hashcat/OpenCL/m25900-pure.cl

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

#define NEW_SIMD_CODE

#ifdef KERNEL_STATIC
#include "inc_vendor.h"
#include "inc_types.h"
#include "inc_platform.cl"
#include "inc_common.cl"
#include "inc_simd.cl"
#include "inc_hash_sha256.cl"
#include "inc_cipher_aes.cl"
#endif

#define COMPARE_S "inc_comp_single.cl"
#define COMPARE_M "inc_comp_multi.cl"

typedef struct blocks
{
  u32 b1[4];
  u32 b2[4];
  u32 b3[4];

} blocks_t;

typedef struct pbkdf2_sha256_tmp
{
  u32 ipad[8];
  u32 opad[8];

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

} pbkdf2_sha256_tmp_t;

DECLSPEC void hmac_sha256_run_V (u32x *w0, u32x *w1, u32x *w2, u32x *w3, u32x *ipad, u32x *opad, 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);
}

DECLSPEC void aes128_encrypt_cbc (const u32 *aes_ks, u32 *aes_iv, const u32 *in, u32 *out, SHM_TYPE u32 *s_te0, SHM_TYPE u32 *s_te1, SHM_TYPE u32 *s_te2, SHM_TYPE u32 *s_te3, SHM_TYPE u32 *s_te4)
{
  u32 in_s[4];

  in_s[0] = in[0];
  in_s[1] = in[1];
  in_s[2] = in[2];
  in_s[3] = in[3];

  in_s[0] ^= aes_iv[0];
  in_s[1] ^= aes_iv[1];
  in_s[2] ^= aes_iv[2];
  in_s[3] ^= aes_iv[3];

  aes128_encrypt (aes_ks, in_s, out, s_te0, s_te1, s_te2, s_te3, s_te4);

  aes_iv[0] = out[0];
  aes_iv[1] = out[1];
  aes_iv[2] = out[2];
  aes_iv[3] = out[3];
}

KERNEL_FQ void m25900_init(KERN_ATTR_TMPS(pbkdf2_sha256_tmp_t))
{
  /**
   * base
   */

  const u64 gid = get_global_id(0);

  if (gid >= gid_max) return;

  sha256_hmac_ctx_t sha256_hmac_ctx;

  sha256_hmac_init_global_swap(&sha256_hmac_ctx, pws[gid].i, pws[gid].pw_len);

  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].salt_buf, salt_bufs[SALT_POS].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 m25900_loop(KERN_ATTR_TMPS(pbkdf2_sha256_tmp_t))
{
  const u64 gid = get_global_id(0);

  if ((gid * VECT_SIZE) >= gid_max) 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 m25900_comp(KERN_ATTR_TMPS_ESALT(pbkdf2_sha256_tmp_t, blocks_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_max) return;

  u32 key[4];

  key[0] = tmps[gid].out[DGST_R0];
  key[1] = tmps[gid].out[DGST_R1];
  key[2] = tmps[gid].out[DGST_R2];
  key[3] = tmps[gid].out[DGST_R3];

  u32 aes_ks[44];

  AES128_set_encrypt_key (aes_ks, key, s_te0, s_te1, s_te2, s_te3);

  u32 b0[4] = { 0 };

  u32 aes_cbc_iv[4] = { 0 };

  u32 yn[4];

  const u32 digest_pos = loop_pos;
  const u32 digest_cur = DIGESTS_OFFSET + digest_pos;

  u32 b1[4];

  b1[0] = esalt_bufs[digest_cur].b1[0];
  b1[1] = esalt_bufs[digest_cur].b1[1];
  b1[2] = esalt_bufs[digest_cur].b1[2];
  b1[3] = esalt_bufs[digest_cur].b1[3];

  u32 b2[4];

  b2[0] = esalt_bufs[digest_cur].b2[0];
  b2[1] = esalt_bufs[digest_cur].b2[1];
  b2[2] = esalt_bufs[digest_cur].b2[2];
  b2[3] = esalt_bufs[digest_cur].b2[3];

  u32 b3[4];

  b3[0] = esalt_bufs[digest_cur].b3[0];
  b3[1] = esalt_bufs[digest_cur].b3[1];
  b3[2] = esalt_bufs[digest_cur].b3[2];
  b3[3] = esalt_bufs[digest_cur].b3[3];

  aes128_encrypt_cbc (aes_ks, aes_cbc_iv, b0, yn, s_te0, s_te1, s_te2, s_te3, s_te4);
  aes128_encrypt_cbc (aes_ks, aes_cbc_iv, b1, yn, s_te0, s_te1, s_te2, s_te3, s_te4);
  aes128_encrypt_cbc (aes_ks, aes_cbc_iv, b2, yn, s_te0, s_te1, s_te2, s_te3, s_te4);
  aes128_encrypt_cbc (aes_ks, aes_cbc_iv, b3, yn, s_te0, s_te1, s_te2, s_te3, s_te4);
 
  u32 nonce[4];

  nonce[0] = 0;
  nonce[1] = 0;
  nonce[2] = 0;
  nonce[3] = 0x00ff0000;  // already swapped

  u32 s0[4];

  aes128_encrypt(aes_ks, nonce, s0, s_te0, s_te1, s_te2, s_te3, s_te4);

  const u32 r0 = yn[0] ^ s0[0];
  const u32 r1 = yn[1] ^ s0[1];
  const u32 r2 = yn[2] ^ s0[2];
  const u32 r3 = yn[3] ^ s0[3];

#define il_pos 0

#ifdef KERNEL_STATIC
#include COMPARE_M
#endif
}