On 8/21/06, malc <malc@pulsesoft.com> wrote:
>
> On Mon, 21 Aug 2006, Igor Kovalenko wrote:
>
> > On 8/21/06, malc <malc@pulsesoft.com> wrote:
> >>
> >> On Mon, 21 Aug 2006, Igor Kovalenko wrote:
> >>
> >> <snip>
> >>
> >> > Right. Here is the real one with correct case labels.
> >>
> >> Sorry, forgot to mention that atop of the hex/bin problem the code also
> >> suffers from lack of proper 16/64 bit modrm decoding.
> >
> >
> > seems to me that 16 and 64 bit cases does not apply here
>
> Well, quick experiment in vm86 environment suggests that i'm right at
> least for 16bit case.
>
> <moo.asm>
>          xor ax, ax
>          int 16h
>          db 0xf, 0x19, 5, 0xde, 0xad
>          db 0xf, 0x19, 5, 0xbe, 0xef
>          mov ah, 9
>          mov dx, erm
>          add dx, 0x100
>          int 21h
>          xor ax, ax
>          int 16h
>          ret
> erm     db     "moo$"
> </moo.asm>
>
> nasm -o moo.com moo.asm
> dosemu moo.com
>
> 64bit case should be investigated by someone who possesses 64bit OS.
>

Here is the updated patch, should implement 16/32/64 modes according to
public intel docs.
Operand size is taken from DisassContext->dflag which is set beforehand; I
assume it is decoded correctly wrt appropriate instruction prefixes.

-- 
Kind Regards,
Igor V. Kovalenko