Assembly Converting MOV / MOVZX and MOVSX to C code (no inline asm)

Question

For the asm emulator i'm trying to write to convert ASM code to equivalent working code just working.. best code would be the one that can either be done in one line or two-three the most, don't care about speed.

From my understanding. MOVZX would be the same as MOV.. if done in C++.

MOV conversion.

MOV ESI,DWORD PTR [ESP+8]

would be like

regs.d.esi = *(unsigned int *)(regs.d.esp+0x00000008);

MOVZX conversion.

MOVZX EAX,BYTE PTR DS:[EDI]

would be like

regs.d.eax = *(unsigned char *)(regs.d.edi);

pretty much the same thing no change what so ever.

Now MOVSX i'm having trouble converting to a simple C code.. seems to be the same as the two above.. except it attempts to append as much fully set bits in front of the value moved as possible.. like

000000C7 becomes FFFFFFC7

Solution

movsx is move with sign-extend. Those set bits are a copy of the sign bit from the original value, and would be clear if the original wasn't negative. It works just like your other conversions, except you need to use a signed type instead of an unsigned one.

regs.d.eax = *(signed char *)(regs.d.edi); // movsx eax, byte ptr ds:[edi]

Comments

[SSpoke]

How about.. MOVSX EAX,AL is this correct? regs.d.eax = (signed char)(regs.h.al);

[ughoavgfhw]

@SSpoke You need to be careful about casting between different integer types, but I'm pretty sure casting between signed and unsigned types of the same size is safe. (You don't need to worry about the implicit cast to the type of regs.d.eax, since that is the actual movsx)

[SSpoke]

al = unsigned char although. still not safe? it's already all applied.

[ughoavgfhw]

@SSpoke No, casting between types of the same size is safe (At least it worked right in my quick tests. Maybe someone can quote the standard?). You could have run into problems if al was some other size.

[SSpoke]

thanks thats what I wanted to know it will never be another size. union _regs { struct { unsigned int eax, ebx, ecx, edx, esi, edi, ebp, esp; } d; struct { unsigned short ax, hax, bx, hbx, cx, hcx, dx, hdx, si, hsi, di, hdi, bp, hbp, sp, hsp; } x; struct { unsigned char al, ah, hax[2], bl, bh, hbx[2], cl, ch, hcx[2], dl, dh, hdx[2], rest[16]; } h; } regs;