Optimize out the swap32() of sha256 in 7zip

[hashcat.git] / OpenCL / m09900_a3.cl

diff --git a/OpenCL/m09900_a3.cl b/OpenCL/m09900_a3.cl
index 2b34c60..a11bf4c 100644 (file)
--- a/OpenCL/m09900_a3.cl
+++ b/OpenCL/m09900_a3.cl
@@ -20,7 +20,7 @@
 #include "OpenCL/common.c"
 #include "OpenCL/simd.c"
 
-static void m09900m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
+void m09900m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
 {
   /**
    * modifier
@@ -101,6 +101,32 @@ static void m09900m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
   const u32 I_w2c3e = w[ 2] + MD5C3e;
   const u32 I_w9c3f = w[ 9] + MD5C3f;
 
+  /**
+   * fixed buffer
+   */
+
+  u32 t0[4];
+  u32 t1[4];
+  u32 t2[4];
+  u32 t3[4];
+
+  t0[0] = 0;
+  t0[1] = 0;
+  t0[2] = 0;
+  t0[3] = 0;
+  t1[0] = 0;
+  t1[1] = 0;
+  t1[2] = 0;
+  t1[3] = 0;
+  t2[0] = 0;
+  t2[1] = 0x80;
+  t2[2] = 0;
+  t2[3] = 0;
+  t3[0] = 0;
+  t3[1] = 0;
+  t3[2] = 100 * 8;
+  t3[3] = 0;
+
   /**
    * loop
    */
@@ -196,28 +222,6 @@ static void m09900m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
     u32x r_c = c;
     u32x r_d = d;
 
-    u32x t0[4];
-    u32x t1[4];
-    u32x t2[4];
-    u32x t3[4];
-
-    t0[0] = 0;
-    t0[1] = 0;
-    t0[2] = 0;
-    t0[3] = 0;
-    t1[0] = 0;
-    t1[1] = 0;
-    t1[2] = 0;
-    t1[3] = 0;
-    t2[0] = 0;
-    t2[1] = 0x80;
-    t2[2] = 0;
-    t2[3] = 0;
-    t3[0] = 0;
-    t3[1] = 0;
-    t3[2] = 100 * 8;
-    t3[3] = 0;
-
     MD5_STEP (MD5_Fo, a, b, c, d, t0[0], MD5C00, MD5S00);
     MD5_STEP (MD5_Fo, d, a, b, c, t0[1], MD5C01, MD5S01);
     MD5_STEP (MD5_Fo, c, d, a, b, t0[2], MD5C02, MD5S02);
@@ -295,7 +299,7 @@ static void m09900m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
   }
 }
 
-static void m09900s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
+void m09900s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
 {
   /**
    * modifier
@@ -376,6 +380,32 @@ static void m09900s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
   const u32 I_w2c3e = w[ 2] + MD5C3e;
   const u32 I_w9c3f = w[ 9] + MD5C3f;
 
+  /**
+   * fixed buffer
+   */
+
+  u32 t0[4];
+  u32 t1[4];
+  u32 t2[4];
+  u32 t3[4];
+
+  t0[0] = 0;
+  t0[1] = 0;
+  t0[2] = 0;
+  t0[3] = 0;
+  t1[0] = 0;
+  t1[1] = 0;
+  t1[2] = 0;
+  t1[3] = 0;
+  t2[0] = 0;
+  t2[1] = 0x80;
+  t2[2] = 0;
+  t2[3] = 0;
+  t3[0] = 0;
+  t3[1] = 0;
+  t3[2] = 100 * 8;
+  t3[3] = 0;
+
   /**
    * digest
    */
@@ -483,28 +513,6 @@ static void m09900s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
     u32x r_c = c;
     u32x r_d = d;
 
-    u32x t0[4];
-    u32x t1[4];
-    u32x t2[4];
-    u32x t3[4];
-
-    t0[0] = 0;
-    t0[1] = 0;
-    t0[2] = 0;
-    t0[3] = 0;
-    t1[0] = 0;
-    t1[1] = 0;
-    t1[2] = 0;
-    t1[3] = 0;
-    t2[0] = 0;
-    t2[1] = 0x80;
-    t2[2] = 0;
-    t2[3] = 0;
-    t3[0] = 0;
-    t3[1] = 0;
-    t3[2] = 100 * 8;
-    t3[3] = 0;
-
     MD5_STEP (MD5_Fo, a, b, c, d, t0[0], MD5C00, MD5S00);
     MD5_STEP (MD5_Fo, d, a, b, c, t0[1], MD5C01, MD5S01);
     MD5_STEP (MD5_Fo, c, d, a, b, t0[2], MD5C02, MD5S02);