From: james.harris.1@gmail.com   
      
   On 20/07/2021 11:29, muta...@gmail.com wrote:   
   > On Monday, July 19, 2021 at 4:57:35 AM UTC+10, James Harris wrote:   
   >   
   >> The difference is not always as simple as an 0x66 prefix byte. For   
   >   
   > The 0x66 goes away completely simply by setting the   
   > D bit appropriately in the descriptor.   
   >   
   >> example, x86-16 and x86-32 can encode memory addresses in completely   
   >> different ways. Consider the SIB byte encoding for 16-bit and 32-bit   
   >> modes at   
   >>   
   >> [BX] in 16-bit mode   
   >> [EDI] in 32-bit mode   
   >   
   > Does this issue also go away if the D bit is set appropriately   
   > in the descriptor table?   
   >   
   > If not, the instruction needs to be eliminated from the C   
   > compiler's repertoire to provide a lowest common   
   > denominator.   
      
   I don't know but it sounds as though you are planning to assume that   
   everything will be OK until you hear otherwise....! Ughhhh!   
      
   If you are really set on getting 8086 and 80386 to run the exact same   
   binaries (which IMO is a bad idea for reasons set out in other posts)   
   wouldn't it be better to find out which instructions have the same   
   binary coding in each mode and only to use those?   
      
   >   
   > And my question is - does that LCD give you a Turing machine?   
   >   
   > If so, I finally know how to build my 8086 executables.   
      
      
      
      
   --   
   James Harris   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)