Limitations Installing
Linux Driver Software
Patching
the Driver onto the Kernel
Unloading/Removing
the Linux Driver
Setting
Values for Optional Properties
The current version of the Linux driver has been tested on selected Linux distributions for ix86–64, 2.4x, and 2.6x kernels. Refer to the Distrib.txt file on the installation CD for a list of the specific Linux distributions on which the driver has been tested .
The Linux driver is released in the following packaging formats (file names):
Identical source files to build the driver are included in both RPM and TAR source packages. The tar file contains additional utilities such as patches and driver diskette images for network installation. The Binary RPM contains only the precompiled object file for Red Hat Linux 2.1 and Linux 3.0 distributions.
Installing the Source RPM Package
Building the Driver from a TAR file
Installing the Binary RPM Package
 |
NOTE: If a BCM5700 driver is loaded and the Linux kernel is updated, the BCM5700 driver module must be recompiled if the driver module was installed using the source RPM or the TAR package. |
rpm -ivh bcm5700-version.src.rpm
cd /usr/src/redhat,OpenLinux,turbo,packages,rpm …
rpm -bb SPECS/bcm5700.spec or rpmbuild -bb SPECS/bcm5700.spec
rpmbuild -bb SPECS/bcm5700.spec (for RPM version 4.x.x)
|
NOTE: During your attempt to install a source RPM package, the following message may be displayed: error: cannot create %sourcedir /usr/src/redhat/SOURCES The most likely cause of the error is that the rpm-build package has not been installed. Locate the rpm-build package on the Red Hat installation media and install it using the following command: rpm -ivh rpm-build-version.i386.rpm Complete the installation of the source RPM. |
rpm -ivh RPMS/i386/bcm5700-version.i386.rpm
The --force option is needed if installing over an existing distribution that may already contain an older version of the driver.
Depending on the kernel, the driver is installed to one of the following paths:
2.2.x kernels:
/lib/modules/kernel_version/net/bcm5700.o
2.4.x kernels:
/lib/modules/kernel_version/kernel/drivers/net/bcm5700.o
2.4.x kernels with the bcm5700 driver patched in:
/lib/modules/kernel_version/kernel/drivers/net/bcm/bcm5700.o
or
/lib/modules/kernel_version/kernel/drivers/addon/bcm5700/bcm5700.o
2.6.0 kernels:
/lib/modules/kernel_version/kernel/drivers/net/bcm5700.ko
2.6.0 kernels with bcm5700 driver patched in:
/lib/modules/kernel_version/kernel/drivers/net/bcm/bcm5700.ko
insmod bcm5700
To configure the network protocol and address, refer to the Linux version-specific documentation.
tar xvzf bcm5700-version.tar.gz
CD src
make
|
NOTE: If you are loading the driver on Red Hat 7.3, 2.1 AS, or other newer kernels that have the tg3 driver, refer to "Remove tg3 Driver" in the Distrib.txt file before loading the driver. |
insmod bcm5700.o
or, for Linux 2.6 kernels:
insmod bcm5700.ko
No message should be returned if this command runs properly
make install
|
NOTE: See the RPM instructions above for the location of the installed driver. |
The binary RPM packages contain precompiled kernel modules for the above Red Hat Linux distributions. These modules are designed only for the default kernels on these distributions. If you have customized or rebuilt your kernel, you may need to install your BCM5700 module using the source RPM package described above.
To install the binary RPM package
rpm -ivh bcm5700-version.i386.rpm
ifconfig eth# down
rmmod bcm5700
insmod bcm5700
ifconfig eth# 192.168.x.x up
|
NOTE: The example above is specific to static addresses. |
To use the Red Hat kudzu hardware detection utility, a number of files containing PCI vendor and device information need to be patched with information on the BCM57XX series adapters. The most recent Red Hat distributions are included. Apply the appropriate patch by running the patch command. For example, on Red Hat Enterprise Linux 2.1 and 3.0 for i386, apply the patch by doing the following:
patch -N -p1 -d /usr < pci-rh80-i386.patch
Run kudzu:
kudzu
Patch files are included for patching the driver into some of the latest 2.4.x kernel source trees. This step is optional and should only be done by users familiar with configuring and building the kernel. The patch modifies the original kernel's source code.
To patch the driver into the kernel
1. Select the patch file that matches your kernel and apply the patch:
patch -p1 -d kernel_src_root < bcm5700-version-2.4.x.patch
where version is the version of the BCM57XX driver and 2.4.x is the version of the kernel to patch (for example, 2.4.10).
|
NOTE: kernel_src_root is usually /usr/src/linux or /usr/src/linux-2.4.x. |
cd kernel_src_root
make menuconfig
make dep
make clean
....
....
For network installations through NFS, FTP, or HTTP (using a network boot disk or PXE), a driver disk that contains the BCM57XX driver may be needed. The driver disk images for the most recent Red Hat versions are included. Boot drivers for other Linux versions can be compiled by modifying the Makefile and the make environment. Further information is available from the Red Hat website, http://www.redhat.com.
To create the driver disk, select the appropriate image file and type the following:
dd if=dd.img of=/dev/fd0H1440
Unloading/Removing the Driver from an RPM Installation
Removing the Driver from a TAR Installation
To unload the driver, use ifconfig to bring down all eth# interfaces opened by the driver, and then type the following:
rmmod bcm5700
If the driver was installed using rpm, do the following to remove it:
rpm -e bcm5700
If the driver was installed using make install from the tar file, the bcm5700.o driver file has to be manually deleted from the operating system. See Installing the Source RPM Package for the location of the installed driver.
Values for optional properties can be set for the Broadcom NetXtreme 57XX Gigabit Ethernet adapter using command line arguments in the insmod command. Typically, the values for the properties are set in the /etc/modules.conf file (see the man page for the modules.conf file). These properties take the following form:
property=value[,value,...]
|
NOTES:
|
The line_speed property selects the line speed of the link. This property is used together with the full_duplex and auto_speed properties to select the speed and duplex operation of the link and the setting of auto-negotiation.
| 0 | Auto-negotiates for the highest speed supported by link partner (default). |
| 10 | Sets the speed at 10 Mbit/s. |
| 100 | Sets the speed at 100 Mbit/s. |
| 1000 | Sets the speed at 1000 Mbit/s. |
If line_speed is set to 10, 100, or 1000 Mbit/s, the adapter auto-negotiates for the selected speed (and selected duplex mode) if auto_speed is set to 1. If auto_speed is set to 0, the selected speed and duplex mode are set without auto-negotiation. The 1000 Mbit/s speed must be negotiated for copper twisted-pair links.
The auto_speed property enables or disables auto-negotiation.
| 0 | Disables auto-negotiation. |
| 1 | Enables auto-negotiation (default). |
This property is ignored and is assumed to be 1 if line_speed is set to 0.
The full_duplex property selects the duplex mode of the link. This property is used together with line_speed to select the speed and duplex mode of the link. This property is ignored if line_speed is 0.
| 0 | Sets the mode to Half-Duplex. |
| 1 | Sets the mode to Full-Duplex (default). |
The rx_flow_control property enables or disables receiving flow control (PAUSE) frames. This property is used together with auto_flow_control.
| 0 | Disables receiving PAUSE frames. |
| 1 | Enables receiving PAUSE frames auto_flow_control is set to 0, or advertises PAUSE frame receive if auto_flow_control is set to 1 (default). |
The tx_flow_control property enables or disables transmitting flow control (PAUSE) frames. This property is used together with auto_flow_control.
| 0 | Disables the transmission of PAUSE frames. |
| 1 | Enables the transmission of PAUSE frames if auto_flow_control is set to 0, or advertises PAUSE frame transmit if auto_flow_control is set to 1 (default). |
The auto_flow_control property enables or disables the auto-negotiation of flow control. This property is used together with rx_flow_control and tx_flow_control to determine the advertised flow control capability.
| 0 | Disables flow control auto-negotiation. |
| 1 | Enables flow control auto-negotiation with the capability specified in rx_flow_control and tx_flow_control (only valid if line_speed is set to 0 or auto_speed is set to 1) (default). |
The mtu property enables jumbo frames up to the specified MTU size. The valid range for this property is 1500 to 9000. The default value is 1500, which is standard Ethernet (non-jumbo) MTU size. Note that the MTU size excludes the Ethernet header size of 14 bytes. The actual frame size is MTU size + 14 bytes. Jumbo MTU sizes are not supported on BCM5705/BCM5721/BCM5751 chips.
The MTU size can also be changed using ifconfig after the driver is loaded. Refer to the ifconfig man page for details.
The tx_checksum property enables or disables hardware transmit TCP/UDP checksum.
| 0 | Disables hardware transmit TCP/UDP checksum. |
| 1 | Enables hardware transmit TCP/UDP checksum (default). |
The rx_checksum property enables or disables hardware receive TCP/UDP checksum.
| 0 | Disables hardware receive TCP/UDP checksum. |
| 1 | Enables hardware receive TCP/UDP checksum (default). |
The scatter_gather property enables or disables scatter/gather and 64-bit DMA on IA32. This option is only useful when running on TUX-enabled kernels or kernels with zero-copy TCP.
| 0 | Disables scatter/gather and 64-bit DMA on IA32. |
| 1 | Enables scatter/gather and 64-bit DMA on IA32 (default). |
The tx_pkt_desc_cnt property configures the number of transmit descriptors. The default value is 100. The valid range of values is from 1 to 600. Depending on kernel and system architecture, the driver may require up to 268 bytes per descriptor. The driver may not be able to allocate the required amount of memory if this property is set too high. This property should not be set to a value of less than 80 if adaptive_coalesce is enabled.
The rx_std_desc_cnt property configures the number of receive descriptors for frames up to 1528 bytes. The default value is 200. The valid range of values is from 1 to 511. This property should not be set with a value less than 80 on systems with high network traffic. Setting the value of this property higher allows the adapter to buffer larger bursts of network traffic without dropping frames, especially on slower systems. Note that the driver may not be able to allocate the required amount of memory if the value of this property is set too high. This property should not be set to a value of less than 50 if adaptive_coalesce is enabled.
The rx_jumbo_desc_cnt property configures the number of receive descriptors for jumbo frames larger than 1528 bytes. The default value is 128 and the valid range of values is from 1 to 255. When jumbo frames larger than 1528 bytes are used, the value of this property should not be set lower than 60 on systems with high network traffic. Setting the value of this property higher allows the adapter to buffer larger bursts of jumbo traffic without dropping frames, especially on slower systems. Depending on kernel and system architecture, the driver may require up to 268 bytes per descriptor. Each descriptor also requires a buffer the size of a maximum jumbo frame. On systems with insufficient memory, it may be necessary to set a lower value for this property. When the maximum frame size is less than 1528 (MTU size less than 1514), this property is not used and always has a value of 0.
The adaptive_coalesce property enables or disables adaptive adjustments to the various interrupt coalescing properties. Enabling this property allows the driver to dynamically adjust the interrupt coalescing properties to achieve high throughput during heavy traffic and low latency during light traffic. The value of rx_std_desc_cnt (and rx_jumbo_desc_cnt if using Jumbo frames) should not be set less than 50, and the value of tx_pkt_desc_cnt should not be set less than 80 when this property is enabled.
| 0 | Disables adaptive adjustments to the various interrupt coalescing properties. |
| 1 | Enables adaptive adjustments to the various interrupt coalescing properties (default). |
The rx_coalesce_ticks property configures the number of 1-microsecond ticks before the adaptr generates a receive interrupt after receiving a frame. This property works in conjunction with the rx_max_coalesce_frames property. An interrupt is generated when either of these thresholds is exceeded. A 0 means this property is ignored and an interrupt is generated when the rx_max_coalesce_frames threshold is reached. The valid range of values is from 0 to 500, and the default value is 80. This property is not used and is adjusted automatically if adaptive_coalesce is set to 1.
The rx_max_coalesce_frames property configures the number of received frames before the adapter generates receive interrupt. The valid range of values is from 0 to 100, and the default value is 15. The values of both this property and rx_coalesce_ticks cannot be set to 0; otherwise, no receive interrupts are generated. The value should also be set significantly lower than the value of rx_std_desc_cnt (and rx_jumbo_desc_cnt if using jumbo frames). This property is not used and is adjusted automatically if adaptive_coalesce is set to 1.
The tx_coalesce_ticks property configures the number of 1-microsecond ticks before the adapter generates a transmit interrupt after transmitting a frame. This property works in conjunction with tx_max_coalesce_frames. An interrupt is generated when either of these thresholds is exceeded. A 0 means this property is ignored and an interrupt is generated when the tx_max_coalesce_frames threshold is reached. The valid range of values is from 0 to 500, and the default value is 200. This property is not used and is adjusted automatically if adaptive_coalesce is set to 1.
The tx_max_coalesce_frames property configures the number of transmitted frames before the adapter generates a transmit interrupt. The valid range of values is from 0 to 100, and the default value is 35. The values of both this property and tx_coalesce_ticks cannot be set to 0; otherwise, no transmit completion interrupt can be generated. This property should always be set to a value lower than the value of tx_pkt_desc_cnt. This property is not used and is adjusted automatically if adaptive_coalesce is set to 1.
The stats_coalesce_ticks property configures the number of 1-microsecond ticks between periodic statistic block DMAs. The valid range of values is from 0 to 3 600 000 000, and the default value is 1 000 000 microseconds (1 second). Setting this property to 0 disables statistics updates. This property is not used and is set to the default value if adaptive_coalesce is set to 1.
The enable_wol property enables or disables Magic Packet™ Wake on LAN when the system is shut down. Note that not all systems support Wake on LAN.
| 0 | Disables Magic Packet Wake on LAN (default). |
| 1 | Enables Magic Packet Wake on LAN. |
The enable_tso property enables or disables the TCP Segmentation Option (TSO) when you are using kernels that support it.
| 0 | Disables TSO (default). |
| 1 | Disables TSO. |
If the delay_link property is set to 1, the driver returns -EOPNOTSUPP when the SIOCGMIIREG or ETHTOOL_GLINK I/O controls are called during the first 6 seconds after driver reset. When the driver resets the adapter during ifconfig, the link is dropped and it may take several seconds for the link to come up after auto-negotiation completes. Some applications, such as ifup, may not wait long enough for the link before giving up. Setting this property to 1 may get around such problems. The default value is 0, which means that the driver always return true link states to all I/O control calls, when applicable.
If the disable_d3hot property is set to 1, the driver never puts the device in the D3Hot power state when the adapter is shut down or is suspended. If set, this property also disables the Wake on Lan setting. A rare D3Hot related problem was seen during repeated shutdown of PCI Express devices on systems running 2.6 kernels.
The following are the most common sample messages that may be logged in the /var/log/messages file. Use dmesg -nlevel to control the level at which messages appear on the console. Most systems are set to level 6 by default.
Broadcom Gigabit Ethernet Driver bcm5700 with Broadcom NIC Extension (NICE) version (date)
Driver Signon
eth#: Broadcom BCM5701 1000Base-T found at mem faff0000,
IRQ 16, node addr 0010180402d8
eth#: Broadcom BCM5701 Integrated Copper transceiver found
eth#: Scatter-gather ON, 64-bit DMA ON, Tx Checksum ON, Rx Checksum ON
NIC Detected
bcm5700: eth# NIC Link is Up, 1000 Mbps full duplex
Link Up and Speed Indication
bcm5700: eth# NIC Link is Down
Link down indication.
Detailed statistics and configuration information can be viewed in the /proc/net/nicinfo/eth#.info file