FAT32 File Carving
How to manually carve a deleted file from a FAT32 file system using FTKimager.
Last updated
How to manually carve a deleted file from a FAT32 file system using FTKimager.
Last updated
Let's begin the process of manually carving a deleted file from a FAT32 filesystem but as always find resources outside of this article to help you further your understanding and remember Google is your friend.
It is recommended having a read of the File Carving page found here to have a baseline understanding of concepts.
The VBR (Volume Boot Record) is found in logical sector 0 of the partition/volume. The VBR contains information about the volume/partition. We are interested in the two pieces of information: Bytes per Sector, and Sectors per Cluster.
The table below shows us the location of these two values.
Offset (Hex)
Size (Bytes)
Description
0x0B
2
Bytes per Sector
0x0D
1
Sectors per Cluster
I have highlighted the two values we are looking at in the picture below.
The Bytes per Sector shows the value 0x0002 which read as little-endian (the bytes are read right to left) is 0x200 and this in decimal is 512 bytes in each sector.
The Sectors per Cluster shows the value 0x02 which as it is only one byte can be read as it appears, and in decimal is 2. We now know that 2 sectors make up a cluster and our sectors are 512 bytes in size meaning a cluster is 512 x 2 which in turn is 1,024 bytes in size.
We can see on the left-hand side the evidence tree and in particular partition 1 with sub-folder ‘root’ which is open, this is the root directory where our files reside on the filesystem. The file list shows us what files are within the root folder.
Notice there is a file named ‘Secret Deal.docx’ with an icon showing a cross; this is how FTKimager shows a deleted file is present, but we can’t open it as a result of it being deleted.
If we turn our attention to the hexadecimal in the bottom pane, we can see how the root directory is set out in 16-byte rows. You can see in the pictures below the short and long file name of the deleted ‘Secret Deal.docx” file.
The long filename is the first 64 bytes in this case but can vary in the number of bytes used taking up more than 1 directory entry, followed by the short filename which is always 32 bytes in length. Other than the icon in the File List view, we know that this file has been deleted because of the status byte 0xE5 at the start of each 32 bytes.
The Long Filename Directory Entry is for storing filenames longer than 8 bytes.
The Short Filename Directory Entry is where we find our other important file information such as Location and File Size etc.
We are mainly interested in the short file name as this is where the filesystem stores the information we require to carve and successfully recover the file. The short file name only stores the first 8 ASCII characters and 3 extension characters; therefore, the long filename is used to store filenames with characters over that limit.
The table below shows how a short filename directory entry is made up.
Offset (Hex)
Size (Bytes)
Description
0x00
8
Filename (the first byte is the status byte)
0x08
3
File Extension
0x0B
1
File Attribute Byte
0x0C
1
Lower Case Flags
0x0D
1
File Creation Time in Milliseconds
0x0E
2
File Creation Time
0x10
2
File Creation Date
0x12
2
Last Accessed Date
0x14
2
High Word Starting Cluster (if zero only the Low Word is used)
0x16
2
Time of Last Write to File
0x18
2
Date of Last Write to File
0x1A
2
Low Word Starting Cluster
0x1C
4
File Size
The high word is 0x0000 meaning we only need the low word value at offset 0x1A to find the starting cluster. The value at offset 0x1A is 0x4300 and this read as little-endian is 0x43 which when converted to decimal is telling us the file starts at cluster 67.
The value at offset 0x1C = 44 62 01 00. This is read as little-endian which is 0x16244 and using the programmer calculator within Windows (which has a hex to dec function) is 90,692 bytes, which is the logical size. Refer to the File Carving Intro here for what the difference between logical and physical size.
We can now divide the file size by cluster size to identify the number of clusters the file uses 90,692/1,024 = 88.57 clusters which we round up to 89 clusters.
The FAT (File Allocation Table) tells us what clusters are in use within the FAT32 filesystem and in the picture below we can see the FAT. The FAT works by showing if a cluster is allocated to a file and how to identify the next cluster it allocates. A FAT32 entry is 4 bytes in length.
For example, the first cluster allocated in the filesystem is cluster 5 at offset 0x14 and it is showing us the next cluster used is cluster 6. This goes on until you reach the hexadecimal value of 0xFFFFFF0F signifying the file ends in this cluster which can be seen at offset 0x108 which is cluster 42 in hexadecimal and cluster 66 in decimal.
Earlier we identified cluster 67 as our starting cluster which is 0x43 in hexadecimal.
We will need to check if the file we are going to carve is contiguous (in sequence) or fragmented meaning we need to locate clusters in different locations.
In the sections File Size and Location, we determined that our file's physical size was 89 clusters and that the starting cluster was 67. I have highlighted cluster 67 within the FAT and the other clusters that make up the file, and as a result, we can say that the file is contiguous.
You may wonder why aren't the highlighted clusters pointing to the next cluster as the example from the initial FAT intro had done so? This is because when a file is deleted the cluster chain is deleted as we can see below.
We have essentially done the heavy lifting now and can finish off with the data carving.
Let's use the "Go to sector/cluster" tool by right-clicking in the hexadecimal of the SLEUTH USB [FAT32] section as shown below. This allows us to navigate to the cluster in question being 67.
Now at cluster 67 which starts at the offset 0x10400, we can use another tool within FTKimager by right-clicking the hexadecimal and selecting "Set Selection Length" as shown below and in this case, we know our logical file size is 90,692 bytes so we will use that.
Once the selection from the starting cluster and the correct size is highlighted then we can right-click once more and use "Save selection..." which allows us to save the raw data as a file and provide an extension so the appropriate program can open it.
In some cases, we can use the extension found in the Short Filename Directory Entry but someone could have given the file in question a different extension for one reason or another. It is best to identify the file signature, also known as a file header, to ensure the correct extension for use with the file.
In this case, I can see 'PK' is visible which already provides me with the clue that it is a compound file and with some comparison work using Gary Kesslers File Signatures website, determined it to be a Microsoft Office file.
We have successfully carved data from a FAT32 filesystem and below we can view the recovered file using Microsoft Word.
I hope you take something away from this article and if you want to follow along to carve the data and recover the file yourself you can download a zip of the image I used at my Github page by clicking here.
