block-sha1: try to use rol/ror appropriately
Use the one with the smaller constant. It _can_ generate slightly smaller code (a constant of 1 is special), but perhaps more importantly it's possibly faster on any uarch that does a rotate with a loop. Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Linus Torvalds committed
Aug 5, 2009 at 19:42 UTC
b8e48a89b8f581eaf95b57782bb8e620ca30e968
1 file changed
+22
-10
block-sha1/sha1.c
+22
-10
@@ -80,7 +80,19 @@ void blk_SHA1_Final(unsigned char hashout[20], blk_SHA_CTX *ctx)
80
((unsigned int *)hashout)[i] = htonl(ctx->H[i]);
81
}
82
83
-#define SHA_ROT(X,n) (((X) << (n)) | ((X) >> (32-(n))))
83
+#if defined(__i386__) || defined(__x86_64__)
84
+
85
+#define SHA_ASM(op, x, n) ({ unsigned int __res; asm(op " %1,%0":"=r" (__res):"i" (n), "0" (x)); __res; })
86
+#define SHA_ROL(x,n) SHA_ASM("rol", x, n)
87
+#define SHA_ROR(x,n) SHA_ASM("ror", x, n)
88
+
89
+#else
90
+
91
+#define SHA_ROT(X,n) (((X) << (l)) | ((X) >> (r)))
92
+#define SHA_ROL(X,n) SHA_ROT(X,n,32-(n))
93
+#define SHA_ROR(X,n) SHA_ROT(X,32-(n),n)
94
+
95
+#endif
96
97
static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
98
{
@@ -93,7 +105,7 @@ static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
105
106
/* Unroll it? */
107
for (t = 16; t <= 79; t++)
96
- W[t] = SHA_ROT(W[t-3] ^ W[t-8] ^ W[t-14] ^ W[t-16], 1);
108
+ W[t] = SHA_ROL(W[t-3] ^ W[t-8] ^ W[t-14] ^ W[t-16], 1);
109
110
A = ctx->H[0];
111
B = ctx->H[1];
@@ -102,8 +114,8 @@ static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
114
E = ctx->H[4];
115
116
#define T_0_19(t) \
105
- TEMP = SHA_ROT(A,5) + (((C^D)&B)^D) + E + W[t] + 0x5a827999; \
106
- E = D; D = C; C = SHA_ROT(B, 30); B = A; A = TEMP;
117
+ TEMP = SHA_ROL(A,5) + (((C^D)&B)^D) + E + W[t] + 0x5a827999; \
118
+ E = D; D = C; C = SHA_ROR(B, 2); B = A; A = TEMP;
119
120
T_0_19( 0); T_0_19( 1); T_0_19( 2); T_0_19( 3); T_0_19( 4);
121
T_0_19( 5); T_0_19( 6); T_0_19( 7); T_0_19( 8); T_0_19( 9);
@@ -111,8 +123,8 @@ static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
123
T_0_19(15); T_0_19(16); T_0_19(17); T_0_19(18); T_0_19(19);
124
125
#define T_20_39(t) \
114
- TEMP = SHA_ROT(A,5) + (B^C^D) + E + W[t] + 0x6ed9eba1; \
115
- E = D; D = C; C = SHA_ROT(B, 30); B = A; A = TEMP;
126
+ TEMP = SHA_ROL(A,5) + (B^C^D) + E + W[t] + 0x6ed9eba1; \
127
+ E = D; D = C; C = SHA_ROR(B, 2); B = A; A = TEMP;
128
129
T_20_39(20); T_20_39(21); T_20_39(22); T_20_39(23); T_20_39(24);
130
T_20_39(25); T_20_39(26); T_20_39(27); T_20_39(28); T_20_39(29);
@@ -120,8 +132,8 @@ static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
132
T_20_39(35); T_20_39(36); T_20_39(37); T_20_39(38); T_20_39(39);
133
134
#define T_40_59(t) \
123
- TEMP = SHA_ROT(A,5) + ((B&C)|(D&(B|C))) + E + W[t] + 0x8f1bbcdc; \
124
- E = D; D = C; C = SHA_ROT(B, 30); B = A; A = TEMP;
135
+ TEMP = SHA_ROL(A,5) + ((B&C)|(D&(B|C))) + E + W[t] + 0x8f1bbcdc; \
136
+ E = D; D = C; C = SHA_ROR(B, 2); B = A; A = TEMP;
137
138
T_40_59(40); T_40_59(41); T_40_59(42); T_40_59(43); T_40_59(44);
139
T_40_59(45); T_40_59(46); T_40_59(47); T_40_59(48); T_40_59(49);
@@ -129,8 +141,8 @@ static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
141
T_40_59(55); T_40_59(56); T_40_59(57); T_40_59(58); T_40_59(59);
142
143
#define T_60_79(t) \
132
- TEMP = SHA_ROT(A,5) + (B^C^D) + E + W[t] + 0xca62c1d6; \
133
- E = D; D = C; C = SHA_ROT(B, 30); B = A; A = TEMP;
144
+ TEMP = SHA_ROL(A,5) + (B^C^D) + E + W[t] + 0xca62c1d6; \
145
+ E = D; D = C; C = SHA_ROR(B, 2); B = A; A = TEMP;
146
147
T_60_79(60); T_60_79(61); T_60_79(62); T_60_79(63); T_60_79(64);
148
T_60_79(65); T_60_79(66); T_60_79(67); T_60_79(68); T_60_79(69);