summaryrefslogtreecommitdiff
path: root/libavutil/ripemd.c
diff options
context:
space:
mode:
Diffstat (limited to 'libavutil/ripemd.c')
-rw-r--r--libavutil/ripemd.c551
1 files changed, 551 insertions, 0 deletions
diff --git a/libavutil/ripemd.c b/libavutil/ripemd.c
new file mode 100644
index 0000000000..b0db2970db
--- /dev/null
+++ b/libavutil/ripemd.c
@@ -0,0 +1,551 @@
+/*
+ * Copyright (C) 2007 Michael Niedermayer <michaelni@gmx.at>
+ * Copyright (C) 2013 James Almer
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <string.h>
+
+#include "attributes.h"
+#include "avutil.h"
+#include "bswap.h"
+#include "intreadwrite.h"
+#include "ripemd.h"
+#include "mem.h"
+
+/** hash context */
+typedef struct AVRIPEMD {
+ uint8_t digest_len; ///< digest length in 32-bit words
+ uint64_t count; ///< number of bytes in buffer
+ uint8_t buffer[64]; ///< 512-bit buffer of input values used in hash updating
+ uint32_t state[10]; ///< current hash value
+ /** function used to update hash for 512-bit input block */
+ void (*transform)(uint32_t *state, const uint8_t buffer[64]);
+} AVRIPEMD;
+
+const int av_ripemd_size = sizeof(AVRIPEMD);
+
+struct AVRIPEMD *av_ripemd_alloc(void)
+{
+ return av_mallocz(sizeof(struct AVRIPEMD));
+}
+
+static const uint32_t KA[4] = {
+ 0x5a827999, 0x6ed9eba1, 0x8f1bbcdc, 0xa953fd4e
+};
+
+static const uint32_t KB[4] = {
+ 0x50a28be6, 0x5c4dd124, 0x6d703ef3, 0x7a6d76e9
+};
+
+static const int ROTA[80] = {
+ 11, 14, 15, 12, 5, 8, 7 , 9, 11, 13, 14, 15, 6, 7, 9, 8,
+ 7 , 6, 8, 13, 11, 9, 7, 15, 7, 12, 15, 9, 11, 7, 13, 12,
+ 11, 13, 6, 7, 14, 9, 13, 15, 14, 8, 13, 6, 5, 12, 7, 5,
+ 11, 12, 14, 15, 14, 15, 9, 8, 9, 14, 5, 6, 8, 6, 5, 12,
+ 9, 15, 5, 11, 6, 8, 13, 12, 5, 12, 13, 14, 11, 8, 5, 6
+};
+
+static const int ROTB[80] = {
+ 8, 9, 9, 11, 13, 15, 15, 5, 7, 7, 8, 11, 14, 14, 12, 6,
+ 9, 13, 15, 7, 12, 8, 9, 11, 7, 7, 12, 7, 6, 15, 13, 11,
+ 9, 7, 15, 11, 8, 6, 6, 14, 12, 13, 5, 14, 13, 13, 7, 5,
+ 15, 5, 8, 11, 14, 14, 6, 14, 6, 9, 12, 9, 12, 5, 15, 8,
+ 8, 5, 12, 9, 12, 5, 14, 6, 8, 13, 6, 5, 15, 13, 11, 11
+};
+
+static const int WA[80] = {
+ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
+ 7, 4, 13, 1, 10, 6, 15, 3, 12, 0, 9, 5, 2, 14, 11, 8,
+ 3, 10, 14, 4, 9, 15, 8, 1, 2, 7, 0, 6, 13, 11, 5, 12,
+ 1, 9, 11, 10, 0, 8, 12, 4, 13, 3, 7, 15, 14, 5, 6, 2,
+ 4, 0, 5, 9, 7, 12, 2, 10, 14, 1, 3, 8, 11, 6, 15, 13
+};
+
+static const int WB[80] = {
+ 5, 14, 7, 0, 9, 2, 11, 4, 13, 6, 15, 8, 1, 10, 3, 12,
+ 6, 11, 3, 7, 0, 13, 5, 10, 14, 15, 8, 12, 4, 9, 1, 2,
+ 15, 5, 1, 3, 7, 14, 6, 9, 11, 8, 12, 2, 10, 0, 4, 13,
+ 8, 6, 4, 1, 3, 11, 15, 0, 5, 12, 2, 13, 9, 7, 10, 14,
+ 12, 15, 10, 4, 1, 5, 8, 7, 6, 2, 13, 14, 0, 3, 9, 11
+};
+
+#define rol(value, bits) (((value) << (bits)) | ((value) >> (32 - (bits))))
+
+#define ROUND128_0_TO_15(a,b,c,d,e,f,g,h) \
+ a = rol(a + (( b ^ c ^ d) + block[WA[n]]), ROTA[n]); \
+ e = rol(e + ((((f ^ g) & h) ^ g) + block[WB[n]] + KB[0]), ROTB[n]); \
+ n++
+
+#define ROUND128_16_TO_31(a,b,c,d,e,f,g,h) \
+ a = rol(a + ((((c ^ d) & b) ^ d) + block[WA[n]] + KA[0]), ROTA[n]); \
+ e = rol(e + (((~g | f) ^ h) + block[WB[n]] + KB[1]), ROTB[n]); \
+ n++
+
+#define ROUND128_32_TO_47(a,b,c,d,e,f,g,h) \
+ a = rol(a + (((~c | b) ^ d) + block[WA[n]] + KA[1]), ROTA[n]); \
+ e = rol(e + ((((g ^ h) & f) ^ h) + block[WB[n]] + KB[2]), ROTB[n]); \
+ n++
+
+#define ROUND128_48_TO_63(a,b,c,d,e,f,g,h) \
+ a = rol(a + ((((b ^ c) & d) ^ c) + block[WA[n]] + KA[2]), ROTA[n]); \
+ e = rol(e + (( f ^ g ^ h) + block[WB[n]]), ROTB[n]); \
+ n++
+
+#define R128_0 \
+ ROUND128_0_TO_15(a,b,c,d,e,f,g,h); \
+ ROUND128_0_TO_15(d,a,b,c,h,e,f,g); \
+ ROUND128_0_TO_15(c,d,a,b,g,h,e,f); \
+ ROUND128_0_TO_15(b,c,d,a,f,g,h,e)
+
+#define R128_16 \
+ ROUND128_16_TO_31(a,b,c,d,e,f,g,h); \
+ ROUND128_16_TO_31(d,a,b,c,h,e,f,g); \
+ ROUND128_16_TO_31(c,d,a,b,g,h,e,f); \
+ ROUND128_16_TO_31(b,c,d,a,f,g,h,e)
+
+#define R128_32 \
+ ROUND128_32_TO_47(a,b,c,d,e,f,g,h); \
+ ROUND128_32_TO_47(d,a,b,c,h,e,f,g); \
+ ROUND128_32_TO_47(c,d,a,b,g,h,e,f); \
+ ROUND128_32_TO_47(b,c,d,a,f,g,h,e)
+
+#define R128_48 \
+ ROUND128_48_TO_63(a,b,c,d,e,f,g,h); \
+ ROUND128_48_TO_63(d,a,b,c,h,e,f,g); \
+ ROUND128_48_TO_63(c,d,a,b,g,h,e,f); \
+ ROUND128_48_TO_63(b,c,d,a,f,g,h,e)
+
+static void ripemd128_transform(uint32_t *state, const uint8_t buffer[64])
+{
+ uint32_t a, b, c, d, e, f, g, h, av_unused t;
+ uint32_t block[16];
+ int n;
+
+ a = e = state[0];
+ b = f = state[1];
+ c = g = state[2];
+ d = h = state[3];
+
+ for (n = 0; n < 16; n++)
+ block[n] = AV_RL32(buffer + 4 * n);
+ n = 0;
+
+#if CONFIG_SMALL
+ for (; n < 16;) {
+ ROUND128_0_TO_15(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+
+ for (; n < 32;) {
+ ROUND128_16_TO_31(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+
+ for (; n < 48;) {
+ ROUND128_32_TO_47(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+
+ for (; n < 64;) {
+ ROUND128_48_TO_63(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+#else
+
+ R128_0; R128_0; R128_0; R128_0;
+
+ R128_16; R128_16; R128_16; R128_16;
+
+ R128_32; R128_32; R128_32; R128_32;
+
+ R128_48; R128_48; R128_48; R128_48;
+#endif
+
+ h += c + state[1];
+ state[1] = state[2] + d + e;
+ state[2] = state[3] + a + f;
+ state[3] = state[0] + b + g;
+ state[0] = h;
+}
+
+static void ripemd256_transform(uint32_t *state, const uint8_t buffer[64])
+{
+ uint32_t a, b, c, d, e, f, g, h, av_unused t;
+ uint32_t block[16];
+ int n;
+
+ a = state[0]; b = state[1]; c = state[2]; d = state[3];
+ e = state[4]; f = state[5]; g = state[6]; h = state[7];
+
+ for (n = 0; n < 16; n++)
+ block[n] = AV_RL32(buffer + 4 * n);
+ n = 0;
+
+#if CONFIG_SMALL
+ for (; n < 16;) {
+ ROUND128_0_TO_15(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+ FFSWAP(uint32_t, a, e);
+
+ for (; n < 32;) {
+ ROUND128_16_TO_31(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+ FFSWAP(uint32_t, b, f);
+
+ for (; n < 48;) {
+ ROUND128_32_TO_47(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+ FFSWAP(uint32_t, c, g);
+
+ for (; n < 64;) {
+ ROUND128_48_TO_63(a,b,c,d,e,f,g,h);
+ t = d; d = c; c = b; b = a; a = t;
+ t = h; h = g; g = f; f = e; e = t;
+ }
+ FFSWAP(uint32_t, d, h);
+#else
+
+ R128_0; R128_0; R128_0; R128_0;
+ FFSWAP(uint32_t, a, e);
+
+ R128_16; R128_16; R128_16; R128_16;
+ FFSWAP(uint32_t, b, f);
+
+ R128_32; R128_32; R128_32; R128_32;
+ FFSWAP(uint32_t, c, g);
+
+ R128_48; R128_48; R128_48; R128_48;
+ FFSWAP(uint32_t, d, h);
+#endif
+
+ state[0] += a; state[1] += b; state[2] += c; state[3] += d;
+ state[4] += e; state[5] += f; state[6] += g; state[7] += h;
+}
+
+#define ROTATE(x,y) \
+ x = rol(x, 10); \
+ y = rol(y, 10); \
+ n++
+
+#define ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j) \
+ a = rol(a + (( b ^ c ^ d) + block[WA[n]]), ROTA[n]) + e; \
+ f = rol(f + (((~i | h) ^ g) + block[WB[n]] + KB[0]), ROTB[n]) + j; \
+ ROTATE(c,h)
+
+#define ROUND160_16_TO_31(a,b,c,d,e,f,g,h,i,j) \
+ a = rol(a + ((((c ^ d) & b) ^ d) + block[WA[n]] + KA[0]), ROTA[n]) + e; \
+ f = rol(f + ((((g ^ h) & i) ^ h) + block[WB[n]] + KB[1]), ROTB[n]) + j; \
+ ROTATE(c,h)
+
+#define ROUND160_32_TO_47(a,b,c,d,e,f,g,h,i,j) \
+ a = rol(a + (((~c | b) ^ d) + block[WA[n]] + KA[1]), ROTA[n]) + e; \
+ f = rol(f + (((~h | g) ^ i) + block[WB[n]] + KB[2]), ROTB[n]) + j; \
+ ROTATE(c,h)
+
+#define ROUND160_48_TO_63(a,b,c,d,e,f,g,h,i,j) \
+ a = rol(a + ((((b ^ c) & d) ^ c) + block[WA[n]] + KA[2]), ROTA[n]) + e; \
+ f = rol(f + ((((h ^ i) & g) ^ i) + block[WB[n]] + KB[3]), ROTB[n]) + j; \
+ ROTATE(c,h)
+
+#define ROUND160_64_TO_79(a,b,c,d,e,f,g,h,i,j) \
+ a = rol(a + (((~d | c) ^ b) + block[WA[n]] + KA[3]), ROTA[n]) + e; \
+ f = rol(f + (( g ^ h ^ i) + block[WB[n]]), ROTB[n]) + j; \
+ ROTATE(c,h)
+
+#define R160_0 \
+ ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j); \
+ ROUND160_0_TO_15(e,a,b,c,d,j,f,g,h,i); \
+ ROUND160_0_TO_15(d,e,a,b,c,i,j,f,g,h); \
+ ROUND160_0_TO_15(c,d,e,a,b,h,i,j,f,g); \
+ ROUND160_0_TO_15(b,c,d,e,a,g,h,i,j,f)
+
+#define R160_16 \
+ ROUND160_16_TO_31(e,a,b,c,d,j,f,g,h,i); \
+ ROUND160_16_TO_31(d,e,a,b,c,i,j,f,g,h); \
+ ROUND160_16_TO_31(c,d,e,a,b,h,i,j,f,g); \
+ ROUND160_16_TO_31(b,c,d,e,a,g,h,i,j,f); \
+ ROUND160_16_TO_31(a,b,c,d,e,f,g,h,i,j)
+
+#define R160_32 \
+ ROUND160_32_TO_47(d,e,a,b,c,i,j,f,g,h); \
+ ROUND160_32_TO_47(c,d,e,a,b,h,i,j,f,g); \
+ ROUND160_32_TO_47(b,c,d,e,a,g,h,i,j,f); \
+ ROUND160_32_TO_47(a,b,c,d,e,f,g,h,i,j); \
+ ROUND160_32_TO_47(e,a,b,c,d,j,f,g,h,i)
+
+#define R160_48 \
+ ROUND160_48_TO_63(c,d,e,a,b,h,i,j,f,g); \
+ ROUND160_48_TO_63(b,c,d,e,a,g,h,i,j,f); \
+ ROUND160_48_TO_63(a,b,c,d,e,f,g,h,i,j); \
+ ROUND160_48_TO_63(e,a,b,c,d,j,f,g,h,i); \
+ ROUND160_48_TO_63(d,e,a,b,c,i,j,f,g,h)
+
+#define R160_64 \
+ ROUND160_64_TO_79(b,c,d,e,a,g,h,i,j,f); \
+ ROUND160_64_TO_79(a,b,c,d,e,f,g,h,i,j); \
+ ROUND160_64_TO_79(e,a,b,c,d,j,f,g,h,i); \
+ ROUND160_64_TO_79(d,e,a,b,c,i,j,f,g,h); \
+ ROUND160_64_TO_79(c,d,e,a,b,h,i,j,f,g)
+
+static void ripemd160_transform(uint32_t *state, const uint8_t buffer[64])
+{
+ uint32_t a, b, c, d, e, f, g, h, i, j, av_unused t;
+ uint32_t block[16];
+ int n;
+
+ a = f = state[0];
+ b = g = state[1];
+ c = h = state[2];
+ d = i = state[3];
+ e = j = state[4];
+
+ for (n = 0; n < 16; n++)
+ block[n] = AV_RL32(buffer + 4 * n);
+ n = 0;
+
+#if CONFIG_SMALL
+ for (; n < 16;) {
+ ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+
+ for (; n < 32;) {
+ ROUND160_16_TO_31(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+
+ for (; n < 48;) {
+ ROUND160_32_TO_47(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+
+ for (; n < 64;) {
+ ROUND160_48_TO_63(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+
+ for (; n < 80;) {
+ ROUND160_64_TO_79(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+#else
+
+ R160_0; R160_0; R160_0;
+ ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j);
+
+ R160_16; R160_16; R160_16;
+ ROUND160_16_TO_31(e,a,b,c,d,j,f,g,h,i);
+
+ R160_32; R160_32; R160_32;
+ ROUND160_32_TO_47(d,e,a,b,c,i,j,f,g,h);
+
+ R160_48; R160_48; R160_48;
+ ROUND160_48_TO_63(c,d,e,a,b,h,i,j,f,g);
+
+ R160_64; R160_64; R160_64;
+ ROUND160_64_TO_79(b,c,d,e,a,g,h,i,j,f);
+#endif
+
+ i += c + state[1];
+ state[1] = state[2] + d + j;
+ state[2] = state[3] + e + f;
+ state[3] = state[4] + a + g;
+ state[4] = state[0] + b + h;
+ state[0] = i;
+}
+
+static void ripemd320_transform(uint32_t *state, const uint8_t buffer[64])
+{
+ uint32_t a, b, c, d, e, f, g, h, i, j, av_unused t;
+ uint32_t block[16];
+ int n;
+
+ a = state[0]; b = state[1]; c = state[2]; d = state[3]; e = state[4];
+ f = state[5]; g = state[6]; h = state[7]; i = state[8]; j = state[9];
+
+ for (n = 0; n < 16; n++)
+ block[n] = AV_RL32(buffer + 4 * n);
+ n = 0;
+
+#if CONFIG_SMALL
+ for (; n < 16;) {
+ ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+ FFSWAP(uint32_t, b, g);
+
+ for (; n < 32;) {
+ ROUND160_16_TO_31(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+ FFSWAP(uint32_t, d, i);
+
+ for (; n < 48;) {
+ ROUND160_32_TO_47(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+ FFSWAP(uint32_t, a, f);
+
+ for (; n < 64;) {
+ ROUND160_48_TO_63(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+ FFSWAP(uint32_t, c, h);
+
+ for (; n < 80;) {
+ ROUND160_64_TO_79(a,b,c,d,e,f,g,h,i,j);
+ t = e; e = d; d = c; c = b; b = a; a = t;
+ t = j; j = i; i = h; h = g; g = f; f = t;
+ }
+ FFSWAP(uint32_t, e, j);
+#else
+
+ R160_0; R160_0; R160_0;
+ ROUND160_0_TO_15(a,b,c,d,e,f,g,h,i,j);
+ FFSWAP(uint32_t, a, f);
+
+ R160_16; R160_16; R160_16;
+ ROUND160_16_TO_31(e,a,b,c,d,j,f,g,h,i);
+ FFSWAP(uint32_t, b, g);
+
+ R160_32; R160_32; R160_32;
+ ROUND160_32_TO_47(d,e,a,b,c,i,j,f,g,h);
+ FFSWAP(uint32_t, c, h);
+
+ R160_48; R160_48; R160_48;
+ ROUND160_48_TO_63(c,d,e,a,b,h,i,j,f,g);
+ FFSWAP(uint32_t, d, i);
+
+ R160_64; R160_64; R160_64;
+ ROUND160_64_TO_79(b,c,d,e,a,g,h,i,j,f);
+ FFSWAP(uint32_t, e, j);
+#endif
+
+ state[0] += a; state[1] += b; state[2] += c; state[3] += d; state[4] += e;
+ state[5] += f; state[6] += g; state[7] += h; state[8] += i; state[9] += j;
+}
+
+av_cold int av_ripemd_init(AVRIPEMD *ctx, int bits)
+{
+ ctx->digest_len = bits >> 5;
+ switch (bits) {
+ case 128: // RIPEMD-128
+ ctx->state[0] = 0x67452301;
+ ctx->state[1] = 0xEFCDAB89;
+ ctx->state[2] = 0x98BADCFE;
+ ctx->state[3] = 0x10325476;
+ ctx->transform = ripemd128_transform;
+ break;
+ case 160: // RIPEMD-160
+ ctx->state[0] = 0x67452301;
+ ctx->state[1] = 0xEFCDAB89;
+ ctx->state[2] = 0x98BADCFE;
+ ctx->state[3] = 0x10325476;
+ ctx->state[4] = 0xC3D2E1F0;
+ ctx->transform = ripemd160_transform;
+ break;
+ case 256: // RIPEMD-256
+ ctx->state[0] = 0x67452301;
+ ctx->state[1] = 0xEFCDAB89;
+ ctx->state[2] = 0x98BADCFE;
+ ctx->state[3] = 0x10325476;
+ ctx->state[4] = 0x76543210;
+ ctx->state[5] = 0xFEDCBA98;
+ ctx->state[6] = 0x89ABCDEF;
+ ctx->state[7] = 0x01234567;
+ ctx->transform = ripemd256_transform;
+ break;
+ case 320: // RIPEMD-320
+ ctx->state[0] = 0x67452301;
+ ctx->state[1] = 0xEFCDAB89;
+ ctx->state[2] = 0x98BADCFE;
+ ctx->state[3] = 0x10325476;
+ ctx->state[4] = 0xC3D2E1F0;
+ ctx->state[5] = 0x76543210;
+ ctx->state[6] = 0xFEDCBA98;
+ ctx->state[7] = 0x89ABCDEF;
+ ctx->state[8] = 0x01234567;
+ ctx->state[9] = 0x3C2D1E0F;
+ ctx->transform = ripemd320_transform;
+ break;
+ default:
+ return AVERROR(EINVAL);
+ }
+ ctx->count = 0;
+ return 0;
+}
+
+void av_ripemd_update(AVRIPEMD* ctx, const uint8_t* data, unsigned int len)
+{
+ unsigned int i, j;
+
+ j = ctx->count & 63;
+ ctx->count += len;
+#if CONFIG_SMALL
+ for (i = 0; i < len; i++) {
+ ctx->buffer[j++] = data[i];
+ if (64 == j) {
+ ctx->transform(ctx->state, ctx->buffer);
+ j = 0;
+ }
+ }
+#else
+ if ((j + len) > 63) {
+ memcpy(&ctx->buffer[j], data, (i = 64 - j));
+ ctx->transform(ctx->state, ctx->buffer);
+ for (; i + 63 < len; i += 64)
+ ctx->transform(ctx->state, &data[i]);
+ j = 0;
+ } else
+ i = 0;
+ memcpy(&ctx->buffer[j], &data[i], len - i);
+#endif
+}
+
+void av_ripemd_final(AVRIPEMD* ctx, uint8_t *digest)
+{
+ int i;
+ uint64_t finalcount = av_le2ne64(ctx->count << 3);
+
+ av_ripemd_update(ctx, "\200", 1);
+ while ((ctx->count & 63) != 56)
+ av_ripemd_update(ctx, "", 1);
+ av_ripemd_update(ctx, (uint8_t *)&finalcount, 8); /* Should cause a transform() */
+ for (i = 0; i < ctx->digest_len; i++)
+ AV_WL32(digest + i*4, ctx->state[i]);
+}