1/*
2chacha-merged.c version 20080118
3D. J. Bernstein
4Public domain.
5*/
6
7#include "includes.h"
8
9#include "chacha.h"
10
11/* $OpenBSD: chacha.c,v 1.1 2013/11/21 00:45:44 djm Exp $ */
12
13typedef unsigned char u8;
14typedef unsigned int u32;
15
16typedef struct chacha_ctx chacha_ctx;
17
18#define U8C(v) (v##U)
19#define U32C(v) (v##U)
20
21#define U8V(v) ((u8)(v) & U8C(0xFF))
22#define U32V(v) ((u32)(v) & U32C(0xFFFFFFFF))
23
24#define ROTL32(v, n) \
25  (U32V((v) << (n)) | ((v) >> (32 - (n))))
26
27#define U8TO32_LITTLE(p) \
28  (((u32)((p)[0])      ) | \
29   ((u32)((p)[1]) <<  8) | \
30   ((u32)((p)[2]) << 16) | \
31   ((u32)((p)[3]) << 24))
32
33#define U32TO8_LITTLE(p, v) \
34  do { \
35    (p)[0] = U8V((v)      ); \
36    (p)[1] = U8V((v) >>  8); \
37    (p)[2] = U8V((v) >> 16); \
38    (p)[3] = U8V((v) >> 24); \
39  } while (0)
40
41#define ROTATE(v,c) (ROTL32(v,c))
42#define XOR(v,w) ((v) ^ (w))
43#define PLUS(v,w) (U32V((v) + (w)))
44#define PLUSONE(v) (PLUS((v),1))
45
46#define QUARTERROUND(a,b,c,d) \
47  a = PLUS(a,b); d = ROTATE(XOR(d,a),16); \
48  c = PLUS(c,d); b = ROTATE(XOR(b,c),12); \
49  a = PLUS(a,b); d = ROTATE(XOR(d,a), 8); \
50  c = PLUS(c,d); b = ROTATE(XOR(b,c), 7);
51
52static const char sigma[16] = "expand 32-byte k";
53static const char tau[16] = "expand 16-byte k";
54
55void
56chacha_keysetup(chacha_ctx *x,const u8 *k,u32 kbits)
57{
58  const char *constants;
59
60  x->input[4] = U8TO32_LITTLE(k + 0);
61  x->input[5] = U8TO32_LITTLE(k + 4);
62  x->input[6] = U8TO32_LITTLE(k + 8);
63  x->input[7] = U8TO32_LITTLE(k + 12);
64  if (kbits == 256) { /* recommended */
65    k += 16;
66    constants = sigma;
67  } else { /* kbits == 128 */
68    constants = tau;
69  }
70  x->input[8] = U8TO32_LITTLE(k + 0);
71  x->input[9] = U8TO32_LITTLE(k + 4);
72  x->input[10] = U8TO32_LITTLE(k + 8);
73  x->input[11] = U8TO32_LITTLE(k + 12);
74  x->input[0] = U8TO32_LITTLE(constants + 0);
75  x->input[1] = U8TO32_LITTLE(constants + 4);
76  x->input[2] = U8TO32_LITTLE(constants + 8);
77  x->input[3] = U8TO32_LITTLE(constants + 12);
78}
79
80void
81chacha_ivsetup(chacha_ctx *x, const u8 *iv, const u8 *counter)
82{
83  x->input[12] = counter == NULL ? 0 : U8TO32_LITTLE(counter + 0);
84  x->input[13] = counter == NULL ? 0 : U8TO32_LITTLE(counter + 4);
85  x->input[14] = U8TO32_LITTLE(iv + 0);
86  x->input[15] = U8TO32_LITTLE(iv + 4);
87}
88
89void
90chacha_encrypt_bytes(chacha_ctx *x,const u8 *m,u8 *c,u32 bytes)
91{
92  u32 x0, x1, x2, x3, x4, x5, x6, x7, x8, x9, x10, x11, x12, x13, x14, x15;
93  u32 j0, j1, j2, j3, j4, j5, j6, j7, j8, j9, j10, j11, j12, j13, j14, j15;
94  u8 *ctarget = NULL;
95  u8 tmp[64];
96  u_int i;
97
98  if (!bytes) return;
99
100  j0 = x->input[0];
101  j1 = x->input[1];
102  j2 = x->input[2];
103  j3 = x->input[3];
104  j4 = x->input[4];
105  j5 = x->input[5];
106  j6 = x->input[6];
107  j7 = x->input[7];
108  j8 = x->input[8];
109  j9 = x->input[9];
110  j10 = x->input[10];
111  j11 = x->input[11];
112  j12 = x->input[12];
113  j13 = x->input[13];
114  j14 = x->input[14];
115  j15 = x->input[15];
116
117  for (;;) {
118    if (bytes < 64) {
119      for (i = 0;i < bytes;++i) tmp[i] = m[i];
120      m = tmp;
121      ctarget = c;
122      c = tmp;
123    }
124    x0 = j0;
125    x1 = j1;
126    x2 = j2;
127    x3 = j3;
128    x4 = j4;
129    x5 = j5;
130    x6 = j6;
131    x7 = j7;
132    x8 = j8;
133    x9 = j9;
134    x10 = j10;
135    x11 = j11;
136    x12 = j12;
137    x13 = j13;
138    x14 = j14;
139    x15 = j15;
140    for (i = 20;i > 0;i -= 2) {
141      QUARTERROUND( x0, x4, x8,x12)
142      QUARTERROUND( x1, x5, x9,x13)
143      QUARTERROUND( x2, x6,x10,x14)
144      QUARTERROUND( x3, x7,x11,x15)
145      QUARTERROUND( x0, x5,x10,x15)
146      QUARTERROUND( x1, x6,x11,x12)
147      QUARTERROUND( x2, x7, x8,x13)
148      QUARTERROUND( x3, x4, x9,x14)
149    }
150    x0 = PLUS(x0,j0);
151    x1 = PLUS(x1,j1);
152    x2 = PLUS(x2,j2);
153    x3 = PLUS(x3,j3);
154    x4 = PLUS(x4,j4);
155    x5 = PLUS(x5,j5);
156    x6 = PLUS(x6,j6);
157    x7 = PLUS(x7,j7);
158    x8 = PLUS(x8,j8);
159    x9 = PLUS(x9,j9);
160    x10 = PLUS(x10,j10);
161    x11 = PLUS(x11,j11);
162    x12 = PLUS(x12,j12);
163    x13 = PLUS(x13,j13);
164    x14 = PLUS(x14,j14);
165    x15 = PLUS(x15,j15);
166
167    x0 = XOR(x0,U8TO32_LITTLE(m + 0));
168    x1 = XOR(x1,U8TO32_LITTLE(m + 4));
169    x2 = XOR(x2,U8TO32_LITTLE(m + 8));
170    x3 = XOR(x3,U8TO32_LITTLE(m + 12));
171    x4 = XOR(x4,U8TO32_LITTLE(m + 16));
172    x5 = XOR(x5,U8TO32_LITTLE(m + 20));
173    x6 = XOR(x6,U8TO32_LITTLE(m + 24));
174    x7 = XOR(x7,U8TO32_LITTLE(m + 28));
175    x8 = XOR(x8,U8TO32_LITTLE(m + 32));
176    x9 = XOR(x9,U8TO32_LITTLE(m + 36));
177    x10 = XOR(x10,U8TO32_LITTLE(m + 40));
178    x11 = XOR(x11,U8TO32_LITTLE(m + 44));
179    x12 = XOR(x12,U8TO32_LITTLE(m + 48));
180    x13 = XOR(x13,U8TO32_LITTLE(m + 52));
181    x14 = XOR(x14,U8TO32_LITTLE(m + 56));
182    x15 = XOR(x15,U8TO32_LITTLE(m + 60));
183
184    j12 = PLUSONE(j12);
185    if (!j12) {
186      j13 = PLUSONE(j13);
187      /* stopping at 2^70 bytes per nonce is user's responsibility */
188    }
189
190    U32TO8_LITTLE(c + 0,x0);
191    U32TO8_LITTLE(c + 4,x1);
192    U32TO8_LITTLE(c + 8,x2);
193    U32TO8_LITTLE(c + 12,x3);
194    U32TO8_LITTLE(c + 16,x4);
195    U32TO8_LITTLE(c + 20,x5);
196    U32TO8_LITTLE(c + 24,x6);
197    U32TO8_LITTLE(c + 28,x7);
198    U32TO8_LITTLE(c + 32,x8);
199    U32TO8_LITTLE(c + 36,x9);
200    U32TO8_LITTLE(c + 40,x10);
201    U32TO8_LITTLE(c + 44,x11);
202    U32TO8_LITTLE(c + 48,x12);
203    U32TO8_LITTLE(c + 52,x13);
204    U32TO8_LITTLE(c + 56,x14);
205    U32TO8_LITTLE(c + 60,x15);
206
207    if (bytes <= 64) {
208      if (bytes < 64) {
209        for (i = 0;i < bytes;++i) ctarget[i] = c[i];
210      }
211      x->input[12] = j12;
212      x->input[13] = j13;
213      return;
214    }
215    bytes -= 64;
216    c += 64;
217    m += 64;
218  }
219}
220