e1000.txt 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461
  1. Linux* Base Driver for Intel(R) Ethernet Network Connection
  2. ===========================================================
  3. Intel Gigabit Linux driver.
  4. Copyright(c) 1999 - 2013 Intel Corporation.
  5. Contents
  6. ========
  7. - Identifying Your Adapter
  8. - Command Line Parameters
  9. - Speed and Duplex Configuration
  10. - Additional Configurations
  11. - Support
  12. Identifying Your Adapter
  13. ========================
  14. For more information on how to identify your adapter, go to the Adapter &
  15. Driver ID Guide at:
  16. http://support.intel.com/support/go/network/adapter/idguide.htm
  17. For the latest Intel network drivers for Linux, refer to the following
  18. website. In the search field, enter your adapter name or type, or use the
  19. networking link on the left to search for your adapter:
  20. http://support.intel.com/support/go/network/adapter/home.htm
  21. Command Line Parameters
  22. =======================
  23. The default value for each parameter is generally the recommended setting,
  24. unless otherwise noted.
  25. NOTES: For more information about the AutoNeg, Duplex, and Speed
  26. parameters, see the "Speed and Duplex Configuration" section in
  27. this document.
  28. For more information about the InterruptThrottleRate,
  29. RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay
  30. parameters, see the application note at:
  31. http://www.intel.com/design/network/applnots/ap450.htm
  32. AutoNeg
  33. -------
  34. (Supported only on adapters with copper connections)
  35. Valid Range: 0x01-0x0F, 0x20-0x2F
  36. Default Value: 0x2F
  37. This parameter is a bit-mask that specifies the speed and duplex settings
  38. advertised by the adapter. When this parameter is used, the Speed and
  39. Duplex parameters must not be specified.
  40. NOTE: Refer to the Speed and Duplex section of this readme for more
  41. information on the AutoNeg parameter.
  42. Duplex
  43. ------
  44. (Supported only on adapters with copper connections)
  45. Valid Range: 0-2 (0=auto-negotiate, 1=half, 2=full)
  46. Default Value: 0
  47. This defines the direction in which data is allowed to flow. Can be
  48. either one or two-directional. If both Duplex and the link partner are
  49. set to auto-negotiate, the board auto-detects the correct duplex. If the
  50. link partner is forced (either full or half), Duplex defaults to half-
  51. duplex.
  52. FlowControl
  53. -----------
  54. Valid Range: 0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx)
  55. Default Value: Reads flow control settings from the EEPROM
  56. This parameter controls the automatic generation(Tx) and response(Rx)
  57. to Ethernet PAUSE frames.
  58. InterruptThrottleRate
  59. ---------------------
  60. (not supported on Intel(R) 82542, 82543 or 82544-based adapters)
  61. Valid Range: 0,1,3,4,100-100000 (0=off, 1=dynamic, 3=dynamic conservative,
  62. 4=simplified balancing)
  63. Default Value: 3
  64. The driver can limit the amount of interrupts per second that the adapter
  65. will generate for incoming packets. It does this by writing a value to the
  66. adapter that is based on the maximum amount of interrupts that the adapter
  67. will generate per second.
  68. Setting InterruptThrottleRate to a value greater or equal to 100
  69. will program the adapter to send out a maximum of that many interrupts
  70. per second, even if more packets have come in. This reduces interrupt
  71. load on the system and can lower CPU utilization under heavy load,
  72. but will increase latency as packets are not processed as quickly.
  73. The default behaviour of the driver previously assumed a static
  74. InterruptThrottleRate value of 8000, providing a good fallback value for
  75. all traffic types,but lacking in small packet performance and latency.
  76. The hardware can handle many more small packets per second however, and
  77. for this reason an adaptive interrupt moderation algorithm was implemented.
  78. Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in which
  79. it dynamically adjusts the InterruptThrottleRate value based on the traffic
  80. that it receives. After determining the type of incoming traffic in the last
  81. timeframe, it will adjust the InterruptThrottleRate to an appropriate value
  82. for that traffic.
  83. The algorithm classifies the incoming traffic every interval into
  84. classes. Once the class is determined, the InterruptThrottleRate value is
  85. adjusted to suit that traffic type the best. There are three classes defined:
  86. "Bulk traffic", for large amounts of packets of normal size; "Low latency",
  87. for small amounts of traffic and/or a significant percentage of small
  88. packets; and "Lowest latency", for almost completely small packets or
  89. minimal traffic.
  90. In dynamic conservative mode, the InterruptThrottleRate value is set to 4000
  91. for traffic that falls in class "Bulk traffic". If traffic falls in the "Low
  92. latency" or "Lowest latency" class, the InterruptThrottleRate is increased
  93. stepwise to 20000. This default mode is suitable for most applications.
  94. For situations where low latency is vital such as cluster or
  95. grid computing, the algorithm can reduce latency even more when
  96. InterruptThrottleRate is set to mode 1. In this mode, which operates
  97. the same as mode 3, the InterruptThrottleRate will be increased stepwise to
  98. 70000 for traffic in class "Lowest latency".
  99. In simplified mode the interrupt rate is based on the ratio of TX and
  100. RX traffic. If the bytes per second rate is approximately equal, the
  101. interrupt rate will drop as low as 2000 interrupts per second. If the
  102. traffic is mostly transmit or mostly receive, the interrupt rate could
  103. be as high as 8000.
  104. Setting InterruptThrottleRate to 0 turns off any interrupt moderation
  105. and may improve small packet latency, but is generally not suitable
  106. for bulk throughput traffic.
  107. NOTE: InterruptThrottleRate takes precedence over the TxAbsIntDelay and
  108. RxAbsIntDelay parameters. In other words, minimizing the receive
  109. and/or transmit absolute delays does not force the controller to
  110. generate more interrupts than what the Interrupt Throttle Rate
  111. allows.
  112. CAUTION: If you are using the Intel(R) PRO/1000 CT Network Connection
  113. (controller 82547), setting InterruptThrottleRate to a value
  114. greater than 75,000, may hang (stop transmitting) adapters
  115. under certain network conditions. If this occurs a NETDEV
  116. WATCHDOG message is logged in the system event log. In
  117. addition, the controller is automatically reset, restoring
  118. the network connection. To eliminate the potential for the
  119. hang, ensure that InterruptThrottleRate is set no greater
  120. than 75,000 and is not set to 0.
  121. NOTE: When e1000 is loaded with default settings and multiple adapters
  122. are in use simultaneously, the CPU utilization may increase non-
  123. linearly. In order to limit the CPU utilization without impacting
  124. the overall throughput, we recommend that you load the driver as
  125. follows:
  126. modprobe e1000 InterruptThrottleRate=3000,3000,3000
  127. This sets the InterruptThrottleRate to 3000 interrupts/sec for
  128. the first, second, and third instances of the driver. The range
  129. of 2000 to 3000 interrupts per second works on a majority of
  130. systems and is a good starting point, but the optimal value will
  131. be platform-specific. If CPU utilization is not a concern, use
  132. RX_POLLING (NAPI) and default driver settings.
  133. RxDescriptors
  134. -------------
  135. Valid Range: 80-256 for 82542 and 82543-based adapters
  136. 80-4096 for all other supported adapters
  137. Default Value: 256
  138. This value specifies the number of receive buffer descriptors allocated
  139. by the driver. Increasing this value allows the driver to buffer more
  140. incoming packets, at the expense of increased system memory utilization.
  141. Each descriptor is 16 bytes. A receive buffer is also allocated for each
  142. descriptor and can be either 2048, 4096, 8192, or 16384 bytes, depending
  143. on the MTU setting. The maximum MTU size is 16110.
  144. NOTE: MTU designates the frame size. It only needs to be set for Jumbo
  145. Frames. Depending on the available system resources, the request
  146. for a higher number of receive descriptors may be denied. In this
  147. case, use a lower number.
  148. RxIntDelay
  149. ----------
  150. Valid Range: 0-65535 (0=off)
  151. Default Value: 0
  152. This value delays the generation of receive interrupts in units of 1.024
  153. microseconds. Receive interrupt reduction can improve CPU efficiency if
  154. properly tuned for specific network traffic. Increasing this value adds
  155. extra latency to frame reception and can end up decreasing the throughput
  156. of TCP traffic. If the system is reporting dropped receives, this value
  157. may be set too high, causing the driver to run out of available receive
  158. descriptors.
  159. CAUTION: When setting RxIntDelay to a value other than 0, adapters may
  160. hang (stop transmitting) under certain network conditions. If
  161. this occurs a NETDEV WATCHDOG message is logged in the system
  162. event log. In addition, the controller is automatically reset,
  163. restoring the network connection. To eliminate the potential
  164. for the hang ensure that RxIntDelay is set to 0.
  165. RxAbsIntDelay
  166. -------------
  167. (This parameter is supported only on 82540, 82545 and later adapters.)
  168. Valid Range: 0-65535 (0=off)
  169. Default Value: 128
  170. This value, in units of 1.024 microseconds, limits the delay in which a
  171. receive interrupt is generated. Useful only if RxIntDelay is non-zero,
  172. this value ensures that an interrupt is generated after the initial
  173. packet is received within the set amount of time. Proper tuning,
  174. along with RxIntDelay, may improve traffic throughput in specific network
  175. conditions.
  176. Speed
  177. -----
  178. (This parameter is supported only on adapters with copper connections.)
  179. Valid Settings: 0, 10, 100, 1000
  180. Default Value: 0 (auto-negotiate at all supported speeds)
  181. Speed forces the line speed to the specified value in megabits per second
  182. (Mbps). If this parameter is not specified or is set to 0 and the link
  183. partner is set to auto-negotiate, the board will auto-detect the correct
  184. speed. Duplex should also be set when Speed is set to either 10 or 100.
  185. TxDescriptors
  186. -------------
  187. Valid Range: 80-256 for 82542 and 82543-based adapters
  188. 80-4096 for all other supported adapters
  189. Default Value: 256
  190. This value is the number of transmit descriptors allocated by the driver.
  191. Increasing this value allows the driver to queue more transmits. Each
  192. descriptor is 16 bytes.
  193. NOTE: Depending on the available system resources, the request for a
  194. higher number of transmit descriptors may be denied. In this case,
  195. use a lower number.
  196. TxDescriptorStep
  197. ----------------
  198. Valid Range: 1 (use every Tx Descriptor)
  199. 4 (use every 4th Tx Descriptor)
  200. Default Value: 1 (use every Tx Descriptor)
  201. On certain non-Intel architectures, it has been observed that intense TX
  202. traffic bursts of short packets may result in an improper descriptor
  203. writeback. If this occurs, the driver will report a "TX Timeout" and reset
  204. the adapter, after which the transmit flow will restart, though data may
  205. have stalled for as much as 10 seconds before it resumes.
  206. The improper writeback does not occur on the first descriptor in a system
  207. memory cache-line, which is typically 32 bytes, or 4 descriptors long.
  208. Setting TxDescriptorStep to a value of 4 will ensure that all TX descriptors
  209. are aligned to the start of a system memory cache line, and so this problem
  210. will not occur.
  211. NOTES: Setting TxDescriptorStep to 4 effectively reduces the number of
  212. TxDescriptors available for transmits to 1/4 of the normal allocation.
  213. This has a possible negative performance impact, which may be
  214. compensated for by allocating more descriptors using the TxDescriptors
  215. module parameter.
  216. There are other conditions which may result in "TX Timeout", which will
  217. not be resolved by the use of the TxDescriptorStep parameter. As the
  218. issue addressed by this parameter has never been observed on Intel
  219. Architecture platforms, it should not be used on Intel platforms.
  220. TxIntDelay
  221. ----------
  222. Valid Range: 0-65535 (0=off)
  223. Default Value: 64
  224. This value delays the generation of transmit interrupts in units of
  225. 1.024 microseconds. Transmit interrupt reduction can improve CPU
  226. efficiency if properly tuned for specific network traffic. If the
  227. system is reporting dropped transmits, this value may be set too high
  228. causing the driver to run out of available transmit descriptors.
  229. TxAbsIntDelay
  230. -------------
  231. (This parameter is supported only on 82540, 82545 and later adapters.)
  232. Valid Range: 0-65535 (0=off)
  233. Default Value: 64
  234. This value, in units of 1.024 microseconds, limits the delay in which a
  235. transmit interrupt is generated. Useful only if TxIntDelay is non-zero,
  236. this value ensures that an interrupt is generated after the initial
  237. packet is sent on the wire within the set amount of time. Proper tuning,
  238. along with TxIntDelay, may improve traffic throughput in specific
  239. network conditions.
  240. XsumRX
  241. ------
  242. (This parameter is NOT supported on the 82542-based adapter.)
  243. Valid Range: 0-1
  244. Default Value: 1
  245. A value of '1' indicates that the driver should enable IP checksum
  246. offload for received packets (both UDP and TCP) to the adapter hardware.
  247. Copybreak
  248. ---------
  249. Valid Range: 0-xxxxxxx (0=off)
  250. Default Value: 256
  251. Usage: insmod e1000.ko copybreak=128
  252. Driver copies all packets below or equaling this size to a fresh RX
  253. buffer before handing it up the stack.
  254. This parameter is different than other parameters, in that it is a
  255. single (not 1,1,1 etc.) parameter applied to all driver instances and
  256. it is also available during runtime at
  257. /sys/module/e1000/parameters/copybreak
  258. SmartPowerDownEnable
  259. --------------------
  260. Valid Range: 0-1
  261. Default Value: 0 (disabled)
  262. Allows PHY to turn off in lower power states. The user can turn off
  263. this parameter in supported chipsets.
  264. KumeranLockLoss
  265. ---------------
  266. Valid Range: 0-1
  267. Default Value: 1 (enabled)
  268. This workaround skips resetting the PHY at shutdown for the initial
  269. silicon releases of ICH8 systems.
  270. Speed and Duplex Configuration
  271. ==============================
  272. Three keywords are used to control the speed and duplex configuration.
  273. These keywords are Speed, Duplex, and AutoNeg.
  274. If the board uses a fiber interface, these keywords are ignored, and the
  275. fiber interface board only links at 1000 Mbps full-duplex.
  276. For copper-based boards, the keywords interact as follows:
  277. The default operation is auto-negotiate. The board advertises all
  278. supported speed and duplex combinations, and it links at the highest
  279. common speed and duplex mode IF the link partner is set to auto-negotiate.
  280. If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps
  281. is advertised (The 1000BaseT spec requires auto-negotiation.)
  282. If Speed = 10 or 100, then both Speed and Duplex should be set. Auto-
  283. negotiation is disabled, and the AutoNeg parameter is ignored. Partner
  284. SHOULD also be forced.
  285. The AutoNeg parameter is used when more control is required over the
  286. auto-negotiation process. It should be used when you wish to control which
  287. speed and duplex combinations are advertised during the auto-negotiation
  288. process.
  289. The parameter may be specified as either a decimal or hexadecimal value as
  290. determined by the bitmap below.
  291. Bit position 7 6 5 4 3 2 1 0
  292. Decimal Value 128 64 32 16 8 4 2 1
  293. Hex value 80 40 20 10 8 4 2 1
  294. Speed (Mbps) N/A N/A 1000 N/A 100 100 10 10
  295. Duplex Full Full Half Full Half
  296. Some examples of using AutoNeg:
  297. modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half)
  298. modprobe e1000 AutoNeg=1 (Same as above)
  299. modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full)
  300. modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full)
  301. modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half)
  302. modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100
  303. Half)
  304. modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full)
  305. modprobe e1000 AutoNeg=32 (Same as above)
  306. Note that when this parameter is used, Speed and Duplex must not be specified.
  307. If the link partner is forced to a specific speed and duplex, then this
  308. parameter should not be used. Instead, use the Speed and Duplex parameters
  309. previously mentioned to force the adapter to the same speed and duplex.
  310. Additional Configurations
  311. =========================
  312. Jumbo Frames
  313. ------------
  314. Jumbo Frames support is enabled by changing the MTU to a value larger than
  315. the default of 1500. Use the ifconfig command to increase the MTU size.
  316. For example:
  317. ifconfig eth<x> mtu 9000 up
  318. This setting is not saved across reboots. It can be made permanent if
  319. you add:
  320. MTU=9000
  321. to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>. This example
  322. applies to the Red Hat distributions; other distributions may store this
  323. setting in a different location.
  324. Notes:
  325. Degradation in throughput performance may be observed in some Jumbo frames
  326. environments. If this is observed, increasing the application's socket buffer
  327. size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values may help.
  328. See the specific application manual and /usr/src/linux*/Documentation/
  329. networking/ip-sysctl.txt for more details.
  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 is not supported and may result in
  333. poor performance or loss of link.
  334. - Adapters based on the Intel(R) 82542 and 82573V/E controller do not
  335. support Jumbo Frames. These correspond to the following product names:
  336. Intel(R) PRO/1000 Gigabit Server Adapter
  337. Intel(R) PRO/1000 PM Network Connection
  338. ethtool
  339. -------
  340. The driver utilizes the ethtool interface for driver configuration and
  341. diagnostics, as well as displaying statistical information. The ethtool
  342. version 1.6 or later is required for this functionality.
  343. The latest release of ethtool can be found from
  344. http://ftp.kernel.org/pub/software/network/ethtool/
  345. Enabling Wake on LAN* (WoL)
  346. ---------------------------
  347. WoL is configured through the ethtool* utility.
  348. WoL will be enabled on the system during the next shut down or reboot.
  349. For this driver version, in order to enable WoL, the e1000 driver must be
  350. loaded when shutting down or rebooting the system.
  351. Support
  352. =======
  353. For general information, go to the Intel support website at:
  354. http://support.intel.com
  355. or the Intel Wired Networking project hosted by Sourceforge at:
  356. http://sourceforge.net/projects/e1000
  357. If an issue is identified with the released source code on the supported
  358. kernel with a supported adapter, email the specific information related
  359. to the issue to e1000-devel@lists.sf.net