/* -*-c-*-
*
- * $Id: skipjack.c,v 1.2 2000/07/15 20:48:45 mdw Exp $
+ * $Id: skipjack.c,v 1.3 2000/08/01 00:28:34 mdw Exp $
*
* The Skipjack block cipher
*
/*----- Revision history --------------------------------------------------*
*
* $Log: skipjack.c,v $
+ * Revision 1.3 2000/08/01 00:28:34 mdw
+ * Performance improvement: read keys in as 32-bit words and deal them out
+ * byte-by-byte.
+ *
* Revision 1.2 2000/07/15 20:48:45 mdw
* Remove some useless tests in the G function.
*
void skipjack_init(skipjack_ctx *k, const void *buf, size_t sz)
{
+ const octet *b = buf;
+ uint32 crud;
KSZ_ASSERT(skipjack, sz);
- memcpy(k->k, buf, sz);
+ k->ka = LOAD32(b);
+ k->kb = LOAD32(b + 4);
+ crud = LOAD16(b + 8);
+ k->kc = U32((crud << 16) | (k->ka >> 16));
+ k->kd = U32((k->ka << 16) | (k->kb >> 16));
+ k->ke = U32((k->kb << 16) | crud);
+ crud = 0;
}
/* --- @skipjack_eblk@, @skipjack_dblk@ --- *
* Use: Low-level block encryption and decryption.
*/
-#define G(x, i) do { \
+#define G(x, k) do { \
octet _x = U8(x >> 8), _y = U8(x); \
- _x ^= f[_y ^ k->k[i++]]; \
- _y ^= f[_x ^ k->k[i++]]; if (i >= 10) i = 0; \
- _x ^= f[_y ^ k->k[i++]]; \
- _y ^= f[_x ^ k->k[i++]]; if (i >= 10) i = 0; \
- x = U16((_x << 8) | _y); \
+ _x ^= f[_y ^ U8(k >> 24)]; \
+ _y ^= f[_x ^ U8(k >> 16)]; \
+ _x ^= f[_y ^ U8(k >> 8)]; \
+ _y ^= f[_x ^ U8(k >> 0)]; \
+ x = (_x << 8) | _y; \
} while (0)
-#define RULE_A(w, x, y, z, n, i) do { \
- G(w, i); z ^= w ^ n++; \
+#define RULE_A(w, x, y, z, n, k) do { \
+ G(w, k); z ^= w ^ n++; \
} while (0)
-#define RULE_B(w, x, y, z, n, i) do { \
- x ^= w ^ n++; G(w, i); \
+#define RULE_B(w, x, y, z, n, k) do { \
+ x ^= w ^ n++; G(w, k); \
} while (0)
void skipjack_eblk(const skipjack_ctx *k, const uint32 *s, uint32 *d)
{
- unsigned i = 0;
unsigned n = 1;
uint16 w = U16(s[0] >> 16), x = U16(s[0]);
uint16 y = U16(s[1] >> 16), z = U16(s[1]);
-
- RULE_A(w, x, y, z, n, i); RULE_A(z, w, x, y, n, i);
- RULE_A(y, z, w, x, n, i); RULE_A(x, y, z, w, n, i);
- RULE_A(w, x, y, z, n, i); RULE_A(z, w, x, y, n, i);
- RULE_A(y, z, w, x, n, i); RULE_A(x, y, z, w, n, i);
- RULE_B(w, x, y, z, n, i); RULE_B(z, w, x, y, n, i);
- RULE_B(y, z, w, x, n, i); RULE_B(x, y, z, w, n, i);
- RULE_B(w, x, y, z, n, i); RULE_B(z, w, x, y, n, i);
- RULE_B(y, z, w, x, n, i); RULE_B(x, y, z, w, n, i);
- RULE_A(w, x, y, z, n, i); RULE_A(z, w, x, y, n, i);
- RULE_A(y, z, w, x, n, i); RULE_A(x, y, z, w, n, i);
- RULE_A(w, x, y, z, n, i); RULE_A(z, w, x, y, n, i);
- RULE_A(y, z, w, x, n, i); RULE_A(x, y, z, w, n, i);
- RULE_B(w, x, y, z, n, i); RULE_B(z, w, x, y, n, i);
- RULE_B(y, z, w, x, n, i); RULE_B(x, y, z, w, n, i);
- RULE_B(w, x, y, z, n, i); RULE_B(z, w, x, y, n, i);
- RULE_B(y, z, w, x, n, i); RULE_B(x, y, z, w, n, i);
+ uint32 ka = k->ka, kb = k->kb, kc = k->kc, kd = k->kd, ke = k->ke;
+
+ RULE_A(w, x, y, z, n, ka); RULE_A(z, w, x, y, n, kb);
+ RULE_A(y, z, w, x, n, kc); RULE_A(x, y, z, w, n, kd);
+ RULE_A(w, x, y, z, n, ke); RULE_A(z, w, x, y, n, ka);
+ RULE_A(y, z, w, x, n, kb); RULE_A(x, y, z, w, n, kc);
+ RULE_B(w, x, y, z, n, kd); RULE_B(z, w, x, y, n, ke);
+ RULE_B(y, z, w, x, n, ka); RULE_B(x, y, z, w, n, kb);
+ RULE_B(w, x, y, z, n, kc); RULE_B(z, w, x, y, n, kd);
+ RULE_B(y, z, w, x, n, ke); RULE_B(x, y, z, w, n, ka);
+ RULE_A(w, x, y, z, n, kb); RULE_A(z, w, x, y, n, kc);
+ RULE_A(y, z, w, x, n, kd); RULE_A(x, y, z, w, n, ke);
+ RULE_A(w, x, y, z, n, ka); RULE_A(z, w, x, y, n, kb);
+ RULE_A(y, z, w, x, n, kc); RULE_A(x, y, z, w, n, kd);
+ RULE_B(w, x, y, z, n, ke); RULE_B(z, w, x, y, n, ka);
+ RULE_B(y, z, w, x, n, kb); RULE_B(x, y, z, w, n, kc);
+ RULE_B(w, x, y, z, n, kd); RULE_B(z, w, x, y, n, ke);
+ RULE_B(y, z, w, x, n, ka); RULE_B(x, y, z, w, n, kb);
d[0] = ((uint32)w << 16) | (uint32)x;
d[1] = ((uint32)y << 16) | (uint32)z;
}
-#define G_INV(x, i) do { \
+#define G_INV(x, k) do { \
octet _x = U8(x >> 8), _y = U8(x); \
- _y ^= f[_x ^ k->k[--i]]; \
- _x ^= f[_y ^ k->k[--i]]; if (i == 0) i = 10; \
- _y ^= f[_x ^ k->k[--i]]; \
- _x ^= f[_y ^ k->k[--i]]; if (i == 0) i = 10; \
- x = U16((_x << 8) | _y); \
+ _y ^= f[_x ^ U8(k >> 0)]; \
+ _x ^= f[_y ^ U8(k >> 8)]; \
+ _y ^= f[_x ^ U8(k >> 16)]; \
+ _x ^= f[_y ^ U8(k >> 24)]; \
+ x = (_x << 8) | _y; \
} while (0)
#define RULE_A_INV(w, x, y, z, n, i) do { \
void skipjack_dblk(const skipjack_ctx *k, const uint32 *s, uint32 *d)
{
- unsigned i = 8;
unsigned n = 33;
uint16 w = U16(s[0] >> 16), x = U16(s[0]);
uint16 y = U16(s[1] >> 16), z = U16(s[1]);
-
- RULE_B_INV(w, x, y, z, n, i); RULE_B_INV(x, y, z, w, n, i);
- RULE_B_INV(y, z, w, x, n, i); RULE_B_INV(z, w, x, y, n, i);
- RULE_B_INV(w, x, y, z, n, i); RULE_B_INV(x, y, z, w, n, i);
- RULE_B_INV(y, z, w, x, n, i); RULE_B_INV(z, w, x, y, n, i);
- RULE_A_INV(w, x, y, z, n, i); RULE_A_INV(x, y, z, w, n, i);
- RULE_A_INV(y, z, w, x, n, i); RULE_A_INV(z, w, x, y, n, i);
- RULE_A_INV(w, x, y, z, n, i); RULE_A_INV(x, y, z, w, n, i);
- RULE_A_INV(y, z, w, x, n, i); RULE_A_INV(z, w, x, y, n, i);
- RULE_B_INV(w, x, y, z, n, i); RULE_B_INV(x, y, z, w, n, i);
- RULE_B_INV(y, z, w, x, n, i); RULE_B_INV(z, w, x, y, n, i);
- RULE_B_INV(w, x, y, z, n, i); RULE_B_INV(x, y, z, w, n, i);
- RULE_B_INV(y, z, w, x, n, i); RULE_B_INV(z, w, x, y, n, i);
- RULE_A_INV(w, x, y, z, n, i); RULE_A_INV(x, y, z, w, n, i);
- RULE_A_INV(y, z, w, x, n, i); RULE_A_INV(z, w, x, y, n, i);
- RULE_A_INV(w, x, y, z, n, i); RULE_A_INV(x, y, z, w, n, i);
- RULE_A_INV(y, z, w, x, n, i); RULE_A_INV(z, w, x, y, n, i);
+ uint32 ka = k->ka, kb = k->kb, kc = k->kc, kd = k->kd, ke = k->ke;
+
+ RULE_B_INV(w, x, y, z, n, kb); RULE_B_INV(x, y, z, w, n, ka);
+ RULE_B_INV(y, z, w, x, n, ke); RULE_B_INV(z, w, x, y, n, kd);
+ RULE_B_INV(w, x, y, z, n, kc); RULE_B_INV(x, y, z, w, n, kb);
+ RULE_B_INV(y, z, w, x, n, ka); RULE_B_INV(z, w, x, y, n, ke);
+ RULE_A_INV(w, x, y, z, n, kd); RULE_A_INV(x, y, z, w, n, kc);
+ RULE_A_INV(y, z, w, x, n, kb); RULE_A_INV(z, w, x, y, n, ka);
+ RULE_A_INV(w, x, y, z, n, ke); RULE_A_INV(x, y, z, w, n, kd);
+ RULE_A_INV(y, z, w, x, n, kc); RULE_A_INV(z, w, x, y, n, kb);
+ RULE_B_INV(w, x, y, z, n, ka); RULE_B_INV(x, y, z, w, n, ke);
+ RULE_B_INV(y, z, w, x, n, kd); RULE_B_INV(z, w, x, y, n, kc);
+ RULE_B_INV(w, x, y, z, n, kb); RULE_B_INV(x, y, z, w, n, ka);
+ RULE_B_INV(y, z, w, x, n, ke); RULE_B_INV(z, w, x, y, n, kd);
+ RULE_A_INV(w, x, y, z, n, kc); RULE_A_INV(x, y, z, w, n, kb);
+ RULE_A_INV(y, z, w, x, n, ka); RULE_A_INV(z, w, x, y, n, ke);
+ RULE_A_INV(w, x, y, z, n, kd); RULE_A_INV(x, y, z, w, n, kc);
+ RULE_A_INV(y, z, w, x, n, kb); RULE_A_INV(z, w, x, y, n, ka);
d[0] = ((uint32)w << 16) | (uint32)x;
d[1] = ((uint32)y << 16) | (uint32)z;