libavutil: optimize camellia cipher

Before the changes :
lavu       CAMELLIA     size: 1048576  runs:   1024  time:   32.541 +- 0.044

After the changes:
lavu       CAMELLIA     size: 1048576  runs:   1024  time:   24.589 +- 0.066

Tested with crypto_bench on a Linux x86_64 OS with Intel Core i5-3210M CPU.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

diff --git a/libavutil/camellia.c b/libavutil/camellia.c
index 7382414..11e57a1 100644 (file)
--- a/libavutil/camellia.c
+++ b/libavutil/camellia.c
@@ -37,6 +37,8 @@
 #define Sigma5  0x10E527FADE682D1D
 #define Sigma6  0xB05688C2B3E6C1FD
 
 #define Sigma5  0x10E527FADE682D1D
 #define Sigma6  0xB05688C2B3E6C1FD
 

+static uint64_t SP[8][256];
+

 typedef struct AVCAMELLIA {
     uint64_t Kw[4];
     uint64_t Ke[6];
 typedef struct AVCAMELLIA {
     uint64_t Kw[4];
     uint64_t Ke[6];


@@ -140,17 +142,11 @@ static void LR128(uint64_t d[2], const uint64_t K[2], int x)
 
 static uint64_t F(uint64_t F_IN, uint64_t KE)
 {
 
 static uint64_t F(uint64_t F_IN, uint64_t KE)
 {

-    uint32_t Zl, Zr;
+    uint8_t y[8];

     KE ^= F_IN;
     KE ^= F_IN;

-    Zl = KE >> 32;
-    Zr = KE & MASK32;
-    Zl = (((uint32_t)SBOX1[(Zl >> 24)] << 24) | ((uint32_t)SBOX2[(Zl >> 16) & MASK8] << 16) | ((uint32_t)SBOX3[(Zl >> 8) & MASK8] << 8) |(SBOX4[Zl & MASK8]));
-    Zr = (((uint32_t)SBOX2[(Zr >> 24)] << 24) | ((uint32_t)SBOX3[(Zr >> 16) & MASK8] << 16) | ((uint32_t)SBOX4[(Zr >> 8) & MASK8] << 8) |(SBOX1[Zr & MASK8]));
-    Zl ^= LR32(Zr, 8);
-    Zr ^= LR32(Zl, 16);
-    Zl ^= RR32(Zr, 8);
-    Zr ^= RR32(Zl, 8);
-    return ((uint64_t)Zr << 32) | (uint64_t)Zl;
+    AV_WB64(y,KE);
+    KE=SP[0][y[0]]^SP[1][y[1]]^SP[2][y[2]]^SP[3][y[3]]^SP[4][y[4]]^SP[5][y[5]]^SP[6][y[6]]^SP[7][y[7]];
+    return KE;

 }
 
 static uint64_t FL(uint64_t FL_IN, uint64_t KE)
 }
 
 static uint64_t FL(uint64_t FL_IN, uint64_t KE)


@@ -330,6 +326,26 @@ static void camellia_decrypt(AVCAMELLIA *cs, uint8_t *dst, const uint8_t *src, u
     AV_WB64(dst + 8, D1);
 }
 
     AV_WB64(dst + 8, D1);
 }
 

+static void computeSP(void)
+{
+    uint64_t z;
+    int i;
+    for (i = 0; i < 256; i++) {
+    z = (uint64_t)SBOX1[i];
+    SP[0][i] = (z << 56) ^ (z << 48) ^ (z << 40) ^ (z << 24) ^ z;
+    SP[7][i] = (z << 56) ^ (z << 48) ^ (z << 40) ^ (z << 24) ^ (z << 16) ^ (z << 8);
+    z = (uint64_t)SBOX2[i];
+    SP[1][i] = (z << 48) ^ (z << 40) ^ (z << 32) ^ (z << 24) ^ (z << 16);
+    SP[4][i] = (z << 48) ^ (z << 40) ^ (z << 32) ^ (z << 16) ^ (z << 8) ^ z;
+    z = (uint64_t)SBOX3[i];
+    SP[2][i] = (z << 56) ^ (z << 40) ^ (z << 32) ^ (z << 16) ^ (z << 8);
+    SP[5][i] = (z << 56) ^ (z << 40) ^ (z << 32) ^ (z << 24) ^ (z << 8) ^ z;
+    z = (uint64_t)SBOX4[i];
+    SP[3][i] = (z << 56) ^ (z << 48) ^ (z << 32) ^ (z << 8) ^ z;
+    SP[6][i] = (z << 56) ^ (z << 48) ^ (z << 32) ^ (z << 24) ^ (z << 16) ^ z;
+    }
+}
+

 struct AVCAMELLIA *av_camellia_alloc(void)
 {
     return av_mallocz(sizeof(struct AVCAMELLIA));
 struct AVCAMELLIA *av_camellia_alloc(void)
 {
     return av_mallocz(sizeof(struct AVCAMELLIA));


@@ -353,6 +369,7 @@ av_cold int av_camellia_init(AVCAMELLIA *cs, const uint8_t *key, int key_bits)
         Kr[0] = AV_RB64(key + 16);
         Kr[1] = AV_RB64(key + 24);
     }
         Kr[0] = AV_RB64(key + 16);
         Kr[1] = AV_RB64(key + 24);
     }

+    computeSP();

     D1 = Kl[0] ^ Kr[0];
     D2 = Kl[1] ^ Kr[1];
     D2 ^= F(D1, Sigma1);
     D1 = Kl[0] ^ Kr[0];
     D2 = Kl[1] ^ Kr[1];
     D2 ^= F(D1, Sigma1);