2ff7d1b635
- Only update the rx ring consumer pointer after running through the rx loop, not with each iteration through the loop. - If possible, use a fast interupt handler instead of an ithread handler. Use the interrupt handler to check and squelch the interrupt, then schedule a taskqueue to do the actual work. This has three benefits: - Eliminates the 'interrupt aliasing' problem found in many chipsets by allowing the driver to mask the interrupt in the NIC instead of the OS masking the interrupt in the APIC. - Allows the driver to control the amount of work done in the interrupt handler. This results in what I call 'adaptive polling', where you get the latency benefits of a quick response to interrupts with the interrupt mitigation and work partitioning of polling. Polling is still an option in the driver, but I consider it orthogonal to this work. - Don't hold the driver lock in the RX handler. The handler and all data associated is effectively serialized already. This eliminates the cost of dropping and reaquiring the lock for every receieved packet. The result is much lower contention for the driver lock, resulting in lower CPU usage and lower latency for interactive workloads. The amount of work done in the taskqueue is controlled by the sysctl dev.em.N.rx_processing_limit and tunable hw.em.rx_process_limit Setting these to -1 effectively removes the limit. The fast interrupt and taskqueue can be disabled by defining NO_EM_FASTINTR. This work has been shown to increase fast-forwarding from ~570 kpps to ~750 kpps (note that the same NIC hardware seems unable to transmit more than 800 kpps, so this increase appears to be limited almost solely by the hardware). Gains have been shown in other workloads, ranging from better performance to elimination of over-saturation livelocks. Thanks to Andre Opperman for his time and resources from his network performance project in performing much of the testing. Thanks to Gleb Smirnoff and Danny Braniss for their help in testing also. |
||
---|---|---|
.. | ||
if_em_hw.c | ||
if_em_hw.h | ||
if_em_osdep.h | ||
if_em.c | ||
if_em.h | ||
LICENSE | ||
README |
$FreeBSD$ FreeBSD* Driver for the Intel(R) PRO/1000 Family of Adapters ============================================================ March 18, 2005 Contents ======== - Overview - Identifying Your Adapter - Building and Installation - Speed and Duplex Configuration - Additional Configurations - Known Limitations - Support - License Overview ======== This file describes the FreeBSD* driver, version 2.1.x, for the Intel(R) PRO/1000 Family of Adapters. This driver has been developed for use with FreeBSD, version 5.x. For questions related to hardware requirements, refer to the documentation supplied with your Intel PRO/1000 adapter. All hardware requirements listed apply to use with FreeBSD. Identifying Your Adapter ======================== For information on how to identify your adapter, go to the Adapter & Driver ID Guide at: http://support.intel.com/support/network/adapter/pro100/21397.htm For the latest Intel network drivers for FreeBSD, see: http://appsr.intel.com/scripts-df/support_intel.asp NOTE: Mobile adapters are not fully supported. Building and Installation ========================= NOTE: The driver can be installed as a dynamic loadable kernel module or compiled into the kernel. You must have kernel sources installed in order to compile the driver module. In the instructions below, x.x.x is the driver version as indicated in the name of the driver tar file. 1. Move the base driver tar file to the directory of your choice. For example, use /home/username/em or /usr/local/src/em. 2. Untar/unzip the archive: tar xvfz em-x.x.x.tar.gz This will create an em-x.x.x directory. 3. To create a loadable module, perform the following steps. NOTE: To compile the driver into the kernel, go directly to step 4. a. To compile the module cd em-x.x.x make b. To install the compiled module in system directory: make install c. If you want the driver to load automatically when the system is booted: 1. Edit /boot/loader.conf, and add the following line: if_em_load="YES" 4. To compile the driver into the kernel: cd em-x.x.x/src cp if_em* /usr/src/sys/dev/em cp Makefile.kernel /usr/src/sys/modules/em/Makefile Edit the /usr/src/sys/conf/files.i386 file, and add the following lines only if they don't already exist: dev/em/if_em.c optional em dev/em/if_em_hw.c optional em Remove the following lines from the /usr/src/sys/conf/files.i386 file, if they exist: dev/em/if_em_fxhw.c optional em dev/em/if_em_phy.c optional em Edit the kernel configuration file (i.e., GENERIC or MYKERNEL) in /usr/src/sys/i386/conf, and ensure the following line is present: device em Compile and install the kernel. The system must be rebooted for the kernel updates to take effect. For additional information on compiling the kernel, consult the FreeBSD operating system documentation. 5. To assign an IP address to the interface, enter the following: ifconfig em<interface_num> <IP_address> 6. Verify that the interface works. Enter the following, where <IP_address> is the IP address for another machine on the same subnet as the interface that is being tested: ping <IP_address> 7. To configure the IP address to remain after reboot, edit /etc/rc.conf, and create the appropriate ifconfig_em<interface_num>entry: ifconfig_em<interface_num>="<ifconfig_settings>" Example usage: ifconfig_em0="inet 192.168.10.1 netmask 255.255.255.0" NOTE: For assistance, see the ifconfig man page. Speed and Duplex Configuration ============================== By default, the adapter auto-negotiates the speed and duplex of the connection. If there is a specific need, the ifconfig utility can be used to configure the speed and duplex settings on the adapter. Example usage: ifconfig em<interface_num> <IP_address> media 100baseTX mediaopt full-duplex NOTE: Only use mediaopt to set the driver to full-duplex. If mediaopt is not specified and you are not running at gigabit speed, the driver defaults to half-duplex. This driver supports the following media type options: autoselect - Enables auto-negotiation for speed and duplex. 10baseT/UTP - Sets speed to 10 Mbps. Use the ifconfig mediaopt option to select full-duplex mode. 100baseTX - Sets speed to 100 Mbps. Use the ifconfig mediaopt option to select full-duplex mode. 1000baseTX - Sets speed to 1000 Mbps. In this case, the driver supports only full-duplex mode. 1000baseSX - Sets speed to 1000 Mbps. In this case, the driver supports only full-duplex mode. For more information on the ifconfig utility, see the ifconfig man page. Additional Configurations ========================= The driver supports Transmit/Receive Checksum Offload and Jumbo Frames on all but the 82542-based adapters. For specific adapters, refer to the Identifying Your Adapter section. Jumbo Frames ------------ To enable Jumbo Frames, use the ifconfig utility to increase the MTU beyond 1500 bytes. NOTES: Only enable Jumbo Frames if your network infrastructure supports them. The Jumbo Frames setting on the switch must be set to at least 22 bytes larger than that of the MTU. The Intel PRO/1000 PM Network Connection does not support jumbo frames. The Jumbo Frames MTU range for Intel Adapters is 1500 to 16114. The default MTU range is 1500. To modify the setting, enter the following: ifconfig em<interface_num> <hostname or IP address> mtu 9000 To confirm the MTU used between two specific devices, use: route get <destination_IP_address> VLANs ----- To create a new VLAN interface: ifconfig <vlan_name> create To associate the VLAN interface with a physical interface and assign a VLAN ID, IP address, and netmask: ifconfig <vlan_name> <ip_address> netmask <subnet_mask> vlan <vlan_id> vlandev <physical_interface> Example: ifconfig vlan10 10.0.0.1 netmask 255.255.255.0 vlan10 vlandev em0 In this example, all packets will be marked on egress with 802.1Q VLAN tags, specifying a VLAN ID of 10. To remove a VLAN interface: ifconfig <vlan_name> destroy Polling ------- To enable polling in the driver, add the following options to the kernel configuration, and then recompile the kernel: options DEVICE_POLLING options HZ=1000 At runtime use: sysctl kern.polling.enable=1 to turn polling on Use: sysctl kern.polling.enable=0 to turn polling off Checksum Offload ---------------- Checksum offloading is not supported on 82542 Gigabit adapters. Checksum offloading supports both TCP and UDP packets and is supported for both transmit and receive. Checksum offloading can be enabled or disabled using ifconfig. Both transmit and receive offloading will be either enabled or disabled together. You cannot enable/disable one without the other. To enable checksum offloading: ifconfig <interface_num> rxcsum To disable checksum offloading: ifconfig <interface_num> -rxcsum To confirm the current setting: ifconfig <interface_num> Look for the presence or absence of the following line: options=3 <RXCSUM,TXCSUM> See the ifconfig man page for further information. Known Limitations ================= There are known performance issues with this driver when running UDP traffic with Jumbo Frames. There is a known compatibility issue where time to link is slow or link is not established between 82541/82547 controllers and some switches. Known switches include: Planex FXG-08TE I-O Data ETG-SH8 The driver can be compiled with the following changes: Edit ./em.x.x.x/src/if_em.h to uncomment the #define EM_MASTER_SLAVE from within the comments. For example, change from: /* #define EM_MASTER_SLAVE 2 */ to: #define EM_MASTER_SLAVE 2 Use one of the following options: 1 = Master mode 2 = Slave mode 3 = Auto master/slave Setting 2 is recommended. Recompile the module: a. To compile the module cd em-x.x.x make clean make b. To install the compiled module in system directory: make install Support ======= For general information and support, go to the Intel support website at: http://support.intel.com If an issue is identified, support is through email only at: freebsdnic@mailbox.intel.com License ======= This software program is released under the terms of a license agreement between you ('Licensee') and Intel. Do not use or load this software or any associated materials (collectively, the 'Software') until you have carefully read the full terms and conditions of the LICENSE located in this software package. By loading or using the Software, you agree to the terms of this Agreement. If you do not agree with the terms of this Agreement, do not install or use the Software. * Other names and brands may be claimed as the property of others.