XXD(1)

🎀 XXD(1) 🎀

General Commands Manual — okay bestie, buckle up, we're talking about xxd, the hex dump queen 👑

---

✨ NAME ✨

xxd — make a hex dump or do the reverse. periodt 💅

---

✨ SYNOPSIS ✨

xxd -h[elp]
xxd [options] [infile [outfile]]
xxd -r[evert] [options] [infile [outfile]]

---

✨ DESCRIPTION ✨

So like, xxd creates a hex dump of a given file or standard input. And not only that — it can also convert a hex dump back to its original binary form 🦋. Like uuencode(1) and uudecode(1), it lets you transmit binary data in a mail-safe ASCII representation, but it's got the glow-up advantage of decoding to standard output. Moreover (yes, moreover, we're being fancy), it can be used to perform binary file patching. iconic behavior honestly 💖

---

✨ OPTIONS ✨

First, the ground rules, bestie:

If no infile is given, standard input is read. If infile is specified as a - character, then input is taken from standard input. If no outfile is given (or a - character is in its place), results are sent to standard output. 🫶

Heads up: a "lazy" parser is used which does not check for more than the first option letter, unless the option is followed by a parameter. Spaces between a single option letter and its parameter are optional. Parameters to options can be specified in decimal, hexadecimal or octal notation. So like — -c8, -c 8, -c 010 and -cols 8 are all equivalent. No cap. 🌈

• -a | -autoskip — Toggle autoskip: a single * replaces NUL-lines. Default off. 💤
• -b | -bits — Switch to bits (binary digits) dump, rather than hex dump. This option writes octets as eight digits "1"s and "0"s instead of a normal hexadecimal dump. Each line is preceded by a line number in hexadecimal and followed by an ASCII (or EBCDIC) representation. The command line switch -p does not work with this mode. Can be combined with -i. 🔢
• -c cols | -cols cols — Format <cols> octets per line. Default 16 (-i: 12, -ps: 30, -b: 6). Max 256. No maximum for -ps. With -ps, 0 results in one long line of output. 📏
• -C | -capitalize — Capitalize variable names in C include file style, when using -i. 🔠
• -d — show offset in decimal instead of hex. 🧮
• -E | -EBCDIC — Change the character encoding in the righthand column from ASCII to EBCDIC. This does not change the hexadecimal representation. The option is meaningless in combinations with -r, -p or -i. 🙈
• -e — Switch to little-endian hex dump. This option treats byte groups as words in little-endian byte order. The default grouping of 4 bytes may be changed using -g. This option only applies to the hex dump, leaving the ASCII (or EBCDIC) representation unchanged. The command line switches -r, -p, -i do not work with this mode. 🌀
• -g bytes | -groupsize bytes — Separate the output of every <bytes> bytes (two hex characters or eight bit digits each) by a whitespace. Specify -g 0 to suppress grouping. <Bytes> defaults to 2 in normal mode, 4 in little-endian mode and 1 in bits mode. Grouping does not apply to PostScript or include style. 🍓
• -h | -help — Print a summary of available commands and exit. No hex dumping is performed. 🚪
• -i | -include — Output in C include file style. A complete static array definition is written (named after the input file), unless xxd reads from stdin. Can be combined with -b. 📦
• -l len | -len len — Stop after writing <len> octets. 🛑
• -n name | -name name — Override the variable name output when -i is used. The array is named name and the length is named name_len. 🏷️
• -o offset — Add <offset> to the displayed file position. ➕
• -p | -ps | -postscript | -plain — Output in PostScript continuous hex dump style. Also known as plain hex dump style. 📜
• -r | -revert — Reverse operation: convert (or patch) hex dump into binary. If not writing to stdout, xxd writes into its output file without truncating it. Use the combination -r -p to read plain hexadecimal dumps without line number information and without a particular column layout. Additional whitespace and line breaks are allowed anywhere. Use the combination -r -b to read a bits dump instead of a hex dump. 🔁
• -R when — In the output both the data column (hex, or bits with -b) and the character column are colored with the same color depending on the byte value. Mostly helping to differentiate printable and non-printable characters. when is never, always, or auto (default: auto). When the $NO_COLOR environment variable is set, colorization will be disabled. 🌈
• -seek offset — When used after -r: revert with <offset> added to file positions found in hex dump. 🔍
• -s [+][-]seek — Start at <seek> bytes abs. (or rel.) infile offset. + indicates that the seek is relative to the current stdin file position (meaningless when not reading from stdin). - indicates that the seek should be that many characters from the end of the input (or if combined with +: before the current stdin file position). Without -s option, xxd starts at the current file position. 🎯
• -t — When outputting in C include file style using -i include the terminating zero. 0️⃣
• -u — Use upper-case hex letters. Default is lower-case. 🔡
• -v | -version — Show version string. 📌

---

✨ CAVEATS ✨

⚠️ Bestie, this is the part where we pay attention. No scrolling past. ⚠️

xxd -r has some built-in magic while evaluating line number information. If the output file is seekable, then the line numbers at the start of each hex dump line may be out of order, lines may be missing, or overlapping. In these cases xxd will lseek(2) to the next position. If the output file is not seekable, only gaps are allowed, which will be filled by null-bytes. 💀

xxd -r never generates parse errors. Garbage is silently skipped. 🗑️

When editing hex dumps, please note that xxd -r skips everything on the input line after reading enough columns of hexadecimal data (see option -c). This also means that changes to the printable ASCII (or EBCDIC) columns are always ignored. Reverting a plain (or PostScript) style hex dump with xxd -r -p does not depend on the correct number of columns. Here, anything that looks like a pair of hex digits is interpreted. 👀

Note the difference between

% xxd -i file

and

% xxd -i < file

xxd -s +seek may be different from xxd -s seek, as lseek(2) is used to "rewind" input. A '+' makes a difference if the input source is stdin, and if stdin's file position is not at the start of the file by the time xxd is started and given its input. The following examples may help to clarify (or further confuse!): 😭

Rewind stdin before reading; needed because the cat has already read to the end of stdin.

% sh -c "cat > plain_copy; xxd -s 0 > hex_copy" < file

Hex dump from file position 0x480 (=1024+128) onwards. The '+' sign means "relative to the current position", thus the '128' adds to the 1k where dd left off.

% sh -c "dd of=plain_snippet bs=1k count=1; xxd -s +128 > hex_snippet" < file

Hex dump from file position 0x100 (=1024-768) onwards.

% sh -c "dd of=plain_snippet bs=1k count=1; xxd -s +-768 > hex_snippet" < file

However, this is a rare situation and the use of '+' is rarely needed. The author prefers to monitor the effect of xxd with strace(1) or truss(1), whenever -s is used. 🔬

---

✨ EXAMPLES ✨

Print everything but the first three lines (hex 0x30 bytes) of file.

% xxd -s 0x30 file

Print 3 lines (hex 0x30 bytes) from the end of file.

% xxd -s -0x30 file

Note: The results of the examples below are relevant to the xxd.1 man page as of March 2026 📅

Print 120 bytes as a continuous hex dump with 20 octets per line.

% xxd -l 120 -ps -c 20 xxd.1
2e544820585844203120224d6172203230323622
20224d616e75616c207061676520666f72207878
64220a2e5c220a2e5c222032317374204d617920
313939360a2e5c22204d616e2070616765206175
74686f723a0a2e5c2220202020546f6e79204e75
67656e74203c746f6e79407363746e7567656e2e

Hex dump the first 120 bytes of this man page with 12 octets per line.

% xxd -l 120 -c 12 xxd.1
00000000: 2e54 4820 5858 4420 3120 224d  .TH XXD 1 "M
0000000c: 6172 2032 3032 3622 2022 4d61  ar 2026" "Ma
00000018: 6e75 616c 2070 6167 6520 666f  nual page fo
00000024: 7220 7878 6422 0a2e 5c22 0a2e  r xxd"..\"..
00000030: 5c22 2032 3173 7420 4d61 7920  \" 21st May
0000003c: 3139 3936 0a2e 5c22 204d 616e  1996..\" Man
00000048: 2070 6167 6520 6175 7468 6f72   page author
00000054: 3a0a 2e5c 2220 2020 2054 6f6e  :..\"    Ton
00000060: 7920 4e75 6765 6e74 203c 746f  y Nugent <to
0000006c: 6e79 4073 6374 6e75 6765 6e2e  ny@sctnugen.

Display just the date from the file xxd.1

% xxd -s 0x33 -l 13 -c 13 xxd.1
00000033: 3231 7374 204d 6179 2031 3939 36  21st May 1996

Copy input_file to output_file and prepend 100 bytes of value 0x00.

% xxd input_file | xxd -r -s 100 > output_file

Patch the date in the file xxd.1

% echo "0000034: 3574 68" | xxd -r - xxd.1
% xxd -s 0x33 -l 13 -c 13 xxd.1
00000033: 3235 7468 204d 6179 2031 3939 36  25th May 1996

Create a 65537 byte file with all bytes 0x00, except for the last one which is 'A' (hex 0x41).

% echo "010000: 41" | xxd -r > file

Hex dump this file with autoskip.

% xxd -a -c 12 file
00000000: 0000 0000 0000 0000 0000 0000  ............
*
0000fffc: 0000 0000 41                   ....A

Create a 1 byte file containing a single 'A' character. The number after '-r -s' adds to the line numbers found in the file; in effect, the leading bytes are suppressed.

% echo "010000: 41" | xxd -r -s -0x10000 > file

Use xxd as a filter within an editor such as vim(1) to hex dump a region marked between 'a' and 'z'.

:'a,'z!xxd

Use xxd as a filter within an editor such as vim(1) to recover a binary hex dump marked between 'a' and 'z'.

:'a,'z!xxd -r

Use xxd as a filter within an editor such as vim(1) to recover one line of a hex dump. Move the cursor over the line and type:

!!xxd -r

Read single characters from a serial line

% xxd -c1 < /dev/term/b &
% stty < /dev/term/b -echo -opost -isig -icanon min 1
% echo -n foo > /dev/term/b

Output in C style include output with terminating zero

% echo -n "0123456789" > digits
% xxd -i -t digits
unsigned char digits[] = {
  0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x00
};
unsigned int digits_len = 10;

---

✨ RETURN VALUES ✨

The following error values are returned:

• 0 — no errors encountered. 🥳
• -1 — operation not supported (xxd -r -i still impossible). 😭
• 1 — error while parsing options. 🤨
• 2 — problems with input file. 📥
• 3 — problems with output file. 📤
• 4,5 — desired seek position is unreachable. 🫠

---

✨ SEE ALSO ✨

uuencode(1), uudecode(1), patch(1) 🌸

---

💀 WARNINGS 💀

okay bestie, listen up, this is the part where i hold your face in my hands and look at you very seriously 🫶

this tool's weirdness matches its creator's brain. that's not me being shady, that's literally the documented truth. it's giving chaotic genius energy and we respect it 💅

Use entirely at your own risk. entirely. all of it. no takebacks. 🙈

then the manpage hits you with three little imperatives and honestly? they're a whole vibe:

• Copy files. 🍓
• Trace it. ✨
• Become a wizard. 🧚👑

that's it. that's the onboarding. no cap.

---

🌈 VERSION 🌈

this manual page documents xxd version 1.7 from 2024-05. yes queen, she's documented, she's dated, she's current 💖

---

👑 AUTHOR 👑

(c) 1990-1997 by Juergen Weigert <jnweiger@informatik.uni-erlangen.de>

and then the most iconic license terms ever written, fr fr:

Distribute freely and credit me,
make money and share with me,
lose money and don't ask me.

not me tearing up a little 😭 that's the whole business model. distribute freely and credit me. make money and share with me. lose money and don't ask me. absolute legend behaviour.

manual page started by Tony Nugent <tony@sctnugen.ppp.gu.edu.au> <T.Nugent@sct.gu.edu.au>

small changes by Bram Moolenaar. edited by Juergen Weigert.

the way that this page had a whole glow-up team behind it 💅🦋

---

so that's the page, bestie 🌸 this one's the xxd manual from Mar 2026, sitting in section XXD(1), and the footer's lil source line is just "Manual page for xxd" with that same date — consider it acknowledged, credited, and kissed on the forehead 😘 go hexdump something responsibly (or don't, per the warnings above). mwah 💖✨