/**
 * 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_optimized.h)
#include M2S(INCLUDE_PATH/inc_rp_optimized.cl)
#include M2S(INCLUDE_PATH/inc_simd.cl)
#include M2S(INCLUDE_PATH/inc_cipher_aes.cl)
#endif

KERNEL_FQ void m26403_m04 (KERN_ATTR_RULES ())
{
  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;

  /**
   * modifier
   */

  u32 pw_buf0[4];
  u32 pw_buf1[4];

  pw_buf0[0] = pws[gid].i[0];
  pw_buf0[1] = pws[gid].i[1];
  pw_buf0[2] = pws[gid].i[2];
  pw_buf0[3] = pws[gid].i[3];
  pw_buf1[0] = pws[gid].i[4];
  pw_buf1[1] = pws[gid].i[5];
  pw_buf1[2] = pws[gid].i[6];
  pw_buf1[3] = pws[gid].i[7];

  const u32 pw_len = pws[gid].pw_len & 63;

  /**
   * Salt prep
   */

  u32 pt[4];

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

  /**
   * loop
   */

  for (u32 il_pos = 0; il_pos < IL_CNT; il_pos += VECT_SIZE)
  {
    u32 w0[4] = { 0 };
    u32 w1[4] = { 0 };

    // ignore output length
    apply_rules_vect_optimized (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);

    u32 ukey[8];

    ukey[0] = w0[0];
    ukey[1] = w0[1];
    ukey[2] = w0[2];
    ukey[3] = w0[3];
    ukey[4] = w1[0];
    ukey[5] = w1[1];
    ukey[6] = w1[2];
    ukey[7] = w1[3];

    #define KEYLEN 60

    u32 ks[KEYLEN];

    aes256_set_encrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3);

    u32 ct[4];

    aes256_encrypt (ks, pt, ct, s_te0, s_te1, s_te2, s_te3, s_te4);

    const u32x r0 = ct[0];
    const u32x r1 = ct[1];
    const u32x r2 = ct[2];
    const u32x r3 = ct[3];

    COMPARE_M_SIMD (r0, r1, r2, r3);
  }
}

KERNEL_FQ void m26403_m08 (KERN_ATTR_RULES ())
{
}

KERNEL_FQ void m26403_m16 (KERN_ATTR_RULES ())
{
}

KERNEL_FQ void m26403_s04 (KERN_ATTR_RULES ())
{
  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;

  /**
   * modifier
   */

  u32 pw_buf0[4];
  u32 pw_buf1[4];

  pw_buf0[0] = pws[gid].i[0];
  pw_buf0[1] = pws[gid].i[1];
  pw_buf0[2] = pws[gid].i[2];
  pw_buf0[3] = pws[gid].i[3];
  pw_buf1[0] = pws[gid].i[4];
  pw_buf1[1] = pws[gid].i[5];
  pw_buf1[2] = pws[gid].i[6];
  pw_buf1[3] = pws[gid].i[7];

  const u32 pw_len = pws[gid].pw_len & 63;

  /**
   * Salt prep
   */

  u32 pt[4];

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

  /**
   * digest
   */

  const u32 search[4] =
  {
    digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R0],
    digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R1],
    digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R2],
    digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R3]
  };

  /**
   * loop
   */

  for (u32 il_pos = 0; il_pos < IL_CNT; il_pos += VECT_SIZE)
  {
    u32 w0[4] = { 0 };
    u32 w1[4] = { 0 };

    // ignore output length
    apply_rules_vect_optimized (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);

    u32 ukey[8];

    ukey[0] = w0[0];
    ukey[1] = w0[1];
    ukey[2] = w0[2];
    ukey[3] = w0[3];
    ukey[4] = w1[0];
    ukey[5] = w1[1];
    ukey[6] = w1[2];
    ukey[7] = w1[3];

    #define KEYLEN 60

    u32 ks[KEYLEN];

    aes256_set_encrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3);

    u32 ct[4];

    aes256_encrypt (ks, pt, ct, s_te0, s_te1, s_te2, s_te3, s_te4);

    const u32x r0 = ct[0];
    const u32x r1 = ct[1];
    const u32x r2 = ct[2];
    const u32x r3 = ct[3];

    COMPARE_S_SIMD (r0, r1, r2, r3);
  }
}

KERNEL_FQ void m26403_s08 (KERN_ATTR_RULES ())
{
}

KERNEL_FQ void m26403_s16 (KERN_ATTR_RULES ())
{
}