rbnacl-libsodium 1.0.11 → 1.0.13

data/vendor/libsodium/src/libsodium/crypto_generichash/blake2/ref/blake2b-compress-sse41.c

@@ -1,80 +0,0 @@
-
-#define BLAKE2_USE_SSSE3
-#define BLAKE2_USE_SSE41
-
-#include <stdint.h>
-#include <string.h>
-
-#if (defined(HAVE_EMMINTRIN_H) && defined(HAVE_TMMINTRIN_H) && defined(HAVE_SMMINTRIN_H)) || \
-    (defined(_MSC_VER) && (defined(_M_X64) || defined(_M_AMD64) || defined(_M_IX86)))
-
-#pragma GCC target("sse2")
-#pragma GCC target("ssse3")
-#pragma GCC target("sse4.1")
-
-#include <emmintrin.h>
-#include <tmmintrin.h>
-#include <smmintrin.h>
-
-#include "blake2.h"
-#include "blake2-impl.h"
-#include "blake2b-compress-sse41.h"
-
-CRYPTO_ALIGN(64) static const uint64_t blake2b_IV[8] =
-{
-  0x6a09e667f3bcc908ULL, 0xbb67ae8584caa73bULL,
-  0x3c6ef372fe94f82bULL, 0xa54ff53a5f1d36f1ULL,
-  0x510e527fade682d1ULL, 0x9b05688c2b3e6c1fULL,
-  0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL
-};
-
-int blake2b_compress_sse41( blake2b_state *S, const uint8_t block[BLAKE2B_BLOCKBYTES] )
-{
-  __m128i row1l, row1h;
-  __m128i row2l, row2h;
-  __m128i row3l, row3h;
-  __m128i row4l, row4h;
-  __m128i b0, b1;
-  __m128i t0, t1;
-  const __m128i r16 = _mm_setr_epi8( 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9 );
-  const __m128i r24 = _mm_setr_epi8( 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10 );
-  const __m128i m0 = LOADU( block + 00 );
-  const __m128i m1 = LOADU( block + 16 );
-  const __m128i m2 = LOADU( block + 32 );
-  const __m128i m3 = LOADU( block + 48 );
-  const __m128i m4 = LOADU( block + 64 );
-  const __m128i m5 = LOADU( block + 80 );
-  const __m128i m6 = LOADU( block + 96 );
-  const __m128i m7 = LOADU( block + 112 );
-  row1l = LOADU( &S->h[0] );
-  row1h = LOADU( &S->h[2] );
-  row2l = LOADU( &S->h[4] );
-  row2h = LOADU( &S->h[6] );
-  row3l = LOADU( &blake2b_IV[0] );
-  row3h = LOADU( &blake2b_IV[2] );
-  row4l = _mm_xor_si128( LOADU( &blake2b_IV[4] ), LOADU( &S->t[0] ) );
-  row4h = _mm_xor_si128( LOADU( &blake2b_IV[6] ), LOADU( &S->f[0] ) );
-  ROUND( 0 );
-  ROUND( 1 );
-  ROUND( 2 );
-  ROUND( 3 );
-  ROUND( 4 );
-  ROUND( 5 );
-  ROUND( 6 );
-  ROUND( 7 );
-  ROUND( 8 );
-  ROUND( 9 );
-  ROUND( 10 );
-  ROUND( 11 );
-  row1l = _mm_xor_si128( row3l, row1l );
-  row1h = _mm_xor_si128( row3h, row1h );
-  STOREU( &S->h[0], _mm_xor_si128( LOADU( &S->h[0] ), row1l ) );
-  STOREU( &S->h[2], _mm_xor_si128( LOADU( &S->h[2] ), row1h ) );
-  row2l = _mm_xor_si128( row4l, row2l );
-  row2h = _mm_xor_si128( row4h, row2h );
-  STOREU( &S->h[4], _mm_xor_si128( LOADU( &S->h[4] ), row2l ) );
-  STOREU( &S->h[6], _mm_xor_si128( LOADU( &S->h[6] ), row2h ) );
-  return 0;
-}
-
-#endif

data/vendor/libsodium/src/libsodium/crypto_generichash/blake2/ref/blake2b-compress-sse41.h

@@ -1,97 +0,0 @@
-
-#ifndef blake2b_compress_sse41_H
-#define blake2b_compress_sse41_H
-
-#define LOADU(p)  _mm_loadu_si128( (const __m128i *)(const void *)(p) )
-#define STOREU(p,r) _mm_storeu_si128((__m128i *)(void *)(p), r)
-
-#define _mm_roti_epi64(x, c) \
-    (-(c) == 32) ? _mm_shuffle_epi32((x), _MM_SHUFFLE(2,3,0,1))  \
-    : (-(c) == 24) ? _mm_shuffle_epi8((x), r24) \
-    : (-(c) == 16) ? _mm_shuffle_epi8((x), r16) \
-    : (-(c) == 63) ? _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_add_epi64((x), (x)))  \
-    : _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_slli_epi64((x), 64-(-(c))))
-
-#define G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -32); \
-  row4h = _mm_roti_epi64(row4h, -32); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -24); \
-  row2h = _mm_roti_epi64(row2h, -24); \
-
-#define G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -16); \
-  row4h = _mm_roti_epi64(row4h, -16); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -63); \
-  row2h = _mm_roti_epi64(row2h, -63); \
-
-#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-  t0 = _mm_alignr_epi8(row2h, row2l, 8); \
-  t1 = _mm_alignr_epi8(row2l, row2h, 8); \
-  row2l = t0; \
-  row2h = t1; \
-  \
-  t0 = row3l; \
-  row3l = row3h; \
-  row3h = t0;    \
-  \
-  t0 = _mm_alignr_epi8(row4h, row4l, 8); \
-  t1 = _mm_alignr_epi8(row4l, row4h, 8); \
-  row4l = t1; \
-  row4h = t0;
-
-#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-  t0 = _mm_alignr_epi8(row2l, row2h, 8); \
-  t1 = _mm_alignr_epi8(row2h, row2l, 8); \
-  row2l = t0; \
-  row2h = t1; \
-  \
-  t0 = row3l; \
-  row3l = row3h; \
-  row3h = t0; \
-  \
-  t0 = _mm_alignr_epi8(row4l, row4h, 8); \
-  t1 = _mm_alignr_epi8(row4h, row4l, 8); \
-  row4l = t1; \
-  row4h = t0;
-
-#include "blake2b-load-sse41.h"
-
-#define ROUND(r) \
-  LOAD_MSG_ ##r ##_1(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_2(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h); \
-  LOAD_MSG_ ##r ##_3(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_4(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h);
-
-#endif

data/vendor/libsodium/src/libsodium/crypto_generichash/blake2/ref/blake2b-compress-ssse3.c

@@ -1,87 +0,0 @@
-
-#include <stdint.h>
-#include <string.h>
-
-#if (defined(HAVE_EMMINTRIN_H) && defined(HAVE_TMMINTRIN_H)) || \
-    (defined(_MSC_VER) && (defined(_M_X64) || defined(_M_AMD64)))
-
-#pragma GCC target("sse2")
-#pragma GCC target("ssse3")
-
-#ifdef _MSC_VER
-# include <intrin.h> /* for _mm_set_epi64x */
-#endif
-#include <emmintrin.h>
-#include <tmmintrin.h>
-
-#include "blake2.h"
-#include "blake2-impl.h"
-#include "blake2b-compress-ssse3.h"
-
-CRYPTO_ALIGN(64) static const uint64_t blake2b_IV[8] =
-{
-  0x6a09e667f3bcc908ULL, 0xbb67ae8584caa73bULL,
-  0x3c6ef372fe94f82bULL, 0xa54ff53a5f1d36f1ULL,
-  0x510e527fade682d1ULL, 0x9b05688c2b3e6c1fULL,
-  0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL
-};
-
-int blake2b_compress_ssse3( blake2b_state *S, const uint8_t block[BLAKE2B_BLOCKBYTES] )
-{
-  __m128i row1l, row1h;
-  __m128i row2l, row2h;
-  __m128i row3l, row3h;
-  __m128i row4l, row4h;
-  __m128i b0, b1;
-  __m128i t0, t1;
-  const __m128i r16 = _mm_setr_epi8( 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9 );
-  const __m128i r24 = _mm_setr_epi8( 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10 );
-  const uint64_t  m0 = ( ( uint64_t * )block )[ 0];
-  const uint64_t  m1 = ( ( uint64_t * )block )[ 1];
-  const uint64_t  m2 = ( ( uint64_t * )block )[ 2];
-  const uint64_t  m3 = ( ( uint64_t * )block )[ 3];
-  const uint64_t  m4 = ( ( uint64_t * )block )[ 4];
-  const uint64_t  m5 = ( ( uint64_t * )block )[ 5];
-  const uint64_t  m6 = ( ( uint64_t * )block )[ 6];
-  const uint64_t  m7 = ( ( uint64_t * )block )[ 7];
-  const uint64_t  m8 = ( ( uint64_t * )block )[ 8];
-  const uint64_t  m9 = ( ( uint64_t * )block )[ 9];
-  const uint64_t m10 = ( ( uint64_t * )block )[10];
-  const uint64_t m11 = ( ( uint64_t * )block )[11];
-  const uint64_t m12 = ( ( uint64_t * )block )[12];
-  const uint64_t m13 = ( ( uint64_t * )block )[13];
-  const uint64_t m14 = ( ( uint64_t * )block )[14];
-  const uint64_t m15 = ( ( uint64_t * )block )[15];
-
-  row1l = LOADU( &S->h[0] );
-  row1h = LOADU( &S->h[2] );
-  row2l = LOADU( &S->h[4] );
-  row2h = LOADU( &S->h[6] );
-  row3l = LOADU( &blake2b_IV[0] );
-  row3h = LOADU( &blake2b_IV[2] );
-  row4l = _mm_xor_si128( LOADU( &blake2b_IV[4] ), LOADU( &S->t[0] ) );
-  row4h = _mm_xor_si128( LOADU( &blake2b_IV[6] ), LOADU( &S->f[0] ) );
-  ROUND( 0 );
-  ROUND( 1 );
-  ROUND( 2 );
-  ROUND( 3 );
-  ROUND( 4 );
-  ROUND( 5 );
-  ROUND( 6 );
-  ROUND( 7 );
-  ROUND( 8 );
-  ROUND( 9 );
-  ROUND( 10 );
-  ROUND( 11 );
-  row1l = _mm_xor_si128( row3l, row1l );
-  row1h = _mm_xor_si128( row3h, row1h );
-  STOREU( &S->h[0], _mm_xor_si128( LOADU( &S->h[0] ), row1l ) );
-  STOREU( &S->h[2], _mm_xor_si128( LOADU( &S->h[2] ), row1h ) );
-  row2l = _mm_xor_si128( row4l, row2l );
-  row2h = _mm_xor_si128( row4h, row2h );
-  STOREU( &S->h[4], _mm_xor_si128( LOADU( &S->h[4] ), row2l ) );
-  STOREU( &S->h[6], _mm_xor_si128( LOADU( &S->h[6] ), row2h ) );
-  return 0;
-}
-
-#endif

data/vendor/libsodium/src/libsodium/crypto_generichash/blake2/ref/blake2b-compress-ssse3.h

@@ -1,97 +0,0 @@
-
-#ifndef blake2b_compress_ssse3_H
-#define blake2b_compress_ssse3_H
-
-#define LOADU(p)  _mm_loadu_si128( (const __m128i *)(const void *)(p) )
-#define STOREU(p,r) _mm_storeu_si128((__m128i *)(void *)(p), r)
-
-#define _mm_roti_epi64(x, c) \
-    (-(c) == 32) ? _mm_shuffle_epi32((x), _MM_SHUFFLE(2,3,0,1))  \
-    : (-(c) == 24) ? _mm_shuffle_epi8((x), r24) \
-    : (-(c) == 16) ? _mm_shuffle_epi8((x), r16) \
-    : (-(c) == 63) ? _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_add_epi64((x), (x)))  \
-    : _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_slli_epi64((x), 64-(-(c))))
-
-#define G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -32); \
-  row4h = _mm_roti_epi64(row4h, -32); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -24); \
-  row2h = _mm_roti_epi64(row2h, -24); \
-
-#define G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -16); \
-  row4h = _mm_roti_epi64(row4h, -16); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -63); \
-  row2h = _mm_roti_epi64(row2h, -63); \
-
-#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-  t0 = _mm_alignr_epi8(row2h, row2l, 8); \
-  t1 = _mm_alignr_epi8(row2l, row2h, 8); \
-  row2l = t0; \
-  row2h = t1; \
-  \
-  t0 = row3l; \
-  row3l = row3h; \
-  row3h = t0;    \
-  \
-  t0 = _mm_alignr_epi8(row4h, row4l, 8); \
-  t1 = _mm_alignr_epi8(row4l, row4h, 8); \
-  row4l = t1; \
-  row4h = t0;
-
-#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-  t0 = _mm_alignr_epi8(row2l, row2h, 8); \
-  t1 = _mm_alignr_epi8(row2h, row2l, 8); \
-  row2l = t0; \
-  row2h = t1; \
-  \
-  t0 = row3l; \
-  row3l = row3h; \
-  row3h = t0; \
-  \
-  t0 = _mm_alignr_epi8(row4l, row4h, 8); \
-  t1 = _mm_alignr_epi8(row4h, row4l, 8); \
-  row4l = t1; \
-  row4h = t0;
-
-#include "blake2b-load-sse2.h"
-
-#define ROUND(r) \
-  LOAD_MSG_ ##r ##_1(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_2(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h); \
-  LOAD_MSG_ ##r ##_3(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_4(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h);
-
-#endif

data/vendor/libsodium/src/libsodium/crypto_generichash/blake2/ref/blake2b-load-avx2.h

@@ -1,339 +0,0 @@
-#ifndef blake2b_load_avx2_H
-#define blake2b_load_avx2_H
-
-#define BLAKE2B_LOAD_MSG_0_1(b0) do { \
-    t0 = _mm256_unpacklo_epi64(m0, m1); \
-    t1 = _mm256_unpacklo_epi64(m2, m3); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_0_2(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m0, m1); \
-    t1 = _mm256_unpackhi_epi64(m2, m3); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_0_3(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m4, m5); \
-    t1 = _mm256_unpacklo_epi64(m6, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_0_4(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m4, m5); \
-    t1 = _mm256_unpackhi_epi64(m6, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_1_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m7, m2); \
-    t1 = _mm256_unpackhi_epi64(m4, m6); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_1_2(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m5, m4); \
-    t1 = _mm256_alignr_epi8(m3, m7, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_1_3(b0) \
-do { \
-    t0 = _mm256_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
-    t1 = _mm256_unpackhi_epi64(m5, m2); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_1_4(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m6, m1); \
-    t1 = _mm256_unpackhi_epi64(m3, m1); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_2_1(b0) \
-do { \
-    t0 = _mm256_alignr_epi8(m6, m5, 8); \
-    t1 = _mm256_unpackhi_epi64(m2, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_2_2(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m4, m0); \
-    t1 = _mm256_blend_epi32(m6, m1, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_2_3(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m1, m5, 0x33); \
-    t1 = _mm256_unpackhi_epi64(m3, m4); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_2_4(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m7, m3); \
-    t1 = _mm256_alignr_epi8(m2, m0, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_3_1(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m3, m1); \
-    t1 = _mm256_unpackhi_epi64(m6, m5); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_3_2(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m4, m0); \
-    t1 = _mm256_unpacklo_epi64(m6, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_3_3(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m2, m1, 0x33); \
-    t1 = _mm256_blend_epi32(m7, m2, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_3_4(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m3, m5); \
-    t1 = _mm256_unpacklo_epi64(m0, m4); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_4_1(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m4, m2); \
-    t1 = _mm256_unpacklo_epi64(m1, m5); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_4_2(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m3, m0, 0x33); \
-    t1 = _mm256_blend_epi32(m7, m2, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_4_3(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m5, m7, 0x33); \
-    t1 = _mm256_blend_epi32(m1, m3, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_4_4(b0) \
-do { \
-    t0 = _mm256_alignr_epi8(m6, m0, 8); \
-    t1 = _mm256_blend_epi32(m6, m4, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_5_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m1, m3); \
-    t1 = _mm256_unpacklo_epi64(m0, m4); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_5_2(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m6, m5); \
-    t1 = _mm256_unpackhi_epi64(m5, m1); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_5_3(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m3, m2, 0x33); \
-    t1 = _mm256_unpackhi_epi64(m7, m0); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_5_4(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m6, m2); \
-    t1 = _mm256_blend_epi32(m4, m7, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_6_1(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m0, m6, 0x33); \
-    t1 = _mm256_unpacklo_epi64(m7, m2); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_6_2(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m2, m7); \
-    t1 = _mm256_alignr_epi8(m5, m6, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_6_3(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m0, m3); \
-    t1 = _mm256_shuffle_epi32(m4, _MM_SHUFFLE(1,0,3,2)); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_6_4(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m3, m1); \
-    t1 = _mm256_blend_epi32(m5, m1, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_7_1(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m6, m3); \
-    t1 = _mm256_blend_epi32(m1, m6, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_7_2(b0) \
-do { \
-    t0 = _mm256_alignr_epi8(m7, m5, 8); \
-    t1 = _mm256_unpackhi_epi64(m0, m4); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_7_3(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m2, m7); \
-    t1 = _mm256_unpacklo_epi64(m4, m1); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_7_4(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m0, m2); \
-    t1 = _mm256_unpacklo_epi64(m3, m5); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_8_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m3, m7); \
-    t1 = _mm256_alignr_epi8(m0, m5, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_8_2(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m7, m4); \
-    t1 = _mm256_alignr_epi8(m4, m1, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_8_3(b0) \
-do { \
-    t0 = m6; \
-    t1 = _mm256_alignr_epi8(m5, m0, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_8_4(b0) \
-do { \
-    t0 = _mm256_blend_epi32(m3, m1, 0x33); \
-    t1 = m2; \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_9_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m5, m4); \
-    t1 = _mm256_unpackhi_epi64(m3, m0); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_9_2(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m1, m2); \
-    t1 = _mm256_blend_epi32(m2, m3, 0x33); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_9_3(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m7, m4); \
-    t1 = _mm256_unpackhi_epi64(m1, m6); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_9_4(b0) \
-do { \
-    t0 = _mm256_alignr_epi8(m7, m5, 8); \
-    t1 = _mm256_unpacklo_epi64(m6, m0); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_10_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m0, m1); \
-    t1 = _mm256_unpacklo_epi64(m2, m3); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_10_2(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m0, m1); \
-    t1 = _mm256_unpackhi_epi64(m2, m3); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_10_3(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m4, m5); \
-    t1 = _mm256_unpacklo_epi64(m6, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_10_4(b0) \
-do { \
-    t0 = _mm256_unpackhi_epi64(m4, m5); \
-    t1 = _mm256_unpackhi_epi64(m6, m7); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_11_1(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m7, m2); \
-    t1 = _mm256_unpackhi_epi64(m4, m6); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_11_2(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m5, m4); \
-    t1 = _mm256_alignr_epi8(m3, m7, 8); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_11_3(b0) \
-do { \
-    t0 = _mm256_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
-    t1 = _mm256_unpackhi_epi64(m5, m2); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#define BLAKE2B_LOAD_MSG_11_4(b0) \
-do { \
-    t0 = _mm256_unpacklo_epi64(m6, m1); \
-    t1 = _mm256_unpackhi_epi64(m3, m1); \
-    b0 = _mm256_blend_epi32(t0, t1, 0xF0); \
-} while(0)
-
-#endif