1 files changed, 34 insertions, 24 deletions
diff --git a/crypto/sha512_generic.c b/crypto/sha512_generic.c
index 8b9035b0189c..88f160b77b1f 100644
--- a/crypto/sha512_generic.c
+++ b/crypto/sha512_generic.c
@@ -78,7 +78,7 @@ static inline void LOAD_OP(int I, u64 *W, const u8 *input)
 static inline void BLEND_OP(int I, u64 *W)
 {
-        W[I] = s1(W[I-2]) + W[I-7] + s0(W[I-15]) + W[I-16];
+        W[I % 16] += s1(W[(I-2) % 16]) + W[(I-7) % 16] + s0(W[(I-15) % 16]);
 }
 static void
@@ -87,38 +87,48 @@ sha512_transform(u64 *state, const u8 *input)
        u64 a, b, c, d, e, f, g, h, t1, t2;
        int i;
-        u64 W[80];
+        u64 W[16];
        /* load the input */
        for (i = 0; i < 16; i++)
                LOAD_OP(i, W, input);
-        for (i = 16; i < 80; i++) {
-                BLEND_OP(i, W);
-        }
        /* load the state into our registers */
        a=state[0];   b=state[1];   c=state[2];   d=state[3];
        e=state[4];   f=state[5];   g=state[6];   h=state[7];
-        /* now iterate */
+#define SHA512_0_15(i, a, b, c, d, e, f, g, h)                  \
-        for (i=0; i<80; i+=8) {
+        t1 = h + e1(e) + Ch(e, f, g) + sha512_K[i] + W[i];      \
-                t1 = h + e1(e) + Ch(e,f,g) + sha512_K[i  ] + W[i  ];
+        t2 = e0(a) + Maj(a, b, c);                              \
-                t2 = e0(a) + Maj(a,b,c);    d+=t1;    h=t1+t2;
+        d += t1;                                                \
-                t1 = g + e1(d) + Ch(d,e,f) + sha512_K[i+1] + W[i+1];
+        h = t1 + t2
-                t2 = e0(h) + Maj(h,a,b);    c+=t1;    g=t1+t2;
-                t1 = f + e1(c) + Ch(c,d,e) + sha512_K[i+2] + W[i+2];
+#define SHA512_16_79(i, a, b, c, d, e, f, g, h)                 \
-                t2 = e0(g) + Maj(g,h,a);    b+=t1;    f=t1+t2;
+        BLEND_OP(i, W);                                         \
-                t1 = e + e1(b) + Ch(b,c,d) + sha512_K[i+3] + W[i+3];
+        t1 = h + e1(e) + Ch(e, f, g) + sha512_K[i] + W[(i)%16]; \
-                t2 = e0(f) + Maj(f,g,h);    a+=t1;    e=t1+t2;
+        t2 = e0(a) + Maj(a, b, c);                              \
-                t1 = d + e1(a) + Ch(a,b,c) + sha512_K[i+4] + W[i+4];
+        d += t1;                                                \
-                t2 = e0(e) + Maj(e,f,g);    h+=t1;    d=t1+t2;
+        h = t1 + t2
-                t1 = c + e1(h) + Ch(h,a,b) + sha512_K[i+5] + W[i+5];
-                t2 = e0(d) + Maj(d,e,f);    g+=t1;    c=t1+t2;
+        for (i = 0; i < 16; i += 8) {
-                t1 = b + e1(g) + Ch(g,h,a) + sha512_K[i+6] + W[i+6];
+                SHA512_0_15(i, a, b, c, d, e, f, g, h);
-                t2 = e0(c) + Maj(c,d,e);    f+=t1;    b=t1+t2;
+                SHA512_0_15(i + 1, h, a, b, c, d, e, f, g);
-                t1 = a + e1(f) + Ch(f,g,h) + sha512_K[i+7] + W[i+7];
+                SHA512_0_15(i + 2, g, h, a, b, c, d, e, f);
-                t2 = e0(b) + Maj(b,c,d);    e+=t1;    a=t1+t2;
+                SHA512_0_15(i + 3, f, g, h, a, b, c, d, e);
+                SHA512_0_15(i + 4, e, f, g, h, a, b, c, d);
+                SHA512_0_15(i + 5, d, e, f, g, h, a, b, c);
+                SHA512_0_15(i + 6, c, d, e, f, g, h, a, b);
+                SHA512_0_15(i + 7, b, c, d, e, f, g, h, a);
+        }
+        for (i = 16; i < 80; i += 8) {
+                SHA512_16_79(i, a, b, c, d, e, f, g, h);
+                SHA512_16_79(i + 1, h, a, b, c, d, e, f, g);
+                SHA512_16_79(i + 2, g, h, a, b, c, d, e, f);
+                SHA512_16_79(i + 3, f, g, h, a, b, c, d, e);
+                SHA512_16_79(i + 4, e, f, g, h, a, b, c, d);
+                SHA512_16_79(i + 5, d, e, f, g, h, a, b, c);
+                SHA512_16_79(i + 6, c, d, e, f, g, h, a, b);
+                SHA512_16_79(i + 7, b, c, d, e, f, g, h, a);
        }
        state[0] += a; state[1] += b; state[2] += c; state[3] += d;

diff --git a/crypto/sha512_generic.c b/crypto/sha512_generic.c index 8b9035b0189c..88f160b77b1f 100644 --- a/crypto/sha512_generic.c +++ b/crypto/sha512_generic.c
@@ -78,7 +78,7 @@ static inline void LOAD_OP(int I, u64 W, const u8 input)
78		78
79	static inline void BLEND_OP(int I, u64 *W)	79	static inline void BLEND_OP(int I, u64 *W)
80	{	80	{
81	W[I] = s1(W[I-2]) + W[I-7] + s0(W[I-15]) + W[I-16];	81	W[I % 16] += s1(W[(I-2) % 16]) + W[(I-7) % 16] + s0(W[(I-15) % 16]);
82	}	82	}
83		83
84	static void	84	static void
@@ -87,38 +87,48 @@ sha512_transform(u64 state, const u8 input)
87	u64 a, b, c, d, e, f, g, h, t1, t2;	87	u64 a, b, c, d, e, f, g, h, t1, t2;
88		88
89	int i;	89	int i;
90	u64 W[80];	90	u64 W[16];
91		91
92	/* load the input */	92	/* load the input */
93	for (i = 0; i < 16; i++)	93	for (i = 0; i < 16; i++)
94	LOAD_OP(i, W, input);	94	LOAD_OP(i, W, input);
95		95
96	for (i = 16; i < 80; i++) {
97	BLEND_OP(i, W);
98	}
99
100	/* load the state into our registers */	96	/* load the state into our registers */
101	a=state[0]; b=state[1]; c=state[2]; d=state[3];	97	a=state[0]; b=state[1]; c=state[2]; d=state[3];
102	e=state[4]; f=state[5]; g=state[6]; h=state[7];	98	e=state[4]; f=state[5]; g=state[6]; h=state[7];
103		99
104	/* now iterate */	100	#define SHA512_0_15(i, a, b, c, d, e, f, g, h) \
105	for (i=0; i<80; i+=8) {	101	t1 = h + e1(e) + Ch(e, f, g) + sha512_K[i] + W[i]; \
106	t1 = h + e1(e) + Ch(e,f,g) + sha512_K[i ] + W[i ];	102	t2 = e0(a) + Maj(a, b, c); \
107	t2 = e0(a) + Maj(a,b,c); d+=t1; h=t1+t2;	103	d += t1; \
108	t1 = g + e1(d) + Ch(d,e,f) + sha512_K[i+1] + W[i+1];	104	h = t1 + t2
109	t2 = e0(h) + Maj(h,a,b); c+=t1; g=t1+t2;	105
110	t1 = f + e1(c) + Ch(c,d,e) + sha512_K[i+2] + W[i+2];	106	#define SHA512_16_79(i, a, b, c, d, e, f, g, h) \
111	t2 = e0(g) + Maj(g,h,a); b+=t1; f=t1+t2;	107	BLEND_OP(i, W); \
112	t1 = e + e1(b) + Ch(b,c,d) + sha512_K[i+3] + W[i+3];	108	t1 = h + e1(e) + Ch(e, f, g) + sha512_K[i] + W[(i)%16]; \
113	t2 = e0(f) + Maj(f,g,h); a+=t1; e=t1+t2;	109	t2 = e0(a) + Maj(a, b, c); \
114	t1 = d + e1(a) + Ch(a,b,c) + sha512_K[i+4] + W[i+4];	110	d += t1; \
115	t2 = e0(e) + Maj(e,f,g); h+=t1; d=t1+t2;	111	h = t1 + t2
116	t1 = c + e1(h) + Ch(h,a,b) + sha512_K[i+5] + W[i+5];	112
117	t2 = e0(d) + Maj(d,e,f); g+=t1; c=t1+t2;	113	for (i = 0; i < 16; i += 8) {
118	t1 = b + e1(g) + Ch(g,h,a) + sha512_K[i+6] + W[i+6];	114	SHA512_0_15(i, a, b, c, d, e, f, g, h);
119	t2 = e0(c) + Maj(c,d,e); f+=t1; b=t1+t2;	115	SHA512_0_15(i + 1, h, a, b, c, d, e, f, g);
120	t1 = a + e1(f) + Ch(f,g,h) + sha512_K[i+7] + W[i+7];	116	SHA512_0_15(i + 2, g, h, a, b, c, d, e, f);
121	t2 = e0(b) + Maj(b,c,d); e+=t1; a=t1+t2;	117	SHA512_0_15(i + 3, f, g, h, a, b, c, d, e);
		118	SHA512_0_15(i + 4, e, f, g, h, a, b, c, d);
		119	SHA512_0_15(i + 5, d, e, f, g, h, a, b, c);
		120	SHA512_0_15(i + 6, c, d, e, f, g, h, a, b);
		121	SHA512_0_15(i + 7, b, c, d, e, f, g, h, a);
		122	}
		123	for (i = 16; i < 80; i += 8) {
		124	SHA512_16_79(i, a, b, c, d, e, f, g, h);
		125	SHA512_16_79(i + 1, h, a, b, c, d, e, f, g);
		126	SHA512_16_79(i + 2, g, h, a, b, c, d, e, f);
		127	SHA512_16_79(i + 3, f, g, h, a, b, c, d, e);
		128	SHA512_16_79(i + 4, e, f, g, h, a, b, c, d);
		129	SHA512_16_79(i + 5, d, e, f, g, h, a, b, c);
		130	SHA512_16_79(i + 6, c, d, e, f, g, h, a, b);
		131	SHA512_16_79(i + 7, b, c, d, e, f, g, h, a);
122	}	132	}
123		133
124	state[0] += a; state[1] += b; state[2] += c; state[3] += d;	134	state[0] += a; state[1] += b; state[2] += c; state[3] += d;