©Copyright 2003 Marvell®.
All rights reserved
yukonuw.htm created 22-Oct-2003
Readme File for yukonuw.pck v6.06
Marvell Yukon Gigabit Ethernet Adapter
This file contains
The yukonuw MDI driver supports the Marvell Yukon Gigabit Ethernet Adapter
on SCO UnixWare 7.1.x and OpenUnix 8.0.
For proper software installation the following package file is required:
yukonuw.pck .
In the text below the expression <package file> refers to this file.
Driver packages downloaded from our web site are available in a compressed
format (.Z files). The downloaded package has to be uncompressed before
installation.
To uncompress the downloaded driver package, proceed as follows:
1. Go to the directory where the compressed driver package yukonuw.pck.Z
is located.
2. Uncompress the file with the following command:
uncompress yukonuw.pck.Z
The uncompressed driver package has the format yukonuw.pck and is
ready for installation.
There are two possibilities for installation:
- pkgadd and netcfg
- scoadmin
pkgadd and netcfg run from the command line, scoadmin uses the graphical
interface.
For removal of the driver, pkgrm is used.
3.1 Installation with pkgadd and netcfg
NOTE: If the Marvell network adapter is not plugged in, shutdown
the system, install the network adapter, and reboot the system
before starting netcfg.
To install the package with pkgadd and netcfg, proceed as follows:
1. Login as root.
2. Go to the directory where the driver <package file> is located.
3. Execute pkgadd with the following command:
pkgadd -d `pwd`/<package file>
4. Follow further instructions of the package installer.
5. If the Marvell network adapter is already plugged in, start netcfg
with the following command:
netcfg
6. Select "Hardware->Add new LAN adapter".
The corresponding device driver entry is listed in the configuration
menu.
7. Select the network device driver.
8. To continue with the installation, select <Continue>.
9. Follow further instructions of netcfg.
3.2 Installation with scoadmin
To install the package with scoadmin, proceed as follows:
1. Start scoadmin in graphical desktop environment.
2. Select "Software_Management->Application Installer".
3. Select "Install from: "other"".
4. In the dialog box "Folder" enter the location of the <package file>
including the complete pathname.
The package panel shows the available software packages.
5. Select the package.
6. Click the <Install> button.
7. Select "Actions->Exit".
8. Select "Networking->Network Configuration Manager".
9. Select "Hardware->Add new LAN adapter".
The corresponding device driver entry is listed in the configuration
menu.
10. Select the network device driver.
11. To continue with the installation, click <Continue>.
12. Follow further instructions.
3.3 Removal with pkgrm
To remove the driver with pkgrm, procees as follows:
1. To view the installed Marvell drivers on your system, execute the
following command in a system terminal:
pkginfo | grep yukonuw
A line similar to the following is displayed:
system <product name> <product description>
2. To remove the driver enter the following command:
pkgrm <product name>
where <product name> is the name of the Marvell network driver
package that is to be removed (see list under 1).
Parameters can be set by means of the "Network Configuration Manager".
There are two possibilities to set the paramters:
- during the installation of the driver, or
- after the installation by selecting "Hardware->Modify Hardware
Configuration" and the <Advanced> button.
NOTE: Once a parameter has been reconfigured for an existing network device
the system must be rebooted for the changes to take effect. This applies to
UnixWare 7.1.1 and 7.1.3 and OpenUnix 8.
- MAC Address override
Change the adapter's current receive address.
Default: no alternative MAC address
By default the network adapter's burnt-in factory address is used as the
station's current network address, which represents the receive address for
the protocol data.
An address may be entered partially: e.g. 00:00:5a:2B:3C:4D or even 5a:2B:3C:4D
- Auto-Negotiation
Values : sense, on, off.
Default: on
Auto-negotiation is the capability to detect the modes of operation
supported by the device at the other end of a link segment, to determine
common abilities, and to configure the network adapter for joint
operation.
- Flow Control Mode
Values : Sym, SymOrRem, LocSend, None
Default: SymOrRem
This parameter can be used to set the flow control capabilities that the
port reports during auto-negotiation.
The following modes are possible:
- Sym (Symmetric) both link partners are allowed to send PAUSE frames.
- SymOrRem (SymmetricOrRemote) both or only remote partner are allowed
to send PAUSE frames.
- LocSend (LocalSend) only local link partner is allowed to send PAUSE
frames.
- None: no link partner is allowed to send PAUSE frames.
NOTE: This parameter is ignored if auto-negotiation is set to Off.
- Duplex Capabilities
Specify the duplex capabilities of the physical port.
Values : both, full, half
Default: both
The duplex capabilities define the link mode of the media access connector
of the physical port. Possible values are: both, full, and half.
Also refer to the parameter "Auto-Negotiation" that allows the following
combinations of settings:
1. Auto Negotation: on Duplex Capabilities: both
This combination forces the port to start the auto-negotiation
process with the best result of full-duplex or half-duplex.
2. Auto Negotation: on Duplex Capabilities: full
This combination forces the port to start the auto-negotiation
process with the result of full-duplex operation mode only.
3. Auto Negotation: on Duplex Capabilities: half
This combination forces the port to start the auto-negotiation
process with the result of half-duplex operation mode only.
4. Auto Negotation: off Duplex Capabilities: both
This combination is not valid and causes the driver to start the
auto-negotiation process with the best result of full-duplex or
half-duplex.
5. Auto Negotation: off Duplex Capabilities: full
If this operation mode is specified, the corresponding MAC is forced
to work in full-duplex mode only. If the remote MAC does not support
full-duplex mode the link will fail.
In full-duplex mode there is no need for the Transmit Media Access
Management to avoid contention with other traffic on the medium. The
frame transmission may be initiated after the interframe delay
regardless of the presence of any receive activity. In full-duplex
mode the MAC sublayer does not perform carrier extension or frame
bursting.
6. Auto Negotation: off Duplex Capabilities: half
If this operation mode is specified, the corresponding MAC is forced
to work in half-duplex mode only. In half-duplex mode the Transmit
Media Access Management attempts to avoid contention with any other
traffic on the medium by monitoring the carrier sense signal
provided by the Physical Layer Signaling (PLS) component and
deferring to passing traffic.
7. Auto Negotation: sense Duplex Capabilities: <any value>
If the auto-negotation parameter is set to `sense` any value of
Duplex Capabilities is ignored. In this operation mode the
corresponding MAC will check the carrier for the best possible mode.
This depends on the capabilities of other stations in the network.
- Role
Set Role in Master-Slave-Negotiation (1000Base-T only).
Values : auto, master, slave
Default: auto
This parameter is only valid for copper adapters. For two 1000Base-T ports
to communicate, one must take the role as master (providing timing
information), while the other must be slave. Usually, this is negotiated
between the two ports during link establishment. If this fails, a port
can be forced to a specific setting with this parameter.
- Link speed
Specify the link speed of the corresponding port. This parameter is only
valid for copper adapters.
Values : auto,10,100,1000
Default: auto
- HwChecksum
Enables computation of IP and TCP checksums by the NIC. This decreases
CPU utilization. The driver supports TCP checksum offloading (pseudo IP
header computation) on the transmit path and TCP- and IP checksum on the
receive path. The enabled checksum mode depends on the capabilities of
the OS and is displayed at the system log at device initialization.
Please note that checksum offload is not supported by all NICs.
Values : on, off
Default: on
- TXD ring size (kb)
- RXD ring size (kb)
These both parameters define the transmit and receive queue size in kilo
byte. Large queue sizes may be used if data bursts from / to the network
may overrrun the receive / transmit capacity of the driver. Transmit and
receive queues buffer outgoing and incoming data.
Values : 1 2 4 8 16 32 64
Default: 8
NOTE: On systems with low RAM the device configuration may fail with the
following system log entry. If this happens the queue sizes above should
be reduced.
WARNING: <device>: pDescrMem error
WARNING: <device>: memory allocation error
WARNING: <device>: configuration command 1 failed, rc: 12
***End of Readme File***