carfield.com.hk: hardware

Tue Jul 03 16:00:00 GMT 2001 From /document/microsoft/hardware

fat

To accommodate the larger number of possible clusters, the directory
entry for each file must have 4
bytes for the starting cluster of the
file, rather than the 2 bytes needed
under FAT16. Traditionally, each
directory entry is a 32-byte record
(as shown in Figure 1). In the
middle of the record are 10 bytes
(bytes 12 to 21) that Microsoft has
reserved for its own future use.
Two of those bytes are now used to accommodate
the extra bytes needed to specify the starting cluster
under FAT32.

The operating system has always kept two copies
of the FAT on disk but used only one. With FAT32,
the OS can now use either copy. Another change is
that the root directory, which used to be of fixed size
and location on disk, is now free to grow as
necessary, like a subdirectory. There is no longer a
limit on the number of directory entries in the root
directory. That's especially important, because each
long filename uses multiple directory entries. (For
details, see "How Windows 95 Stores Long
Filenames," PC Magazine, June 25, 1996.)

The combination of a movable root directory and the
use of both FAT copies offer the potential of
smooth, dynamic resizing of a hard disk--to shrink a
partition, for example, to make room for some other
operating system. The new approach is safer than
ones used by third-party programs to resize
partitions under FAT16. (While Windows doesn't
include a utility that takes advantage of the potential
of FAT32, third-party products that do have been
announced. They include PowerQuest's Partition
Magic 3.0 and Quarterdeck's PartitionIt.)

(google search) (amazon search)

hardware

Fat

Fat_dir

fat

ntfs