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

#ifdef KERNEL_STATIC
#include "inc_vendor.h"
#include "inc_types.h"
#include "inc_platform.cl"
#include "inc_common.cl"
#include "inc_hash_sha1.cl"
#include "inc_cipher_aes.cl"
#endif

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

#define ROUNDS 0x40000

typedef struct rar3_tmp
{
  u32 dgst[5];

  u32 w[66]; // 256 byte pass + 8 byte salt

  u32 iv[4];

} rar3_tmp_t;

DECLSPEC void memcat8c_be (u32 *w0, u32 *w1, u32 *w2, u32 *w3, const u32 len, const u32 append, u32 *digest)
{
  const u32 func_len = len & 63;

  //const u32 mod = func_len & 3;
  const u32 div = func_len / 4;

  u32 tmp0;
  u32 tmp1;

  #if defined IS_AMD || defined IS_GENERIC
  tmp0 = hc_bytealign_be (0, append, func_len);
  tmp1 = hc_bytealign_be (append, 0, func_len);
  #endif

  #ifdef IS_NV
  const int selector = (0x76543210 >> ((func_len & 3) * 4)) & 0xffff;

  tmp0 = hc_byte_perm (append, 0, selector);
  tmp1 = hc_byte_perm (0, append, selector);
  #endif

  u32 carry = 0;

  switch (div)
  {
    case  0:  w0[0] |= tmp0;
              w0[1]  = tmp1;
              break;
    case  1:  w0[1] |= tmp0;
              w0[2]  = tmp1;
              break;
    case  2:  w0[2] |= tmp0;
              w0[3]  = tmp1;
              break;
    case  3:  w0[3] |= tmp0;
              w1[0]  = tmp1;
              break;
    case  4:  w1[0] |= tmp0;
              w1[1]  = tmp1;
              break;
    case  5:  w1[1] |= tmp0;
              w1[2]  = tmp1;
              break;
    case  6:  w1[2] |= tmp0;
              w1[3]  = tmp1;
              break;
    case  7:  w1[3] |= tmp0;
              w2[0]  = tmp1;
              break;
    case  8:  w2[0] |= tmp0;
              w2[1]  = tmp1;
              break;
    case  9:  w2[1] |= tmp0;
              w2[2]  = tmp1;
              break;
    case 10:  w2[2] |= tmp0;
              w2[3]  = tmp1;
              break;
    case 11:  w2[3] |= tmp0;
              w3[0]  = tmp1;
              break;
    case 12:  w3[0] |= tmp0;
              w3[1]  = tmp1;
              break;
    case 13:  w3[1] |= tmp0;
              w3[2]  = tmp1;
              break;
    case 14:  w3[2] |= tmp0;
              w3[3]  = tmp1;
              break;
    case 15:  w3[3] |= tmp0;
              carry  = tmp1;
              break;
  }

  const u32 new_len = func_len + 3;

  if (new_len >= 64)
  {
    sha1_transform (w0, w1, w2, w3, digest);

    w0[0] = carry;
    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;
  }
}

// only change in this function compared to OpenCL/inc_hash_sha1.cl is that it returns
// the expanded 64 byte buffer w0_t..wf_t in t[]:

DECLSPEC void sha1_transform_rar29 (const u32 *w0, const u32 *w1, const u32 *w2, const u32 *w3, u32 *digest, u32 *t)
{
  u32 a = digest[0];
  u32 b = digest[1];
  u32 c = digest[2];
  u32 d = digest[3];
  u32 e = digest[4];

  #ifdef IS_CPU

  u32 w0_t = w0[0];
  u32 w1_t = w0[1];
  u32 w2_t = w0[2];
  u32 w3_t = w0[3];
  u32 w4_t = w1[0];
  u32 w5_t = w1[1];
  u32 w6_t = w1[2];
  u32 w7_t = w1[3];
  u32 w8_t = w2[0];
  u32 w9_t = w2[1];
  u32 wa_t = w2[2];
  u32 wb_t = w2[3];
  u32 wc_t = w3[0];
  u32 wd_t = w3[1];
  u32 we_t = w3[2];
  u32 wf_t = w3[3];

  #define K SHA1C00

  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w0_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w1_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w2_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w3_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w4_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w5_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w6_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w7_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w8_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w9_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, wa_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, wb_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, wc_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, wd_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, we_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, wf_t);
  w0_t = hc_rotl32_S ((wd_t ^ w8_t ^ w2_t ^ w0_t), 1u); SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w0_t);
  w1_t = hc_rotl32_S ((we_t ^ w9_t ^ w3_t ^ w1_t), 1u); SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w1_t);
  w2_t = hc_rotl32_S ((wf_t ^ wa_t ^ w4_t ^ w2_t), 1u); SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w2_t);
  w3_t = hc_rotl32_S ((w0_t ^ wb_t ^ w5_t ^ w3_t), 1u); SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w3_t);

  #undef K
  #define K SHA1C01

  w4_t = hc_rotl32_S ((w1_t ^ wc_t ^ w6_t ^ w4_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w4_t);
  w5_t = hc_rotl32_S ((w2_t ^ wd_t ^ w7_t ^ w5_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w5_t);
  w6_t = hc_rotl32_S ((w3_t ^ we_t ^ w8_t ^ w6_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w6_t);
  w7_t = hc_rotl32_S ((w4_t ^ wf_t ^ w9_t ^ w7_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w7_t);
  w8_t = hc_rotl32_S ((w5_t ^ w0_t ^ wa_t ^ w8_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w8_t);
  w9_t = hc_rotl32_S ((w6_t ^ w1_t ^ wb_t ^ w9_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w9_t);
  wa_t = hc_rotl32_S ((w7_t ^ w2_t ^ wc_t ^ wa_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, wa_t);
  wb_t = hc_rotl32_S ((w8_t ^ w3_t ^ wd_t ^ wb_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, wb_t);
  wc_t = hc_rotl32_S ((w9_t ^ w4_t ^ we_t ^ wc_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, wc_t);
  wd_t = hc_rotl32_S ((wa_t ^ w5_t ^ wf_t ^ wd_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, wd_t);
  we_t = hc_rotl32_S ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, we_t);
  wf_t = hc_rotl32_S ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, wf_t);
  w0_t = hc_rotl32_S ((wd_t ^ w8_t ^ w2_t ^ w0_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w0_t);
  w1_t = hc_rotl32_S ((we_t ^ w9_t ^ w3_t ^ w1_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w1_t);
  w2_t = hc_rotl32_S ((wf_t ^ wa_t ^ w4_t ^ w2_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w2_t);
  w3_t = hc_rotl32_S ((w0_t ^ wb_t ^ w5_t ^ w3_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w3_t);
  w4_t = hc_rotl32_S ((w1_t ^ wc_t ^ w6_t ^ w4_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w4_t);
  w5_t = hc_rotl32_S ((w2_t ^ wd_t ^ w7_t ^ w5_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w5_t);
  w6_t = hc_rotl32_S ((w3_t ^ we_t ^ w8_t ^ w6_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w6_t);
  w7_t = hc_rotl32_S ((w4_t ^ wf_t ^ w9_t ^ w7_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w7_t);

  #undef K
  #define K SHA1C02

  w8_t = hc_rotl32_S ((w5_t ^ w0_t ^ wa_t ^ w8_t), 1u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w8_t);
  w9_t = hc_rotl32_S ((w6_t ^ w1_t ^ wb_t ^ w9_t), 1u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w9_t);
  wa_t = hc_rotl32_S ((w7_t ^ w2_t ^ wc_t ^ wa_t), 1u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, wa_t);
  wb_t = hc_rotl32_S ((w8_t ^ w3_t ^ wd_t ^ wb_t), 1u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, wb_t);
  wc_t = hc_rotl32_S ((w9_t ^ w4_t ^ we_t ^ wc_t), 1u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, wc_t);
  wd_t = hc_rotl32_S ((wa_t ^ w5_t ^ wf_t ^ wd_t), 1u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, wd_t);
  we_t = hc_rotl32_S ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, we_t);
  wf_t = hc_rotl32_S ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, wf_t);
  w0_t = hc_rotl32_S ((wd_t ^ w8_t ^ w2_t ^ w0_t), 1u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w0_t);
  w1_t = hc_rotl32_S ((we_t ^ w9_t ^ w3_t ^ w1_t), 1u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w1_t);
  w2_t = hc_rotl32_S ((wf_t ^ wa_t ^ w4_t ^ w2_t), 1u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w2_t);
  w3_t = hc_rotl32_S ((w0_t ^ wb_t ^ w5_t ^ w3_t), 1u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w3_t);
  w4_t = hc_rotl32_S ((w1_t ^ wc_t ^ w6_t ^ w4_t), 1u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w4_t);
  w5_t = hc_rotl32_S ((w2_t ^ wd_t ^ w7_t ^ w5_t), 1u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w5_t);
  w6_t = hc_rotl32_S ((w3_t ^ we_t ^ w8_t ^ w6_t), 1u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w6_t);
  w7_t = hc_rotl32_S ((w4_t ^ wf_t ^ w9_t ^ w7_t), 1u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w7_t);
  w8_t = hc_rotl32_S ((w5_t ^ w0_t ^ wa_t ^ w8_t), 1u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w8_t);
  w9_t = hc_rotl32_S ((w6_t ^ w1_t ^ wb_t ^ w9_t), 1u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w9_t);
  wa_t = hc_rotl32_S ((w7_t ^ w2_t ^ wc_t ^ wa_t), 1u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, wa_t);
  wb_t = hc_rotl32_S ((w8_t ^ w3_t ^ wd_t ^ wb_t), 1u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, wb_t);

  #undef K
  #define K SHA1C03

  wc_t = hc_rotl32_S ((w9_t ^ w4_t ^ we_t ^ wc_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, wc_t);
  wd_t = hc_rotl32_S ((wa_t ^ w5_t ^ wf_t ^ wd_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, wd_t);
  we_t = hc_rotl32_S ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, we_t);
  wf_t = hc_rotl32_S ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, wf_t);
  w0_t = hc_rotl32_S ((wd_t ^ w8_t ^ w2_t ^ w0_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w0_t);
  w1_t = hc_rotl32_S ((we_t ^ w9_t ^ w3_t ^ w1_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w1_t);
  w2_t = hc_rotl32_S ((wf_t ^ wa_t ^ w4_t ^ w2_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w2_t);
  w3_t = hc_rotl32_S ((w0_t ^ wb_t ^ w5_t ^ w3_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w3_t);
  w4_t = hc_rotl32_S ((w1_t ^ wc_t ^ w6_t ^ w4_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w4_t);
  w5_t = hc_rotl32_S ((w2_t ^ wd_t ^ w7_t ^ w5_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w5_t);
  w6_t = hc_rotl32_S ((w3_t ^ we_t ^ w8_t ^ w6_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w6_t);
  w7_t = hc_rotl32_S ((w4_t ^ wf_t ^ w9_t ^ w7_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w7_t);
  w8_t = hc_rotl32_S ((w5_t ^ w0_t ^ wa_t ^ w8_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w8_t);
  w9_t = hc_rotl32_S ((w6_t ^ w1_t ^ wb_t ^ w9_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w9_t);
  wa_t = hc_rotl32_S ((w7_t ^ w2_t ^ wc_t ^ wa_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, wa_t);
  wb_t = hc_rotl32_S ((w8_t ^ w3_t ^ wd_t ^ wb_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, wb_t);
  wc_t = hc_rotl32_S ((w9_t ^ w4_t ^ we_t ^ wc_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, wc_t);
  wd_t = hc_rotl32_S ((wa_t ^ w5_t ^ wf_t ^ wd_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, wd_t);
  we_t = hc_rotl32_S ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, we_t);
  wf_t = hc_rotl32_S ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, wf_t);

  t[ 0] = w0_t;
  t[ 1] = w1_t;
  t[ 2] = w2_t;
  t[ 3] = w3_t;
  t[ 4] = w4_t;
  t[ 5] = w5_t;
  t[ 6] = w6_t;
  t[ 7] = w7_t;
  t[ 8] = w8_t;
  t[ 9] = w9_t;
  t[10] = wa_t;
  t[11] = wb_t;
  t[12] = wc_t;
  t[13] = wd_t;
  t[14] = we_t;
  t[15] = wf_t;

  #undef K

  #else

  u32 w00_t = w0[0];
  u32 w01_t = w0[1];
  u32 w02_t = w0[2];
  u32 w03_t = w0[3];
  u32 w04_t = w1[0];
  u32 w05_t = w1[1];
  u32 w06_t = w1[2];
  u32 w07_t = w1[3];
  u32 w08_t = w2[0];
  u32 w09_t = w2[1];
  u32 w0a_t = w2[2];
  u32 w0b_t = w2[3];
  u32 w0c_t = w3[0];
  u32 w0d_t = w3[1];
  u32 w0e_t = w3[2];
  u32 w0f_t = w3[3];
  u32 w10_t;
  u32 w11_t;
  u32 w12_t;
  u32 w13_t;
  u32 w14_t;
  u32 w15_t;
  u32 w16_t;
  u32 w17_t;
  u32 w18_t;
  u32 w19_t;
  u32 w1a_t;
  u32 w1b_t;
  u32 w1c_t;
  u32 w1d_t;
  u32 w1e_t;
  u32 w1f_t;
  u32 w20_t;
  u32 w21_t;
  u32 w22_t;
  u32 w23_t;
  u32 w24_t;
  u32 w25_t;
  u32 w26_t;
  u32 w27_t;
  u32 w28_t;
  u32 w29_t;
  u32 w2a_t;
  u32 w2b_t;
  u32 w2c_t;
  u32 w2d_t;
  u32 w2e_t;
  u32 w2f_t;
  u32 w30_t;
  u32 w31_t;
  u32 w32_t;
  u32 w33_t;
  u32 w34_t;
  u32 w35_t;
  u32 w36_t;
  u32 w37_t;
  u32 w38_t;
  u32 w39_t;
  u32 w3a_t;
  u32 w3b_t;
  u32 w3c_t;
  u32 w3d_t;
  u32 w3e_t;
  u32 w3f_t;
  u32 w40_t;
  u32 w41_t;
  u32 w42_t;
  u32 w43_t;
  u32 w44_t;
  u32 w45_t;
  u32 w46_t;
  u32 w47_t;
  u32 w48_t;
  u32 w49_t;
  u32 w4a_t;
  u32 w4b_t;
  u32 w4c_t;
  u32 w4d_t;
  u32 w4e_t;
  u32 w4f_t;

  #define K SHA1C00

  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w00_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w01_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w02_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w03_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w04_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w05_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w06_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w07_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w08_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w09_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w0a_t);
  SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w0b_t);
  SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w0c_t);
  SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w0d_t);
  SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w0e_t);
  SHA1_STEP_S (SHA1_F0o, a, b, c, d, e, w0f_t);
  w10_t = hc_rotl32_S ((w0d_t ^ w08_t ^ w02_t ^ w00_t), 1u); SHA1_STEP_S (SHA1_F0o, e, a, b, c, d, w10_t);
  w11_t = hc_rotl32_S ((w0e_t ^ w09_t ^ w03_t ^ w01_t), 1u); SHA1_STEP_S (SHA1_F0o, d, e, a, b, c, w11_t);
  w12_t = hc_rotl32_S ((w0f_t ^ w0a_t ^ w04_t ^ w02_t), 1u); SHA1_STEP_S (SHA1_F0o, c, d, e, a, b, w12_t);
  w13_t = hc_rotl32_S ((w10_t ^ w0b_t ^ w05_t ^ w03_t), 1u); SHA1_STEP_S (SHA1_F0o, b, c, d, e, a, w13_t);

  #undef K
  #define K SHA1C01

  w14_t = hc_rotl32_S ((w11_t ^ w0c_t ^ w06_t ^ w04_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w14_t);
  w15_t = hc_rotl32_S ((w12_t ^ w0d_t ^ w07_t ^ w05_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w15_t);
  w16_t = hc_rotl32_S ((w13_t ^ w0e_t ^ w08_t ^ w06_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w16_t);
  w17_t = hc_rotl32_S ((w14_t ^ w0f_t ^ w09_t ^ w07_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w17_t);
  w18_t = hc_rotl32_S ((w15_t ^ w10_t ^ w0a_t ^ w08_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w18_t);
  w19_t = hc_rotl32_S ((w16_t ^ w11_t ^ w0b_t ^ w09_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w19_t);
  w1a_t = hc_rotl32_S ((w17_t ^ w12_t ^ w0c_t ^ w0a_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w1a_t);
  w1b_t = hc_rotl32_S ((w18_t ^ w13_t ^ w0d_t ^ w0b_t), 1u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w1b_t);
  w1c_t = hc_rotl32_S ((w19_t ^ w14_t ^ w0e_t ^ w0c_t), 1u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w1c_t);
  w1d_t = hc_rotl32_S ((w1a_t ^ w15_t ^ w0f_t ^ w0d_t), 1u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w1d_t);
  w1e_t = hc_rotl32_S ((w1b_t ^ w16_t ^ w10_t ^ w0e_t), 1u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w1e_t);
  w1f_t = hc_rotl32_S ((w1c_t ^ w17_t ^ w11_t ^ w0f_t), 1u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w1f_t);
  w20_t = hc_rotl32_S ((w1a_t ^ w10_t ^ w04_t ^ w00_t), 2u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w20_t);
  w21_t = hc_rotl32_S ((w1b_t ^ w11_t ^ w05_t ^ w01_t), 2u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w21_t);
  w22_t = hc_rotl32_S ((w1c_t ^ w12_t ^ w06_t ^ w02_t), 2u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w22_t);
  w23_t = hc_rotl32_S ((w1d_t ^ w13_t ^ w07_t ^ w03_t), 2u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w23_t);
  w24_t = hc_rotl32_S ((w1e_t ^ w14_t ^ w08_t ^ w04_t), 2u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w24_t);
  w25_t = hc_rotl32_S ((w1f_t ^ w15_t ^ w09_t ^ w05_t), 2u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w25_t);
  w26_t = hc_rotl32_S ((w20_t ^ w16_t ^ w0a_t ^ w06_t), 2u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w26_t);
  w27_t = hc_rotl32_S ((w21_t ^ w17_t ^ w0b_t ^ w07_t), 2u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w27_t);

  #undef K
  #define K SHA1C02

  w28_t = hc_rotl32_S ((w22_t ^ w18_t ^ w0c_t ^ w08_t), 2u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w28_t);
  w29_t = hc_rotl32_S ((w23_t ^ w19_t ^ w0d_t ^ w09_t), 2u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w29_t);
  w2a_t = hc_rotl32_S ((w24_t ^ w1a_t ^ w0e_t ^ w0a_t), 2u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w2a_t);
  w2b_t = hc_rotl32_S ((w25_t ^ w1b_t ^ w0f_t ^ w0b_t), 2u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w2b_t);
  w2c_t = hc_rotl32_S ((w26_t ^ w1c_t ^ w10_t ^ w0c_t), 2u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w2c_t);
  w2d_t = hc_rotl32_S ((w27_t ^ w1d_t ^ w11_t ^ w0d_t), 2u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w2d_t);
  w2e_t = hc_rotl32_S ((w28_t ^ w1e_t ^ w12_t ^ w0e_t), 2u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w2e_t);
  w2f_t = hc_rotl32_S ((w29_t ^ w1f_t ^ w13_t ^ w0f_t), 2u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w2f_t);
  w30_t = hc_rotl32_S ((w2a_t ^ w20_t ^ w14_t ^ w10_t), 2u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w30_t);
  w31_t = hc_rotl32_S ((w2b_t ^ w21_t ^ w15_t ^ w11_t), 2u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w31_t);
  w32_t = hc_rotl32_S ((w2c_t ^ w22_t ^ w16_t ^ w12_t), 2u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w32_t);
  w33_t = hc_rotl32_S ((w2d_t ^ w23_t ^ w17_t ^ w13_t), 2u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w33_t);
  w34_t = hc_rotl32_S ((w2e_t ^ w24_t ^ w18_t ^ w14_t), 2u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w34_t);
  w35_t = hc_rotl32_S ((w2f_t ^ w25_t ^ w19_t ^ w15_t), 2u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w35_t);
  w36_t = hc_rotl32_S ((w30_t ^ w26_t ^ w1a_t ^ w16_t), 2u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w36_t);
  w37_t = hc_rotl32_S ((w31_t ^ w27_t ^ w1b_t ^ w17_t), 2u); SHA1_STEP_S (SHA1_F2o, a, b, c, d, e, w37_t);
  w38_t = hc_rotl32_S ((w32_t ^ w28_t ^ w1c_t ^ w18_t), 2u); SHA1_STEP_S (SHA1_F2o, e, a, b, c, d, w38_t);
  w39_t = hc_rotl32_S ((w33_t ^ w29_t ^ w1d_t ^ w19_t), 2u); SHA1_STEP_S (SHA1_F2o, d, e, a, b, c, w39_t);
  w3a_t = hc_rotl32_S ((w34_t ^ w2a_t ^ w1e_t ^ w1a_t), 2u); SHA1_STEP_S (SHA1_F2o, c, d, e, a, b, w3a_t);
  w3b_t = hc_rotl32_S ((w35_t ^ w2b_t ^ w1f_t ^ w1b_t), 2u); SHA1_STEP_S (SHA1_F2o, b, c, d, e, a, w3b_t);

  #undef K
  #define K SHA1C03

  w3c_t = hc_rotl32_S ((w36_t ^ w2c_t ^ w20_t ^ w1c_t), 2u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w3c_t);
  w3d_t = hc_rotl32_S ((w37_t ^ w2d_t ^ w21_t ^ w1d_t), 2u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w3d_t);
  w3e_t = hc_rotl32_S ((w38_t ^ w2e_t ^ w22_t ^ w1e_t), 2u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w3e_t);
  w3f_t = hc_rotl32_S ((w39_t ^ w2f_t ^ w23_t ^ w1f_t), 2u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w3f_t);
  w40_t = hc_rotl32_S ((w34_t ^ w20_t ^ w08_t ^ w00_t), 4u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w40_t);
  w41_t = hc_rotl32_S ((w35_t ^ w21_t ^ w09_t ^ w01_t), 4u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w41_t);
  w42_t = hc_rotl32_S ((w36_t ^ w22_t ^ w0a_t ^ w02_t), 4u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w42_t);
  w43_t = hc_rotl32_S ((w37_t ^ w23_t ^ w0b_t ^ w03_t), 4u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w43_t);
  w44_t = hc_rotl32_S ((w38_t ^ w24_t ^ w0c_t ^ w04_t), 4u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w44_t);
  w45_t = hc_rotl32_S ((w39_t ^ w25_t ^ w0d_t ^ w05_t), 4u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w45_t);
  w46_t = hc_rotl32_S ((w3a_t ^ w26_t ^ w0e_t ^ w06_t), 4u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w46_t);
  w47_t = hc_rotl32_S ((w3b_t ^ w27_t ^ w0f_t ^ w07_t), 4u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w47_t);
  w48_t = hc_rotl32_S ((w3c_t ^ w28_t ^ w10_t ^ w08_t), 4u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w48_t);
  w49_t = hc_rotl32_S ((w3d_t ^ w29_t ^ w11_t ^ w09_t), 4u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w49_t);
  w4a_t = hc_rotl32_S ((w3e_t ^ w2a_t ^ w12_t ^ w0a_t), 4u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w4a_t);
  w4b_t = hc_rotl32_S ((w3f_t ^ w2b_t ^ w13_t ^ w0b_t), 4u); SHA1_STEP_S (SHA1_F1, a, b, c, d, e, w4b_t);
  w4c_t = hc_rotl32_S ((w40_t ^ w2c_t ^ w14_t ^ w0c_t), 4u); SHA1_STEP_S (SHA1_F1, e, a, b, c, d, w4c_t);
  w4d_t = hc_rotl32_S ((w41_t ^ w2d_t ^ w15_t ^ w0d_t), 4u); SHA1_STEP_S (SHA1_F1, d, e, a, b, c, w4d_t);
  w4e_t = hc_rotl32_S ((w42_t ^ w2e_t ^ w16_t ^ w0e_t), 4u); SHA1_STEP_S (SHA1_F1, c, d, e, a, b, w4e_t);
  w4f_t = hc_rotl32_S ((w43_t ^ w2f_t ^ w17_t ^ w0f_t), 4u); SHA1_STEP_S (SHA1_F1, b, c, d, e, a, w4f_t);

  t[ 0] = w40_t;
  t[ 1] = w41_t;
  t[ 2] = w42_t;
  t[ 3] = w43_t;
  t[ 4] = w44_t;
  t[ 5] = w45_t;
  t[ 6] = w46_t;
  t[ 7] = w47_t;
  t[ 8] = w48_t;
  t[ 9] = w49_t;
  t[10] = w4a_t;
  t[11] = w4b_t;
  t[12] = w4c_t;
  t[13] = w4d_t;
  t[14] = w4e_t;
  t[15] = w4f_t;

  #undef K
  #endif

  digest[0] += a;
  digest[1] += b;
  digest[2] += c;
  digest[3] += d;
  digest[4] += e;
}

// only change in this function compared to OpenCL/inc_hash_sha1.cl is that
// it calls our modified sha1_transform_rar29 () function

DECLSPEC void sha1_update_64_rar29 (sha1_ctx_t *ctx, u32 *w0, u32 *w1, u32 *w2, u32 *w3, const int bytes, u32 *t)
{
  MAYBE_VOLATILE const int pos = ctx->len & 63;

  int len = 64;

  if (bytes < 64)
  {
    len = bytes;
  }

  ctx->len += len;

  if (pos == 0)
  {
    ctx->w0[0] = w0[0];
    ctx->w0[1] = w0[1];
    ctx->w0[2] = w0[2];
    ctx->w0[3] = w0[3];
    ctx->w1[0] = w1[0];
    ctx->w1[1] = w1[1];
    ctx->w1[2] = w1[2];
    ctx->w1[3] = w1[3];
    ctx->w2[0] = w2[0];
    ctx->w2[1] = w2[1];
    ctx->w2[2] = w2[2];
    ctx->w2[3] = w2[3];
    ctx->w3[0] = w3[0];
    ctx->w3[1] = w3[1];
    ctx->w3[2] = w3[2];
    ctx->w3[3] = w3[3];

    if (len == 64)
    {
      sha1_transform_rar29 (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h, t);

      ctx->w0[0] = 0;
      ctx->w0[1] = 0;
      ctx->w0[2] = 0;
      ctx->w0[3] = 0;
      ctx->w1[0] = 0;
      ctx->w1[1] = 0;
      ctx->w1[2] = 0;
      ctx->w1[3] = 0;
      ctx->w2[0] = 0;
      ctx->w2[1] = 0;
      ctx->w2[2] = 0;
      ctx->w2[3] = 0;
      ctx->w3[0] = 0;
      ctx->w3[1] = 0;
      ctx->w3[2] = 0;
      ctx->w3[3] = 0;
    }
  }
  else
  {
    if ((pos + len) < 64)
    {
      switch_buffer_by_offset_be_S (w0, w1, w2, w3, pos);

      ctx->w0[0] |= w0[0];
      ctx->w0[1] |= w0[1];
      ctx->w0[2] |= w0[2];
      ctx->w0[3] |= w0[3];
      ctx->w1[0] |= w1[0];
      ctx->w1[1] |= w1[1];
      ctx->w1[2] |= w1[2];
      ctx->w1[3] |= w1[3];
      ctx->w2[0] |= w2[0];
      ctx->w2[1] |= w2[1];
      ctx->w2[2] |= w2[2];
      ctx->w2[3] |= w2[3];
      ctx->w3[0] |= w3[0];
      ctx->w3[1] |= w3[1];
      ctx->w3[2] |= w3[2];
      ctx->w3[3] |= w3[3];
    }
    else
    {
      u32 c0[4] = { 0 };
      u32 c1[4] = { 0 };
      u32 c2[4] = { 0 };
      u32 c3[4] = { 0 };

      switch_buffer_by_offset_carry_be_S (w0, w1, w2, w3, c0, c1, c2, c3, pos);

      ctx->w0[0] |= w0[0];
      ctx->w0[1] |= w0[1];
      ctx->w0[2] |= w0[2];
      ctx->w0[3] |= w0[3];
      ctx->w1[0] |= w1[0];
      ctx->w1[1] |= w1[1];
      ctx->w1[2] |= w1[2];
      ctx->w1[3] |= w1[3];
      ctx->w2[0] |= w2[0];
      ctx->w2[1] |= w2[1];
      ctx->w2[2] |= w2[2];
      ctx->w2[3] |= w2[3];
      ctx->w3[0] |= w3[0];
      ctx->w3[1] |= w3[1];
      ctx->w3[2] |= w3[2];
      ctx->w3[3] |= w3[3];

      sha1_transform_rar29 (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h, t);

      ctx->w0[0] = c0[0];
      ctx->w0[1] = c0[1];
      ctx->w0[2] = c0[2];
      ctx->w0[3] = c0[3];
      ctx->w1[0] = c1[0];
      ctx->w1[1] = c1[1];
      ctx->w1[2] = c1[2];
      ctx->w1[3] = c1[3];
      ctx->w2[0] = c2[0];
      ctx->w2[1] = c2[1];
      ctx->w2[2] = c2[2];
      ctx->w2[3] = c2[3];
      ctx->w3[0] = c3[0];
      ctx->w3[1] = c3[1];
      ctx->w3[2] = c3[2];
      ctx->w3[3] = c3[3];
    }
  }
}

// main change in this function compared to OpenCL/inc_hash_sha1.cl is that
// we call sha1_update_64_rar29 () and sometimes replace w[]

DECLSPEC void sha1_update_rar29 (sha1_ctx_t *ctx, u32 *w, const int len)
{
  u32 w0[4];
  u32 w1[4];
  u32 w2[4];
  u32 w3[4];

  MAYBE_VOLATILE const int pos = ctx->len & 63;

  int pos1 = 0;
  int pos4 = 0;

  if (len > 64) // or: if (pos1 < (len - 64))
  {
    w0[0] = w[pos4 +  0];
    w0[1] = w[pos4 +  1];
    w0[2] = w[pos4 +  2];
    w0[3] = w[pos4 +  3];
    w1[0] = w[pos4 +  4];
    w1[1] = w[pos4 +  5];
    w1[2] = w[pos4 +  6];
    w1[3] = w[pos4 +  7];
    w2[0] = w[pos4 +  8];
    w2[1] = w[pos4 +  9];
    w2[2] = w[pos4 + 10];
    w2[3] = w[pos4 + 11];
    w3[0] = w[pos4 + 12];
    w3[1] = w[pos4 + 13];
    w3[2] = w[pos4 + 14];
    w3[3] = w[pos4 + 15];

    sha1_update_64 (ctx, w0, w1, w2, w3, 64);

    pos1 += 64;
    pos4 += 16;
  }

  for (int diff = 64 - pos; pos1 < len; pos1 += 64, pos4 += 16, diff += 64)
  {
    w0[0] = w[pos4 +  0];
    w0[1] = w[pos4 +  1];
    w0[2] = w[pos4 +  2];
    w0[3] = w[pos4 +  3];
    w1[0] = w[pos4 +  4];
    w1[1] = w[pos4 +  5];
    w1[2] = w[pos4 +  6];
    w1[3] = w[pos4 +  7];
    w2[0] = w[pos4 +  8];
    w2[1] = w[pos4 +  9];
    w2[2] = w[pos4 + 10];
    w2[3] = w[pos4 + 11];
    w3[0] = w[pos4 + 12];
    w3[1] = w[pos4 + 13];
    w3[2] = w[pos4 + 14];
    w3[3] = w[pos4 + 15];

    // only major change in this function compared to OpenCL/inc_hash_sha1.cl:

    u32 t[17] = { 0 };

    sha1_update_64_rar29 (ctx, w0, w1, w2, w3, len - pos1, t);


    if ((diff + 63) >= len) break;

    // replaces 64 bytes (with offset diff) of the underlying data w[] with t[]:

    // for (int i = 0; i < 16; i++) t[i] = hc_swap32_S (t[i]);

    t[ 0] = hc_swap32_S (t[ 0]); // unroll seems to be faster
    t[ 1] = hc_swap32_S (t[ 1]);
    t[ 2] = hc_swap32_S (t[ 2]);
    t[ 3] = hc_swap32_S (t[ 3]);
    t[ 4] = hc_swap32_S (t[ 4]);
    t[ 5] = hc_swap32_S (t[ 5]);
    t[ 6] = hc_swap32_S (t[ 6]);
    t[ 7] = hc_swap32_S (t[ 7]);
    t[ 8] = hc_swap32_S (t[ 8]);
    t[ 9] = hc_swap32_S (t[ 9]);
    t[10] = hc_swap32_S (t[10]);
    t[11] = hc_swap32_S (t[11]);
    t[12] = hc_swap32_S (t[12]);
    t[13] = hc_swap32_S (t[13]);
    t[14] = hc_swap32_S (t[14]);
    t[15] = hc_swap32_S (t[15]);

    const u32 n_idx = diff / 4;
    const u32 n_off = diff % 4;

    if (n_off)
    {
      const u32 off_mul = n_off * 8;
      const u32 off_sub = 32 - off_mul;

      t[16] =                      (t[15] << off_sub);
      t[15] = (t[15] >> off_mul) | (t[14] << off_sub);
      t[14] = (t[14] >> off_mul) | (t[13] << off_sub);
      t[13] = (t[13] >> off_mul) | (t[12] << off_sub);
      t[12] = (t[12] >> off_mul) | (t[11] << off_sub);
      t[11] = (t[11] >> off_mul) | (t[10] << off_sub);
      t[10] = (t[10] >> off_mul) | (t[ 9] << off_sub);
      t[ 9] = (t[ 9] >> off_mul) | (t[ 8] << off_sub);
      t[ 8] = (t[ 8] >> off_mul) | (t[ 7] << off_sub);
      t[ 7] = (t[ 7] >> off_mul) | (t[ 6] << off_sub);
      t[ 6] = (t[ 6] >> off_mul) | (t[ 5] << off_sub);
      t[ 5] = (t[ 5] >> off_mul) | (t[ 4] << off_sub);
      t[ 4] = (t[ 4] >> off_mul) | (t[ 3] << off_sub);
      t[ 3] = (t[ 3] >> off_mul) | (t[ 2] << off_sub);
      t[ 2] = (t[ 2] >> off_mul) | (t[ 1] << off_sub);
      t[ 1] = (t[ 1] >> off_mul) | (t[ 0] << off_sub);
      t[ 0] = (t[ 0] >> off_mul);
    }

    w[n_idx] &= 0xffffff00 << ((3 - n_off) * 8);

    w[n_idx] |= t[0];

    w[n_idx +  1] = t[ 1];
    w[n_idx +  2] = t[ 2];
    w[n_idx +  3] = t[ 3];
    w[n_idx +  4] = t[ 4];
    w[n_idx +  5] = t[ 5];
    w[n_idx +  6] = t[ 6];
    w[n_idx +  7] = t[ 7];
    w[n_idx +  8] = t[ 8];
    w[n_idx +  9] = t[ 9];
    w[n_idx + 10] = t[10];
    w[n_idx + 11] = t[11];
    w[n_idx + 12] = t[12];
    w[n_idx + 13] = t[13];
    w[n_idx + 14] = t[14];
    w[n_idx + 15] = t[15];

    // the final set is only meaningful: if (n_off)

    w[n_idx + 16] &= 0xffffffff >> (n_off * 8);

    w[n_idx + 16] |= t[16];
  }
}

KERNEL_FQ void m12500_init (KERN_ATTR_TMPS (rar3_tmp_t))
{
  /**
   * base
   */

  const u64 gid = get_global_id (0);

  if (gid >= gid_max) return;

  tmps[gid].dgst[0] = SHA1M_A;
  tmps[gid].dgst[1] = SHA1M_B;
  tmps[gid].dgst[2] = SHA1M_C;
  tmps[gid].dgst[3] = SHA1M_D;
  tmps[gid].dgst[4] = SHA1M_E;

  // store pass and salt in tmps:

  const u32 pw_len = pws[gid].pw_len;

  // first set the utf16le pass:

  u32 w[80] = { 0 };

  for (u32 i = 0, j = 0, k = 0; i < pw_len; i += 16, j += 4, k += 8)
  {
    u32 a[4];

    a[0] = pws[gid].i[j + 0];
    a[1] = pws[gid].i[j + 1];
    a[2] = pws[gid].i[j + 2];
    a[3] = pws[gid].i[j + 3];

    u32 b[4];
    u32 c[4];

    make_utf16le (a, b, c);

    w[k + 0] = hc_swap32_S (b[0]);
    w[k + 1] = hc_swap32_S (b[1]);
    w[k + 2] = hc_swap32_S (b[2]);
    w[k + 3] = hc_swap32_S (b[3]);
    w[k + 4] = hc_swap32_S (c[0]);
    w[k + 5] = hc_swap32_S (c[1]);
    w[k + 6] = hc_swap32_S (c[2]);
    w[k + 7] = hc_swap32_S (c[3]);
  }

  // append salt:

  const u32 salt_idx = (pw_len * 2) / 4;
  const u32 salt_off = (pw_len * 2) & 3;

  u32 salt_buf[3];

  salt_buf[0] = hc_swap32_S (salt_bufs[salt_pos].salt_buf[0]); // swap needed due to -O kernel
  salt_buf[1] = hc_swap32_S (salt_bufs[salt_pos].salt_buf[1]);
  salt_buf[2] = 0;

  // switch buffer by offset (can only be 0 or 2 because of utf16):

  if (salt_off == 2) // or just: if (salt_off)
  {
    salt_buf[2] =                       (salt_buf[1] << 16);
    salt_buf[1] = (salt_buf[1] >> 16) | (salt_buf[0] << 16);
    salt_buf[0] = (salt_buf[0] >> 16);
  }

  w[salt_idx] |= salt_buf[0];

  w[salt_idx + 1] = salt_buf[1];
  w[salt_idx + 2] = salt_buf[2];

  // store initial w[] (pass and salt) in tmps:

  for (u32 i = 0; i < 66; i++) // unroll ?
  {
    tmps[gid].w[i] = w[i];
  }

  // iv:

  tmps[gid].iv[0] = 0;
  tmps[gid].iv[1] = 0;
  tmps[gid].iv[2] = 0;
  tmps[gid].iv[3] = 0;
}

KERNEL_FQ void m12500_loop (KERN_ATTR_TMPS (rar3_tmp_t))
{
  const u64 gid = get_global_id (0);

  if (gid >= gid_max) return;

  /**
   * base
   */

  const u32 pw_len = pws[gid].pw_len;

  const u32 salt_len = 8;

  const u32 pw_salt_len = (pw_len * 2) + salt_len;

  const u32 p3 = pw_salt_len + 3;

  u32 w[80] = { 0 }; // 64 byte aligned

  for (u32 i = 0; i < 66; i++) // unroll ?
  {
    w[i] = tmps[gid].w[i];
  }

  // update IV:

  const u32 init_pos = loop_pos / (ROUNDS / 16);

  sha1_ctx_t ctx_iv;

  sha1_init (&ctx_iv);

  ctx_iv.h[0] = tmps[gid].dgst[0];
  ctx_iv.h[1] = tmps[gid].dgst[1];
  ctx_iv.h[2] = tmps[gid].dgst[2];
  ctx_iv.h[3] = tmps[gid].dgst[3];
  ctx_iv.h[4] = tmps[gid].dgst[4];

  ctx_iv.len = loop_pos * p3;

  sha1_update_rar29 (&ctx_iv, w, pw_salt_len);

  memcat8c_be (ctx_iv.w0, ctx_iv.w1, ctx_iv.w2, ctx_iv.w3, ctx_iv.len, hc_swap32_S (loop_pos), ctx_iv.h);

  ctx_iv.len += 3;


  // copy the context from ctx_iv to ctx:

  sha1_ctx_t ctx;

  ctx.h[0] = ctx_iv.h[0];
  ctx.h[1] = ctx_iv.h[1];
  ctx.h[2] = ctx_iv.h[2];
  ctx.h[3] = ctx_iv.h[3];
  ctx.h[4] = ctx_iv.h[4];

  ctx.w0[0] = ctx_iv.w0[0];
  ctx.w0[1] = ctx_iv.w0[1];
  ctx.w0[2] = ctx_iv.w0[2];
  ctx.w0[3] = ctx_iv.w0[3];

  ctx.w1[0] = ctx_iv.w1[0];
  ctx.w1[1] = ctx_iv.w1[1];
  ctx.w1[2] = ctx_iv.w1[2];
  ctx.w1[3] = ctx_iv.w1[3];

  ctx.w2[0] = ctx_iv.w2[0];
  ctx.w2[1] = ctx_iv.w2[1];
  ctx.w2[2] = ctx_iv.w2[2];
  ctx.w2[3] = ctx_iv.w2[3];

  ctx.w3[0] = ctx_iv.w3[0];
  ctx.w3[1] = ctx_iv.w3[1];
  ctx.w3[2] = ctx_iv.w3[2];
  ctx.w3[3] = ctx_iv.w3[3];

  ctx.len = p3; // or ctx_iv.len ?

  // final () for the IV byte:

  sha1_final (&ctx_iv);

  const u32 iv_idx = init_pos / 4;
  const u32 iv_off = init_pos % 4;

  tmps[gid].iv[iv_idx] |= (ctx_iv.h[4] & 0xff) << (iv_off * 8);

  // main loop:

  for (u32 i = 0, j = (loop_pos + 1); i < 16383; i++, j++)
  {
    sha1_update_rar29 (&ctx, w, pw_salt_len);

    memcat8c_be (ctx.w0, ctx.w1, ctx.w2, ctx.w3, ctx.len, hc_swap32_S (j), ctx.h);

    ctx.len += 3;
  }

  tmps[gid].dgst[0] = ctx.h[0];
  tmps[gid].dgst[1] = ctx.h[1];
  tmps[gid].dgst[2] = ctx.h[2];
  tmps[gid].dgst[3] = ctx.h[3];
  tmps[gid].dgst[4] = ctx.h[4];

  // only needed if pw_len > 28:

  for (u32 i = 0; i < 66; i++) // unroll ?
  {
    tmps[gid].w[i] = w[i];
  }
}

KERNEL_FQ void m12500_comp (KERN_ATTR_TMPS (rar3_tmp_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_max) return;

  /**
   * base
   */

  const u32 pw_len = pws[gid].pw_len;

  const u32 salt_len = 8;

  const u32 pw_salt_len = (pw_len * 2) + salt_len;

  const u32 p3 = pw_salt_len + 3;

  u32 h[5];

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

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

  w0[0] = 0x80000000;
  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] = (ROUNDS * p3) * 8;

  sha1_transform (w0, w1, w2, w3, h);

  u32 ukey[4];

  ukey[0] = hc_swap32_S (h[0]);
  ukey[1] = hc_swap32_S (h[1]);
  ukey[2] = hc_swap32_S (h[2]);
  ukey[3] = hc_swap32_S (h[3]);

  u32 ks[44];

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

  u32 data[4];

  data[0] = salt_bufs[salt_pos].salt_buf[2];
  data[1] = salt_bufs[salt_pos].salt_buf[3];
  data[2] = salt_bufs[salt_pos].salt_buf[4];
  data[3] = salt_bufs[salt_pos].salt_buf[5];

  u32 out[4];

  AES128_decrypt (ks, data, out, s_td0, s_td1, s_td2, s_td3, s_td4);

  u32 iv[2];

  iv[0] = tmps[gid].iv[0];
  iv[1] = tmps[gid].iv[1];

  out[0] ^= hc_swap32_S (iv[0]);
  out[1] ^= hc_swap32_S (iv[1]);

  const u32 r0 = out[0];
  const u32 r1 = out[1];
  const u32 r2 = 0;
  const u32 r3 = 0;

  #define il_pos 0

  #ifdef KERNEL_STATIC
  #include COMPARE_M
  #endif
}