Hello,
This is the tiger transform routine/procedure/method converted from the
Crypt++ 5.5.2 C/C++/Asm code to DCPCrypt2 Delphi/Basm code.
It works with optimizations turned off.
As soon as optimizations are turned on, it crashes with an access
violation
as soon as Index := 0; is executed.
I am not sure what the problem is because:
1. I didn't write the asm code.
2. I am not an asm expert.
I did my best converting it.
I think some possible causes might be:
1. Mis-aligned of byte boundaries ? (Maybe the C/C++/Asm had to do special
memory alignment ?)
2. Stack issue's ? (Not enough pushes or pops ?) (Framing issue?)
3. Compiler optimization bug ?
Who can tell for sure ?
procedure TDCPOptimizedTiger.Compress( Digest : Pint64; const X : Pint64
);
// procedure TDCPOptimizedTiger.Compress;
//var
// digest : pointer;
// LocalX: array[0..7] of int64;
// x : pointer;
begin
// Move(HashBuffer,LocalX,Sizeof(LocalX));
// Digest := @[EMAIL PROTECTED]
X := @[EMAIL PROTECTED]
asm
lea edx, [TigerTable]
mov eax, digest
mov esi, X
movq mm0, [eax]
movq mm1, [eax+1*8]
movq mm5, mm1
movq mm2, [eax+2*8]
movq mm7, [edx+4*2048+0*8]
movq mm6, [edx+4*2048+1*8]
mov ecx, esp
and esp, $fffffff0
sub esp, 8*8
push ecx
xor ebx, ebx
@[EMAIL PROTECTED]
pxor mm2, [esi+0*8+ebx]
movd ecx, mm2
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm2, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm2, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm0, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm1, mm4
movq mm3, mm1
psllq mm1, 2
paddq mm1, mm3
pxor mm0, [esi+1*8+ebx]
movd ecx, mm0
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm0, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm0, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm1, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm2, mm4
movq mm3, mm2
psllq mm2, 2
paddq mm2, mm3
cmp ebx, 6*8
je @[EMAIL PROTECTED]
pxor mm1, [esi+2*8+ebx]
movd ecx, mm1
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm1, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm1, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm2, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm0, mm4
movq mm3, mm0
psllq mm0, 2
paddq mm0, mm3
add ebx, 3*8
jmp @[EMAIL PROTECTED]
@[EMAIL PROTECTED]
movq mm3, [esi+7*8]
pxor mm3, mm6
movq mm4, [esi+0*8]
psubq mm4, mm3
movq [esp+4+0*8], mm4
pxor mm4, [esi+1*8]
movq mm3, mm4
movq [esp+4+1*8], mm4
paddq mm4, [esi+2*8]
pxor mm3, mm7
psllq mm3, 19
movq [esp+4+2*8], mm4
pxor mm3, mm4
movq mm4, [esi+3*8]
psubq mm4, mm3
movq [esp+4+3*8], mm4
pxor mm4, [esi+4*8]
movq mm3, mm4
movq [esp+4+4*8], mm4
paddq mm4, [esi+5*8]
pxor mm3, mm7
psrlq mm3, 23
movq [esp+4+5*8], mm4
pxor mm3, mm4
movq mm4, [esi+6*8]
psubq mm4, mm3
movq [esp+4+6*8], mm4
pxor mm4, [esi+7*8]
movq mm3, mm4
movq [esp+4+7*8], mm4
paddq mm4, [esp+4+0*8]
pxor mm3, mm7
psllq mm3, 19
movq [esp+4+0*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+1*8]
psubq mm4, mm3
movq [esp+4+1*8], mm4
pxor mm4, [esp+4+2*8]
movq mm3, mm4
movq [esp+4+2*8], mm4
paddq mm4, [esp+4+3*8]
pxor mm3, mm7
psrlq mm3, 23
movq [esp+4+3*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+4*8]
psubq mm4, mm3
movq [esp+4+4*8], mm4
pxor mm4, [esp+4+5*8]
movq [esp+4+5*8], mm4
paddq mm4, [esp+4+6*8]
movq [esp+4+6*8], mm4
pxor mm4, [edx+4*2048+2*8]
movq mm3, [esp+4+7*8]
psubq mm3, mm4
movq [esp+4+7*8], mm3
xor ebx, ebx
@[EMAIL PROTECTED]
pxor mm1, [esp+4+0*8+ebx]
movd ecx, mm1
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm1, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm1, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm2, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm0, mm4
movq mm3, mm0
psllq mm0, 3
psubq mm0, mm3
pxor mm2, [esp+4+1*8+ebx]
movd ecx, mm2
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm2, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm2, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm0, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm1, mm4
movq mm3, mm1
psllq mm1, 3
psubq mm1, mm3
cmp ebx, 6*8
je @[EMAIL PROTECTED]
pxor mm0, [esp+4+2*8+ebx]
movd ecx, mm0
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm0, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm0, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm1, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm2, mm4
movq mm3, mm2
psllq mm2, 3
psubq mm2, mm3
add ebx, 3*8
jmp @[EMAIL PROTECTED]
@[EMAIL PROTECTED]
movq mm3, [esp+4+7*8]
pxor mm3, mm6
movq mm4, [esp+4+0*8]
psubq mm4, mm3
movq [esp+4+0*8], mm4
pxor mm4, [esp+4+1*8]
movq mm3, mm4
movq [esp+4+1*8], mm4
paddq mm4, [esp+4+2*8]
pxor mm3, mm7
psllq mm3, 19
movq [esp+4+2*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+3*8]
psubq mm4, mm3
movq [esp+4+3*8], mm4
pxor mm4, [esp+4+4*8]
movq mm3, mm4
movq [esp+4+4*8], mm4
paddq mm4, [esp+4+5*8]
pxor mm3, mm7
psrlq mm3, 23
movq [esp+4+5*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+6*8]
psubq mm4, mm3
movq [esp+4+6*8], mm4
pxor mm4, [esp+4+7*8]
movq mm3, mm4
movq [esp+4+7*8], mm4
paddq mm4, [esp+4+0*8]
pxor mm3, mm7
psllq mm3, 19
movq [esp+4+0*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+1*8]
psubq mm4, mm3
movq [esp+4+1*8], mm4
pxor mm4, [esp+4+2*8]
movq mm3, mm4
movq [esp+4+2*8], mm4
paddq mm4, [esp+4+3*8]
pxor mm3, mm7
psrlq mm3, 23
movq [esp+4+3*8], mm4
pxor mm3, mm4
movq mm4, [esp+4+4*8]
psubq mm4, mm3
movq [esp+4+4*8], mm4
pxor mm4, [esp+4+5*8]
movq [esp+4+5*8], mm4
paddq mm4, [esp+4+6*8]
movq [esp+4+6*8], mm4
pxor mm4, [edx+4*2048+2*8]
movq mm3, [esp+4+7*8]
psubq mm3, mm4
movq [esp+4+7*8], mm3
xor ebx, ebx
@[EMAIL PROTECTED]
pxor mm0, [esp+4+0*8+ebx]
movd ecx, mm0
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm0, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm0, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm1, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm2, mm4
movq mm3, mm2
psllq mm2, 3
paddq mm2, mm3
pxor mm1, [esp+4+1*8+ebx]
movd ecx, mm1
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm1, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm1, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm2, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm0, mm4
movq mm3, mm0
psllq mm0, 3
paddq mm0, mm3
cmp ebx, 6*8
je @[EMAIL PROTECTED]
pxor mm2, [esp+4+2*8+ebx]
movd ecx, mm2
movzx edi, cl
movq mm3, [edx+0*2048+edi*8]
movzx edi, ch
movq mm4, [edx+3*2048+edi*8]
shr ecx, 16
movzx edi, cl
pxor mm3, [edx+1*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+2*2048+edi*8]
pextrw ecx, mm2, 2
movzx edi, cl
pxor mm3, [edx+2*2048+edi*8]
movzx edi, ch
pxor mm4, [edx+1*2048+edi*8]
pextrw ecx, mm2, 3
movzx edi, cl
pxor mm3, [edx+3*2048+edi*8]
psubq mm0, mm3
movzx edi, ch
pxor mm4, [edx+0*2048+edi*8]
paddq mm1, mm4
movq mm3, mm1
psllq mm1, 3
paddq mm1, mm3
add ebx, 3*8
jmp @[EMAIL PROTECTED]
@[EMAIL PROTECTED]
pxor mm0, [eax+0*8]
movq [eax+0*8], mm0
psubq mm1, mm5
movq [eax+1*8], mm1
paddq mm2, [eax+2*8]
movq [eax+2*8], mm2
pop esp
emms
end;
// CurrentHash[0]:= a xor aa;
// CurrentHash[1]:= b - bb;
// CurrentHash[2]:= c + cc;
Index:= 0;
FillChar(HashBuffer,Sizeof(HashBuffer),0);
end;
Please helpppppppp ?
Bye,
Skybuck.
|
