From: rick.c.hodgin@nospicedham.gmail.com   
      
   On Friday, September 15, 2017 at 5:46:11 PM UTC-4, patric...@nos   
   icedham.gmail.com wrote:   
   > In 8086 Assembly does "mov ax, [es:bx + di]" specify that "di" is relative   
   > to the es segment or ds segment?   
   >   
   > In other words, is bx relative to es and di relative to es (same override),   
   >                 or is bx relative to es and di relative to ds (only one   
   override)?   
   >   
   > Thanks in advance,   
   >   
   > Patrick   
      
   On the 8086 (and later 80386 and beyond), by default there are natural   
   relationships between certain registers and data segments.   
      
   sp, bp     -- Always relate to ss: by default.   
      
   di, si, bx -- Always relate to ds: by default, or es: if there are   
                 two operations, such as movsb.   
      
   ip         -- Always relates to cs: and cannot be overridden.  There   
                 is a POP CS instruction, and a JMP FAR and CALL FAR   
                 instructions, but these still load CS with the new value.   
      
   So whenever you use an override, you're overriding the default segment   
   so that all of the memory operands in the rest of the instruction relate   
   to it.   
      
   In your example:   
      
       mov ax, [es:bx + di]   
      
   By default, di and bx fall into the default ds: segment, so with the   
   override you are specifying, "Do not use ds:, but instead use es:, for   
   this one instruction."   
      
   If you needed to do more than one instruction, you can save some space   
   by doing this:   
      
       push    ds       ; Save ds for later use   
       push    es       ; Load es onto stack   
       pop     ds       ; Get stack value into ds   
       ; Note:  This is effectively the instruction "mov ds,es" which   
       ;        does not exist on the 8086.   
      
       ; Do your multiple instructions here   
      
       pop     ds       ; Restore the original saved ds above   
      
   Segment overrides consume one additional byte of opcode space as   
   well, and they affect performance somewhat, so you have to gauge   
   when to use them, when to not, etc.  This often involves quite a   
   bit of experimentation with different formats on different machines   
   to see which works best.   
      
   This syntax:   
      
       mov ax, [es:bx + di]   
      
   Is more often seen this way in my experience, with the index register   
   shown first:   
      
       mov ax, es:[di + bx]   
      
   It also makes it more clear that the entire memory location of [di+bx]   
   is being affected by the es: override.   
      
   Thank you,   
   Rick C. Hodgin   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)