diff --git a/lib/sha256.c b/lib/sha256.c
index e262227b..bfd6a29c 100644
--- a/lib/sha256.c
+++ b/lib/sha256.c
@@ -32,6 +32,7 @@ sha256_init(sha256_context *ctx)
   ctx->nblocks = 0;
   ctx->nblocks_high = 0;
   ctx->count = 0;
+  ctx->blocksize = 64;
 }
 
 void
@@ -49,6 +50,7 @@ sha224_init(sha224_context *ctx)
   ctx->nblocks = 0;
   ctx->nblocks_high = 0;
   ctx->count = 0;
+  ctx->blocksize = 64;
 }
 
 /* (4.2) same as SHA-1's F1.  */
@@ -110,7 +112,7 @@ sum1(u32 x)
     32-bit-words. See FIPS 180-2 for details.
  */
 static unsigned int
-transform(sha256_context *ctx, const unsigned char *data)
+sha256_transform_block(sha256_context *ctx, const unsigned char *data)
 {
   static const u32 K[64] = {
       0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
@@ -217,7 +219,7 @@ sha256_transform(void *ctx, const unsigned char *data, size_t nblks)
 
   do
   {
-    burn = transform(hd, data);
+    burn = sha256_transform_block(hd, data);
     data += 64;
   }
   while (--nblks);
@@ -234,8 +236,7 @@ sha256_transform(void *ctx, const unsigned char *data, size_t nblks)
 void
 sha256_update(sha256_context *ctx, const byte *in_buf, size_t in_len)
 {
-  unsigned int stack_burn = 0;
-  const unsigned int blocksize = 64;
+  const unsigned int blocksize = ctx->blocksize;
   size_t inblocks;
 
   if (sizeof(ctx->buf) < blocksize)
@@ -243,8 +244,7 @@ sha256_update(sha256_context *ctx, const byte *in_buf, size_t in_len)
 
   if (ctx->count == blocksize)  /* Flush the buffer. */
   {
-    stack_burn = sha256_transform(ctx, ctx->buf, 1);
-    stack_burn = 0;
+    sha256_transform(ctx, ctx->buf, 1);
     ctx->count = 0;
     if (!++ctx->nblocks)
       ctx->nblocks_high++;
@@ -264,7 +264,7 @@ sha256_update(sha256_context *ctx, const byte *in_buf, size_t in_len)
   if (in_len >= blocksize)
   {
     inblocks = in_len / blocksize;
-    stack_burn = sha256_transform(ctx, in_buf, inblocks);
+    sha256_transform(ctx, in_buf, inblocks);
     ctx->count = 0;
     ctx->nblocks_high += (ctx->nblocks + inblocks < inblocks);
     ctx->nblocks += inblocks;
@@ -285,7 +285,6 @@ sha256_final(sha256_context *ctx)
 {
   u32 t, th, msb, lsb;
   byte *p;
-  unsigned int burn;
 
   sha256_update(ctx, NULL, 0); /* flush */;
 
@@ -325,7 +324,7 @@ sha256_final(sha256_context *ctx)
   /* append the 64 bit count */
   put_u32(ctx->buf + 56, msb);
   put_u32(ctx->buf + 60, lsb);
-  burn = sha256_transform(ctx, ctx->buf, 1);
+  sha256_transform(ctx, ctx->buf, 1);
 
   p = ctx->buf;
 
@@ -443,7 +442,7 @@ sha256_hmac_update(sha256_hmac_context *ctx, const void *buffer, size_t length)
    the length of the digest will be stored at that address.  The
    returned value is valid as long as the context exists.  On error
    NULL is returned. */
-const byte *
+byte *
 sha256_hmac_final(sha256_hmac_context *ctx)
 {
   sha256_final(&ctx->ctx);
diff --git a/lib/sha256.h b/lib/sha256.h
index a054ff90..1cc49a79 100644
--- a/lib/sha256.h
+++ b/lib/sha256.h
@@ -20,10 +20,11 @@
 
 typedef struct {
   u32  h0,h1,h2,h3,h4,h5,h6,h7;
-  byte buf[64];
+  byte buf[128];
   u32 nblocks;
   u32 nblocks_high;
   int count;
+  u32 blocksize;
 } sha256_context;
 typedef sha256_context sha224_context;
 
@@ -63,8 +64,8 @@ void sha224_hmac_update(sha224_hmac_context *ctx, const void *buf, size_t buflen
   sha256_hmac_update(ctx, buf, buflen);
 }
 
-const byte *sha256_hmac_final(sha256_hmac_context *ctx);
-const byte *sha224_hmac_final(sha224_hmac_context *ctx)
+byte *sha256_hmac_final(sha256_hmac_context *ctx);
+byte *sha224_hmac_final(sha224_hmac_context *ctx)
 {
   return sha256_hmac_final(ctx);
 }
diff --git a/lib/sha256_test.c b/lib/sha256_test.c
index f7b93f0a..3aeab677 100644
--- a/lib/sha256_test.c
+++ b/lib/sha256_test.c
@@ -165,10 +165,7 @@ get_sha256_hmac(const struct hmac_data_in in, char (*out_hash)[SHA256_HEX_SIZE])
   sha256_hmac_init(&ctx, in.key, in.key_len);
   sha256_hmac_update(&ctx, in.data, in.data_len);
   byte *hash_byte = sha256_hmac_final(&ctx);
-
-  int i;
-  for (i = 0; i < SHA256_SIZE; i++)
-    sprintf(*out_hash + i*2, "%02x", hash_byte[i]);
+  byte_to_hex((char*)out_hash, hash_byte, SHA256_SIZE);
 }
 
 static void
@@ -178,10 +175,7 @@ get_sha224_hmac(const struct hmac_data_in in, char (*out_hash)[SHA256_HEX_SIZE])
   sha224_hmac_init(&ctx, in.key, in.key_len);
   sha224_hmac_update(&ctx, in.data, in.data_len);
   byte *hash_byte = sha224_hmac_final(&ctx);
-
-  int i;
-  for (i = 0; i < SHA224_SIZE; i++)
-    sprintf(*out_hash + i*2, "%02x", hash_byte[i]);
+  byte_to_hex((char*)out_hash, hash_byte, SHA224_HEX_SIZE);
 }
 
 
diff --git a/lib/sha512.c b/lib/sha512.c
new file mode 100644
index 00000000..f610bba9
--- /dev/null
+++ b/lib/sha512.c
@@ -0,0 +1,493 @@
+/*
+ *	BIRD -- SHA512 and SHA384 Hash Functions
+ *
+ *	(c) 2015 CZ.NIC z.s.p.o.
+ *
+ *	Based on the code from libgcrypt-1.6.0, which is
+ *	(c) 2003, 2006, 2008, 2009 Free Software Foundation, Inc.
+ *
+ *	Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <netinet/in.h>
+
+#include "lib/sha256.h"
+#include "lib/sha256.c"	/* REMOVE ME */
+#include "lib/sha512.h"
+#include "lib/unaligned.h"
+
+#define U64_C(c) (c ## UL) /* Maybe is system dependent */
+
+void
+sha512_init(sha512_context *ctx)
+{
+  sha512_state *hd = &ctx->state;
+
+  hd->h0 = U64_C(0x6a09e667f3bcc908);
+  hd->h1 = U64_C(0xbb67ae8584caa73b);
+  hd->h2 = U64_C(0x3c6ef372fe94f82b);
+  hd->h3 = U64_C(0xa54ff53a5f1d36f1);
+  hd->h4 = U64_C(0x510e527fade682d1);
+  hd->h5 = U64_C(0x9b05688c2b3e6c1f);
+  hd->h6 = U64_C(0x1f83d9abfb41bd6b);
+  hd->h7 = U64_C(0x5be0cd19137e2179);
+
+  ctx->bctx.nblocks = 0;
+  ctx->bctx.nblocks_high = 0;
+  ctx->bctx.count = 0;
+  ctx->bctx.blocksize = 128;
+}
+
+void
+sha384_init(sha384_context *ctx)
+{
+  sha512_state *hd = &ctx->state;
+
+  hd->h0 = U64_C(0xcbbb9d5dc1059ed8);
+  hd->h1 = U64_C(0x629a292a367cd507);
+  hd->h2 = U64_C(0x9159015a3070dd17);
+  hd->h3 = U64_C(0x152fecd8f70e5939);
+  hd->h4 = U64_C(0x67332667ffc00b31);
+  hd->h5 = U64_C(0x8eb44a8768581511);
+  hd->h6 = U64_C(0xdb0c2e0d64f98fa7);
+  hd->h7 = U64_C(0x47b5481dbefa4fa4);
+
+  ctx->bctx.nblocks = 0;
+  ctx->bctx.nblocks_high = 0;
+  ctx->bctx.count = 0;
+  ctx->bctx.blocksize = 128;
+}
+
+void sha512_update(sha512_context *ctx, const byte *in_buf, size_t in_len)
+{
+  sha256_update(&ctx->bctx, in_buf, in_len);
+}
+
+static inline u64
+ROTR (u64 x, u64 n)
+{
+  return((x >> n) | (x << (64 - n)));
+}
+
+static inline u64
+Ch(u64 x, u64 y, u64 z)
+{
+  return((x & y) ^ ( ~x & z));
+}
+
+static inline u64
+Maj(u64 x, u64 y, u64 z)
+{
+  return((x & y) ^ (x & z) ^ (y & z));
+}
+
+static inline u64
+Sum0 (u64 x)
+{
+  return(ROTR (x, 28) ^ ROTR (x, 34) ^ ROTR (x, 39));
+}
+
+static inline u64
+Sum1 (u64 x)
+{
+  return(ROTR (x, 14) ^ ROTR (x, 18) ^ ROTR (x, 41));
+}
+
+static const u64 k[] =
+{
+    U64_C(0x428a2f98d728ae22), U64_C(0x7137449123ef65cd),
+    U64_C(0xb5c0fbcfec4d3b2f), U64_C(0xe9b5dba58189dbbc),
+    U64_C(0x3956c25bf348b538), U64_C(0x59f111f1b605d019),
+    U64_C(0x923f82a4af194f9b), U64_C(0xab1c5ed5da6d8118),
+    U64_C(0xd807aa98a3030242), U64_C(0x12835b0145706fbe),
+    U64_C(0x243185be4ee4b28c), U64_C(0x550c7dc3d5ffb4e2),
+    U64_C(0x72be5d74f27b896f), U64_C(0x80deb1fe3b1696b1),
+    U64_C(0x9bdc06a725c71235), U64_C(0xc19bf174cf692694),
+    U64_C(0xe49b69c19ef14ad2), U64_C(0xefbe4786384f25e3),
+    U64_C(0x0fc19dc68b8cd5b5), U64_C(0x240ca1cc77ac9c65),
+    U64_C(0x2de92c6f592b0275), U64_C(0x4a7484aa6ea6e483),
+    U64_C(0x5cb0a9dcbd41fbd4), U64_C(0x76f988da831153b5),
+    U64_C(0x983e5152ee66dfab), U64_C(0xa831c66d2db43210),
+    U64_C(0xb00327c898fb213f), U64_C(0xbf597fc7beef0ee4),
+    U64_C(0xc6e00bf33da88fc2), U64_C(0xd5a79147930aa725),
+    U64_C(0x06ca6351e003826f), U64_C(0x142929670a0e6e70),
+    U64_C(0x27b70a8546d22ffc), U64_C(0x2e1b21385c26c926),
+    U64_C(0x4d2c6dfc5ac42aed), U64_C(0x53380d139d95b3df),
+    U64_C(0x650a73548baf63de), U64_C(0x766a0abb3c77b2a8),
+    U64_C(0x81c2c92e47edaee6), U64_C(0x92722c851482353b),
+    U64_C(0xa2bfe8a14cf10364), U64_C(0xa81a664bbc423001),
+    U64_C(0xc24b8b70d0f89791), U64_C(0xc76c51a30654be30),
+    U64_C(0xd192e819d6ef5218), U64_C(0xd69906245565a910),
+    U64_C(0xf40e35855771202a), U64_C(0x106aa07032bbd1b8),
+    U64_C(0x19a4c116b8d2d0c8), U64_C(0x1e376c085141ab53),
+    U64_C(0x2748774cdf8eeb99), U64_C(0x34b0bcb5e19b48a8),
+    U64_C(0x391c0cb3c5c95a63), U64_C(0x4ed8aa4ae3418acb),
+    U64_C(0x5b9cca4f7763e373), U64_C(0x682e6ff3d6b2b8a3),
+    U64_C(0x748f82ee5defb2fc), U64_C(0x78a5636f43172f60),
+    U64_C(0x84c87814a1f0ab72), U64_C(0x8cc702081a6439ec),
+    U64_C(0x90befffa23631e28), U64_C(0xa4506cebde82bde9),
+    U64_C(0xbef9a3f7b2c67915), U64_C(0xc67178f2e372532b),
+    U64_C(0xca273eceea26619c), U64_C(0xd186b8c721c0c207),
+    U64_C(0xeada7dd6cde0eb1e), U64_C(0xf57d4f7fee6ed178),
+    U64_C(0x06f067aa72176fba), U64_C(0x0a637dc5a2c898a6),
+    U64_C(0x113f9804bef90dae), U64_C(0x1b710b35131c471b),
+    U64_C(0x28db77f523047d84), U64_C(0x32caab7b40c72493),
+    U64_C(0x3c9ebe0a15c9bebc), U64_C(0x431d67c49c100d4c),
+    U64_C(0x4cc5d4becb3e42b6), U64_C(0x597f299cfc657e2a),
+    U64_C(0x5fcb6fab3ad6faec), U64_C(0x6c44198c4a475817)
+};
+
+/*
+ * Transform the message W which consists of 16 64-bit-words
+ */
+static unsigned int
+sha512_transform_block(sha512_state *hd, const unsigned char *data)
+{
+  u64 a, b, c, d, e, f, g, h;
+  u64 w[16];
+  int t;
+
+  /* get values from the chaining vars */
+  a = hd->h0;
+  b = hd->h1;
+  c = hd->h2;
+  d = hd->h3;
+  e = hd->h4;
+  f = hd->h5;
+  g = hd->h6;
+  h = hd->h7;
+
+  for ( t = 0; t < 16; t++ )
+    w[t] = get_u64(data + t * 8);
+
+#define S0(x) (ROTR((x),1) ^ ROTR((x),8) ^ ((x)>>7))
+#define S1(x) (ROTR((x),19) ^ ROTR((x),61) ^ ((x)>>6))
+
+  for (t = 0; t < 80 - 16; )
+  {
+    u64 t1, t2;
+
+    /* Performance on a AMD Athlon(tm) Dual Core Processor 4050e
+         with gcc 4.3.3 using gcry_md_hash_buffer of each 10000 bytes
+         initialized to 0,1,2,3...255,0,... and 1000 iterations:
+
+         Not unrolled with macros:  440ms
+         Unrolled with macros:      350ms
+         Unrolled with inline:      330ms
+     */
+#if 0 /* Not unrolled.  */
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t] + w[t%16];
+    w[t%16] += S1 (w[(t - 2)%16]) + w[(t - 7)%16] + S0 (w[(t - 15)%16]);
+    t2 = Sum0 (a) + Maj(a, b, c);
+    h = g;
+    g = f;
+    f = e;
+    e = d + t1;
+    d = c;
+    c = b;
+    b = a;
+    a = t1 + t2;
+    t++;
+#else /* Unrolled to interweave the chain variables.  */
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t] + w[0];
+    w[0] += S1 (w[14]) + w[9] + S0 (w[1]);
+    t2 = Sum0 (a) + Maj(a, b, c);
+    d += t1;
+    h = t1 + t2;
+
+    t1 = g + Sum1 (d) + Ch(d, e, f) + k[t+1] + w[1];
+    w[1] += S1 (w[15]) + w[10] + S0 (w[2]);
+    t2 = Sum0 (h) + Maj(h, a, b);
+    c += t1;
+    g  = t1 + t2;
+
+    t1 = f + Sum1 (c) + Ch(c, d, e) + k[t+2] + w[2];
+    w[2] += S1 (w[0]) + w[11] + S0 (w[3]);
+    t2 = Sum0 (g) + Maj(g, h, a);
+    b += t1;
+    f  = t1 + t2;
+
+    t1 = e + Sum1 (b) + Ch(b, c, d) + k[t+3] + w[3];
+    w[3] += S1 (w[1]) + w[12] + S0 (w[4]);
+    t2 = Sum0 (f) + Maj(f, g, h);
+    a += t1;
+    e  = t1 + t2;
+
+    t1 = d + Sum1 (a) + Ch(a, b, c) + k[t+4] + w[4];
+    w[4] += S1 (w[2]) + w[13] + S0 (w[5]);
+    t2 = Sum0 (e) + Maj(e, f, g);
+    h += t1;
+    d  = t1 + t2;
+
+    t1 = c + Sum1 (h) + Ch(h, a, b) + k[t+5] + w[5];
+    w[5] += S1 (w[3]) + w[14] + S0 (w[6]);
+    t2 = Sum0 (d) + Maj(d, e, f);
+    g += t1;
+    c  = t1 + t2;
+
+    t1 = b + Sum1 (g) + Ch(g, h, a) + k[t+6] + w[6];
+    w[6] += S1 (w[4]) + w[15] + S0 (w[7]);
+    t2 = Sum0 (c) + Maj(c, d, e);
+    f += t1;
+    b  = t1 + t2;
+
+    t1 = a + Sum1 (f) + Ch(f, g, h) + k[t+7] + w[7];
+    w[7] += S1 (w[5]) + w[0] + S0 (w[8]);
+    t2 = Sum0 (b) + Maj(b, c, d);
+    e += t1;
+    a  = t1 + t2;
+
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t+8] + w[8];
+    w[8] += S1 (w[6]) + w[1] + S0 (w[9]);
+    t2 = Sum0 (a) + Maj(a, b, c);
+    d += t1;
+    h  = t1 + t2;
+
+    t1 = g + Sum1 (d) + Ch(d, e, f) + k[t+9] + w[9];
+    w[9] += S1 (w[7]) + w[2] + S0 (w[10]);
+    t2 = Sum0 (h) + Maj(h, a, b);
+    c += t1;
+    g  = t1 + t2;
+
+    t1 = f + Sum1 (c) + Ch(c, d, e) + k[t+10] + w[10];
+    w[10] += S1 (w[8]) + w[3] + S0 (w[11]);
+    t2 = Sum0 (g) + Maj(g, h, a);
+    b += t1;
+    f  = t1 + t2;
+
+    t1 = e + Sum1 (b) + Ch(b, c, d) + k[t+11] + w[11];
+    w[11] += S1 (w[9]) + w[4] + S0 (w[12]);
+    t2 = Sum0 (f) + Maj(f, g, h);
+    a += t1;
+    e  = t1 + t2;
+
+    t1 = d + Sum1 (a) + Ch(a, b, c) + k[t+12] + w[12];
+    w[12] += S1 (w[10]) + w[5] + S0 (w[13]);
+    t2 = Sum0 (e) + Maj(e, f, g);
+    h += t1;
+    d  = t1 + t2;
+
+    t1 = c + Sum1 (h) + Ch(h, a, b) + k[t+13] + w[13];
+    w[13] += S1 (w[11]) + w[6] + S0 (w[14]);
+    t2 = Sum0 (d) + Maj(d, e, f);
+    g += t1;
+    c  = t1 + t2;
+
+    t1 = b + Sum1 (g) + Ch(g, h, a) + k[t+14] + w[14];
+    w[14] += S1 (w[12]) + w[7] + S0 (w[15]);
+    t2 = Sum0 (c) + Maj(c, d, e);
+    f += t1;
+    b  = t1 + t2;
+
+    t1 = a + Sum1 (f) + Ch(f, g, h) + k[t+15] + w[15];
+    w[15] += S1 (w[13]) + w[8] + S0 (w[0]);
+    t2 = Sum0 (b) + Maj(b, c, d);
+    e += t1;
+    a  = t1 + t2;
+
+    t += 16;
+#endif
+  }
+
+  for (; t < 80; )
+  {
+    u64 t1, t2;
+
+#if 0 /* Not unrolled.  */
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t] + w[t%16];
+    t2 = Sum0 (a) + Maj(a, b, c);
+    h = g;
+    g = f;
+    f = e;
+    e = d + t1;
+    d = c;
+    c = b;
+    b = a;
+    a = t1 + t2;
+    t++;
+#else /* Unrolled to interweave the chain variables.  */
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t] + w[0];
+    t2 = Sum0 (a) + Maj(a, b, c);
+    d += t1;
+    h  = t1 + t2;
+
+    t1 = g + Sum1 (d) + Ch(d, e, f) + k[t+1] + w[1];
+    t2 = Sum0 (h) + Maj(h, a, b);
+    c += t1;
+    g  = t1 + t2;
+
+    t1 = f + Sum1 (c) + Ch(c, d, e) + k[t+2] + w[2];
+    t2 = Sum0 (g) + Maj(g, h, a);
+    b += t1;
+    f  = t1 + t2;
+
+    t1 = e + Sum1 (b) + Ch(b, c, d) + k[t+3] + w[3];
+    t2 = Sum0 (f) + Maj(f, g, h);
+    a += t1;
+    e  = t1 + t2;
+
+    t1 = d + Sum1 (a) + Ch(a, b, c) + k[t+4] + w[4];
+    t2 = Sum0 (e) + Maj(e, f, g);
+    h += t1;
+    d  = t1 + t2;
+
+    t1 = c + Sum1 (h) + Ch(h, a, b) + k[t+5] + w[5];
+    t2 = Sum0 (d) + Maj(d, e, f);
+    g += t1;
+    c  = t1 + t2;
+
+    t1 = b + Sum1 (g) + Ch(g, h, a) + k[t+6] + w[6];
+    t2 = Sum0 (c) + Maj(c, d, e);
+    f += t1;
+    b  = t1 + t2;
+
+    t1 = a + Sum1 (f) + Ch(f, g, h) + k[t+7] + w[7];
+    t2 = Sum0 (b) + Maj(b, c, d);
+    e += t1;
+    a  = t1 + t2;
+
+    t1 = h + Sum1 (e) + Ch(e, f, g) + k[t+8] + w[8];
+    t2 = Sum0 (a) + Maj(a, b, c);
+    d += t1;
+    h  = t1 + t2;
+
+    t1 = g + Sum1 (d) + Ch(d, e, f) + k[t+9] + w[9];
+    t2 = Sum0 (h) + Maj(h, a, b);
+    c += t1;
+    g  = t1 + t2;
+
+    t1 = f + Sum1 (c) + Ch(c, d, e) + k[t+10] + w[10];
+    t2 = Sum0 (g) + Maj(g, h, a);
+    b += t1;
+    f  = t1 + t2;
+
+    t1 = e + Sum1 (b) + Ch(b, c, d) + k[t+11] + w[11];
+    t2 = Sum0 (f) + Maj(f, g, h);
+    a += t1;
+    e  = t1 + t2;
+
+    t1 = d + Sum1 (a) + Ch(a, b, c) + k[t+12] + w[12];
+    t2 = Sum0 (e) + Maj(e, f, g);
+    h += t1;
+    d  = t1 + t2;
+
+    t1 = c + Sum1 (h) + Ch(h, a, b) + k[t+13] + w[13];
+    t2 = Sum0 (d) + Maj(d, e, f);
+    g += t1;
+    c  = t1 + t2;
+
+    t1 = b + Sum1 (g) + Ch(g, h, a) + k[t+14] + w[14];
+    t2 = Sum0 (c) + Maj(c, d, e);
+    f += t1;
+    b  = t1 + t2;
+
+    t1 = a + Sum1 (f) + Ch(f, g, h) + k[t+15] + w[15];
+    t2 = Sum0 (b) + Maj(b, c, d);
+    e += t1;
+    a  = t1 + t2;
+
+    t += 16;
+#endif
+  }
+
+  /* Update chaining vars.  */
+  hd->h0 += a;
+  hd->h1 += b;
+  hd->h2 += c;
+  hd->h3 += d;
+  hd->h4 += e;
+  hd->h5 += f;
+  hd->h6 += g;
+  hd->h7 += h;
+
+  return /* burn_stack */ (8 + 16) * sizeof(u64) + sizeof(u32) + 3 * sizeof(void*);
+}
+
+static unsigned int
+sha512_transform(void *context, const unsigned char *data, size_t nblks)
+{
+  sha512_context *ctx = context;
+  unsigned int burn;
+
+  do
+  {
+    burn = sha512_transform_block(&ctx->state, data) + 3 * sizeof(void*);
+    data += 128;
+  }
+  while(--nblks);
+
+  return burn;
+}
+
+/* The routine final terminates the computation and
+ * returns the digest.
+ * The handle is prepared for a new cycle, but adding bytes to the
+ * handle will the destroy the returned buffer.
+ * Returns: 64 bytes representing the digest.  When used for sha384,
+ * we take the leftmost 48 of those bytes.
+ */
+byte *
+sha512_final(sha512_context *ctx)
+{
+  u64 t, th, msb, lsb;
+  byte *p;
+
+  sha256_update(&ctx->bctx, NULL, 0); /* flush */ ;
+
+  t = ctx->bctx.nblocks;
+  /* if (sizeof t == sizeof ctx->bctx.nblocks) */
+  th = ctx->bctx.nblocks_high;
+  /* else */
+  /*   th = ctx->bctx.nblocks >> 64; In case we ever use u128  */
+
+  /* multiply by 128 to make a byte count */
+  lsb = t << 7;
+  msb = (th << 7) | (t >> 57);
+  /* add the count */
+  t = lsb;
+  if ((lsb += ctx->bctx.count) < t)
+    msb++;
+  /* multiply by 8 to make a bit count */
+  t = lsb;
+  lsb <<= 3;
+  msb <<= 3;
+  msb |= t >> 61;
+
+  if (ctx->bctx.count < 112)
+  {				/* enough room */
+    ctx->bctx.buf[ctx->bctx.count++] = 0x80;	/* pad */
+    while(ctx->bctx.count < 112)
+      ctx->bctx.buf[ctx->bctx.count++] = 0;	/* pad */
+  }
+  else
+  {				/* need one extra block */
+    ctx->bctx.buf[ctx->bctx.count++] = 0x80;	/* pad character */
+    while(ctx->bctx.count < 128)
+      ctx->bctx.buf[ctx->bctx.count++] = 0;
+    sha256_update(&ctx->bctx, NULL, 0); /* flush */ ;
+    memset(ctx->bctx.buf, 0, 112);	/* fill next block with zeroes */
+  }
+  /* append the 128 bit count */
+  put_u64(ctx->bctx.buf + 112, msb);
+  put_u64(ctx->bctx.buf + 120, lsb);
+  sha512_transform(ctx, ctx->bctx.buf, 1);
+
+  p = ctx->bctx.buf;
+#define X(a) do { put_u64(p, ctx->state.h##a); p += 8; } while(0)
+  X (0);
+  X (1);
+  X (2);
+  X (3);
+  X (4);
+  X (5);
+  /* Note that these last two chunks are included even for SHA384.
+     We just ignore them. */
+  X (6);
+  X (7);
+#undef X
+
+  return ctx->bctx.buf;
+}
diff --git a/lib/sha512.h b/lib/sha512.h
new file mode 100644
index 00000000..b5da77b6
--- /dev/null
+++ b/lib/sha512.h
@@ -0,0 +1,53 @@
+/*
+ *	BIRD -- SHA512 and SHA384 Hash Functions
+ *
+ *	(c) 2015 CZ.NIC z.s.p.o.
+ *
+ *	Based on the code from libgcrypt-1.6.0, which is
+ *	(c) 2003, 2006, 2008, 2009 Free Software Foundation, Inc.
+ *
+ *	Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+#ifndef _BIRD_SHA512_H_
+#define _BIRD_SHA512_H_
+
+#include "lib/sha256.h"
+
+#define SHA512_SIZE 	64
+#define SHA512_HEX_SIZE	129
+
+#define SHA384_SIZE 	48
+#define SHA384_HEX_SIZE	97
+
+typedef struct
+{
+  u64 h0, h1, h2, h3, h4, h5, h6, h7;
+} sha512_state;
+
+typedef struct
+{
+  sha256_context bctx;
+  sha512_state state;
+} sha512_context;
+typedef sha512_context sha384_context;
+
+
+void sha512_init(sha512_context *ctx);
+void sha384_init(sha384_context *ctx);
+
+void sha512_update(sha512_context *ctx, const byte *in_buf, size_t in_len);
+void sha384_update(sha384_context *ctx, const byte *in_buf, size_t in_len)
+{
+  sha512_update(ctx, in_buf, in_len);
+}
+
+byte* sha512_final(sha512_context *ctx);
+byte* sha384_final(sha384_context *ctx)
+{
+  return sha512_final(ctx);
+}
+
+static unsigned int sha512_transform(void *context, const unsigned char *data, size_t nblks);
+
+#endif /* _BIRD_SHA512_H_ */
diff --git a/lib/sha512_test.c b/lib/sha512_test.c
new file mode 100644
index 00000000..73fdd5e8
--- /dev/null
+++ b/lib/sha512_test.c
@@ -0,0 +1,123 @@
+/*
+ *	BIRD Library -- SHA512 and SHA384 Hash Functions Tests
+ *
+ *	(c) 2015 CZ.NIC z.s.p.o.
+ *
+ *	Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+#include <stdlib.h>
+
+#include "test/birdtest.h"
+#include "test/birdtest_support.h"
+#include "sysdep/config.h"
+#include "lib/sha512.h"
+#include "lib/sha512.c" /* REMOVE ME */
+
+static void
+byte_to_hex(char *out, const byte *in, uint len)
+{
+  int i;
+  for (i = 0; i < len; i++)
+    sprintf(out + i*2, "%02x", in[i]);
+}
+
+static void
+get_sha512(const char *str, char (*out_hash)[SHA512_HEX_SIZE])
+{
+  sha512_context ctx;
+  sha512_init(&ctx);
+  sha512_update(&ctx, str, strlen(str));
+  byte *hash = sha512_final(&ctx);
+  byte_to_hex((char*)out_hash, hash, SHA512_SIZE);
+}
+
+static void
+get_sha384(const char *str, char (*out_hash)[SHA384_HEX_SIZE])
+{
+  sha384_context ctx;
+  sha384_init(&ctx);
+  sha384_update(&ctx, str, strlen(str));
+  byte *hash = sha384_final(&ctx);
+  byte_to_hex((char*)out_hash, hash, SHA384_SIZE);
+}
+
+static int
+t_sha512(void)
+{
+  struct in_out {
+    char *in;
+    char out[SHA512_HEX_SIZE];
+  } in_out[] = {
+      {
+	  .in  = "",
+	  .out = "cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e",
+      },
+      {
+	  .in  = "a",
+	  .out = "1f40fc92da241694750979ee6cf582f2d5d7d28e18335de05abc54d0560e0f5302860c652bf08d560252aa5e74210546f369fbbbce8c12cfc7957b2652fe9a75",
+      },
+      {
+	  .in  = "abc",
+	  .out = "ddaf35a193617abacc417349ae20413112e6fa4e89a97ea20a9eeee64b55d39a2192992a274fc1a836ba3c23a3feebbd454d4423643ce80e2a9ac94fa54ca49f",
+      },
+      {
+	  .in  = "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
+	  .out = "204a8fc6dda82f0a0ced7beb8e08a41657c16ef468b228a8279be331a703c33596fd15c13b1b07f9aa1d3bea57789ca031ad85c7a71dd70354ec631238ca3445",
+      },
+      {
+	  .in  = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
+	  .out = "86497b815f64702e2ac6aca1f1d16f7159b4f0b34f6e92a41e632982a7291465957e0ef171042b9630bb66c6e35051613f99bdc95c371eeb46bff8c897eba6e9",
+      },
+  };
+
+  bt_assert_fn_in_out(get_sha512, in_out, "'%s'", "'%s'");
+
+  return BT_SUCCESS;
+}
+
+static int
+t_sha384(void)
+{
+
+  struct in_out {
+    char *in;
+    char out[SHA384_HEX_SIZE];
+  } in_out[] = {
+      {
+	  .in  = "",
+	  .out = "38b060a751ac96384cd9327eb1b1e36a21fdb71114be07434c0cc7bf63f6e1da274edebfe76f65fbd51ad2f14898b95b",
+      },
+      {
+	  .in  = "a",
+	  .out = "54a59b9f22b0b80880d8427e548b7c23abd873486e1f035dce9cd697e85175033caa88e6d57bc35efae0b5afd3145f31",
+      },
+      {
+	  .in  = "abc",
+	  .out = "cb00753f45a35e8bb5a03d699ac65007272c32ab0eded1631a8b605a43ff5bed8086072ba1e7cc2358baeca134c825a7",
+      },
+      {
+	  .in  = "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
+	  .out = "3391fdddfc8dc7393707a65b1b4709397cf8b1d162af05abfe8f450de5f36bc6b0455a8520bc4e6f5fe95b1fe3c8452b",
+      },
+      {
+	  .in  = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
+	  .out = "7b2cc66b0f621c1cb45c8dec93becf425d08f48d0e652154f8fffdde3ac7b1d2c6b19e9e507867301a3b604a8dafd3ba",
+      },
+  };
+
+  bt_assert_fn_in_out(get_sha384, in_out, "'%s'", "'%s'");
+
+  return BT_SUCCESS;
+}
+
+int
+main(int argc, char *argv[])
+{
+  bt_init(argc, argv);
+
+  bt_test_suite(t_sha512, "Testing SHA512");
+  bt_test_suite(t_sha384, "Testing SHA384");
+
+  return bt_end();
+}
diff --git a/lib/unaligned.h b/lib/unaligned.h
index 73fe2d8a..f686b9c8 100644
--- a/lib/unaligned.h
+++ b/lib/unaligned.h
@@ -36,6 +36,15 @@ get_u32(const void *p)
   return ntohl(x);
 }
 
+static inline u64
+get_u64(const void *p)
+{
+  u32 xh, xl;
+  memcpy(&xh, p, 4);
+  memcpy(&xl, p+4, 4);
+  return (((u64) ntohl(xh)) << 32) | ntohl(xl);
+}
+
 static inline void
 put_u16(void *p, u16 x)
 {
@@ -50,4 +59,14 @@ put_u32(void *p, u32 x)
   memcpy(p, &x, 4);
 }
 
+static inline void
+put_u64(void *p, u64 x)
+{
+  u32 xh, xl;
+  xh = htonl(x >> 32);
+  xl = htonl((u32) x);
+  memcpy(p, &xh, 4);
+  memcpy(p+4, &xl, 4);
+}
+
 #endif