fix compile for x64, DLL and VC 6

1 files changed, 68 insertions, 69 deletions

diff --git a/panama.cpp b/panama.cpp
index 89a5aeaa..a60e1670 100644
--- a/panama.cpp
+++ b/panama.cpp
@@ -26,31 +26,31 @@ void Panama_SSE2_Pull(size_t count, word32 *state, word32 *z, const word32 *y)
 	__asm__ __volatile__
 	(
 		".intel_syntax noprefix;"
-	AS1(	push	ebx)
+	AS_PUSH(		bx)
 #else
-	AS2(	mov		ecx, count)
-	AS2(	mov		esi, state)
-	AS2(	mov		edi, z)
-	AS2(	mov		edx, y)
+	AS2(	mov		WORD_REG(cx), count)
+	AS2(	mov		WORD_REG(si), state)
+	AS2(	mov		WORD_REG(di), z)
+	AS2(	mov		WORD_REG(dx), y)
 #endif
-	AS2(	shl		ecx, 5)
+	AS2(	shl		WORD_REG(cx), 5)
 	ASJ(	jz,		5, f)
-	AS2(	mov		ebx, [esi+4*17])
-	AS2(	add		ecx, ebx)
+	AS2(	mov		ebx, [WORD_REG(si)+4*17])
+	AS2(	add		WORD_REG(cx), WORD_REG(bx))
 
-	AS1(	push	ebp)
-	AS1(	push	ecx)
+	AS_PUSH(		bp)
+	AS_PUSH(		cx)
 
-	AS2(	movdqa	xmm0, [esi+0*16])
-	AS2(	movdqa	xmm1, [esi+1*16])
-	AS2(	movdqa	xmm2, [esi+2*16])
-	AS2(	movdqa	xmm3, [esi+3*16])
-	AS2(	mov		eax, [esi+4*16])
+	AS2(	movdqa	xmm0, [WORD_REG(si)+0*16])
+	AS2(	movdqa	xmm1, [WORD_REG(si)+1*16])
+	AS2(	movdqa	xmm2, [WORD_REG(si)+2*16])
+	AS2(	movdqa	xmm3, [WORD_REG(si)+3*16])
+	AS2(	mov		eax, [WORD_REG(si)+4*16])
 
 	ASL(4)
 	// gamma and pi
 #if CRYPTOPP_BOOL_SSSE3_ASM_AVAILABLE
-	AS2(	test	ebx, 1)
+	AS2(	test	WORD_REG(bx), 1)
 	ASJ(	jnz,	6, f)
 #endif
 	AS2(	movdqa	xmm6, xmm2)
@@ -81,7 +81,7 @@ void Panama_SSE2_Pull(size_t count, word32 *state, word32 *z, const word32 *y)
 #define pi(i)	\
 	AS2(	movd	ecx, xmm7)\
 	AS2(	rol		ecx, ASM_MOD((ASM_MOD(5*i,17)*(ASM_MOD(5*i,17)+1)/2), 32))\
-	AS2(	mov		[esi+SSE2_Index(ASM_MOD(5*(i), 17))*4], ecx)
+	AS2(	mov		[WORD_REG(si)+SSE2_Index(ASM_MOD(5*(i), 17))*4], ecx)
 
 #define pi4(x, y, z, a, b, c, d)	\
 	AS2(	pcmpeqb	xmm7, xmm7)\
@@ -110,65 +110,65 @@ void Panama_SSE2_Pull(size_t count, word32 *state, word32 *z, const word32 *y)
 	AS2(	punpckhdq	xmm2, xmm0)		// 11 12 15 16
 
 	// keystream
-	AS2(	test	edi, edi)
+	AS2(	test	WORD_REG(di), WORD_REG(di))
 	ASJ(	jz,		0, f)
 	AS2(	movdqa	xmm6, xmm4)
 	AS2(	punpcklqdq	xmm4, xmm2)
 	AS2(	punpckhqdq	xmm6, xmm2)
-	AS2(	test	edx, 0xf)
+	AS2(	test	WORD_REG(dx), 0xf)
 	ASJ(	jnz,	2, f)
-	AS2(	test	edx, edx)
+	AS2(	test	WORD_REG(dx), WORD_REG(dx))
 	ASJ(	jz,		1, f)
-	AS2(	pxor	xmm4, [edx])
-	AS2(	pxor	xmm6, [edx+16])
-	AS2(	add		edx, 32)
+	AS2(	pxor	xmm4, [WORD_REG(dx)])
+	AS2(	pxor	xmm6, [WORD_REG(dx)+16])
+	AS2(	add		WORD_REG(dx), 32)
 	ASJ(	jmp,	1, f)
 	ASL(2)
-	AS2(	movdqu	xmm0, [edx])
-	AS2(	movdqu	xmm2, [edx+16])
+	AS2(	movdqu	xmm0, [WORD_REG(dx)])
+	AS2(	movdqu	xmm2, [WORD_REG(dx)+16])
 	AS2(	pxor	xmm4, xmm0)
 	AS2(	pxor	xmm6, xmm2)
-	AS2(	add		edx, 32)
+	AS2(	add		WORD_REG(dx), 32)
 	ASL(1)
-	AS2(	test	edi, 0xf)
+	AS2(	test	WORD_REG(di), 0xf)
 	ASJ(	jnz,	3, f)
-	AS2(	movdqa	[edi], xmm4)
-	AS2(	movdqa	[edi+16], xmm6)
-	AS2(	add		edi, 32)
+	AS2(	movdqa	[WORD_REG(di)], xmm4)
+	AS2(	movdqa	[WORD_REG(di)+16], xmm6)
+	AS2(	add		WORD_REG(di), 32)
 	ASJ(	jmp,	0, f)
 	ASL(3)
-	AS2(	movdqu	[edi], xmm4)
-	AS2(	movdqu	[edi+16], xmm6)
-	AS2(	add		edi, 32)
+	AS2(	movdqu	[WORD_REG(di)], xmm4)
+	AS2(	movdqu	[WORD_REG(di)+16], xmm6)
+	AS2(	add		WORD_REG(di), 32)
 	ASL(0)
 
 	// buffer update
-	AS2(	lea		ecx, [ebx + 32])
-	AS2(	and		ecx, 31*32)
-	AS2(	lea		ebp, [ebx + (32-24)*32])
-	AS2(	and		ebp, 31*32)
+	AS2(	lea		WORD_REG(cx), [WORD_REG(bx) + 32])
+	AS2(	and		WORD_REG(cx), 31*32)
+	AS2(	lea		WORD_REG(bp), [WORD_REG(bx) + (32-24)*32])
+	AS2(	and		WORD_REG(bp), 31*32)
 
-	AS2(	movdqa	xmm0, [esi+20*4+ecx+0*8])
+	AS2(	movdqa	xmm0, [WORD_REG(si)+20*4+WORD_REG(cx)+0*8])
 	AS2(	pxor	xmm3, xmm0)
 	ASS(	pshufd	xmm0, xmm0, 2, 3, 0, 1)
-	AS2(	movdqa	[esi+20*4+ecx+0*8], xmm3)
-	AS2(	pxor	xmm0, [esi+20*4+ebp+2*8])
-	AS2(	movdqa	[esi+20*4+ebp+2*8], xmm0)
+	AS2(	movdqa	[WORD_REG(si)+20*4+WORD_REG(cx)+0*8], xmm3)
+	AS2(	pxor	xmm0, [WORD_REG(si)+20*4+WORD_REG(bp)+2*8])
+	AS2(	movdqa	[WORD_REG(si)+20*4+WORD_REG(bp)+2*8], xmm0)
 
-	AS2(	movdqa	xmm4, [esi+20*4+ecx+2*8])
+	AS2(	movdqa	xmm4, [WORD_REG(si)+20*4+WORD_REG(cx)+2*8])
 	AS2(	pxor	xmm1, xmm4)
-	AS2(	movdqa	[esi+20*4+ecx+2*8], xmm1)
-	AS2(	pxor	xmm4, [esi+20*4+ebp+0*8])
-	AS2(	movdqa	[esi+20*4+ebp+0*8], xmm4)
+	AS2(	movdqa	[WORD_REG(si)+20*4+WORD_REG(cx)+2*8], xmm1)
+	AS2(	pxor	xmm4, [WORD_REG(si)+20*4+WORD_REG(bp)+0*8])
+	AS2(	movdqa	[WORD_REG(si)+20*4+WORD_REG(bp)+0*8], xmm4)
 
 	// theta
-	AS2(	movdqa	xmm3, [esi+3*16])
-	AS2(	movdqa	xmm2, [esi+2*16])
-	AS2(	movdqa	xmm1, [esi+1*16])
-	AS2(	movdqa	xmm0, [esi+0*16])
+	AS2(	movdqa	xmm3, [WORD_REG(si)+3*16])
+	AS2(	movdqa	xmm2, [WORD_REG(si)+2*16])
+	AS2(	movdqa	xmm1, [WORD_REG(si)+1*16])
+	AS2(	movdqa	xmm0, [WORD_REG(si)+0*16])
 
 #if CRYPTOPP_BOOL_SSSE3_ASM_AVAILABLE
-	AS2(	test	ebx, 1)
+	AS2(	test	WORD_REG(bx), 1)
 	ASJ(	jnz,	8, f)
 #endif
 	AS2(	movd	xmm6, eax)
@@ -214,21 +214,21 @@ void Panama_SSE2_Pull(size_t count, word32 *state, word32 *z, const word32 *y)
 	AS2(	pxor	xmm0, xmm4)
 
 	// sigma
-	AS2(	lea		ecx, [ebx + (32-4)*32])
-	AS2(	and		ecx, 31*32)
-	AS2(	lea		ebp, [ebx + 16*32])
-	AS2(	and		ebp, 31*32)
+	AS2(	lea		WORD_REG(cx), [WORD_REG(bx) + (32-4)*32])
+	AS2(	and		WORD_REG(cx), 31*32)
+	AS2(	lea		WORD_REG(bp), [WORD_REG(bx) + 16*32])
+	AS2(	and		WORD_REG(bp), 31*32)
 
-	AS2(	movdqa	xmm4, [esi+20*4+ecx+0*16])
-	AS2(	movdqa	xmm5, [esi+20*4+ebp+0*16])
+	AS2(	movdqa	xmm4, [WORD_REG(si)+20*4+WORD_REG(cx)+0*16])
+	AS2(	movdqa	xmm5, [WORD_REG(si)+20*4+WORD_REG(bp)+0*16])
 	AS2(	movdqa	xmm6, xmm4)
 	AS2(	punpcklqdq	xmm4, xmm5)
 	AS2(	punpckhqdq	xmm6, xmm5)
 	AS2(	pxor	xmm3, xmm4)
 	AS2(	pxor	xmm2, xmm6)
 
-	AS2(	movdqa	xmm4, [esi+20*4+ecx+1*16])
-	AS2(	movdqa	xmm5, [esi+20*4+ebp+1*16])
+	AS2(	movdqa	xmm4, [WORD_REG(si)+20*4+WORD_REG(cx)+1*16])
+	AS2(	movdqa	xmm5, [WORD_REG(si)+20*4+WORD_REG(bp)+1*16])
 	AS2(	movdqa	xmm6, xmm4)
 	AS2(	punpcklqdq	xmm4, xmm5)
 	AS2(	punpckhqdq	xmm6, xmm5)
@@ -236,23 +236,22 @@ void Panama_SSE2_Pull(size_t count, word32 *state, word32 *z, const word32 *y)
 	AS2(	pxor	xmm0, xmm6)
 
 	// loop
-	AS2(	add		ebx, 32)
-	AS2(	cmp		ebx, [esp])
+	AS2(	add		WORD_REG(bx), 32)
+	AS2(	cmp		WORD_REG(bx), [WORD_REG(sp)])
 	ASJ(	jne,	4, b)
 
 	// save state
-	AS2(	mov		ebp, [esp+4])
-	AS2(	add		esp, 8)
-	AS2(	mov		[esi+4*17], ebx)
-	AS2(	mov		[esi+4*16], eax)
-	AS2(	movdqa	[esi+3*16], xmm3)
-	AS2(	movdqa	[esi+2*16], xmm2)
-	AS2(	movdqa	[esi+1*16], xmm1)
-	AS2(	movdqa	[esi+0*16], xmm0)
+	AS2(	add		WORD_REG(sp), WORD_SZ)
+	AS_POP(			bp)
+	AS2(	mov		[WORD_REG(si)+4*16], eax)
+	AS2(	movdqa	[WORD_REG(si)+3*16], xmm3)
+	AS2(	movdqa	[WORD_REG(si)+2*16], xmm2)
+	AS2(	movdqa	[WORD_REG(si)+1*16], xmm1)
+	AS2(	movdqa	[WORD_REG(si)+0*16], xmm0)
 	ASL(5)
 
 #ifdef __GNUC__
-	AS1(	pop		ebx)
+	AS_POP(			bx)
 	".att_syntax prefix;"
 		:
 		: "c" (count), "S" (state), "D" (z), "d" (y)