Re: 16-bit vs. 32-bit code segment check -- wanting comments good and bad...

From: Bx.C (invalid-email-address_at_invalid.shiragajin)
Date: 03/26/04 

Date: Fri, 26 Mar 2004 13:19:28 +0000 (UTC)

"Tim Roberts" <timr@probo.com> wrote in message
news:4pl760hne3gqmk975a51vmlllnchdpng6h@4ax.com...
> "Bx.C" <invalid-email-address@invalid.shiragajin> wrote:
> >
> >binary 16-bit 32-bit
> >--------------------------------
> >55 push bp push ebp
> >89 E5 mov bp,sp mov ebp,esp
> >55 push bp push ebp
> >29 E5 sub bp,sp sub ebp,esp
> >01 EC add sp,bp add esp,ebp
> >83 FD 04 cmp bp,+4 cmp ebp,+4
> >19 C0 sbb ax,ax sbb ax,ax
> >5D pop bp pop ebp
> >74 xx jz cs32b jz cs32b
>
> Here's a shorter one:
>
> b8 00 00 fe c0 mov ax,0 mov eax,0c0fe0000h
> inc al
> 84 c0 test al,al test al,al
> 74 xx jz cs16b jz cs16b
>
> >i think people making debuggers to work in real-mode and protected-mode
> >might like this code snippet...
>
> We used to use something like this in the Cinepak codec in Windows 3.1.
> Windows 3.1 had the ability to load 32-bit code segments from a
> protect-mode EXE, but it would load them as 16-bit segments. We'd use
that
> kind of code to decide whether we needed to alter the segment descriptor.
> --
> - Tim Roberts, timr@probo.com
> Providenza & Boekelheide, Inc.
>
>

hehe... i KNEW it was useful for something...

how about this even shorter version that Herbert Kleebauer came up w/ in
alt.lang.asm last night... this solution looks much better...

(good thing he supplied his binary, i still can't make heads or tails of the
mnemonics he uses)

------------------------------------
b1 01 mov al,001h mov al,001h
c1 e0 10 shl ax,010h shl eax,010h
74 xx jz cs16b jz cs16b
------------------------------------

jumps if in 16-bit... doesn't if in 32-bit...

Relevant Pages

  • Re: Using inline asm
    ... stackframe, say instead of push ebp; pop ebp, or used some other ... That might mess up returns... ... The intention of the asm is to tell gcc exactly what the opcode does, ...
    (comp.os.msdos.djgpp)
  • Re: Startup routine initialization
    ... Shorter than the explicit instructions. ... At one time, "enter" was shorter, but slower, than the explicit instructions, but "leave" was shorter and equally fast... ... sub esp, ??? ...
    (alt.lang.asm)