[abeetha.m]

File Systems


Although most operating systems will follow the same rules
for creating MBRs and partition tables, they can do whatever they want to
within the partitions that they own. This is in fact the reason that this whole
system of partitions was developed.


As discussed above, the partition type code shown in the
5th byte of a partition table entry identifies the operating system and/or file
system used by a partition. An operating system must have a way to keep track
of the files on its volumes (partitions). This is called a file system and will
vary greatly from one operating system to another. Some OSes are capable of
reading file systems created by other OSes and some OSes can install more than
one type of file system.


These are a few of the more common file systems:


FAT - MS-DOS, WIN95, Windows NT, OS/2


VFAT - variant
of FAT16 used by Windows 95


FAT32 - Windows
95B (OSR2)


NTFS - Windows
NT


HPFS - OS/2
(High Performance File System)


FFS - Free BSD
Unix


EXT2 - Linux

---

FAT


FAT is the file system used by DOS. Windows NT and OS/2 can
also use the FAT system. FAT is an acronym for File Allocation Table. The FAT
table is like a map used to keep track of the location of all of the files in a
partition. Although the sector is the smallest unit that a BIOS or operating
system can work with on a drive (as far as reading and writing data), DOS
stores files in units called allocation units (or clusters, an older and
interchangeable term), which are defined as 2, 4, 8, 16, or 32 sectors,
depending on the size of the partition.


The cluster size is based on the partition size as follows:





Volume Size (MB) Allocation Unit Size FAT Type









0 - 16 4096 (4KB)
8 sectors 12-bit







16 - 128 2048 (2KB)
4 sectors 16-bit




128 - 256 4096
(4KB) 8 sectors 16-bit




256 - 512 8192 (8KB)
16 sectors 16-bit




512 - 1024 16384
(16KB) 32 sectors 16-bit




1024 - 2048 32768
(32KB) 64 sectors 16-bit


The volume sizes above are in
binary megabytes as used by FDISK.


When the DOS FORMAT command is used to perform a high-level
format on a partition of a hard drive, it creates a boot sector, a FAT table, a
second copy of the FAT table, and a root directory. The remainder of the drive
is the data area used for storing files. DOS will use either a 12-bit FAT
(FAT12) system or a 16-bit FAT (FAT16) system, depending on the size of the
partition. Partitions less than 16MB are automatically formatted using FAT12.
Partitions 16MB or larger are formatted with FAT16.


FAT12 uses 12-bit entries to point to file locations. This
means that it can handle only 4096 clusters (each having a unique 12-bit ID
number). FAT16 uses 16-bit entries, so it can handle 65,536 clusters.


Actually, the total numbers are 4078 and 65,518, due to
some values being reserved. The values FFF0-FFFF are reserved. (FFF-FF0 with
FAT12). FFF7 (or FF7) indicates a bad sector. FFFF (or FFF) indicates end of
cluster chain. Values from FFF8-FFFF (or FF8-FFF) may also be used to indicate
end of cluster chain. Also, the value 0000 (or 000) indicates an unused
cluster. The value 0001 (or 001) is never used.


The first two entries in the FAT table are reserved. These
entries contain the media descriptor byte and 3 FF bytes (two FF bytes with
FAT12, since each entry is one and a half bytes). This is the reason that the
value 0001 (or 001) is not used.


The design of the FAT file system is responsible for the
2GB limit on partition sizes under DOS.


A file's directory entry indicates what cluster it begins
in. That location in the FAT table shows either that the file also ends there
or that it also uses another cluster, which it points to. The location pointed
to for the cluster used by the file will also indicate either that the file ends
in that location or that it continues in the cluster indicated, and so on. So
the FAT will show a chain of all of the clusters occupied by a file. This
system allows a file to be fragmented, that is occupying non-contiguous
clusters. As files of various sizes are deleted from and copied to the volume,
file fragmentation can get out of control. Tools such as DEFRAG are used to
move parts of files around so that they occupy contiguous clusters. This
improves the efficiency with which a drive can read files from its platters.

---

VFAT


VFAT stands for Virtual File Allocation Table. This is a
variant on FAT16, which is used by Windows 95. VFAT is actually what is called
an Installable File System (IFS).


The main noticeable difference between FAT16 and VFAT is
the ability to use long filenames.

---

FAT32


FAT32 is introduced with the OSR2 version of Windows 95,
also called Windows 95B. The FDISK program provided with WIN95B will ask if you
want support for large drives. What it really means is do you want to use
FAT32. At this time (OCT97) FAT32 partitions can not be read by any OS besides
Windows 95.

---

---

BIOS


This section discusses briefly some aspects of the PC BIOS
that pertain to hard drives, just to give an idea as to what all is involved.


BIOS is an acronym for Basic Input Output System. The BIOS
on a PC resides in a ROM chip on the motherboard. Newer motherboards will use a
flash ROM chip for the BIOS so that it can be updated easily.


A PC can be looked at as a series of layers. The hardware
is the lowest layer. The BIOS layer provides an interface between the hardware
and the operating system, which is the next layer. The highest layer is the
application programs run by the user.


The BIOS on every PC is based largely on the BIOS that IBM
developed for the first PC, the IBM PC, in 1981.


Besides handling the bootup process, the BIOS handles
hardware and software interrupts. Computers based on Intel 80x86 CPUs are
mainly controlled by interrupts, which can be issued by the CPU, by hardware,
or by software. Hardware interrupts are invoked through IRQ lines, which are
actual wires on the ISA bus. Software interrupts are issued by the INT
instruction (part of the CPU's instruction set).


Each interrupt has a unique number between 00h and FFh.
INTs 00-04 are reserved for the CPU. INTs 08-0F and 70-77 are reserved for
hardware interrupts. The rest are for software interrupts. The BIOS sets up a
table called the Interrupt Vector Table located between 0000:0000 and 0000:0400
- the very beginning of RAM. This table contains addresses for the BIOS
routines associated with each interrupt. Interrupts issued by hardware devices
are handled by Interrupt Service Routines (ISRs). Interupts issued by software
are handled by Device Service Routines (DSRs). Most BIOS Device Service
Routines include various functions, which are identified by a number stored in
register AH.


INT 13 is the BIOS DSR for disk services, for both diskette
drives and hard drives. Here is a list of the functions used by the INT13 Fixed
Disk DSR: (some of these are obsolete or rarely used)


Parameter: Function:


AH = 00h Reset System Fixed Disk


AH = 01h Read Disk Staus


AH = 02h Read Disk Sectors


AH = 03h Write Disk Sectors


AH = 04h Verify Disk Sectors


AH = 05h Format Disk Cylinder


AH = 06h Format Bad Track


AH = 07h Format Drive


AH = 08h Read Drive Parameters


AH = 09h Initialize Drive Parameters


AH = 0Ah Read Long Sector


AH = 0Bh Write Long Sector


AH = 0Ch Seek to Cylinder


AH = 0Dh Alternate Reset Fixed Disk


AH = 0Eh Diagnostics 1: Read test Buffer


AH = 0Fh Diagnostics 2: Write Test Buffer


AH = 10h Test for Drive Ready


AH = 11h Recalibrate Drive


AH = 12h Controller RAM Diagnostic


AH = 13h Controller Drive Diagnostic


AH = 14h Controller Internal Diagnostic


AH = 15h Read Disk Type


Drive Numbers


The BIOS assigns drive numbers to diskette drives and hard
drives. 00h through 7Fh are reserved for floppy drives. 80h through FFh are
reserved for hard drives. All of the functions used by INT13 reference the
drive number in register DL. 00 is the first floppy drive, 01 is the second
floppy drive, 80 is the first hard drive, 81 is the second hard drive. The
BIOSes on the original IBM PC and XT systems supported 4 floppy drives numbered
00-03 and no hard drives. The AT BIOS added support for 2 hard drives and
reduced the floppy drive support to 2 drives. Some newer BIOSes (such as
Phoenix 4.x) support up to 4 hard drives numbered 80, 81, 82 and 83. BIOS
extensions such as on an EIDE interface card may support additional drives.