Password Algorithms: AOL Instant Messenger


AOL instant messenger is used by 0.73% of IM market according to OPSWAT
It’s still in development and because of changes in how the password is stored, this analysis will be presented in 2 entries.


In 6.x, when a user checks the ‘Save Password’ box, the application will store the current username and password in NTUSER.DAT

From command prompt, you can run REG.exe to dump the entries.

C:\reg query "HKCU\Software\America Online\AIM6\Passwords"

HKEY_CURRENT_USER\Software\America Online\AIM6\Passwords  REG_SZ  zo+VVoi9LWCtc0B8z9ZnfojNdjVuv08DXid8yK++LYI=

Based on the characters in the string, it appears to be base64 encoded.
Here, i’m using openssl and redirect the output to a file before hexdumping.

echo zo+VVoi9LWCtc0B8z9ZnfojNdjVuv08DXid8yK++LYI= | openssl enc -base64 -d > aim.bin & hexdump aim.bin

00000000: ce 8f 95 56 88 bd 2d 60 - ad 73 40 7c cf d6 67 7e ...V.... .s....g.
00000010: 88 cd 76 35 6e bf 4f 03 - 5e 27 7c c8 af be 2d 82 ..v5n.O. ........
00000020: 82

This doesn’t appear to contain anything intelligible to the eye so we’ll need to look deeper in the application itself.


After some digging in the binaries, the blob stored in the registry contains a salt and ciphertext derived from Blowfish, here’s the structure.

#define MAX_SALT_LEN 8
#define MAX_PASS_LEN 16

typedef struct _AIM_PASSWORD_BLOB {
  u8 Salt[MAX_SALT_LEN];
  u8 Password[2*MAX_PASS_LEN+1];

The password is stored in UNICODE format.
The salt is generated by first initializing the seed to current time, then calling rand() 8 times to fill 64-bit buffer.
Everytime the user logs in and out, the password entry is updated with new salt.

// generate 64-bit salt using time(0) as seed
void gen_salt(PAIM_PASSWORD_BLOB pBlob) {

    for (int i = 0; i < MAX_SALT_LEN; i++) {
      pBlob->Salt[i] = rand() & 0xff;

Blowfish has a variable key length and the encryption algorithm uses all 448-bits of key to encrypt the users password.
Below only illustrates how the static key is created.

// used to create the 56 bytes of key
const u8 aim_kbox[256] =
  { 0x59, 0x3a, 0x7c, 0x77, 0xf3, 0x2b, 0xab, 0x1f,
    0x99, 0x98, 0x86, 0x6c, 0x59, 0xaa, 0x9d, 0x7f,
    0x58, 0x3f, 0x6a, 0xb9, 0x0b, 0x47, 0x29, 0x35,
    0xaa, 0x6d, 0xea, 0x95, 0xe2, 0xfb, 0xe4, 0x02,
    0xcb, 0xf7, 0x0c, 0x6e, 0x19, 0x92, 0xe6, 0x1c,
    0x96, 0xc4, 0x9b, 0x63, 0xd0, 0x30, 0x4d, 0xaf,
    0x0e, 0x4d, 0xa7, 0xc8, 0x89, 0xc7, 0xb8, 0x57,
    0xd9, 0x23, 0x01, 0xa6, 0xae, 0xa3, 0xcc, 0xa7,
    0xc0, 0x69, 0xc0, 0x38, 0x09, 0xde, 0xb3, 0xa5,
    0x31, 0x55, 0xbf, 0x6e, 0x4a, 0xec, 0x98, 0x4b,
    0xbd, 0xb3, 0x1c, 0x6e, 0x84, 0x11, 0x2c, 0x08,
    0x9a, 0x63, 0xbb, 0x0e, 0xb0, 0xe5, 0x24, 0x3d,
    0x22, 0xd6, 0xc1, 0x5c, 0x29, 0xd7, 0xb9, 0xc1,
    0x52, 0x95, 0x19, 0x16, 0x2f, 0xa7, 0x27, 0x5d,
    0x4c, 0xba, 0xf3, 0x32, 0x64, 0xeb, 0x2e, 0x50,
    0xd5, 0x74, 0x3f, 0x57, 0x52, 0x8b, 0x94, 0xcd,
    0xd8, 0x87, 0x36, 0x62, 0xe3, 0x45, 0xa1, 0x78,
    0xe1, 0xca, 0xd2, 0xe2, 0xe7, 0x29, 0xa1, 0xec,
    0xa3, 0xa7, 0x51, 0x9c, 0x92, 0x1e, 0x66, 0x38,
    0x72, 0x9f, 0xb6, 0x08, 0xfb, 0x5b, 0xc3, 0x5d,
    0xca, 0xc4, 0x48, 0xd3, 0x9e, 0xef, 0x12, 0xd2,
    0xc6, 0x3d, 0x11, 0x5b, 0xda, 0x3a, 0xaf, 0x01,
    0xaa, 0xc5, 0x60, 0x35, 0x74, 0x72, 0x7b, 0xc7,
    0x5a, 0x2c, 0x48, 0xaa, 0x12, 0x1f, 0x5a, 0xf8,
    0xe0, 0xd5, 0x1f, 0x35, 0xc1, 0x9e, 0xb5, 0xd8,
    0xe9, 0x36, 0xff, 0x07, 0x70, 0xa2, 0xf9, 0xa3,
    0x49, 0x5b, 0x48, 0x84, 0x73, 0xae, 0x16, 0x06,
    0x73, 0x63, 0x1c, 0xc4, 0x01, 0x9d, 0x00, 0xa3,
    0x02, 0x01, 0xe2, 0x13, 0xb6, 0x1a, 0x32, 0x91,
    0x33, 0xfc, 0xc4, 0x70, 0x12, 0x28, 0x26, 0xda,
    0x68, 0xb0, 0x31, 0x12, 0xf8, 0x9c, 0xde, 0xfb,
    0xa6, 0x8b, 0x5b, 0xde, 0x2f, 0x9e, 0x5e, 0x68 };

#define SWAP(x, y) u8 t; t = x;x = y;y = t;

void gen_static_key(u8 output[]) {

  u8 kbox[256];

  memcpy(kbox, aim_kbox, sizeof(aim_kbox));

  for (int i = 0; i < 128; i++) {
    SWAP(kbox[ kbox[i] ], kbox[ kbox[i+1] ])
  memcpy(output, kbox, 56);

The end result of gen_static_key() is a 56-byte blowfish key.
The user’s 8-byte salt will replace the first 8-bytes of this static key before being passed to BF_set_key()

It’s not necessary to use gen_static_key() after the first time since it doesn’t use any entropy at all.


The final decoding routine using C++ looks something like:

#define MAX_SALT_LEN 8
#define MAX_PASS_LEN 16
#define MAX_KEY_LEN 56

typedef struct _AIM_PASSWORD_BLOB {
  u8 Salt[MAX_SALT_LEN];
  u8 Password[2*MAX_PASS_LEN+1];

void base64_decode(PAIM_PASSWORD_BLOB output, char blob_string[]) {
    BIO *b64, *mem;

    b64 = BIO_new(BIO_f_base64());
    BIO_set_flags(b64, BIO_FLAGS_BASE64_NO_NL);
    mem = BIO_new_mem_buf(blob_string, strlen(blob_string));
    mem = BIO_push(b64, mem);

    BIO_read(mem, output, strlen(blob_string));

const u8 static_key[48] =
  { 0x99, 0x00, 0x86, 0xa5, 0x27, 0xaa, 0x9d, 0x7f,
    0x58, 0xaa, 0xae, 0xb9, 0x0b, 0x47, 0x3a, 0x35,
    0xaa, 0xe0, 0xea, 0x95, 0x66, 0xfb, 0xe4, 0x9f,
    0xcb, 0xf7, 0x16, 0x1c, 0xa3, 0x92, 0xe6, 0x1c,
    0x96, 0x06, 0x9b, 0x5b, 0x29, 0x30, 0xbf, 0xaf,
    0xec, 0x11, 0x29, 0xc8, 0x89, 0x5b, 0xb8, 0x57 };

std::wstring AIM_Decrypt(char base64_blob[]) {
    BF_KEY key;
    u8 aim_key[MAX_KEY_LEN];
    wchar_t password[2*MAX_PASS_LEN+1];

    memset(&blob, 0, sizeof(AIM_PASSWORD_BLOB));
    memset(password, 0, sizeof(password));

    base64_decode(&blob, base64_blob);

    memcpy(&aim_key, blob.Salt, MAX_SALT_LEN);
    memcpy(&aim_key[MAX_SALT_LEN], static_key, sizeof(static_key));

    BF_set_key(&key, sizeof(aim_key), aim_key);

    BF_LONG *in = (BF_LONG*)blob.Password;
    BF_LONG *out = (BF_LONG*)password;

    for (int i = 0;i < MAX_PASS_LEN * 2 / sizeof(BF_LONG);i += 2) {
      memcpy(&out[i], &in[i], sizeof(BF_LONG)*2);
      BF_decrypt(&out[i], &key);
    return std::wstring(password);

The string returned from AIM_Decrypt is in UNICODE format.


This protection isn’t great since you could easily decrypt entries just by reading NTUSER.DAT in the windows profile of AIM user.

A better solution might be CryptProtectData() and Base64.

Comments and questions welcome.

Password Algorithms: Google Talk


Lots of people that use Gmail use Google Talk
It functions pretty much like any other IM application such as those available from MSN, Yahoo, AOL

OPSWAT reported in June 2011 that Google Talk accounts for 3.56% of market share in Instant Messaging.

That’s not a big percentage of users but the algorithm described here is probably used in other google applications.


If the user chooses to ‘Remember Password’ an account entry will be created in NTUSER.DAT
So for example if the google id is the registry entry would be

HKEY_CURRENT_USER\Software\Google\Google Talk\Accounts\

Password values are stored under a string value called pw
Without showing string for my own password, it looks something like


The value itself is much much longer but I just want to give you an idea of what to look for.
It doesn’t use any conventional encoding algorithm like Base64.


After some digging around in the binaries, Gtalk first initializes some entropy using a static key, the domain and username.

DWORD gtalk_entropy[4];
DWORD static_key[4] = { 0x69F31EA3, 0x1FD96207, 0x7D35E91E, 0x487DD24F };

 * Retrieve the current domainusername from thread or process token
 * return TRUE if successful
BOOL GetUserInfo(std::wstring &domain, std::wstring &username) {
    HANDLE hToken;
    DWORD dwTokenSize = 0, dwUserName = 64, dwDomain = 64;
    WCHAR UserName[64], Domain[64];
    SID_NAME_USE peUse;
    PSID pSid = NULL;
    BOOL bResult = FALSE;

    OpenThreadToken(GetCurrentThread(), TOKEN_QUERY, TRUE, &hToken);

    if (GetLastError() == ERROR_NO_TOKEN) {
      if (!OpenProcessToken(GetCurrentProcess(), TOKEN_QUERY, &hToken)) {
        return FALSE;

    if (!GetTokenInformation(hToken, TokenUser, 0, 0, &dwTokenSize)) {
      if (GetLastError() == ERROR_INSUFFICIENT_BUFFER) {
        pSid = new BYTE[dwTokenSize];
        if (pSid != NULL) {
          if (GetTokenInformation(hToken, TokenUser, pSid, dwTokenSize, &dwTokenSize)) {
            bResult = LookupAccountSid(NULL, reinterpret_cast(pSid)->User.Sid,
                UserName, &dwUserName, Domain, &dwDomain, &peUse);
            if (bResult) {
              domain = Domain;
              username = UserName;
          delete []pSid;
    return bResult;

All the above code does is retrieve from the process/thread security token the domain and username.

The seed value appears to be derived from a random number generator developed
by R.Park and K.Miller, however this is merely speculation as no generator exists in the binary itself.
48271 and 69621 are common multiplier values for RNG.

 * initialize the entropy that will used to encrypt/decrypt passwords
BOOL init_entropy() {
    std::wstring domain, username;

    BOOL bResult = GetUserInfo(domain, username);

    if (bResult) {
      memcpy(gtalk_entropy, static_key, sizeof(static_key));

      long M = 2147483647; // modulus
      long A = 48271;      // multiplier
      long Q = M / A;
      long R = M % A;
      long seed = 387822687;   // this could be wrong but does work
                               // and is exactly 9 digits

      seed = A * (seed % Q) - R * (seed / Q);
      seed += M;

      long idx = 0;

      // mix with username
      for (std::wstring::size_type i = 0;i < username.length();i++) {
        gtalk_entropy[idx++ % 4] ^= username[i] * seed;
        seed *= A;

      // mix with domain
      for (std::wstring::size_type i = 0;i < domain.length();i++) {
        gtalk_entropy[idx++ % 4] ^= domain[i] * seed;
        seed *= A;
    return bResult;


Once the entropy is initialized, it’s possible to decode the pw value into binary which is just a DPAPI blob.

Instead of the usual Base64 encoding, Google Talk uses Base16 with a custom alphabet and a multiplier value.

// convert base16 string into binary
void gtalk_decode(BYTE blob[], std::wstring input) {
    std::wstring alphabet = L"!"#$%&'()*+,-./0";
    long seed = gtalk_entropy[0] | 1;
    long A = 69621;
    PBYTE p = blob;

    for (size_t i = 4;i < input.length();i += 2) {
      int c;

      c  = (alphabet.find_first_of( + 0))) << 4;
      c |= (alphabet.find_first_of( + 1))) & 0x0f;

      *p++ = c - (seed & 0xff);
      seed *= A;

Finally, CryptUnprotectData() is used to decrypt the UNICODE password.

// decrypt blob using DPAPI
BOOL gtalk_decrypt(BYTE password[], BYTE blob_data[], size_t blob_size) {
    DATA_BLOB DataIn, DataEntropy, DataOut;

    DataEntropy.cbData = sizeof(gtalk_entropy);
    DataEntropy.pbData = (BYTE*)gtalk_entropy;

    DataIn.cbData = blob_size;
    DataIn.pbData = blob_data;

    BOOL bResult = CryptUnprotectData(&DataIn, NULL, &DataEntropy, NULL, NULL, 1, &DataOut);

    if (bResult) {
      memcpy(password, DataOut.pbData, DataOut.cbData);
      password[DataOut.cbData] = 0;
    return bResult;


The password is protected reasonably well although the entropy generation is a little redundant and I dare say a classic example of Security through Obscurity :)
CryptProtectData and Base64 would provide the same level of protection.

Password Algorithms: K9 Web Protection Admin

K9 Logo


K9 Web Protection by Blue Coat software is a cheap but effective solution for parents monitoring internet activity of their children.
It’s also a cheap solution for internet cafes and companies to monitor customers and staff.

According to the vendors website, it’s installed on some 3 million computers and presumably most of those are using the freeware version for home use which is what I’ve downloaded.

Although there’s a commercial version, I’m not aware of any differences between the 2 with respect to the password algorithm.

Storage Method

Installed on XP, most files are kept in:

C:\Program Files\Blue Coat K9 Web Protection

The license file is always 944 bytes in size and also where the Admin password is stored.

Here’s the actual password “hash” and offset of where it appears.

000001d0: c4 b8 b5 b5 b7 b7 bd b1 - be 29 6b d6 eb 2c a9 00

The offset was discovered through analysing one of the binaries.

So how is this “hash” generated? :)

You can recover this password with simple subtraction. (at least in my own case) The example you see above is “theeggman” and will be revealed if you subtract 0x50 from each byte until you reach an invalid character.

The subtraction value may be different for you since I haven’t checked beyond my own XP install.
However, it’s possible to get the value from the license file from offset 0x2c

00000020: 75 a5 ce 51 00 00 00 00 - c3 e1 5a 78 50 00 00 00


Below is rough code for how a password is generated.

#define MAX_PASS_LEN 15

void encode(unsigned char k9_hash[], char password[]) {
    size_t pass_len = strlen(password);
    size_t i;

    for (i = 0;i < pass_len && i < MAX_PASS_LEN;i++) {
      k9_hash[i] = password[i] + license_data[44];

    for (;i < MAX_PASS_LEN;i++) {
      k9_hash[i] = rand() % 256;
    k9_hash[MAX_PASS_LEN] = 0;


This uses hardcoded values from my own license file and successfully decodes to plaintext.

#include <stdio.h>
#include <ctype.h>

#define MAX_PASS_LEN 15

char* decode(char password[], unsigned char k9_hash[]) {
  char sym[] = "!@#$%^*(){}";
  size_t pass_len = 0;

  for (;pass_len < MAX_PASS_LEN;pass_len++) {
    int c = k9_hash[pass_len] - 0x50;
    if (!isalnum(c)) {
      if (!strchr(sym, c)) {
    password[pass_len] = c;
  password[pass_len] = 0;
  return password;

unsigned char k9_admin_pass[] = { 0xc4, 0xb8, 0xb5, 0xb5,
                                  0xb7, 0xb7, 0xbd, 0xb1,
                                  0xbe, 0x29, 0x6b, 0xd6,
                                  0xeb, 0x2c, 0xa9, 0x00 };

void main(void) {
  char password[MAX_PASS_LEN+1];

  memset(password, 0, sizeof(password));
  printf("\nPassword = %s\n", decode(password, k9_admin_pass));


In order to take advantage of the weak hashing algorithm, you need at least read access to the license file and after checking permissions, it’s only available to the owner, SYSTEM and other local administrators.

It doesn’t provide much protection for the password but controlling admin access at OS level negates this.