Home » What is the 6th character of the password hash on Linux, and why is it often a slash?

What is the 6th character of the password hash on Linux, and why is it often a slash?


Hash format and source

The format of the password hash is $<type>$<salt>$<hash>, where <type> 5 is an SHA-256 based hash. The salt is usually at least 8 characters, (and is in the examples in the question) so the sixth character is part of the salt.

Those hashes are likely generated by a version of the shadow tool suite (src package shadow in Debian, shadow-utils in CentOS)

I tried to find out why, exactly, the code biases the slash. (thanks to @thrig for originally digging up the code.)

TLDR: It’s a bit interesting, but doesn’t matter.

The code generating the salt

In libmisc/salt.c, we find the gensalt function that calls l64a in a loop:

strcat (salt, l64a (random()));
do {
       strcat (salt, l64a (random()));
} while (strlen (salt) < salt_size);

The loop takes a random number from random(), turns it into a piece of a string, and concatenates that to the string forming the salt. Repeat until enough characters are collected.

What happens in l64a is more interesting though. The inner loop generates one character at a time from the input value (which came from random()):

for (i = 0; value != 0 && i < 6; i++) {
    digit = value & 0x3f;

    if (digit < 2) {
        *s = digit + '.';
    } else if (digit < 12) {
        *s = digit + '0' - 2;
    } else if (digit < 38) {
        *s = digit + 'A' - 12;
    } else {
        *s = digit + 'a' - 38;

    value >>= 6;

The first line of the loop (digit = value & 0x3f) picks six bits from the input value, and the if clauses turn the value formed by those into a character. (. for zero, / for a one, 0 for a two, etc.)

l64a takes a long but the values output by random() are limited to RAND_MAX, which appears to be 2147483647 or 2^31 – 1 on glibc. So, the value that goes to l64a is a random number of 31 bits. By taking 6 bits at a time or a 31 bit value, we get five reasonably evenly distributed characters, plus a sixth that only comes from one bit!

The last character generated by l64a cannot be a ., however, since the loop also has the condition value != 0, and instead of a . as sixth character, l64a returns only five characters. Hence, half the time, the sixth character is a /, and half the time l64a returns five or fewer characters. In the latter case, a following l64a can also generate a slash in the first positions, so in a full salt, the sixth character should be a slash a bit more than half the time.

The code also has a function to randomize the length of the salt, it’s 8 to 16 bytes. The same bias for the slash character happens also with further calls to l64a which would cause the 11th and 12th character to also have a slash more often than anything else. The 100 salts presented in the question have 46 slashes in the sixth position, and 13 and 15 in the 11th and 12th position, respectively. (a bit less than half of the salts are shorter than 11 characters).

On Debian

On Debian, I couldn’t reproduce this with a straight chpasswd as shown in the question. But chpasswd -c SHA256 shows the same behaviour. According to the manual, the default action, without -c, is to let PAM handle the hashing, so apparently PAM on Debian at least uses a different code to generate the salt. I didn’t look at the PAM code on any distribution, however.

(The previous version of this answer stated the effect didn’t appear on Debian. That wasn’t correct.)

Significance, and requirements for salts

Does it matter, though? As @RemcoGerlich commented, it’s pretty much only a question of encoding. It will effectively fix some bits of the salt to zero, but it’s likely that this will have no significant effect in this case, since the origin of those bits is this call to srandom in seedRNG:

srandom (tv.tv_sec ^ tv.tv_usec ^ getpid ());

This is a variant of ye olde custom of seeding an RNG with the current time. (tv_sec and tv_usec are the seconds and microseconds of the current time, getpid() gives the process id if the running process.)
As the time and PIDs are not very unpredictable, the amount of randomness here is likely not larger than what the encoding can hold.

The time and PID is not something you’d like to create keys with, but might be unpredictable enough for salts. Salts must be distinct to prevent brute-force testing multiple password hashes with a single calculation, but should also be unpredictable, to prevent or slow down targeted precomputation, which could be used to shorten the time from getting the password hashes to getting the actual passwords.

Even with the slight issues, as long as the algorithm doesn’t generate the same salt for different passwords, it should be fine. And it doesn’t seem to, even when generating a couple dozen in a loop, as the list in the question shows.

Also, the code in question isn’t used for anything but generating salts for passwords, so there are no implications about problems elsewhere.

For salts, see also, e.g. this on Stack Overflow and this on security.SE.


In conclusion, there’s nothing wrong with your system. Making sure your passwords are any good, and not used on unrelated systems is more useful to think about.

That character is part of the salt per the crypt(3) manual. Given that the length of the salt (the string between the $5$ ID and subsequent $) varies for the exhibited hashes, I’m not exactly sure what picking a random character from that particular column for a few passwords illustrates.

On the other hand, the / is rather more prevalent (102 instances) in the entire salt compared to the other possible characters (around 18), so something in chpasswd does appear to favor that character in the salt;

for x in `seq 1 100000`; do
  echo testacct:asdfasdfasdf | chpasswd -c SHA256
  awk -F: '/testacct/{print $2}' /etc/shadow | awk -F$ '{print $3}' >> salts
perl -nle 'print for m/(.)/g' salts | sort | uniq -c | sort -nr | head -5

on a RedHat EL 6 system run turns up:

   1006 /
    195 X
    193 U
    193 q
    193 e

And, yes, the code in shadow-utils- exhibits a bias towards / which might make dictionary attacks easier:

#include <sys/time.h>

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

// these next two borrowed from libmisc/salt.c of shadow- from
// Centos 6.8 RPM at http://vault.centos.org/6.8/os/Source/SPackages/shadow-utils-
char *l64a(long value)
    static char buf[8];
    char *s = buf;
    int digit;
    int i;

    if (value < 0) {

    for (i = 0; value != 0 && i < 6; i++) {
        digit = value & 0x3f;

        if (digit < 2) {
            *s = digit + '.';
        } else if (digit < 12) {
            *s = digit + '0' - 2;
        } else if (digit < 38) {
            *s = digit + 'A' - 12;
        } else {
            *s = digit + 'a' - 38;

        value >>= 6;


    return (buf);

static void seedRNG(void)
    struct timeval tv;
    static int seeded = 0;

    if (0 == seeded) {
        (void) gettimeofday(&tv, NULL);
        srandom(tv.tv_sec ^ tv.tv_usec ^ getpid());
        seeded = 1;

int main(void)
    for (int x = 0; x < 1000; x++) {
        printf("%sn", l64a(random()));


Which results in:

% ./salttest | perl -nle 'print for m/(.)/g' | sort | uniq -c | sort -nr | head -3
 593 /
  96 8
  93 3

And then using the same routines from the latest https://github.com/shadow-maint/shadow/blob/master/libmisc/salt.c we find that there’s still a bias towards /. So uh yeah this is a bug that should be patched so / is not favored so much, as ideally the salt characters should be equally weighted.

mkpasswd(1) might be a front-end to crypt(3), but it’s not the same as running chpasswd(1), which is part of the “shadow-utils” package on CentOS and “passwd” on Debian. Instead, you should compare apples-to-apples. Consider the following script:


# This repeatedly changes a `saltuser' password
# and grabs the salt out of /etc/shadow.
# Requires root and the existence of `saltuser' user.

if [ $EUID -ne 0 ]; then
    echo "This script requires root access to read /etc/shadow."
    exit 1

grep -q saltuser /etc/passwd

if [ $? -ne 0 ]; then
    echo "This script requires the 'saltuser' to be present."
    exit 2

: > /tmp/salts.txt

for i in {1..1000}; do
    PW=$(tr -cd '[[:print:]]' < /dev/urandom | head -c 64)
    echo "saltuser:${PW}" | chpasswd -c SHA256 -s 0 2> /dev/urandom
    awk -F '$' '/^saltuser/ {print $3}' /etc/shadow >> /tmp/salts.txt

while read LINE; do
    # 6th character in the salt
    echo ${LINE:5:1}
done < /tmp/salts.txt | sort | uniq -c | sort -rn

Output from Debian Sid:

512 /
 14 T
 13 W
 13 v
 13 t
 12 x
 12 m
 12 d
 11 p
 11 L
 11 F
 11 4
 10 s
 10 l
 10 g
 10 f
 10 7
 10 6
  9 Z
  9 w
  9 N
  9 H
  9 G
  9 E
  9 A
  8 Y
  8 X
  8 r
  8 O
  8 j
  8 c
  8 B
  8 b
  8 9
  7 u
  7 R
  7 q
  7 P
  7 M
  7 k
  7 D
  6 z
  6 y
  6 U
  6 S
  6 K
  6 5
  5 V
  5 Q
  5 o
  5 J
  5 I
  5 i
  5 C
  5 a
  5 3
  4 n
  4 h
  4 e
  4 2
  4 0
  4 .
  3 8
  3 1

Output from CentOS 7:

504 /
 13 P
 13 B
 12 s
 12 Z
 11 e
 11 Y
 11 O
 11 L
 11 G
 10 w
 10 u
 10 q
 10 i
 10 h
 10 X
 10 I
 10 E
  9 x
  9 g
  9 f
  9 W
  9 F
  9 C
  9 9
  9 8
  8 v
  8 t
  8 c
  8 b
  8 S
  8 H
  8 D
  8 0
  7 z
  7 y
  7 o
  7 k
  7 U
  7 T
  7 R
  7 M
  7 A
  7 6
  7 4
  7 1
  6 p
  6 d
  6 a
  6 Q
  6 J
  6 5
  6 .
  5 r
  5 m
  5 j
  5 V
  5 3
  5 2
  4 n
  4 l
  4 N
  4 K
  3 7

So, the problem isn’t unique to CentOS, but likely coming from upstream where both projects are pulling from.

Related Solutions

Joining bash arguments into single string with spaces

[*] I believe that this does what you want. It will put all the arguments in one string, separated by spaces, with single quotes around all: str="'$*'" $* produces all the scripts arguments separated by the first character of $IFS which, by default, is a space....

AddTransient, AddScoped and AddSingleton Services Differences

TL;DR Transient objects are always different; a new instance is provided to every controller and every service. Scoped objects are the same within a request, but different across different requests. Singleton objects are the same for every object and every...

How to download package not install it with apt-get command?

Use --download-only: sudo apt-get install --download-only pppoe This will download pppoe and any dependencies you need, and place them in /var/cache/apt/archives. That way a subsequent apt-get install pppoe will be able to complete without any extra downloads....

What defines the maximum size for a command single argument?

Answers Definitely not a bug. The parameter which defines the maximum size for one argument is MAX_ARG_STRLEN. There is no documentation for this parameter other than the comments in binfmts.h: /* * These are the maximum length and maximum number of strings...

Bulk rename, change prefix

I'd say the simplest it to just use the rename command which is common on many Linux distributions. There are two common versions of this command so check its man page to find which one you have: ## rename from Perl (common in Debian systems -- Ubuntu, Mint,...

Output from ls has newlines but displays on a single line. Why?

When you pipe the output, ls acts differently. This fact is hidden away in the info documentation: If standard output is a terminal, the output is in columns (sorted vertically) and control characters are output as question marks; otherwise, the output is...

mv: Move file only if destination does not exist

mv -vn file1 file2. This command will do what you want. You can skip -v if you want. -v makes it verbose - mv will tell you that it moved file if it moves it(useful, since there is possibility that file will not be moved) -n moves only if file2 does not exist....

Is it possible to store and query JSON in SQLite?

SQLite 3.9 introduced a new extension (JSON1) that allows you to easily work with JSON data . Also, it introduced support for indexes on expressions, which (in my understanding) should allow you to define indexes on your JSON data as well. PostgreSQL has some...

Combining tail && journalctl

You could use: journalctl -u service-name -f -f, --follow Show only the most recent journal entries, and continuously print new entries as they are appended to the journal. Here I've added "service-name" to distinguish this answer from others; you substitute...

how can shellshock be exploited over SSH?

One example where this can be exploited is on servers with an authorized_keys forced command. When adding an entry to ~/.ssh/authorized_keys, you can prefix the line with command="foo" to force foo to be run any time that ssh public key is used. With this...

Why doesn’t the tilde (~) expand inside double quotes?

The reason, because inside double quotes, tilde ~ has no special meaning, it's treated as literal. POSIX defines Double-Quotes as: Enclosing characters in double-quotes ( "" ) shall preserve the literal value of all characters within the double-quotes, with the...

What is GNU Info for?

GNU Info was designed to offer documentation that was comprehensive, hyperlinked, and possible to output to multiple formats. Man pages were available, and they were great at providing printed output. However, they were designed such that each man page had a...

Set systemd service to execute after fstab mount

a CIFS network location is mounted via /etc/fstab to /mnt/ on boot-up. No, it is not. Get this right, and the rest falls into place naturally. The mount is handled by a (generated) systemd mount unit that will be named something like mnt-wibble.mount. You can...

Merge two video clips into one, placing them next to each other

To be honest, using the accepted answer resulted in a lot of dropped frames for me. However, using the hstack filter_complex produced perfectly fluid output: ffmpeg -i left.mp4 -i right.mp4 -filter_complex hstack output.mp4 ffmpeg -i input1.mp4 -i input2.mp4...

How portable are /dev/stdin, /dev/stdout and /dev/stderr?

It's been available on Linux back into its prehistory. It is not POSIX, although many actual shells (including AT&T ksh and bash) will simulate it if it's not present in the OS; note that this simulation only works at the shell level (i.e. redirection or...

How can I increase the number of inodes in an ext4 filesystem?

It seems that you have a lot more files than normal expectation. I don't know whether there is a solution to change the inode table size dynamically. I'm afraid that you need to back-up your data, and create new filesystem, and restore your data. To create new...

Why doesn’t cp have a progress bar like wget?

The tradition in unix tools is to display messages only if something goes wrong. I think this is both for design and practical reasons. The design is intended to make it obvious when something goes wrong: you get an error message, and it's not drowned in...