hashcat/OpenCL/m32500-pure.cl

/**
 * Author......: See docs/credits.txt
 * License.....: MIT
 */

// #define NEW_SIMD_CODE

#ifdef KERNEL_STATIC
#include M2S(INCLUDE_PATH/inc_vendor.h)
#include M2S(INCLUDE_PATH/inc_types.h)
#include M2S(INCLUDE_PATH/inc_platform.cl)
#include M2S(INCLUDE_PATH/inc_common.cl)
#include M2S(INCLUDE_PATH/inc_rp.h)
#include M2S(INCLUDE_PATH/inc_rp.cl)
#include M2S(INCLUDE_PATH/inc_scalar.cl)
#include M2S(INCLUDE_PATH/inc_simd.cl)
#include M2S(INCLUDE_PATH/inc_hash_sha256.cl)
#include M2S(INCLUDE_PATH/inc_cipher_aes.cl)
#endif

#define COMPARE_S M2S(INCLUDE_PATH/inc_comp_single.cl)
#define COMPARE_M M2S(INCLUDE_PATH/inc_comp_multi.cl)

typedef struct payload
{
  u32 pl_buf[64];
  u32 pl_len;

} payload_t;

typedef struct doge_tmp
{
  u32  ipad[8];
  u32  opad[8];

  u32  dgst[32];
  u32  out[32];

} doge_tmp_t;

DECLSPEC void hmac_sha256_run_V (PRIVATE_AS u32x *w0, PRIVATE_AS u32x *w1, PRIVATE_AS u32x *w2, PRIVATE_AS u32x *w3, PRIVATE_AS u32x *ipad, PRIVATE_AS u32x *opad, PRIVATE_AS u32x *digest)
{
  digest[0] = ipad[0];
  digest[1] = ipad[1];
  digest[2] = ipad[2];
  digest[3] = ipad[3];
  digest[4] = ipad[4];
  digest[5] = ipad[5];
  digest[6] = ipad[6];
  digest[7] = ipad[7];

  sha256_transform_vector (w0, w1, w2, w3, digest);

  w0[0] = digest[0];
  w0[1] = digest[1];
  w0[2] = digest[2];
  w0[3] = digest[3];
  w1[0] = digest[4];
  w1[1] = digest[5];
  w1[2] = digest[6];
  w1[3] = digest[7];
  w2[0] = 0x80000000;
  w2[1] = 0;
  w2[2] = 0;
  w2[3] = 0;
  w3[0] = 0;
  w3[1] = 0;
  w3[2] = 0;
  w3[3] = (64 + 32) * 8;

  digest[0] = opad[0];
  digest[1] = opad[1];
  digest[2] = opad[2];
  digest[3] = opad[3];
  digest[4] = opad[4];
  digest[5] = opad[5];
  digest[6] = opad[6];
  digest[7] = opad[7];

  sha256_transform_vector (w0, w1, w2, w3, digest);
}

//---------------------------------------------------------------------------------------
CONSTANT_VK u32 base64_table[64] =
{
  'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
  'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
  'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
  'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
  'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
  'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
  'w', 'x', 'y', 'z', '0', '1', '2', '3',
  '4', '5', '6', '7', '8', '9', '+', '/',
};

// Wow it's the right file
DECLSPEC u32 base64_encode_three_bytes_better (u32 in)
{
  //in has 3 u8s in, first u8 is not set)
  u32 out = 0;

  out |= base64_table[(in >> 18) & 0x3F] << 24;
  out |= base64_table[(in >> 12) & 0x3F] << 16;
  out |= base64_table[(in >>  6) & 0x3F] << 8;
  out |= base64_table[(in >>  0) & 0x3F] << 0;

  return out;
}

DECLSPEC void base64_encode_sha256 (u32 *out, const u32 *in)
{
  out[0] = base64_encode_three_bytes_better(                (in[0] >>  8));
  out[1] = base64_encode_three_bytes_better((in[0] << 16) | (in[1] >> 16));
  out[2] = base64_encode_three_bytes_better((in[1] <<  8) | (in[2] >> 24));
  out[3] = base64_encode_three_bytes_better((in[2] <<  0));

  out[4] = base64_encode_three_bytes_better(                (in[3] >>  8));
  out[5] = base64_encode_three_bytes_better((in[3] << 16) | (in[4] >> 16));
  out[6] = base64_encode_three_bytes_better((in[4] <<  8) | (in[5] >> 24));
  out[7] = base64_encode_three_bytes_better((in[5] <<  0));

  out[8] = base64_encode_three_bytes_better(                (in[6] >>  8));
  out[9] = base64_encode_three_bytes_better((in[6] << 16) | (in[7] >> 16));

  // 0x7c = ord('A') ^ ord('=') so replaces the A that we'll get at the end with an =
  out[10] = base64_encode_three_bytes_better(in[7] <<  8) ^ 0x7c;
}
//---------------------------------------------------------------------------------------

KERNEL_FQ void m32500_init (KERN_ATTR_TMPS_ESALT (doge_tmp_t, payload_t))
{
  const u64 gid = get_global_id (0);

  if (gid >= GID_CNT) return;

  sha256_ctx_t ctx;

  sha256_init (&ctx);
  sha256_update_global_swap (&ctx, pws[gid].i, pws[gid].pw_len);
  sha256_final (&ctx);

  u32 w[16] = { 0 }; // only uses 11, but have to be 16 for sha256_hmac_init function

  base64_encode_sha256 (w, ctx.h);

  // pbkdf

  sha256_hmac_ctx_t sha256_hmac_ctx;

  sha256_hmac_init (&sha256_hmac_ctx, w, 44);

  tmps[gid].ipad[0] = sha256_hmac_ctx.ipad.h[0];
  tmps[gid].ipad[1] = sha256_hmac_ctx.ipad.h[1];
  tmps[gid].ipad[2] = sha256_hmac_ctx.ipad.h[2];
  tmps[gid].ipad[3] = sha256_hmac_ctx.ipad.h[3];
  tmps[gid].ipad[4] = sha256_hmac_ctx.ipad.h[4];
  tmps[gid].ipad[5] = sha256_hmac_ctx.ipad.h[5];
  tmps[gid].ipad[6] = sha256_hmac_ctx.ipad.h[6];
  tmps[gid].ipad[7] = sha256_hmac_ctx.ipad.h[7];

  tmps[gid].opad[0] = sha256_hmac_ctx.opad.h[0];
  tmps[gid].opad[1] = sha256_hmac_ctx.opad.h[1];
  tmps[gid].opad[2] = sha256_hmac_ctx.opad.h[2];
  tmps[gid].opad[3] = sha256_hmac_ctx.opad.h[3];
  tmps[gid].opad[4] = sha256_hmac_ctx.opad.h[4];
  tmps[gid].opad[5] = sha256_hmac_ctx.opad.h[5];
  tmps[gid].opad[6] = sha256_hmac_ctx.opad.h[6];
  tmps[gid].opad[7] = sha256_hmac_ctx.opad.h[7];

  sha256_hmac_update_global_swap (&sha256_hmac_ctx, salt_bufs[SALT_POS_HOST].salt_buf, salt_bufs[SALT_POS_HOST].salt_len);

  for (u32 i = 0, j = 1; i < 8; i += 8, j += 1)
  {
    sha256_hmac_ctx_t sha256_hmac_ctx2 = sha256_hmac_ctx;

    u32 w0[4];
    u32 w1[4];
    u32 w2[4];
    u32 w3[4];

    w0[0] = j;
    w0[1] = 0;
    w0[2] = 0;
    w0[3] = 0;
    w1[0] = 0;
    w1[1] = 0;
    w1[2] = 0;
    w1[3] = 0;
    w2[0] = 0;
    w2[1] = 0;
    w2[2] = 0;
    w2[3] = 0;
    w3[0] = 0;
    w3[1] = 0;
    w3[2] = 0;
    w3[3] = 0;

    sha256_hmac_update_64 (&sha256_hmac_ctx2, w0, w1, w2, w3, 4);

    sha256_hmac_final (&sha256_hmac_ctx2);

    tmps[gid].dgst[i + 0] = sha256_hmac_ctx2.opad.h[0];
    tmps[gid].dgst[i + 1] = sha256_hmac_ctx2.opad.h[1];
    tmps[gid].dgst[i + 2] = sha256_hmac_ctx2.opad.h[2];
    tmps[gid].dgst[i + 3] = sha256_hmac_ctx2.opad.h[3];
    tmps[gid].dgst[i + 4] = sha256_hmac_ctx2.opad.h[4];
    tmps[gid].dgst[i + 5] = sha256_hmac_ctx2.opad.h[5];
    tmps[gid].dgst[i + 6] = sha256_hmac_ctx2.opad.h[6];
    tmps[gid].dgst[i + 7] = sha256_hmac_ctx2.opad.h[7];

    tmps[gid].out[i + 0] = tmps[gid].dgst[i + 0];
    tmps[gid].out[i + 1] = tmps[gid].dgst[i + 1];
    tmps[gid].out[i + 2] = tmps[gid].dgst[i + 2];
    tmps[gid].out[i + 3] = tmps[gid].dgst[i + 3];
    tmps[gid].out[i + 4] = tmps[gid].dgst[i + 4];
    tmps[gid].out[i + 5] = tmps[gid].dgst[i + 5];
    tmps[gid].out[i + 6] = tmps[gid].dgst[i + 6];
    tmps[gid].out[i + 7] = tmps[gid].dgst[i + 7];
  }

}

KERNEL_FQ void m32500_loop (KERN_ATTR_TMPS_ESALT (doge_tmp_t, payload_t))
{
  //pbkdf2hmac here

  const u64 gid = get_global_id (0);

  if ((gid * VECT_SIZE) >= GID_CNT) return;

  u32x ipad[8];
  u32x opad[8];

  ipad[0] = packv (tmps, ipad, gid, 0);
  ipad[1] = packv (tmps, ipad, gid, 1);
  ipad[2] = packv (tmps, ipad, gid, 2);
  ipad[3] = packv (tmps, ipad, gid, 3);
  ipad[4] = packv (tmps, ipad, gid, 4);
  ipad[5] = packv (tmps, ipad, gid, 5);
  ipad[6] = packv (tmps, ipad, gid, 6);
  ipad[7] = packv (tmps, ipad, gid, 7);

  opad[0] = packv (tmps, opad, gid, 0);
  opad[1] = packv (tmps, opad, gid, 1);
  opad[2] = packv (tmps, opad, gid, 2);
  opad[3] = packv (tmps, opad, gid, 3);
  opad[4] = packv (tmps, opad, gid, 4);
  opad[5] = packv (tmps, opad, gid, 5);
  opad[6] = packv (tmps, opad, gid, 6);
  opad[7] = packv (tmps, opad, gid, 7);

  for (u32 i = 0; i < 8; i += 8)
  {
    u32x dgst[8];
    u32x out[8];

    dgst[0] = packv (tmps, dgst, gid, i + 0);
    dgst[1] = packv (tmps, dgst, gid, i + 1);
    dgst[2] = packv (tmps, dgst, gid, i + 2);
    dgst[3] = packv (tmps, dgst, gid, i + 3);
    dgst[4] = packv (tmps, dgst, gid, i + 4);
    dgst[5] = packv (tmps, dgst, gid, i + 5);
    dgst[6] = packv (tmps, dgst, gid, i + 6);
    dgst[7] = packv (tmps, dgst, gid, i + 7);

    out[0] = packv (tmps, out, gid, i + 0);
    out[1] = packv (tmps, out, gid, i + 1);
    out[2] = packv (tmps, out, gid, i + 2);
    out[3] = packv (tmps, out, gid, i + 3);
    out[4] = packv (tmps, out, gid, i + 4);
    out[5] = packv (tmps, out, gid, i + 5);
    out[6] = packv (tmps, out, gid, i + 6);
    out[7] = packv (tmps, out, gid, i + 7);

    for (u32 j = 0; j < LOOP_CNT; j++)
    {
      u32x w0[4];
      u32x w1[4];
      u32x w2[4];
      u32x w3[4];

      w0[0] = dgst[0];
      w0[1] = dgst[1];
      w0[2] = dgst[2];
      w0[3] = dgst[3];
      w1[0] = dgst[4];
      w1[1] = dgst[5];
      w1[2] = dgst[6];
      w1[3] = dgst[7];
      w2[0] = 0x80000000;
      w2[1] = 0;
      w2[2] = 0;
      w2[3] = 0;
      w3[0] = 0;
      w3[1] = 0;
      w3[2] = 0;
      w3[3] = (64 + 32) * 8;

      hmac_sha256_run_V (w0, w1, w2, w3, ipad, opad, dgst);

      out[0] ^= dgst[0];
      out[1] ^= dgst[1];
      out[2] ^= dgst[2];
      out[3] ^= dgst[3];
      out[4] ^= dgst[4];
      out[5] ^= dgst[5];
      out[6] ^= dgst[6];
      out[7] ^= dgst[7];
    }

    unpackv (tmps, dgst, gid, i + 0, dgst[0]);
    unpackv (tmps, dgst, gid, i + 1, dgst[1]);
    unpackv (tmps, dgst, gid, i + 2, dgst[2]);
    unpackv (tmps, dgst, gid, i + 3, dgst[3]);
    unpackv (tmps, dgst, gid, i + 4, dgst[4]);
    unpackv (tmps, dgst, gid, i + 5, dgst[5]);
    unpackv (tmps, dgst, gid, i + 6, dgst[6]);
    unpackv (tmps, dgst, gid, i + 7, dgst[7]);

    unpackv (tmps, out, gid, i + 0, out[0]);
    unpackv (tmps, out, gid, i + 1, out[1]);
    unpackv (tmps, out, gid, i + 2, out[2]);
    unpackv (tmps, out, gid, i + 3, out[3]);
    unpackv (tmps, out, gid, i + 4, out[4]);
    unpackv (tmps, out, gid, i + 5, out[5]);
    unpackv (tmps, out, gid, i + 6, out[6]);
    unpackv (tmps, out, gid, i + 7, out[7]);
  }
}

KERNEL_FQ void m32500_comp (KERN_ATTR_TMPS_ESALT (doge_tmp_t, payload_t))
{
   /**
   * base
   */

  const u64 gid = get_global_id (0);
  const u64 lid = get_local_id(0);
  const u64 lsz = get_local_size(0);

  /**
   * aes shared
   */

  #ifdef REAL_SHM

  LOCAL_VK u32 s_td0[256];
  LOCAL_VK u32 s_td1[256];
  LOCAL_VK u32 s_td2[256];
  LOCAL_VK u32 s_td3[256];
  LOCAL_VK u32 s_td4[256];

  LOCAL_VK u32 s_te0[256];
  LOCAL_VK u32 s_te1[256];
  LOCAL_VK u32 s_te2[256];
  LOCAL_VK u32 s_te3[256];
  LOCAL_VK u32 s_te4[256];

  for (u32 i = lid; i < 256; i += lsz)
  {
    s_td0[i] = td0[i];
    s_td1[i] = td1[i];
    s_td2[i] = td2[i];
    s_td3[i] = td3[i];
    s_td4[i] = td4[i];

    s_te0[i] = te0[i];
    s_te1[i] = te1[i];
    s_te2[i] = te2[i];
    s_te3[i] = te3[i];
    s_te4[i] = te4[i];
  }

  SYNC_THREADS();

  #else

  CONSTANT_AS u32a* s_td0 = td0;
  CONSTANT_AS u32a* s_td1 = td1;
  CONSTANT_AS u32a* s_td2 = td2;
  CONSTANT_AS u32a* s_td3 = td3;
  CONSTANT_AS u32a* s_td4 = td4;

  CONSTANT_AS u32a* s_te0 = te0;
  CONSTANT_AS u32a* s_te1 = te1;
  CONSTANT_AS u32a* s_te2 = te2;
  CONSTANT_AS u32a* s_te3 = te3;
  CONSTANT_AS u32a* s_te4 = te4;

  #endif

  if (gid >= GID_CNT) return;

  u32 ukey[8];

  ukey[0] = tmps[gid].out[0];
  ukey[1] = tmps[gid].out[1];
  ukey[2] = tmps[gid].out[2];
  ukey[3] = tmps[gid].out[3];
  ukey[4] = tmps[gid].out[4];
  ukey[5] = tmps[gid].out[5];
  ukey[6] = tmps[gid].out[6];
  ukey[7] = tmps[gid].out[7];

  u32 ks[60];

  AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3);

  // iv

  u32 prev_ct[4]; // iv is the first 4 u32s -> needs to be prev ct for cbc encryption (each block used prior ct)
                  // todo: might want to swap in module

  prev_ct[0] = hc_swap32 (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[0]);
  prev_ct[1] = hc_swap32 (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[1]);
  prev_ct[2] = hc_swap32 (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[2]);
  prev_ct[3] = hc_swap32 (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[3]);

  u32 isAscii = 0;

  // ct
  u32 ct_buf[4];

  // Padding is Crypto.pad.iso10126 -pads with random bytes until the last byte, and which defines the number of padding bytes
  // So knocking off last block to not account for any non-ascii padding
  // todo: pkcs_padding_bs16() might be able to replace this

  for (u32 i = 4, j = 16; j < esalt_bufs[DIGESTS_OFFSET_HOST].pl_len - 16; i += 4, j += 16)
  {
    ct_buf[0] = hc_swap32_S (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[i + 0]);
    ct_buf[1] = hc_swap32_S (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[i + 1]);
    ct_buf[2] = hc_swap32_S (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[i + 2]);
    ct_buf[3] = hc_swap32_S (esalt_bufs[DIGESTS_OFFSET_HOST].pl_buf[i + 3]);

    u32 pt_buf[4];

    AES256_decrypt (ks, ct_buf, pt_buf, s_td0, s_td1, s_td2, s_td3, s_td4);

    for (u32 x = 0; x < 4; x++)
    {
      pt_buf[x] ^= prev_ct[x];
      isAscii |= pt_buf[x] & 0x80808080; //check the ciphertext is human readable
      prev_ct[x] = ct_buf[x];            //set previous CT as the new IV for the next block
    }
  }

  const u32 r0 = isAscii;
  const u32 r1 = 0;
  const u32 r2 = 0;
  const u32 r3 = 0;

  #define il_pos 0

  #ifdef KERNEL_STATIC
  #include COMPARE_M
  #endif
}