e1000.txt 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652
  1. Linux* Base Driver for the Intel(R) PRO/1000 Family of Adapters
  2. ===============================================================
  3. September 26, 2006
  4. Contents
  5. ========
  6. - In This Release
  7. - Identifying Your Adapter
  8. - Building and Installation
  9. - Command Line Parameters
  10. - Speed and Duplex Configuration
  11. - Additional Configurations
  12. - Known Issues
  13. - Support
  14. In This Release
  15. ===============
  16. This file describes the Linux* Base Driver for the Intel(R) PRO/1000 Family
  17. of Adapters. This driver includes support for Itanium(R)2-based systems.
  18. For questions related to hardware requirements, refer to the documentation
  19. supplied with your Intel PRO/1000 adapter. All hardware requirements listed
  20. apply to use with Linux.
  21. The following features are now available in supported kernels:
  22. - Native VLANs
  23. - Channel Bonding (teaming)
  24. - SNMP
  25. Channel Bonding documentation can be found in the Linux kernel source:
  26. /Documentation/networking/bonding.txt
  27. The driver information previously displayed in the /proc filesystem is not
  28. supported in this release. Alternatively, you can use ethtool (version 1.6
  29. or later), lspci, and ifconfig to obtain the same information.
  30. Instructions on updating ethtool can be found in the section "Additional
  31. Configurations" later in this document.
  32. NOTE: The Intel(R) 82562v 10/100 Network Connection only provides 10/100
  33. support.
  34. Identifying Your Adapter
  35. ========================
  36. For more information on how to identify your adapter, go to the Adapter &
  37. Driver ID Guide at:
  38. http://support.intel.com/support/network/adapter/pro100/21397.htm
  39. For the latest Intel network drivers for Linux, refer to the following
  40. website. In the search field, enter your adapter name or type, or use the
  41. networking link on the left to search for your adapter:
  42. http://downloadfinder.intel.com/scripts-df/support_intel.asp
  43. Command Line Parameters
  44. =======================
  45. If the driver is built as a module, the following optional parameters
  46. are used by entering them on the command line with the modprobe command
  47. using this syntax:
  48. modprobe e1000 [<option>=<VAL1>,<VAL2>,...]
  49. For example, with two PRO/1000 PCI adapters, entering:
  50. modprobe e1000 TxDescriptors=80,128
  51. loads the e1000 driver with 80 TX descriptors for the first adapter and
  52. 128 TX descriptors for the second adapter.
  53. The default value for each parameter is generally the recommended setting,
  54. unless otherwise noted.
  55. NOTES: For more information about the AutoNeg, Duplex, and Speed
  56. parameters, see the "Speed and Duplex Configuration" section in
  57. this document.
  58. For more information about the InterruptThrottleRate,
  59. RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay
  60. parameters, see the application note at:
  61. http://www.intel.com/design/network/applnots/ap450.htm
  62. A descriptor describes a data buffer and attributes related to
  63. the data buffer. This information is accessed by the hardware.
  64. AutoNeg
  65. -------
  66. (Supported only on adapters with copper connections)
  67. Valid Range: 0x01-0x0F, 0x20-0x2F
  68. Default Value: 0x2F
  69. This parameter is a bit-mask that specifies the speed and duplex settings
  70. advertised by the adapter. When this parameter is used, the Speed and
  71. Duplex parameters must not be specified.
  72. NOTE: Refer to the Speed and Duplex section of this readme for more
  73. information on the AutoNeg parameter.
  74. Duplex
  75. ------
  76. (Supported only on adapters with copper connections)
  77. Valid Range: 0-2 (0=auto-negotiate, 1=half, 2=full)
  78. Default Value: 0
  79. This defines the direction in which data is allowed to flow. Can be
  80. either one or two-directional. If both Duplex and the link partner are
  81. set to auto-negotiate, the board auto-detects the correct duplex. If the
  82. link partner is forced (either full or half), Duplex defaults to half-
  83. duplex.
  84. FlowControl
  85. -----------
  86. Valid Range: 0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx)
  87. Default Value: Reads flow control settings from the EEPROM
  88. This parameter controls the automatic generation(Tx) and response(Rx)
  89. to Ethernet PAUSE frames.
  90. InterruptThrottleRate
  91. ---------------------
  92. (not supported on Intel(R) 82542, 82543 or 82544-based adapters)
  93. Valid Range: 0,1,3,100-100000 (0=off, 1=dynamic, 3=dynamic conservative)
  94. Default Value: 3
  95. The driver can limit the amount of interrupts per second that the adapter
  96. will generate for incoming packets. It does this by writing a value to the
  97. adapter that is based on the maximum amount of interrupts that the adapter
  98. will generate per second.
  99. Setting InterruptThrottleRate to a value greater or equal to 100
  100. will program the adapter to send out a maximum of that many interrupts
  101. per second, even if more packets have come in. This reduces interrupt
  102. load on the system and can lower CPU utilization under heavy load,
  103. but will increase latency as packets are not processed as quickly.
  104. The default behaviour of the driver previously assumed a static
  105. InterruptThrottleRate value of 8000, providing a good fallback value for
  106. all traffic types,but lacking in small packet performance and latency.
  107. The hardware can handle many more small packets per second however, and
  108. for this reason an adaptive interrupt moderation algorithm was implemented.
  109. Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in which
  110. it dynamically adjusts the InterruptThrottleRate value based on the traffic
  111. that it receives. After determining the type of incoming traffic in the last
  112. timeframe, it will adjust the InterruptThrottleRate to an appropriate value
  113. for that traffic.
  114. The algorithm classifies the incoming traffic every interval into
  115. classes. Once the class is determined, the InterruptThrottleRate value is
  116. adjusted to suit that traffic type the best. There are three classes defined:
  117. "Bulk traffic", for large amounts of packets of normal size; "Low latency",
  118. for small amounts of traffic and/or a significant percentage of small
  119. packets; and "Lowest latency", for almost completely small packets or
  120. minimal traffic.
  121. In dynamic conservative mode, the InterruptThrottleRate value is set to 4000
  122. for traffic that falls in class "Bulk traffic". If traffic falls in the "Low
  123. latency" or "Lowest latency" class, the InterruptThrottleRate is increased
  124. stepwise to 20000. This default mode is suitable for most applications.
  125. For situations where low latency is vital such as cluster or
  126. grid computing, the algorithm can reduce latency even more when
  127. InterruptThrottleRate is set to mode 1. In this mode, which operates
  128. the same as mode 3, the InterruptThrottleRate will be increased stepwise to
  129. 70000 for traffic in class "Lowest latency".
  130. Setting InterruptThrottleRate to 0 turns off any interrupt moderation
  131. and may improve small packet latency, but is generally not suitable
  132. for bulk throughput traffic.
  133. NOTE: InterruptThrottleRate takes precedence over the TxAbsIntDelay and
  134. RxAbsIntDelay parameters. In other words, minimizing the receive
  135. and/or transmit absolute delays does not force the controller to
  136. generate more interrupts than what the Interrupt Throttle Rate
  137. allows.
  138. CAUTION: If you are using the Intel(R) PRO/1000 CT Network Connection
  139. (controller 82547), setting InterruptThrottleRate to a value
  140. greater than 75,000, may hang (stop transmitting) adapters
  141. under certain network conditions. If this occurs a NETDEV
  142. WATCHDOG message is logged in the system event log. In
  143. addition, the controller is automatically reset, restoring
  144. the network connection. To eliminate the potential for the
  145. hang, ensure that InterruptThrottleRate is set no greater
  146. than 75,000 and is not set to 0.
  147. NOTE: When e1000 is loaded with default settings and multiple adapters
  148. are in use simultaneously, the CPU utilization may increase non-
  149. linearly. In order to limit the CPU utilization without impacting
  150. the overall throughput, we recommend that you load the driver as
  151. follows:
  152. modprobe e1000 InterruptThrottleRate=3000,3000,3000
  153. This sets the InterruptThrottleRate to 3000 interrupts/sec for
  154. the first, second, and third instances of the driver. The range
  155. of 2000 to 3000 interrupts per second works on a majority of
  156. systems and is a good starting point, but the optimal value will
  157. be platform-specific. If CPU utilization is not a concern, use
  158. RX_POLLING (NAPI) and default driver settings.
  159. RxDescriptors
  160. -------------
  161. Valid Range: 80-256 for 82542 and 82543-based adapters
  162. 80-4096 for all other supported adapters
  163. Default Value: 256
  164. This value specifies the number of receive buffer descriptors allocated
  165. by the driver. Increasing this value allows the driver to buffer more
  166. incoming packets, at the expense of increased system memory utilization.
  167. Each descriptor is 16 bytes. A receive buffer is also allocated for each
  168. descriptor and can be either 2048, 4096, 8192, or 16384 bytes, depending
  169. on the MTU setting. The maximum MTU size is 16110.
  170. NOTE: MTU designates the frame size. It only needs to be set for Jumbo
  171. Frames. Depending on the available system resources, the request
  172. for a higher number of receive descriptors may be denied. In this
  173. case, use a lower number.
  174. RxIntDelay
  175. ----------
  176. Valid Range: 0-65535 (0=off)
  177. Default Value: 0
  178. This value delays the generation of receive interrupts in units of 1.024
  179. microseconds. Receive interrupt reduction can improve CPU efficiency if
  180. properly tuned for specific network traffic. Increasing this value adds
  181. extra latency to frame reception and can end up decreasing the throughput
  182. of TCP traffic. If the system is reporting dropped receives, this value
  183. may be set too high, causing the driver to run out of available receive
  184. descriptors.
  185. CAUTION: When setting RxIntDelay to a value other than 0, adapters may
  186. hang (stop transmitting) under certain network conditions. If
  187. this occurs a NETDEV WATCHDOG message is logged in the system
  188. event log. In addition, the controller is automatically reset,
  189. restoring the network connection. To eliminate the potential
  190. for the hang ensure that RxIntDelay is set to 0.
  191. RxAbsIntDelay
  192. -------------
  193. (This parameter is supported only on 82540, 82545 and later adapters.)
  194. Valid Range: 0-65535 (0=off)
  195. Default Value: 128
  196. This value, in units of 1.024 microseconds, limits the delay in which a
  197. receive interrupt is generated. Useful only if RxIntDelay is non-zero,
  198. this value ensures that an interrupt is generated after the initial
  199. packet is received within the set amount of time. Proper tuning,
  200. along with RxIntDelay, may improve traffic throughput in specific network
  201. conditions.
  202. Speed
  203. -----
  204. (This parameter is supported only on adapters with copper connections.)
  205. Valid Settings: 0, 10, 100, 1000
  206. Default Value: 0 (auto-negotiate at all supported speeds)
  207. Speed forces the line speed to the specified value in megabits per second
  208. (Mbps). If this parameter is not specified or is set to 0 and the link
  209. partner is set to auto-negotiate, the board will auto-detect the correct
  210. speed. Duplex should also be set when Speed is set to either 10 or 100.
  211. TxDescriptors
  212. -------------
  213. Valid Range: 80-256 for 82542 and 82543-based adapters
  214. 80-4096 for all other supported adapters
  215. Default Value: 256
  216. This value is the number of transmit descriptors allocated by the driver.
  217. Increasing this value allows the driver to queue more transmits. Each
  218. descriptor is 16 bytes.
  219. NOTE: Depending on the available system resources, the request for a
  220. higher number of transmit descriptors may be denied. In this case,
  221. use a lower number.
  222. TxIntDelay
  223. ----------
  224. Valid Range: 0-65535 (0=off)
  225. Default Value: 64
  226. This value delays the generation of transmit interrupts in units of
  227. 1.024 microseconds. Transmit interrupt reduction can improve CPU
  228. efficiency if properly tuned for specific network traffic. If the
  229. system is reporting dropped transmits, this value may be set too high
  230. causing the driver to run out of available transmit descriptors.
  231. TxAbsIntDelay
  232. -------------
  233. (This parameter is supported only on 82540, 82545 and later adapters.)
  234. Valid Range: 0-65535 (0=off)
  235. Default Value: 64
  236. This value, in units of 1.024 microseconds, limits the delay in which a
  237. transmit interrupt is generated. Useful only if TxIntDelay is non-zero,
  238. this value ensures that an interrupt is generated after the initial
  239. packet is sent on the wire within the set amount of time. Proper tuning,
  240. along with TxIntDelay, may improve traffic throughput in specific
  241. network conditions.
  242. XsumRX
  243. ------
  244. (This parameter is NOT supported on the 82542-based adapter.)
  245. Valid Range: 0-1
  246. Default Value: 1
  247. A value of '1' indicates that the driver should enable IP checksum
  248. offload for received packets (both UDP and TCP) to the adapter hardware.
  249. Speed and Duplex Configuration
  250. ==============================
  251. Three keywords are used to control the speed and duplex configuration.
  252. These keywords are Speed, Duplex, and AutoNeg.
  253. If the board uses a fiber interface, these keywords are ignored, and the
  254. fiber interface board only links at 1000 Mbps full-duplex.
  255. For copper-based boards, the keywords interact as follows:
  256. The default operation is auto-negotiate. The board advertises all
  257. supported speed and duplex combinations, and it links at the highest
  258. common speed and duplex mode IF the link partner is set to auto-negotiate.
  259. If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps
  260. is advertised (The 1000BaseT spec requires auto-negotiation.)
  261. If Speed = 10 or 100, then both Speed and Duplex should be set. Auto-
  262. negotiation is disabled, and the AutoNeg parameter is ignored. Partner
  263. SHOULD also be forced.
  264. The AutoNeg parameter is used when more control is required over the
  265. auto-negotiation process. It should be used when you wish to control which
  266. speed and duplex combinations are advertised during the auto-negotiation
  267. process.
  268. The parameter may be specified as either a decimal or hexadecimal value as
  269. determined by the bitmap below.
  270. Bit position 7 6 5 4 3 2 1 0
  271. Decimal Value 128 64 32 16 8 4 2 1
  272. Hex value 80 40 20 10 8 4 2 1
  273. Speed (Mbps) N/A N/A 1000 N/A 100 100 10 10
  274. Duplex Full Full Half Full Half
  275. Some examples of using AutoNeg:
  276. modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half)
  277. modprobe e1000 AutoNeg=1 (Same as above)
  278. modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full)
  279. modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full)
  280. modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half)
  281. modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100
  282. Half)
  283. modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full)
  284. modprobe e1000 AutoNeg=32 (Same as above)
  285. Note that when this parameter is used, Speed and Duplex must not be specified.
  286. If the link partner is forced to a specific speed and duplex, then this
  287. parameter should not be used. Instead, use the Speed and Duplex parameters
  288. previously mentioned to force the adapter to the same speed and duplex.
  289. Additional Configurations
  290. =========================
  291. Configuring the Driver on Different Distributions
  292. -------------------------------------------------
  293. Configuring a network driver to load properly when the system is started
  294. is distribution dependent. Typically, the configuration process involves
  295. adding an alias line to /etc/modules.conf or /etc/modprobe.conf as well
  296. as editing other system startup scripts and/or configuration files. Many
  297. popular Linux distributions ship with tools to make these changes for you.
  298. To learn the proper way to configure a network device for your system,
  299. refer to your distribution documentation. If during this process you are
  300. asked for the driver or module name, the name for the Linux Base Driver
  301. for the Intel(R) PRO/1000 Family of Adapters is e1000.
  302. As an example, if you install the e1000 driver for two PRO/1000 adapters
  303. (eth0 and eth1) and set the speed and duplex to 10full and 100half, add
  304. the following to modules.conf or or modprobe.conf:
  305. alias eth0 e1000
  306. alias eth1 e1000
  307. options e1000 Speed=10,100 Duplex=2,1
  308. Viewing Link Messages
  309. ---------------------
  310. Link messages will not be displayed to the console if the distribution is
  311. restricting system messages. In order to see network driver link messages
  312. on your console, set dmesg to eight by entering the following:
  313. dmesg -n 8
  314. NOTE: This setting is not saved across reboots.
  315. Jumbo Frames
  316. ------------
  317. Jumbo Frames support is enabled by changing the MTU to a value larger than
  318. the default of 1500. Use the ifconfig command to increase the MTU size.
  319. For example:
  320. ifconfig eth<x> mtu 9000 up
  321. This setting is not saved across reboots. It can be made permanent if
  322. you add:
  323. MTU=9000
  324. to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>. This example
  325. applies to the Red Hat distributions; other distributions may store this
  326. setting in a different location.
  327. Notes:
  328. - To enable Jumbo Frames, increase the MTU size on the interface beyond
  329. 1500.
  330. - The maximum MTU setting for Jumbo Frames is 16110. This value coincides
  331. with the maximum Jumbo Frames size of 16128.
  332. - Using Jumbo Frames at 10 or 100 Mbps may result in poor performance or
  333. loss of link.
  334. - Some Intel gigabit adapters that support Jumbo Frames have a frame size
  335. limit of 9238 bytes, with a corresponding MTU size limit of 9216 bytes.
  336. The adapters with this limitation are based on the Intel(R) 82571EB,
  337. 82572EI, 82573L and 80003ES2LAN controller. These correspond to the
  338. following product names:
  339. Intel(R) PRO/1000 PT Server Adapter
  340. Intel(R) PRO/1000 PT Desktop Adapter
  341. Intel(R) PRO/1000 PT Network Connection
  342. Intel(R) PRO/1000 PT Dual Port Server Adapter
  343. Intel(R) PRO/1000 PT Dual Port Network Connection
  344. Intel(R) PRO/1000 PF Server Adapter
  345. Intel(R) PRO/1000 PF Network Connection
  346. Intel(R) PRO/1000 PF Dual Port Server Adapter
  347. Intel(R) PRO/1000 PB Server Connection
  348. Intel(R) PRO/1000 PL Network Connection
  349. Intel(R) PRO/1000 EB Network Connection with I/O Acceleration
  350. Intel(R) PRO/1000 EB Backplane Connection with I/O Acceleration
  351. Intel(R) PRO/1000 PT Quad Port Server Adapter
  352. - Adapters based on the Intel(R) 82542 and 82573V/E controller do not
  353. support Jumbo Frames. These correspond to the following product names:
  354. Intel(R) PRO/1000 Gigabit Server Adapter
  355. Intel(R) PRO/1000 PM Network Connection
  356. - The following adapters do not support Jumbo Frames:
  357. Intel(R) 82562V 10/100 Network Connection
  358. Intel(R) 82566DM Gigabit Network Connection
  359. Intel(R) 82566DC Gigabit Network Connection
  360. Intel(R) 82566MM Gigabit Network Connection
  361. Intel(R) 82566MC Gigabit Network Connection
  362. Intel(R) 82562GT 10/100 Network Connection
  363. Intel(R) 82562G 10/100 Network Connection
  364. Ethtool
  365. -------
  366. The driver utilizes the ethtool interface for driver configuration and
  367. diagnostics, as well as displaying statistical information. Ethtool
  368. version 1.6 or later is required for this functionality.
  369. The latest release of ethtool can be found from
  370. http://sourceforge.net/projects/gkernel.
  371. NOTE: Ethtool 1.6 only supports a limited set of ethtool options. Support
  372. for a more complete ethtool feature set can be enabled by upgrading
  373. ethtool to ethtool-1.8.1.
  374. Enabling Wake on LAN* (WoL)
  375. ---------------------------
  376. WoL is configured through the Ethtool* utility. Ethtool is included with
  377. all versions of Red Hat after Red Hat 7.2. For other Linux distributions,
  378. download and install Ethtool from the following website:
  379. http://sourceforge.net/projects/gkernel.
  380. For instructions on enabling WoL with Ethtool, refer to the website listed
  381. above.
  382. WoL will be enabled on the system during the next shut down or reboot.
  383. For this driver version, in order to enable WoL, the e1000 driver must be
  384. loaded when shutting down or rebooting the system.
  385. Wake On LAN is only supported on port A for the following devices:
  386. Intel(R) PRO/1000 PT Dual Port Network Connection
  387. Intel(R) PRO/1000 PT Dual Port Server Connection
  388. Intel(R) PRO/1000 PT Dual Port Server Adapter
  389. Intel(R) PRO/1000 PF Dual Port Server Adapter
  390. Intel(R) PRO/1000 PT Quad Port Server Adapter
  391. NAPI
  392. ----
  393. NAPI (Rx polling mode) is supported in the e1000 driver. NAPI is enabled
  394. or disabled based on the configuration of the kernel. To override
  395. the default, use the following compile-time flags.
  396. To enable NAPI, compile the driver module, passing in a configuration option:
  397. make CFLAGS_EXTRA=-DE1000_NAPI install
  398. To disable NAPI, compile the driver module, passing in a configuration option:
  399. make CFLAGS_EXTRA=-DE1000_NO_NAPI install
  400. See www.cyberus.ca/~hadi/usenix-paper.tgz for more information on NAPI.
  401. Known Issues
  402. ============
  403. Dropped Receive Packets on Half-duplex 10/100 Networks
  404. ------------------------------------------------------
  405. If you have an Intel PCI Express adapter running at 10mbps or 100mbps, half-
  406. duplex, you may observe occasional dropped receive packets. There are no
  407. workarounds for this problem in this network configuration. The network must
  408. be updated to operate in full-duplex, and/or 1000mbps only.
  409. Jumbo Frames System Requirement
  410. -------------------------------
  411. Memory allocation failures have been observed on Linux systems with 64 MB
  412. of RAM or less that are running Jumbo Frames. If you are using Jumbo
  413. Frames, your system may require more than the advertised minimum
  414. requirement of 64 MB of system memory.
  415. Performance Degradation with Jumbo Frames
  416. -----------------------------------------
  417. Degradation in throughput performance may be observed in some Jumbo frames
  418. environments. If this is observed, increasing the application's socket
  419. buffer size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values
  420. may help. See the specific application manual and
  421. /usr/src/linux*/Documentation/
  422. networking/ip-sysctl.txt for more details.
  423. Jumbo Frames on Foundry BigIron 8000 switch
  424. -------------------------------------------
  425. There is a known issue using Jumbo frames when connected to a Foundry
  426. BigIron 8000 switch. This is a 3rd party limitation. If you experience
  427. loss of packets, lower the MTU size.
  428. Allocating Rx Buffers when Using Jumbo Frames
  429. ---------------------------------------------
  430. Allocating Rx buffers when using Jumbo Frames on 2.6.x kernels may fail if
  431. the available memory is heavily fragmented. This issue may be seen with PCI-X
  432. adapters or with packet split disabled. This can be reduced or eliminated
  433. by changing the amount of available memory for receive buffer allocation, by
  434. increasing /proc/sys/vm/min_free_kbytes.
  435. Multiple Interfaces on Same Ethernet Broadcast Network
  436. ------------------------------------------------------
  437. Due to the default ARP behavior on Linux, it is not possible to have
  438. one system on two IP networks in the same Ethernet broadcast domain
  439. (non-partitioned switch) behave as expected. All Ethernet interfaces
  440. will respond to IP traffic for any IP address assigned to the system.
  441. This results in unbalanced receive traffic.
  442. If you have multiple interfaces in a server, either turn on ARP
  443. filtering by entering:
  444. echo 1 > /proc/sys/net/ipv4/conf/all/arp_filter
  445. (this only works if your kernel's version is higher than 2.4.5),
  446. NOTE: This setting is not saved across reboots. The configuration
  447. change can be made permanent by adding the line:
  448. net.ipv4.conf.all.arp_filter = 1
  449. to the file /etc/sysctl.conf
  450. or,
  451. install the interfaces in separate broadcast domains (either in
  452. different switches or in a switch partitioned to VLANs).
  453. 82541/82547 can't link or are slow to link with some link partners
  454. -----------------------------------------------------------------
  455. There is a known compatibility issue with 82541/82547 and some
  456. low-end switches where the link will not be established, or will
  457. be slow to establish. In particular, these switches are known to
  458. be incompatible with 82541/82547:
  459. Planex FXG-08TE
  460. I-O Data ETG-SH8
  461. To workaround this issue, the driver can be compiled with an override
  462. of the PHY's master/slave setting. Forcing master or forcing slave
  463. mode will improve time-to-link.
  464. # make CFLAGS_EXTRA=-DE1000_MASTER_SLAVE=<n>
  465. Where <n> is:
  466. 0 = Hardware default
  467. 1 = Master mode
  468. 2 = Slave mode
  469. 3 = Auto master/slave
  470. Disable rx flow control with ethtool
  471. ------------------------------------
  472. In order to disable receive flow control using ethtool, you must turn
  473. off auto-negotiation on the same command line.
  474. For example:
  475. ethtool -A eth? autoneg off rx off
  476. Unplugging network cable while ethtool -p is running
  477. ----------------------------------------------------
  478. In kernel versions 2.5.50 and later (including 2.6 kernel), unplugging
  479. the network cable while ethtool -p is running will cause the system to
  480. become unresponsive to keyboard commands, except for control-alt-delete.
  481. Restarting the system appears to be the only remedy.
  482. Support
  483. =======
  484. For general information, go to the Intel support website at:
  485. http://support.intel.com
  486. or the Intel Wired Networking project hosted by Sourceforge at:
  487. http://sourceforge.net/projects/e1000
  488. If an issue is identified with the released source code on the supported
  489. kernel with a supported adapter, email the specific information related
  490. to the issue to e1000-devel@lists.sf.net