Making Sense of Hexdump
I often work with binary data that has a format I could interpret if only I could see it in a human readable form. Most text editors aren’t much use. One way (of many) to solve that is to use the hexdump utility. hexdump is very versatile and allows you to look at the structure inside binary files as you see fit and once you learn how to use it you can apply it quickly to many problems. I think that easily qualifies as cool.
If you’ve used the hexdump man page then you may have found the formatting syntax quite intimidating. Part of the format syntax is pretty much the same as a commonly used function in the C programming language. I found hexdump syntax difficult and I am a C programmer. In truth it isn’t really difficult to understand but without an example to use while reading the documentation it may not be obvious how the formatting options are applied.
Contents:
- Basic usage
- Viewing only part of a file
- Example dumping a partition table using hexdump
- Introducing the format string
- Getting output in lines
- Dumping blocks of data (m/n)
- Splitting out individual elements of the structured data
- Decimal output
- 32 bit data
- Signed/unsigned
- Adding extra text to the output
- Put it in a shell script with some layout
- It’s a 64 bit world!
- Text in a file
- Files with text and numeric data
- Numbers with a fractional part
- I no longer know where I am in the file
- Custom field widths
- Putting it all together
The place to start is with a guess and just give hexdump the name of a data file and get whatever default format hexdump uses:
# hexdump mydata 0000000 457f 464c 0101 0001 0000 0000 0000 0000 0000010 0002 0003 0001 0000 8430 0804 0034 0000 0000020 22ec 0000 0000 0000 0034 0020 0008 0028 * 0001030 0027 0024 0006 0000 0034 0000 8034 0804
What you are seeing there is the binary contents of the mydata file shown as lines of 8 individual 16-bit values in hexadecimal. The first number on each line is the starting offset in the file for the first of the 8 following values on that line. The * indicates that all of the lines from 0000030 to 0001020 inclusive would be the same value as the 0000020 line and that keeps the output condensed for large files with long and aligned repeating sequences. Be aware that, with no command line options, hexdump will dump the entire file to screen so use it with care on large files.
The first 16 bits of the example file are the hexadecimal value 457f. It’s important to note that the value shown is as interpreted on an Intel x86 CPU. On some other types of CPU that value could be output as 7f45 due to differences in something called byte ordering. If you are not familiar with the term it’s probably worth reading about it, for example on Wikipedia.
If 16-bit values isn’t really what you were looking for then a quick review of the hexdump man page indicates there are several switches that select some other pre-defined output formats. There’s only one other option, -C, that produces hexadecimal output:
# hexdump -C mydata 00000000 7f 45 4c 46 01 01 01 00 00 00 00 00 00 00 00 00 |.ELF............| 00000010 02 00 03 00 01 00 00 00 30 84 04 08 34 00 00 00 |........0...4...| 00000020 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| * 00001030 27 00 24 00 06 00 00 00 34 00 00 00 34 80 04 08 |'.$.....4...4...|
Now you can see individual byte values and even see the text for those parts of the file that are normal text characters. So, now I would guess from the “ELF” signature that this file is probably a program executable.
Viewing only part of a file
If you have a really big file where you are only interested in seeing some of the data you can use the -n option to specify how many bytes to dump. The -n 32 in the following example causes hexdump to show only 32 bytes of the file.
# hexdump -C
-n 32
mydata 00000000 7f 45 4c 46 01 01 01 00 00 00 00 00 00 00 00 00 |.ELF............| 00000010 02 00 03 00 01 00 00 00 30 84 04 08 34 00 00 00 |........0...4...| 00000020
If you are only interested in the data some distance into the file you can use the -s option to specify how far into the file to start dumping from. The -s 16 in the following example causes hexdump to show the content of the file beginning 16 bytes in from the start and continuing to the end of the file. :
# hexdump -C
-s 16
mydata 00000010 02 00 03 00 01 00 00 00 30 84 04 08 34 00 00 00 |........0...4...| 00000020 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| * 00001030 27 00 24 00 06 00 00 00 34 00 00 00 34 80 04 08 |'.$.....4...4...|
You can also combine these two switches to see a subset of the data some offset into the file. In this example, using both -s 16 and -n 32 causes hexdump to show 32 bytes of the file beginning 16 bytes in from the start.:
# hexdump -C
-s 16 -n 32
mydata 00000010 02 00 03 00 01 00 00 00 30 84 04 08 34 00 00 00 |........0...4...| 00000020 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| 00000030
You can stop hexdump from replacing duplicate lines with a * by using the -v option. In the following example -v causes the previously collapsed duplicate rows at offsets 30 hex and 40 hex to be displayed:
# hexdump -C -s 16 -n 64
-v
mydata 00000010 02 00 03 00 01 00 00 00 30 84 04 08 34 00 00 00 |........0...4...| 00000020 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| 00000030 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| 00000040 ec 22 00 00 00 00 00 00 34 00 20 00 08 00 28 00 |."......4. ...(.| 00000050
That covers the basic usage of hexdump and it has been sufficient for all my needs until recently.
Example dumping a partition table using hexdump
I wanted to dump the partition table on a disk so that I could mount it via a loopback device. It’s easy to get the necessary information from fdisk by switching it to show sector sized units but I thought I’d play with hexdump to force myself to use the format syntax. This is where we are headed:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3d" 1/1 " %02x" 3/1 " %3d" 2/4 " %9d" "\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 63 16771797 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0
The partition table consists of four 16-byte entries at offset 446 bytes into the disk (i.e. in the first sector of a typical disk). Here’s an example from an image of a disk that had a single partition:
# hexdump -C -s 446 -n 64 diskdump.ddimg 000001be 00 01 01 00 83 fe ff ff 3f 00 00 00 d5 ea ff 00 |........?.......| 000001ce 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| * 000001fe
By coincidence the partition table entry length and the default line length of hexdump are 16 bytes so each line is a complete partition table entry.
It’s probably better to have hexdump show the whole of the block rather than condense the duplicate lines:
# hexdump -v -s 446 -n 64 diskdump.ddimg 000001be 00 01 01 00 83 fe ff ff 3f 00 00 00 d5 ea ff 00 |........?.......| 000001ce 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 000001de 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 000001ee 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 000001fe
It’s not especially difficult to decode that by hand based on published information about the format of partition table entries but you can make hexdump do the work on its own and learn enough about hexdump to be able to apply it to other things in the future. For that it is necessary to dig into the formatting syntax.
Introducing the format string
A format string is specified with the -e command line option. The hexdump man page describes or references all the syntax elements and gives some examples but it takes some work to determine how to do something not in the examples. Here’s something really simple that just outputs the whole table as space separated single byte values in hexadecimal:
# hexdump -s 446 -n 64 -v
-e ‘1/1 ” %02X”‘
diskdump.ddimg 00 01 01 00 83 FE FF FF 3F 00 00 00 D5 EA FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00#
You’ll recognize the use of -n, -s and -v so that it dumps 64 bytes at offset 446 with no collapsing of duplicates. The rest is the format expression and the file to dump. Note that the format expression is enclosed in single quotes: ‘. The 1/1 means process one byte in groups of one. The n/m syntax is specific to hexdump and there are a few other hexdump specifics but the element beginning with a % comes from the C printf function family, it is referred to in the hexdump man page and you can look at the syntax in more detail with man 3 printf.
In the example, the “ %02X” means output a space followed by the value of a single byte as two hexadecimal digits with a leading zero. The % means insert values from the file at this point in the output and what follows the % describes how to do that. The X means output in hexadecimal and the 2 means output as a two digit number and the zero means output leading zeros when the number is less than two digits long.
The use of an uppercase X means that the A through F digits in hexadecimal numbers will be in uppercase. Using a lowercase x causes those digits to be output as a through f in lowercase.
Getting output in lines
The # at the end of the output in the example above is the next shell prompt, I didn’t include any end of line formatting in the example so I’ll fix that first. The \n character sequence is used to specify a new line and also comes from the C printf family. But I’ll get 64 lines of output with one byte value on each line if I just add a \n:
# hexdump -s 446 -n 64 -v -e '1/1 " %02X"
“\n”
' diskdump.ddimg 00 01 01 00 83 FE FF FF 3F 00 etc
That’s because the whole quoted expression is parsed in full repeatedly until the whole of the value for the -n switch is satisfied. The example above the expression processes one byte one time and adds a line feed. That needs to be done 64 times to meet the -n 64 requirement.
Dumping blocks of data (m/n)
If we go back to that 1/1 we can tell hexdump to output bytes in groups of 16 (the size of one partition table entry) by specifying 16/1 instead of 1/1:
# hexdump -s 446 -n 64 -v -e '
16/1
" %02X" "\n"' diskdump.ddimg 00 01 01 00 83 FE FF FF 3F 00 00 00 D5 EA FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
The 16/1 isn’t reversible, the first number is the number of elements to process in a group and the second number is the element length. Sixteen single byte values is what I am looking for. The other way around, 1/16, would be a single sixteen byte value. That results in a “bad byte count for conversion character X” error. The value after the / is the number of bytes to process for each %02X and the value before the / is the number of times to repeat that operation. %02X will not decode a 128 bit (16 byte) value, hence the error. Therefore, 16/1 is the appropriate syntax.
Splitting out individual elements of the structured data
Now I need to fill in some details. The first byte of a partition table is a status value that indicates if the partition is bootable or not. I’ll separate it out as its own decode option by handling the first byte in isolation and the remaining 15 bytes per partition as a group:
# hexdump -s 446 -n 64 -v -e '
1/1 “%02x” 15/1
" %02X" "\n"' diskdump.ddimg 00 01 01 00 83 FE FF FF 3F 00 00 00 D5 EA FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
That doesn’t look any different of course but I have barely started. Next in a partition table entry is a relic of the first PCs with hard disks. There are three individual byte values that specify the starting head, sector and cylinder (least significant 8 bits only) location for the partition. That’s a group of three single byte values:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x"
3/1 ” %02x” 12/1
" %02X" "\n"' diskdump.ddimg 00 01 01 00 83 FE FF FF 3F 00 00 00 D5 EA FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Decimal output
That’s still no different, the head, sector and cylinder have numerical significance so I’ll show them in decimal for human readability:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 "
%3d
" 12/1 " %02x" "\n"' diskdump.ddimg 00 1 1 0 83 fe ff ff 3f 00 00 00 d5 ea ff 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00
The new format element is “ %3d”, that means output a space followed by a three character wide field in decimal. Byte values can be between zero and 255 in decimal so can’t be longer than three digits. Since the numeric value is relevant I dropped the zero prefix without dropping the length prefix and get a space padded, right justified field. You can see that partition 1 has a zero status code and begins on cylinder zero, head 1, sector 1. That is the normal location for the first partition on a disk.
Next is the partition type, another single byte:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3d"
1/1 ” %02x” 11/1
" %02x" "\n"' diskdump.ddimg 00 1 1 0 83 fe ff ff 3f 00 00 00 d5 ea ff 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00 00 0 0 0 00 00 00 00 00 00 00 00 00 00 00 00
Next is the head, sector and cylinder (least significant 8 bits only) location of the end of the partition. I’ll do those in decimal again:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3d" 1/1 " %02x"
3/1 ” %3d” 8/1
" %02x" "\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 3f 00 00 00 d5 ea ff 00 00 0 0 0 00 0 0 0 00 00 00 00 00 00 00 00 00 0 0 0 00 0 0 0 00 00 00 00 00 00 00 00 00 0 0 0 00 0 0 0 00 00 00 00 00 00 00 00
32 bit data
Now the structure is beginning to show. After that I get to the values that will be useful in mounting the file systems and for which single byte output is not useful. The start sector number (logical block address, or LBA) and length (in sectors) for the partition are given as 32 bit values:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3d" 1/1 " %02x" 3/1 " %3d"
2/4 ” %08x”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 0000003f 00ffead5 00 0 0 0 00 0 0 0 00000000 00000000 00 0 0 0 00 0 0 0 00000000 00000000 00 0 0 0 00 0 0 0 00000000 00000000
Note that the latest change adds a 2/4 count specifier, that means 2 instances of a 4 byte (32 bit) value. The %08x format means output the value as an 8 digit hexadecimal number with leading zeros. Since the LBA values have numeric significance it will be more useful in decimal and without the zero paddingl
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3d" 1/1 " %02x" 3/1 " %3d" 2/4 "
%9d”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 63 16771797 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0
The longest 32 bit decimal number has nine digits, versus eight digits for the same number in hexadecimal so I changed the “ %08x” to “ %9d”.
Signed/unsigned
That’s all I needed but I was a little careless, though it didn’t affect the output this time. The d format character says the value is signed (can be positive or negative). Partitions can’t have negative sector, head, cylinder or LBA values. On most platforms an integer value can’t be identified as signed or unsigned based only on the values of the bytes. External knowledge is required that states whether to interpret a value as signed or unsigned. Since I know the values are all unsigned then everywhere I had a d in a % format I should replace it with a u to indicate the value is unsigned (positive values only):
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1
” %3u”
1/1 " %02x" 3/1
” %3u”
2/4
” %9u”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 63 16771797 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0 00 0 0 0 00 0 0 0 0 0
Adding extra text to the output
You can have literal text of your choosing in the output from hexdump. We’ve actually already had lots of it in the spaces and \n values that we’ve used. You can add other things and make the output more meaningful:
# hexdump -s 446 -n 64 -v -e '1/1
“Partition:| %02x”
3/1
” | %3u”
1/1 "
| %02x”
3/1 "
| %3u”
2/4 "
| %9u”
"\n"' /data/vms/vmware/vmdk/2raw/rawfromvmdk.img Partition:| 00 | 1 | 1 | 0 | 83 | 254 | 255 | 255 | 63 | 16771797 Partition:| 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0 Partition:| 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0 Partition:| 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0
Those changes gave us column borders and an introductory word on each line.
Put it in a shell script with some layout
So, now I have a hexdump command line that will dump a partition table. Here it is in a shell script with some cell borders added:
#!/bin/bash # Dump what would be the partition table of the command line specified # file if that file was partitioned media echo echo 'PARTITION TABLE IN' $1 echo echo ' -------Start------- --------End--------' echo 'St | Hd | Cyl | Sec | Tp | Hd | Cyl | Sec | LBA Start | LBA End' echo '---|-----|-----|-----|----|-----|-----|----------------------------' hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " | %3u" 1/1 " | %02x" 3/1 " | %3u" 2/4 " | %9u" "\n"' $1 echo '-------------------------------------------------------------------' echo echo
Save it as partdump somewhere in the path, change the mode to executable and you have a cool new tool:
# partdump diskdump.ddimg PARTITION TABLE IN diskdump.ddimg -------Start------- --------End-------- St | Hd | Cyl | Sec | Tp | Hd | Cyl | Sec | LBA Start | LBA End ---|-----|-----|-----|----|-----|-----|---------------------------- 00 | 1 | 1 | 0 | 83 | 254 | 255 | 255 | 63 | 16771797 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0 00 | 0 | 0 | 0 | 00 | 0 | 0 | 0 | 0 | 0 -------------------------------------------------------------------
It’s a 64 bit world!
As nice as that is, it barely scratches the surface of what you can do with hexdump. All we’ve done is handle 8 bit and 32 bit values in hexadecimal and signed/unsigned decimal with various field widths and value layouts such as with leading zeros or spaces. These are integer operations. These days you have to be able to deal with values longer than 32 bits. The printf syntax for 64 bit values isn’t necessary in hexdump. You can just specify that you are working with a 64 bit (8 byte) value and what output format you want. For example, if the LBA values happened to be a single 64 bit you could do this to see them in hexadecimal – note that this is no longer a valid interpretation of a partition table:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3u" 1/1 " %02x" 3/1 " %3u"
1/8 “
%x" "\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 ffead50000003f 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0
Note the value ffead50000003f at the end was formatted as a single 8 byte value (1/8) and output in hexadecimal %x. You could use %16x to preserve the numeric column alignment:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3u" 1/1 " %02x" 3/1 " %3u" 1/8 "
%16x”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 ffead50000003f 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0
And you can include leading zeros:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3u" 1/1 " %02x" 3/1 " %3u" 1/8 " %016x" "\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 00ffead50000003f 00 0 0 0 00 0 0 0 0000000000000000 00 0 0 0 00 0 0 0 0000000000000000 00 0 0 0 00 0 0 0 0000000000000000
And you can use signed decimal instead of hexadecimal:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3u" 1/1 " %02x" 3/1 " %3u" 1/8 "
%20d”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 72034319610150975 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0
And you can use unsigned decimal instead of hexadecimal:
# hexdump -s 446 -n 64 -v -e '1/1 "%02x" 3/1 " %3u" 1/1 " %02x" 3/1 " %3u" 1/8 "
%20u”
"\n"' diskdump.ddimg 00 1 1 0 83 254 255 255 72034319610150975 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0 0
Text in a file
Not all values in a binary file are numeric, the file may also contain text. hexdump provides a way to show that too. Unlike the C printf family of functions, you have to know in advance how long the text is and there are many ways to achieve the same thing. Going back to the ELF executable file that I started with, you could just dump the signature as-is:
# hexdump
-s 1 -n 3
-v -e
‘1/3 “%s” “\n”‘
mydata ELF
You get the same effect with every one of these:
# hexdump -s 1 -n 3 -v -e '3/1
“%c”
"\n"' mydata ELF # hexdump -s 1 -n 3 -v -e '3/1
“%_p”
"\n"' mydata ELF # hexdump -s 1 -n 3 -v -e '3/1
“%_c”
"\n"' mydata ELF # hexdump -s 1 -n 3 -v -e '3/1
“%_u”
"\n"' mydata ELF
There are many other examples that will output the same three characters but with different arrangements of line feeds. The _p, _c and _u examples are specific to hexdump and not part of the printf syntax. The _p causes characters that have no standard glyph to be output as a . – we can see that by going a couple of bytes beyond the F in the ELF signature:
# hexdump -s 1 -n 5 -v -e '5/1 "%c" "\n"' mydata ELF # hexdump -s 1 -n 5 -v -e '5/1 "%_p" "\n"' mydata ELF..
Note that the first one, using the printf standard %c displayed nothing for the two characters following the ELF but the second one displayed a dot for each character after the ELF. That helps with alignment and it indicates when the data changes from text (printing characters) to non-text (non-printing characters).
The _c example outputs the text in the host’s default character set and displays non-printing characters by value in octal (base 8):
# hexdump -s 1 -n 5 -v -e '5/1 "%_c" "\n"' mydata ELF002001
So, the dots after the ELF are the byte values 2 and 1 (the octal is output here as three digits).
The _u example outputs the text in US ASCll with control characters (those with decimal byte values less than 32 and the value 255) displayed by their three-character control name:
# hexdump -s 1 -n 5 -v -e '5/1 "%_u" "\n"' mydata ELFstxsoh
That’s a Start of TeXt control character followed by a Start Of Heading control character. In this file the byte values don’t represent those ASCII control characters but you can see how you could use %s, %_p, %_c and %_u as appropriate for different types of input files.
Files with text and numeric data
Just as we combined 8, 32 and 64 bit %x output with each other and with 8, 32 and 64 bit decimal values earlier, we can combine numeric and text values on the same command line:
# hexdump -n 6 -v -e
‘1/1 “%02x ” 3/1 “%_u” 2/1 ” %02x”
"\n"' mydata 7f ELF 02 01
I removed the -s 1 so that you could see the text in between some other types of data.
Numbers with a fractional part
In the programming world they are known as floating point numbers. hexdump can display them but there is no guarantee that a floating point value in a file is in the native floating point format of the host. Nevertheless. If you knew there was a floating point value 5213 bytes into that ELF file (which there isn’t in reality):
# hexdump -s 5213
-n 8 -v -e ‘1/8 “%f “
"\n"' mydata 311143424.000000
Once again you can combine it with other format options to look at sequential elements of the file that have different formats. The default length of the %f conversion is 8 bytes so you can actually drop the 1/8 from the example because there are no other format values and the dump length is 8. If the number has too many digits then you can show it in exponential format:
hexdump -s 5295 -n 8 -v -e '1/8
“%g”
"\n"' mydata 1.07497e-255
If you prefer the e to be uppercase you can use %G:
hexdump -s 5295 -n 8 -v -e '1/8
“%G”
"\n"' mydata 1.07497E-255
All three of these floating point formats can also operate on 4 byte floating point types:
hexdump -s 5255 -n 8 -v -e '1/4 "%G" "\n"' mydata 1.67511E-10 1.07861E-38
I no longer know where I am in the file
When you decode a large number of elements in a file you can end up with so much output that you can’t tell where in the file a given value is. In the first examples you saw hexdump output the offset into the file in the first column:
# hexdump mydata 0000000 457f 464c 0101 0001 0000 0000 0000 0000 0000010 0002 0003 0001 0000 8430 0804 0034 0000
You can add the same to your own decoding adding a _a prefix to any of the integer numeric format types: x, d, o. If the sample file had 8 floating point values 4091 bytes into the file and I wanted to know the location of each one:
# hexdump -s 4091 -n 64 -v -e '
“%_ax | “
1/8 "%G" "\n"' mydata ffb | 2.56934 1003 | 2.57715 100b | 2.58496 1013 | 2.59277 101b | 0 1023 | 0 102b | 1.29478E+16 1033 | 9.34901E+25
I included the bar so that you could see the problem when the value overflows to add a new digit. We can use the same prefixes to %_ax and %_ad as we could to the normal %x and %d:
# hexdump -s 4091 -n 64 -v -e '
“%08_ax | “
1/8 "%G" "\n"' mydata 00000ffb | 2.56934 00001003 | 2.57715 0000100b | 2.58496 00001013 | 2.59277 0000101b | 0 00001023 | 0 0000102b | 1.29478E+16 00001033 | 9.34901E+25
Decimal is probably more useful for human readability:
# hexdump -s 4091 -n 64 -v -e '
“%9_ad | “
1/8 "%G" "\n"' mydata
4091 | 2.56934
4099 | 2.57715
4107 | 2.58496
4115 | 2.59277
4123 | 0
4131 | 0
4139 | 1.29478E+16
4147 | 9.34901E+25
There’s a special version of the file-position option that will show the position of the next byte in the file after the last value output, _A:
hexdump -s 4091 -n 64 -v -e '"%9_ad | " 1/8 "%G" "\n"
“%_Ad \n”
' mydata
4091 | 2.56934
4099 | 2.57715
4107 | 2.58496
4115 | 2.59277
4123 | 0
4131 | 0
4139 | 1.29478E+16
4147 | 9.34901E+25
4155
The 4155 is the effect of the %_Ad, the extra indent was deliberate. The _A option can be useful when you are building output from multiple hexdump command lines and want to know what -s value to use on the next command without having to calculate it.
Custom field widths
When specifying the field widths after a % you can use another syntax to describe the overall width to output the value in and how many digits of output to include. For example, the following are equivalent:
# hexdump -s 5091 -n 32 -v -e '1/4 "%08x" "\n"' mydata 2e697472 752f0053 732f7273 702f6372 616b6361 2f736567 4c495542 6c672f44 # hexdump -s 5091 -n 32 -v -e '1/4
“%8.8x”
"\n"' mydata 2e697472 752f0053 732f7273 702f6372 616b6361 2f736567 4c495542 6c672f44
In the second one I replaced %08x with %8.8x. In the second syntax, the number before the dot is the overall width for the field and the number after the dot is the number of digits of output to generate. The value is output right justified if the first number is larger than the second. So, we could have the 8 digit hex value in a 10 character field:
# hexdump -s 5299 -n 32 -v -e '1/4
“%10.8x”
"\n"' mydata 0b002402 030b3e0b 0300000e 0b0b0024 08030b3e 24040000 3e0b0b00 0500000b
Whatever the value of the first or second number, the whole length implied by the input length and the format character is output, so you can’t fool hexdump with %0.0x:
# hexdump -s 5299 -n 32 -v -e '1/4 "%0.0x" "\n"' mydata b002402 30b3e0b 300000e b0b0024 8030b3e 24040000 3e0b0b00 500000b
The values were output in full and left justified.
Putting it all together
The following is a really big example that uses lots of hexdump commands with shell variables, arithmetic and logic operations to dump a packet from a tcpdump file down to the tcp header – inserting the correct offset for some other tcpdump file is left as an exercise for the reader:
#!/bin/bash
# Decode a UDP packet raw binary dataDone
let offs=40
declare -i ws
declare -i end
declare -i rmndr
declare -i byteval
declare -i bitws
NetworkOrder16Bit() {
ws=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws*=256))
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws+=byteval))
}
NetworkOrder32Bit() {
ws=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws*=256))
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws+=byteval))
((ws*=256))
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws+=byteval))
((ws*=256))
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws+=byteval))
}
echo
echo
end=$offs+74
#Start with the destination MAC address
hexdump -s $offs -n 6 -e '"Ether Dest: " 5/1 "%02x:" 1/1 "%02x\n"' test2.pkts
((offs+=6))
# Then the source
hexdump -s $offs -n 6 -e '"Ether Src: " 5/1 "%02x:" 1/1 "%02x\n"' test2.pkts
((offs+=6))
# Next is the packet type, but its in network byte order so we can't
# use '1/2 "%04x"' - we may be run on a system with another byte order and you'll
# get different output
hexdump -s $offs -n 2 -e '"Ether Type: " 2/1 "%02x" "\n"' test2.pkts
((offs+=2))
# IP header
# Get the version/len value into a variable we can do arithmetic with
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
# The most significant 4 bits are the IP version
((bitws=byteval&0xF0))
((bitws/=16))
echo "IP Version: $bitws"
# Least significant 4 bits are the header length
((bitws=byteval&0x0F))
((bitws*=4))
echo "Header Length: $bitws"
# Type of service
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
# Precedence
((bitws=byteval&0xE0))
((bitws/=32))
echo "Precedence: $bitws"
# Delay
((bitws=byteval&0x10))
if (( bitws ))
then
echo "Minimize delay"
fi
# Throughput
((bitws=byteval&0x08))
if (( bitws ))
then
echo "High throughput"
fi
# Reliability
((bitws=byteval&0x08))
if (( bitws ))
then
echo "High reliability"
fi
# Cost
((bitws=byteval&0x08))
if (( bitws ))
then
echo "Minimize cost"
fi
# Message length - in network byte order
NetworkOrder16Bit
echo "Message Length: $ws"
# ID
hexdump -s $offs -n 2 -e '"ID: " 2/1 "%02x" "\n"' test2.pkts
((offs+=2))
# Flags
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
# DF
((bitws=byteval&0x40))
if (( bitws != 0))
then
echo "Don't fragment"
else
# More fragments
((bitws=byteval&0x20))
if (( bitws ))
then
echo "More fragments"
fi
# Fragment number
((ws=bitws&0x1F))
((ws*=256))
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((ws+=byteval))
echo "Fragment number: $ws"
fi
((offs+=1))
hexdump -s $offs -n 1 -e '"TTL: " 1/1 "%u\n"' test2.pkts
((offs+=1))
hexdump -s $offs -n 1 -e '"Protocol: " 1/1 "%u\n"' test2.pkts
((offs+=1))
hexdump -s $offs -n 2 -e '"Header checksum: " 2/1 "%02x" "\n"' test2.pkts
((offs+=2))
hexdump -s $offs -n 4 -e '"Source IP Addr: " 3/1 "%02u." 1/1 "%02u\n"' test2.pkts
((offs+=4))
hexdump -s $offs -n 4 -e '"Dest IP Addr: " 3/1 "%02u." 1/1 "%02u\n"' test2.pkts
((offs+=4))
# TCP header
NetworkOrder16Bit
echo "Source Port: $ws"
NetworkOrder16Bit
echo "Dest Port: $ws"
hexdump -s $offs -n 4 -e '"Sequence Number: " 4/1 "%02x" "\n"' test2.pkts
((offs+=4))
hexdump -s $offs -n 4 -e '"ACK Number: " 4/1 "%02x" "\n"' test2.pkts
((offs+=4))
ws=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((ws/=16))
((ws*=4))
echo "Header Length: $ws"
byteval=`hexdump -s $offs -n 1 -e '1/1 "%u"' test2.pkts`
((offs+=1))
((bitws=byteval&0x80))
echo -n "Flags: "
if (( bitws ))
then
echo -n "CWR "
fi
((bitws=byteval&0x80))
if (( bitws ))
then
echo -n "CWR "
fi
((bitws=byteval&0x40))
if (( bitws ))
then
echo -n "ECN "
fi
((bitws=byteval&0x20))
if (( bitws ))
then
echo -n "URG "
fi
((bitws=byteval&0x10))
if (( bitws ))
then
echo -n "ACK "
fi
((bitws=byteval&0x08))
if (( bitws ))
then
echo -n "PSH "
fi
((bitws=byteval&0x04))
if (( bitws ))
then
echo -n "RES "
fi
((bitws=byteval&0x02))
if (( bitws ))
then
echo -n "SYN "
fi
((bitws=byteval&0x01))
if (( bitws ))
then
echo -n "FIN "
fi
echo
NetworkOrder16Bit
echo "Window Size: $ws"
hexdump -s $offs -n 2 -e '"Checksum: " 2/1 "%02x" "\n"' test2.pkts
((offs+=1))
# Some more of the packet
echo
echo
hexdump -s $offs -n 128 -C test2.pkts
echo
echo
That shouldn’t be considered as a viable packet analyzer but it shows most of the elements necessary to build a custom tool to investigate some arbitrary binary data. Hopefully there’s something in that you will find useful when applying hexdump to your own binary file decode problems.
Related Articles
Mar 27th, 2023
SUSE Joins the Confidential Computing Consortium
Comments
I’ve been looking for a while for some good examples of using the full power of hexdump, and you certainly did a fantastic job of doing just that! Thank you!
It was a real great article showing how best you can make use of existing utilities.
Wonderful job ! Thanks.
The man page for hexdump is difficult to understand. The author of this article, Dmair – was very polite by ascribing the man page as intimidating. I think this article should be made available in the Linux Documentation Project, or at least a link.
This artilce is an example of how to give succinct and detailed explanation of a utility with an awkward syntax.
It would have been a trial and error process to work out how to use the -e formatting capability without this. Thankyou, thankyou Dmair.
Thanks so much for the detailed description and examples of hexdump. Like you said, without examples, the manual is really confusing.
Thanks a million!
PS: I intentionally created an account here to thank you. 🙂 Nonetheless I think there should be many more helpful articles in this communities site.
Grateful to the article. This helped me to browse an on disk FS B+Tree without needing to develop tools using high level languages.
Thank you for taking the time to write this very comprehensive and easy-to-understand write-up on using HexDump.
Cheers maid
How can I deal with byte order issues?
If I do a hexdump of my file the first 8 bytes are
03 14 47 40 b7 4e 48 00
If I do a hexdump -C the first 8 bytes are
14 03 40 47 4e b7 00 48
In my case, I need to use hexdump -C, but I need to skip the first three bytes, then I need to reverse the order of each word. So I am looking for
00 b7 4e 47
echo -ne “\x00\xb7\x4e\x47” | hexdump -e ‘1/4 “%f” “\n”‘ gives me 52919.000000 which is a date of 5/29/2019.
Is there a way to deal with these byte order issues from within hexdump?
Worst case, there’s nothing to prevent doing it the way it was done in the packet dump (network byte order) example in the document. Use a shell script, keep a position pointer, like $0ffs in the document and hexdump -n 1 at a time using your preferred order for -s. It would make -C output a mess (one character per-line) but there’s nothing to prevent re-implementing -C in your preferred order from a script calling hexdump using -n 1 twice per-line, once at a time with %02X for each hex byte output and once at a time afterwards with %_c or %_p.
Great article. Thanks.
…or use a tool like od instead. It has an endian choice: –endian={big|little}, e.g.:
> echo “hello” | od -A x -t x2z –endian=big -v
000000 6865 6c6c 6f0a >hello.<
> echo “hello” | od -A x -t x2z –endian=little -v
000000 6568 6c6c 0a6f >hello.<
000006