Netbooting NetBSD/pmax

Please note: this page applies only to NetBSD-1.3 and later releases of NetBSD/pmax. For earlier revisions the console and second stage bootstraps and the kernel load are the same, but the kernel will then use RARP and bootparams to determine its IP address and NFS root and swap locations. For more information on this, see the diskless(8) manual page.

Possible problems

There are some problems with netbooting some DECstations. This can depend on both the model and specific PROM version you have. Decstation PROMs support booting via either the DecNET MOP protocol, or via bootp and tftp.

The biggest hurdle with diskless booting is that the PROMs of each model of DECstation come in a variety of different versions, with bugs that break diskless booting in various ways. Some versions of PROMs only boot reliably with MOP, others only boot reliably with tftp. Experimentally, smaller kernels seem to boot where bigger kernels fail. (The generic installation kernel counts as "bigger" in this context).

The long-term solution is probably to write a small, secondary bootloader, boot that via the PROMS, and use it to download a real kernel. No-one has yet written such a bootloader.

Here is a list of known PROM versions.

There's a new Diskless NetBSD HOW-TO available that has NetBSD/pmax specific information in it.

What to do

Booting and running NetBSD/pmax over a network requires a BOOTP server, a TFTP server and an NFS server. (These are usually all run on the same machine.) There are two basic stages to the boot:

1. The PROM software sends a BOOTP request to get its own address, the address of the TFTP server and the kernel to download. After loading the kernel into memory, it executes it.

2. The kernel probes and configures the devices, and then sends out another BOOTP request so it can find out its address, the NFS server, and path. It then mounts its root via NFS and continues.

1. Setting Up the Server

You will need to set up your server to serve BOOTP, TFTP and NFS.

The NFS setup is quite simple. Untar the NetBSD snapshot or distribution into a directory on your server and NFS export that directory to the client. You'll probably want to map root to `root' (rather than the default `nobody') when you export it. A typical /etc/exports line on a NetBSD system would be:

	/usr/export/pmax -maproot=0 myclient.mydom.com

For the TFTP setup, copy the ECOFF format kernel netbsd.ecoff from the extracted snapshot into an appropriately named file (I use netbsd.hostname) in the directory used by your TFTP server.

For the BOOTP server you need to specify the:

• hardware type (Ethernet)
• hardware address (Ethernet MAC address)
• IP address of the client
• subnet mask of the client
• address of of the TFTP/NFS server
• name of the kernel loaded via TFTP
• path to the root for the client (mounted via NFS)

Here's an example for a Unix system running bootpd:

	myhost.mydom.com:\
	        :ht=ethernet:ha=08002b123456:\
	        :ip=192.168.1.2:sm=255.255.255.0:\
	        :sa=192.168.1.1:bf=netbsd.myhost:rp=/usr/export/pmax:

2. The PMAX PROM

This section incomplete!

The only Ethernet device the PROM on most PMAX systems knows how to boot from is the onboard Ethernet interface. The PROM knows how to boot from a FDDI card with tftp, but there is no known reports of this actually working with NetBSD/pmax. How about an extra PMAD-A??.

Notes on various models' problems TFTP booting... Refer to the PROM list for more info

Model	Boot command
5000/xx, 5000/1xx, 5000/240, 5000/260	boot 3/tftp
5000/200	boot 6/tftp
2100, 3100	boot -f tftp()

3. The Kernel

Once loaded, the kernel will probe and configure devices. It will then send out another BOOTP request to get its IP address and the path to mount its root filesystem.

If you specified the n flag on boot, the kernel prompt on the console for a root device, dump device and file system type. You may reply to these with a question mark (?) to get a list of valid options. Currently there is no support for crash dumps on a system without a disk.

4. Notes

1. One or more of the following options must be specified in your kernel config file. See options(4).
   • NFS_BOOT_BOOTP
   • NFS_BOOT_DHCP

2. You will probably need to run swapctl -A -t noblk again in /etc/rc.local (at which point NFS is up) to activate NFS swap files in /etc/fstab. (Check this! /etc/rc looks ok).

3. There used to be problems setting up Digital UNIX (formerly DEC OSF/1) systems as NFS servers containing NetBSD/pmax diskless root filesystems. This should no longer be a problem - NFSv3 handles this correctly.

5. Credits

This page was based on the NetBSD/alpha Netbooting page written by Curt Sampson <cjs@NetBSD.org>.

Home Page

Supported Architectures