sha.cpp

AlgorithmType: SymmetricCipher Name: AES/ECB Source: NIST Special Publication 800-38A Plaintext:

	W64LIT(0x0fc19dc68b8cd5b5), W64LIT(0x240ca1cc77ac9c65),
	W64LIT(0x2de92c6f592b0275), W64LIT(0x4a7484aa6ea6e483),
	W64LIT(0x5cb0a9dcbd41fbd4), W64LIT(0x76f988da831153b5),
	W64LIT(0x983e5152ee66dfab), W64LIT(0xa831c66d2db43210),
	W64LIT(0xb00327c898fb213f), W64LIT(0xbf597fc7beef0ee4),
	W64LIT(0xc6e00bf33da88fc2), W64LIT(0xd5a79147930aa725),
	W64LIT(0x06ca6351e003826f), W64LIT(0x142929670a0e6e70),
	W64LIT(0x27b70a8546d22ffc), W64LIT(0x2e1b21385c26c926),
	W64LIT(0x4d2c6dfc5ac42aed), W64LIT(0x53380d139d95b3df),
	W64LIT(0x650a73548baf63de), W64LIT(0x766a0abb3c77b2a8),
	W64LIT(0x81c2c92e47edaee6), W64LIT(0x92722c851482353b),
	W64LIT(0xa2bfe8a14cf10364), W64LIT(0xa81a664bbc423001),
	W64LIT(0xc24b8b70d0f89791), W64LIT(0xc76c51a30654be30),
	W64LIT(0xd192e819d6ef5218), W64LIT(0xd69906245565a910),
	W64LIT(0xf40e35855771202a), W64LIT(0x106aa07032bbd1b8),
	W64LIT(0x19a4c116b8d2d0c8), W64LIT(0x1e376c085141ab53),
	W64LIT(0x2748774cdf8eeb99), W64LIT(0x34b0bcb5e19b48a8),
	W64LIT(0x391c0cb3c5c95a63), W64LIT(0x4ed8aa4ae3418acb),
	W64LIT(0x5b9cca4f7763e373), W64LIT(0x682e6ff3d6b2b8a3),
	W64LIT(0x748f82ee5defb2fc), W64LIT(0x78a5636f43172f60),
	W64LIT(0x84c87814a1f0ab72), W64LIT(0x8cc702081a6439ec),
	W64LIT(0x90befffa23631e28), W64LIT(0xa4506cebde82bde9),
	W64LIT(0xbef9a3f7b2c67915), W64LIT(0xc67178f2e372532b),
	W64LIT(0xca273eceea26619c), W64LIT(0xd186b8c721c0c207),
	W64LIT(0xeada7dd6cde0eb1e), W64LIT(0xf57d4f7fee6ed178),
	W64LIT(0x06f067aa72176fba), W64LIT(0x0a637dc5a2c898a6),
	W64LIT(0x113f9804bef90dae), W64LIT(0x1b710b35131c471b),
	W64LIT(0x28db77f523047d84), W64LIT(0x32caab7b40c72493),
	W64LIT(0x3c9ebe0a15c9bebc), W64LIT(0x431d67c49c100d4c),
	W64LIT(0x4cc5d4becb3e42b6), W64LIT(0x597f299cfc657e2a),
	W64LIT(0x5fcb6fab3ad6faec), W64LIT(0x6c44198c4a475817)
};

#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE && CRYPTOPP_BOOL_X86
// put assembly version in separate function, otherwise MSVC 2005 SP1 doesn't generate correct code for the non-assembly version
CRYPTOPP_NAKED static void CRYPTOPP_FASTCALL SHA512_SSE2_Transform(word64 *state, const word64 *data)
{
#ifdef __GNUC__
	__asm__ __volatile__
	(
		".intel_syntax noprefix;"
	AS1(	push	ebx)
	AS2(	mov		ebx, eax)
#else
	AS1(	push	ebx)
	AS1(	push	esi)
	AS1(	push	edi)
	AS2(	lea		ebx, SHA512_K)
#endif

	AS2(	mov		eax, esp)
	AS2(	and		esp, 0xfffffff0)
	AS2(	sub		esp, 27*16)				// 17*16 for expanded data, 20*8 for state
	AS1(	push	eax)
	AS2(	xor		eax, eax)
	AS2(	lea		edi, [esp+4+8*8])		// start at middle of state buffer. will decrement pointer each round to avoid copying
	AS2(	lea		esi, [esp+4+20*8+8])	// 16-byte alignment, then add 8

	AS2(	movq	mm4, [ecx+0*8])
	AS2(	movq	[edi+0*8], mm4)
	AS2(	movq	mm0, [ecx+1*8])
	AS2(	movq	[edi+1*8], mm0)
	AS2(	movq	mm0, [ecx+2*8])
	AS2(	movq	[edi+2*8], mm0)
	AS2(	movq	mm0, [ecx+3*8])
	AS2(	movq	[edi+3*8], mm0)
	AS2(	movq	mm5, [ecx+4*8])
	AS2(	movq	[edi+4*8], mm5)
	AS2(	movq	mm0, [ecx+5*8])
	AS2(	movq	[edi+5*8], mm0)
	AS2(	movq	mm0, [ecx+6*8])
	AS2(	movq	[edi+6*8], mm0)
	AS2(	movq	mm0, [ecx+7*8])
	AS2(	movq	[edi+7*8], mm0)
	ASJ(	jmp,	0, f)

#define SSE2_S0_S1(r, a, b, c)	\
	AS2(	movq	mm6, r)\
	AS2(	psrlq	r, a)\
	AS2(	movq	mm7, r)\
	AS2(	psllq	mm6, 64-c)\
	AS2(	pxor	mm7, mm6)\
	AS2(	psrlq	r, b-a)\
	AS2(	pxor	mm7, r)\
	AS2(	psllq	mm6, c-b)\
	AS2(	pxor	mm7, mm6)\
	AS2(	psrlq	r, c-b)\
	AS2(	pxor	r, mm7)\
	AS2(	psllq	mm6, b-a)\
	AS2(	pxor	r, mm6)

#define SSE2_s0(r, a, b, c)	\
	AS2(	movdqa	xmm6, r)\
	AS2(	psrlq	r, a)\
	AS2(	movdqa	xmm7, r)\
	AS2(	psllq	xmm6, 64-c)\
	AS2(	pxor	xmm7, xmm6)\
	AS2(	psrlq	r, b-a)\
	AS2(	pxor	xmm7, r)\
	AS2(	psrlq	r, c-b)\
	AS2(	pxor	r, xmm7)\
	AS2(	psllq	xmm6, c-a)\
	AS2(	pxor	r, xmm6)

#define SSE2_s1(r, a, b, c)	\
	AS2(	movdqa	xmm6, r)\
	AS2(	psrlq	r, a)\
	AS2(	movdqa	xmm7, r)\
	AS2(	psllq	xmm6, 64-c)\
	AS2(	pxor	xmm7, xmm6)\
	AS2(	psrlq	r, b-a)\
	AS2(	pxor	xmm7, r)\
	AS2(	psllq	xmm6, c-b)\
	AS2(	pxor	xmm7, xmm6)\
	AS2(	psrlq	r, c-b)\
	AS2(	pxor	r, xmm7)

	ASL(SHA512_Round)
	// k + w is in mm0, a is in mm4, e is in mm5
	AS2(	paddq	mm0, [edi+7*8])		// h
	AS2(	movq	mm2, [edi+5*8])		// f
	AS2(	movq	mm3, [edi+6*8])		// g
	AS2(	pxor	mm2, mm3)
	AS2(	pand	mm2, mm5)
	SSE2_S0_S1(mm5,14,18,41)
	AS2(	pxor	mm2, mm3)
	AS2(	paddq	mm0, mm2)			// h += Ch(e,f,g)
	AS2(	paddq	mm5, mm0)			// h += S1(e)
	AS2(	movq	mm2, [edi+1*8])		// b
	AS2(	movq	mm1, mm2)
	AS2(	por		mm2, mm4)
	AS2(	pand	mm2, [edi+2*8])		// c
	AS2(	pand	mm1, mm4)
	AS2(	por		mm1, mm2)
	AS2(	paddq	mm1, mm5)			// temp = h + Maj(a,b,c)
	AS2(	paddq	mm5, [edi+3*8])		// e = d + h
	AS2(	movq	[edi+3*8], mm5)
	AS2(	movq	[edi+11*8], mm5)
	SSE2_S0_S1(mm4,28,34,39)			// S0(a)
	AS2(	paddq	mm4, mm1)			// a = temp + S0(a)
	AS2(	movq	[edi-8], mm4)
	AS2(	movq	[edi+7*8], mm4)
	AS1(	ret)

	// first 16 rounds
	ASL(0)
	AS2(	movq	mm0, [edx+eax*8])
	AS2(	movq	[esi+eax*8], mm0)
	AS2(	movq	[esi+eax*8+16*8], mm0)
	AS2(	paddq	mm0, [ebx+eax*8])
	ASC(	call,	SHA512_Round)
	AS1(	inc		eax)
	AS2(	sub		edi, 8)
	AS2(	test	eax, 7)
	ASJ(	jnz,	0, b)
	AS2(	add		edi, 8*8)
	AS2(	cmp		eax, 16)
	ASJ(	jne,	0, b)

	// rest of the rounds
	AS2(	movdqu	xmm0, [esi+(16-2)*8])
	ASL(1)
	// data expansion, W[i-2] already in xmm0
	AS2(	movdqu	xmm3, [esi])
	AS2(	paddq	xmm3, [esi+(16-7)*8])
	AS2(	movdqa	xmm2, [esi+(16-15)*8])
	SSE2_s1(xmm0, 6, 19, 61)
	AS2(	paddq	xmm0, xmm3)
	SSE2_s0(xmm2, 1, 7, 8)
	AS2(	paddq	xmm0, xmm2)
	AS2(	movdq2q	mm0, xmm0)
	AS2(	movhlps	xmm1, xmm0)
	AS2(	paddq	mm0, [ebx+eax*8])
	AS2(	movlps	[esi], xmm0)
	AS2(	movlps	[esi+8], xmm1)
	AS2(	movlps	[esi+8*16], xmm0)
	AS2(	movlps	[esi+8*17], xmm1)
	// 2 rounds
	ASC(	call,	SHA512_Round)
	AS2(	sub		edi, 8)
	AS2(	movdq2q	mm0, xmm1)
	AS2(	paddq	mm0, [ebx+eax*8+8])
	ASC(	call,	SHA512_Round)
	// update indices and loop
	AS2(	add		esi, 16)
	AS2(	add		eax, 2)
	AS2(	sub		edi, 8)
	AS2(	test	eax, 7)
	ASJ(	jnz,	1, b)
	// do housekeeping every 8 rounds
	AS2(	mov		esi, 0xf)
	AS2(	and		esi, eax)
	AS2(	lea		esi, [esp+4+20*8+8+esi*8])
	AS2(	add		edi, 8*8)
	AS2(	cmp		eax, 80)
	ASJ(	jne,	1, b)

#define SSE2_CombineState(i)	\
	AS2(	movq	mm0, [edi+i*8])\
	AS2(	paddq	mm0, [ecx+i*8])\
	AS2(	movq	[ecx+i*8], mm0)

	SSE2_CombineState(0)
	SSE2_CombineState(1)
	SSE2_CombineState(2)
	SSE2_CombineState(3)
	SSE2_CombineState(4)
	SSE2_CombineState(5)
	SSE2_CombineState(6)
	SSE2_CombineState(7)

	AS1(	pop		esp)
	AS1(	emms)

#if defined(__GNUC__)
	AS1(	pop		ebx)
	".att_syntax prefix;"
		:
		: "a" (SHA512_K), "c" (state), "d" (data)
		: "%esi", "%edi", "memory", "cc"
	);
#else
	AS1(	pop		edi)
	AS1(	pop		esi)
	AS1(	pop		ebx)
	AS1(	ret)
#endif
}
#endif	// #if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE

void SHA512::Transform(word64 *state, const word64 *data)
{
#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE && CRYPTOPP_BOOL_X86
	if (HasSSE2())
	{
		SHA512_SSE2_Transform(state, data);
		return;
	}
#endif

#define S0(x) (rotrFixed(x,28)^rotrFixed(x,34)^rotrFixed(x,39))
#define S1(x) (rotrFixed(x,14)^rotrFixed(x,18)^rotrFixed(x,41))
#define s0(x) (rotrFixed(x,1)^rotrFixed(x,8)^(x>>7))
#define s1(x) (rotrFixed(x,19)^rotrFixed(x,61)^(x>>6))

#define R(i) h(i)+=S1(e(i))+Ch(e(i),f(i),g(i))+SHA512_K[i+j]+(j?blk2(i):blk0(i));\
	d(i)+=h(i);h(i)+=S0(a(i))+Maj(a(i),b(i),c(i))

	word64 W[16];
	word64 T[8];
    /* Copy context->state[] to working vars */
	memcpy(T, state, sizeof(T));
    /* 80 operations, partially loop unrolled */
	for (unsigned int j=0; j<80; j+=16)
	{
		R( 0); R( 1); R( 2); R( 3);
		R( 4); R( 5); R( 6); R( 7);
		R( 8); R( 9); R(10); R(11);
		R(12); R(13); R(14); R(15);
	}
    /* Add the working vars back into context.state[] */
    state[0] += a(0);
    state[1] += b(0);
    state[2] += c(0);
    state[3] += d(0);
    state[4] += e(0);
    state[5] += f(0);
    state[6] += g(0);
    state[7] += h(0);
}

#endif

NAMESPACE_END

#endif	// #ifndef CRYPTOPP_IMPORTS