VERITAS Volume Manager 3.1 Administrator's Guide

Volume Manager Operations
Volume Manager Rootability
Chapter 3120
Volume Manager Rootability
Rootability is the term used to indicate that the logical volumes
containing the root file system and the system swap area are under
Volume Manager control. Normally the Volume Manager is started
following a successful boot after the operating system has passed control
to the initial user mode process. However, when the volume containing
the root file system is under Volume Manager control, portions of the
Volume Manager must be started early from the operating system kernel
before the operating system starts the first user process. Thus, the
Volume Manager code that enables rootability is contained within the
operating system kernel.
An HP-UX boot disk is set up to contain a Logical Interchchange format
(LIF) area. Part of the LIF structure is a LIF LABEL record, which
contains information about the starting block number and length of the
volumes containing the stand and root filesystems as well as the
volume containing the system swap area. Part of the procedure for
making a disk VxVM-rootable is to initialize the LIF LABEL record with
volume extent information for the stand, root, swap, and, optionally,
dump volumes. This is done with vxbootsetup (1M).
Booting With Root Volumes
Before the kernel mounts the root file system, it determines if the boot
disk was a rootable VxVM disk. If so, then the kernel passes control to
the kernel Volume Manager rootability code. The kernel rootability code
extracts the starting block number and length of the root and swap
volumes from the LIF LABEL record and builds “fake” volume and disk
configuration objects for these volumes and then loads this fake
configuration into the VxVM kernel driver. At this point, I/O can proceed
on these fake root and swap volumes by simply referencing the device
number that was set up by the rootability code.
Once the kernel passes control to the initial user procedure
(pre_init_rc()), the Volume Manager daemon starts and reads the
configuration of the volumes in the root disk group and loads them into
the kernel. At this time the fake root and swap volume hierarchy can be
discarded, as further I/O to these volumes will be done through the
normal configuration objects just loaded into the kernel.