ed25519_blake2b 0.1.1

data/ext/ed25519_blake2b/blake2b-ref.c

@@ -0,0 +1,373 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+
+#include <stdint.h>
+#include <string.h>
+#include <stdio.h>
+
+#include "blake2.h"
+#include "blake2-impl.h"
+
+#include "blake2-config.h"
+
+#ifdef _MSC_VER
+#include <intrin.h> /* for _mm_set_epi64x */
+#endif
+#include <emmintrin.h>
+#if defined(HAVE_SSSE3)
+#include <tmmintrin.h>
+#endif
+#if defined(HAVE_SSE41)
+#include <smmintrin.h>
+#endif
+#if defined(HAVE_AVX)
+#include <immintrin.h>
+#endif
+#if defined(HAVE_XOP)
+#include <x86intrin.h>
+#endif
+
+#include "blake2b-round.h"
+
+static const uint64_t blake2b_IV[8] =
+{
+  0x6a09e667f3bcc908ULL, 0xbb67ae8584caa73bULL,
+  0x3c6ef372fe94f82bULL, 0xa54ff53a5f1d36f1ULL,
+  0x510e527fade682d1ULL, 0x9b05688c2b3e6c1fULL,
+  0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL
+};
+
+/* Some helper functions */
+static void blake2b_set_lastnode( blake2b_state *S )
+{
+  S->f[1] = (uint64_t)-1;
+}
+
+static int blake2b_is_lastblock( const blake2b_state *S )
+{
+  return S->f[0] != 0;
+}
+
+static void blake2b_set_lastblock( blake2b_state *S )
+{
+  if( S->last_node ) blake2b_set_lastnode( S );
+
+  S->f[0] = (uint64_t)-1;
+}
+
+static void blake2b_increment_counter( blake2b_state *S, const uint64_t inc )
+{
+  S->t[0] += inc;
+  S->t[1] += ( S->t[0] < inc );
+}
+
+/* init xors IV with input parameter block */
+int blake2b_init_param( blake2b_state *S, const blake2b_param *P )
+{
+  size_t i;
+  /*blake2b_init0( S ); */
+  const unsigned char * v = ( const unsigned char * )( blake2b_IV );
+  const unsigned char * p = ( const unsigned char * )( P );
+  unsigned char * h = ( unsigned char * )( S->h );
+  /* IV XOR ParamBlock */
+  memset( S, 0, sizeof( blake2b_state ) );
+
+  for( i = 0; i < BLAKE2B_OUTBYTES; ++i ) h[i] = v[i] ^ p[i];
+
+  S->outlen = P->digest_length;
+  return 0;
+}
+
+
+/* Some sort of default parameter block initialization, for sequential blake2b */
+int blake2b_init( blake2b_state *S, size_t outlen )
+{
+  blake2b_param P[1];
+
+  if ( ( !outlen ) || ( outlen > BLAKE2B_OUTBYTES ) ) return -1;
+
+  P->digest_length = (uint8_t)outlen;
+  P->key_length    = 0;
+  P->fanout        = 1;
+  P->depth         = 1;
+  store32( &P->leaf_length, 0 );
+  store32( &P->node_offset, 0 );
+  store32( &P->xof_length, 0 );
+  P->node_depth    = 0;
+  P->inner_length  = 0;
+  memset( P->reserved, 0, sizeof( P->reserved ) );
+  memset( P->salt,     0, sizeof( P->salt ) );
+  memset( P->personal, 0, sizeof( P->personal ) );
+
+  return blake2b_init_param( S, P );
+}
+
+int blake2b_init_key( blake2b_state *S, size_t outlen, const void *key, size_t keylen )
+{
+  blake2b_param P[1];
+
+  if ( ( !outlen ) || ( outlen > BLAKE2B_OUTBYTES ) ) return -1;
+
+  if ( ( !keylen ) || keylen > BLAKE2B_KEYBYTES ) return -1;
+
+  P->digest_length = (uint8_t)outlen;
+  P->key_length    = (uint8_t)keylen;
+  P->fanout        = 1;
+  P->depth         = 1;
+  store32( &P->leaf_length, 0 );
+  store32( &P->node_offset, 0 );
+  store32( &P->xof_length, 0 );
+  P->node_depth    = 0;
+  P->inner_length  = 0;
+  memset( P->reserved, 0, sizeof( P->reserved ) );
+  memset( P->salt,     0, sizeof( P->salt ) );
+  memset( P->personal, 0, sizeof( P->personal ) );
+
+  if( blake2b_init_param( S, P ) < 0 )
+    return 0;
+
+  {
+    uint8_t block[BLAKE2B_BLOCKBYTES];
+    memset( block, 0, BLAKE2B_BLOCKBYTES );
+    memcpy( block, key, keylen );
+    blake2b_update( S, block, BLAKE2B_BLOCKBYTES );
+    secure_zero_memory( block, BLAKE2B_BLOCKBYTES ); /* Burn the key from stack */
+  }
+  return 0;
+}
+
+static void blake2b_compress( 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;
+#if defined(HAVE_SSSE3) && !defined(HAVE_XOP)
+  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 );
+#endif
+#if defined(HAVE_SSE41)
+  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 );
+#else
+  const uint64_t  m0 = load64(block +  0 * sizeof(uint64_t));
+  const uint64_t  m1 = load64(block +  1 * sizeof(uint64_t));
+  const uint64_t  m2 = load64(block +  2 * sizeof(uint64_t));
+  const uint64_t  m3 = load64(block +  3 * sizeof(uint64_t));
+  const uint64_t  m4 = load64(block +  4 * sizeof(uint64_t));
+  const uint64_t  m5 = load64(block +  5 * sizeof(uint64_t));
+  const uint64_t  m6 = load64(block +  6 * sizeof(uint64_t));
+  const uint64_t  m7 = load64(block +  7 * sizeof(uint64_t));
+  const uint64_t  m8 = load64(block +  8 * sizeof(uint64_t));
+  const uint64_t  m9 = load64(block +  9 * sizeof(uint64_t));
+  const uint64_t m10 = load64(block + 10 * sizeof(uint64_t));
+  const uint64_t m11 = load64(block + 11 * sizeof(uint64_t));
+  const uint64_t m12 = load64(block + 12 * sizeof(uint64_t));
+  const uint64_t m13 = load64(block + 13 * sizeof(uint64_t));
+  const uint64_t m14 = load64(block + 14 * sizeof(uint64_t));
+  const uint64_t m15 = load64(block + 15 * sizeof(uint64_t));
+#endif
+  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 ) );
+}
+
+
+int blake2b_update( blake2b_state *S, const void *pin, size_t inlen )
+{
+  const unsigned char * in = (const unsigned char *)pin;
+  if( inlen > 0 )
+  {
+    size_t left = S->buflen;
+    size_t fill = BLAKE2B_BLOCKBYTES - left;
+    if( inlen > fill )
+    {
+      S->buflen = 0;
+      memcpy( S->buf + left, in, fill ); /* Fill buffer */
+      blake2b_increment_counter( S, BLAKE2B_BLOCKBYTES );
+      blake2b_compress( S, S->buf ); /* Compress */
+      in += fill; inlen -= fill;
+      while(inlen > BLAKE2B_BLOCKBYTES) {
+        blake2b_increment_counter(S, BLAKE2B_BLOCKBYTES);
+        blake2b_compress( S, in );
+        in += BLAKE2B_BLOCKBYTES;
+        inlen -= BLAKE2B_BLOCKBYTES;
+      }
+    }
+    memcpy( S->buf + S->buflen, in, inlen );
+    S->buflen += inlen;
+  }
+  return 0;
+}
+
+
+int blake2b_final( blake2b_state *S, void *out, size_t outlen )
+{
+  if( out == NULL || outlen < S->outlen )
+    return -1;
+
+  if( blake2b_is_lastblock( S ) )
+    return -1;
+
+  blake2b_increment_counter( S, S->buflen );
+  blake2b_set_lastblock( S );
+  memset( S->buf + S->buflen, 0, BLAKE2B_BLOCKBYTES - S->buflen ); /* Padding */
+  blake2b_compress( S, S->buf );
+
+  memcpy( out, &S->h[0], S->outlen );
+  return 0;
+}
+
+
+int blake2b( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen )
+{
+  blake2b_state S[1];
+
+  /* Verify parameters */
+  if ( NULL == in && inlen > 0 ) return -1;
+
+  if ( NULL == out ) return -1;
+
+  if( NULL == key && keylen > 0 ) return -1;
+
+  if( !outlen || outlen > BLAKE2B_OUTBYTES ) return -1;
+
+  if( keylen > BLAKE2B_KEYBYTES ) return -1;
+
+  if( keylen )
+  {
+    if( blake2b_init_key( S, outlen, key, keylen ) < 0 ) return -1;
+  }
+  else
+  {
+    if( blake2b_init( S, outlen ) < 0 ) return -1;
+  }
+
+  blake2b_update( S, ( const uint8_t * )in, inlen );
+  blake2b_final( S, out, outlen );
+  return 0;
+}
+
+int blake2( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ) {
+  return blake2b(out, outlen, in, inlen, key, keylen);
+}
+
+#if defined(SUPERCOP)
+int crypto_hash( unsigned char *out, unsigned char *in, unsigned long long inlen )
+{
+  return blake2b( out, BLAKE2B_OUTBYTES, in, inlen, NULL, 0 );
+}
+#endif
+
+#if defined(BLAKE2B_SELFTEST)
+#include <string.h>
+#include "blake2-kat.h"
+int main( void )
+{
+  uint8_t key[BLAKE2B_KEYBYTES];
+  uint8_t buf[BLAKE2_KAT_LENGTH];
+  size_t i, step;
+
+  for( i = 0; i < BLAKE2B_KEYBYTES; ++i )
+    key[i] = ( uint8_t )i;
+
+  for( i = 0; i < BLAKE2_KAT_LENGTH; ++i )
+    buf[i] = ( uint8_t )i;
+
+  /* Test simple API */
+  for( i = 0; i < BLAKE2_KAT_LENGTH; ++i )
+  {
+    uint8_t hash[BLAKE2B_OUTBYTES];
+    blake2b( hash, BLAKE2B_OUTBYTES, buf, i, key, BLAKE2B_KEYBYTES );
+
+    if( 0 != memcmp( hash, blake2b_keyed_kat[i], BLAKE2B_OUTBYTES ) )
+    {
+      goto fail;
+    }
+  }
+
+  /* Test streaming API */
+  for(step = 1; step < BLAKE2B_BLOCKBYTES; ++step) {
+    for (i = 0; i < BLAKE2_KAT_LENGTH; ++i) {
+      uint8_t hash[BLAKE2B_OUTBYTES];
+      blake2b_state S;
+      uint8_t * p = buf;
+      size_t mlen = i;
+      int err = 0;
+
+      if( (err = blake2b_init_key(&S, BLAKE2B_OUTBYTES, key, BLAKE2B_KEYBYTES)) < 0 ) {
+        goto fail;
+      }
+
+      while (mlen >= step) {
+        if ( (err = blake2b_update(&S, p, step)) < 0 ) {
+          goto fail;
+        }
+        mlen -= step;
+        p += step;
+      }
+      if ( (err = blake2b_update(&S, p, mlen)) < 0) {
+        goto fail;
+      }
+      if ( (err = blake2b_final(&S, hash, BLAKE2B_OUTBYTES)) < 0) {
+        goto fail;
+      }
+
+      if (0 != memcmp(hash, blake2b_keyed_kat[i], BLAKE2B_OUTBYTES)) {
+        goto fail;
+      }
+    }
+  }
+
+  puts( "ok" );
+  return 0;
+fail:
+  puts("error");
+  return -1;
+}
+#endif

data/ext/ed25519_blake2b/blake2b-round.h

@@ -0,0 +1,157 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2B_ROUND_H
+#define BLAKE2B_ROUND_H
+
+#define LOADU(p)  _mm_loadu_si128( (const __m128i *)(p) )
+#define STOREU(p,r) _mm_storeu_si128((__m128i *)(p), r)
+
+#define TOF(reg) _mm_castsi128_ps((reg))
+#define TOI(reg) _mm_castps_si128((reg))
+
+#define LIKELY(x) __builtin_expect((x),1)
+
+
+/* Microarchitecture-specific macros */
+#ifndef HAVE_XOP
+#ifdef HAVE_SSSE3
+#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))))
+#else
+#define _mm_roti_epi64(r, c) _mm_xor_si128(_mm_srli_epi64( (r), -(c) ),_mm_slli_epi64( (r), 64-(-(c)) ))
+#endif
+#else
+/* ... */
+#endif
+
+
+
+#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); \
+
+#if defined(HAVE_SSSE3)
+#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;
+#else
+
+#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = row4l;\
+  t1 = row2l;\
+  row4l = row3l;\
+  row3l = row3h;\
+  row3h = row4l;\
+  row4l = _mm_unpackhi_epi64(row4h, _mm_unpacklo_epi64(t0, t0)); \
+  row4h = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(row4h, row4h)); \
+  row2l = _mm_unpackhi_epi64(row2l, _mm_unpacklo_epi64(row2h, row2h)); \
+  row2h = _mm_unpackhi_epi64(row2h, _mm_unpacklo_epi64(t1, t1))
+
+#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = row3l;\
+  row3l = row3h;\
+  row3h = t0;\
+  t0 = row2l;\
+  t1 = row4l;\
+  row2l = _mm_unpackhi_epi64(row2h, _mm_unpacklo_epi64(row2l, row2l)); \
+  row2h = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(row2h, row2h)); \
+  row4l = _mm_unpackhi_epi64(row4l, _mm_unpacklo_epi64(row4h, row4h)); \
+  row4h = _mm_unpackhi_epi64(row4h, _mm_unpacklo_epi64(t1, t1))
+
+#endif
+
+#if defined(HAVE_SSE41)
+#include "blake2b-load-sse41.h"
+#else
+#include "blake2b-load-sse2.h"
+#endif
+
+#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