Move volatiles for AMD closer to the problem

OpenCL/inc_common.cl

@@ -1266,6 +1266,12 @@ static void switch_buffer_by_offset_le (u32x w0[4], u32x w1[4], u32x w2[4], u32x
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
   w0[0] = swap32 (w0[0]);
   w0[1] = swap32 (w0[1]);
@@ -1284,7 +1290,7 @@ static void switch_buffer_by_offset_le (u32x w0[4], u32x w1[4], u32x w2[4], u32x
   w3[2] = swap32 (w3[2]);
   w3[3] = swap32 (w3[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = amd_bytealign (w3[2], w3[3], offset);
@@ -1635,7 +1641,7 @@ static void switch_buffer_by_offset_le (u32x w0[4], u32x w1[4], u32x w2[4], u32x
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = __byte_perm (w3[2], w3[3], selector);
@@ -1967,6 +1973,12 @@ static void switch_buffer_by_offset_carry_le (u32x w0[4], u32x w1[4], u32x w2[4]
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if defined IS_AMD || defined IS_GENERIC
   w0[0] = swap32 (w0[0]);
   w0[1] = swap32 (w0[1]);
@@ -1985,7 +1997,7 @@ static void switch_buffer_by_offset_carry_le (u32x w0[4], u32x w1[4], u32x w2[4]
   w3[2] = swap32 (w3[2]);
   w3[3] = swap32 (w3[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign (w3[3],     0, offset);
@@ -2480,7 +2492,7 @@ static void switch_buffer_by_offset_carry_le (u32x w0[4], u32x w1[4], u32x w2[4]
 
   #ifdef IS_NV
   // todo
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       c0[0] = amd_bytealign (    0, w3[3], offset_minus_4);
@@ -3279,9 +3291,15 @@ static void switch_buffer_by_offset_carry_le (u32x w0[4], u32x w1[4], u32x w2[4]
 
 static void switch_buffer_by_offset_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = amd_bytealign (w3[2], w3[3], offset);
@@ -3616,7 +3634,7 @@ static void switch_buffer_by_offset_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = __byte_perm (w3[3], w3[2], selector);
@@ -3944,8 +3962,14 @@ static void switch_buffer_by_offset_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x
 
 static void switch_buffer_by_offset_carry_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x c0[4], u32x c1[4], u32x c2[4], u32x c3[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign (w3[3],     0, offset);
@@ -4415,7 +4439,7 @@ static void switch_buffer_by_offset_carry_be (u32x w0[4], u32x w1[4], u32x w2[4]
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = __byte_perm (    0, w3[3], selector);
@@ -4882,6 +4906,12 @@ static void switch_buffer_by_offset_8x4_le (u32x w0[4], u32x w1[4], u32x w2[4],
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
   w0[0] = swap32 (w0[0]);
   w0[1] = swap32 (w0[1]);
@@ -4916,7 +4946,7 @@ static void switch_buffer_by_offset_8x4_le (u32x w0[4], u32x w1[4], u32x w2[4],
   w7[2] = swap32 (w7[2]);
   w7[3] = swap32 (w7[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = amd_bytealign (w7[2], w7[3], offset);
@@ -6115,7 +6145,7 @@ static void switch_buffer_by_offset_8x4_le (u32x w0[4], u32x w1[4], u32x w2[4],
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       w7[3] = __byte_perm (w7[2], w7[3], selector);
@@ -6682,8 +6712,14 @@ static void switch_buffer_by_offset_8x4_le (u32x w0[4], u32x w1[4], u32x w2[4],
 
 static void switch_buffer_by_offset_8x4_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = amd_bytealign (w7[2], w7[3], offset);
@@ -7849,7 +7885,7 @@ static void switch_buffer_by_offset_8x4_be (u32x w0[4], u32x w1[4], u32x w2[4],
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = __byte_perm (w7[3], w7[2], selector);
@@ -9008,8 +9044,14 @@ static void switch_buffer_by_offset_8x4_be (u32x w0[4], u32x w1[4], u32x w2[4],
 
 static void switch_buffer_by_offset_8x4_carry_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], u32x c0[4], u32x c1[4], u32x c2[4], u32x c3[4], u32x c4[4], u32x c5[4], u32x c6[4], u32x c7[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign (w7[3],     0, offset);
@@ -10703,7 +10745,7 @@ static void switch_buffer_by_offset_8x4_carry_be (u32x w0[4], u32x w1[4], u32x w
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = __byte_perm (    0, w7[3], selector);
@@ -12394,12 +12436,18 @@ static void switch_buffer_by_offset_1x64_le (u32x w[64], const u32 offset)
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
 
   #pragma unroll
   for (int i = 0; i < 64; i++) w[i] = swap32 (w[i]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = amd_bytealign (w[62], w[63], offset);
@@ -16769,7 +16817,7 @@ static void switch_buffer_by_offset_1x64_le (u32x w[64], const u32 offset)
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = __byte_perm (w[62], w[63], selector);
@@ -21128,8 +21176,14 @@ static void switch_buffer_by_offset_1x64_le (u32x w[64], const u32 offset)
 
 static void switch_buffer_by_offset_1x64_be (u32x w[64], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = amd_bytealign (w[62], w[63], offset);
@@ -25495,7 +25549,7 @@ static void switch_buffer_by_offset_1x64_be (u32x w[64], const u32 offset)
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = __byte_perm (w[63], w[62], selector);
@@ -32416,6 +32470,12 @@ static void switch_buffer_by_offset_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
   w0[0] = swap32_S (w0[0]);
   w0[1] = swap32_S (w0[1]);
@@ -32434,7 +32494,7 @@ static void switch_buffer_by_offset_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w
   w3[2] = swap32_S (w3[2]);
   w3[3] = swap32_S (w3[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = amd_bytealign_S (w3[2], w3[3], offset);
@@ -32785,7 +32845,7 @@ static void switch_buffer_by_offset_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = __byte_perm_S (w3[2], w3[3], selector);
@@ -33116,6 +33176,12 @@ static void switch_buffer_by_offset_carry_le_S (u32 w0[4], u32 w1[4], u32 w2[4],
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if defined IS_AMD || defined IS_GENERIC
   w0[0] = swap32_S (w0[0]);
   w0[1] = swap32_S (w0[1]);
@@ -33134,7 +33200,7 @@ static void switch_buffer_by_offset_carry_le_S (u32 w0[4], u32 w1[4], u32 w2[4],
   w3[2] = swap32_S (w3[2]);
   w3[3] = swap32_S (w3[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign_S (w3[3],     0, offset);
@@ -33629,7 +33695,7 @@ static void switch_buffer_by_offset_carry_le_S (u32 w0[4], u32 w1[4], u32 w2[4],
 
   #ifdef IS_NV
   // todo
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       c0[0] = amd_bytealign_S (    0, w3[3], offset_minus_4);
@@ -34428,8 +34494,14 @@ static void switch_buffer_by_offset_carry_le_S (u32 w0[4], u32 w1[4], u32 w2[4],
 
 static void switch_buffer_by_offset_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = amd_bytealign_S (w3[2], w3[3], offset);
@@ -34763,7 +34835,7 @@ static void switch_buffer_by_offset_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w3[3] = __byte_perm_S (w3[3], w3[2], selector);
@@ -35090,8 +35162,14 @@ static void switch_buffer_by_offset_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w
 
 static void switch_buffer_by_offset_carry_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 c0[4], u32 c1[4], u32 c2[4], u32 c3[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign_S (w3[3],     0, offset);
@@ -35561,7 +35639,7 @@ static void switch_buffer_by_offset_carry_be_S (u32 w0[4], u32 w1[4], u32 w2[4],
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = __byte_perm_S (    0, w3[3], selector);
@@ -36028,6 +36106,12 @@ static void switch_buffer_by_offset_8x4_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
   w0[0] = swap32_S (w0[0]);
   w0[1] = swap32_S (w0[1]);
@@ -36062,7 +36146,7 @@ static void switch_buffer_by_offset_8x4_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u
   w7[2] = swap32_S (w7[2]);
   w7[3] = swap32_S (w7[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = amd_bytealign_S (w7[2], w7[3], offset);
@@ -37261,7 +37345,7 @@ static void switch_buffer_by_offset_8x4_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       w7[3] = __byte_perm_S (w7[2], w7[3], selector);
@@ -37828,8 +37912,14 @@ static void switch_buffer_by_offset_8x4_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u
 
 static void switch_buffer_by_offset_8x4_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = amd_bytealign_S (w7[2], w7[3], offset);
@@ -38995,7 +39085,7 @@ static void switch_buffer_by_offset_8x4_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w7[3] = __byte_perm_S (w7[3], w7[2], selector);
@@ -40154,8 +40244,14 @@ static void switch_buffer_by_offset_8x4_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u
 
 static void switch_buffer_by_offset_8x4_carry_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], u32 c0[4], u32 c1[4], u32 c2[4], u32 c3[4], u32 c4[4], u32 c5[4], u32 c6[4], u32 c7[4], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = amd_bytealign_S (w7[3],     0, offset);
@@ -41849,7 +41945,7 @@ static void switch_buffer_by_offset_8x4_carry_be_S (u32 w0[4], u32 w1[4], u32 w2
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       c0[0] = __byte_perm_S (    0, w7[3], selector);
@@ -43540,12 +43636,18 @@ static void switch_buffer_by_offset_1x64_le_S (u32 w[64], const u32 offset)
 
   const int offset_minus_4 = 4 - offset_mod_4;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
 
   #pragma unroll
   for (int i = 0; i < 64; i++) w[i] = swap32_S (w[i]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = amd_bytealign_S (w[62], w[63], offset);
@@ -47915,7 +48017,7 @@ static void switch_buffer_by_offset_1x64_le_S (u32 w[64], const u32 offset)
   const int selector = 0x0706050403020100 >> (offset_minus_4 * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = __byte_perm_S (w[62], w[63], selector);
@@ -52274,8 +52376,14 @@ static void switch_buffer_by_offset_1x64_le_S (u32 w[64], const u32 offset)
 
 static void switch_buffer_by_offset_1x64_be_S (u32 w[64], const u32 offset)
 {
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = amd_bytealign_S (w[62], w[63], offset);
@@ -56641,7 +56749,7 @@ static void switch_buffer_by_offset_1x64_be_S (u32 w[64], const u32 offset)
   const int selector = 0x0706050403020100 >> ((offset & 3) * 8);
   #endif
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case  0:
       w[63] = __byte_perm_S (w[63], w[62], selector);

OpenCL/inc_hash_md4.cl

@@ -110,11 +110,7 @@ static void md4_init (md4_ctx_t *ctx)
 
 static void md4_update_64 (md4_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1234,11 +1230,7 @@ static void md4_init_vector_from_scalar (md4_ctx_vector_t *ctx, md4_ctx_t *ctx0)
 
 static void md4_update_vector_64 (md4_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_md5.cl

@@ -146,11 +146,7 @@ static void md5_init (md5_ctx_t *ctx)
 
 static void md5_update_64 (md5_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1306,11 +1302,7 @@ static void md5_init_vector_from_scalar (md5_ctx_vector_t *ctx, md5_ctx_t *ctx0)
 
 static void md5_update_vector_64 (md5_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_ripemd160.cl

@@ -244,11 +244,7 @@ static void ripemd160_init (ripemd160_ctx_t *ctx)
 
 static void ripemd160_update_64 (ripemd160_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1503,11 +1499,7 @@ static void ripemd160_init_vector_from_scalar (ripemd160_ctx_vector_t *ctx, ripe
 
 static void ripemd160_update_vector_64 (ripemd160_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_sha1.cl

@@ -176,11 +176,7 @@ static void sha1_init (sha1_ctx_t *ctx)
 
 static void sha1_update_64 (sha1_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1603,11 +1599,7 @@ static void sha1_init_vector_from_scalar (sha1_ctx_vector_t *ctx, sha1_ctx_t *ct
 
 static void sha1_update_vector_64 (sha1_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_sha224.cl

@@ -161,11 +161,7 @@ static void sha224_init (sha224_ctx_t *ctx)
 
 static void sha224_update_64 (sha224_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1320,11 +1316,7 @@ static void sha224_init_vector_from_scalar (sha224_ctx_vector_t *ctx, sha224_ctx
 
 static void sha224_update_vector_64 (sha224_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_sha256.cl

@@ -161,11 +161,7 @@ static void sha256_init (sha256_ctx_t *ctx)
 
 static void sha256_update_64 (sha256_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 
@@ -1320,11 +1316,7 @@ static void sha256_init_vector_from_scalar (sha256_ctx_vector_t *ctx, sha256_ctx
 
 static void sha256_update_vector_64 (sha256_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_sha384.cl

@@ -185,11 +185,7 @@ static void sha384_init (sha384_ctx_t *ctx)
 
 static void sha384_update_128 (sha384_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 127;
-  #else
   const int pos = ctx->len & 127;
-  #endif
 
   ctx->len += len;
 
@@ -2016,11 +2012,7 @@ static void sha384_init_vector_from_scalar (sha384_ctx_vector_t *ctx, sha384_ctx
 
 static void sha384_update_vector_128 (sha384_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 127;
-  #else
   const int pos = ctx->len & 127;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_sha512.cl

@@ -185,11 +185,7 @@ static void sha512_init (sha512_ctx_t *ctx)
 
 static void sha512_update_128 (sha512_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 127;
-  #else
   const int pos = ctx->len & 127;
-  #endif
 
   ctx->len += len;
 
@@ -2016,11 +2012,7 @@ static void sha512_init_vector_from_scalar (sha512_ctx_vector_t *ctx, sha512_ctx
 
 static void sha512_update_vector_128 (sha512_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 127;
-  #else
   const int pos = ctx->len & 127;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_hash_whirlpool.cl

@@ -2607,11 +2607,7 @@ static void whirlpool_init_vector_from_scalar (whirlpool_ctx_vector_t *ctx, whir
 
 static void whirlpool_update_vector_64 (whirlpool_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
 {
-  #ifdef IS_AMD
-  volatile const int pos = ctx->len & 63;
-  #else
   const int pos = ctx->len & 63;
-  #endif
 
   ctx->len += len;
 

OpenCL/inc_rp_optimized.cl

@@ -20,7 +20,13 @@ static void truncate_right (u32 buf0[4], u32 buf1[4], const u32 offset)
 {
   const u32 tmp = (1u << ((offset & 3u) * 8u)) - 1u;
 
-  switch (offset / 4)
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
+  switch (offset_switch)
   {
     case  0:  buf0[0] &= tmp;
               buf0[1]  = 0;
@@ -73,7 +79,13 @@ static void truncate_left (u32 buf0[4], u32 buf1[4], const u32 offset)
 {
   const u32 tmp = ~((1u << ((offset & 3u) * 8u)) - 1u);
 
-  switch (offset / 4)
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
+  switch (offset_switch)
   {
     case  0:  buf0[0] &= tmp;
               break;
@@ -767,6 +779,12 @@ static void append_block8 (const u32 offset, u32 buf0[4], u32 buf1[4], const u32
   u32 s6 = 0;
   u32 s7 = 0;
 
+  #ifdef IS_AMD
+  volatile const int offset_switch = offset / 4;
+  #else
+  const int offset_switch = offset / 4;
+  #endif
+
   #if (defined IS_AMD && AMD_GCN < 3) || defined IS_GENERIC
   const u32 src_r00 = swap32_S (src_r0[0]);
   const u32 src_r01 = swap32_S (src_r0[1]);
@@ -777,7 +795,7 @@ static void append_block8 (const u32 offset, u32 buf0[4], u32 buf1[4], const u32
   const u32 src_r12 = swap32_S (src_r1[2]);
   const u32 src_r13 = swap32_S (src_r1[3]);
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       s7 = amd_bytealign_S (src_r12, src_r13, offset);
@@ -902,7 +920,7 @@ static void append_block8 (const u32 offset, u32 buf0[4], u32 buf1[4], const u32
   const u32 src_r12 = src_r1[2];
   const u32 src_r13 = src_r1[3];
 
-  switch (offset / 4)
+  switch (offset_switch)
   {
     case 0:
       s7 = __byte_perm_S (src_r12, src_r13, selector);
@@ -1340,7 +1358,13 @@ static u32 rule_op_mangle_delete_at (MAYBE_UNUSED const u32 p0, MAYBE_UNUSED con
   const u32 ml = (1 << ((p0 & 3) * 8)) - 1;
   const u32 mr = ~ml;
 
-  switch (p0 / 4)
+  #ifdef IS_AMD
+  volatile const int p0_switch = p0 / 4;
+  #else
+  const int p0_switch = p0 / 4;
+  #endif
+
+  switch (p0_switch)
   {
     case  0:  buf0[0] =  (buf0[0] & ml)
                       | (tib40[0] & mr);
@@ -1441,7 +1465,13 @@ static u32 rule_op_mangle_omit (MAYBE_UNUSED const u32 p0, MAYBE_UNUSED const u3
   const u32 ml = (1 << ((p0 & 3) * 8)) - 1;
   const u32 mr = ~ml;
 
-  switch (p0 / 4)
+  #ifdef IS_AMD
+  volatile const int p0_switch = p0 / 4;
+  #else
+  const int p0_switch = p0 / 4;
+  #endif
+
+  switch (p0_switch)
   {
     case  0:  buf0[0] =  (buf0[0] & ml)
                       | (tib40[0] & mr);
@@ -1521,7 +1551,13 @@ static u32 rule_op_mangle_insert (MAYBE_UNUSED const u32 p0, MAYBE_UNUSED const
 
   const u32 mr = 0xffffff00 << ((p0 & 3) * 8);
 
-  switch (p0 / 4)
+  #ifdef IS_AMD
+  volatile const int p0_switch = p0 / 4;
+  #else
+  const int p0_switch = p0 / 4;
+  #endif
+
+  switch (p0_switch)
   {
     case  0:  buf0[0] =  (buf0[0] & ml) | p1n | (tib40[0] & mr);
               buf0[1] =  tib40[1];