/**
 * 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_sha1.cl"
#else
#include "inc_vendor.h"
#include "inc_types.h"
#include "inc_platform.h"
#include "inc_common.h"
#include "inc_simd.h"
#include "inc_hash_sha1.h"
#endif

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

typedef struct wpa_pbkdf2_tmp
{
  u32 ipad[5];
  u32 opad[5];

  u32 dgst[10];
  u32 out[10];

} wpa_pbkdf2_tmp_t;

typedef struct wpa_pmkid
{
  u32  pmkid[4];
  u32  pmkid_data[16];
  u8   orig_mac_ap[6];
  u8   orig_mac_sta[6];
  u8   essid_len;
  u32  essid_buf[16];

} wpa_pmkid_t;

DECLSPEC void hmac_sha1_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];

  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 m16800_init (KERN_ATTR_TMPS_ESALT (wpa_pbkdf2_tmp_t, wpa_pmkid_t))
{
  /**
   * base
   */

  const u64 gid = get_global_id (0);

  if (gid >= gid_max) 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];

  sha1_hmac_update_global_swap (&sha1_hmac_ctx, esalt_bufs[DIGESTS_OFFSET].essid_buf, esalt_bufs[DIGESTS_OFFSET].essid_len);

  for (u32 i = 0, j = 1; i < 8; i += 5, j += 1)
  {
    sha1_hmac_ctx_t sha1_hmac_ctx2 = sha1_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;

    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 m16800_loop (KERN_ATTR_TMPS_ESALT (wpa_pbkdf2_tmp_t, wpa_pmkid_t))
{
  const u64 gid = get_global_id (0);

  if ((gid * VECT_SIZE) >= gid_max) 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 < 8; 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 m16800_comp (KERN_ATTR_TMPS_ESALT (wpa_pbkdf2_tmp_t, wpa_pmkid_t))
{
  // not in use here, special case...
}

KERNEL_FQ void m16800_aux1 (KERN_ATTR_TMPS_ESALT (wpa_pbkdf2_tmp_t, wpa_pmkid_t))
{
  const u64 gid = get_global_id (0);

  if (gid >= gid_max) return;

  u32 w[16];

  w[ 0] = tmps[gid].out[0];
  w[ 1] = tmps[gid].out[1];
  w[ 2] = tmps[gid].out[2];
  w[ 3] = tmps[gid].out[3];
  w[ 4] = tmps[gid].out[4];
  w[ 5] = tmps[gid].out[5];
  w[ 6] = tmps[gid].out[6];
  w[ 7] = tmps[gid].out[7];
  w[ 8] = 0;
  w[ 9] = 0;
  w[10] = 0;
  w[11] = 0;
  w[12] = 0;
  w[13] = 0;
  w[14] = 0;
  w[15] = 0;

  const u32 digest_pos = loop_pos;

  const u32 digest_cur = DIGESTS_OFFSET + digest_pos;

  GLOBAL_AS const wpa_pmkid_t *wpa_pmkid = &esalt_bufs[digest_cur];

  sha1_hmac_ctx_t sha1_hmac_ctx;

  sha1_hmac_init (&sha1_hmac_ctx, w, 32);

  sha1_hmac_update_global_swap (&sha1_hmac_ctx, wpa_pmkid->pmkid_data, 20);

  sha1_hmac_final (&sha1_hmac_ctx);

  const u32 r0 = sha1_hmac_ctx.opad.h[0];
  const u32 r1 = sha1_hmac_ctx.opad.h[1];
  const u32 r2 = sha1_hmac_ctx.opad.h[2];
  const u32 r3 = sha1_hmac_ctx.opad.h[3];

  #ifdef KERNEL_STATIC

  #define il_pos 0
  #include COMPARE_M

  #else

  if ((hc_swap32_S (r0) == wpa_pmkid->pmkid[0])
   && (hc_swap32_S (r1) == wpa_pmkid->pmkid[1])
   && (hc_swap32_S (r2) == wpa_pmkid->pmkid[2])
   && (hc_swap32_S (r3) == wpa_pmkid->pmkid[3]))
  {
    if (hc_atomic_inc (&hashes_shown[digest_cur]) == 0)
    {
      mark_hash (plains_buf, d_return_buf, SALT_POS, digests_cnt, digest_pos, digest_cur, gid, 0, 0, 0);
    }
  }

  #endif
}