Notebook systems are becoming more like desktop systems with each generation, and many notebooks have CPU, memory, disk, and video capabilities that make them true desktop replacements. But the compact, mobile nature of notebooks requires more complex configurations in order to handle power management, mobile networking, and frequently changing hardware configurations.
Many of the topics in this chapter apply to both desktop and notebook systems (and, to a lesser extent, server systems), but become more complex in a mobile environment.
When you're on the go, you have to carry your power with you. Notebook power management therefore receives a lot more attention than desktop power management, even though attention to these issues on the desktop can result in significant savings in electrical costs, system wear, and heat production.
For many years, power-management interfaces have been proprietary and required custom software supplied by the hardware vendor in order to function well (even when they purportedly adhered to industry standards). The situation is slowly improving, and Fedora contains good tools for power management on well-behaved systems.
Fedora uses the Advanced Configuation and Power Interface (ACPI) specification to monitor and manage the current power configuration. This approach requires support from the motherboard and CPU as well as the operating system; fortunately, most systems built in the last decade have some level of ACPI support, though many BIOS implementations are nonstandard.
Fedora's main power-management tool is gnome-power-manager. You can access the gnome-power-manager configuration window using the menu option System→Preferences→More Preferences→Power Management.
Fedora Core does not include the KDE ACPI modules. However, you can use gnome-power-manager in KDE by starting it manually: press Alt-F2 or open a terminal, and type:
$ gnome-power-manager
A second power-management icon will appear in the KDE panel. You can disable KDE's icon through the Control Center menu option; go to Power Control→Laptop Battery and deselect the checkbox labeled "Show battery monitor."
gnome-power-manager will automatically be started next time you enter KDE.
The same program controls the power-management icon in the panel bar, which may or may not appear depending on the system configuration (it will usually appear by default on a system that has a battery, including most notebooks, but will not appear by default on a system with no battery). The symbol used for the icon will change according to the power supply: it will show a battery when discharging the battery, a battery plus a power cord when charging the battery, or just a power cord on a system with no battery. You can access the gnome-power-manager configuration window by right-clicking on the icon and selecting Preferences.
Regardless of how you access the configuration window, you will see the display shown in Figure 3-1 .
Figure 3-1. Power Manager Preferences window
This window contains three tabs: one that configures power management when the system is running on AC or charging the battery, one that configures power management when the system is running on (and discharging) the battery, and one for general power-management settings.
The controls on first two tabs are almost identical:
Sleep
Configures the conditions under which the display and the whole system is put to sleep. For the display, "sleep" is defined as a low-power mode; for the system, the definition of "sleep" is taken from the setting on the General tab. You can set either to a value between 11 and 60 minutes (in one-minute intervals), or you can disable sleep by moving the slider all the way to the right (Never).
When laptop lid is closed
Specifies the action to be taken when a lid closure event is signaled by ACPI. The options include: do nothing, which causes the system to continue to use full power, to be available the moment you open the screen; blank the screen, which is similar to doing nothing but will use slightly less power and delay system startup for a few seconds; or suspend or hibernate the system.
Configuring the screen to blank without suspending or hibernating the system whenever the lid is closed is ideal for listening to digital music.
Prefer power savings over performance
If you are performing a task that is not very demanding, such as editing a document, instruct gnome-power-manager to maximize the battery life (or, on AC, reduce power consumption) by selecting the checkbox. On the other hand, if you need optimal performance without regard to the power consumptionsuch as when you are giving an important presentationleave the checkbox for this option unselected.
When battery power critical
This button appears only on the "Running on Battery" tab, and it configures the action to be taken if the battery runs down to the point that it will power the system for only a few minutes. The options are to do nothing, to suspend or hibernate, or to shut down the system. For most users, doing nothing is a poor choice because the system will abruptly shut off within a few minutes; suspend continues to use power, although at a much lower rate than usual, so the battery will ultimately go dead in due course (causing the loss of any data in memory). Shutdown in an option, but the shutdown procedure itself may take a few minutes and uses a fair bit of power. Therefore, the best choice (if your system supports it) is to hibernate the system when the battery power becomes critical. Hibernation takes less than a minute to complete and even less time to resume, yet it uses no power during the time that the system is in hibernation.
Depending on the hardware installed on your system, you may see different combinations of controls on the "Running on AC" and "Running on Battery" tabs.
The third tab in this window controls general power-management options, as shown in Figure 3-2 .
Figure 3-2. General tab on Power Management Preferences window
The options in this window control how the system is put to sleep when the inactivity period set in the other tabs is reached (do nothing, suspend, or hibernate). It also configures the times that the the power-management icon is displayed in the Notification Area; the default is to display it only when the battery is present, so if you want to easily access the power-management controls on an AC-only system, set this to "Always display icon."
It's important to understand the difference between the suspend and hibernate options: in both cases, the system is effectively off, but in suspend mode the memory is still operating (and consuming power), while in hibernate mode the memory is transferred to swapspace. This means that suspend state will eventually run your battery down, and then the information in RAM will be lostbut until that time, the system will be able to quickly resume its operation. On the other hand, your system can be in hibernate state forever without any power and still resume successfully, but it will take a little longer to do so.
In order to hibernate or suspend successfully, the drivers for all of the devices attached to the system must be able to save and restore the device state. There's no simple way to predict if this will be the case, so it's best to try suspend and resume operations when you have nothing critical happening on the system. For hibernation, it's also necessary to have a swap partition instead of (or in addition to) swapspace on a logical volume.
To manually provoke a suspend or hibernate operation, right-click on the power-management icon in the status bar and use the menu options that appear. To resume from a suspend, use the wake-up key defined for your system (you may have to consult the system documentation or experiment to find this). To resume from hibernation, turn your system power on with the power button, then let it start a normal boot cycle. The kernel will recognize the hibernation state and attempt to resume; if this fails, a normal boot will take place (but you may lose changes to any files that were opened when you hibernated your computer, so it's a good idea to save changes before you hibernate).
A password is required to unlock the screen after a resume if passwords are configured in the screensaver (which is the default).
Resuming from hibernation will fail if the kernel has been upgraded since the last boot.
If your system has a battery, gnome-power-manager provides a number of interesting graphs showing your power state. To view them, right-click on the power management icon on the panel, then select Information. The window shown in Figure 3-3 will be presented.
Figure 3-3. Power Information window
The tabs across the top provide access to the available information and graphs:
Device Information
Displays the current battery state. Clicking on More presents the battery technology, serial number, and a capacity rating showing the percentage of the original design capacity that the battery is now capable of holding (the higher the number, the better the battery condition).
Charge History
This graph shows the battery charge (as a percentage of its current maximum) over time. The graphs are automatically scaled horizontally.
Power History
Displays the battery charge and discharge rates over time, as shown in Figure 3-4 . If the battery is fully charged and the system is on external power, the discharge rate will be shown as zero, but if the battery is charging, the system will show the rate at which it is charging. When running on battery power, this graph shows the rate of discharge. Power events are marked on this graph, including transitions to and from AC power, suspend and hibernate events, lid closures, and display idle periods. You can use the information in this graph to see the impact of your usage patterns on the system's power consumption.
Figure 3-4. Power History graph
Estimated Time History
Shows the history of the calculated time to charge the battery, when you're on AC power, or to discharge the battery, when you're running only on the battery. Power events are also shown on this log.
Event Log
Displays a more verbose listing of the same power events shown on the graphs.
On an AC-only system, the graphs will all be flat lines, but power events will still be displayed.
If CMOS semiconductors were perfect, they would consume power only when they changed state. They're not perfect, so they leak energy and consume power when idle, but that's a tiny fraction of the power they consume when changing state.
The system clock is a pulse generator that controls the speed at which the CPU changes state, and therefore controls the amount of energy used by the CPU and related system components. Therefore, there is a trade-off between performance and power consumption.
Fedora can balance power usage against performance automatically according to system workload. This feature is provided by the cpuspeed service and is enabled by default.
The default parameters used by this service work well in most cases, but can be adjusted by editing the file /etc/cpuspeed.conf , which looks like this:
VMAJOR=1
VMINOR=1
# uncomment this and set to the name of your CPUFreq module
#DRIVER="powernow-k7"
# Let background (nice) processes speed up the cpu
OPTS="$OPTS -n"
# Add your favorite options here
#OPTS="$OPTS -s 0 -i 10 -r"
# uncomment and modify this to check the state of the AC adapter
#OPTS="$OPTS -a /proc/acpi/ac_adapter/*/state"
# uncomment and modify this to check the system temperature
#OPTS="$OPTS -t /proc/acpi/thermal_zone/*/temperature 75"
Usually the DRIVER, VMAJOR, and VMINOR lines should not be changed.
Uncomment the OPTS lines that contain additional options you wish to use. The pre-configured lines have these meanings:
-n
Allow processes that have been marked as low priority using the nice command to run at full speed. The default is to slow down the processor when only low-priority processes are running.
-s 0 -i 10 -r
Manages only CPU 0 ( -s 0 ), making speed change decisions once a second instead of the default of once every two seconds ( -i 10 ), and restore the original speed when cpuspeed exits ( -r ).
-a /proc/acpi/ac_adapter/*/state
Monitors the AC adapter and switches to minimum speed when AC power is removed. Adding -C will force the system to operate at maximum speed when AC power is connected.
-t /proc/acpi/thermal_zone/*/temperature 75
Watches the system temperature and slows down the CPU when the temperature exceeds 75 degrees Celsius (167 degrees Fahrenheit).
To see a list of all available options, run this command:
$ /usr/sbin/cpuspeed --help 2>&1|less
After editing /etc/cpuspeed , restart the cpuspeed service:
# service cpuspeed restart
Stopping cpuspeed: [ OK ]
Starting cpuspeed: [ OK ]
You can monitor the CPU speed by installing a monitor on your panel. Right-click on your GNOME panel, then select "Add to Panel." Choose the CPU Frequency Scaling Monitor and click Add. The icon shown in Figure 3-5 will appear; the bar graph will rise and fall as the CPU clock speed is adjusted, and hovering your mouse cursor over the display will show the current frequency setting (the box that appears below the icon in Figure 3-5).
Figure 3-5. CPU Frequency Scaling Monitor
If you are using your system through a text-mode console, you can still access the important power-management tools.
To suspend the system, execute the pm-suspend script:
$ pm-suspend
To hibernate, use the pm-hibernate script:
$ pm-hibernate
To view the battery status, access ACPI through the /proc filesystem:
$ cat /proc/acpi/battery/*/state
present: yes
capacity state: ok
charging state: charging
present rate: 3079 mA
remaining capacity: 2912 mAh
present voltage: 16273 mV
Battery details are also available from ACPI:
$ cat /proc/acpi/battery/*/info
present: yes
design capacity: 6450 mAh
last full capacity: 5154 mAh
battery technology: rechargeable
design voltage: 14800 mV
design capacity warning: 515 mAh
design capacity low: 156 mAh
capacity granularity 1: 5 mAh
capacity granularity 2: 5 mAh
model number: DELL 0017F
serial number: 14639
battery type: LION
OEM info: Sony
By dividing the last full capacity by the design capacity, you can determine the condition of the battery. In this case, the battery can presently store 5154/6450 mAh, or about 80 percent of its design capacity, indicating that it's in reasonably good condition.
To further reduce power consumption:
Dim your screen as much as the ambient lighting conditions will allow. The backlight for the screen draws a huge amount of power; one of my systems draws 31 percent more power in total when the screen is at maximum brightness than when it is set to minimum brightness.
Turn off all unneeded services, reducing CPU and disk activity.
Use a lightweight desktop environment such as Xfce instead of GNOME or KDE.
Xfce can be easily installed (see Lab 5.3, "Using Repositories"). To select your desktop environment, use the Session menu on the graphical login screen.
Unplug any unneeded external devices, including USB devices, headphones, mice, and keyboards. Each USB device can consume up to 2.5 watts of power, increasing your power consumption by as much as 20 percent.
Turn off your wireless network radio when it is not in use. Most laptops have a wireless kill switch (sometimes labeled airplane or flight mode ) that turns off the radio portion of the wireless card. This can cut your power consumption by up to 5 percent.
The ACPI specification enables a system's BIOS to supply data and program code to the operating system. The code is written in a unique, processor-independent pseudo-machine language called ACPI Machine Langauge (AML). The Linux kernel interpretively executes the AML code to access certain features of the host hardware system. You can think of the AML code as a type of device driver that is downloaded from the BIOS to the operating system.
The Linux kernel uses the ACPI data and code to collect data that is then exposed through the /proc/acpi directory. Information received through the /proc/acpi directorysuch as instructions to change the CPU frequencyis passed to the ACPI code.
Applications such as cpuspeed and gnome-power-manager monitor the ACPI information and combine it with other information (such as current process load) and then make power management decisions. These decisions cause actions to be taken by various subsystems such as the ACPI, the X Window server, storage, and loadable modules.
The Fedora power-management system is in active development. To stay informed of the latest developments, subscribe to the fedora-laptop-list (see Lab 9.1, "Participating in the Fedora Mailing Lists").
The hdparm program can be used to stop a drive immediately or to configure it to stop if it is idle for a certain length of time; this is called a spin-down (and restarting the drive is called a spin-up ). This does save some power; however, the drive is rarely idle for very long and the length of time (and amount of energy) required to spin-up the drive is significant (and hard on some hardware), so opinion is divided on whether it makes sense to use this feature.
In the case of a two-drive system where the second drive is rarely used, an idle spin-down timeout is a good idea. Configure it with the hdparm command:
# hdparm -S 6 /dev/[hs]d[a-z]
The -S option configures the amount of time that the drive must be idle before spin-down is triggered. The scheme used to encode the timeout period is a bit convoluted (it is described in detail on the manpage for hdparm ), but 0 means that spin-down is disabled, and a value from 1 to 240 sets the idle timeout in multiples of 5 seconds (5 seconds to 20 minutes). The value of 6 used here indicates a 30-second idle timeout.
Some important operations are best performed when a system is not in normal use, such as backup, software updating, and data indexing. It's not necessary to keep a system running 24x7 in order to schedule nighttime tasks.
Most modern system BIOSes contain an alarm wake feature, which causes the system to turn on at a preset time. This feature can be used with the hibernate state or a full system shutdown to automate nocturnal activities.
The BIOS configuration utility varies from system to system; consult your system or motherboard documentation.
Some BIOS versions permit the wake time to be set through ACPI. To see if this is possible on your system, enable the alarm wake feature using your BIOS configuration utility, then boot Fedora and examine the /proc/acpi/alarm file:
$ cat /proc/acpi/alarm
2007-03-17 00:00:00
Attempt to set this to a time in the future (the date portion of the time may be ignored by your system). If your system clock is maintained in UTC, be sure to also specify this time in UTC:
# echo " 2007-03-17 16:45:00" >/proc/acpi/alarm
$ cat /proc/acpi/alarm
2007-03-17 16:45:00
Shut down your system and see if it turns on at the specified time.
Regardless of whether you can set the alarm time using Linux, you can use the wake-up feature in conjunction with cron to configure automatic boot and shutdown.
Configure your system to turn on at a specified time using either the BIOS or Linux ACPI methods. Create a nocturnal script that performs the work you wish to do at boot time:
#!/bin/bash
#
# /usr/local/bin/nocturnal :: script for nighttime processing
# Place whatever commands you wish to execute at night here
/usr/local/bin/backup-scp
yum -y update
# Shut the system back off - you can use pm-hibernate here
# if it works on your system
shutdown -h now
Replace the last line with pm-hibernate if hibernation works on your system and you wish to reduce the startup time in the morning.
Edit your crontab :
# crontab -e
Add a line to execute the nocturnal script a few minutes after your preset wake-up time:
# Perform nighttime processing after the 4:30 wake-up
35 4 * * * /usr/local/bin/nocturnal
If you shut down the system at the end of your working day, it will wake up at night, perform the processing you have configured, and then shut down.
If your system supports changing the alarm time through ACPI, you can schedule multiple wake-up times: have your system start up in the middle of the night and perform the operations described above, and then have it schedule the next wake-up time before shutting down:
#!/bin/bash
#
# /usr/local/bin/nocturnal :: script for nighttime processing
# Please whatever commands you wish to execute at night here
/usr/local/bin/backup-scp
yum -y update
# Schedule another wake-up
date "+%Y-%m-%d
07:50:00 " >/proc/acpi/alarm
# Shut the system back off; you can use pm-hibernate here
# if it works on your system
shutdown -h now
This sets the next wake-up for 7:50 a.m. the same day, just in time to start work at 8 a.m. If your nocturnal processing takes place before midnight, schedule the wake-up for the following day:
# Schedule another wake-up
date +"%Y-%m-%d 07:50:00" -d tomorrow >/proc/acpi/alarm
gnome-power-manager home page: http://www.gnome.org/projects/gnome-power-manager/
ACPI Promoters' web site, including the ACPI specification: http://www.acpi.info/
"Linux ACPI-Howto, The Sequel": http://www.columbia.edu/~ariel/acpi/acpi_howto.html
The manpage for hdparm
The majority of modern computer system are connected to a network. While server and desktop systems are often configured for one network at installation time and remain plugged into that same network for weeks, months, or years, laptop systems are frequently on the move and may connect to several different networks in one day. Fortunately, Fedora provides a good set of network configuration tools that enable you to easily swing from one network to another like a digital Tarzan.
There are three ways to configure networking on Fedora. Each approach has its advantages and disadvantages:
Graphical configuration tool
The best approach for desktop and server systems that will connect to one or two networks and rarely require changes to the network configuration
NetworkManager
Excellent for laptops that will be connecting to a variety of different networks, but only compatible with certain network hardware
Network configuration commands
Good for experimentation, remote administration, and very complex configurations
Select the menu option System→Administration→Networking to access the GUI network configuration tool shown in Figure 3-6. Alternatively, you can type the command system-config-network into a shell (or use the traditional nickname for this program, neat).
Figure 3-6. Network Configuration window
To add a network connection, click on the New icon. The window in Figure 3-7 will appear, enabling you to select the connection type.
Figure 3-7. New Device Type window
Use the default Ethernet connection option for any LAN connection, including cable modem connections as well as all DSL connections made through a router or gateway device. Click Forward to proceed to the device-selection window in Figure 3-8 .
Figure 3-8. Device-selection window
All of the Ethernet devices that have been automatically detected (or previously configured manually) will be listed, with the device name in parentheses (such as eth0 ). Select the device you wish to configure and click Forward.
If the device is not in the list, select Other Ethernet Card from the bottom of the list and click Forward. The window shown in Figure 3-9 will appear; select the Adapter type (which selects the device driver to be used), the device name to be used (the default is usually OK), and any resources the card will use (this area should almost always be left blank). Click Forward.
Figure 3-9. New device-setup window
You will now see the window shown in Figure 3-10 . Select one of the two options to assign the IP address for this network connection:
Automatically obtain IP address settings
Use this option if you wish to use an IP address assigned by a system on your network (such as another computer running a DHCP server, a gateway or router device, or a cable modem). Set the adjacent protocol control to DHCP unless your network uses the older BOOTP protocol (rare).
Statically set IP addresses
Select this option to manually configure the IP address. Fill in the IP address, subnet mask, and gateway (router) address in the labeled fields.
Click Forward to proceed.
Figure 3-10. IP address configuration
Figure 3-11 shows the final confirmation window that appears. Review the information for accuracy and then click Apply.
Figure 3-11. Confirmation window
To edit an existing network device, double-click on it in the main Network Configuration window (shown earlier in Figure 3-6), or select it and click the Edit icon. Figure 3-12 shows the editing window that appears.
The editing window includes some options that are unavailable when the device is first created.
Figure 3-12. Network Configuration device-editing window
There are three tabs in this window. The first tab, General, provides fields for basic device configuration:
Nickname
Any name of your choice. This is provided only for your reference; for example, if you have two Ethernet cards, you might nickname one Internet and the other Corporate to identify the networks to which they are connected.
Activate device when computer starts
Most network devices will have this box checked, but you should leave it unchecked for unused devices and for devices that are used only in certain contextsfor example, a wireless card on a laptop that is used only at home.
Allow all users to enable and disable this device
If selected, this feature enables any user to activate or deactivate the network device without the root password.
Enable IPv6 configuration for this interface
If your network supports IP version 6 (which is rare but becoming more common), select this box.
IP address settings
These fields replicate the fields used during the initial device setup.
Changing the hostname field and then saving the network configuration will prevent you from opening any new GUI applications because the X Window System uses the hostname in the security key used to control access to the display. To correct this problem, log out and then log in again after changing the hostname, or enter this command before doing so:
$ xhost +localhost
Figure 3-13 shows the Route tab, which is used to configure network routes when there is more than one gateway (router) available to your system.
For single-gateway networksincluding most home and office networksuse the Default Gateway field on the Devices tab and do not fill in any information on the Route tab.
Figure 3-13. Routing configuration
To add a new route, click the Edit button, then enter the network to be routed as an IP network address and a netmask, and then enter the Gateway ( router) to which packets destined for that network are to be sent.
Figure 3-13 shows a new routing entry for an 8-bit subnet: the network number is 172.16.4.0, and the netmask is 255.255.255.0, which means that any packets addressed to an IP address that starts with 172.16.4 must be sent through the gateway associated with this route, which has been set to 172.16.97.200.
The gateway must be on the same subnet as the network interface device.
To edit or delete existing routes, select the route and click on the Edit or Delete buttons.
Figure 3-14 shows the third tab, labeled Hardware Device, which contains three groups of fields:
Figure 3-14. Hardware Device tab
Hardware
Selects the physical network device used for this interface.
Device alias number
Used to configure multiple IP addresses for one device. If you have one Ethernet card and wish to assign it the IP addresses 192.168.4.13 and 10.0.17.42, one IP address could be assigned to the base device ( eth0 ), and the other IP address could be assigned to a different interface device entry connected to the same hardware (which would result in a device alias, such as eth0:1 ). This field is used to set the alias number ( 1 in this example).
To create alias devices, use the same procedure that you use to create physical interfaces. The configuration tool will set the alias number for you automatically.
Bind to MAC address
This field associates this device entry with a specific physical network interface card via the MAC address (electronic serial number) of the network card. This is desirable in most cases because it prevents the network interfaces from being renumbered if network hardware is later added to the system, shifting (for example) eth0 into eth1 . However, this can cause problems if you replace a network card with a different card of the same model, or if you are using removable disks that are moved between systems (such as hard drives in removable drive trays, which are often used in testing environments and schools). If you do expect the MAC address to change, deselect this checkbox.
The main network configuration window also contains tabs for Hardware, IPSec, DNS, and the Hosts table. I'm going to focus on the DNS and Hosts tabs in this lab.
The DNS tab shown in Figure 3-15 is used to configure nameservers , computers that are configured to translate hostnames such as google.com into IP addresses. When using DHCP, this information is usually configured automatically. If your DHCP server does not supply this information or you're not using DHCP, enter up to three DNS servers using the Primary, Secondary, and Tertiary DNS fields. In the Hostname field, enter the name of the server you are configuring; this should be part of a valid domain, or localhost.localdomain if you don't have a domain.
Figure 3-15. DNS tab
If you are running a nameserver on the same computer, enter 127.0.0.1 (the loopback address that connects to the local host) as the Primary nameserver. Optionally, enter another nameserver as the Secondary DNS just in case your local nameserver is ever down.
If a nameserver times out on you frequently (a problem that is common on slow or congested links, and which causes name lookups to fail the first time and succeed the second time they are requested), enter the nameserver twice (for example, make it both the primary and secondary nameserver).
The " DNS search path" field is used to specify the domains to be searched when looking up a hostname that does not have a domain-name component. For example, if this field contained myorganization.ca fedorabook.com , then the hostname bluesky would be looked up as bluesky.myorganization.ca , and if no IP address could be found for that name, it would be looked up as bluesky.fedorabook.com .
The Hosts tab shown in Figure 3-16 is used to configure static mappings , which define the relationship between hostnames and IP addresses without using DNS. This is useful if you have only a handful of hosts and wish to refer to them by hostname without going to the trouble of configuring a DNS nameserver.
Figure 3-16. Hosts tab
To add a static mapping, click the New button, and then enter the IP address, hostname, and (optionally) any aliases or alternate hostnames for that particular IP address. In Figure 3-16 , the address 172.16.97.143 is being associated with the hostname darkday.fedorabook.com as well as the (shorter) alias of just darkday . Click OK to save the mapping.
As you would expect, the Edit and Delete buttons may be used to change or remove a selected static mapping.
Do not delete the entry for 127.0.0.1 or ::1, or your system will not work properly. Both the system's hostname and localhost must appear as the hostname or alias for this entry. If you change this entry and save the configuration, it may not be possible to open new programs on the graphical display until you log out and then log in again.
To save the network configuration that you have configured using the GUI tool, select the menu option File→Save. This will update the network configuration files with your changes but will not immediately activate those changes.
To enable or disable network interfaces, go to the Devices tab ( Figure 3-6 ), select the interface you wish to change, and click Activate or Deactivate. Alternately, you can exit from the network configuration tool and restart the Network service graphically (see Lab 4.6, "Managing and Configuring Services ") or by entering this command:
# service network restart
Shutting down interface eth0: [ OK ]
Shutting down loopback interface: [ OK ]
Bringing up loopback interface: [ OK ]
Bringing up interface eth0: [ OK ]
Network interface devices which are set to activate at boot will automatically start up when the network interface is started. Other devices may be started and stopped using the network configuration tool.
However, if the option labeled "Allow all users to enable and disable the device" has been set in the interface device configuration ( Figure 3-12 ), any user can activate or deactivate the device using the GNOME menu option Application→System Tools→Network Device Control (in KDE, System→Network Device Control).
This option may not appear on your GNOME menu; to make it available, right-click on the GNOME application menu and select Edit Menus. Figure 3-17 shows the menu editor that appears.
Figure 3-17. The GNOME menu editor
Select Applications→System Tools in the left pane, select the checkbox labeled Network Device Control in the right pane, and then click on the Close button.
When you start the Network Device Control program, either through the menu or by typing system-control-network in a shell, the window shown in Figure 3-18 will appear.
Figure 3-18. Network Device Control window
To activate or deactivate network interface devices using this program, select the interface and click on the Activate or Deactivate buttons. Click Close when you are done with the window.
You can also activate and deactivate network interfaces from the command line using the ifup and ifdown commands:
# ifconfig eth1
eth1 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:794 errors:0 dropped:0 overruns:0 frame:0
TX packets:195 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:115671 (112.9 KiB) TX bytes:19491 (19.0 KiB)
# ifup eth1
Determining IP information for eth1... done.
# ifconfig eth1
eth1 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:172.16.97.101 Bcast:172.16.97.255 Mask:255.255.255.0
inet6 addr: fe80::20c:2dff:fe00:2bdb/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:802 errors:0 dropped:0 overruns:0 frame:0
TX packets:213 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:117520 (114.7 KiB) TX bytes:22579 (22.0 KiB)
# ifdown eth1
# ifconfig eth1
eth1 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:802 errors:0 dropped:0 overruns:0 frame:0
TX packets:213 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:117520 (114.7 KiB) TX bytes:22579 (22.0 KiB)
The graphical network configuration tool supports the concept of profiles to permit easy roaming between different networks. You can configure a separate profile for each network that you use (such as home , office , and coffeeshop ).
A standard profile named Common is created when your system is installed. The Common profile is automatically selected at boot time, and it's the only profile that can define interfaces that will be started when the system boots.
To create additional profiles, use the Profile→New menu option. The dialog shown in Figure 3-19 will appear; enter the name of the new profile. You can then select a specific profile from the Profile menu, use the checkboxes in front of the network devices to configure which interfaces are active in that profile, and then select File→Save to save the configuration of that profile.
Figure 3-19. Creating a new Network Configuration profile
Once you have created at least one profile in addition to the standard Common profile, the Network Device Control window will gain a new pull-down menu that enables you to select the active profile, as shown in Figure 3-20 (compare with Figure 3-18).
Figure 3-20. The Active Profile control has been added
To change profiles, select the desired profile from the pull-down menu, and then click on the Activate button. The list of configured network interface devices will change to reflect the interfaces configured as part of the selected profile. Interfaces that are part of the profile but that cannot be activated or deactivated by ordinary users will appear in the interface list, but will be grayed out and unselectable.
The procedure for configuring an ISDN or analog telephone dial-up connection, a token-ring connection, or an xDSL (DSL, ADSL, SDSL, or IDSL) connection is identical to the procedure for configuring an Ethernet connection, except that additional (or slightly different) details are required in order to configure the connection:
Token Ring
This configuration is almost identical to Ethernet configuration, except that the configured device name starts with tr instead of eth (for example, the first token ring interface is called TR0 instead of eth0 ).
Wireless
Wireless configuration adds a window right after the Ethernet device driver selection and resource settings. This window requests the management mode (Auto, Managed [for normal communication with a permanent network], or Ad Hoc [for temporary networks of peer systems]), network name (set this to the SSID or ESSID of the network to which you are connecting), channel, transmit rate, and WEP encryption key. In most cases, entering the network name and WEP key is all that is required.
xDSL
All varieties of DSL lines require just one screen to set up. The configuration tool will ask you for the Ethernet device (such as eth0 ) that is connected to the DSL modem, the Internet service provider name (such as Bell Sympatico, which is recorded only for your reference), the account type (normal or T-Online), and the PPOE login name and password provided by your ISP.
ISDN and Modem Dial-Up Connections
Although ISDN and analog modems use very different technology, the configuration process is similar. First, you'll need to enter the device details. For an ISDN line, this is the ISDN adapter type and resources (usually left blank), as well as the D-channel type (leave this set to Euro ISDN in all countries except Germany, where it should be set to 1TR6). For a modem, this includes the modem device, baud rate, and flow control (these settings are usually all left at their defaults), plus volume and touch-tone settings (adjust these according to your preferences). For both types of dial-up connections, the next window asks you to select your Internet service provider and enter the phone number, username, and password that were provided by your ISP; the last window asks for the encapsulation mode (usually Sync PPP) and IP settings (whether to obtain these from the provider or manually configure them).
The GNOME NetworkManager facility provides an alternative to the use of profiles. NetworkManager is a service that automatically detects network configuration changes and determines available options. A small GUI application permits you to quickly select the network option you wish to use. It's ideal for laptop users because it permits rapid switching between multiple wired and wireless networks.
At this time, NetworkManager is a work in progress. When it works, it works well, but some hardware that works fine with other configuration techniques does not work at all with NetworkManager. It is proposed that NetworkManager will eventually become the configuration tool of choice.
The NetworkManager service is enabled in the same way as every other service (see Lab 4.6, "Managing and Configuring Services "). You can configure it to start at the next boot by choosing the menu option System→Administration→Services, and then selecting the checkboxes labeled NetworkManger (the related NetworkManagerDispatcher service is not usually required); to start the services immediately, select each of them in turn and then click the Start button.
From the command line, you can start the service immediately using the service command, and you can configure it to start at boot time using chkconfig :
# service NetworkManager start
Setting network parameters...
Starting NetworkManager daemon: [ OK ]
# chkconfig NetworkManager on
Notice the nonstandard use of capital letters in the name NetworkManager.
As soon as the NetworkManager starts, an icon should appear in your panel's notification area, which is shown in Figure 3-21 . This icon will indicate the type of IP connection currently in use. To switch to a different connection, click on the icon; a list of available connections will appear (including all broadcast local network names if you have a wireless card, plus a little signal-strength bar graph for each wireless network). Click on the network you wish to connect to, and NetworkManager will attempt to make the connection. You will be prompted to enter WEP or WPA encryption keys if necessary.
Figure 3-21. NetworkManager icon (left), showing that a wired Ethernet connection is active
NetworkManager will establish connections with wired networks automatically, but it will not automatically connect to wireless networks to which you have not previously connected because they may belong to your neighbor, or to companies that you are passing if you are in a vehicle. It will also detect the loss of network connectivity when a wireless signal is no longer usable or a network cable is unplugged.
You can also view the NetworkManager status from the command line by using nm-tool :
# nm-tool
NetworkManager Tool
State: connected
- Device: eth0 ----------------------------------------------------------------
NM Path: /org/freedesktop/NetworkManager/Devices/eth0
Type: Wired
Driver: b44
Active: yes
HW Address: 00:0D:56:33:D7:18
Capabilities:
Supported: yes
Carrier Detect: yes
Speed: 100 Mb/s
Wired Settings
Hardware Link: yes
IP Settings:
IP Address: 172.16.97.100
Subnet Mask: 255.255.255.0
Broadcast: 172.16.97.255
Gateway: 172.16.97.254
Primary DNS: 24.153.23.66
Secondary DNS: 24.153.22.67
The GUI network configuration tool and NetworkManager both work well for desktop users, but when you're logged in to a server that is a few time zones away or need to make a fast change, it's useful to be able to configure networking from the command line.
The main interface configuration command is ifconfig (for interface configuration ). Executed by itself, it displays the basic configuration of active interfaces:
$ /sbin/ifconfig
eth0 Link encap:Ethernet HWaddr 00:0D:56:33:D7:18
inet addr:172.16.97.100 Bcast:172.16.97.255 Mask:255.255.255.0
inet6 addr: fe80::20d:56ff:fe33:d718/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:289 errors:0 dropped:0 overruns:0 frame:0
TX packets:228 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:45844 (44.7 KiB) TX bytes:27193 (26.5 KiB)
Interrupt:177
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:2258 errors:0 dropped:0 overruns:0 frame:0
TX packets:2258 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:2884024 (2.7 MiB) TX bytes:2884024 (2.7 MiB)
The two interfaces displayed here are eth0, the first Ethernet interface, and lo, the loopback interface used when a client and a server that are both on the local machine need to communicate. For each interface, the information displayed includes the IP version 4 address ( inet addr ), IP version 6 address ( inet6 addr ), netmask ( Mask ), status flags (such as UP and RUNNING ), and transmit, receive, and error statistics.
You can narrow down the report to a single interface by specifying that interface name as an argument:
$ /sbin/ifconfig eth0
eth0 Link encap:Ethernet HWaddr 00:0D:56:33:D7:18
inet addr:172.16.97.100 Bcast:172.16.97.255 Mask:255.255.255.0
inet6 addr: fe80::20d:56ff:fe33:d718/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:331 errors:0 dropped:0 overruns:0 frame:0
TX packets:261 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:49667 (48.5 KiB) TX bytes:32047 (31.2 KiB)
Interrupt:177
To see both active and inactive interfaces, use the -a option:
$ /sbin/ifconfig -a
eth0 Link encap:Ethernet HWaddr 00:0D:56:33:D7:18
...(Lines snipped)...
lo Link encap:Local Loopback
...(Lines snipped)...
sit0 Link encap:IPv6-in-IPv4
NOARP MTU:1480 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
In this case, there are two interfaces here that didn't show up when ifconfig was run without the -a argument: sit0 , used for IPv6 tunneling, and wlan0 , a Wi-Fi (802.11a/b/g) interface.
ifconfig is also used to configure interfaces. The wlan0 interface can be given an IP address, broadcast address, netmask, and state ( up or down ):
# ifconfig wlan0 up 192.168.9.37 netmask 255.255.255.0 broadcast 192.168.9.255
# ifconfig wlan0
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:192.168.9.37 Bcast:192.168.9.255 Mask:255.255.255.0
inet6 addr: fe80::20c:2dff:fe00:2bdb/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1 errors:0 dropped:0 overruns:0 frame:0
TX packets:18 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:268 (268.0 b) TX bytes:2922 (2.8 KiB)
In almost all cases, the broadcast address can be left out because it can be determined from the IP address and netmask. The netmask can also be omitted if it is the default value for the network class indicated by the IP address. Table 3-1 lists the standard netmasks.
Although the use of network classes has been made obsolete by the introduction and widespread use of classless interdomain routing (CIDR), the network class values are still used to determine the default netmask. This generates the correct value for most private networks.
Table 3-1. Netmasks by IP address class (not including Multicast addresses)
| First octet of IP address | Example | Network class | Netmask | Number of IP addresses in subnet |
|---|---|---|---|---|
| 0127 | 3.15.97.4 | A | 255.0.0.0 | 16,777,216 |
| 128191 | 132.2.2.9 | B | 255.255.0.0 | 65,536 |
| 192255 | 204.99.3.8 | C | 255.255.255.0 | 256 |
The up argument is also unnecessary if an IP address is being specified.
The previous command can therefore be written much more simply:
# ifconfig wlan0 192.168.9.37
# ifconfig wlan0
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:192.168.9.37 Bcast:192.168.9.255 Mask:255.255.255.0
inet6 addr: fe80::20c:2dff:fe00:2bdb/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1 errors:0 dropped:0 overruns:0 frame:0
TX packets:18 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:268 (268.0 b) TX bytes:2922 (2.8 KiB)
For wireless interfaces, there are two additional commands that are useful for configuration: iwconfig , which sets wireless parameterssuch as the channel, encryption, and ESSIDand iwlist , which can be used to scan for available networks.
iwconfig will display current settings if no parameters are given:
# iwconfig
lo no wireless extensions.
eth0 no wireless extensions.
sit0 no wireless extensions.
wlan0 IEEE 802.11b ESSID:"" Nickname:"zd1201"
Mode:Managed Channel:6 Access Point: 00:90:4C:7E:00:29
Bit Rate:11 Mb/s
Retry:off RTS thr:off Fragment thr:off
Encryption key:off
Power Management:off
Link Quality:0/128 Signal level=60/128 Noise level:0/128
Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
Tx excessive retries:0 Invalid misc:0 Missed beacon:0
Here you can see that this interface has associated with an access point with a MAC address of 00:90:4C:7E:00:29, and which is apparently an 802.11b network (11 Mb/s).
To scan for available networks, use iwlist with the scan argument:
# iwlist scan
lo Interface doesn't support scanning.
eth0 Interface doesn't support scanning.
sit0 Interface doesn't support scanning.
wlan0 Scan completed :
Cell 01 - Address: 00:90:4C:7E:00:29
ESSID:"fedorabook"
Mode:Master
Channel:11
Bit Rates:1 Mb/s
Bit Rates:2 Mb/s
Bit Rates:5.5 Mb/s
Bit Rates:11 Mb/s
Encryption key:off
Quality=60/128 Signal level=-76 dBm Noise level=-100 dBm
Cell 02 - Address: 00:87:29:13:c0:71
ESSID:"tylers"
Mode:Master
Channel:2
Bit Rates:1 Mb/s
Bit Rates:2 Mb/s
Bit Rates:5.5 Mb/s
Bit Rates:11 Mb/s
Bit Rates:18 Mb/s
Bit Rates:24 Mb/s
Bit Rates:36 Mb/s
Bit Rates:54 Mb/s
Encryption key:on
Quality=59/128 Signal level=-76 dBm Noise level=-99 dBm
To select which network wlan0 associates with, set the extended service set identifier (ESSID) of the interface:
# iwconfig wlan0 essid fedorabook
# iwconfig wlan0
wlan0 IEEE 802.11b ESSID:"fedorabook" Nickname:"zd1201"
Mode:Managed Channel:6 Access Point: 00:90:4C:7E:00:29
Bit Rate:11 Mb/s
Retry:off RTS thr:off Fragment thr:off
Encryption key:off
Power Management:off
Link Quality:0/128 Signal level=76/128 Noise level:0/128
Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
Tx excessive retries:0 Invalid misc:0 Missed beacon:0
To disable an interface, turn it down using ifconfig :
# ifconfig wlan0 down
# ifconfig wlan0
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:192.168.9.37 Bcast:192.168.9.37 Mask:255.255.255.0
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:476 errors:0 dropped:0 overruns:0 frame:0
TX packets:18 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:84965 (82.9 KiB) TX bytes:2922 (2.8 KiB)
Notice that the ifconfig display no longer shows the UP flag for the interface.
The netmask is used to determine which computers are on the local network, and which ones are remote and must therefore be reached through a gateway or router. These rules create the default routing table, which can be displayed with route :
# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
172.16.97.0 * 255.255.255.0 U 0 0 0 eth0
default 172.16.97.254 0.0.0.0 UG 0 0 0 eth0
Notice that two routes have been configured. The first one states that local machines (those with IP addresses starting with 172.16.97) can be reached directly on the local network (gateway * and no G in the Flags column), and the second entry states that packets destined to any other IP address are to be sent through the router 172.16.97.254 (which is on the local network and therefore directly reachable).
The default route can be removed and added back in, pointing to a different gateway/router:
# route delete default
# route add default gw 172.16.97.253
# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
172.16.97.0 * 255.255.255.0 U 0 0 0 eth0
default 172.16.97.253 0.0.0.0 UG 0 0 0 eth0
When the wlan0 interface is configured, a new route is added for hosts directly accessible through that interface:
# ifconfig wlan0 192.168.9.37
# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
172.16.97.0 * 255.255.255.0 U 0 0 0 eth0
192.168.9.0 * 255.255.255.0 U 0 0 0 wlan0
default 172.16.97.253 0.0.0.0 UG 0 0 0 eth0
If other networks are available through additional gateways, these can be configured by using route with the -net and netmask arguments. For example, if your corporate network 10.x.x.x in London were accessible through the gateway 192.168.9.1 on your wireless network, you could configure the route with this command:
# route add -net 10.0.0.0 netmask 255.0.0.0 gw 192.168.9.1
# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
172.16.97.0 * 255.255.255.0 U 0 0 0 eth0
192.168.9.0 * 255.255.255.0 U 0 0 0 wlan0
10.0.0.0 192.168.9.1 255.0.0.0 UG 0 0 0 wlan0
default 172.16.97.253 0.0.0.0 UG 0 0 0 eth0
DNS resolution is controlled by the file /etc/resolv.conf , which looks something like this:
search fedorabook.com oreilly.com
nameserver 127.0.0.1
nameserver 216.183.93.224
There are three common option keywords used in this file:
search
A space- or tab-delimited list of domains to be searched when attempting to resolve a hostname without a domain component. In this example, if the DNS resolver were given the hostname bluesky , it would attempt to resolve the hostname bluesky.fedorabook.com , and if that failed, it would attempt to resolve bluesky.oreilly.com . There is a limit of six domains in the search list.
domain
A rarely used alternative to search that can specify only one domain. The domain and search options are mutually exclusive and cannot be used at the same time.
nameserver
The IP address of a nameserver available to resolve DNS queries. Listing multiple nameservers provides redundancy in case one of the servers is unavailable. In this example, the address for localhost (this computer) is given first, with a second nameserver entry providing the IP address of a remote nameserver as backup.
If the DNS settings are configured by DHCP, this file is overwritten automatically with the values provided by the DNS server. In that case, an additional comment line will appear at the top of the file:
; generated by /sbin/dhclient-script
To change the DNS configuration, simply edit this file with a text editor, adding or removing domains in the search line or adding or removing nameserver lines as necessary.
The /etc/hosts file contains a list of IP and hostname mappings. As initially set up by Anaconda (the Fedora installation system), the file will look like this:
# Do not remove the following line, or various programs
# that require network functionality will fail.
::1 bluesky.fedorabook.com localhost
The one entry in this file associates the system's name and the localhost alias with the loopback device (which may be expressed as ::1 in IPv6 notation, or 127.0.0.1 in IPv4 notation). This entry must exist in the file, or many system services will fail to operate.
You can add additional entries to /etc/hosts if you want to refer to local computers by name but don't want to go through the effort of setting up DNS (see Lab 7.3, "Configuring a Domain Name Server"). Simply place the IP address at the start of the line and then list the names and aliases for that host, separated by spaces or tabs:
# Do not remove the following line, or various programs
# that require network functionality will fail.
::1 bluesky.fedorabook.com localhost
172.16.97.60 darkday.fedorabook.com darkday frank
172.16.97.73 accounting.fedorabook.com accounting susan
172.16.97.207 samba.fedorabook.com
To change the system's hostname, edit the /etc/hosts file and change the entry for the loopback line (do not remove the localhost alias):
# Do not remove the following line, or various programs
# that require network functionality will fail.
::1 beige.fedorabook.com localhost
Then edit the HOSTNAME entry in /etc/sysconfig/network :
NETWORKING=yes
NETWORKING_IPV6=yes
HOSTNAME=beige.fedorabook.com
The change will take effect next time you boot. To make the change take effect immediately, use the hostname command:
# hostname beige.fedorabook.com
# hostname
beige.fedorabook.com
Fedora Core provides the dhclient program to configure network interfaces based on information received from Dynamic Host Configuration Protocol (DHCP) servers. Simply run this program as root , specifying the interface(s) that you wish to configure:
# dhclient wlan0
# ifconfig wlan0
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:10.144.12.160 Bcast:10.144.255.255 Mask:255.255.0.0
inet6 addr: fe80::20c:2dff:fe00:2bdb/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:3 errors:0 dropped:0 overruns:0 frame:0
TX packets:18 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1222 (1.1 KiB) TX bytes:3442 (3.3 KiB)
Since information supplied by a DHCP server is considered a lease that expires after a preset time, dhclient continues to run in the background so that it can renew the lease when necessary. If you move the machine to a new network and attempt to run dhclient again, the existing background process will be detected, and the new copy of dhclient will exit immediately without obtaining a new network configuration. To work around this problem, kill the background copy of dhclient before running it for the second time:
# killall dhclient
# dhclient wlan0
dhclient may be running even if you didn't start it manually, since boot-time network activation or activation through system-config-network or system-control-network may have launched it.
Fedora's distribution policies do not permit the inclusion of binary software without source code, and that includes firmware. Unfortunately, some very popular wireless network cards require firmware for which the vendor will not release source code.
The most common wireless adapter family affected by firmware issues is the Intel Pro Wireless (IPW) seriesoften integrated into systems under the Centrino moniker, but also sold as add-on units with Mini-PCI, CardBus, or USB interfaces.
If you find that your wireless network card is not working, it is possible that a driver is present, but the firmware file is not. Use grep to search the system logfile for messages related to firmware:
# grep firmware /var/log/messages
Jun 29 04:11:57 beige kernel: usb 2-1: Failed to load zd1201.fw firmware file!
Jun 29 04:11:57 beige kernel: usb 2-1: Make sure the hotplug firmware loader is installed.
Jun 29 04:11:57 beige kernel: usb 2-1: zd1201 firmware upload failed: -2
Jun 29 04:11:57 beige firmware_helper[14394]: Loading of /lib/firmware/zd1201.fw for usb driver failed: No such file or directory
These messages clearly show that the system attempted to load firmware for a USB wireless adapter but failed because the firmware file was not found ( No such file or directory ).
To find more information, view the /var/log/messages file using a text editor or the less program, and search for the date and time identified by the previous grep command:
# less /var/log/messages
...(Lines skipped)...
Jun 29 04:11:57 beige kernel: usb 2-1: new full speed USB device using uhci_hcd and address 5
Jun 29 04:11:57 beige kernel: usb 2-1: configuration #1 chosen from 1 choice
Jun 29 04:11:57 beige kernel: usb 2-1: Failed to load zd1201.fw firmware file!
Jun 29 04:11:57 beige kernel: usb 2-1: Make sure the hotplug firmware loader is installed.
Jun 29 04:11:57 beige kernel: usb 2-1: Goto http://linux-lc100020.sourceforge.net for more info
Jun 29 04:11:57 beige kernel: usb 2-1: zd1201 firmware upload failed: -2
Notice the message directing you to the driver web site. Visit that web site and download the firmware file provided (in this case, the file was named zd1201-0.14-fw.tar.gz , which was downloaded to the /tmp directory through a web browser). The next step is to unpack this file and then install the firmware by copying the *.fw files to /lib/firmware :
# cd /tmp
# tar xvzf zd1201-0.14-fw.tar.gz
zd1201-0.14-fw/
zd1201-0.14-fw/zd1201.fw
zd1201-0.14-fw/README
zd1201-0.14-fw/makefile
zd1201-0.14-fw/zd1201-ap.fw
# cd zd1201-0.14-fw
# cp *.fw /lib/firmware
You can now use the wireless device after resetting the device driver, which you can do by physically disconnecting and reconnecting the adapter (if it is a removable device, such as a USB or CardBus adapter), rebooting the system, or using modprobe to remove and then reload the device driver.
In this case, the driver name is zd1201 , so the driver can be reloaded with these commands:
# modprobe -r zd1201
# modprobe zd1201
After pausing for a moment to permit the interface to be configured, you can view /var/log/ messages and the output of ifconfig to see whether the driver loaded successfully and brought up the interface:
# tail - 50 /var/log/messages
...(Lines snipped)...
Jun 29 04:25:58 beige kernel: usbcore: deregistering driver zd1201
Jun 29 04:26:04 beige kernel: usb 2-1: wlan0: ZD1201 USB Wireless interface
Jun 29 04:26:04 beige kernel: usbcore: registered new driver zd1201
Jun 29 04:26:05 beige dhclient: DHCPREQUEST on eth1 to 255.255.255.255 port 67
Jun 29 04:26:10 beige dhclient: DHCPREQUEST on eth1 to 255.255.255.255 port 67
Jun 29 04:26:11 beige dhclient: DHCPACK from 172.16.97.254
Jun 29 04:26:11 beige NET[15776]: /sbin/dhclient-script : updated /etc/resolv.conf
Jun 29 04:26:11 beige dhclient: bound to 172.16.97.101 -- renewal in 39113 seconds.
# ifconfig
eth0 Link encap:Ethernet HWaddr 00:0D:56:33:D7:18
inet addr:172.16.97.100 Bcast:172.16.97.255 Mask:255.255.255.0
inet6 addr: fe80::20d:56ff:fe33:d718/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:18181 errors:0 dropped:0 overruns:0 frame:0
TX packets:3263 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:2561730 (2.4 MiB) TX bytes:375878 (367.0 KiB)
Interrupt:177
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:4936 errors:0 dropped:0 overruns:0 frame:0
TX packets:4936 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:2973825 (2.8 MiB) TX bytes:2973825 (2.8 MiB)
wlan0 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
inet addr:172.16.97.101 Bcast:172.16.97.255 Mask:255.255.255.0
inet6 addr: fe80::20c:2dff:fe00:2bdb/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:10 errors:0 dropped:0 overruns:0 frame:0
TX packets:13 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1026 (1.0 KiB) TX bytes:2384 (2.3 KiB)
Any IP address that is publicly accessible must be assigned by a central authority in order to prevent address collisions. For private networks, addresses can be freely assigned from pools reserved for this purpose, as listed in Table 3-2 ; these addresses are guaranteed not to conflict with any public IP addresses. Many home and small business gateway products default to assigning addresses in the 192.168.1.x range. If your computer is assigned one of these addresses, it will not normally be accessible from machines that are on different networks (unless you are using the port forwarding feature of your router or wireless access point).
Table 3-2. Private network address pools
| Address range | Available IP addresses | Treatment using default netmask |
|---|---|---|
| 10.0.x.x | 16,777,216 | One Class A network of 16,777,216 addresses |
| 172.16.x.x172.31.x.x | 1,048,576 | 16 class B networks of 65,536 addresses each |
| 192.168.x.x | 65,536 | 256 class C networks of 256 addresses each |
The Linux kernel keeps track of network devices in the form of interfaces . Each interface is assigned a unique name, such as eth0 , etH1 , eth2 , tr0 , tr1 , tr2 , and so forth. The interface name is initially assigned by the device driver and kernel but may be overridden using ifrename . Each interface has a number of flags, statistics, and configuration settings; these values are exposed through /proc/net and /sys/class/net , and are displayed and manipulated by ifconfig . Interface aliases such as eth0:1 and eth0:2 permit multiple interfaces to be assigned to one physical network device.
Packet routing is also performed by the kernel; the kernel routing table is exposed through /proc/net/route and /proc/net/ipv6_route , and is displayed and manipulated by the route command ( netstat -r displays the same information).
The DNS information stored in /etc/ resolv.conf is used by the resolver libraries loaded by applications. Since most resolvers load the contents of /etc/resolv.conf only when an application is started, it is usually necessary to restart your applications after changing the resolver configuration.
As an exception to the rule, Firefox does not need to be restarted after /etc/resolv.conf is changed.
The system-config-network tool is a Python script that manipulates the file /etc/sysconfig/network and the contents of the directory /etc/sysconfig/network-scripts (in addition to the /etc/hosts and /etc/resolv.conf files). These files are then read by the system init script to configure the boot-time network environment, and they are also used by other utilities such as ifup , ifdown , and system-control-network .
The NetworkManager service consists of a single binary, /usr/sbin/NetworkManager, which attempts to keep a network running at all times. The NetworkManagerDispatcher service can be used to launch scripts in /etc/NetworkManager/dispatcher.d as interfaces are brought up and down, but this capability is not used by Fedora Core, so you can leave this service disabled. The GUI component of NetworkManager is provided by nm-applet ( /usr/bin/nm-applet ); the two components communicate through the desktop bus (dbus) mechanism.
Wireless interface drivers provide an extended set of control and monitoring functions called the wireless extensions . There are many versions of the wireless extensions in use, and not all of the extensions are supported by each driver. The current version of the wireless extensions is version 20, and it is expected that this version will remain stable for some time. When a wireless card can be configured by the GUI or command line but not by NetworkManger, it's usually due to incomplete or out-of-date wireless extension support in the interface device.
A router passes packets from one interface to another, and Fedora is capable of doing this. /proc/sys/net/ipv4/ip_forward controls packet forwarding; writing a 1 to this path enables forwarding:
# echo 1 >/proc/sys/net/ipv4/ip_forward
Likewise, writing a 0 disables forwarding:
# echo 0 >/proc/sys/net/ipv4/ip_forward
The ifrename utility can be used to rename a network interface. The -i option specifies the old interface name, and -n sets the new name. To rename wlan0 to eth1 , for example:
# ifrename -i wlan0 -n eth1
eth1
# ifconfig wlan0
wlan0: error fetching interface information: Device not found
# ifconfig eth1
eth1 Link encap:Ethernet HWaddr 00:0C:2D:00:2B:DB
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
The GUI network configuration program ( system-config-network ) will sometimes refuse to configure interfaces that start with wlan ; after renaming the interface, you can use the GUI configuration tool to permanently set up the interface.
The manpages for ifconfig , route , netstat , iwconfig , iwlist , resolver , resolv.conf , hosts , dhclient , NetworkManager , NetworkManagerDispatcher , nm-applet , nm-tool , and modprobe .
The home pages for wireless firmware projects on SourceForge, including http://ipw2100.sf.net/ , http://ipw2200.sourceforge.net/ , http://ipw3945.sf.net/ , and http://linux-lc100020.sourceforge.net/
RFC1918, Address Allocation for Private Internets: http://tools.ietf.org/html/1918
Documentation on the files in /etc/sysconfig : /usr/share/doc/initscripts*/sysconfig.txt
Many laptops use a Synaptics TouchPad (or an Alps GlidePoint, which can use the same driver). By default, a touchpad will emulate a PS/2 mouse, so it should work fine with the default driver, but if you use the Synaptics-specific driver, you can exquisitely fine-tune the touchpad's extended features.
You will need to manually edit the X server configuration file, /etc/X11/ xorg.conf , to make two changes.
It's a good idea to get into the habit of making a backup of configuration files before modifying them, just in case something goes wrong:
# cp /etc/X11/Xorg.conf /etc/X11/Xorg.conf.backup
First, add an InputDevice line to the ServerLayout section:
Section "ServerLayout"
Identifier "Default Layout"
Screen 0 "Screen0" 0 0
InputDevice "Mouse0" "CorePointer"
InputDevice "TouchPad0" "AlwaysCore"
InputDevice "Keyboard0" "CoreKeyboard"
EndSection
Next, add a new InputDevice section (you can add this to any part of the file that is not between Section and EndSection lines):
Section "InputDevice"
Identifier "Touchpad0"
Driver "synaptics"
Option "SHMConfig" "on"
EndSection
When you restart the X server by restarting the system or pressing Ctrl-Alt-Backspace (save any work first!), the Synaptics driver will be loaded with a default configuration that will permit you to:
Click the left mouse button by tapping one finger in the middle area or by tapping the upper-left corner.
Drag with the left mouse button by tapping and then dragging one finger (touch-release-touch, then drag).
Click the middle mouse button by tapping two fingers in the middle area or by tapping the upper-right corner (this will usually perform a fast-paste of selected text).
Click the right mouse button by tapping three fingers in the middle area or by tapping the lower-right corner.
Scroll up and down by running your finger up and down the right side (if supported by your application).
Scroll left and right by running your finger across the bottom (if supported by your application). Some web browsers, such as Firefox, use this for history navigation (left for previous page, right for next page).
These default options work well for most users, but the driver is incredibly customizable. It's also one of the few X drivers that can be adjusted without restarting the X server, through the use of the synclient program.
Running synclient with the -l option will list the current driver settings:
$ synclient -l
Parameter settings:
LeftEdge = 1900
RightEdge = 5400
TopEdge = 1900
BottomEdge = 4000
FingerLow = 25
FingerHigh = 30
MaxTapTime = 180
MaxTapMove = 220
MaxDoubleTapTime = 180
ClickTime = 100
FastTaps = 0
EmulateMidButtonTime = 75
VertScrollDelta = 100
HorizScrollDelta = 100
MinSpeed = 0.09
MaxSpeed = 0.18
AccelFactor = 0.0015
EdgeMotionMinZ = 30
EdgeMotionMaxZ = 160
EdgeMotionMinSpeed = 1
EdgeMotionMaxSpeed = 400
EdgeMotionUseAlways = 0
UpDownScrolling = 1
LeftRightScrolling = 1
UpDownRepeat = 1
LeftRightRepeat = 1
ScrollButtonRepeat = 100
TouchpadOff = 0
GuestMouseOff = 0
LockedDrags = 0
RTCornerButton = 2
RBCornerButton = 3
LTCornerButton = 0
LBCornerButton = 0
TapButton1 = 1
TapButton2 = 2
TapButton3 = 3
CircularScrolling = 0
CircScrollDelta = 0.1
CircScrollTrigger = 0
CircularPad = 0
PalmDetect = 1
PalmMinWidth = 10
PalmMinZ = 200
CoastingSpeed = 0
These parameters are fully explained in the manpage for synaptics , but the most commonly altered values are described in Table 3-3 .
Table 3-3. Commonly altered Synaptics driver values
| Options | Description | Reason for change |
|---|---|---|
| LeftEdge, TopEdge, RightEdge, BottomEdge | Define the border between edge/corner and middle regions of the touchpad | Used to shrink or expand the regions used for scrolling and corner-taps. |
| PalmDetect, PalmMinWidth, PalmMinZ | Configure the touchpad to ignore broad touches | Prevents accidental touches of the user's palm from registering as touchpad events. PalmDetect enables/disables, PamMinWidth sets the minimum touch diameter that will be considered a palm, and PalmMinZ is the minimum pressure required to register a palm. |
| RTCornerButton, LTCornerButton, LBCornerButton, RBCornerButton | Define the mouse-button clicks that will be registered when the user touches the corner of the keypad | By default, the top-right corner can be tapped to produce a middle mouse click, and the bottom-right corner can be tapped to produce a right mouse click. If you are finding that these clicks are being accidentally registered, set these button values to zero (0); alternately, you can select a specific button you wish to associate with a corner tap (1=left, 2=middle, 3=right, 4=down, 5=up, 6=left, 7=right) |
| TapButton1, TapButton2, TapButton3 | Define the mouse-button clicks that will be registered when the user taps in the middle region of the touchpad with one, two, or three fingers | If you find that you accidentally tap the touchpad with more fingers than you intend to use, you can change the two- and three-finger tap buttons. To disable a tap altogether, set the appropriate value to 0. |
| VertScrollDelta, HorizScrollDelta | Define the amount of finger motion required to register a scroll event. | Set these values to adjust the scroll rates, or set them to 0 to disable scrolling altogether. Note that smaller values increase sensitivity. If you're finding that Firefox keeps moving back and forth in its History when you accidentally sweep your finger across the touchpad, try setting HorizScrollDelta to 0. |
| MaxTapTime | Defines the maximum time (in milliseconds) in which to detect a tap (which is interpreted as a left mouse click). | If tap-to-click drives you bonkers, set this to 0. Otherwise, you can use it to adjust how sensitive your touchpad is to tapping. |
To test an option value, use the synclient program:
$ synclient VertScrollDelta=10
$ synclient PalmDetect=0
If you need to find position or pressure values for these options, you can use the monitor ( -m ) option of synclient to experiment:
$ synclient -m 100
time x y z f w l r u d m multi gl gm gr gdx gdy
0.000 3277 2899 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
1.247 3687 3172 65 1 5 0 0 0 0 0 00000000 0 0 0 0 0
1.351 3402 3070 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
1.871 3926 3650 62 1 4 0 0 0 0 0 00000000 0 0 0 0 0
1.975 4337 3339 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
2.495 3932 3133 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
2.599 3816 3245 1 1 9 0 0 0 0 0 00000000 0 0 0 0 0
2.703 3810 3286 1 1 7 0 0 0 0 0 00000000 0 0 0 0 0
2.807 3923 3224 1 1 14 0 0 0 0 0 00000000 0 0 0 0 0
2.911 3923 3224 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
8.423 4018 3986 131 2 5 0 0 0 0 0 00000000 0 0 0 0 0
8.527 4104 3933 134 2 5 0 0 0 0 0 00000000 0 0 0 0 0
8.631 4653 3827 0 0 0 0 0 0 0 0 00000000 0 0 0 0 0
...(
Ctrl-C to terminate )...
In this output, the following options indicate:
x , y , and z
Position and pressure of touch.
f
Finger count.
w
Finger width.
l , r , u , d , m , and multi
Button state (most touchpads have only the l and r buttons). 0 indicates the button is up; 1 indicates the button is down.
All columns starting with g
Guest (secondary) pointing device information, such as a finger button.
Once you have tweaked the values to suit your needs, add your preferred values to the Synaptics InputDevice section of the /etc/X11/xorg.conf file, using the Option keyword:
Section "InputDevice"
Identifier "Touchpad0"
Driver "synaptics"
Option "SHMConfig" "on"
Option "VertScrollDelta" "10"
Option "PalmDetect" "0"
EndSection
These settings will then take effect when the system is restarted.
The Synaptics TouchPad (or Alps GlidePad) device is connected through a PS/2 or USB interface and contains a microcontroller, touch sensor, buttons, and interface electronics. The firmware is programmed to emulate a standard mouse unless (or until) the device driver sends it codes to switch it into native mode.
The Synaptics driver works with the microcontroller in the touchpad to translate user activity into standard mouse signals. There are only 16 types of events that are reported to the X server: button down and button up for buttons 1 through 7, and horizontal and vertical motion. All of the possible tap, press, and slide gestures are translated into combinations of these 16 events.
For example, touching the pad with one finger on the right side (X position greater than RightEdge , and Y position between TopEdge and BottomEdge ), and then moving your finger up more than VertScrollDelta units will be interpreted as a vertical scroll action, which will result in a button-down event on button 4, followed by a button-up event on button 4 (this corresponds to rotating the scroll wheel forward one click on a standard mouse).
Not all touchpad models have the hardware or firmware to provide all of the features supported by the driver; for example, most models do not have all of the possible buttons, and some lack multiple-finger or finger-width detection.
This can be done if the Synaptics device is defined in addition to a traditional mouse, as recommended in the X11 configuration shown in this lab.
Even with palm detection, some typists and touchpads register false information during typing (especially true with certain laptop case designs, which transmit case stress to the touchpad). To help in these situations, the Synaptics software includes the syndaemon program, which will automatically disable the touchpad when the keyboard is in use. To start this daemon, simply type its name as a command:
$ syndaemon
The manpages for synaptics , synclient , and syndaemon
The driver home page: http://web.telia.com/~u89404340/touchpad/
The video circuits of most laptops support two video outputs: one to the built-in LCD panel and one for external devices. The external output is not enabled by default, but you can switch between the internal and external displays using a function key.
More advanced configurations of the two outputs are supported by some video cards. For example, you can configure a single large desktop spanning both displays, or set up the external display to show a subset of what is shown on the internal display.
Output from one card to multiple monitors is supported only by some drivers. The most commonly used ones are the NVIDIA and ATI proprietary drivers, and the open source Radeon driver. All of these drivers have some limitations; for example, the NVIDIA driver, when used on a laptop, will always configure an external monitor as primary (screen 0), and the ATI drivers permit only general monitor positioning (you can specify that one monitor is to the right of another, but you can't indicate a difference in their vertical alignment).
Note that for the changes described in this section to take effect, you will need to restart the X server in one of these ways: changing to runlevel 3 and then back to 5, restarting the system, or terminating the running X server with Ctrl-Alt-Backspace (save your work first!).
First, configure and test your system using the NVIDIA proprietary driver (see Lab 5.3, "Using Repositories "), and then add the highlighted lines to the Device section of your /etc/X11/ xorg.conf file:
Section "Device"
Identifier "Card0"
Driver "nvidia"
VendorName "nVidia Corporation"
BoardName "NV34 [GeForce FX 5200]"
BusID "PCI:1:0:0"
Option "TwinView"
Option "SecondMonitorHorizSync" "31.0 - 80.0"
Option "SecondMonitorVertRefresh" "50.0 - 75.0"
Option "TwinViewXineramaInfo" "on"
Option "MetaModes" "800x600,1024x768"
Option "TwinViewOrientation" "LeftOf"
EndSection
The SecondMonitorHorizSync and SecondMonitorVertRefresh options configure the horizontal and vertical frequency ranges for the second monitor (the settings for the primary monitor are in the Monitor section of the file). The values given here are reasonable for most small monitors. The TwinViewXineramaInfo line configures the driver to inform applications about the fact that the desktop is on two screens, so that windows can be placed intelligently (avoiding dialog boxes that span both screens, for example).
The MetaModes option configures the relative sizes of the two screens; the first screen is always the external display. The TwinViewOrientation specifies the position of the second display relative to the first for the purpose of mouse movement and window positioning (possible values are Above , Below , LeftOf , or RightOf ).
For finer control over the monitor relationships, the positions of each monitor may be specified in the MetaModes line as an absolute position within the entire desktop. These values are given in the form + X + Y after each resolution; for example, "800x600+0+0,1024x768+800+0" specifies that the primary monitor is to the left of the secondary monitor and that the two monitors are aligned at the top, and "800x600+1024+168,1024x768+0+0" specifies that the primary monitor is on the right and that the monitors are aligned at the bottom (168 pixels is the difference between the two monitor heights, 768 and 600 pixels).
When using the external monitor connection for projection, it can be convenient to project a subset of what is shown on the laptop's panel. Using the NVIDIA driver, this can be configured by overlapping the two display regions using the MetaModes option. On a laptop with a 1400x1050 display, for example, you can project an 800x600 pixel subset to your audience:
Section "Device"
Identifier "Card0"
Driver "nvidia"
VendorName "nVidia Corporation"
BoardName "NV34 [GeForce FX 5200]"
BusID "PCI:1:0:0"
Option "TwinView"
Option "SecondMonitorHorizSync" "31.0 - 80.0"
Option "SecondMonitorVertRefresh" "50.0 - 75.0"
Option "TwinViewXineramaInfo" "on"
Option "MetaModes" "800x600+200+200,1400x1050+0+0"
EndSection
When you restart the X server, the GNOME panel bars will be located in the middle of the screen, because the smaller 800x600 subset display is considered "Primary." Drag the panels to the top and bottom of the laptop display. You can then start an application and position the portion of the window that you wish to display into the 800x600 pixel area that starts 200 pixels down and 200 pixels to the right of the upper-left corner of the screen.
This works well with OpenOffice.org Impress in "Normal" mode, which displays three panes, including the current image in the center and a preview of slides on the left (the panes can be rearranged if it is more convenient to see the preview on the right). Reposition the dividing line between the panes and scroll the center pane until the current slide completely fills the external display.
This configuration enables you to preview the slides using the OpenOffice.org preview plane as shown in Figure 3-22 without changing the projected slide. Once you select and then click on a slide, it becomes the current slide and is displayed both in the center plane of the LCD image and on the external projector, shown in Figure 3-23 .
Figure 3-22. LCD display, including slide preview pane (right)
Figure 3-23. Projected 800 600 subset of LCD image, showing only the current slide
First, configure and test your system using the ATI open source driver radeon , and then add the lines highlighted in bold to the Device section of your /etc/X11/ xorg.conf :
Section "Device"
Identifier "Card0"
Driver "radeon"
VendorName "ATI"
BoardName "ATI 7500"
BusID "PCI:1:0:0"
Option "MergedFB" "on"
Option "CRT2HSync" "31.0 - 80.0"
Option "CRT2VRefresh" "50.0 - 75.0"
Option "MergedXinerama" "on"
Option "MetaModes" "800x600-1024x768"
Option "CRT2Position" "RightOf"
EndSection
The MergedFB option enables dual video output. CRT2HSync and CRT2VRefresh set the horizontal and vertical frequency ranges for the second monitor (the settings for the primary monitor are in the Monitor section of the file), and the MetaModes option sets the resolution of the two displays. MergedXinerama enables sending display hints to applications, and CRT2Position sets the the position of the second display relative to the first for the purpose of mouse movement and window positioning (possible values are Above , Below , LeftOf , or RightOf ).
Restart the X server by changing to runlevel 3 and then back to 5, by restarting the system, or by terminating the running X server with Ctrl-Alt-Backspace (save your work first!). When the X server restarts, both displays should be active.
The ATI closed source driver is configured in much the same way as the other drivers. First, configure and test your system using the ATI closed source driver fglrx , and then add the lines highlighted in bold to the Device section of your /etc/X11/xorg.conf :
Section "Device"
Identifier "Card0"
Driver "fglrx"
VendorName "ATI"
BoardName "ATI 7500"
BusID "PCI:1:0:0"
Option "DesktopSetup" "0x00000200"
Option "HSync2" "31.0 - 80.0"
Option "VRefresh2" "50.0 - 75.0"
EndSection
The DesktopSetup option enables dual video output and specifies that the monitor attached to connector 1 is on the left; change this value to "0x00000201" if the monitors are reversed (it's not always apparent whether the internal or external monitor is attached to connector 1). HSync2 and VRefresh2 set the horizontal and vertical frequency ranges for the second monitor (the settings for the primary monitor are in the Monitor section of the file). The resolutions will automatically be selected from available options by the device driver.
Restart the X server by changing to runlevel 3 and then back to 5, by restarting the system, or by terminating the running X server with Ctrl-Alt-Backspace (save your work first!). When the X server restarts, the two monitors should display two halves of the desktop.
The X server contains code to manage two (or more) separate video cards, combining them into a single desktop. However, the X.org server and the configuration file layout were never really designed for multiple outputs from one card controlled by one driver; for example, there is no way to associate more than one Monitor section in the configuration file with a single Device (video card) section.
Some video card drivers have been enhanced to support multiple video outputs, and in order to fit within the X.org configuration file format, the configuration information for the second monitor is placed in the Device section.
There are two ways to use multiple monitors as part of a single display in X:
Each monitor can be given a distinct Screen number, which enables output to be sent to a specific monitor but prevents windows from spanning displays or being moved from one display to another.
Use Xinerama, which is named after the old Cinerama movie technology and that combines multiple monitors into a single large display, permitting windows to span monitors and to be moved between monitors.
All of the X.org video card drivers that support multiple video outputs use the Xinerama approach, but the Xinerama extensions used to inform applications of the underlying monitor geometry may be enabled or disabled using configuration options (with the exception of the ATI driver, which does not offer this capability). This information is particularly useful to window managers because it enables the window manager to correctly center dialogs in the middle of the monitor instead of the middle of the virtual desktop, and to make maximized windows fill a monitor instead of spanning monitors.
Very few of the other X.org video drivers support multiple video outputs. If you have another driver and want to see the options supported, look for a manpage for your driver. For example, to see the driver options for the Intel 810 adapter:
$ man i810
The manpages for the radeon driver
NVIDIA closed-source driver information from /usr/share/doc/NVIDIA_GLX-1.0/README.txt