ia64/linux-2.6.18-xen.hg

diff Documentation/networking/e1000.txt @ 0:831230e53067

Import 2.6.18 from kernel.org tarball.
author Ian Campbell <ian.campbell@xensource.com>
date Wed Apr 11 14:15:44 2007 +0100 (2007-04-11)
parents
children
line diff
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/Documentation/networking/e1000.txt	Wed Apr 11 14:15:44 2007 +0100
     1.3 @@ -0,0 +1,571 @@
     1.4 +Linux* Base Driver for the Intel(R) PRO/1000 Family of Adapters
     1.5 +===============================================================
     1.6 +
     1.7 +November 15, 2005
     1.8 +
     1.9 +
    1.10 +Contents
    1.11 +========
    1.12 +
    1.13 +- In This Release
    1.14 +- Identifying Your Adapter
    1.15 +- Command Line Parameters
    1.16 +- Speed and Duplex Configuration
    1.17 +- Additional Configurations
    1.18 +- Known Issues
    1.19 +- Support
    1.20 +
    1.21 +
    1.22 +In This Release
    1.23 +===============
    1.24 +
    1.25 +This file describes the Linux* Base Driver for the Intel(R) PRO/1000 Family
    1.26 +of Adapters.  This driver includes support for Itanium(R)2-based systems.
    1.27 +
    1.28 +For questions related to hardware requirements, refer to the documentation
    1.29 +supplied with your Intel PRO/1000 adapter. All hardware requirements listed
    1.30 +apply to use with Linux.
    1.31 +
    1.32 +The following features are now available in supported kernels:
    1.33 + - Native VLANs
    1.34 + - Channel Bonding (teaming)
    1.35 + - SNMP
    1.36 +
    1.37 +Channel Bonding documentation can be found in the Linux kernel source:
    1.38 +/Documentation/networking/bonding.txt
    1.39 +
    1.40 +The driver information previously displayed in the /proc filesystem is not
    1.41 +supported in this release.  Alternatively, you can use ethtool (version 1.6
    1.42 +or later), lspci, and ifconfig to obtain the same information.
    1.43 +
    1.44 +Instructions on updating ethtool can be found in the section "Additional
    1.45 +Configurations" later in this document.
    1.46 +
    1.47 +
    1.48 +Identifying Your Adapter
    1.49 +========================
    1.50 +
    1.51 +For more information on how to identify your adapter, go to the Adapter &
    1.52 +Driver ID Guide at:
    1.53 +
    1.54 +    http://support.intel.com/support/network/adapter/pro100/21397.htm
    1.55 +
    1.56 +For the latest Intel network drivers for Linux, refer to the following
    1.57 +website. In the search field, enter your adapter name or type, or use the
    1.58 +networking link on the left to search for your adapter:
    1.59 +
    1.60 +    http://downloadfinder.intel.com/scripts-df/support_intel.asp
    1.61 +
    1.62 +
    1.63 +Command Line Parameters =======================
    1.64 +
    1.65 +If the driver is built as a module, the  following optional parameters
    1.66 +are used by entering them on the command line with the modprobe or insmod
    1.67 +command using this syntax:
    1.68 +
    1.69 +     modprobe e1000 [<option>=<VAL1>,<VAL2>,...]
    1.70 +
    1.71 +     insmod e1000 [<option>=<VAL1>,<VAL2>,...]
    1.72 +
    1.73 +For example, with two PRO/1000 PCI adapters, entering:
    1.74 +
    1.75 +     insmod e1000 TxDescriptors=80,128
    1.76 +
    1.77 +loads the e1000 driver with 80 TX descriptors for the first adapter and 128
    1.78 +TX descriptors for the second adapter.
    1.79 +
    1.80 +The default value for each parameter is generally the recommended setting,
    1.81 +unless otherwise noted.
    1.82 +
    1.83 +NOTES:  For more information about the AutoNeg, Duplex, and Speed
    1.84 +        parameters, see the "Speed and Duplex Configuration" section in
    1.85 +        this document.
    1.86 +
    1.87 +        For more information about the InterruptThrottleRate,
    1.88 +        RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay
    1.89 +        parameters, see the application note at:
    1.90 +        http://www.intel.com/design/network/applnots/ap450.htm
    1.91 +
    1.92 +        A descriptor describes a data buffer and attributes related to
    1.93 +        the data buffer. This information is accessed by the hardware.
    1.94 +
    1.95 +
    1.96 +AutoNeg
    1.97 +-------
    1.98 +(Supported only on adapters with copper connections)
    1.99 +Valid Range:   0x01-0x0F, 0x20-0x2F
   1.100 +Default Value: 0x2F
   1.101 +
   1.102 +This parameter is a bit mask that specifies which speed and duplex
   1.103 +settings the board advertises. When this parameter is used, the Speed
   1.104 +and Duplex parameters must not be specified.
   1.105 +
   1.106 +NOTE:  Refer to the Speed and Duplex section of this readme for more
   1.107 +       information on the AutoNeg parameter.
   1.108 +
   1.109 +
   1.110 +Duplex
   1.111 +------
   1.112 +(Supported only on adapters with copper connections)
   1.113 +Valid Range:   0-2 (0=auto-negotiate, 1=half, 2=full)
   1.114 +Default Value: 0
   1.115 +
   1.116 +Defines the direction in which data is allowed to flow. Can be either
   1.117 +one or two-directional. If both Duplex and the link partner are set to
   1.118 +auto-negotiate, the board auto-detects the correct duplex. If the link
   1.119 +partner is forced (either full or half), Duplex defaults to half-duplex.
   1.120 +
   1.121 +
   1.122 +FlowControl
   1.123 +----------
   1.124 +Valid Range:   0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx)
   1.125 +Default Value: Reads flow control settings from the EEPROM
   1.126 +
   1.127 +This parameter controls the automatic generation(Tx) and response(Rx)
   1.128 +to Ethernet PAUSE frames.
   1.129 +
   1.130 +
   1.131 +InterruptThrottleRate
   1.132 +---------------------
   1.133 +(not supported on Intel 82542, 82543 or 82544-based adapters)
   1.134 +Valid Range:   100-100000 (0=off, 1=dynamic)
   1.135 +Default Value: 8000
   1.136 +
   1.137 +This value represents the maximum number of interrupts per second the
   1.138 +controller generates. InterruptThrottleRate is another setting used in
   1.139 +interrupt moderation. Dynamic mode uses a heuristic algorithm to adjust
   1.140 +InterruptThrottleRate based on the current traffic load.
   1.141 +
   1.142 +NOTE:  InterruptThrottleRate takes precedence over the TxAbsIntDelay and
   1.143 +       RxAbsIntDelay parameters. In other words, minimizing the receive
   1.144 +       and/or transmit absolute delays does not force the controller to
   1.145 +       generate more interrupts than what the Interrupt Throttle Rate
   1.146 +       allows.
   1.147 +
   1.148 +CAUTION:  If you are using the Intel PRO/1000 CT Network Connection
   1.149 +          (controller 82547), setting InterruptThrottleRate to a value
   1.150 +          greater than 75,000, may hang (stop transmitting) adapters
   1.151 +          under certain network conditions. If this occurs a NETDEV
   1.152 +          WATCHDOG message is logged in the system event log. In
   1.153 +          addition, the controller is automatically reset, restoring
   1.154 +          the network connection. To eliminate the potential for the
   1.155 +          hang, ensure that InterruptThrottleRate is set no greater
   1.156 +          than 75,000 and is not set to 0.
   1.157 +
   1.158 +NOTE:  When e1000 is loaded with default settings and multiple adapters
   1.159 +       are in use simultaneously, the CPU utilization may increase non-
   1.160 +       linearly. In order to limit the CPU utilization without impacting
   1.161 +       the overall throughput, we recommend that you load the driver as
   1.162 +       follows:
   1.163 +
   1.164 +           insmod e1000.o InterruptThrottleRate=3000,3000,3000
   1.165 +
   1.166 +       This sets the InterruptThrottleRate to 3000 interrupts/sec for
   1.167 +       the first, second, and third instances of the driver. The range
   1.168 +       of 2000 to 3000 interrupts per second works on a majority of
   1.169 +       systems and is a good starting point, but the optimal value will
   1.170 +       be platform-specific. If CPU utilization is not a concern, use
   1.171 +       RX_POLLING (NAPI) and default driver settings.
   1.172 +
   1.173 +
   1.174 +RxDescriptors
   1.175 +-------------
   1.176 +Valid Range:   80-256 for 82542 and 82543-based adapters
   1.177 +               80-4096 for all other supported adapters
   1.178 +Default Value: 256
   1.179 +
   1.180 +This value specifies the number of receive descriptors allocated by the
   1.181 +driver. Increasing this value allows the driver to buffer more incoming
   1.182 +packets.  Each descriptor is 16 bytes.  A receive buffer is also
   1.183 +allocated for each descriptor and is 2048.
   1.184 +
   1.185 +
   1.186 +RxIntDelay
   1.187 +----------
   1.188 +Valid Range:   0-65535 (0=off)
   1.189 +Default Value: 0
   1.190 +
   1.191 +This value delays the generation of receive interrupts in units of 1.024
   1.192 +microseconds.  Receive interrupt reduction can improve CPU efficiency if
   1.193 +properly tuned for specific network traffic. Increasing this value adds
   1.194 +extra latency to frame reception and can end up decreasing the throughput
   1.195 +of TCP traffic. If the system is reporting dropped receives, this value
   1.196 +may be set too high, causing the driver to run out of available receive
   1.197 +descriptors.
   1.198 +
   1.199 +CAUTION:  When setting RxIntDelay to a value other than 0, adapters may
   1.200 +          hang (stop transmitting) under certain network conditions. If
   1.201 +          this occurs a NETDEV WATCHDOG message is logged in the system
   1.202 +          event log. In addition, the controller is automatically reset,
   1.203 +          restoring the network connection. To eliminate the potential
   1.204 +          for the hang ensure that RxIntDelay is set to 0.
   1.205 +
   1.206 +
   1.207 +RxAbsIntDelay
   1.208 +-------------
   1.209 +(This parameter is supported only on 82540, 82545 and later adapters.)
   1.210 +Valid Range:   0-65535 (0=off)
   1.211 +Default Value: 128
   1.212 +
   1.213 +This value, in units of 1.024 microseconds, limits the delay in which a
   1.214 +receive interrupt is generated. Useful only if RxIntDelay is non-zero,
   1.215 +this value ensures that an interrupt is generated after the initial
   1.216 +packet is received within the set amount of time.  Proper tuning,
   1.217 +along with RxIntDelay, may improve traffic throughput in specific network
   1.218 +conditions.
   1.219 +
   1.220 +
   1.221 +Speed
   1.222 +-----
   1.223 +(This parameter is supported only on adapters with copper connections.)
   1.224 +Valid Settings: 0, 10, 100, 1000
   1.225 +Default Value:  0 (auto-negotiate at all supported speeds)
   1.226 +
   1.227 +Speed forces the line speed to the specified value in megabits per second
   1.228 +(Mbps). If this parameter is not specified or is set to 0 and the link
   1.229 +partner is set to auto-negotiate, the board will auto-detect the correct
   1.230 +speed. Duplex should also be set when Speed is set to either 10 or 100.
   1.231 +
   1.232 +
   1.233 +TxDescriptors
   1.234 +-------------
   1.235 +Valid Range:   80-256 for 82542 and 82543-based adapters
   1.236 +               80-4096 for all other supported adapters
   1.237 +Default Value: 256
   1.238 +
   1.239 +This value is the number of transmit descriptors allocated by the driver.
   1.240 +Increasing this value allows the driver to queue more transmits. Each
   1.241 +descriptor is 16 bytes.
   1.242 +
   1.243 +NOTE:  Depending on the available system resources, the request for a
   1.244 +       higher number of transmit descriptors may be denied.  In this case,
   1.245 +       use a lower number.
   1.246 +
   1.247 +
   1.248 +TxIntDelay
   1.249 +----------
   1.250 +Valid Range:   0-65535 (0=off)
   1.251 +Default Value: 64
   1.252 +
   1.253 +This value delays the generation of transmit interrupts in units of
   1.254 +1.024 microseconds. Transmit interrupt reduction can improve CPU
   1.255 +efficiency if properly tuned for specific network traffic. If the
   1.256 +system is reporting dropped transmits, this value may be set too high
   1.257 +causing the driver to run out of available transmit descriptors.
   1.258 +
   1.259 +
   1.260 +TxAbsIntDelay
   1.261 +-------------
   1.262 +(This parameter is supported only on 82540, 82545 and later adapters.)
   1.263 +Valid Range:   0-65535 (0=off)
   1.264 +Default Value: 64
   1.265 +
   1.266 +This value, in units of 1.024 microseconds, limits the delay in which a
   1.267 +transmit interrupt is generated. Useful only if TxIntDelay is non-zero,
   1.268 +this value ensures that an interrupt is generated after the initial
   1.269 +packet is sent on the wire within the set amount of time.  Proper tuning,
   1.270 +along with TxIntDelay, may improve traffic throughput in specific
   1.271 +network conditions.
   1.272 +
   1.273 +XsumRX
   1.274 +------
   1.275 +(This parameter is NOT supported on the 82542-based adapter.)
   1.276 +Valid Range:   0-1
   1.277 +Default Value: 1
   1.278 +
   1.279 +A value of '1' indicates that the driver should enable IP checksum
   1.280 +offload for received packets (both UDP and TCP) to the adapter hardware.
   1.281 +
   1.282 +
   1.283 +Speed and Duplex Configuration
   1.284 +==============================
   1.285 +
   1.286 +Three keywords are used to control the speed and duplex configuration.
   1.287 +These keywords are Speed, Duplex, and AutoNeg.
   1.288 +
   1.289 +If the board uses a fiber interface, these keywords are ignored, and the
   1.290 +fiber interface board only links at 1000 Mbps full-duplex.
   1.291 +
   1.292 +For copper-based boards, the keywords interact as follows:
   1.293 +
   1.294 +  The default operation is auto-negotiate. The board advertises all
   1.295 +  supported speed and duplex combinations, and it links at the highest
   1.296 +  common speed and duplex mode IF the link partner is set to auto-negotiate.
   1.297 +
   1.298 +  If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps
   1.299 +  is advertised (The 1000BaseT spec requires auto-negotiation.)
   1.300 +
   1.301 +  If Speed = 10 or 100, then both Speed and Duplex should be set. Auto-
   1.302 +  negotiation is disabled, and the AutoNeg parameter is ignored. Partner
   1.303 +  SHOULD also be forced.
   1.304 +
   1.305 +The AutoNeg parameter is used when more control is required over the
   1.306 +auto-negotiation process.  It should be used when you wish to control which
   1.307 +speed and duplex combinations are advertised during the auto-negotiation
   1.308 +process.
   1.309 +
   1.310 +The parameter may be specified as either a decimal or hexidecimal value as
   1.311 +determined by the bitmap below.
   1.312 +
   1.313 +Bit position   7      6      5       4       3      2      1       0
   1.314 +Decimal Value  128    64     32      16      8      4      2       1
   1.315 +Hex value      80     40     20      10      8      4      2       1
   1.316 +Speed (Mbps)   N/A    N/A    1000    N/A     100    100    10      10
   1.317 +Duplex                       Full            Full   Half   Full    Half
   1.318 +
   1.319 +Some examples of using AutoNeg:
   1.320 +
   1.321 +  modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half)
   1.322 +  modprobe e1000 AutoNeg=1 (Same as above)
   1.323 +  modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full)
   1.324 +  modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full)
   1.325 +  modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half)
   1.326 +  modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100
   1.327 +  Half)
   1.328 +  modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full)
   1.329 +  modprobe e1000 AutoNeg=32 (Same as above)
   1.330 +
   1.331 +Note that when this parameter is used, Speed and Duplex must not be specified.
   1.332 +
   1.333 +If the link partner is forced to a specific speed and duplex, then this
   1.334 +parameter should not be used.  Instead, use the Speed and Duplex parameters
   1.335 +previously mentioned to force the adapter to the same speed and duplex.
   1.336 +
   1.337 +
   1.338 +Additional Configurations
   1.339 +=========================
   1.340 +
   1.341 +  Configuring the Driver on Different Distributions
   1.342 +  -------------------------------------------------
   1.343 +
   1.344 +  Configuring a network driver to load properly when the system is started
   1.345 +  is distribution dependent. Typically, the configuration process involves
   1.346 +  adding an alias line to /etc/modules.conf or /etc/modprobe.conf as well
   1.347 +  as editing other system startup scripts and/or configuration files. Many
   1.348 +  popular Linux distributions ship with tools to make these changes for you.
   1.349 +  To learn the proper way to configure a network device for your system,
   1.350 +  refer to your distribution documentation. If during this process you are
   1.351 +  asked for the driver or module name, the name for the Linux Base Driver
   1.352 +  for the Intel PRO/1000 Family of Adapters is e1000.
   1.353 +
   1.354 +  As an example, if you install the e1000 driver for two PRO/1000 adapters
   1.355 +  (eth0 and eth1) and set the speed and duplex to 10full and 100half, add
   1.356 +  the following to modules.conf or or modprobe.conf:
   1.357 +
   1.358 +       alias eth0 e1000
   1.359 +       alias eth1 e1000
   1.360 +       options e1000 Speed=10,100 Duplex=2,1
   1.361 +
   1.362 +  Viewing Link Messages
   1.363 +  ---------------------
   1.364 +
   1.365 +  Link messages will not be displayed to the console if the distribution is
   1.366 +  restricting system messages. In order to see network driver link messages
   1.367 +  on your console, set dmesg to eight by entering the following:
   1.368 +
   1.369 +       dmesg -n 8
   1.370 +
   1.371 +  NOTE: This setting is not saved across reboots.
   1.372 +
   1.373 +  Jumbo Frames
   1.374 +  ------------
   1.375 +
   1.376 +  The driver supports Jumbo Frames for all adapters except 82542 and
   1.377 +  82573-based adapters. Jumbo Frames support is enabled by changing the
   1.378 +  MTU to a value larger than the default of 1500. Use the ifconfig command
   1.379 +  to increase the MTU size. For example:
   1.380 +
   1.381 +       ifconfig eth<x> mtu 9000 up
   1.382 +
   1.383 +  This setting is not saved across reboots.  It can be made permanent if
   1.384 +  you add:
   1.385 +
   1.386 +       MTU=9000
   1.387 +
   1.388 +   to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>.  This example
   1.389 +   applies to the Red Hat distributions; other distributions may store this
   1.390 +   setting in a different location.
   1.391 +
   1.392 +  Notes:
   1.393 +
   1.394 +  - To enable Jumbo Frames, increase the MTU size on the interface beyond
   1.395 +    1500.
   1.396 +  - The maximum MTU setting for Jumbo Frames is 16110. This value coincides
   1.397 +    with the maximum Jumbo Frames size of 16128.
   1.398 +  - Using Jumbo Frames at 10 or 100 Mbps may result in poor performance or
   1.399 +    loss of link.
   1.400 +  - Some Intel gigabit adapters that support Jumbo Frames have a frame size
   1.401 +    limit of 9238 bytes, with a corresponding MTU size limit of 9216 bytes.
   1.402 +    The adapters with this limitation are based on the Intel 82571EB and
   1.403 +    82572EI controllers, which correspond to these product names:
   1.404 +     Intel® PRO/1000 PT Dual Port Server Adapter
   1.405 +     Intel® PRO/1000 PF Dual Port Server Adapter
   1.406 +     Intel® PRO/1000 PT Server Adapter
   1.407 +     Intel® PRO/1000 PT Desktop Adapter
   1.408 +     Intel® PRO/1000 PF Server Adapter
   1.409 +
   1.410 +  - The Intel PRO/1000 PM Network Connection does not support jumbo frames.
   1.411 +
   1.412 +
   1.413 +  Ethtool
   1.414 +  -------
   1.415 +
   1.416 +  The driver utilizes the ethtool interface for driver configuration and
   1.417 +  diagnostics, as well as displaying statistical information.  Ethtool
   1.418 +  version 1.6 or later is required for this functionality.
   1.419 +
   1.420 +  The latest release of ethtool can be found from
   1.421 +  http://sourceforge.net/projects/gkernel.
   1.422 +
   1.423 +  NOTE: Ethtool 1.6 only supports a limited set of ethtool options. Support
   1.424 +  for a more complete ethtool feature set can be enabled by upgrading
   1.425 +  ethtool to ethtool-1.8.1.
   1.426 +
   1.427 +  Enabling Wake on LAN* (WoL)
   1.428 +  ---------------------------
   1.429 +
   1.430 +  WoL is configured through the Ethtool* utility. Ethtool is included with
   1.431 +  all versions of Red Hat after Red Hat 7.2. For other Linux distributions,
   1.432 +  download and install Ethtool from the following website:
   1.433 +  http://sourceforge.net/projects/gkernel.
   1.434 +
   1.435 +  For instructions on enabling WoL with Ethtool, refer to the website listed
   1.436 +  above.
   1.437 +
   1.438 +  WoL will be enabled on the system during the next shut down or reboot.
   1.439 +  For this driver version, in order to enable WoL, the e1000 driver must be
   1.440 +  loaded when shutting down or rebooting the system.
   1.441 +
   1.442 +  NAPI
   1.443 +  ----
   1.444 +
   1.445 +  NAPI (Rx polling mode) is supported in the e1000 driver. NAPI is enabled
   1.446 +  or disabled based on the configuration of the kernel. To override
   1.447 +  the default, use the following compile-time flags.
   1.448 +
   1.449 +  To enable NAPI, compile the driver module, passing in a configuration option:
   1.450 +
   1.451 +       make CFLAGS_EXTRA=-DE1000_NAPI install
   1.452 +
   1.453 +  To disable NAPI, compile the driver module, passing in a configuration option:
   1.454 +
   1.455 +       make CFLAGS_EXTRA=-DE1000_NO_NAPI install
   1.456 +
   1.457 +  See www.cyberus.ca/~hadi/usenix-paper.tgz for more information on NAPI.
   1.458 +
   1.459 +
   1.460 +Known Issues
   1.461 +============
   1.462 +
   1.463 +  Jumbo Frames System Requirement
   1.464 +  -------------------------------
   1.465 +
   1.466 +  Memory allocation failures have been observed on Linux systems with 64 MB
   1.467 +  of RAM or less that are running Jumbo Frames. If you are using Jumbo
   1.468 +  Frames, your system may require more than the advertised minimum
   1.469 +  requirement of 64 MB of system memory.
   1.470 +
   1.471 +  Performance Degradation with Jumbo Frames
   1.472 +  -----------------------------------------
   1.473 +
   1.474 +  Degradation in throughput performance may be observed in some Jumbo frames
   1.475 +  environments. If this is observed, increasing the application's socket
   1.476 +  buffer size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values
   1.477 +  may help. See the specific application manual and
   1.478 +  /usr/src/linux*/Documentation/
   1.479 +  networking/ip-sysctl.txt for more details.
   1.480 +
   1.481 +  Jumbo frames on Foundry BigIron 8000 switch
   1.482 +  -------------------------------------------
   1.483 +  There is a known issue using Jumbo frames when connected to a Foundry
   1.484 +  BigIron 8000 switch. This is a 3rd party limitation. If you experience
   1.485 +  loss of packets, lower the MTU size.
   1.486 +
   1.487 +  Multiple Interfaces on Same Ethernet Broadcast Network
   1.488 +  ------------------------------------------------------
   1.489 +
   1.490 +  Due to the default ARP behavior on Linux, it is not possible to have
   1.491 +  one system on two IP networks in the same Ethernet broadcast domain
   1.492 +  (non-partitioned switch) behave as expected. All Ethernet interfaces
   1.493 +  will respond to IP traffic for any IP address assigned to the system.
   1.494 +  This results in unbalanced receive traffic.
   1.495 +
   1.496 +  If you have multiple interfaces in a server, either turn on ARP
   1.497 +  filtering by entering:
   1.498 +
   1.499 +      echo 1 > /proc/sys/net/ipv4/conf/all/arp_filter
   1.500 +  (this only works if your kernel's version is higher than 2.4.5),
   1.501 +
   1.502 +  NOTE: This setting is not saved across reboots. The configuration
   1.503 +  change can be made permanent by adding the line:
   1.504 +      net.ipv4.conf.all.arp_filter = 1
   1.505 +  to the file /etc/sysctl.conf
   1.506 +
   1.507 +        or,
   1.508 +
   1.509 +  install the interfaces in separate broadcast domains (either in
   1.510 +  different switches or in a switch partitioned to VLANs).
   1.511 +
   1.512 +  82541/82547 can't link or are slow to link with some link partners
   1.513 +  -----------------------------------------------------------------
   1.514 +
   1.515 +  There is a known compatibility issue with 82541/82547 and some
   1.516 +  low-end switches where the link will not be established, or will
   1.517 +  be slow to establish.  In particular, these switches are known to
   1.518 +  be incompatible with 82541/82547:
   1.519 +
   1.520 +      Planex FXG-08TE
   1.521 +      I-O Data ETG-SH8
   1.522 +
   1.523 +  To workaround this issue, the driver can be compiled with an override
   1.524 +  of the PHY's master/slave setting.  Forcing master or forcing slave
   1.525 +  mode will improve time-to-link.
   1.526 +
   1.527 +      # make EXTRA_CFLAGS=-DE1000_MASTER_SLAVE=<n>
   1.528 +
   1.529 +  Where <n> is:
   1.530 +
   1.531 +      0 = Hardware default
   1.532 +      1 = Master mode
   1.533 +      2 = Slave mode
   1.534 +      3 = Auto master/slave
   1.535 +
   1.536 +  Disable rx flow control with ethtool
   1.537 +  ------------------------------------
   1.538 +
   1.539 +  In order to disable receive flow control using ethtool, you must turn
   1.540 +  off auto-negotiation on the same command line.
   1.541 +
   1.542 +  For example:
   1.543 +
   1.544 +     ethtool -A eth? autoneg off rx off
   1.545 +
   1.546 +
   1.547 +Support
   1.548 +=======
   1.549 +
   1.550 +For general information, go to the Intel support website at:
   1.551 +
   1.552 +    http://support.intel.com
   1.553 +
   1.554 +    or the Intel Wired Networking project hosted by Sourceforge at:
   1.555 +
   1.556 +    http://sourceforge.net/projects/e1000
   1.557 +
   1.558 +If an issue is identified with the released source code on the supported
   1.559 +kernel with a supported adapter, email the specific information related
   1.560 +to the issue to e1000-devel@lists.sourceforge.net
   1.561 +
   1.562 +
   1.563 +License
   1.564 +=======
   1.565 +
   1.566 +This software program is released under the terms of a license agreement
   1.567 +between you ('Licensee') and Intel. Do not use or load this software or any
   1.568 +associated materials (collectively, the 'Software') until you have carefully
   1.569 +read the full terms and conditions of the file COPYING located in this software
   1.570 +package. By loading or using the Software, you agree to the terms of this
   1.571 +Agreement. If you do not agree with the terms of this Agreement, do not
   1.572 +install or use the Software.
   1.573 +
   1.574 +* Other names and brands may be claimed as the property of others.