/**
 * 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_simd.cl)
#include M2S(INCLUDE_PATH/inc_hash_sha1.cl)
#include M2S(INCLUDE_PATH/inc_hash_sha256.cl)
#include M2S(INCLUDE_PATH/inc_cipher_aes.cl)
#endif

typedef struct iwork_tmp
{
  u32 ipad[5];
  u32 opad[5];

  u32 dgst[5];
  u32 out[5];

} iwork_tmp_t;

typedef struct iwork
{
  u32 iv[4];
  u32 data[16];

} iwork_t;

DECLSPEC void hmac_sha1_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];

  sha1_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] = 0x80000000;
  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] = (64 + 20) * 8;

  digest[0] = opad[0];
  digest[1] = opad[1];
  digest[2] = opad[2];
  digest[3] = opad[3];
  digest[4] = opad[4];

  sha1_transform_vector (w0, w1, w2, w3, digest);
}

KERNEL_FQ void m23300_init (KERN_ATTR_TMPS_ESALT (iwork_tmp_t, iwork_t))
{
  /**
   * base
   */

  const u64 gid = get_global_id (0);

  if (gid >= GID_CNT) return;

  sha1_hmac_ctx_t sha1_hmac_ctx;

  sha1_hmac_init_global_swap (&sha1_hmac_ctx, pws[gid].i, pws[gid].pw_len);

  tmps[gid].ipad[0] = sha1_hmac_ctx.ipad.h[0];
  tmps[gid].ipad[1] = sha1_hmac_ctx.ipad.h[1];
  tmps[gid].ipad[2] = sha1_hmac_ctx.ipad.h[2];
  tmps[gid].ipad[3] = sha1_hmac_ctx.ipad.h[3];
  tmps[gid].ipad[4] = sha1_hmac_ctx.ipad.h[4];

  tmps[gid].opad[0] = sha1_hmac_ctx.opad.h[0];
  tmps[gid].opad[1] = sha1_hmac_ctx.opad.h[1];
  tmps[gid].opad[2] = sha1_hmac_ctx.opad.h[2];
  tmps[gid].opad[3] = sha1_hmac_ctx.opad.h[3];
  tmps[gid].opad[4] = sha1_hmac_ctx.opad.h[4];

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

  w0[0] = salt_bufs[SALT_POS_HOST].salt_buf[0];
  w0[1] = salt_bufs[SALT_POS_HOST].salt_buf[1];
  w0[2] = salt_bufs[SALT_POS_HOST].salt_buf[2];
  w0[3] = salt_bufs[SALT_POS_HOST].salt_buf[3];
  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;

  sha1_hmac_update_64 (&sha1_hmac_ctx, w0, w1, w2, w3, salt_bufs[SALT_POS_HOST].salt_len);

  for (u32 i = 0, j = 1; i < 4; i += 5, j += 1)
  {
    sha1_hmac_ctx_t sha1_hmac_ctx2 = sha1_hmac_ctx;

    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;

    sha1_hmac_update_64 (&sha1_hmac_ctx2, w0, w1, w2, w3, 4);

    sha1_hmac_final (&sha1_hmac_ctx2);

    tmps[gid].dgst[i + 0] = sha1_hmac_ctx2.opad.h[0];
    tmps[gid].dgst[i + 1] = sha1_hmac_ctx2.opad.h[1];
    tmps[gid].dgst[i + 2] = sha1_hmac_ctx2.opad.h[2];
    tmps[gid].dgst[i + 3] = sha1_hmac_ctx2.opad.h[3];
    tmps[gid].dgst[i + 4] = sha1_hmac_ctx2.opad.h[4];

    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];
  }
}

KERNEL_FQ void m23300_loop (KERN_ATTR_TMPS_ESALT (iwork_tmp_t, iwork_t))
{
  const u64 gid = get_global_id (0);

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

  u32x ipad[5];
  u32x opad[5];

  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);

  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);

  for (u32 i = 0; i < 4; i += 5)
  {
    u32x dgst[5];
    u32x out[5];

    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);

    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);

    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] = 0x80000000;
      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] = (64 + 20) * 8;

      hmac_sha1_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];
    }

    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, 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]);
  }
}

KERNEL_FQ void m23300_comp (KERN_ATTR_TMPS_ESALT (iwork_tmp_t, iwork_t))
{
  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;

  /**
   * AES part
   */

  u32 ukey[4];

  ukey[0] = tmps[gid].out[0];
  ukey[1] = tmps[gid].out[1];
  ukey[2] = tmps[gid].out[2];
  ukey[3] = tmps[gid].out[3];

  #define KEYLEN 44

  u32 ks[KEYLEN];

  AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3);

  u32 iv[4];

  iv[0] = esalt_bufs[DIGESTS_OFFSET_HOST].iv[0];
  iv[1] = esalt_bufs[DIGESTS_OFFSET_HOST].iv[1];
  iv[2] = esalt_bufs[DIGESTS_OFFSET_HOST].iv[2];
  iv[3] = esalt_bufs[DIGESTS_OFFSET_HOST].iv[3];

  u32 res[12]; // actually res[16], but we don't need the full 64 bytes output

  for (u32 i = 0; i < 12; i += 4)
  {
    u32 data[4];

    data[0] = esalt_bufs[DIGESTS_OFFSET_HOST].data[i + 0];
    data[1] = esalt_bufs[DIGESTS_OFFSET_HOST].data[i + 1];
    data[2] = esalt_bufs[DIGESTS_OFFSET_HOST].data[i + 2];
    data[3] = esalt_bufs[DIGESTS_OFFSET_HOST].data[i + 3];

    u32 out[4];

    aes128_decrypt (ks, data, out, s_td0, s_td1, s_td2, s_td3, s_td4);

    res[i + 0] = out[0] ^ iv[0];
    res[i + 1] = out[1] ^ iv[1];
    res[i + 2] = out[2] ^ iv[2];
    res[i + 3] = out[3] ^ iv[3];

    iv[0] = data[0];
    iv[1] = data[1];
    iv[2] = data[2];
    iv[3] = data[3];
  }

  sha256_ctx_t ctx;

  sha256_init (&ctx);

  u32 w0[4];

  w0[0] = hc_swap32_S (res[0]);
  w0[1] = hc_swap32_S (res[1]);
  w0[2] = hc_swap32_S (res[2]);
  w0[3] = hc_swap32_S (res[3]);

  u32 w1[4];

  w1[0] = hc_swap32_S (res[4]);
  w1[1] = hc_swap32_S (res[5]);
  w1[2] = hc_swap32_S (res[6]);
  w1[3] = hc_swap32_S (res[7]);

  u32 w2[4];

  w2[0] = 0;
  w2[1] = 0;
  w2[2] = 0;
  w2[3] = 0;

  u32 w3[4];

  w3[0] = 0;
  w3[1] = 0;
  w3[2] = 0;
  w3[3] = 0;

  sha256_update_64 (&ctx, w0, w1, w2, w3, 32);

  sha256_final (&ctx);

  u32 checksum[4];

  checksum[0] = hc_swap32_S (ctx.h[0]);
  checksum[1] = hc_swap32_S (ctx.h[1]);
  checksum[2] = hc_swap32_S (ctx.h[2]);
  checksum[3] = hc_swap32_S (ctx.h[3]);

  if ((res[ 8] == checksum[0]) &&
      (res[ 9] == checksum[1]) &&
      (res[10] == checksum[2]) &&
      (res[11] == checksum[3]))
  {
    if (hc_atomic_inc (&hashes_shown[DIGESTS_OFFSET_HOST]) == 0)
    {
      mark_hash (plains_buf, d_return_buf, SALT_POS_HOST, DIGESTS_CNT, 0, DIGESTS_OFFSET_HOST + 0, gid, 0, 0, 0);
    }

    return;
  }
}