Note – this section of the manual is for information only. You do not need to read and understand this large and in depth section to use the driver! However you may want to if you wish to gain an understanding of disk access, the FAT filing system and how this driver works.

Terms used for hard disks and therefore MMC / SD memory cards

Remember when understanding these terms that hard disks uses multiple disks of magnetic material with a read/write head for each side of each disk. Bytes are read from and written to a disks surface in circular paths.

Track

The circular track on one surface of a disk (numbered 0 – #). This is not usually referred to.

Cylinder

All of the tracks in the same position on all of the surfaces (numbered 0 – #). This is not usually referred to other than when determining the parameters of a disk during initialisation.

Head

Each side of a disk has a read / write head (numbered 0 – #). This is not usually referred to other than when determining the parameters of a disk during initialisation.

Sector

This is the fundamental unit of disk mapping – all reading and writing to disks is carried out in sectors. A sector is usually 512 bytes in size, but can be 128 – 1024 bytes.

(Numbered as 1 – # (0 is reserved for identification purposes)).

Cluster

A cluster is a specified group of sectors. It is clusters that are the addressing unit when reading and writing files using the FAT system (i.e. a directory will point to a particular file using the cluster number that contains the start of the file). A cluster may only be used by one file, and large files will use multiple clusters to hold their data. A disk with a large cluster size (lots of sectors per cluster) will mean that disk space is wasted as any unused bytes after the end of a file in its final cluster will not be available for anything else. A disk with a small cluster size means less wastage. However, a small cluster size means a larger FAT table as a FAT table contains an entry for every cluster on a disk (or in the partition if the disk is partitioned), hence the need to FAT32 instead of FAT16 for larger volumes.

The valid range is 1 – 64 sectors per cluster. The first cluster that may be used is number 2 (clusters 0 & 1 are reserved).

The FAT filing system was developed for DOS and DOS thinks of a disk as a linear object, not as it is actually constructed. This means that DOS treats the sectors of a disk as a sequential list of sectors, from the first on the disk to the last. Whilst this made things more complex when writing drivers for hard disks, it makes things easier when dealing with modern flash memory cards as these are linear memory objects.

Byte Ordering

The FAT file system uses ‘little endian’. That is that the first byte read is the least significant byte of a large value, the next byte read is more significant than the last and so on. For example this is how a 32bit value would be stored (with the bit numbers shown):-

byte[3] 31 30 29 28 27 26 25 24 //This is the last byte read from the disk

byte[2] 23 22 21 20 19 18 17 16

byte[1] 15 14 13 12 11 10 09 08

byte[0] 07 06 05 04 03 02 01 00 //This is the first byte read from the disk

The following sections show how the different sections of a disk are organised for FAT16 and FAT32, looking at the disk as a linear memory object (which is how it is addressed). See the following sections for an in depth description of each block.

The Layout of a FAT16 Volume

Then follows further partitions if present:-

Repeated for each partition

Note1 – These cells may repeat or not be present at all.

The Layout of a FAT32 Volume

This is basically the same as for a FAT16 volume, but without the root directory included (and with each block using a different amount of space).

Then if there is more than 1 partition, the additional partitions follow:-

Repeated for each partition

*Note 1 – Shaded cells may repeat or not be present at all.

The first sector of a hard disk is set aside for the Master Boot Record. This is operating system independent. It is located on the first Sector of the disk, at Cylinder 0, Head 0, Sector 1. It contains the partition table, which defines the different sections of your hard drive and if this section of a disk is corrupted it can mean that the disk is dead!

Note – if trying to view the master boot record using PC disk viewing software ensure that you have selected the correct section of the disk. Some software will show you the contents of the first partition by default, not the first sector containing the master boot record.

The first sector of a partition contains a boot record. There are differences between the FAT16 and FAT32 boot records.

FAT16

FAT32

The FAT table (whether FAT16 or FAT32) contains an entry for every cluster on the disk (or partition if the disk is partitioned). Each entry is either 16 bits in size for FAT16, or 32bits in size for FAT32. The contents of an entry may be as follows:-

FAT16 Table Entry Values:-

0×0000 The cluster is free.

0×0001 Reserved

0×0002 – 0xFFF0 This cluster is used. The value indicates the next cluster number for the file.

0xFFF7 Cluster is bad

0xFFF8 – 0xFFFF EOC (End Of Clusterchain) (typically you should use 0xFFFF)

FAT32 Table Entry Values:-

0x#0000000 The cluster is free.

0×0001 Reserved

0×0002 – 0xFFF0 This cluster is used. The value indicates the next cluster number for the file.

0x#FFFFFF7 Cluster is bad

0x#FFFFFF8 – 0x#FFFFFFF EOC (End Of Clusterchain) (typically you should use 0x#FFFFFFF

(The top 4 bits are reserved and will not necessarily be zero. They must be ignored when reading a cluster number but maintained when writing a new value to an entry)

When a file is stored the first available free cluster is found from the FAT table and stored in the files directory entry (see later in this manual). The file is written to the cluster. If it doesn’t fit within the cluster then the next free cluster is found and the new cluster number is written in the previous clusters FAT table entry. This continues until the last cluster that is required for the file (which may be the first cluster if the file will fit within one cluster). The EOC marker is written to the FAT tables for the last cluster to indicate that no further clusters are used.

Therefore when reading a file the start cluster number is determined from the files entry in the directory the file is located in. Then the FAT table is used to find the next cluster that holds the next block of the files data, then the next etc. Whilst the EOC marker indicates that a cluster is the last cluster used to store a file, the exact file size is stored in the files directory entry so that the last used byte number of the file can be determined.

FAT16 FAT Table

Byte address is (Partition Start Address + 512 + #)

FAT32 FAT Table

Byte address is (Partition Start Address + 512 + #)

FAT16 uses 2 FAT tables, one after the other, and FAT32 uses up to 4 FAT tables. This provides a backup in case of corruption of one of the tables. If you change the contents of the FAT table, ensure that all copies are updated (checking for FAT32 to see which tables should be updated).

Location & Size

The first FAT table starts straight after the Boot Record. Therefore the start address of the first FAT table:

= Start address of partition + No of reserved sectors

Each additional FAT table follows straight on after the last. The number of FAT tables is recommended to be 2 due to old systems that assume a value of 2. However the number of FAT tables does not have to be 2 and for flash drives where a backup of the FAT table is redundant only a single table may be used. It is also possible to have more than 2 FAT tables.

A FAT directory is simply a ‘file’ containing a linear list of 32 byte entries. The only special directory, which must always be present, is the root directory. For FAT16 volumes the root directory is located in a fixed location on the disk immediately following the last FAT and is a fixed size in sectors as specified in the Boot Record.

For FAT16 the first sector of the root directory is sector number relative to the first sector of the FAT volume:

For FAT32 the root directory can be of variable size and is a cluster chain just like any other directory. The first cluster of the root directory is specified in the Boot Record.

Each directory entry is 32 bytes and formatted as follows:

Note that bytes 0x0C to 0×15 we’re unused in the original DOS specification and may still be left unused if desired.

Special Markers

If the first byte of a directory entry is 0xE5 then the entry has been erased. If the first byte is 0×00 then the entry has never been used (this can be used to detect the end of the table as all following entries will also be 0×00).

Location & Size

For FAT16 the root directory is located directly after the 2^nd FAT table:

= Start address of partition + No of reserved sectors + (Number of FAT tables x FAT table size)

Its size is specified by the boot record:

= maximum number of root directory entries x 32 bytes per entry

The data area starts straight after the root directory. The only difference between the root folder and any other folders is that the root folder is at a specified location and has a fixed number of entries.

For FAT32 the root directory can be of variable size and is a cluster chain, just like any other directory is. The first cluster of the root directory on a FAT32 volume is stored in the sector specified in the boot record.

For both FAT16 and FAT23, unlike other directories, the root directory itself does not have any date or time stamps, does not have a file name (other than the implied file name “\”), and does not contain “.” and “..” files as the first two directory entries in the directory. The only other special aspect of the root directory is that it is the only directory on the FAT volume for which it is valid to have a file that has only the ‘Volume ID’ attribute bit set.

Date and Time Formats

If date and time are not supported then they should be written as zero. Bytes 22 – 25, time of last write and date of last write, must be supported according to the FAT specification but if a device has no real time clock then this isn’t possible.

Date field

A 16-bit field that is a date relative to 01/01/1980:-

Bits 15:9 Count of years from 1980, valid range 0 – 127 (=1980–2107).

Bits 8:5 Month of year, valid range 1–12 (1 = January)

Bits 4:0 Day of month, valid range 1-31

Time Format.

A 16-bit field with a valid range from Midnight 00:00:00 to 23:59:58:-

Bits 15:11 Hours, valid range 0 – 23

Bits 10:5 Minutes, valid range 0 – 59

Bits 4:0 2-second count, valid range 0–29 (= 0 – 58 seconds)

The remainder of the volume is the data area, which may contain files and directories. It is this area that the FAT tables relate to.

Start Address

For FAT16 the start address of the data area is:-

Start address of partition + Number of reserved sectors + (Number of FAT tables x FAT table size) + Number of root directory sectors

For FAT32 the start address of the data area is:-

Start address of partition + Number of reserved sectors + (Number of FAT tables x FAT table size)

For a given cluster number in the FAT table, the start address of that sector is:-

data area start address + ((FAT table cluster number – 2) x sectors per cluster)

Because sectors per cluster is restricted to powers of 2 (1, 2, 4, 8, 16, 32…), division and multiplication by sectors per cluster can actually be performed via shift operations which is often faster than multiply or divide instructions

(Not applicable to FAT16)

The partition boot record specifies the sector that contains this information block, which can be utilised by a FAT driver to speed up write operations.

Byte address is (Sector Start + #)