包含在查找更多文档专项支持资源 | 以 PDF 格式下载本书 (538 KB)
Chapter 1 Configuring DevicesThis chapter describes how to use Solaris 8 Intel Platform Edition Device Configuration Assistant software, manufacturers' device configuration media, and documentation to solve configuration problems. The next chapter contains Device Reference Pages and describes how to use them to configure your Intel 32-bit processor architecture (IA) system to run in the Solaris operating environment. Identifying and Correcting ProblemsUse the Solaris 8 Intel Platform Edition Device Configuration Assistant program to identify devices and the resources each device uses. If there is a problem, you must provide device names and resource parameters so the Configuration Assistant can pass this information to the Solaris kernel. Physical Address Extension (PAE) ModeWith the release of Pentium Pro, Intel introduced a mode called PAE (physical address extension) on its advanced processors. By using PAE, Solaris Intel Platform Edition can address up to 32 Gbytes of physical memory. Individual processes are still limited to a maximum of 3.5 Gbytes of virtual address space. PAE mode enables you to run multiple instances of databases and memory-intensive applications, and to support large numbers of online users on your machine. It is best to use PCI disk controllers that support dual address cycle (DAC) in your machine because they can transfer data to and from any physical location. Other cards are limited to 4 Gbytes of physical memory; as a result, performance might slow down because the system needs to copy additional memory to transfer data. Some device drivers are not yet able to take advantage of PAE mode. PCI device drivers written by Sun have been tested on IA machines with more than 4 Gbytes of memory. Its OEM partners intend to test their machines with devices they supply on IA machines with more than 4 Gbytes of memory. In some cases, however, if you add a third-party device driver to your system, it might become unstable, and panics and data corruption might result. If your system becomes unstable and you need that driver, you must disable PAE mode.
|
|
Issue |
What to Do |
|
How a known ISA or EISA device can be recognized by the Configuration Assistant software |
The I/O port address chosen for the unrecognized device might conflict with the I/O port adress of another system device. Provide nonconflicting address information for the unrecognized device using the system BIOS, the EISA configuration utility (ECU), or the configuration programs supplied by the hardware manufacturer. The Configuration Assistant uses that information to identify that device for the Solaris environment. |
To set the configuration parameters for an ISA or EISA adapter, run the manufacturer's ECU. This utility must be run every time an ISA or EISA board is added, removed, or moved to a different bus slot. Although the function of this utility is standardized, implementations vary among manufacturers, each providing unique user interface screens or menus.
Boot DOS.
Back up the board manufacturer's EISA configuration diskette before using it to configure your hardware.
For each ISA or EISA add-in adapter to be configured, copy the EISA .cfg and .ovl configuration files from the board manufacturer's configuration diskette to the system EISA configuration diskette.
Run the ECU.
The program is called CF.EXE or CFG.EXE.
For each device, set the appropriate configuration parameters and any special operating modes.
|
Issue |
What to Do |
|
How to recover if your machine fails to autoboot |
If you have a loopback serial cable between COM1 and COM2 with autobooting enabled, use the eeprom command to set one of the following properties: eeprom com1-noprobe=true or eeprom com2-noprobe=true |
|
Issue |
What to Do |
|
How to activate Plug and Play mode |
Set the switch to Plug and Play mode, and plug the device into the system. The software will configure it automatically. |
|
Issue |
What to Do |
|
How to control the assignments of IRQs to PCI devices |
In the chipset configuration of the system, verify that an IRQ is enabled for PCI bus use. After checking IRQs used by ISA devices, assign as many available IRQs to PCI devices as possible so the PCI bus can resolve device conflicts. |
|
Issue |
What to Do |
|
How to use the system BIOS to change device settings |
Refer to the manufacturer's documentation for the method of accessing the BIOS setup for your system and the features it provides. |
|
How to set up a cachable region in system memory for American Megatrends, Inc. (AMI) BIOS |
For the best performance, make the cachable region equal to the total memory installed in the system. |
The kdmconfig program attempts to identify and configure the hardware devices required for running the Common Desktop Environment (CDE) or other windowing system. These devices include the video controller, monitor, keyboard, and pointing device. kdmconfig is automatically run during system startup and if it detects that any of the required hardware has changed, it gives you the chance to test or change the new configuration. kdmconfig can also be run from the command line.
After all the required devices have been specified, kdmconfig lets you test the configuration by putting up a simple test display. Click Yes to accept the current configuration; otherwise, click No or press any key to reconfigure.
kdmconfig might not be able to properly identify some supported hardware devices. In these instances, use kdmconfig to manually specify the devices.
An Ethernet adapter and its link partner (such as a hub, switch, or another network adapter connected via a crossover cable) must operate at the same duplex settings.
If the adapter and link partner support NWay media autonegotiation, both devices should automatically select optimal speed and duplex mode.
If NWay autonegotiation is not supported or is not configured on either the adapter or its link partner, both devices must be explicitly set to run at the same duplex mode. A device usually defaults to half-duplex operation if it can't determine the link partner's duplex capabilities.
A hub or switch that supports full-duplex operation usually has a mechanism that sets duplex mode on a per-device or per-port basis. Setting speed, duplex mode, or both in this manner usually disables NWay autonegotiation for the device or port.
A network adapter supported by the dnet device driver must have its duplex mode set in the driver's .conf file. See the device driver man page for details.
Operating speed can sometimes be set in the driver's .conf file, but NWay autonegotiation might be disabled when using this method.
A device is usually able to detect the speed (but not the duplex mode) of its link partner, even without NWay autonegotiation.
Device Reference Pages specify the supported connector type where appropriate. All network devices are assumed to work at 10 Mbps only, unless otherwise specified in the Device Reference Pages. Following are network connectors and the media they support.
|
Connector |
Supported Media |
Comments |
Speed |
|---|---|---|---|
|
RJ-45 |
10BASE-T |
Category-3 Twisted Pair cable |
10 Mbps |
|
RJ-45 |
100BASE-TX |
Category-5 Twisted Pair cable |
100 Mbps |
|
BNC |
10BASE2 |
Coax cable ("Thin" Ethernet cable) |
10 Mbps |
|
AUI |
10BASE5 |
Shielded Twisted Pair ("Thick" Ethernet cable) |
10 Mbps |
Some PCI motherboards contain DMA chipsets that are unable to support 100-Mbps Fast Ethernet. The Solaris environment does not support 100-Mbps PCI network operation on systems containing the slow chipsets. This problem affects PCI cards only.
These chipsets are known to exhibit this problem:
82430LX (Mercury)
82450GX (Orion) (A and B steppings only)
These chipsets do not exhibit this problem:
82430NX (Neptune)
82430FX (Triton)
82430HX (Triton II)
82440FX (Natoma)
82450GX (Orion) (C0 stepping and later)
In particular, PCI cards supported by the dnet and iprb drivers don't perform well on machines with the problem chipsets. You must decide whether the performance on a particular machine is adequate for the intended purpose.
If you replace your network adapter with one that uses a different network driver, before rebooting the second time, rename the /etc/hostname.olddriver0 file as follows:
# mv /etc/hostname.olddriver0 /etc/hostname.newdriver0
Any time you add, remove, or replace hardware, run the Configuration Assistant utility.
After renaming the driver, perform a reconfiguration boot for your changes to take effect:
# touch /reconfigure # reboot