arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2024-11-26 18:08:20 +00:00
Code Issues Releases Wiki Activity
7c2dadab17
hashcat/OpenCL/inc_hash_sha256.cl
jsteube 51dd982b12 Bring back some volatile for AMD
2017-09-08 14:08:21 +02:00

1891 lines
44 KiB
Common Lisp
Raw Blame History

// important notes on this:
// input buf unused bytes needs to be set to zero
// input buf needs to be in algorithm native byte order (md5 = LE, sha256 = BE, etc)
// input buf needs to be 64 byte aligned when using md5_update()
__constant u32a k_sha256[64] =
{
SHA256C00, SHA256C01, SHA256C02, SHA256C03,
SHA256C04, SHA256C05, SHA256C06, SHA256C07,
SHA256C08, SHA256C09, SHA256C0a, SHA256C0b,
SHA256C0c, SHA256C0d, SHA256C0e, SHA256C0f,
SHA256C10, SHA256C11, SHA256C12, SHA256C13,
SHA256C14, SHA256C15, SHA256C16, SHA256C17,
SHA256C18, SHA256C19, SHA256C1a, SHA256C1b,
SHA256C1c, SHA256C1d, SHA256C1e, SHA256C1f,
SHA256C20, SHA256C21, SHA256C22, SHA256C23,
SHA256C24, SHA256C25, SHA256C26, SHA256C27,
SHA256C28, SHA256C29, SHA256C2a, SHA256C2b,
SHA256C2c, SHA256C2d, SHA256C2e, SHA256C2f,
SHA256C30, SHA256C31, SHA256C32, SHA256C33,
SHA256C34, SHA256C35, SHA256C36, SHA256C37,
SHA256C38, SHA256C39, SHA256C3a, SHA256C3b,
SHA256C3c, SHA256C3d, SHA256C3e, SHA256C3f,
};
typedef struct sha256_ctx
{
u32 h[8];
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int len;
} sha256_ctx_t;
static void sha256_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[8])
{
u32 a = digest[0];
u32 b = digest[1];
u32 c = digest[2];
u32 d = digest[3];
u32 e = digest[4];
u32 f = digest[5];
u32 g = digest[6];
u32 h = digest[7];
u32 w0_t = w0[0];
u32 w1_t = w0[1];
u32 w2_t = w0[2];
u32 w3_t = w0[3];
u32 w4_t = w1[0];
u32 w5_t = w1[1];
u32 w6_t = w1[2];
u32 w7_t = w1[3];
u32 w8_t = w2[0];
u32 w9_t = w2[1];
u32 wa_t = w2[2];
u32 wb_t = w2[3];
u32 wc_t = w3[0];
u32 wd_t = w3[1];
u32 we_t = w3[2];
u32 wf_t = w3[3];
#define ROUND_EXPAND_S() \
{ \
w0_t = SHA256_EXPAND_S (we_t, w9_t, w1_t, w0_t); \
w1_t = SHA256_EXPAND_S (wf_t, wa_t, w2_t, w1_t); \
w2_t = SHA256_EXPAND_S (w0_t, wb_t, w3_t, w2_t); \
w3_t = SHA256_EXPAND_S (w1_t, wc_t, w4_t, w3_t); \
w4_t = SHA256_EXPAND_S (w2_t, wd_t, w5_t, w4_t); \
w5_t = SHA256_EXPAND_S (w3_t, we_t, w6_t, w5_t); \
w6_t = SHA256_EXPAND_S (w4_t, wf_t, w7_t, w6_t); \
w7_t = SHA256_EXPAND_S (w5_t, w0_t, w8_t, w7_t); \
w8_t = SHA256_EXPAND_S (w6_t, w1_t, w9_t, w8_t); \
w9_t = SHA256_EXPAND_S (w7_t, w2_t, wa_t, w9_t); \
wa_t = SHA256_EXPAND_S (w8_t, w3_t, wb_t, wa_t); \
wb_t = SHA256_EXPAND_S (w9_t, w4_t, wc_t, wb_t); \
wc_t = SHA256_EXPAND_S (wa_t, w5_t, wd_t, wc_t); \
wd_t = SHA256_EXPAND_S (wb_t, w6_t, we_t, wd_t); \
we_t = SHA256_EXPAND_S (wc_t, w7_t, wf_t, we_t); \
wf_t = SHA256_EXPAND_S (wd_t, w8_t, w0_t, wf_t); \
}
#define ROUND_STEP_S(i) \
{ \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha256[i + 0]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha256[i + 1]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha256[i + 2]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha256[i + 3]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha256[i + 4]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha256[i + 5]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha256[i + 6]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha256[i + 7]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha256[i + 8]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha256[i + 9]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha256[i + 10]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha256[i + 11]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha256[i + 12]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha256[i + 13]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, k_sha256[i + 14]); \
SHA256_STEP_S (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha256[i + 15]); \
}
ROUND_STEP_S (0);
#ifdef _unroll
#pragma unroll
#endif
for (int i = 16; i < 64; i += 16)
{
ROUND_EXPAND_S (); ROUND_STEP_S (i);
}
#undef ROUND_EXPAND_S
#undef ROUND_STEP_S
digest[0] += a;
digest[1] += b;
digest[2] += c;
digest[3] += d;
digest[4] += e;
digest[5] += f;
digest[6] += g;
digest[7] += h;
}
static void sha256_init (sha256_ctx_t *ctx)
{
ctx->h[0] = SHA256M_A;
ctx->h[1] = SHA256M_B;
ctx->h[2] = SHA256M_C;
ctx->h[3] = SHA256M_D;
ctx->h[4] = SHA256M_E;
ctx->h[5] = SHA256M_F;
ctx->h[6] = SHA256M_G;
ctx->h[7] = SHA256M_H;
ctx->w0[0] = 0;
ctx->w0[1] = 0;
ctx->w0[2] = 0;
ctx->w0[3] = 0;
ctx->w1[0] = 0;
ctx->w1[1] = 0;
ctx->w1[2] = 0;
ctx->w1[3] = 0;
ctx->w2[0] = 0;
ctx->w2[1] = 0;
ctx->w2[2] = 0;
ctx->w2[3] = 0;
ctx->w3[0] = 0;
ctx->w3[1] = 0;
ctx->w3[2] = 0;
ctx->w3[3] = 0;
ctx->len = 0;
}
static void sha256_update_64 (sha256_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
{
#ifdef IS_AMD
volatile const int pos = ctx->len & 63;
#else
const int pos = ctx->len & 63;
#endif
ctx->len += len;
if ((pos + len) < 64)
{
switch_buffer_by_offset_be_S (w0, w1, w2, w3, pos);
ctx->w0[0] |= w0[0];
ctx->w0[1] |= w0[1];
ctx->w0[2] |= w0[2];
ctx->w0[3] |= w0[3];
ctx->w1[0] |= w1[0];
ctx->w1[1] |= w1[1];
ctx->w1[2] |= w1[2];
ctx->w1[3] |= w1[3];
ctx->w2[0] |= w2[0];
ctx->w2[1] |= w2[1];
ctx->w2[2] |= w2[2];
ctx->w2[3] |= w2[3];
ctx->w3[0] |= w3[0];
ctx->w3[1] |= w3[1];
ctx->w3[2] |= w3[2];
ctx->w3[3] |= w3[3];
}
else
{
u32 c0[4] = { 0 };
u32 c1[4] = { 0 };
u32 c2[4] = { 0 };
u32 c3[4] = { 0 };
switch_buffer_by_offset_carry_be_S (w0, w1, w2, w3, c0, c1, c2, c3, pos);
ctx->w0[0] |= w0[0];
ctx->w0[1] |= w0[1];
ctx->w0[2] |= w0[2];
ctx->w0[3] |= w0[3];
ctx->w1[0] |= w1[0];
ctx->w1[1] |= w1[1];
ctx->w1[2] |= w1[2];
ctx->w1[3] |= w1[3];
ctx->w2[0] |= w2[0];
ctx->w2[1] |= w2[1];
ctx->w2[2] |= w2[2];
ctx->w2[3] |= w2[3];
ctx->w3[0] |= w3[0];
ctx->w3[1] |= w3[1];
ctx->w3[2] |= w3[2];
ctx->w3[3] |= w3[3];
sha256_transform (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
ctx->w0[0] = c0[0];
ctx->w0[1] = c0[1];
ctx->w0[2] = c0[2];
ctx->w0[3] = c0[3];
ctx->w1[0] = c1[0];
ctx->w1[1] = c1[1];
ctx->w1[2] = c1[2];
ctx->w1[3] = c1[3];
ctx->w2[0] = c2[0];
ctx->w2[1] = c2[1];
ctx->w2[2] = c2[2];
ctx->w2[3] = c2[3];
ctx->w3[0] = c3[0];
ctx->w3[1] = c3[1];
ctx->w3[2] = c3[2];
ctx->w3[3] = c3[3];
}
}
static void sha256_update (sha256_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_swap (sha256_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_utf16le (sha256_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
sha256_update_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
sha256_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_update_utf16le_swap (sha256_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_update_global (sha256_ctx_t *ctx, const __global u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_global_swap (sha256_ctx_t *ctx, const __global u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_global_utf16le (sha256_ctx_t *ctx, const __global u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
sha256_update_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
sha256_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_update_global_utf16le_swap (sha256_ctx_t *ctx, const __global u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le_S (w1, w2, w3);
make_utf16le_S (w0, w0, w1);
w0[0] = swap32_S (w0[0]);
w0[1] = swap32_S (w0[1]);
w0[2] = swap32_S (w0[2]);
w0[3] = swap32_S (w0[3]);
w1[0] = swap32_S (w1[0]);
w1[1] = swap32_S (w1[1]);
w1[2] = swap32_S (w1[2]);
w1[3] = swap32_S (w1[3]);
w2[0] = swap32_S (w2[0]);
w2[1] = swap32_S (w2[1]);
w2[2] = swap32_S (w2[2]);
w2[3] = swap32_S (w2[3]);
w3[0] = swap32_S (w3[0]);
w3[1] = swap32_S (w3[1]);
w3[2] = swap32_S (w3[2]);
w3[3] = swap32_S (w3[3]);
sha256_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_final (sha256_ctx_t *ctx)
{
const int pos = ctx->len & 63;
append_0x80_4x4_S (ctx->w0, ctx->w1, ctx->w2, ctx->w3, pos ^ 3);
if (pos >= 56)
{
sha256_transform (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
ctx->w0[0] = 0;
ctx->w0[1] = 0;
ctx->w0[2] = 0;
ctx->w0[3] = 0;
ctx->w1[0] = 0;
ctx->w1[1] = 0;
ctx->w1[2] = 0;
ctx->w1[3] = 0;
ctx->w2[0] = 0;
ctx->w2[1] = 0;
ctx->w2[2] = 0;
ctx->w2[3] = 0;
ctx->w3[0] = 0;
ctx->w3[1] = 0;
ctx->w3[2] = 0;
ctx->w3[3] = 0;
}
ctx->w3[2] = 0;
ctx->w3[3] = ctx->len * 8;
sha256_transform (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
}
// sha256_hmac
typedef struct sha256_hmac_ctx
{
sha256_ctx_t ipad;
sha256_ctx_t opad;
} sha256_hmac_ctx_t;
static void sha256_hmac_init_64 (sha256_hmac_ctx_t *ctx, const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4])
{
u32 t0[4];
u32 t1[4];
u32 t2[4];
u32 t3[4];
// ipad
t0[0] = w0[0] ^ 0x36363636;
t0[1] = w0[1] ^ 0x36363636;
t0[2] = w0[2] ^ 0x36363636;
t0[3] = w0[3] ^ 0x36363636;
t1[0] = w1[0] ^ 0x36363636;
t1[1] = w1[1] ^ 0x36363636;
t1[2] = w1[2] ^ 0x36363636;
t1[3] = w1[3] ^ 0x36363636;
t2[0] = w2[0] ^ 0x36363636;
t2[1] = w2[1] ^ 0x36363636;
t2[2] = w2[2] ^ 0x36363636;
t2[3] = w2[3] ^ 0x36363636;
t3[0] = w3[0] ^ 0x36363636;
t3[1] = w3[1] ^ 0x36363636;
t3[2] = w3[2] ^ 0x36363636;
t3[3] = w3[3] ^ 0x36363636;
sha256_init (&ctx->ipad);
sha256_update_64 (&ctx->ipad, t0, t1, t2, t3, 64);
// opad
t0[0] = w0[0] ^ 0x5c5c5c5c;
t0[1] = w0[1] ^ 0x5c5c5c5c;
t0[2] = w0[2] ^ 0x5c5c5c5c;
t0[3] = w0[3] ^ 0x5c5c5c5c;
t1[0] = w1[0] ^ 0x5c5c5c5c;
t1[1] = w1[1] ^ 0x5c5c5c5c;
t1[2] = w1[2] ^ 0x5c5c5c5c;
t1[3] = w1[3] ^ 0x5c5c5c5c;
t2[0] = w2[0] ^ 0x5c5c5c5c;
t2[1] = w2[1] ^ 0x5c5c5c5c;
t2[2] = w2[2] ^ 0x5c5c5c5c;
t2[3] = w2[3] ^ 0x5c5c5c5c;
t3[0] = w3[0] ^ 0x5c5c5c5c;
t3[1] = w3[1] ^ 0x5c5c5c5c;
t3[2] = w3[2] ^ 0x5c5c5c5c;
t3[3] = w3[3] ^ 0x5c5c5c5c;
sha256_init (&ctx->opad);
sha256_update_64 (&ctx->opad, t0, t1, t2, t3, 64);
}
static void sha256_hmac_init (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
if (len > 64)
{
sha256_ctx_t tmp;
sha256_init (&tmp);
sha256_update (&tmp, w, len);
sha256_final (&tmp);
w0[0] = tmp.h[0];
w0[1] = tmp.h[1];
w0[2] = tmp.h[2];
w0[3] = tmp.h[3];
w1[0] = tmp.h[4];
w1[1] = tmp.h[5];
w1[2] = tmp.h[6];
w1[3] = tmp.h[7];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
}
else
{
w0[0] = w[ 0];
w0[1] = w[ 1];
w0[2] = w[ 2];
w0[3] = w[ 3];
w1[0] = w[ 4];
w1[1] = w[ 5];
w1[2] = w[ 6];
w1[3] = w[ 7];
w2[0] = w[ 8];
w2[1] = w[ 9];
w2[2] = w[10];
w2[3] = w[11];
w3[0] = w[12];
w3[1] = w[13];
w3[2] = w[14];
w3[3] = w[15];
}
sha256_hmac_init_64 (ctx, w0, w1, w2, w3);
}
static void sha256_hmac_init_swap (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
if (len > 64)
{
sha256_ctx_t tmp;
sha256_init (&tmp);
sha256_update_swap (&tmp, w, len);
sha256_final (&tmp);
w0[0] = tmp.h[0];
w0[1] = tmp.h[1];
w0[2] = tmp.h[2];
w0[3] = tmp.h[3];
w1[0] = tmp.h[4];
w1[1] = tmp.h[5];
w1[2] = tmp.h[6];
w1[3] = tmp.h[7];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
}
else
{
w0[0] = swap32_S (w[ 0]);
w0[1] = swap32_S (w[ 1]);
w0[2] = swap32_S (w[ 2]);
w0[3] = swap32_S (w[ 3]);
w1[0] = swap32_S (w[ 4]);
w1[1] = swap32_S (w[ 5]);
w1[2] = swap32_S (w[ 6]);
w1[3] = swap32_S (w[ 7]);
w2[0] = swap32_S (w[ 8]);
w2[1] = swap32_S (w[ 9]);
w2[2] = swap32_S (w[10]);
w2[3] = swap32_S (w[11]);
w3[0] = swap32_S (w[12]);
w3[1] = swap32_S (w[13]);
w3[2] = swap32_S (w[14]);
w3[3] = swap32_S (w[15]);
}
sha256_hmac_init_64 (ctx, w0, w1, w2, w3);
}
static void sha256_hmac_init_global (sha256_hmac_ctx_t *ctx, __global const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
if (len > 64)
{
sha256_ctx_t tmp;
sha256_init (&tmp);
sha256_update_global (&tmp, w, len);
sha256_final (&tmp);
w0[0] = tmp.h[0];
w0[1] = tmp.h[1];
w0[2] = tmp.h[2];
w0[3] = tmp.h[3];
w1[0] = tmp.h[4];
w1[1] = tmp.h[5];
w1[2] = tmp.h[6];
w1[3] = tmp.h[7];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
}
else
{
w0[0] = w[ 0];
w0[1] = w[ 1];
w0[2] = w[ 2];
w0[3] = w[ 3];
w1[0] = w[ 4];
w1[1] = w[ 5];
w1[2] = w[ 6];
w1[3] = w[ 7];
w2[0] = w[ 8];
w2[1] = w[ 9];
w2[2] = w[10];
w2[3] = w[11];
w3[0] = w[12];
w3[1] = w[13];
w3[2] = w[14];
w3[3] = w[15];
}
sha256_hmac_init_64 (ctx, w0, w1, w2, w3);
}
static void sha256_hmac_init_global_swap (sha256_hmac_ctx_t *ctx, __global const u32 *w, const int len)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
if (len > 64)
{
sha256_ctx_t tmp;
sha256_init (&tmp);
sha256_update_global_swap (&tmp, w, len);
sha256_final (&tmp);
w0[0] = tmp.h[0];
w0[1] = tmp.h[1];
w0[2] = tmp.h[2];
w0[3] = tmp.h[3];
w1[0] = tmp.h[4];
w1[1] = tmp.h[5];
w1[2] = tmp.h[6];
w1[3] = tmp.h[7];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
}
else
{
w0[0] = swap32_S (w[ 0]);
w0[1] = swap32_S (w[ 1]);
w0[2] = swap32_S (w[ 2]);
w0[3] = swap32_S (w[ 3]);
w1[0] = swap32_S (w[ 4]);
w1[1] = swap32_S (w[ 5]);
w1[2] = swap32_S (w[ 6]);
w1[3] = swap32_S (w[ 7]);
w2[0] = swap32_S (w[ 8]);
w2[1] = swap32_S (w[ 9]);
w2[2] = swap32_S (w[10]);
w2[3] = swap32_S (w[11]);
w3[0] = swap32_S (w[12]);
w3[1] = swap32_S (w[13]);
w3[2] = swap32_S (w[14]);
w3[3] = swap32_S (w[15]);
}
sha256_hmac_init_64 (ctx, w0, w1, w2, w3);
}
static void sha256_hmac_update_64 (sha256_hmac_ctx_t *ctx, u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const int len)
{
sha256_update_64 (&ctx->ipad, w0, w1, w2, w3, len);
}
static void sha256_hmac_update (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
sha256_update (&ctx->ipad, w, len);
}
static void sha256_hmac_update_swap (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
sha256_update_swap (&ctx->ipad, w, len);
}
static void sha256_hmac_update_utf16le (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
sha256_update_utf16le (&ctx->ipad, w, len);
}
static void sha256_hmac_update_utf16le_swap (sha256_hmac_ctx_t *ctx, const u32 *w, const int len)
{
sha256_update_utf16le_swap (&ctx->ipad, w, len);
}
static void sha256_hmac_update_global (sha256_hmac_ctx_t *ctx, const __global u32 *w, const int len)
{
sha256_update_global (&ctx->ipad, w, len);
}
static void sha256_hmac_update_global_swap (sha256_hmac_ctx_t *ctx, const __global u32 *w, const int len)
{
sha256_update_global_swap (&ctx->ipad, w, len);
}
static void sha256_hmac_update_global_utf16le (sha256_hmac_ctx_t *ctx, const __global u32 *w, const int len)
{
sha256_update_global_utf16le (&ctx->ipad, w, len);
}
static void sha256_hmac_update_global_utf16le_swap (sha256_hmac_ctx_t *ctx, const __global u32 *w, const int len)
{
sha256_update_global_utf16le_swap (&ctx->ipad, w, len);
}
static void sha256_hmac_final (sha256_hmac_ctx_t *ctx)
{
sha256_final (&ctx->ipad);
u32 t0[4];
u32 t1[4];
u32 t2[4];
u32 t3[4];
t0[0] = ctx->ipad.h[0];
t0[1] = ctx->ipad.h[1];
t0[2] = ctx->ipad.h[2];
t0[3] = ctx->ipad.h[3];
t1[0] = ctx->ipad.h[4];
t1[1] = ctx->ipad.h[5];
t1[2] = ctx->ipad.h[6];
t1[3] = ctx->ipad.h[7];
t2[0] = 0;
t2[1] = 0;
t2[2] = 0;
t2[3] = 0;
t3[0] = 0;
t3[1] = 0;
t3[2] = 0;
t3[3] = 0;
sha256_update_64 (&ctx->opad, t0, t1, t2, t3, 32);
sha256_final (&ctx->opad);
}
// while input buf can be a vector datatype, the length of the different elements can not
typedef struct sha256_ctx_vector
{
u32x h[8];
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int len;
} sha256_ctx_vector_t;
static void sha256_transform_vector (const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4], u32x digest[8])
{
u32x a = digest[0];
u32x b = digest[1];
u32x c = digest[2];
u32x d = digest[3];
u32x e = digest[4];
u32x f = digest[5];
u32x g = digest[6];
u32x h = digest[7];
u32x w0_t = w0[0];
u32x w1_t = w0[1];
u32x w2_t = w0[2];
u32x w3_t = w0[3];
u32x w4_t = w1[0];
u32x w5_t = w1[1];
u32x w6_t = w1[2];
u32x w7_t = w1[3];
u32x w8_t = w2[0];
u32x w9_t = w2[1];
u32x wa_t = w2[2];
u32x wb_t = w2[3];
u32x wc_t = w3[0];
u32x wd_t = w3[1];
u32x we_t = w3[2];
u32x wf_t = w3[3];
#define ROUND_EXPAND() \
{ \
w0_t = SHA256_EXPAND (we_t, w9_t, w1_t, w0_t); \
w1_t = SHA256_EXPAND (wf_t, wa_t, w2_t, w1_t); \
w2_t = SHA256_EXPAND (w0_t, wb_t, w3_t, w2_t); \
w3_t = SHA256_EXPAND (w1_t, wc_t, w4_t, w3_t); \
w4_t = SHA256_EXPAND (w2_t, wd_t, w5_t, w4_t); \
w5_t = SHA256_EXPAND (w3_t, we_t, w6_t, w5_t); \
w6_t = SHA256_EXPAND (w4_t, wf_t, w7_t, w6_t); \
w7_t = SHA256_EXPAND (w5_t, w0_t, w8_t, w7_t); \
w8_t = SHA256_EXPAND (w6_t, w1_t, w9_t, w8_t); \
w9_t = SHA256_EXPAND (w7_t, w2_t, wa_t, w9_t); \
wa_t = SHA256_EXPAND (w8_t, w3_t, wb_t, wa_t); \
wb_t = SHA256_EXPAND (w9_t, w4_t, wc_t, wb_t); \
wc_t = SHA256_EXPAND (wa_t, w5_t, wd_t, wc_t); \
wd_t = SHA256_EXPAND (wb_t, w6_t, we_t, wd_t); \
we_t = SHA256_EXPAND (wc_t, w7_t, wf_t, we_t); \
wf_t = SHA256_EXPAND (wd_t, w8_t, w0_t, wf_t); \
}
#define ROUND_STEP(i) \
{ \
SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha256[i + 0]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha256[i + 1]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha256[i + 2]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha256[i + 3]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha256[i + 4]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha256[i + 5]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha256[i + 6]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha256[i + 7]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha256[i + 8]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha256[i + 9]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha256[i + 10]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha256[i + 11]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha256[i + 12]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha256[i + 13]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, k_sha256[i + 14]); \
SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha256[i + 15]); \
}
ROUND_STEP (0);
#ifdef _unroll
#pragma unroll
#endif
for (int i = 16; i < 64; i += 16)
{
ROUND_EXPAND (); ROUND_STEP (i);
}
#undef ROUND_EXPAND
#undef ROUND_STEP
digest[0] += a;
digest[1] += b;
digest[2] += c;
digest[3] += d;
digest[4] += e;
digest[5] += f;
digest[6] += g;
digest[7] += h;
}
static void sha256_init_vector (sha256_ctx_vector_t *ctx)
{
ctx->h[0] = SHA256M_A;
ctx->h[1] = SHA256M_B;
ctx->h[2] = SHA256M_C;
ctx->h[3] = SHA256M_D;
ctx->h[4] = SHA256M_E;
ctx->h[5] = SHA256M_F;
ctx->h[6] = SHA256M_G;
ctx->h[7] = SHA256M_H;
ctx->w0[0] = 0;
ctx->w0[1] = 0;
ctx->w0[2] = 0;
ctx->w0[3] = 0;
ctx->w1[0] = 0;
ctx->w1[1] = 0;
ctx->w1[2] = 0;
ctx->w1[3] = 0;
ctx->w2[0] = 0;
ctx->w2[1] = 0;
ctx->w2[2] = 0;
ctx->w2[3] = 0;
ctx->w3[0] = 0;
ctx->w3[1] = 0;
ctx->w3[2] = 0;
ctx->w3[3] = 0;
ctx->len = 0;
}
static void sha256_init_vector_from_scalar (sha256_ctx_vector_t *ctx, sha256_ctx_t *ctx0)
{
ctx->h[0] = ctx0->h[0];
ctx->h[1] = ctx0->h[1];
ctx->h[2] = ctx0->h[2];
ctx->h[3] = ctx0->h[3];
ctx->h[4] = ctx0->h[4];
ctx->h[5] = ctx0->h[5];
ctx->h[6] = ctx0->h[6];
ctx->h[7] = ctx0->h[7];
ctx->w0[0] = ctx0->w0[0];
ctx->w0[1] = ctx0->w0[1];
ctx->w0[2] = ctx0->w0[2];
ctx->w0[3] = ctx0->w0[3];
ctx->w1[0] = ctx0->w1[0];
ctx->w1[1] = ctx0->w1[1];
ctx->w1[2] = ctx0->w1[2];
ctx->w1[3] = ctx0->w1[3];
ctx->w2[0] = ctx0->w2[0];
ctx->w2[1] = ctx0->w2[1];
ctx->w2[2] = ctx0->w2[2];
ctx->w2[3] = ctx0->w2[3];
ctx->w3[0] = ctx0->w3[0];
ctx->w3[1] = ctx0->w3[1];
ctx->w3[2] = ctx0->w3[2];
ctx->w3[3] = ctx0->w3[3];
ctx->len = ctx0->len;
}
static void sha256_update_vector_64 (sha256_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
{
#ifdef IS_AMD
volatile const int pos = ctx->len & 63;
#else
const int pos = ctx->len & 63;
#endif
ctx->len += len;
if ((pos + len) < 64)
{
switch_buffer_by_offset_be (w0, w1, w2, w3, pos);
ctx->w0[0] |= w0[0];
ctx->w0[1] |= w0[1];
ctx->w0[2] |= w0[2];
ctx->w0[3] |= w0[3];
ctx->w1[0] |= w1[0];
ctx->w1[1] |= w1[1];
ctx->w1[2] |= w1[2];
ctx->w1[3] |= w1[3];
ctx->w2[0] |= w2[0];
ctx->w2[1] |= w2[1];
ctx->w2[2] |= w2[2];
ctx->w2[3] |= w2[3];
ctx->w3[0] |= w3[0];
ctx->w3[1] |= w3[1];
ctx->w3[2] |= w3[2];
ctx->w3[3] |= w3[3];
}
else
{
u32x c0[4] = { 0 };
u32x c1[4] = { 0 };
u32x c2[4] = { 0 };
u32x c3[4] = { 0 };
switch_buffer_by_offset_carry_be (w0, w1, w2, w3, c0, c1, c2, c3, pos);
ctx->w0[0] |= w0[0];
ctx->w0[1] |= w0[1];
ctx->w0[2] |= w0[2];
ctx->w0[3] |= w0[3];
ctx->w1[0] |= w1[0];
ctx->w1[1] |= w1[1];
ctx->w1[2] |= w1[2];
ctx->w1[3] |= w1[3];
ctx->w2[0] |= w2[0];
ctx->w2[1] |= w2[1];
ctx->w2[2] |= w2[2];
ctx->w2[3] |= w2[3];
ctx->w3[0] |= w3[0];
ctx->w3[1] |= w3[1];
ctx->w3[2] |= w3[2];
ctx->w3[3] |= w3[3];
sha256_transform_vector (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
ctx->w0[0] = c0[0];
ctx->w0[1] = c0[1];
ctx->w0[2] = c0[2];
ctx->w0[3] = c0[3];
ctx->w1[0] = c1[0];
ctx->w1[1] = c1[1];
ctx->w1[2] = c1[2];
ctx->w1[3] = c1[3];
ctx->w2[0] = c2[0];
ctx->w2[1] = c2[1];
ctx->w2[2] = c2[2];
ctx->w2[3] = c2[3];
ctx->w3[0] = c3[0];
ctx->w3[1] = c3[1];
ctx->w3[2] = c3[2];
ctx->w3[3] = c3[3];
}
}
static void sha256_update_vector (sha256_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_vector_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
sha256_update_vector_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_vector_swap (sha256_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32 (w0[0]);
w0[1] = swap32 (w0[1]);
w0[2] = swap32 (w0[2]);
w0[3] = swap32 (w0[3]);
w1[0] = swap32 (w1[0]);
w1[1] = swap32 (w1[1]);
w1[2] = swap32 (w1[2]);
w1[3] = swap32 (w1[3]);
w2[0] = swap32 (w2[0]);
w2[1] = swap32 (w2[1]);
w2[2] = swap32 (w2[2]);
w2[3] = swap32 (w2[3]);
w3[0] = swap32 (w3[0]);
w3[1] = swap32 (w3[1]);
w3[2] = swap32 (w3[2]);
w3[3] = swap32 (w3[3]);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, 64);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
w2[0] = w[pos4 + 8];
w2[1] = w[pos4 + 9];
w2[2] = w[pos4 + 10];
w2[3] = w[pos4 + 11];
w3[0] = w[pos4 + 12];
w3[1] = w[pos4 + 13];
w3[2] = w[pos4 + 14];
w3[3] = w[pos4 + 15];
w0[0] = swap32 (w0[0]);
w0[1] = swap32 (w0[1]);
w0[2] = swap32 (w0[2]);
w0[3] = swap32 (w0[3]);
w1[0] = swap32 (w1[0]);
w1[1] = swap32 (w1[1]);
w1[2] = swap32 (w1[2]);
w1[3] = swap32 (w1[3]);
w2[0] = swap32 (w2[0]);
w2[1] = swap32 (w2[1]);
w2[2] = swap32 (w2[2]);
w2[3] = swap32 (w2[3]);
w3[0] = swap32 (w3[0]);
w3[1] = swap32 (w3[1]);
w3[2] = swap32 (w3[2]);
w3[3] = swap32 (w3[3]);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, len - pos1);
}
static void sha256_update_vector_utf16le (sha256_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le (w1, w2, w3);
make_utf16le (w0, w0, w1);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le (w1, w2, w3);
make_utf16le (w0, w0, w1);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_update_vector_utf16le_swap (sha256_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le (w1, w2, w3);
make_utf16le (w0, w0, w1);
w0[0] = swap32 (w0[0]);
w0[1] = swap32 (w0[1]);
w0[2] = swap32 (w0[2]);
w0[3] = swap32 (w0[3]);
w1[0] = swap32 (w1[0]);
w1[1] = swap32 (w1[1]);
w1[2] = swap32 (w1[2]);
w1[3] = swap32 (w1[3]);
w2[0] = swap32 (w2[0]);
w2[1] = swap32 (w2[1]);
w2[2] = swap32 (w2[2]);
w2[3] = swap32 (w2[3]);
w3[0] = swap32 (w3[0]);
w3[1] = swap32 (w3[1]);
w3[2] = swap32 (w3[2]);
w3[3] = swap32 (w3[3]);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16le (w1, w2, w3);
make_utf16le (w0, w0, w1);
w0[0] = swap32 (w0[0]);
w0[1] = swap32 (w0[1]);
w0[2] = swap32 (w0[2]);
w0[3] = swap32 (w0[3]);
w1[0] = swap32 (w1[0]);
w1[1] = swap32 (w1[1]);
w1[2] = swap32 (w1[2]);
w1[3] = swap32 (w1[3]);
w2[0] = swap32 (w2[0]);
w2[1] = swap32 (w2[1]);
w2[2] = swap32 (w2[2]);
w2[3] = swap32 (w2[3]);
w3[0] = swap32 (w3[0]);
w3[1] = swap32 (w3[1]);
w3[2] = swap32 (w3[2]);
w3[3] = swap32 (w3[3]);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_update_vector_utf16beN (sha256_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
int pos1;
int pos4;
for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8)
{
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16beN (w1, w2, w3);
make_utf16beN (w0, w0, w1);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, 32 * 2);
}
w0[0] = w[pos4 + 0];
w0[1] = w[pos4 + 1];
w0[2] = w[pos4 + 2];
w0[3] = w[pos4 + 3];
w1[0] = w[pos4 + 4];
w1[1] = w[pos4 + 5];
w1[2] = w[pos4 + 6];
w1[3] = w[pos4 + 7];
make_utf16beN (w1, w2, w3);
make_utf16beN (w0, w0, w1);
sha256_update_vector_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2);
}
static void sha256_final_vector (sha256_ctx_vector_t *ctx)
{
const int pos = ctx->len & 63;
append_0x80_4x4 (ctx->w0, ctx->w1, ctx->w2, ctx->w3, pos ^ 3);
if (pos >= 56)
{
sha256_transform_vector (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
ctx->w0[0] = 0;
ctx->w0[1] = 0;
ctx->w0[2] = 0;
ctx->w0[3] = 0;
ctx->w1[0] = 0;
ctx->w1[1] = 0;
ctx->w1[2] = 0;
ctx->w1[3] = 0;
ctx->w2[0] = 0;
ctx->w2[1] = 0;
ctx->w2[2] = 0;
ctx->w2[3] = 0;
ctx->w3[0] = 0;
ctx->w3[1] = 0;
ctx->w3[2] = 0;
ctx->w3[3] = 0;
}
ctx->w3[2] = 0;
ctx->w3[3] = ctx->len * 8;
sha256_transform_vector (ctx->w0, ctx->w1, ctx->w2, ctx->w3, ctx->h);
}
// HMAC + Vector
typedef struct sha256_hmac_ctx_vector
{
sha256_ctx_vector_t ipad;
sha256_ctx_vector_t opad;
} sha256_hmac_ctx_vector_t;
static void sha256_hmac_init_vector_64 (sha256_hmac_ctx_vector_t *ctx, const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4])
{
u32x t0[4];
u32x t1[4];
u32x t2[4];
u32x t3[4];
// ipad
t0[0] = w0[0] ^ 0x36363636;
t0[1] = w0[1] ^ 0x36363636;
t0[2] = w0[2] ^ 0x36363636;
t0[3] = w0[3] ^ 0x36363636;
t1[0] = w1[0] ^ 0x36363636;
t1[1] = w1[1] ^ 0x36363636;
t1[2] = w1[2] ^ 0x36363636;
t1[3] = w1[3] ^ 0x36363636;
t2[0] = w2[0] ^ 0x36363636;
t2[1] = w2[1] ^ 0x36363636;
t2[2] = w2[2] ^ 0x36363636;
t2[3] = w2[3] ^ 0x36363636;
t3[0] = w3[0] ^ 0x36363636;
t3[1] = w3[1] ^ 0x36363636;
t3[2] = w3[2] ^ 0x36363636;
t3[3] = w3[3] ^ 0x36363636;
sha256_init_vector (&ctx->ipad);
sha256_update_vector_64 (&ctx->ipad, t0, t1, t2, t3, 64);
// opad
t0[0] = w0[0] ^ 0x5c5c5c5c;
t0[1] = w0[1] ^ 0x5c5c5c5c;
t0[2] = w0[2] ^ 0x5c5c5c5c;
t0[3] = w0[3] ^ 0x5c5c5c5c;
t1[0] = w1[0] ^ 0x5c5c5c5c;
t1[1] = w1[1] ^ 0x5c5c5c5c;
t1[2] = w1[2] ^ 0x5c5c5c5c;
t1[3] = w1[3] ^ 0x5c5c5c5c;
t2[0] = w2[0] ^ 0x5c5c5c5c;
t2[1] = w2[1] ^ 0x5c5c5c5c;
t2[2] = w2[2] ^ 0x5c5c5c5c;
t2[3] = w2[3] ^ 0x5c5c5c5c;
t3[0] = w3[0] ^ 0x5c5c5c5c;
t3[1] = w3[1] ^ 0x5c5c5c5c;
t3[2] = w3[2] ^ 0x5c5c5c5c;
t3[3] = w3[3] ^ 0x5c5c5c5c;
sha256_init_vector (&ctx->opad);
sha256_update_vector_64 (&ctx->opad, t0, t1, t2, t3, 64);
}
static void sha256_hmac_init_vector (sha256_hmac_ctx_vector_t *ctx, const u32x *w, const int len)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
if (len > 64)
{
sha256_ctx_vector_t tmp;
sha256_init_vector (&tmp);
sha256_update_vector (&tmp, w, len);
sha256_final_vector (&tmp);
w0[0] = tmp.h[0];
w0[1] = tmp.h[1];
w0[2] = tmp.h[2];
w0[3] = tmp.h[3];
w1[0] = tmp.h[4];
w1[1] = tmp.h[5];
w1[2] = tmp.h[6];
w1[3] = tmp.h[7];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
}
else
{
w0[0] = w[ 0];
w0[1] = w[ 1];
w0[2] = w[ 2];
w0[3] = w[ 3];
w1[0] = w[ 4];
w1[1] = w[ 5];
w1[2] = w[ 6];
w1[3] = w[ 7];
w2[0] = w[ 8];
w2[1] = w[ 9];
w2[2] = w[10];
w2[3] = w[11];
w3[0] = w[12];
w3[1] = w[13];
w3[2] = w[14];
w3[3] = w[15];
}
sha256_hmac_init_vector_64 (ctx, w0, w1, w2, w3);
}
static void sha256_hmac_update_vector_64 (sha256_hmac_ctx_vector_t *ctx, u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const int len)
{
sha256_update_vector_64 (&ctx->ipad, w0, w1, w2, w3, len);
}
static void sha256_hmac_update_vector (sha256_hmac_ctx_vector_t *ctx, const u32x *w, const int len)
{
sha256_update_vector (&ctx->ipad, w, len);
}
static void sha256_hmac_final_vector (sha256_hmac_ctx_vector_t *ctx)
{
sha256_final_vector (&ctx->ipad);
u32x t0[4];
u32x t1[4];
u32x t2[4];
u32x t3[4];
t0[0] = ctx->ipad.h[0];
t0[1] = ctx->ipad.h[1];
t0[2] = ctx->ipad.h[2];
t0[3] = ctx->ipad.h[3];
t1[0] = ctx->ipad.h[4];
t1[1] = ctx->ipad.h[5];
t1[2] = ctx->ipad.h[6];
t1[3] = ctx->ipad.h[7];
t2[0] = 0;
t2[1] = 0;
t2[2] = 0;
t2[3] = 0;
t3[0] = 0;
t3[1] = 0;
t3[2] = 0;
t3[3] = 0;
sha256_update_vector_64 (&ctx->opad, t0, t1, t2, t3, 32);
sha256_final_vector (&ctx->opad);
}
Reference in New Issue View Git Blame Copy Permalink