1 .. SPDX-License-Identifier: GPL-2.0+
3 Linux* Base Driver for Intel(R) Ethernet Network Connection
4 ===========================================================
6 Intel Gigabit Linux driver.
7 Copyright(c) 1999 - 2013 Intel Corporation.
12 - Identifying Your Adapter
13 - Command Line Parameters
14 - Speed and Duplex Configuration
15 - Additional Configurations
18 Identifying Your Adapter
19 ========================
21 For more information on how to identify your adapter, go to the Adapter &
24 http://support.intel.com/support/go/network/adapter/idguide.htm
26 For the latest Intel network drivers for Linux, refer to the following
27 website. In the search field, enter your adapter name or type, or use the
28 networking link on the left to search for your adapter:
30 http://support.intel.com/support/go/network/adapter/home.htm
32 Command Line Parameters
33 =======================
35 The default value for each parameter is generally the recommended setting,
36 unless otherwise noted.
39 For more information about the AutoNeg, Duplex, and Speed
40 parameters, see the "Speed and Duplex Configuration" section in
43 For more information about the InterruptThrottleRate,
44 RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay
45 parameters, see the application note at:
46 http://www.intel.com/design/network/applnots/ap450.htm
51 (Supported only on adapters with copper connections)
53 :Valid Range: 0x01-0x0F, 0x20-0x2F
56 This parameter is a bit-mask that specifies the speed and duplex settings
57 advertised by the adapter. When this parameter is used, the Speed and
58 Duplex parameters must not be specified.
61 Refer to the Speed and Duplex section of this readme for more
62 information on the AutoNeg parameter.
67 (Supported only on adapters with copper connections)
69 :Valid Range: 0-2 (0=auto-negotiate, 1=half, 2=full)
72 This defines the direction in which data is allowed to flow. Can be
73 either one or two-directional. If both Duplex and the link partner are
74 set to auto-negotiate, the board auto-detects the correct duplex. If the
75 link partner is forced (either full or half), Duplex defaults to half-
81 :Valid Range: 0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx)
82 :Default Value: Reads flow control settings from the EEPROM
84 This parameter controls the automatic generation(Tx) and response(Rx)
85 to Ethernet PAUSE frames.
90 (not supported on Intel(R) 82542, 82543 or 82544-based adapters)
93 0,1,3,4,100-100000 (0=off, 1=dynamic, 3=dynamic conservative,
94 4=simplified balancing)
97 The driver can limit the amount of interrupts per second that the adapter
98 will generate for incoming packets. It does this by writing a value to the
99 adapter that is based on the maximum amount of interrupts that the adapter
100 will generate per second.
102 Setting InterruptThrottleRate to a value greater or equal to 100
103 will program the adapter to send out a maximum of that many interrupts
104 per second, even if more packets have come in. This reduces interrupt
105 load on the system and can lower CPU utilization under heavy load,
106 but will increase latency as packets are not processed as quickly.
108 The default behaviour of the driver previously assumed a static
109 InterruptThrottleRate value of 8000, providing a good fallback value for
110 all traffic types,but lacking in small packet performance and latency.
111 The hardware can handle many more small packets per second however, and
112 for this reason an adaptive interrupt moderation algorithm was implemented.
114 Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in which
115 it dynamically adjusts the InterruptThrottleRate value based on the traffic
116 that it receives. After determining the type of incoming traffic in the last
117 timeframe, it will adjust the InterruptThrottleRate to an appropriate value
120 The algorithm classifies the incoming traffic every interval into
121 classes. Once the class is determined, the InterruptThrottleRate value is
122 adjusted to suit that traffic type the best. There are three classes defined:
123 "Bulk traffic", for large amounts of packets of normal size; "Low latency",
124 for small amounts of traffic and/or a significant percentage of small
125 packets; and "Lowest latency", for almost completely small packets or
128 In dynamic conservative mode, the InterruptThrottleRate value is set to 4000
129 for traffic that falls in class "Bulk traffic". If traffic falls in the "Low
130 latency" or "Lowest latency" class, the InterruptThrottleRate is increased
131 stepwise to 20000. This default mode is suitable for most applications.
133 For situations where low latency is vital such as cluster or
134 grid computing, the algorithm can reduce latency even more when
135 InterruptThrottleRate is set to mode 1. In this mode, which operates
136 the same as mode 3, the InterruptThrottleRate will be increased stepwise to
137 70000 for traffic in class "Lowest latency".
139 In simplified mode the interrupt rate is based on the ratio of TX and
140 RX traffic. If the bytes per second rate is approximately equal, the
141 interrupt rate will drop as low as 2000 interrupts per second. If the
142 traffic is mostly transmit or mostly receive, the interrupt rate could
145 Setting InterruptThrottleRate to 0 turns off any interrupt moderation
146 and may improve small packet latency, but is generally not suitable
147 for bulk throughput traffic.
150 InterruptThrottleRate takes precedence over the TxAbsIntDelay and
151 RxAbsIntDelay parameters. In other words, minimizing the receive
152 and/or transmit absolute delays does not force the controller to
153 generate more interrupts than what the Interrupt Throttle Rate
157 If you are using the Intel(R) PRO/1000 CT Network Connection
158 (controller 82547), setting InterruptThrottleRate to a value
159 greater than 75,000, may hang (stop transmitting) adapters
160 under certain network conditions. If this occurs a NETDEV
161 WATCHDOG message is logged in the system event log. In
162 addition, the controller is automatically reset, restoring
163 the network connection. To eliminate the potential for the
164 hang, ensure that InterruptThrottleRate is set no greater
165 than 75,000 and is not set to 0.
168 When e1000 is loaded with default settings and multiple adapters
169 are in use simultaneously, the CPU utilization may increase non-
170 linearly. In order to limit the CPU utilization without impacting
171 the overall throughput, we recommend that you load the driver as
174 modprobe e1000 InterruptThrottleRate=3000,3000,3000
176 This sets the InterruptThrottleRate to 3000 interrupts/sec for
177 the first, second, and third instances of the driver. The range
178 of 2000 to 3000 interrupts per second works on a majority of
179 systems and is a good starting point, but the optimal value will
180 be platform-specific. If CPU utilization is not a concern, use
181 RX_POLLING (NAPI) and default driver settings.
187 - 48-256 for 82542 and 82543-based adapters
188 - 48-4096 for all other supported adapters
191 This value specifies the number of receive buffer descriptors allocated
192 by the driver. Increasing this value allows the driver to buffer more
193 incoming packets, at the expense of increased system memory utilization.
195 Each descriptor is 16 bytes. A receive buffer is also allocated for each
196 descriptor and can be either 2048, 4096, 8192, or 16384 bytes, depending
197 on the MTU setting. The maximum MTU size is 16110.
200 MTU designates the frame size. It only needs to be set for Jumbo
201 Frames. Depending on the available system resources, the request
202 for a higher number of receive descriptors may be denied. In this
203 case, use a lower number.
208 :Valid Range: 0-65535 (0=off)
211 This value delays the generation of receive interrupts in units of 1.024
212 microseconds. Receive interrupt reduction can improve CPU efficiency if
213 properly tuned for specific network traffic. Increasing this value adds
214 extra latency to frame reception and can end up decreasing the throughput
215 of TCP traffic. If the system is reporting dropped receives, this value
216 may be set too high, causing the driver to run out of available receive
220 When setting RxIntDelay to a value other than 0, adapters may
221 hang (stop transmitting) under certain network conditions. If
222 this occurs a NETDEV WATCHDOG message is logged in the system
223 event log. In addition, the controller is automatically reset,
224 restoring the network connection. To eliminate the potential
225 for the hang ensure that RxIntDelay is set to 0.
230 (This parameter is supported only on 82540, 82545 and later adapters.)
232 :Valid Range: 0-65535 (0=off)
235 This value, in units of 1.024 microseconds, limits the delay in which a
236 receive interrupt is generated. Useful only if RxIntDelay is non-zero,
237 this value ensures that an interrupt is generated after the initial
238 packet is received within the set amount of time. Proper tuning,
239 along with RxIntDelay, may improve traffic throughput in specific network
245 (This parameter is supported only on adapters with copper connections.)
247 :Valid Settings: 0, 10, 100, 1000
248 :Default Value: 0 (auto-negotiate at all supported speeds)
250 Speed forces the line speed to the specified value in megabits per second
251 (Mbps). If this parameter is not specified or is set to 0 and the link
252 partner is set to auto-negotiate, the board will auto-detect the correct
253 speed. Duplex should also be set when Speed is set to either 10 or 100.
259 - 48-256 for 82542 and 82543-based adapters
260 - 48-4096 for all other supported adapters
263 This value is the number of transmit descriptors allocated by the driver.
264 Increasing this value allows the driver to queue more transmits. Each
265 descriptor is 16 bytes.
268 Depending on the available system resources, the request for a
269 higher number of transmit descriptors may be denied. In this case,
275 :Valid Range: 0-65535 (0=off)
278 This value delays the generation of transmit interrupts in units of
279 1.024 microseconds. Transmit interrupt reduction can improve CPU
280 efficiency if properly tuned for specific network traffic. If the
281 system is reporting dropped transmits, this value may be set too high
282 causing the driver to run out of available transmit descriptors.
287 (This parameter is supported only on 82540, 82545 and later adapters.)
289 :Valid Range: 0-65535 (0=off)
292 This value, in units of 1.024 microseconds, limits the delay in which a
293 transmit interrupt is generated. Useful only if TxIntDelay is non-zero,
294 this value ensures that an interrupt is generated after the initial
295 packet is sent on the wire within the set amount of time. Proper tuning,
296 along with TxIntDelay, may improve traffic throughput in specific
302 (This parameter is NOT supported on the 82542-based adapter.)
307 A value of '1' indicates that the driver should enable IP checksum
308 offload for received packets (both UDP and TCP) to the adapter hardware.
313 :Valid Range: 0-xxxxxxx (0=off)
315 :Usage: modprobe e1000.ko copybreak=128
317 Driver copies all packets below or equaling this size to a fresh RX
318 buffer before handing it up the stack.
320 This parameter is different than other parameters, in that it is a
321 single (not 1,1,1 etc.) parameter applied to all driver instances and
322 it is also available during runtime at
323 /sys/module/e1000/parameters/copybreak
329 :Default Value: 0 (disabled)
331 Allows PHY to turn off in lower power states. The user can turn off
332 this parameter in supported chipsets.
334 Speed and Duplex Configuration
335 ==============================
337 Three keywords are used to control the speed and duplex configuration.
338 These keywords are Speed, Duplex, and AutoNeg.
340 If the board uses a fiber interface, these keywords are ignored, and the
341 fiber interface board only links at 1000 Mbps full-duplex.
343 For copper-based boards, the keywords interact as follows:
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.
349 - If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps
350 is advertised (The 1000BaseT spec requires auto-negotiation.)
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.
356 The AutoNeg parameter is used when more control is required over the
357 auto-negotiation process. It should be used when you wish to control which
358 speed and duplex combinations are advertised during the auto-negotiation
361 The parameter may be specified as either a decimal or hexadecimal value as
362 determined by the bitmap below.
364 ============== ====== ====== ======= ======= ====== ====== ======= ======
365 Bit position 7 6 5 4 3 2 1 0
366 Decimal Value 128 64 32 16 8 4 2 1
367 Hex value 80 40 20 10 8 4 2 1
368 Speed (Mbps) N/A N/A 1000 N/A 100 100 10 10
369 Duplex Full Full Half Full Half
370 ============== ====== ====== ======= ======= ====== ====== ======= ======
372 Some examples of using AutoNeg::
374 modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half)
375 modprobe e1000 AutoNeg=1 (Same as above)
376 modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full)
377 modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full)
378 modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half)
379 modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100
381 modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full)
382 modprobe e1000 AutoNeg=32 (Same as above)
384 Note that when this parameter is used, Speed and Duplex must not be specified.
386 If the link partner is forced to a specific speed and duplex, then this
387 parameter should not be used. Instead, use the Speed and Duplex parameters
388 previously mentioned to force the adapter to the same speed and duplex.
390 Additional Configurations
391 =========================
396 Jumbo Frames support is enabled by changing the MTU to a value larger than
397 the default of 1500. Use the ifconfig command to increase the MTU size.
400 ifconfig eth<x> mtu 9000 up
402 This setting is not saved across reboots. It can be made permanent if
407 to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>. This example
408 applies to the Red Hat distributions; other distributions may store this
409 setting in a different location.
412 Degradation in throughput performance may be observed in some Jumbo frames
413 environments. If this is observed, increasing the application's socket buffer
414 size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values may help.
415 See the specific application manual and /usr/src/linux*/Documentation/
416 networking/ip-sysctl.txt for more details.
418 - The maximum MTU setting for Jumbo Frames is 16110. This value coincides
419 with the maximum Jumbo Frames size of 16128.
421 - Using Jumbo frames at 10 or 100 Mbps is not supported and may result in
422 poor performance or loss of link.
424 - Adapters based on the Intel(R) 82542 and 82573V/E controller do not
425 support Jumbo Frames. These correspond to the following product names::
427 Intel(R) PRO/1000 Gigabit Server Adapter
428 Intel(R) PRO/1000 PM Network Connection
433 The driver utilizes the ethtool interface for driver configuration and
434 diagnostics, as well as displaying statistical information. The ethtool
435 version 1.6 or later is required for this functionality.
437 The latest release of ethtool can be found from
438 https://www.kernel.org/pub/software/network/ethtool/
440 Enabling Wake on LAN* (WoL)
441 ---------------------------
443 WoL is configured through the ethtool* utility.
445 WoL will be enabled on the system during the next shut down or reboot.
446 For this driver version, in order to enable WoL, the e1000 driver must be
447 loaded when shutting down or rebooting the system.
452 For general information, go to the Intel support website at:
454 http://support.intel.com
456 or the Intel Wired Networking project hosted by Sourceforge at:
458 http://sourceforge.net/projects/e1000
460 If an issue is identified with the released source code on the supported
461 kernel with a supported adapter, email the specific information related
462 to the issue to e1000-devel@lists.sf.net