gfmul.c

使用visual studio 2005 开发的开源文件、磁盘加密软件。这是6.1a版。加密自己资料的好工具。也是学习的优秀范本。结成了众多加密算法。

    xor_8k( 8); xor_8k( 9); xor_8k(10); xor_8k(11);
    xor_8k(12); xor_8k(13); xor_8k(14); xor_8k(15);
    move_block_aligned(a, r);
}

#else

void GfMul128Tab (unsigned char a[CBLK_LEN], GfCtx8k *ctx)
{   unsigned __int32 r[CBLK_LEN >> 2], *p;
    int i;

    p = ctx->gf_t8k[0][a[0] & 15];
    memcpy(r, p, CBLK_LEN);
    p = ctx->gf_t8k[1][a[0] >> 4];
    xor_block_aligned(r, p);
    for(i = 1; i < CBLK_LEN; ++i)
    {
        xor_block_aligned(r, ctx->gf_t8k[i + i][a[i] & 15]);
        xor_block_aligned(r, ctx->gf_t8k[i + i + 1][a[i] >> 4]);
    }
    memcpy(a, r, CBLK_LEN);
}

#endif

void compile_8k_table(unsigned __int8 *a, GfCtx8k *ctx)
{   int i, j, k;

    memset(ctx->gf_t8k, 0, 32 * 16 * 16);
    for(i = 0; i < 2 * CBLK_LEN; ++i)
    {
        if(i == 0)
        {
            memcpy(ctx->gf_t8k[1][8], a, CBLK_LEN);
            for(j = 4; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t8k[1][j], ctx->gf_t8k[1][j + j], CBLK_LEN);
                mul_x(ctx->gf_t8k[1][j]);
            }
            memcpy(ctx->gf_t8k[0][8], ctx->gf_t8k[1][1], CBLK_LEN);
            mul_x(ctx->gf_t8k[0][8]);
            for(j = 4; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t8k[0][j], ctx->gf_t8k[0][j + j], CBLK_LEN);
                mul_x(ctx->gf_t8k[0][j]);
            }
        }
        else if(i > 1)
            for(j = 8; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t8k[i][j], ctx->gf_t8k[i - 2][j], CBLK_LEN);
                mul_x8(ctx->gf_t8k[i][j]);
            }

        for(j = 2; j < 16; j += j)
        {
            mode(32t) *pj = ctx->gf_t8k[i][j];
            mode(32t) *pk = ctx->gf_t8k[i][1];
            mode(32t) *pl = ctx->gf_t8k[i][j + 1];

            for(k = 1; k < j; ++k)
            {
                *pl++ = pj[0] ^ *pk++;
                *pl++ = pj[1] ^ *pk++;
                *pl++ = pj[2] ^ *pk++;
                *pl++ = pj[3] ^ *pk++;
            }
        }
    }
}


void compile_4k_table64(unsigned __int8 *a, GfCtx4k64 *ctx)
{   int i, j, k;

    memset(ctx->gf_t4k, 0, sizeof(ctx->gf_t4k));
    for(i = 0; i < 2 * CBLK_LEN8; ++i)
    {
        if(i == 0)
        {
            memcpy(ctx->gf_t4k[1][8], a, CBLK_LEN8);
            for(j = 4; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t4k[1][j], ctx->gf_t4k[1][j + j], CBLK_LEN8);
                mul_x64(ctx->gf_t4k[1][j]);
            }
            memcpy(ctx->gf_t4k[0][8], ctx->gf_t4k[1][1], CBLK_LEN8);
            mul_x64(ctx->gf_t4k[0][8]);
            for(j = 4; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t4k[0][j], ctx->gf_t4k[0][j + j], CBLK_LEN8);
                mul_x64(ctx->gf_t4k[0][j]);
            }
        }
        else if(i > 1)
            for(j = 8; j > 0; j >>= 1)
            {
                memcpy(ctx->gf_t4k[i][j], ctx->gf_t4k[i - 2][j], CBLK_LEN8);
                mul_x8_64(ctx->gf_t4k[i][j]);
            }

        for(j = 2; j < 16; j += j)
        {
            mode(32t) *pj = ctx->gf_t4k[i][j];
            mode(32t) *pk = ctx->gf_t4k[i][1];
            mode(32t) *pl = ctx->gf_t4k[i][j + 1];

            for(k = 1; k < j; ++k)
            {
                *pl++ = pj[0] ^ *pk++;
                *pl++ = pj[1] ^ *pk++;
                *pl++ = pj[2] ^ *pk++;
                *pl++ = pj[3] ^ *pk++;
            }
        }
    }
}

static int IsBitSet128 (unsigned int bit, unsigned __int8 *a)
{
	return a[(127 - bit) / 8] & (0x80 >> ((127 - bit) % 8));
}

static int IsBitSet64 (unsigned int bit, unsigned __int8 *a)
{
	return a[(63 - bit) / 8] & (0x80 >> ((63 - bit) % 8));
}

static void SetBit128 (unsigned int bit, unsigned __int8 *a)
{
	a[(127 - bit) / 8] |= 0x80 >> ((127 - bit) % 8);
}

static void SetBit64 (unsigned int bit, unsigned __int8 *a)
{
	a[(63 - bit) / 8] |= 0x80 >> ((63 - bit) % 8);
}

void MirrorBits128 (unsigned __int8 *a)
{
	unsigned __int8 t[128 / 8];
	int i;
	memset (t,0,16);
	for (i = 0; i < 128; i++)
	{
		if (IsBitSet128(i, a))
			SetBit128 (127 - i, t);
	}
	memcpy (a, t, sizeof (t));
	burn (t,sizeof (t));
}

void MirrorBits64 (unsigned __int8 *a)
{
	unsigned __int8 t[64 / 8];
	int i;
	memset (t,0,8);
	for (i = 0; i < 64; i++)
	{
		if (IsBitSet64(i, a))
			SetBit64 (63 - i, t);
	}
	memcpy (a, t, sizeof (t));
	burn (t,sizeof (t));
}

/* Allocate and initialize speed optimization table
   for multiplication by 64-bit operand in MSB-first mode */
int Gf128Tab64Init (unsigned __int8 *a, GfCtx *ctx)
{
	GfCtx8k *ctx8k;
	unsigned __int8 am[16];
	int i, j;

	ctx8k = (GfCtx8k *) TCalloc (sizeof (GfCtx8k));
	if (!ctx8k)
		return FALSE;

	memcpy (am, a, 16);
	MirrorBits128 (am);
    compile_8k_table (am, ctx8k);

	/* Convert 8k LSB-first table to 4k MSB-first */
	for (i = 16; i < 32; i++) 
	{
		for (j = 0; j < 16; j++) 
		{
			int jm = 0;
			jm |= (j & 0x1) << 3;
			jm |= (j & 0x2) << 1;
			jm |= (j & 0x4) >> 1;
			jm |= (j & 0x8) >> 3;

			memcpy (&ctx->gf_t128[i-16][jm], (unsigned char *)&ctx8k->gf_t8k[31-i][j], 16);
			MirrorBits128 ((unsigned char *)&ctx->gf_t128[i-16][jm]);
		}
	}

	burn (ctx8k ,sizeof (*ctx8k));
	burn (am, sizeof (am));
	TCfree (ctx8k);
	return TRUE;
}

int Gf64TabInit (unsigned __int8 *a, GfCtx *ctx)
{
	/* Deprecated/legacy */

	GfCtx4k64 *ctx4k;
	unsigned __int8 am[8];
	int i, j;

	ctx4k = (GfCtx4k64 *) TCalloc (sizeof (GfCtx4k64));
	if (!ctx4k)
		return FALSE;

	memcpy (am, a, 8);
	MirrorBits64 (am);
    compile_4k_table64 (am, ctx4k);

	/* Convert LSB-first table to MSB-first */
	for (i = 0; i < 16; i++) 
	{
		for (j = 0; j < 16; j++) 
		{
			int jm = 0;
			jm |= (j & 0x1) << 3;
			jm |= (j & 0x2) << 1;
			jm |= (j & 0x4) >> 1;
			jm |= (j & 0x8) >> 3;

			memcpy (&ctx->gf_t64[i][jm], (unsigned char *)&ctx4k->gf_t4k[15-i][j], 8);
			MirrorBits64 ((unsigned char *)&ctx->gf_t64[i][jm]);
		}
	}

	burn (ctx4k,sizeof (*ctx4k));
	burn (am, sizeof (am));
	TCfree (ctx4k);
	return TRUE;
}

#define xor_8kt64(i)   \
    xor_block_aligned(r, ctx->gf_t128[i + i][a[i] & 15]); \
    xor_block_aligned(r, ctx->gf_t128[i + i + 1][a[i] >> 4])

/* Multiply a 128-bit number by a 64-bit number in the finite field GF(2^128) */
void Gf128MulBy64Tab (unsigned __int8 a[8], unsigned __int8 p[16], GfCtx *ctx)
{  
	unsigned __int32 r[CBLK_LEN >> 2];

	move_block_aligned(r, ctx->gf_t128[7*2][a[7] & 15]);
    xor_block_aligned(r,  ctx->gf_t128[7*2+1][a[7] >> 4]);

	if (*(unsigned __int16 *)a)
	{
		xor_8kt64(0);
		xor_8kt64(1);
	}
	if (a[2])
	{
		xor_8kt64(2);
	}
	xor_8kt64(3);
    xor_8kt64(4);
	xor_8kt64(5);
	xor_8kt64(6);

    move_block_aligned(p, r);
}

#define xor_8k64(i)   \
    xor_block_aligned64(r, ctx->gf_t64[i + i][a[i] & 15]); \
    xor_block_aligned64(r, ctx->gf_t64[i + i + 1][a[i] >> 4])

/* Multiply two 64-bit numbers in the finite field GF(2^64) */
void Gf64MulTab (unsigned char a[8], unsigned char p[8], GfCtx *ctx)
{  
	/* Deprecated/legacy */

	unsigned __int32 r[CBLK_LEN8 >> 2];

	move_block_aligned64(r, ctx->gf_t64[7*2][a[7] & 15]);
    xor_block_aligned64(r,  ctx->gf_t64[7*2+1][a[7] >> 4]);

	if (*(unsigned __int16 *)a)
	{
		xor_8k64(0);
		xor_8k64(1);
	}
	if (a[2])
	{
		xor_8k64(2);
	}
	xor_8k64(3);
    xor_8k64(4);
	xor_8k64(5);
	xor_8k64(6);

    move_block_aligned64(p, r);
}


/* Basic algorithms for testing of optimized algorithms */

static void xor128 (unsigned __int64 *a, unsigned __int64 *b)
{
	*a++ ^= *b++;
	*a ^= *b;
}

static void shl128 (unsigned __int8 *a)
{
	int i, x = 0, xx;
	for (i = 15; i >= 0; i--)
	{
		xx = (a[i] & 0x80) >> 7;
		a[i] = (a[i] << 1) | x;
		x = xx;
	}
}

static void GfMul128Basic (unsigned __int8 *a, unsigned __int8 *b, unsigned __int8 *p)
{
	int i;
	unsigned __int8 la[16];
	memcpy (la, a, 16);
	memset (p, 0, 16);

	for (i = 0; i < 128; i++)
	{
		if (IsBitSet128 (i, b))
			xor128 ((unsigned __int64 *)p, (unsigned __int64 *)la);

		if (la[0] & 0x80)
		{
			shl128 (la);
			la[15] ^= 0x87;
		}
		else
		{
			shl128 (la);
		}
	}
}

static void xor64 (unsigned __int64 *a, unsigned __int64 *b)
{
	*a ^= *b;
}

static void shl64 (unsigned __int8 *a)
{
	int i, x = 0, xx;
	for (i = 7; i >= 0; i--)
	{
		xx = (a[i] & 0x80) >> 7;
		a[i] = (a[i] << 1) | x;
		x = xx;
	}
}

static void GfMul64Basic (unsigned __int8 *a, unsigned __int8 *b, unsigned __int8* p)
{
	/* Deprecated/legacy */

	int i;
	unsigned __int8 la[8];
	memcpy (la, a, 8);
	memset (p, 0, 8);

	for (i = 0; i < 64; i++)
	{
		if (IsBitSet64 (i, b))
			xor64 ((unsigned __int64 *)p, (unsigned __int64 *)la);

		if (la[0] & 0x80)
		{
			shl64 (la);
			la[7] ^= 0x1b;
		}
		else
		{
			shl64 (la);
		}
	}
}


BOOL GfMulSelfTest ()
{
	BOOL result = TRUE;
	unsigned __int8 a[16];
	unsigned __int8 b[16];
	unsigned __int8 p1[16];
	unsigned __int8 p2[16];
	GfCtx *gfCtx = (GfCtx *) TCalloc (sizeof (GfCtx));
	int i, j;

	if (!gfCtx)
		return FALSE;

	/* GF(2^64) - deprecated/legacy */
	for (i = 0; i < 0x100; i++)
	{
		for (j = 0; j < 8; j++)
		{
			a[j] = (unsigned __int8) i;
			b[j] = a[j] ^ 0xff;
		}

		GfMul64Basic (a, b, p1);
	
		Gf64TabInit (a, gfCtx);
		Gf64MulTab (b, p2, gfCtx);

		if (memcmp (p1, p2, 8) != 0)
			result = FALSE;
	}

	/* GF(2^128) */
	for (i = 0; i < 0x100; i++)
	{
		for (j = 0; j < 16; j++)
		{
			a[j] = (unsigned __int8) i;
			b[j] = j < 8 ? 0 : a[j] ^ 0xff;
		}

		GfMul128Basic (a, b, p1);
	
		Gf128Tab64Init (a, gfCtx);
		Gf128MulBy64Tab (b + 8, p2, gfCtx);

		if (memcmp (p1, p2, 16) != 0)
			result = FALSE;
	}

	TCfree (gfCtx);
	return result;
}

#if defined(__cplusplus)
}
#endif