Write functions to handle loading and saving words a block at a time, taking into

account endian differences.

The current code does not take advantage of the knowledge of which endianness
we are running on; an optimization suggested by Yves Jerschow is to use (unsafe)
casts to speed up the load/store operations. This turns out to provide large
performance increases (30% or more) in some cases.

Even without the unsafe casts, this version seems to average a few percent
faster, probably because the longer loading loops have been partially or
fully unrolled.

This also makes the code implementing low-level algorithms like ciphers and
hashes a bit more succint.

1 files changed, 5 insertions, 13 deletions

diff --git a/src/gost.cpp b/src/gost.cpp
index a8a295c4b..d999d0d2d 100644
--- a/src/gost.cpp
+++ b/src/gost.cpp
@@ -13,8 +13,7 @@ namespace Botan {
 *************************************************/
 void GOST::enc(const byte in[], byte out[]) const
    {
-   u32bit N1 = make_u32bit(in[3], in[2], in[1], in[0]),
-          N2 = make_u32bit(in[7], in[6], in[5], in[4]);
+   u32bit N1 = load_le<u32bit>(in, 0), N2 = load_le<u32bit>(in, 1);
 
    for(u32bit j = 0; j != 32; j += 2)
       {
@@ -29,10 +28,7 @@ void GOST::enc(const byte in[], byte out[]) const
             SBOX3[get_byte(2, T0)] | SBOX4[get_byte(3, T0)];
       }
 
-   out[0] = get_byte(3, N2); out[1] = get_byte(2, N2);
-   out[2] = get_byte(1, N2); out[3] = get_byte(0, N2);
-   out[4] = get_byte(3, N1); out[5] = get_byte(2, N1);
-   out[6] = get_byte(1, N1); out[7] = get_byte(0, N1);
+   store_le(out, N2, N1);
    }
 
 /*************************************************
@@ -40,8 +36,7 @@ void GOST::enc(const byte in[], byte out[]) const
 *************************************************/
 void GOST::dec(const byte in[], byte out[]) const
    {
-   u32bit N1 = make_u32bit(in[3], in[2], in[1], in[0]),
-          N2 = make_u32bit(in[7], in[6], in[5], in[4]);
+   u32bit N1 = load_le<u32bit>(in, 0), N2 = load_le<u32bit>(in, 1);
 
    for(u32bit j = 0; j != 32; j += 2)
       {
@@ -56,10 +51,7 @@ void GOST::dec(const byte in[], byte out[]) const
             SBOX3[get_byte(2, T0)] | SBOX4[get_byte(3, T0)];
       }
 
-   out[0] = get_byte(3, N2); out[1] = get_byte(2, N2);
-   out[2] = get_byte(1, N2); out[3] = get_byte(0, N2);
-   out[4] = get_byte(3, N1); out[5] = get_byte(2, N1);
-   out[6] = get_byte(1, N1); out[7] = get_byte(0, N1);
+   store_le(out, N2, N1);
    }
 
 /*************************************************
@@ -69,7 +61,7 @@ void GOST::key(const byte key[], u32bit)
    {
    for(u32bit j = 0; j != 8; ++j)
       {
-      u32bit K = make_u32bit(key[4*j+3], key[4*j+2], key[4*j+1], key[4*j]);
+      u32bit K = load_le<u32bit>(key, j);
       EK[j] = EK[j+8] = EK[j+16] = K;
       }