range: 80-256 for 82542 and 82543-based adapters 287 80-4096 for all other supported adapters 288Default Value: 256 289 290This value is the number of transmit descriptors allocated by the driver. 291Increasing this value allows the driver to queue more transmits. Each 292descriptor is 16 bytes. 293 294NOTE: Depending on the available system resources, the request for a 295 higher number of transmit descriptors may be denied. In this case, 296 use a lower number. 297 298 299TxIntDelay 300---------- 301Valid Range: 0-65535 (0=off) 302Default Value: 64 303 304This value delays the generation of transmit interrupts in units of 3051.024 microseconds. Transmit interrupt reduction can improve CPU 306efficiency if properly tuned for specific network traffic. If the 307system is reporting dropped transmits, this value may be set too high 308causing the driver to run out of available transmit descriptors. 309 310 311TxAbsIntDelay 312------------- 313(This parameter is supported only on 82540, 82545 and later adapters.) 314Valid Range: 0-65535 (0=off) 315Default Value: 64 316 317This value, in units of 1.024 microseconds, limits the delay in which a 318transmit interrupt is generated. Useful only if TxIntDelay is non-zero, 319this value ensures that an interrupt is generated after the initial 320packet is sent on the wire within the set amount of time. Proper tuning, 321along with TxIntDelay, may improve traffic throughput in specific 322network conditions. 323 324XsumRX 325------ 326(This parameter is NOT supported on the 82542-based adapter.) 327Valid Range: 0-1 328Default Value: 1 329 330A value of '1' indicates that the driver should enable IP checksum 331offload for received packets (both UDP and TCP) to the adapter hardware. 332 333 334Speed and Duplex Configuration 335============================== 336 337Three keywords are used to control the speed and duplex configuration. 338These keywords are Speed, Duplex, and AutoNeg. 339 340If the board uses a fiber interface, these keywords are ignored, and the 341fiber interface board only links at 1000 Mbps full-duplex. 342 343For copper-based boards, the keywords interact as follows: 344 345 The default operation is auto-negotiate. The board advertises all 346 supported speed and duplex combinations, and it links at the highest 347 common speed and duplex mode IF the link partner is set to auto-negotiate. 348 349 If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps 350 is advertised (The 1000BaseT spec requires auto-negotiation.) 351 352 If Speed = 10 or 100, then both Speed and Duplex should be set. Auto- 353 negotiation is disabled, and the AutoNeg parameter is ignored. Partner 354 SHOULD also be forced. 355 356The AutoNeg parameter is used when more control is required over the 357auto-negotiation process. It should be used when you wish to control which 358speed and duplex combinations are advertised during the auto-negotiation 359process. 360 361The parameter may be specified as either a decimal or hexadecimal value as 362determined by the bitmap below. 363 364Bit position 7 6 5 4 3 2 1 0 365Decimal Value 128 64 32 16 8 4 2 1 366Hex value 80 40 20 10 8 4 2 1 367Speed (Mbps) N/A N/A 1000 N/A 100 100 10 10 368Duplex Full Full Half Full Half 369 370Some examples of using AutoNeg: 371 372 modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half) 373 modprobe e1000 AutoNeg=1 (Same as above) 374 modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full) 375 modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full) 376 modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half) 377 modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100 378 Half) 379 modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full) 380 modprobe e1000 AutoNeg=32 (Same as above) 381 382Note that when this parameter is used, Speed and Duplex must not be specified. 383 384If the link partner is forced to a specific speed and duplex, then this 385parameter should not be used. Instead, use the Speed and Duplex parameters 386previously mentioned to force the adapter to the same speed and duplex. 387 388 389Additional Configurations 390========================= 391 392 Configuring the Driver on Different Distributions 393 ------------------------------------------------- 394 Configuring a network driver to load properly when the system is started 395 is distribution dependent. Typically, the configuration process involves 396 adding an alias line to /etc/modules.conf or /etc/modprobe.conf as well 397 as editing other system startup scripts and/or configuration files. Many 398 popular Linux distributions ship with tools to make these changes for you. 399 To learn the proper way to configure a network device for your system, 400 refer to your distribution documentation. If during this process you are 401 asked for the driver or module name, the name for the Linux Base Driver 402 for the Intel(R) PRO/1000 Family of Adapters is e1000. 403 404 As an example, if you install the e1000 driver for two PRO/1000 adapters 405 (eth0 and eth1) and set the speed and duplex to 10full and 100half, add 406 the following to modules.conf or or modprobe.conf: 407 408 alias eth0 e1000 409 alias eth1 e1000 410 options e1000 Speed=10,100 Duplex=2,1 411 412 Viewing Link Messages 413 --------------------- 414 Link messages will not be displayed to the console if the distribution is 415 restricting system messages. In order to see network driver link messages 416 on your console, set dmesg to eight by entering the following: 417 418 dmesg -n 8 419 420 NOTE: This setting is not saved across reboots. 421 422 Jumbo Frames 423 ------------ 424 Jumbo Frames support is enabled by changing the MTU to a value larger than 425 the default of 1500. Use the ifconfig command to increase the MTU size. 426 For example: 427 428 ifconfig eth<x> mtu 9000 up 429 430 This setting is not saved across reboots. It can be made permanent if 431 you add: 432 433 MTU=9000 434 435 to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>. This example 436 applies to the Red Hat distributions; other distributions may store this 437 setting in a different location. 438 439 Notes: 440 441 - To enable Jumbo Frames, increase the MTU size on the interface beyond 442 1500. 443 444 - The maximum MTU setting for Jumbo Frames is 16110. This value coincides 445 with the maximum Jumbo Frames size of 16128. 446 447 - Using Jumbo Frames at 10 or 100 Mbps may result in poor performance or 448 loss of link. 449 450 - Some Intel gigabit adapters that support Jumbo Frames have a frame size 451 limit of 9238 bytes, with a corresponding MTU size limit of 9216 bytes. 452 The adapters with this limitation are based on the Intel(R) 82571EB, 453 82572EI, 82573L and 80003ES2LAN controller. These correspond to the 454 following product names: 455 Intel(R) PRO/1000 PT Server Adapter 456 Intel(R) PRO/1000 PT Desktop Adapter 457 Intel(R) PRO/1000 PT Network Connection 458 Intel(R) PRO/1000 PT Dual Port Server Adapter 459 Intel(R) PRO/1000 PT Dual Port Network Connection 460 Intel(R) PRO/1000 PF Server Adapter 461 Intel(R) PRO/1000 PF Network Connection 462 Intel(R) PRO/1000 PF Dual Port Server Adapter 463 Intel(R) PRO/1000 PB Server Connection 464 Intel(R) PRO/1000 PL Network Connection 465 Intel(R) PRO/1000 EB Network Connection with I/O Acceleration 466 Intel(R) PRO/1000 EB Backplane Connection with I/O Acceleration 467 Intel(R) PRO/1000 PT Quad Port Server Adapter 468 469 - Adapters based on the Intel(R) 82542 and 82573V/E controller do not 470 support Jumbo Frames. These correspond to the following product names: 471 Intel(R) PRO/1000 Gigabit Server Adapter 472 Intel(R) PRO/1000 PM Network Connection 473 474 - The following adapters do not support Jumbo Frames: 475 Intel(R) 82562V 10/100 Network Connection 476 Intel(R) 82566DM Gigabit Network Connection 477 Intel(R) 82566DC Gigabit Network Connection 478 Intel(R) 82566MM Gigabit Network Connection 479 Intel(R) 82566MC Gigabit Network Connection 480 Intel(R) 82562GT 10/100 Network Connection 481 Intel(R) 82562G 10/100 Network Connection 482 483 484 Ethtool 485 ------- 486 The driver utilizes the ethtool interface for driver configuration and 487 diagnostics, as well as displaying statistical information. Ethtool 488 version 1.6 or later is required for this functionality. 489 490 The latest release of ethtool can be found from 491 http://sourceforge.net/projects/gkernel. 492 493 NOTE: Ethtool 1.6 only supports a limited set of ethtool options. Support 494 for a more complete ethtool feature set can be enabled by upgrading 495 ethtool to ethtool-1.8.1. 496 497 Enabling Wake on LAN* (WoL) 498 --------------------------- 499 WoL is configured through the Ethtool* utility. Ethtool is included with 500 all versions of Red Hat after Red Hat 7.2. For other Linux distributions, 501 download and install Ethtool from the following website: 502 http://sourceforge.net/projects/gkernel. 503 504 For instructions on enabling WoL with Ethtool, refer to the website listed 505 above. 506 507 WoL will be enabled on the system during the next shut down or reboot. 508 For this driver version, in order to enable WoL, the e1000 driver must be 509 loaded when shutting down or rebooting the system. 510 511 Wake On LAN is only supported on port A for the following devices: 512 Intel(R) PRO/1000 PT Dual Port Network Connection 513 Intel(R) PRO/1000 PT Dual Port Server Connection 514 Intel(R) PRO/1000 PT Dual Port Server Adapter 515 Intel(R) PRO/1000 PF Dual Port Server Adapter 516 Intel(R) PRO/1000 PT Quad Port Server Adapter 517 518 NAPI 519 ---- 520 NAPI (Rx polling mode) is enabled in the e1000 driver. 521 522 See www.cyberus.ca/~hadi/usenix-paper.tgz for more information on NAPI. 523 524 525Known Issues 526============ 527 528Dropped Receive Packets on Half-duplex 10/100 Networks 529------------------------------------------------------ 530If you have an Intel PCI Express adapter running at 10mbps or 100mbps, half- 531duplex, you may observe occasional dropped receive packets. There are no 532workarounds for this problem in this network configuration. The network must 533be updated to operate in full-duplex, and/or 1000mbps only. 534 535Jumbo Frames System Requirement 536------------------------------- 537Memory allocation failures have been observed on Linux systems with 64 MB 538of RAM or less that are running Jumbo Frames. If you are using Jumbo 539Frames, your system may require more than the advertised minimum 540requirement of 64 MB of system memory. 541 542Performance Degradation with Jumbo Frames 543----------------------------------------- 544Degradation in throughput performance may be observed in some Jumbo frames 545environments. If this is observed, increasing the application's socket 546buffer size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values 547may help. See the specific application manual and 548/usr/src/linux*/Documentation/ 549networking/ip-sysctl.txt for more details. 550 551Jumbo Frames on Foundry BigIron 8000 switch 552------------------------------------------- 553There is a known issue using Jumbo frames when connected to a Foundry 554BigIron 8000 switch. This is a 3rd party limitation. If you experience 555loss of packets, lower the MTU size. 556 557Allocating Rx Buffers when Using Jumbo Frames 558--------------------------------------------- 559Allocating Rx buffers when using Jumbo Frames on 2.6.x kernels may fail if 560the available memory is heavily fragmented. This issue may be seen with PCI-X 561adapters or with packet split disabled. This can be reduced or eliminated 562by changing the amount of available memory for receive buffer allocation, by 563increasing /proc/sys/vm/min_free_kbytes. 564 565Multiple Interfaces on Same Ethernet Broadcast Network 566------------------------------------------------------ 567Due to the default ARP behavior on Linux, it is not possible to have 568one system on two IP networks in the same Ethernet broadcast domain 569(non-partitioned switch) behave as expected. All Ethernet interfaces 570will respond to IP traffic for any IP address assigned to the system. 571This results in unbalanced receive traffic. 572 573If you have multiple interfaces in a server, either turn on ARP 574filtering by entering: 575 576 echo 1 > /proc/sys/net/ipv4/conf/all/arp_filter 577(this only works if your kernel's version is higher than 2.4.5), 578 579NOTE: This setting is not saved across reboots. The configuration 580change can be made permanent by adding the line: 581 net.ipv4.conf.all.arp_filter = 1 582to the file /etc/sysctl.conf 583 584 or, 585 586install the interfaces in separate broadcast domains (either in 587different switches or in a switch partitioned to VLANs). 588 58982541/82547 can't link or are slow to link with some link partners 590----------------------------------------------------------------- 591There is a known compatibility issue with 82541/82547 and some 592low-end switches where the link will not be established, or will 593be slow to establish. In particular, these switches are known to 594be incompatible with 82541/82547: 595 596 Planex FXG-08TE 597 I-O Data ETG-SH8 598 599To workaround this issue, the driver can be compiled with an override 600of the PHY's master/slave setting. Forcing master or forcing slave 601mode will improve time-to-link. 602 603 # make CFLAGS_EXTRA=-DE1000_MASTER_SLAVE=<n> 604 605Where <n> is: 606 607 0 = Hardware default 608 1 = Master mode 609 2 = Slave mode 610 3 = Auto master/slave 611 612Disable rx flow control with ethtool 613------------------------------------ 614In order to disable receive flow control using ethtool, you must turn 615off auto-negotiation on the same command line. 616 617For example: 618 619 ethtool -A eth? autoneg off rx off 620 621Unplugging network cable while ethtool -p is running 622---------------------------------------------------- 623In kernel versions 2.5.50 and later (including 2.6 kernel), unplugging 624the network cable while ethtool -p is running will cause the system to 625become unresponsive to keyboard commands, except for control-alt-delete. 626Restarting the system appears to be the only remedy. 627 628 629Support 630======= 631 632For general information, go to the Intel support website at: 633 634 http://support.intel.com 635 636or the Intel Wired Networking project hosted by Sourceforge at: 637 638 http://sourceforge.net/projects/e1000 639 640If an issue is identified with the released source code on the supported 641kernel with a supported adapter, email the specific information related 642to the issue to e1000-devel@lists.sf.net 643