looking for smallest code to set only the 31st bit of a 32 bit reg

[Ratch]

ThoughtCriminal,
     

Code Select Expand

  MOV AL,080H
  SHL EAX,24

     The above uses 5 bytes, BUT it can be used for EAX,EBX, and EDX too.  Also bit patterns other than 080H can be easily be sent to the leftmost byte.  Ratch

[LimoDriver]

Am I the only one taking crazy pills?

Whatever happened to MOV ?!

Code Select Expand


mov e?x, WHATEVER


is ALWAYS 5 bytes!

I like to mov it, mov it.

[Ratch]

LimoDriver,

     Right you are.  Works for all registers too.  Sometimes when we try to be tricky, we overlook the obvious.  Ratch

[PBrennick]

Of course it would work for all registers! What a strange statement.

Paul

[Ratch]

PBrennick,

Of course it would work for all registers! What a strange statement.

     Is it?  When you consider that the methods submitted by dioxin and I do not?  dioxin's method does not work for EAX,EBX,EDX,EBP,ESI,EDI and mine won't work for EBP,ESI,EDI.  Ratch

[mercifier]

4 bytes:

Code Select Expand

004010DF   B1 1F            MOV CL,1F
004010E1   D3E1             SHL ECX,CL
Paul.

Why not:

Code Select Expand

            MOV CL,1
            ROR ECX,CL

?

[LimoDriver]

Why not

Code Select Expand


mov ebx, 12872h
cli
aad
hlt
lodsw


Because ... it doesn't WORK! :wink

[MichaelW]

mercifier,

That would work only if the upper three bytes of ECX were zero.

[dsouza123]

Dioxin's method though at 5 bytes will work for eax, ebx, ecx, edx

Code Select Expand


mov  al, 31   ; any odd byte value from 1 to 255 will give the same result
shl eax, 31

mov  bl,  1
shl ebx, 31

mov  cl,  3
shl ecx, 31

mov  dl,  7
shl edx, 31


[mercifier]

That would work only if the upper three bytes of ECX were zero.

You're right. The MSB would be set, but the upper three bytes would just be shifted down one position and unless they were all zeroes, that wouldn't be very useful.

[ThoughtCriminal]

Wow what Paul offered works exactly as I hoped and 1 byte less than just moving 80000000h into ecx.

My goal was not speed just the smallest code size possible.  I was interesting to see a lot of the less used instructions.

Thanks everyone.