v3.23.0

2025-09-17 23:44:27 +00:00 · 2023-08-30 20:15:48 -04:00
parent 57a6b7b58b
commit 4378d2f841
72 changed files with 10184 additions and 2182 deletions
--- a/algo/groestl/aes_ni/groestl-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl-intr-aes.h
@@ -139,7 +139,7 @@ static const __m128i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003 };
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
+  b1 = _mm_set1_epi64x( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -237,7 +237,7 @@ static const __m128i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003 };
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
+  b1 = _mm_set1_epi64x( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
--- a/algo/groestl/aes_ni/groestl256-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl256-intr-aes.h
@@ -128,7 +128,7 @@ static const __m128i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
+  b1 = _mm_set1_epi64x( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -226,7 +226,7 @@ static const __m128i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
+  b1 = _mm_set1_epi64x( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -275,7 +275,7 @@ static const __m128i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };
 */
 #define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
  /* AddRoundConstant */\
-  b1 = m128_const_64( 0xffffffffffffffff, 0 ); \
+  b1 = _mm_set_epi64x( 0xffffffffffffffff, 0 ); \
  a0 = _mm_xor_si128( a0, casti_m128i( round_const_l0, i ) ); \
  a1 = _mm_xor_si128( a1, b1 ); \
  a2 = _mm_xor_si128( a2, b1 ); \
--- a/algo/groestl/aes_ni/hash-groestl.c
+++ b/algo/groestl/aes_ni/hash-groestl.c
@@ -31,7 +31,7 @@ HashReturn_gr init_groestl( hashState_groestl* ctx, int hashlen )
  }

  // The only non-zero in the IV is len. It can be hard coded.
-  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+  ctx->chaining[ 6 ] = _mm_set_epi64x( 0x0200000000000000, 0 );

  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;
@@ -48,7 +48,7 @@ HashReturn_gr reinit_groestl( hashState_groestl* ctx )
     ctx->chaining[i] = _mm_setzero_si128();
     ctx->buffer[i]   = _mm_setzero_si128();
  }
-  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+  ctx->chaining[ 6 ] = _mm_set_epi64x( 0x0200000000000000, 0 );
  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;

@@ -116,7 +116,7 @@ HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
   else
   {
       // add first padding
-       ctx->buffer[rem_ptr] = m128_const_64( 0, 0x80 );
+       ctx->buffer[rem_ptr] = _mm_set_epi64x( 0, 0x80 );
       // add zero padding
       for ( i = rem_ptr + 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();
@@ -148,7 +148,7 @@ int groestl512_full( hashState_groestl* ctx, void* output,
      ctx->chaining[i] = _mm_setzero_si128();
      ctx->buffer[i]   = _mm_setzero_si128();
   }
-   ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+   ctx->chaining[ 6 ] = _mm_set_epi64x( 0x0200000000000000, 0 );
   ctx->buf_ptr = 0;

   // --- update ---
@@ -182,7 +182,7 @@ int groestl512_full( hashState_groestl* ctx, void* output,
   else
   {
       // add first padding
-       ctx->buffer[i] = m128_const_64( 0, 0x80 );
+       ctx->buffer[i] = _mm_set_epi64x( 0, 0x80 );
       // add zero padding
       for ( i += 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();
@@ -239,7 +239,7 @@ HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
   else
   {
       // add first padding
-       ctx->buffer[i] = m128_const_64( 0, 0x80 );
+       ctx->buffer[i] = _mm_set_epi64x( 0, 0x80 );
       // add zero padding
       for ( i += 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();
--- a/algo/groestl/aes_ni/hash-groestl256.c
+++ b/algo/groestl/aes_ni/hash-groestl256.c
@@ -46,7 +46,7 @@ HashReturn_gr reinit_groestl256(hashState_groestl256* ctx)
     ctx->buffer[i]   = _mm_setzero_si128();
  }

-  ctx->chaining[ 3 ] = m128_const_64( 0, 0x0100000000000000 );
+  ctx->chaining[ 3 ] = _mm_set_epi64x( 0, 0x0100000000000000 );

  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;
--- a/algo/groestl/groestl256-intr-4way.h
+++ b/algo/groestl/groestl256-intr-4way.h
@@ -539,7 +539,7 @@ static const __m256i SUBSH_MASK7_2WAY =
  j = _mm256_cmpgt_epi8(j, i );\
  i = _mm256_add_epi8(i, i);\
  j = _mm256_and_si256(j, k);\
-  i = _mm256_xor_si256(i, j);\
+  i = mm256_xorand( i, j, k );\
 }

 #define MixBytes_2way(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
@@ -550,7 +550,7 @@ static const __m256i SUBSH_MASK7_2WAY =
  b0 = a2;\
  a1 = _mm256_xor_si256(a1, a2);\
  b1 = a3;\
-  a2 = _mm256_xor_si256(a2, a3);\
+  TEMP2 = _mm256_xor_si256(a2, a3);\
  b2 = a4;\
  a3 = _mm256_xor_si256(a3, a4);\
  b3 = a5;\
@@ -562,34 +562,20 @@ static const __m256i SUBSH_MASK7_2WAY =
  a7 = _mm256_xor_si256(a7, b6);\
  \
  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  b0 = _mm256_xor_si256(b0, a4);\
-  b6 = _mm256_xor_si256(b6, a4);\
-  b1 = _mm256_xor_si256(b1, a5);\
-  b7 = _mm256_xor_si256(b7, a5);\
-  b2 = _mm256_xor_si256(b2, a6);\
-  b0 = _mm256_xor_si256(b0, a6);\
-  /* spill values y_4, y_5 to memory */\
-  TEMP0 = b0;\
-  b3 = _mm256_xor_si256(b3, a7);\
-  b1 = _mm256_xor_si256(b1, a7);\
-  TEMP1 = b1;\
-  b4 = _mm256_xor_si256(b4, a0);\
-  b2 = _mm256_xor_si256(b2, a0);\
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  b0 = a0;\
-  b5 = _mm256_xor_si256(b5, a1);\
-  b3 = _mm256_xor_si256(b3, a1);\
-  b1 = a1;\
-  b6 = _mm256_xor_si256(b6, a2);\
-  b4 = _mm256_xor_si256(b4, a2);\
-  TEMP2 = a2;\
-  b7 = _mm256_xor_si256(b7, a3);\
-  b5 = _mm256_xor_si256(b5, a3);\
-  \
+  TEMP0 = mm256_xor3( b0, a4, a6 ); \
+  TEMP1 = mm256_xor3( b1, a5, a7 ); \
+  b2 = mm256_xor3( b2, a6, a0 ); \
+  b0 = a0; \
+  b3 = mm256_xor3( b3, a7, a1 ); \
+  b1 = a1; \
+  b6 = mm256_xor3( b6, a4, TEMP2 ); \
+  b4 = mm256_xor3( b4, a0, TEMP2 ); \
+  b7 = mm256_xor3( b7, a5, a3 ); \
+  b5 = mm256_xor3( b5, a1, a3 ); \
  /* compute x_i = t_i + t_{i+3} */\
  a0 = _mm256_xor_si256(a0, a3);\
  a1 = _mm256_xor_si256(a1, a4);\
-  a2 = _mm256_xor_si256(a2, a5);\
+  a2 = _mm256_xor_si256( TEMP2, a5);\
  a3 = _mm256_xor_si256(a3, a6);\
  a4 = _mm256_xor_si256(a4, a7);\
  a5 = _mm256_xor_si256(a5, b0);\
@@ -671,7 +657,6 @@ static const __m256i SUBSH_MASK7_2WAY =
  \
  /* MixBytes */\
  MixBytes_2way(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-\
 }

 /* 10 rounds, P and Q in parallel */
--- a/algo/groestl/groestl512-intr-4way.h
+++ b/algo/groestl/groestl512-intr-4way.h
@@ -710,7 +710,7 @@ static const __m256i SUBSH_MASK7_2WAY =
  b0 = a2;\
  a1 = _mm256_xor_si256(a1, a2);\
  b1 = a3;\
-  a2 = _mm256_xor_si256(a2, a3);\
+  TEMP2 = _mm256_xor_si256(a2, a3);\
  b2 = a4;\
  a3 = _mm256_xor_si256(a3, a4);\
  b3 = a5;\
@@ -722,34 +722,23 @@ static const __m256i SUBSH_MASK7_2WAY =
  a7 = _mm256_xor_si256(a7, b6);\
  \
  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  b0 = _mm256_xor_si256(b0, a4);\
-  b6 = _mm256_xor_si256(b6, a4);\
-  b1 = _mm256_xor_si256(b1, a5);\
-  b7 = _mm256_xor_si256(b7, a5);\
-  b2 = _mm256_xor_si256(b2, a6);\
-  b0 = _mm256_xor_si256(b0, a6);\
+  TEMP0 = mm256_xor3( b0, a4, a6 ); \
  /* spill values y_4, y_5 to memory */\
-  TEMP0 = b0;\
-  b3 = _mm256_xor_si256(b3, a7);\
-  b1 = _mm256_xor_si256(b1, a7);\
-  TEMP1 = b1;\
-  b4 = _mm256_xor_si256(b4, a0);\
-  b2 = _mm256_xor_si256(b2, a0);\
+  TEMP1 = mm256_xor3( b1, a5, a7 ); \
+  b2 = mm256_xor3( b2, a6, a0 ); \
  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  b0 = a0;\
-  b5 = _mm256_xor_si256(b5, a1);\
-  b3 = _mm256_xor_si256(b3, a1);\
-  b1 = a1;\
-  b6 = _mm256_xor_si256(b6, a2);\
-  b4 = _mm256_xor_si256(b4, a2);\
-  TEMP2 = a2;\
-  b7 = _mm256_xor_si256(b7, a3);\
-  b5 = _mm256_xor_si256(b5, a3);\
+  b0 = a0; \
+  b3 = mm256_xor3( b3, a7, a1 ); \
+  b1 = a1; \
+  b6 = mm256_xor3( b6, a4, TEMP2 ); \
+  b4 = mm256_xor3( b4, a0, TEMP2 ); \
+  b7 = mm256_xor3( b7, a5, a3 ); \
+  b5 = mm256_xor3( b5, a1, a3 ); \
  \
  /* compute x_i = t_i + t_{i+3} */\
  a0 = _mm256_xor_si256(a0, a3);\
  a1 = _mm256_xor_si256(a1, a4);\
-  a2 = _mm256_xor_si256(a2, a5);\
+  a2 = _mm256_xor_si256( TEMP2, a5);\
  a3 = _mm256_xor_si256(a3, a6);\
  a4 = _mm256_xor_si256(a4, a7);\
  a5 = _mm256_xor_si256(a5, b0);\