3.1.1. Entering the BOOT_ADMIN interface

Entering the BOOT_ADMIN management tool isn't that awful :

1. Turn your PA-RISC box on.

2. During the boot process, the following message will appear on the current console (see Section 3.2) :

   Searching for Potential Boot Devices. To terminate search, press and hold the ESCAPE key.

   When this message appears, press and hold the Esc key until an option menu appears. This can take a while, be patient.

3. By default, you should enter the BOOT_ADMIN console. Though on some 715s and 725s, an option menu looking like this may appear :

   b) Boot from specified device s) Search for bootable devices a) Enter Boot Administration mode x) Exit and continue boot sequence ?) Help Select from menu:

   Select 'a) Enter Boot Administration mode'. This will bring up a 'BOOT_ADMIN>' prompt.

Once you have the 'BOOT_ADMIN>' prompt, you can pat yourself on the back : you are in BOOT_ADMIN mode !

3.1.2. BOOT_ADMIN commands

BOOT_ADMIN is an early boot subsystem where you can execute some precise commands. You should find here everything you need to know about them.

All HP-PA systems have BOOT_ADMIN. The display can be different but the idea remains the same. That's why the following list is not complete but consistent enough. Another important thing is that for each command, you have a shorter way to invoke it. You can see the shortcut shown as uppercase letters in the command name. Full names will be used in these sections.

Some commands may appear in several different menus, this is normal.

3.2.1. Using graphic console

To use the graphic console, you must first ensure that the Linux kernel supports your system's graphic card. There are two ways to deal with the graphic console. If you think about bug-reporting any trouble, you must know how to differenciate both. First, the STI console is the classical video text console, like VGA on a common PC for example. This name is due to the fact that each PA-RISC box features the Standard Text Interface which defines some standardized ways to access the video memory. The other graphic console is the well known framebuffer console (which on HP-PA uses STI in a special manner, hence the name stifb). In this case, when booting, you will see a characteristic little penguin appearing on the top-left corner. This is the easiest way to differenciate the two graphic modes.

Obviously, if you can use graphic console, it is the easiest way to proceed. Nevertheless, you must be sure that your hardware is supported.

3.2.2. Using serial console

The serial console is a good way to get all console messages handy, including the BOOT_ADMIN ones. It is very useful for bug reports, as its output can be easily dumped. Moreover, most of the servers can only be managed with serial console. Anyway, the only cases where you will HAVE TO use serial console is either if you don't have a monitor for your PA-RISC machine, or if your machine doesn't support graphics. It is also possible that the kernel can NOT handle some specific graphic hardware present in your model.

Here follows the procedure to setup serial console support.

3.2.3. Switching consoles

It might prove useful that you learn how to manage the console mode on your PA-RISC box. The following sections will explain the various operations on console modes.

---

3.2.3.1. Checking current console mode

Type : path console to see the current console mode.

If it's graphic console mode, it will return something like : 'Console path = graphic_1'.

If it's serial console, it will return : 'Console path = rs232_a.9600.8.none' or something similar.

For some models, you can find slight differences but the idea stays the same. If you want to see more descriptions here, please send us a message describing the box you use and what you get.

4.3.1. Making a lifimage with RAMDISK

First things first : when should you go this way ?

At an earlier step of the PA/Linux project, the lifimage was very useful. In fact, simply putting this file in a boot server tree allows you to boot your HP box via the boot lan instruction without any further involvement (see Section 5.3). The main advantage of a RAMDISK is that it unpacks its own file system in RAM, and therefore is completely independant from the machine I/O capabilities (hard drives, etc). The main drawback is that you have to build your own RAMDISK if you have memory constraints or some customized files. Now, let's see how to obtain a lifimage with RAMDISK.

If you don't want to mess with building your own RAMDISK, you can use root.bin that can be found on Debian Boot-Floppies.

We assume you got the latest source of the PA/Linux kernel. Mainly, you will need a (cross-)compiler, the linux/ directory and the PALO package installed. If you do not have it, run as root apt-get install palo. Everything can be found at http://www.parisc-linux.org/. Go through the make menuconfig step. Then, run make palo and if you have the PALO installed, you should get this message at the end of the compilation :

A generic palo config file (./palo.conf) has been created for you. You should check it and re-run "make palo". WARNING: the "lifimage" file is now placed in this directory by default!

So, edit the palo.conf file :

# This a generic Palo configuration file. For more information about how # it works try 'palo -?'. # # Most people using 'make palo' want a bootable file, usable for # network or tape booting for example. --init-tape=lifimage --recoverykernel=vmlinux ########## Pick your ROOT here! ########## # You need at least one 'root='! # # If you want a root ramdisk, use the next 2 lines # (Edit the ramdisk image name!!!!) --ramdisk=ram-disk-image-file --commandline=0/vmlinux HOME=/ root=/dev/ram initrd=0/ramdisk # If you want NFS root, use the following command line (Edit the HOSTNAME!!!) #--commandline=0/vmlinux HOME=/ root=/dev/nfs nfsroot=HOSTNAME ip=bootp # If you have root on a disk partition, use this (Edit the partition name!!!) #--commandline=0/vmlinux HOME=/ root=/dev/sda1

As you can see, the RAMDISK mode is the default. The string ram-disk-image-file should give to PALO the path of your RAMDISK file. You shouldn't change anything else to this file. After configuring the palo.conf, you can go back to your make palo. The result, a lifimage file, is waiting for you in the linux/ directory.

4.3.2. Making a lifimage with NFSROOT

This method is widely used because the kernel and the file system are directly accessible on your boot server. It is also very easy to test a new kernel. You just have to generate the kernel and put it in the correct directory. When starting up, the PA-RISC box will boot via the boot lan instruction its new kernel.

Getting the NFSROOT support is easier than RAMDISK. You have to edit the palo.conf to specify the boot server IP address instead of the string HOSTNAME. In fact, if your server has 10.10.10.2 as its IP adress, then the palo.conf file should contain :

# This a generic Palo configuration file. For more information about how # it works try 'palo -?'. # # Most people using 'make palo' want a bootable file, usable for # network or tape booting for example. --init-tape=lifimage --recoverykernel=vmlinux ########## Pick your ROOT here! ########## # You need at least one 'root='! # # If you want a root ramdisk, use the next 2 lines # (Edit the ramdisk image name!!!!) #--ramdisk=ram-disk-image-file #--commandline=0/vmlinux HOME=/ root=/dev/ram initrd=0/ramdisk # If you want NFS root, use the following command line (Edit the HOSTNAME!!!) --commandline=0/vmlinux HOME=/ root=/dev/nfs nfsroot=10.10.10.2 ip=bootp # If you have root on a disk partition, use this (Edit the partition name!!!) #--commandline=0/vmlinux HOME=/ root=/dev/sda1

If you have another IP, this field must be filled in with the correct data. You shouldn't change anything else to this file. After having configured the palo.conf, you can go into the linux/ directory and issue a make palo. The result, a lifimage file, is as usual waiting for you in the linux/ directory.

For advanced details on NFSROOT management, take a look at Bibliography for the appropriate HOWTOs.

4.3.3. Making a bootable partition

This part is where PALO can be seen as a LILO clone. PALO is mainly a program that enables a PA box to boot a kernel present on its own hard disk drive. This section is going to explain how to make it work. When installing the PALO package, Paul Bame, the author and maintainer, put a copy of the default /etc/palo.conf in /usr/share/doc/palo/palo.conf. If you want to understand how PALO works, you just have to read this file !

To setup a bootable hard disk, you have to partition properly your hard drive (if any, and if you want to use it as your primary boot device). This implies that this step can only be achieved either if you have already booted a minimal system on your PA-RISC box (via CD or network, see Chapter 5), or if you intend to prepare your hard disk using another computer than the target (which can be useful to unpack and setup a downloaded file system for example). The point of this HOWTO is not to teach you how to use fdisk or other, so here are the few things you HAVE TO know :

• A partition within the first 2GB of your target device has to be of partition type 'f0', which is the reserved partition type for PALO boot loader.

• It does not need to be huge. This is were PALO will save its configuration, recovery kernel(s) - about 5MB each - and optional ramdisk. 16-32MB seems far sufficient.

Beware ! Your vmlinux has also to be located within the first 2GB of the hard disk. We strongly recommend to create a separated /boot partition at the front of the disk if your '/' is bigger than that, because if ever your vmlinux goes above the first 2GB of the disk (like when filling up '/' with data), the box won't boot anymore.

Here is the output of fdisk which represents the hard drive of a box with 16MB PALO space, 128MB swap space and about 1GB '/' partition :

bash# fdisk -l /dev/sda

Disk /dev/sda: 133 heads, 62 sectors, 1017 cylinders
Units = cylinders of 8246 * 512 bytes

   Device Boot    Start       End    Blocks   Id  System
/dev/sda1   *         1         4     16461   f0  Linux/PA-RISC boot
/dev/sda2             5        34    123690   82  Linux swap
/dev/sda3   *        35       277   1001889   83  Linux

Now let's deal with PALO configuration. Here are the various parameters you can change :

• recoverykernel is the path to the kernel that you want to boot within a failsafe session, it will be stored in the 'f0' partition.

• bootloader is the path to the iplboot boot loader utility which is created by PALO when you issue a make iplboot.

• init-partitioned is used to indicate the pre-partionned device where palo will write its boot parameters. The effect is immediate. It means that PALO is going to write on the first octets of the first partition of this device, which partition-type must be 'f0', as shown above.

• commandline : the first digit is the number of your ext2 partition where the kernel file is located, as reported by fdisk. Logically, the next string is the absolute path to the kernel. The following space separated parameters will be passed to the kernel as its arguments. e.g. : HOME= and TERM= are environmental parameters passed to init when booting. They are not compulsory but they can be useful. root= tells the kernel which partition it must mount as the root file system while booting. It can be tricky when you have more than one disk.

  You can also add console=, to force the designation of the output console. You should remind that console=ttyS0 is for a serial console and console=tty0 is for a STI-console. Recently, support for the PDC console (if enabled in the kernel, see Section 6.2.3) has been added, using console=ttyB0. Indeed, the latest versions of PALO autodetect the right console path (except for PDC), and can figure out whether a 32bit or 64bit kernel should be used. If not, please mail to the mailing list.

In fact, this third usage of PALO is the most common but the default /etc/palo.conf makes it easy to configure. Just choose your root partition. It should the partition containing your root directory. According our fdisk example, we want /dev/sda3. Thus, the configuration file should look like that :

# The following arguments are set up for booting from /dev/sda3, specifically # mounting partition 3 as root, and using /boot/vmlinux as both the # recovery kernel, and the default dynamically-booted kernel. --recoverykernel=/boot/vmlinux --bootloader=/boot/iplboot --init-partitioned=/dev/sda --commandline=3/boot/vmlinux HOME=/ TERM=linux root=/dev/sda3

4.4.1. The theory

You have setup everything, rebooted your box, and suddendly you want to change something to the kernel boot arguments, or even boot another kernel. Damn it ! How could you, now that the box is booting ? Well, stay calm and relax, we have the solution !

First, you must learn how to interact with PALO during the startup sequence. You have to enter BOOT_ADMIN, as explained in Section 3.1.1. For some old models (up to 712 or so), you must add the ipl (or isl) string to your boot command in the BOOT_ADMIN console :

BOOT_ADMIN> boot pri ipl

On most PA-RISC boxes, the system will ask you if you want to interact with IPL anyway. You just have to answer by a "y". You will fall back to PALO configuration display, with the list of all parameters and their corresponding numbers.

You just have to enter the number corresponding to the parameter you want to change. Hit ENTER, modify it and validate the changes by hiting ENTER again. The system will redisplay the new list. This modification is not permanent ! To save your changes, you will have to run /sbin/palo when your system will be up and running, and it will write on the disk all the parameters contained in the default file, (/etc/palo.conf), which you will have properly modified if needed. If you want to add a supplementary parameter, select any one and write yours on the editing line, beginning with a space :

Edit which field? (or 'b' to boot with this command line)? 0 3/boot/vmlinux initrd=root.bin

After validation, the list will count one more parameter. If you want to delete one, select it and erase the complete entry. You will see that the list counts one less parameter.

For more informations about PALO, please take look at the PALO readme. This section is mostly inspired from Paul Bame's file as well as the page about PALO you can find at http://www.pateam.org/palo.html.

4.4.2. A complete example

This example has been suggested by Michael Damaschke. So, let's go for the story of the happy PA/Linux user booting a kernel, also called "I don't know how to configure my workstation to use the kernel I want during boot sequence !".

After switching your workstation and monitor on, a message on the screen will tell you that the workstation is about to start automatically the boot sequence, except if you hold the Esc key to stop the auto-booting. This is a very difficult step : you must hold the Esc key down ;o)

Depending on your model, you might need to press this key during a quite long time.

In some cases, the monitor can be too slow to get on, and won't allow you to see the warning message. A good workaround is to keep a close eye on the keyboard's lights : when they blink, this is the right time to press and hold the Esc key. If you still have troubles, please refer to the Consoles section.

There are few differences about the way to get access to BOOT_ADMIN (see Section 3.1.1). If you have an old box, you will get an information message displayed, where the workstation's firmware tells you that it will start searching for all bootable devices, or that you can break this by holding down the Esc key. This is the same procedure as before, you must press the Esc key.

You might then get a menu where you must press the a-key followed by ENTER-key. You are now facing the deadly 'BOOT_ADMIN>' prompt :^). First, we will turn off auto boot process by entering the following lines :

BOOT_ADMIN> auto boot off

then hit the ENTER key.

After that, you must tell the system from which boot device you would like to boot. If it's a hard drive, it must have a 'f0' partition at the beginning (see Chapter 5).

In this example, the old kernel is vmlinux and the new one is vmlinux-2.4.17-pa3. The chosen SCSI boot device is designed by : SCSI.X.0, where X is the SCSI-ID of the disk you want to boot from. e.g.

BOOT_ADMIN> boot SCSI.5.0

At the end of the previous command line, you must add the IPL token if you have a HP 9000/7xx system to specify that you want to interact with IPL. If you have a more recent hardware, the system will ask if you want to interact with IPL anyway :

Interact with IPL (Y or N)?>

Now, you can manually configure the PALO booting parameters. You can see a new menu, where you can configure on line '0' (selected by default) the boot partition number, and the path of your boot kernel.

Here is the complete session log of a A500 serial console output :

Main Menu: Enter command or menu > bo scsi.5.0 ipl Interact with IPL (Y, N, or Cancel)?> y Booting... Boot IO Dependent Code (IODC) revision 1 HARD Booted. palo ipl 0.97 root@c3k Tue Nov 27 14:51:48 MST 2001 Information: Boot device can't seek past 2Gb (ignore next error). byteio_read: seekread() returned -1 expected 2048 Partition Start(MB) End(MB) Id Type 1 1 15 f0 Palo 2 16 503 82 swap 3 504 2887 83 ext2 PALO(F0) partition contains: 0/vmlinux64 3990942 bytes @ 0x44000 Information: No console specified on kernel command line. This is normal. PALO will choose the console currently used by firmware (serial). Current command line: 3/boot/vmlinux root=/dev/sda3 HOME=/ console=ttyS0 TERM=vt102 0: 3/boot/vmlinux 1: root=/dev/sda3 2: HOME=/ 3: console=ttyS0 4: TERM=vt102 Edit which field? (or 'b' to boot with this command line)? 0 3/boot/vmlinux-2.4.17-pa3 initrd=0/root.bin Current command line: 3/boot/vmlinux-2.4.17-pa3 initrd=root.bin root=/dev/sda3 HOME=/ console=ttyS0 TERM=vt102 0: 3/boot/vmlinux-2.4.17-pa3 1: initrd=0/root.bin 2: root=/dev/sda3 3: HOME=/ 4: console=ttyS0 5: TERM=vt102 Edit which field? (or 'b' to boot with this command line)? 1 Current command line: 3/boot/vmlinux-2.4.17-pa3 root=/dev/sda3 HOME=/ console=ttyS0 TERM=vt102 0: 3/boot/vmlinux-2.4.17-pa3 1: root=/dev/sda3 2: HOME=/ 3: console=ttyS0 4: TERM=vt102 Edit which field? (or 'b' to boot with this command line)? b

PALO was first setup to boot the kernel file vmlinux located on the third partition of the SCSI device ID 5 LUN 0. (We know this since we have asked BOOT_ADMIN to boot on this device). But we wanted another kernel this time. We have pressed the ENTER key (to validate the default choice '0') and modify the text to match our needs, here vmlinux-2.4.17-pa3. We have also added an initrd=0/root.bin argument to the command line. We have validated our changes by hitting the ENTER key. Finally we have decided that we didn't want this additional argument, so we have selected it and erased it. At the end it asked again which field we wanted to edit, we have just put 'b' instead of any number and hit ENTER to boot our new kernel.

Please don't change any other parameter unless you really know what you do !

That's it ! PALO has no more secrets for you :-)

As you might have noticed, the BOOT_ADMIN interface can take several apparences, so don't be disappointed if yours does not match our examples.

5.3.1. Preparing to boot from network

This is a very old way to operate but it used to be the only one available for a long time. Usually you won't need to boot from network, except in some very specific cases (e.g. unsupported I/O devices). That's why it is detailed here.

You will need a lifimage to perform a network boot. See Section 4.2 to learn how to create one. You can also download one at http://www.pateam.org/cd-images/lifimages/

5.3.2. rboot or bootp?

All 'recent' machines can boot using bootp, starting from 715/100, 715/120, and 712s. Older ones, mostly early 715s, 710s and 725s need rboot.

To use BOOTP you have to enable the 'IP: Kernel level autoconfiguration -> IP: BOOTP support' within the 'Networking options' section of the kernel configuration, if you want to use a home-made kernel. See Chapter 6 for details.

5.3.3. Using rboot

ethernet_addr            bootfile

5.3.4. Using dhcp/tftp

We will see here how to use a DHCP server as a BOOTP one.

---

5.3.4.1. Obtaining dhcp/tftp

Debian users will just have to install the packages via these commands as root :

bash# apt-get install dhcp bash# apt-get install tftpd

If you need rpm packages (for the ISC dhcp server), the best way is to go to http://rpmfind.net/. It seems that Red Hat users need to create the user "nobody" belonging to the group "nogroup". The files present in your /tftpboot/ directory (see below) should have these user/group privileges.

The dhcp package can do far more than a simple bootp daemon. Nevertheless, it is also known to be far easier to configure. If you really want to try bootp, skip this and go to Section 5.3.5.

---

5.3.4.2. Configuring dhcp/tftp

Here are the instructions to set up dhcp on your boot server. To keep this explanation simple, we will assume that you want to assign a fixed IP to your box, without DNS update. Your subnet will be 192.168.1.0/24, with optional : gateway at 192.168.1.1, domain name foo.com and DNS at 192.168.1.4.

This section is dedicated to Debian users. For others distributions, it should be similar though there can be some differences like default directories. Since Debian maintains a dhcp package, we will focus on it only.

1. Edit /etc/inetd.conf on your boot server to add the following line, if it doesn't already exist :

   tftp dgram udp wait nobody /usr/sbin/tcpd \ /usr/sbin/in.tftpd /tftpboot

   Here, /tftpboot/ is being used as tftpd server's root. You can choose another directory if you want. According to man tftpd, this is the usual default directory.

   When this is done, restart inetd with : /etc/init.d/inetd restart. You can also issue a killall -HUP inetd.

2. According to man 5 dhcpd.conf, edit the /etc/dhcpd.conf file to contain something like :

   allow bootp; default-lease-time 600; max-lease-time 7200; # This will tell the box its hostname while booting: use-host-decl-names on; subnet 192.168.1.0 netmask 255.255.255.0 { option routers 192.168.1.1; option domain-name "foo.com"; option domain-name-server 192.168.1.4; } host [hostname] { hardware ethernet [mac address]; fixed-address [ip address]; filename "[boot filename]"; option root-path "[root path]"; }

   You have to fill in the [hostname], [mac address], [ip address], [boot filename] and [root path] fields with the appropriate information, where :

   • [hostname] is the name of the PA-RISC system.

   • [mac address] is the colon-delimited ethernet address of the PA-RISC system, which can be obtained by typing lanaddress at the 'BOOT_ADMIN>' prompt (see Section 3.1.2.3).

   • [ip address] is the IP address you wish to assign to the PA-RISC system.

   • [boot filename] is the name of the bootable kernel image you want to boot your system with.

   • [root path] is the path to the NFS root filesystem exported by the server.

   You'll end up with something like this for each box you want to netboot :

   host tatooine { hardware ethernet 00:40:05:18:0c:dd; fixed-address 192.168.1.22; filename "lifimage-tatooine"; option root-path "/exports/tatooineroot"; }

5.3.5. Using bootp/tftp

5.3.5.1. Obtaining bootp/tftp

For Debian users, you just have to install the packages via these commands as root :

bash# apt-get install bootp tfptd

If you need rpm packages, the best way is to go to http://rpmfind.net/. It seems that Red Hat users need to create the user "nobody" belonging to the group "nogroup". The files present in your /tftpboot/ directory (see below) should have these user/group privileges.

You'll have been warned ! This daemon is far more obfuscated in its configuration.

---

5.3.5.2. Configuring bootp/tftp

Follow these instructions to use bootp on your boot server :

This section is dedicated to Debian users. For others distributions, it should be similar though there can be some differences like default directories. Since Debian maintains a bootp package, we will focus on it only.

1. Edit /etc/inetd.conf on your boot server to add the following lines, if they don't already exist :

   tftp dgram udp wait nobody /usr/sbin/tcpd \ /usr/sbin/in.tftpd /tftpboot bootps dgram udp wait root /usr/sbin/bootpd \ bootpd -i -t 120

   Here, /tftpboot/ is being used as tftpd server's root. You can choose another directory if you want. According to man tftpd, this is the usual default directory.

   When this is done, restart inetd with : /etc/init.d/inetd restart. You can also issue a killall -HUP inetd.

2. According to man 5 bootptab, edit the /etc/bootptab file to contain :

   [hostname]:hd=/tftpboot:\ :rp=[root path]:\ :ht=ethernet:\ :ha=[mac address]:\ :ip=[ip address]:\ :bf=[boot filename]:\ :sm=255.255.255.0:\ :to=7200:

   You have to fill in the [hostname], [mac address], [ip address] and [root path] fields with the appropriate information, where :

   • [hostname] is the name of the PA-RISC system.

   • [mac address] is the NOT-delimited ethernet address of the PA-RISC system, which can be obtained by typing lanaddress at the 'BOOT_ADMIN>' prompt (see Section 3.1.2.3).

   • [ip address] is the IP address you wish to assign to the PA-RISC system.

   • [boot filename] is the name of the bootable kernel image you want to boot your system with.

   • [root path] is the path to the NFS root filesystem exported by the server.

   You'll end up with something like this :

   vodka:hd=/tftpboot:\ :rp=/usr/src/parisc/:\ :ht=ethernet:\ :ha=080069088717:\ :ip=140.244.9.208:\ :bf=lifimage:\ :sm=255.255.255.0:\ :to=7200:

5.3.6. Booting your PA/Linux system from network

To conclude with the developers' way to boot the kernel, this section will tell you how to actually boot your system from a network server. But it tends to be less and less used. Most users will prefer to stick to Section 5.2.

Here we are. These are just some tips to get the boot for those who tried the network way. We assume that you've done everything outlined above, your development PC is on the same subnet than your PA-RISC machine, you've got a bootable PA/Linux kernel lifimage on your boot server, and you're willing to give it a try. If everything is ready, as well as you, the following procedure will introduce you to the joy of network booting your PA box into Linux.

1. Fire up your PA-RISC system.

2. Watch your PA-RISC box starting up. When the following message appears during the PA-RISC machine's boot process, press and hold the Esc key :

   Searching for Potential Boot Devices. To terminate search, press and hold the ESCAPE key.

3. If needed, select 'a) Enter Boot Administration mode' from the menu. This brings up the 'BOOT_ADMIN>' prompt.

4. Type the following at the prompt : boot lan.

5. Watch your PA-RISC system magically becoming a PA/Linux system. Ta dah !

Of course your are supposed to run only one boot server at a time on your network, in order to avoid conflicts...

6.1.1. native build

Since Debian is the only distribution supporting PA-RISC architecture, if you want to use the Super Cow powers, you need to have some basic knowledge about the Debian packaging system. We will explain here how to quickly get a gcc compiler ready on your PA-RISC box.

# Binary packages
deb http://ftp.de.debian.org/debian unstable main contrib non-free
# non-US packages
deb http://ftp.de.debian.org/debian-non-US unstable/non-US main non-free contrib
# source packages
deb-src http://ftp.de.debian.org/debian unstable main contrib non-free
# non-US source packages
deb-src http://ftp.de.debian.org/debian-non-US unstable/non-US main contrib non-free

bash# apt-get update
bash# apt-get upgrade

6.1.2. cross compiled build

In this kernel build method, everything depends on the architecture of your building machine. For x86 computers, you can download a ready-to-use cross compiler archive on the PA/Linux FTP server. You can also find some "exotic" (like for MacOS X) cross-compilers archives on the PA/Linux ESIEE Team website. For other architectures or if you want to compile your own toolchain, please refer to Carlos O'Donell's HOWTO.

As there is not yet a 64bit userspace on HP-PA, you have to cross-compile 64bit kernel even if you are building on a 64bit PA-RISC box. You can get unofficial debs for hppa64 compilers and binutils at ftp://ftp.parisc-linux.org/unofficial-debs/. See PA-RISC Linux Website for details.

6.2.1. HIL Support

Since kernel-2.4.18-pa45, there is a full HIL support, for mice, tablets and keyboards. It is based on the Linux Input Driver model. See the PA-RISC/Linux FAQ and the mail posted on the mailing list by Helge Deller. Here is what it says :

1. Make sure you have a 2.4.18-pa45 or higher kernel source.

2. Look at your kernel configuration for the following options :

   CONFIG_INPUT=y CONFIG_INPUT_KEYBDEV=y CONFIG_INPUT_MOUSEDEV=y CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024 CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768 CONFIG_INPUT_EVDEV=y CONFIG_INPUT_SERIO=y CONFIG_HIL=y CONFIG_HP_SDC=y CONFIG_HIL_MLC=y CONFIG_HP_SDC_MLC=y CONFIG_HIL_KBD=y CONFIG_HIL_PTR=y

   There is no more CONFIG_HIL_KBD_BASIC.

3. On your target system, check that the following devices are available :

   /dev/input/mice /dev/input/mouseX /dev/input/eventX

   If they are not yet present, create them as root by running :

   bash# cd /dev; ./MAKEDEV input

4. Configure gpm with the following options in /etc/gpm.conf :

   device=/dev/input/mice type=imps2

5. Here is a sample /etc/X11/XF86Config-4 :

   Section "InputDevice" Identifier "HIL Keyboard" Driver "keyboard" Option "CoreKeyboard" EndSection Section "InputDevice" Identifier "HIL Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ImPS/2" Option "ZAxisMapping" "4 5" EndSection Section "ServerLayout" Identifier "Default Layout" Screen "Default Screen" InputDevice "HIL Keyboard" InputDevice "HIL Mouse" EndSection

   You can also download a sample XF86Config-4 here : ftp://ftp.parisc-linux.org/XFree86/XF86Config-4, adjust color depth and resolution, and put it in your /etc/X11/.

6.2.2. USB Support

USB support on HP-PA is still experimental, therefore it is only proposed as modules in default kernel configuration. We have tried to install a B2000 with builtin USB support, both 32 and 64bit, and it worked fine, despite some keyboard problems. Don't worry, nothing critical : the range of keys located between the main part of the keyboard (the letters, backspace, enter...) and the numeric pad are spoiled. They do not behave as expected at all.

You can use the numeric pad as arrow keys : when NumLock is not activated, it behaves as a navigation pad. e.g. 8 is Up Arrow, 4 is Left Arrow and so on.

1. Make sure you have a 2.4.18 or higher kernel source.

2. Look at your kernel configuration for the following options :

   CONFIG_SUPERIO=y CONFIG_HOTPLUG=y CONFIG_INPUT=y CONFIG_INPUT_KEYBDEV=y CONFIG_INPUT_MOUSEDEV=y CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024 CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768 CONFIG_INPUT_EVDEV=y CONFIG_USB=y CONFIG_USB_DEVICEFS=y CONFIG_USB_OHCI=y CONFIG_HID=y

3. On your target system, check that the following devices are available :

   /dev/input/mice /dev/input/mouseX /dev/input/eventX

   If they are not yet present, create them as root by running :

   bash# cd /dev; ./MAKEDEV input

4. Configure gpm with the following options in /etc/gpm.conf :

   device=/dev/input/mice type=imps2

5. The XF86-Config-4 is similar to the HIL, as it is also using the Linux Input Driver.

6.2.3. PDC Console Support

PDC Console has been improved by Richard Hirst in pa37 kernel, though it is still a very experimental feature. It is expected to provide adequate PDC Console support to E- and K-Class machines. Feedback would be really appreciated.

Now follow these steps to get it to work :

1. Make sure you have a 2.4.18-pa37 or higher kernel source.

2. Look at your kernel configuration for the following options :

   CONFIG_SERIAL_CONSOLE=y CONFIG_SERIAL_GSC=y CONFIG_SERIAL_NONSTANDARD=y CONFIG_PDC_CONSOLE=y

3. On your target system, check that the following devices are available :

   /dev/ttyB0

   If they are not yet present, create them as root by running :

   bash# cd /dev; ./MAKEDEV ttyB0

   It needs a recent MAKEDEV package to be created this way. By the time this HOWTO was written, such a package could only be found on netinst ISO provided at http://www.pateam.org/download.html.

4. Now you can boot your system, taking care that PALO uses console=ttyB0.

6.2.4. IDE Devices Support

There is nothing really special about IDE support. You have to check that your IDE Chipset is supported by the kernel. A common chipset found on PA-RISC hardware is NS87415. You can find it on B2000, J5000 and C3000 for instance. You will need IDE support to use some CD devices.

Here is an example to get IDE to work with this chipset :

1. Make sure you have a recent kernel source.

2. Look at your kernel configuration for the following options :

   CONFIG_IOMMU_CCIO=y CONFIG_PCI=y CONFIG_PCI_LBA=y CONFIG_IOSAPIC=y CONFIG_IOMMU_SBA=y CONFIG_SUPERIO=y CONFIG_IDE=y CONFIG_BLK_DEV_IDE=y CONFIG_BLK_DEV_IDEPCI=y CONFIG_BLK_DEV_IDEDMA=y CONFIG_BLK_DEV_ADMA=y CONFIG_BLK_DEV_IDEDMA=y CONFIG_BLK_DEV_NS87415=y

3. On your target system, check that the following devices are available :

   /dev/hd*

   If they are not yet present, create them as root by running :

   bash# cd dev; ./MAKEDEV hda hdb hdc hdd hde

   It needs a recent MAKEDEV package to be created this way. By the time this HOWTO was written, such a package could only be found on netinst ISO provided at http://www.pateam.org/download.html.

Of course we didn't mention much of the non architecture independant options. Moreover, the above settings may vary depending on your hardware. This is just an example.