CRYPTOLOGY - Hacking: The Art of Exploitation - page 419

reader@hacking:~/booksrc $ gcc -o crypt_test crypt_test.c
/tmp/cccrSvYU.o: In function `main':
crypt_test.c:(.text+0x73): undefined reference to `crypt'
collect2: ld returned 1 exit status
reader@hacking:~/booksrc $ gcc -o crypt_test crypt_test.c -l crypt
reader@hacking:~/booksrc $ ./crypt_test testing je
password "testing" with salt "je" hashes to ==> jeLu9ckBgvgX.
reader@hacking:~/booksrc $ ./crypt_test test je
password "test" with salt "je" hashes to ==> jeHEAX1m66RV.
reader@hacking:~/booksrc $ ./crypt_test test xy
password "test" with salt "xy" hashes to ==> xyVSuHLjceD92
reader@hacking:~/booksrc $

Notice that in the last two runs, the same password is encrypted, but
using different salt values. The salt value is used to perturb the algorithm
further, so there can be multiple hash values for the same plaintext value if
different salt values are used. The hash value (including the prepended salt)
is stored in the password file under the premise that if an attacker were to
steal the password file, the hashes would be useless.
When a legitimate user needs to authenticate using the password hash,
that user's hash is looked up in the password file. The user is prompted to
enter her password, the original salt value is extracted from the password file,
and whatever the user types is sent through the same one-way hash function
with the salt value. If the correct password was entered, the one-way hashing
function will produce the same hash output as is stored in the password file.
This allows authentication to function as expected, without ever having to
store the plaintext password.

0x761 Dictionary Attacks

It turns out, however, that the encrypted passwords in the password file aren't
so useless after all. Sure, it's mathematically impossible to reverse the hash,
but it is possible to just quickly hash every word in a dictionary, using the salt
value for a specific hash, and then compare the result with that hash. If the
hashes match, then that word from the dictionary must be the plaintext
password.
A simple dictionary attack program can be whipped up fairly easily. It just
needs to read words from a file, hash each one using the proper salt value,
and display the word if there is a match. The following source code does this
using filestream functions, which are included with stdio.h. These functions
are easier to work with, since they wrap up the messiness of open() calls and
file descriptors, using FILE structure pointers, instead. In the source below,
the fopen() call's r argument tells it to open the file for reading. It returns
NULL on failure, or a pointer to the open filestream. The fgets() call gets
a string from the filestream, up to a maximum length or when it reaches the
end of a line. In this case, it's used to read each line from the word-list file.
This function also returns NULL on failure, which is used to detect then
end of the file.