7.1.1 Disk Partitions
A disk consists of storage units called sectors. Each sector is usually 512 bytes. A sector is addressed by the logical block number (LBN). The LBN is the basic unit of the disk's user-accessible data area that you can address. The first LBN is numbered 0, and the highest LBN is numbered one less than the number of LBNs in the user-accessible area of the disk.
Sectors are grouped together to form up to eight disk partitions.
However, disks differ in the number and size of partitions.
The /etc/disktab file contains a list of supported disks and the default
partition sizes for the system. Refer to the disktab(4) reference page
for more information.
Disk partitions are logical divisions of a disk that allow you to organize files by putting them into separate areas of varying sizes. Partitions hold data in structures called file systems and can also be used for system operations such as paging and swapping. File systems have a hierarchical structure of directories and files, which is described in Section 7.1.3. By selecting the file systems to be placed in a partition, you can monitor the growth and activity of the disk.
Disk partitions have default sizes that depend on the type of disk and that can be altered by using the disklabel command. Partitions are named a to h. While the allocated space for a partition can overlap another partition, a properly partitioned disk should not have file systems on overlapping partitions.
Figure 7-1 shows the default partitions and starting
(offset) sectors for an RZ73 disk:
Figure 7-1: RZ73 Default Disk Partitions
The disk label is located in block 0 (zero) in one of the first
sectors of the disk. The disk label provides detailed information about the
geometry of the disk and the partitions into which the disk is divided. The
system disk driver and the boot program use the disk label information to
recognize the drive, the disk partitions, and the file systems. Other information
is used by the operating system to use the disk most efficiently and to locate
important file system information.
The disk label description of each partition contains an identifier
for the partition type (for example, standard file system, swap space, and
so on). There are two copies of a disk label, one located on the disk and
one located in system memory. Because it is faster to access system memory
than to perform I/O, when a system recognizes a disk, it copies the disk label
into memory. The file system updates the in-memory copy of the label if it
contains incomplete information about the file system. If a disk is not labeled
and does not contain an ULTRIX-style partition table, the partitions are obtained
from the /etc/disktab file. You can change the label with
the disklabel command. Refer to Section 7.8
and to the
The basic unit of virtual memory and physical memory is the page. Virtual
memory attempts to keep a process' most recently referenced virtual pages
in physical memory. When a process references virtual pages, they are brought
into physical memory from their storage locations on disk. Modified virtual
pages can be moved to a temporary location on the disk called swap space if
the physical pages (the pages in physical memory) that contain the virtual
pages are needed by either a newly referenced virtual page or by a page with
a higher priority. Therefore, a process' virtual address space can consist
of pages that are located in physical memory, stored temporarily in swap space,
and stored permanently on disk in executable or data files. The operating
system uses two operations to move virtual pages between physical memory and
disk: paging and swapping.
Paging involves moving a single virtual page or a small cluster
of pages between disk and physical memory. If a process references a virtual
page that is not in physical memory, the operating system reads a copy of
the virtual page from its permanent location on disk or from swap space into
physical memory. This operation is called a pagein. Pageins typically occur
when a process executes a new image and references locations in the executable
image that have not been referenced before.
If a physical page is needed to hold a newly referenced virtual page
or a page with a higher priority, the operating system writes a modified virtual
page (or a small cluster of pages) that has not been recently referenced to
the swap space. This operation is called modified page writing or a pageout.
Note that only modified virtual pages are written to swap space because there
is always a copy of the unmodified pages in their permanent locations on disk.
Swapping involves moving a large number of virtual pages between physical
memory and disk. The operating system requires a certain amount of physical
memory for efficient operation. If the number of free physical pages drops
below the system-defined limit, and if the system is unable to reclaim enough
physical memory by paging out individual virtual pages or clusters of pages,
the operating system selects a low priority process and reclaims all the physical
pages that it is using. It does this by writing all of its modified virtual
pages to swap space. This operation is called a swapout. Swapouts typically
occur on systems that are memory constrained.
See Section 7.10 for information about how this
command interacts with overlapping partitions. The amount of swap space that
your system requires depends on the swap space allocation strategy that you
use and your system workload. Strategies are described in the following section.
Immediate mode is more conservative than deferred mode because each
modifiable virtual page is assigned a page of swap space when it is created.
If you use the immediate mode of swap space allocation, you must allocate
a swap space that is at least as large as the total amount of modifiable virtual
address space that will be created on your system. Immediate mode requires
significantly more swap space than deferred mode because it guarantees that
there will be enough swap space if every modifiable virtual page is modified.
If you use the deferred mode of swap space allocation, you must estimate
the total amount of virtual address space that will be both created and modified,
and compare that total amount with the size of your system's physical memory.
If this total amount is greater than the size of physical memory, the swap
space must be large enough to hold the modified virtual pages that do not
fit into your physical memory. If your system's workload is complex and you
are unable to estimate the appropriate amount of swap space by using this
method, you should first use the default amount of swap space and adjust the
swap space as needed.
You should always monitor your system's use of swap space. If
the system issues messages that indicate that swap space is almost depleted,
you can use the swapon command to allocate additional swap
space. If you use the immediate mode, swap space depletion prevents you from
creating additional modifiable virtual address space. If you use the deferred
mode, swap space depletion may result in one or more processes being involuntarily
terminated.
You may receive the following informational messages when you remove
the /sbin/swapdefault file and when you boot a system that
is using the deferred method:
If the /sbin/swapdefault file does not exist and
you want to enable the immediate method of swap allocation, become the root
user and create the file by using the following command syntax:
ln -s../dev/rzxy /sbin/swapdefault
The x variable specifies the device number
for the device that holds the primary swap partition, and the y variable specifies the swap partition. Usually, the swap
device number is the same as the boot device number, and the primary swap
partition is partition b.
You must reboot the system for the new method to take effect.
A UFS file system has four major parts:
The first block of every file system (block 0) is reserved for a boot,
or initialization, program.
Block 1 of every file system is called the superblock and contains the
following information:
When new blocks are allocated to a file, they are obtained from the
free-block list. When a file is deleted, its blocks are returned to the free-block
list.
A group of blocks follows the superblock. Each of these blocks contains
a number of inodes. Each inode has an associated inumber. An inode describes
an individual file in the file system. There is one inode for each possible
file in the file system. File systems have a maximum number of inodes; therefore
there is a maximum number of files that a file system can contain. The maximum
number of inodes depends on the size of the file system.
The first inode (inode 1) on each file system is unnamed and unused.
The second inode (inode 2) must correspond to the root directory for the file
system. All other files in the file system are under the file system's root
directory. After inode 2, you can assign any inode to any file. You can
also assign any data block to any file. The inodes and blocks are not allocated
in any particular order.
If an inode is assigned to a file, the inode can contain the following
information:
The possible types are regular, device, named pipes, socket, and symbolic
link files.
The inode contains the user and group identification numbers that are
associated with the owner of the file.
Protection information specifies read, write, and execute access for
the file owner, members of the group associated with the file, and others.
The protection information also includes other mode information specified
by the chmod command.
A directory entry (link) consists of a name and the inumber (inode number)
that represents the file. The link count indicates the number of directory
entries that refer to the file. A file is deleted if the link count is zero;
the file's inode is returned to the list of free inodes, and its associated
data blocks are returned to the free-block list.
These pointers indicate the actual location of the data blocks on the
physical disk.
Data blocks contain user data or system files.
disklabel(8) reference page for more information.
7.1.2 Adding Swap Space
The Digital UNIX
operating system uses a combination of physical memory and disk space to create
virtual memory, which can be much larger than the physical memory. Virtual
memory can support more processes than the physical memory alone. This section
and the sections that follow describe important virtual memory concepts that
you should consider when configuring swap space.
7.1.2.1 How Swap Space is Allocated
Swap space is initially allocated during system
installation. You can add swap space after the installation by including
swap space entries in the /etc/fstab file and then rebooting.
Additionally, you can use the swapon command to add more
swap space--overriding the /etc/fstab definitions--until
the next time the system is rebooted. Refer to Section 7.4
and the swapon(8) reference page for more information.
7.1.2.2 Estimating Swap Space Requirements
There are two strategies for swap space allocation: immediate
mode and deferred or over-commitment mode. The two strategies differ in the
point in time at which swap space is allocated. In immediate mode, swap space
is allocated when modifiable virtual address space is created. In deferred
mode, swap space is not allocated until the system needs to write a modified
virtual page to swap space.
Note
The operating system will terminate a process if it attempts to write
a modified virtual page to swap space that is depleted.
7.1.2.3 Selecting the Swap Space Allocation Method
To determine which swap space allocation method is being used, check
for the existence of a soft link named /sbin/swapdefault,
which points to the primary
swap partition. If the /sbin/swapdefault file
exists, the system uses the immediate method of swap space allocation. To
enable the deferred method, rename or delete this soft link.
vm_swap_init: warning sbin/swapdefault swap device not found
vm_swap_init: in swap over-commitment mode
7.1.3 UNIX File System Structure
This
section discusses the UNIX File System (UFS). For information on the Advanced
File System (AdvFS) structure, refer to Chapter 8.
7.1.4 File System and Directory Hierarchy
The standard Digital UNIX system
directory hierarchy is set up for efficient organization. It separates files
by function and intended use. Effective use of the file system includes placing
command files in directories that are in the normal search path as specified
by the users' .profile or .login file,
as appropriate. Figure 7-2 shows the major directories in the
file system. Not all of the directories in the Digital UNIX hierarchy are
shown; you should use those shown in Figure 7-2 to ensure that
your product will be portable to other systems. Some of the directories are
actually symbolic links.
Figure 7-2: Partial Digital UNIX Directory Hierarchy
Table 7-1
describes the contents and purposes of the directories shown in Figure 7-2.