This chapter describes the Command (PROM) Monitor programs, which control the boot environment for all Silicon Graphics workstations or servers. With the Command Monitor, you can boot and operate the CPU under controlled conditions, run the CPU in Command Monitor mode, and load programs (for example, the operating system kernel, /unix or special debugging and execution versions of the kernel).
The prom(1) reference page, accessible with the command:
man prom |
contains valuable information about the PROM monitor on your system.
PROM stands for Programmable Read-Only Memory. Most PROM chips are placed in your computer at the factory with software programmed into them that allows the CPU to boot and allows you to perform system administration and software installations. The PROMs are not part of your disk or your operating system; they are the lowest level of access available for your system. You cannot erase them or bypass them.
Since PROMs are not normally changed after the manufacture of the system, newly added features will not be present on older systems. Some systems have PROM firmware that responds to new programming when the operating system is updated. See your hardware Owner's Guide for more information on these types of PROMs.
Note that there are numerous minor differences between machines, and you should refer to your Owner's Guide and Release Notes for information specific to your machine.
Newer systems use a PROM called the ARCS prom. ARCS stands for Advanced Risc Computing Standard. This PROM provides a graphical interface and allows the use of the mouse to control booting and execution. ARCS systems allow the use of the keyboard, and the older key syntaxes have been retained for compatibility. Systems that use the ARCS prom include the Indigo R4K, Indy, Indigo2, CHALLENGE, Onyx, some Crimson systems, and others.
This chapter contains information on the following topics:
Basic instruction on entering the Command Monitor. See “How to Enter the Command (PROM) Monitor”.
A summary of the commands available through the general Command Monitor. See “Summary of Command Monitor Commands”.
How to get help while using the Command Monitor. See “Getting Help in the Command Monitor”.
Instructions for convenient use of the Command Monitor. See “Using Command Monitor Commands” and “Running the Command Monitor”.
Instructions for manipulating the Command Monitor Environment. See “The Command Monitor Environment”.
Instructions for booting programs from the Command Monitor. See “Booting a Program from the Command Monitor”.
To get into the Command Monitor on most machines, follow these steps:
Reboot the system with the reboot command, or if it is already switched off, turn it on.
On server systems without graphics capability, you see the following prompt:
Starting up the system.... To perform system maintenance instead, press <Esc> |
On systems with graphics, you see similar messages, displayed on your screen as shown in Figure 10-1.
The procedures are substantially the same for graphical or text usage, except that you need not press the Escape key, you can use your mouse cursor to click a button labeled Stop for Maintenance.
Press the <Esc> key or click the button. You see the following menu, or a similar menu:
System Maintenance Menu 1 Start System 2 Install System Software 3 Run Diagnostics 4 Recover System 5 Enter Command Monitor 6 Select Keyboard Layout |
The menu items have the following effects:
| Start System | This option starts the default operating system. | |
| Install System Software |
| |
| Run Diagnostics |
| |
| Recover System |
| |
| Enter Command Monitor |
| |
| Select Keyboard Layout |
|
Enter the numeral 5 and press <Return> or click on the appropriate button. You see the Command Monitor prompt:
>> |
You have entered the Command Monitor.
Table 4-1 summarizes the Command Monitor commands and gives each command's syntax.
Table 10-1. Command Monitor Command Summary
Command | Description | Syntax |
|---|---|---|
auto | Boots default operating system (no arguments). This has the same effect as selecting Start System from the PROM Monitor initial menu. | auto |
boot | Boots the named file with the given arguments. | boot [-f ][-n] pathname |
date | Displays or sets the date and time. | date [mmddhhmm[ccyy|yy] [.ss]] |
eaddr | Prints the ethernet address of the built-in ethernet controller on this system. | eaddr |
exit | leave Command Monitor and return to the PROM menu. | exit |
help | Prints a Command Monitor command summary | help [command] |
hinv | Prints an inventory of known hardware on the system. Some optional boards may not be known to the PROM monitor. | hinv |
init | Partially restarts the Command Monitor noting changed environment variables. | init |
ls | List files on a specified device. | ls devicename |
off | Turns off power to the system. | off |
passwd | Sets PROM password | passwd |
pathname | Given a valid file pathname, the system attempts to find and execute any program found at that path. | pathname |
printenv | Displays the current environment variables | printenv [env_var_list] |
resetenv | Resets all environment variables to default | resetenv |
resetpw | Resets the PROM password to null (no password required). | resetpw |
setenv | Sets environment variables. Using the -p flag makes the variable setting persistent, that is, the setting will remain through reboot cycles. | setenv [-p] variable value |
single | Boots the system into single user mode. | single |
unsetenv | Unsets an environment variable | unsetenv variable |
version | Displays Command Monitor version | version |
The question mark (?) command displays a short description of a specified command. If you do not specify a command, the ? command displays a summary of all Command Monitor commands. To get help, type either help or a question mark (?).
help [command] ? [command] |
The following sections cover these subjects:
The command syntax notation that this chapter uses
The function of the commands listed in Table 10-1
You can edit on the command line by using the commands shown in
Table 4-2.
Table 10-2. Command Monitor Command Line Editor
Command | Description |
|---|---|
<ctrl-h>, <del>, | Deletes previous character |
<ctrl-u> | Deletes entire line; question mark (?) prompts for corrected line |
<ctrl-c> | If a command is executing, kills current command |
!! | Repeat the last command |
The Command Monitor command syntax is designed to resemble the syntax of commands used with the IRIX operating system. This chapter uses IRIX notation for command descriptions:
Boldface words are literals. Type them as they are shown.
Square brackets ([]) surrounding an argument means that the argument is optional.
Vertical lines (|) separating arguments mean that you can specify only one optional argument within a set of brackets.
file means that you must specify a file name. A file name includes a device specification as described in “Syntax of Command Monitor File Names”.
When you specify file names for Command Monitor commands, use this syntax:
device([cntrlr,[unit[,partition]]])file |
device specifies a device driver name known to the PROM.
cntrlr specifies a controller number for devices that may have multiple controllers.
unit specifies a unit number on the specified controller.
partition specifies a partition number within a unit.
file specifies a pathname for the file to be accessed.
If you do not specify cntrlr, unit, and partition, they default to zero. The notation shows that you can specify only a cntrlr, a cntrlr and unit, or all three variables. The commas are significant as place markers. For example, the root partition (partition 0) on a single SCSI disk system is shown as:
dksc(0,1,0) |
where:
dksc indicates the SCSI driver
The first 0 indicates SCSI controller 0
The 1 indicates drive number 1 on SCSI controller 0
The final 0 indicates partition 0 (root partition) on drive 1 on SCSI controller 0.
The /usr partition (partition 3) on the same disk would be written as:
dksc(0,1,3) |
The Command Monitor defines the devices shown in Table 4-3.
Table 10-3. Device Names for Command Monitor Commands
Device Name | Description |
|---|---|
dkip | the ESDI disk controller (ips in IRIX) |
dksc | the SCSI disk controller (dks in IRIX) |
tpsc | the SCSI tape controller (tps in IRIX) |
xyl | the SMD disk controller (xyl in IRIX) |
ipi | the IPI disk controller (ipi in IRIX) |
tty | CPU board duart |
tty(0) | the local console |
tty(1) | the remote console |
gfx | the graphics console |
console | the ``pseudo console'' which may be one of gfx(0), tty(0), or tty(1). |
bootp | Ethernet controller using bootp and TFTP protocols |
tpqic | the quarter-inch QIC02 tape drive |
The PROM device notation is different from IRIX device notation. Certain environment variables (such as root and swap) are passed to higher level programs, and often require IRIX notation for the /dev device name. For example, in PROM notation, an ESDI disk partition most commonly used for swap is written:
dkip(0,0,1) |
In IRIX notation, the same disk is:
ips0d0s1 |
Systems that use the ARCS prom (including Indy, Indigo2, Indigo R4K, CHALLENGE, Onyx, and others) use a slightly different syntax for specifying pathnames and disk partitions.
ARCS pathnames use the same syntax as the hardware inventory. The pathnames are written as a series of "type(unit)" components that parallel the hardware inventory format.
Old style pathnames are automatically converted to new style pathnames, so the old names can still be used. The PROM will match the first device described by the pathname, so full pathnames are not always required. Some examples of common pathnames are shown in Table 10-4.
ARCS Naming Convention | Pathname or Device |
|---|---|
scsi(0)disk(1)partition(1) | dksc(0,1,1) |
disk(1)part(1) | dksc(0,1,1) |
scsi(0)cdrom(5)partition(7) | dksc(0,5,7) |
network(0)bootp()host:file | bootp()host:file |
serial(0) | first serial port |
keyboard() | graphics keyboard |
video() | graphics display |
This section describes the commands that you use to run the Command Monitor. The Command Monitor accepts the commands listed in Table 10-1, “ Command Monitor Command Summary,”.
The init command reinitializes the processor from PROM memory, and returns you to the monitor program.
Your system has a facility that allows you to require a password from users who attempt to gain access to the Command Monitor. To set the PROM password, perform the following steps:
Select option 5 from the System Maintenance Menu to enter the Command Monitor. You see the Command Monitor prompt:
Command Monitor. Type "exit" to return to the menu. >> |
Enter the command:
help |
Issue the passwd command:
passwd |
You see the prompt:
Enter new password: |
Enter the password you want for your machine and press <Return>. You see the prompt:
Confirm new password: |
Enter the password again, exactly as you typed it before. If you typed the password the same as the first time, you next see the Command Monitor prompt again. If you made a mistake, the system prints an error message and you must begin again. If you see no error message, your password is now set. Whenever you access the Command Monitor, you will be required to enter this password.
It is very important that you choose and enter your password carefully, because if it is entered incorrectly or forgotten, you may have to remove a jumper on the CPU board of your system. This procedure is different for each system type, and is described in your Owner's Guide. Some systems, though, allow you to reset the PROM password from IRIX by logging in as root and issuing the following command:
nvram passwd_key "" |
The quotation marks with no characters or space between them are essential to remove the PROM password. You must be root to perform this operation.
The resetpw command within the Command Monitor also resets the PROM password.
You can copy a hard disk onto another hard disk easily through the Command Monitor. You may want to do this to create a backup disk in case of failure, or perhaps you have a specific software setup that you wish to copy for a new system. In order for this procedure to work correctly, and for the new disk to be useful, the disks must be of identical size and manufacture. Also, the system that is to receive the new disk must use the same CPU and graphics board set that the existing system uses, because the kernel is custom configured for CPU and graphics type, and the kernel will be copied exactly to the new disk.
Follow these steps:
Bring your system down and install the new disk in the space provided for an additional disk. Select a SCSI device number that is not currently in use for the new disk. For this example, we will use device 2 on SCSI controller 0 (the integral SCSI controller) for the new disk, and device 1 on SCSI controller 0 for the disk to be copied.
Boot the system to the System Maintenance Menu.
Select option 5 from the System Maintenance Menu. You will see the Command Monitor prompt:
>> |
Give the sash or boot command to load the sash:
sash |
From the sash prompt, give the command:
cp -b 128k disk(1)part(10) disk(2)part(10) |
You see a ``read error'' message when the copy is complete. This is the normal message that tells you that the copying software has read the entire disk. If you see a ``write error'' message, there was an error copying the disk and you will probably have to start over or install the disk by more conventional means.
The Command Monitor maintains an environment, which is a list of variable names and corresponding values (the values are actually text strings). These environment variables contain information that the Command Monitor either uses itself or passes to booted programs. The system stores some environment variables—those that are important and unlikely to change frequently—in non-volatile RAM (nvram). If you turn off power to the machine or press the Reset button, the system remembers these variables. When you change the setting of these variables using the setenv command, the PROM code automatically stores the new values in non-volatile RAM.
You can also use the /etc/nvram command to set or print the values of non-volatile RAM variables on your system. For complete information on the nvram command, see the nvram(1) reference page.
Table 10-5, “ Variables Stored in Non-volatile RAM,” shows a list of the environment variables that the system stores in non-volatile RAM.
The ARCS Prom defines some variables not found in older PROMS, and so an additional list is provided in Table 10-7.
Several environment variables also exist that affect IRIX's operation. These are not stored in non-volatile RAM, but they do affect the operation of the PROM and of IRIX. See Table 10-6.
Table 10-5 lists non-volatile RAM variables:
Table 10-5. Variables Stored in Non-volatile RAM
Variable | Description |
|---|---|
netaddr | Specifies the local network address for booting across the Ethernet. See the bootp protocol. |
dbaud | Specifies the diagnostics console baud rate. You can change it by setting this variable (acceptable rates include 75, 110, 134, 150, 300, 600, 1200, 2400, 4800, 9600, and 19200), or by pressing the <Break> key. IRIS uses the dbaud rate for the diagnostics console during the entire system start-up. Pressing the <Break> key changes the baud rate only temporarily; the baud rate reverts to the value specified in dbaud or rbaud when you press the reset switch or issue an init command. |
rbaud | Specifies the remote console baud rate. The list of acceptable baud rates is the same as for dbaud, above. |
bootfile | Specifies the name of the file to use for autobooting, normally a stand-alone shell (sash). This variable is valid for pre-ARCS PROMs only. ARCS proms store this information in the OSLoader variable. |
bootmode | Specifies the type of boot in pre-ARCS PROMs. ARCS PROMs store this information in the AutoLoad variable. The options have these meanings: c - performs a complete cold autoboot, using the file pointed to by the bootfile variable to boot the kernel; boots sash, then boots kernel; runs power-on diagnostics. m - (default) goes straight to the Command Monitor; clears memory; runs power-on diagnostics. d - go straight to the Command Monitor; do not clear memory; do not run power-on diagnostics (on IRIS-4D 100, 200 and 300 series systems, this has the same effect as bootmode m). |
boottune | Selects the boot music string. A value of 0 randomizes the selection each time. 1 is the default value. (Supported only on Power Indigo2) |
autopower | Allows systems with software power control to automatically reset after a power failure if set to y. |
console | Specifies which console to use. The
options have these meanings: |
keybd | Specifies the type of keyboard used. The default is ``df.'' Available settings depend on the exact PROM revision, but may include some or all of: USA, DEU, FRA, ITA, DNK, ESP, CHE-D, SWE, FIN, GBR, BEL, NOR, PRT, CHE-F. or US, DE, FR, IT, DK, ES, deCH, SE, FI, GB, BE, NO, PT, frCH on systems with the keyboard layout selector. On some systems, JP is also acceptable to specify a Japanese keyboard. |
diskless | Specifies that the system is diskless and must be booted over the network. On ARCS systems, diskless system environment parameters should be set as follows: diskless=1 SystemPartition=bootp() host:/path OSLoader= kernelname |
monitor | Specifies the monitor resolution on Indy systems when an unrecognized brand of monitor is used. Set this variable to h or H to specify a high resolution monitor, the default is a low resolution monitor. |
nogfxkeybd | Specifies that the keyboard is not required to be connected if set to 1. |
notape | Specifies that no tape drive is attached to the system. If a tape drive is attached to the system, this variable must be set to 1 (true) in order to access a tape drive on another system on the network. |
volume | Specifies the system speaker volume numerically. |
pagecolor | Specifies the background color of the textport using a set of 6 hexadecimal RGB values. |
prompoweroff | On Indy systems only, this variable specifies that the system should return to the PROM monitor before powering down on shutdown if set to y. |
rebound | Specifies that the system should automatically reboot after a kernel panic if set to y. |
sgilogo | Specifies that the Silicon Graphics logo and related information will be displayed on the PROM monitor graphical screen if set to y. |
diagmode | Specifies the mode of power-on diagnostics. If set to v, then diagnostics are verbose and extensive. . |
Table 10-6 lists Command Monitor environment variables that directly affect the operating system. Note that these variables are not stored in non-volatile RAM and are discarded if the machine is powered down.
Table 10-6. Environment Variables That Affect the IRIX Operating System
Variable | Description |
|---|---|
showconfig | Prints extra information as IRIX boots. If set through setenv, its value must be istrue. |
initstate | Passed to IRIX, where it overrides the initdefault line in /etc/inittab. Permitted values are s and the numbers 0-6. See init(1M). |
swap | Specifies in IRIX notation the swap partition to use. If not set, it defaults to the partition configured into the operating system, which is normally partition 1 on the drive specified by the root environment variable. |
path | Specifies a list of device prefixes that tell the Command Monitor where to look for a file, if no device is specified. |
verbose | Tells the system to display detailed error messages. |
When you boot a program from the Command Monitor, it passes the current settings of all the environment variables to the booted program.
The environment variables specific to ARCS Proms are described in Table 10-7.
Table 10-7. ARCS PROM Environment Variables
Variable | Description |
|---|---|
ConsoleIn/ConsoleOut | These variables are set automatically at system startup. |
OSLoadPartition | The disk partition where the operating system kernel is located. This is also used as the default root partition and is set automatically at system startup. |
OSLoader | The operating system loading program. By default, this is SASH (The Stand-Alone Shell). This is set automatically at system startup. |
SystemPartition | The disk partition where the operating system loading program is found. This is set automatically at system startup. |
OSLoadFilename | The file name of the operating system kernel. By default, this is /unix. This variable is automatically set at system startup. |
OSLoadOptions | This variable specifies options to the boot command used to load the Operating System. For more information on boot options, see “Booting a Program from the Command Monitor”. |
AutoLoad | This variable specifies whether the operating system will boot automatically after a reset or power cycle. This variable supercedes bootmode and can be set to yes or no. |
The printenv command displays the Command Monitor's current environment variables.
printenv [env_var_list] |
To change (reset) the variables, see the next section.
The setenv command changes the values of existing environment variables or creates new environment variables.
setenv env_var string |
env_var is the variable you're setting, and string is the value you assign to that variable. To see the current monitor settings, use printenv.
When you use setenv to change the value of one of the stored environment variables in Table 10-5, the system automatically saves the new value in non-volatile RAM. You do not need to re-enter the change the next time the machine is turned off and then on again.
If the keybd variable is set to anything but the default df, the appropriate keyboard translation table is loaded from the volume header of the hard disk. If the table is missing or unable to load, then the default table stored in the PROMs is used. The keybd variable can be set to any value, but the keyboard translation table should be loaded from the volume header on the hard disk. This variable overrides the normal system mechanism for determining the kind of keyboard installed in the system. You should not change this variable unless you are performing keyboard diagnostics. Table 10-8 lists keybd variables suggested for international keyboards:
Table 10-8. keybd Variables for International Keyboards
Variable | Description |
|---|---|
BEL or BE | Belgian |
DNK or DK | Danish |
DEU or DE | German |
DF | The default |
FRA or FR | French |
FIN or FI | Finnish |
ITA or IT | Italian |
JP | Japanese |
NOR or NO | Norwegian |
PRT or PT | Portuguese |
CHE-F or freCH | Swiss-French |
CHE-D or deCH | Swiss-German |
ESP or ES | Spanish |
SE or SWE | Swedish |
GB or GBR | United Kingdom (Great Britain) |
US or USA | United States (available on all models) |
This section describes each Command Monitor boot command and shows you how to use it. When you reboot or press the Reset button, you start up the Command Monitor. Do not press the Reset button under normal circumstances, that is, when the workstation is running IRIX.
The auto command reboots the operating system. It uses the default boot file as though you were powering up the CPU. At the Command Monitor prompt (>>), type:
auto |
The PROM's environment variable bootfile specifies the default boot file. In addition, you must set the environment variable root to the disk partition that IRIX uses as its root file system. The auto command assumes that the desired image of IRIX resides on the partition specified by root of the drive specified in the environment variable bootfile.
The bootfile name can contain no more than 14 characters. To select a different boot file, see “Changing Environment Variables”.
The boot command starts the system when you want to use a specific boot program and give optional arguments to that program. The syntax of the boot command is:
boot[-f program][-n][args] |
-f specifies the program you want to boot. The program name must contain fewer than 20 characters. If you do not specify this option, the environment variable bootfile specifies the default program. boot normally loads sash.
When you specify a program, you can include a device specification. If you don't, the Command Monitor uses the device specifications in the environment variable path. The Command Monitor tries in turn each device that you specify in path, until it finds the program you request, or until it has tried all the devices listed in path.
-n means no go: it loads the specified program, but does not transfer control to it. Instead, -n returns you to the Command Monitor command environment.
args are variables that the Command Monitor passes to the program you're booting. For an arg that starts with a hyphen (-), you must prepend an additional hyphen so that the Command Monitor doesn't think that the argument is intended for itself. The Command Monitor removes the extra hyphen before it passes the argument to the booted program. For more information, see “Booting the Standalone Shell”.
For example, to boot the disk formatter/exerciser program (fx) from the cartridge tape drive, use this command:
boot -f SCSI(0)tape(7)partition(0)fx |
Without any arguments, boot loads the program specified in bootfile.
The Command Monitor has been designed to keep it independent of operating systems and as small as possible. Therefore, the Command Monitor cannot directly boot files residing in IRIX or other operating system file trees. However, the Command Monitor does provide a two-level boot mechanism that lets it load an intermediary program that does understand file systems; this program can then find and load the desired boot file. The program is called the standalone shell, and is referred to as sash. sash is a reconfigured and expanded version of the Command Monitor program, and includes the modules needed to handle operating system file structures. It also has enhanced knowledge about devices.
After the system software is installed, a copy of sash is located in the volume header of the first disk. The header contains a very simple file structure that the Command Monitor understands. You can also boot sash from tape or across the network if need be. To boot sash from your disk, shut down the system, and when you see the message:
Starting up the system... To perform system maintenance instead, press Esc |
Press the escape key. You may have to enter your system's Command Monitor password, if your system has one. Next, you see a menu similar to the following:
System Maintenance Menu (1) Start System (2) Install System Software (3) Run Diagnostics (4) Recover System (5) Enter Command Monitor |
Select option 5, ``Enter Command Monitor'' from the System Maintenance Menu. You see the following message and prompt:
Command Monitor. Type "exit" to return to the menu. >> |
To boot the standalone shell (sash), enter the command:
boot -f sash |
sash operates in interactive command mode. You see the sash prompt:
sash: |
To use the multi-level boot feature, set the PROM environment variable bootfile to refer to a specific copy of sash. In normal configurations, setting bootfile to dkip(0,0,8)sash tells the Command Monitor to load sash from the ESDI disk controller 0, disk unit 0, partition 8 (the volume header). Use this syntax:
| ESDI drives | setenv bootfile "dkip(0)disk(0)partition(8)sash" | |
| SCSI drives | setenv bootfile "scsi(0)disk(1)partition(8)sash" | |
| SMD drives | setenv bootfile "xyl(0)disk(0)partition(8)sash" | |
| IPI drives | setenv bootfile "ipi(0)disk(0)partition(8)sash" |
Then issue a boot command, as in this example for an ESDI drive:
boot dkip()unix initstate=s |
The following actions take place:
boot loads dkip(0)disk(0)partition(8)sash, as specified by bootfile, since the boot command doesn't contain a -f argument. (A -f argument would override the default specified by bootfile.)
sash gets two arguments: dkip()unix and initstate=s, which brings the IRIS up in single-user mode. (Note that the Command Monitor removes the leading hyphen [-] from any argument, so if you use the next layer of software, and need an argument with a leading hyphen, you should put two hyphens in front of it.)
sash loads the file specified by the first argument (dkip()unix) and passes the next argument to that file.
Do not issue the auto command from sash with the bootfile set as shown above. If you do, the system tries to boot sash over itself and will exit with an error.
To be able to use the auto command from sash, set bootfile to refer to the kernel, for example, dkip()unix. Even better, return to the PROM level to use the auto command.
At the heart of the operation of diskless workstations is the bootp protocol. The bootp protocol is a DARPA standard protocol supported on all Silicon Graphics servers and workstations. One of the devices that the Command Monitor can use for booting is the network. Silicon Graphics provides a TCP/IP boot protocol that lets you boot files that reside on another host in the network, if the other host supports the booting protocol. The network booting protocol is the bootp protocol. It is a datagram protocol that uses the User Datagram Protocol (UDP) of TCP/IP to transfer files across the Ethernet network.
To boot across the network, you must first determine the Internet address of the machine you want to boot. The Internet address is a number assigned by the network administrator of the network to which the system is attached. The format of the number is four decimal numbers between 0 and 255, separated by periods; for example:
192.20.0.2 |
Use the setenv command to set the netaddr environment variable to this address; for example:
setenv netaddr 192.20.0.2 |
Once you have set the netaddr environment variable, you can use bootp to refer to a remote file by using a file name of the form:
bootp()[hostname:] path |
hostname is the name of the host where the file resides. The specified host must run the bootp server daemon, bootp. If you omit hostname, bootp broadcasts to get the file from any of the hosts on the same network as the machine making the request. The first host that answers fills the request. Only hosts that support bootp can respond to the request. It is safe to omit the hostname only when you know that the path is unique to a particular host, or when you know that all the copies of the file are interchangeable.
hostname can be the name of a host on a different Ethernet network from the machine that you are booting, if a gateway on the local Ethernet network provides a route to the remote host. The gateway must be an IRIS workstation running a bootp server that you have configured to do cross-network forwarding.
For more information about booting through gateways, see bootp(1M). For more information about the /etc/inetd.conf configuration file, see inetd(1M).
path is the pathname of a file on the remote host. For example, this command:
boot -f bootp()wheeler:/usr/local/boot/unix |
boots the file /usr/local/boot/unix from the remote host wheeler. The command:
boot -f bootp()/usr/alice/help |
boots the file /usr/alice/help from any host on the network responding to the bootp broadcast request that has a file of that name.
To configure the gateway to permit cross-network forwarding, follow these steps:
Log in as root or become the superuser by issuing the su command.
Edit the file /etc/inetd.conf on the gateway machine. This file configures the bootp server, which is started by the inetd(1M) daemon.
Change the bootp description so that inetd invokes bootp with the -f flag. Find this line:
bootp dgram udp wait root /usr/etc/bootp bootp |
Add the -f flag to the final bootp on the line:
bootp dgram udp wait root /usr/etc/bootp bootp -f |
Change the tftp configuration line in one of the following ways:
Remove the -s flag from the argument list for tftpd:
tftp dgram udp wait guest /usr/etc/tftpd tftpd -s |
This allows tftpd access to all publicly readable directories. If you are concerned about a possible security compromise, you can instead explicitly list the directories to which tftpd needs access. In this case, you need to add /usr/etc:
tftp dgram udp wait guest /usr/etc/tftpd tftpd -s /usr/etc |
See tftpd(1M) and tftp(1C) for more information.
Signal inetd to re-read its configuration file.
killall -1 inetd |
If you have access to a larger network, and the bootable file you need is sufficiently remote on the network that the tftp and bootp timeouts and network delays are keeping you from booting successfully, it is possible to use an intermediary host as a bootp server.
As an example, consider the following situation. You have a host named local_host that needs to boot a kernel found on the remote system far_host. But the network is heavily used, resulting in bootp and tftp timing out before the boot operation can take place. However, a third host, near_host, has the optional NFS software and has automount(1M) running, allowing access to the files on far_host. To boot through this method, perform the following steps:
On near_host, the system acting as intermediary, log in as root and edit the file /etc/inetd.conf. This file configures the bootp server, which is started by the inetd(1M) daemon.
On near_host, change the bootp description in the /etc/inetd.conf file so that inetd invokes bootp with the -f flag. Find this line:
bootp dgram udp wait root /usr/etc/bootp bootp |
Add the -f flag to the final bootp on the line:
bootp dgram udp wait root /usr/etc/bootp bootp -f |
On near_host, change the tftp configuration line in the /etc/inetd.conf file in one of the following two ways:
Remove the -s /usr/local/boot string from the argument list for tftpd, so that the entry matches the following:
tftp dgram udp wait guest /usr/etc/tftpd tftpd |
This allows tftpd access to all publicly readable directories.
If you are concerned about a possible security compromise, you can instead explicitly list the directories to which tftpd needs access. In this case, you need to add /hosts:
tftp dgram udp wait guest /usr/etc/tftpd tftpd -s /hosts |
See tftpd(1M) and tftp(1C) for more information.
Signal inetd to re-read its configuration file.
killall -1 inetd |
On far_host, the system on the distant subnetwork, use NFS to export the directory containing the needed bootable kernel (in this case, the file is /usr/local/boot/unix). If you need help exporting a directory, see the export(1M) reference page.
On local_host, the system you are trying to boot, give the command:
boot -f bootp()near_host:/hosts/far_host/usr/local/boot/unix |
If bootp times out, try the command again, as automount may require a bit of time to retrieve the files from the remote system.
To tell the Command Monitor to load standalone commands from various resources (such as a disk or CD-ROM device), set the path environment variable. (See “Changing Environment Variables”.) Set the path variable as follows:
setenv path "device_name alternate_path" |
For example, issue the following command:
setenv path "dkip(0)disk(0)partition(8)bootp()/altdir/altbootfile" |
This causes the Command Monitor to boot the file dkip(0)disk(0)part(8)altbootfile. If that file fails, the Command Monitor boots bootp()/altdir/altbootfile. If that file also fails, the Command Monitor prints the message ``command not found''. Note that pathnames are separated with spaces. If the device specification is contained within a command or by bootfile, the Command Monitor ignores path. Only bootp or volume headers are understood by the PROM.