Hi all,
I'd like to use the encryption method used to generate the /etc/shadow passwords.
The goal is to write a script that get a plain-text password as argument and returns an encrycped one.
Can you help me, please?
hai
To encrypt the password you have a function called crypt.. I have used the function in c. The prototype of the function is
char * crypt(const char * pass, const char * key);
the first argument is the password to encrypt
the second one is the key to encrypt the password
the function returns the encrypted password
example program
/* cryp.c */
#include <stdio.h>
#include <unistd.h>
#include <crypt.h>
main()
{
printf("%s\n",crypt("mypassword","B"));
}
compilation:
gcc -lcrypt cryp.c
Output:
BBVV5v7k.VS2.
returns the excrypted password of the string "mypassword" with the encryption key as B
Note: the key to encrypt will always be the first char in the output
Goodluck
Collins
It works !!! Many thanks.
A question, I've tried to run with a 1 character encryption key (e.g."X") and the output string was like "X.xxxxxxxxxxx".
To eliminate the . between key and crypted string I had to use a 2 characters key, e.g.:
"AB" -> "ABxxxxxxxxxxx"
Is this a normal behaviour?
Thanks again.
hai nisant
The length of the key should be two characters. If the user passes only one character for the key field the function puts the same character twice.
so if the key is X ,
the output would be something like
XX<remaining encrypted pass>
if the key is two char length, then the first two char of the encrypted password will be the same as that of the passed argument... Hope i am right
Have a nice time
Collins
Actually, in a call like:
printf("%s\n",crypt("mypassword","B"));
That first argument, the password to be encrypted, is called the "key". The second argument is called the "salt". The salt is required to be two characters from the set [a-zA-Z0-9./]. Calling crypt() with anything other than a properly constructed two character salt is an error. The results of such a call are undefined. One implementation may use "BB" and another "B.". Still another might return an error. So please use two character salts.
I've tried using this code:
char k[2];
char str[255];
scanf("%s",k);
scanf("%s",str);
printf("%s\n",crypt(str,k));
Entering only 1 character to k variable I verify that encrypted string begins with key character repeated twice.
Tested on Solaris and HP-UX SO.
The previous test was executed on Mac OS Linux environment, maybe there are some little differences in gcc and/or libcrypto libraries.
However, it seems I can change the password in unattended mode using a C pgm tha returns the encrypted pwd because I also verified that both the encrypted pwds returned by C pgm and passwd command are the same.
So many thanks, Collins, for your help 
.
Bye
nisant
Thank you, Perderabo, for specifying the possible behaviours in coding an only 1 character "salt" in crypt() call.
Bye
nisant
Can i get back the clear text password by copy the encrypted value from /etc/passwd and put this value in the function 'decrypt()'......is that possible ???
No. crypt() is actually a bad name, since it doesn't do any encryption, but only runs your password through a hashing function (MD5/Blowfish/...).
Is there any way to get back clear text password from /etc/shadow encrypted data ??
Yeah, there are several programs that do this. The best available is called "john the ripper", JtR for short. It works by trying all possible passwords until it hits a match. Last November I started JtR running to try and crack a list of 168 passwords. (This is a security test and its part of my job. My manager knows I am doing this.) So far it has broken 9 passwords. It is slowing down. It breaks the easy passwords first. It's been over a month since password 9 was broken. My guess it that it will take several decades to break all 168. I'm using a somewhat souped up sunblade-1000. With a top of the line overclocked quad extreme rig, it could probably crack them all in under a decade.