Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!seismo!ll-xn!cit-vax!mangler
From: mangler@cit-vax.Caltech.Edu (System Mangler)
Newsgroups: net.unix-wizards
Subject: Re: Location of partition table on disk
Message-ID: <1073@cit-vax.Caltech.Edu>
Date: Sun, 26-Oct-86 03:44:24 EST
Article-I.D.: cit-vax.1073
Posted: Sun Oct 26 03:44:24 1986
Date-Received: Mon, 27-Oct-86 01:05:25 EST
References: <8102@sun.uucp>
Organization: California Institute of Technology
Lines: 26
Summary: swap partitions, standalone programs

In article <8102@sun.uucp>, mojo@sun.UUCP writes:
> In the inode partitioning approach you have to have a file system open
> (which means you have to know partitioning info) before you can read the
> inode to get the partitioning info.

The only partitioning info you really need to know about the root
filesystem is the location of the superblock.  The superblock
will then tell you everything you want to know about the partition:
size, geometry, ...

The problem arises earlier in the boot process.  The first call to
a 4.2bsd disk driver is not an open(), as you might expect; rather,
it's an xxsize() call to determine the size of the primary swap
partition.

Putting the partition info in the inode is a great idea (no limit
on the number of partitions) but the bootstrapping problems are real.
It is dangerous to have one of your partitions defined separately,
e.g. the swap partition defined in the kernel; it causes accidents.
And then there's the standalone programs, which need to understand
partitions without being required to mount a root filesystem.

Looks like we need a "partitions" block, someplace that won't get
overwritten.  So where?  What sector number does Ultrix use?

Don Speck   speck@vlsi.caltech.edu  seismo!cit-vax!speck