That might be too slow. A lookup table is appropriate here.

   mov  r4,#6
   call setb_ind
   ...
setb_ind:
   xch  a,r4
   push acc
   add  a,#setb_ind_T-setb_ind_X
   movc a,@a+pc
setb_ind_X:
   orl  a,r4
   mov  r4,a
   pop  acc
   xch  a,r4
   ret

setb_ind_T:  
   db   1,2,4,8,16,32,64,128

Hmmmm... Does not seem too fast either :-)
Maybe a smartly devised algo would do...

   mov  r4,#6
   call setb_ind
   ...
setb_ind:
   push b
   mov  b,r4
   jb   b.2,setb_in_1xx
   jb   b.1,setb_in_01x
   jb   b.0,setb_in_001
   orl  a,#1
   pop  b
   ret
setb_in_001:
   orl  a,#2
   pop  b
   ret
setb_in_01x:
   jb   b.0,setb_in_011
   orl  a,#4
   pop  b
   ret
setb_in_011:
   orl  a,#8
   pop  b
   ret
... etc.etc.

Or a jumptable.

Code memory/FLASH is cheap today.

You can even have toggle indirect (or call it XOR if you want).


Jan Waclawek


[edit]Oh, @#$%^, now I realized what did you mean by counter... and shift register... yep, that's the smartest way; do the hard way only if r4 can be an arbitrary value, of course...[/edit]