Rotating Bits

Index > Main > Rotating Bits

Goto page 1, 2 Next

Author

Thread

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 00:19

What's the best way to get the lower half of eax into the higher half while moving the higher into the lower, preferably w/o the possibility of ending up with the carry flag set? I'm doing some multiplying and I need to get dx into the higher half of eax while maintaining ax(not necessarily in ax, just the value of ax needs to keep from being lost).

27 Apr 2010, 00:19

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20526
Location: In your JS exploiting you and your system

revolution 27 Apr 2010, 00:32

"best" in what sense? Smallest code size? Fastest execution speed? Easiest for you to understand? Least amount of typing?

27 Apr 2010, 00:32

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 01:00

Sorry for the ambiguity, easiest to understand(less typing usually comes with this one). Is there anything that will rotate bits w/o setting the carry flag when it rotates out a 1 and will rotate bits from one end to the other(bits rotated out, rotate in at opposite end). If not, what would you use to accomplish my task I mentioned in my above post..

27 Apr 2010, 01:00

windwakr

Joined: 30 Jun 2004
Posts: 827

windwakr 27 Apr 2010, 01:22

ROL/ROR?

rol eax,16

EDIT:
Wait no, that messes with carry.

_________________
----> * <---- My star, won HERE

27 Apr 2010, 01:22

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20526
Location: In your JS exploiting you and your system

revolution 27 Apr 2010, 01:34

In X86 AFAIK there are no non-flags-setting rotation instructions or groups of instructions.

LEA would be the closest but that will discard the upper bits.

Why not just save the carry flag, do the rotations, then restore the flags?

27 Apr 2010, 01:34

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 01:42

revolution wrote:

Why not just save the carry flag, do the rotations, then restore the flags?

I guess that's what I'll do.

27 Apr 2010, 01:42

edfed

Joined: 20 Feb 2006
Posts: 4354
Location: Now

edfed 27 Apr 2010, 01:49

Code:

push eax
push eax
pop ax
pop eax
lea esp,[ep+2]

will not mess the flags. Laughing

the same, but now, with fixed memory

Code:

mem dd ?,?
...
mov [mem],eax
mov [mem+4],eax
mov eax,[mem+2]

personnaly, i prefer do it with ROR and ROL.

about DX to high part...
push dx ax
pop eax

27 Apr 2010, 01:49

sinsi

Joined: 10 Aug 2007
Posts: 794
Location: Adelaide

sinsi 27 Apr 2010, 01:53

Code:

  xchg al,ah
  bswap eax
  xchg al,ah

messy, but so is saving/restoring flags.

27 Apr 2010, 01:53

ass0

Joined: 31 Dec 2008
Posts: 518
Location: ( . Y . )

ass0 27 Apr 2010, 01:54

I think you are confusing rol/ror with rcl/rcr

rol doesn't use carry flag while rcl does.

_________________

Nombre: Aquiles Castro.
Location2: about:robots

27 Apr 2010, 01:54

ass0

Joined: 31 Dec 2008
Posts: 518
Location: ( . Y . )

ass0 27 Apr 2010, 02:02

http://en.wikipedia.org/wiki/X86_assembly_language wrote:

rotate with and without carry, rcl/rcr, rol/ror

Code:

mov cl,16
rol eax,cl

_________________

Nombre: Aquiles Castro.
Location2: about:robots

27 Apr 2010, 02:02

edfed

Joined: 20 Feb 2006
Posts: 4354
Location: Now

edfed 27 Apr 2010, 02:06

indead, that's pretty logic...
what should be passed ot carry in case of roX?
the fisrt bit?
the last bit?

then, nothing pass through carry and it is clear. Wink

27 Apr 2010, 02:06

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 02:07

rol shifts bit to carry and low bit while rcl uses the carry flag as if it were a 17th bit.

http://webster.cs.ucr.edu/AoA/DOS/ch06/CH06-3.html#HEADING3-294

27 Apr 2010, 02:07

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20526
Location: In your JS exploiting you and your system

revolution 27 Apr 2010, 02:17

Tyler: Download the manuals from Intel/AMD. Don't just rely on random websites to tell you about X86 instructions, get it from the original source.

27 Apr 2010, 02:17

ass0

Joined: 31 Dec 2008
Posts: 518
Location: ( . Y . )

ass0 27 Apr 2010, 02:21

http://www.intel.com/software/products/documentation/vlin/mergedprojects/analyzer_ec/mergedprojects/reference_olh/mergedProjects/instructions/instruct32_hh/vc270.htm wrote:

For the ROL and ROR instructions, the original value of the CF flag is not a part of the result, but the CF flag receives a copy of the bit that was shifted from one end to the other.

Yeah a bit confusing.

_________________

Nombre: Aquiles Castro.
Location2: about:robots

27 Apr 2010, 02:21

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 03:05

Was that a pun?

revolution: Okay, I'll reference a manual next time. That site was the first that come up with a search of "rol asm" on google and it agreed with what I'd already thought, so I assumed it was right, and it is an edu. Was I wrong?

27 Apr 2010, 03:05

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20526
Location: In your JS exploiting you and your system

revolution 27 Apr 2010, 03:18

I can't see how being "an edu" helps. Try intel.com or amd.com, anything else is just rehashing the originals and opening up the possibility of introducing mistakes.

27 Apr 2010, 03:18

LocoDelAssembly
Your code has a bug

Joined: 06 May 2005
Posts: 4624
Location: Argentina

LocoDelAssembly 27 Apr 2010, 05:21

Just wondering, why do you need CF untouched? Can you show us the code you're working on?

27 Apr 2010, 05:21

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 05:30

I just didn't want it tampered with, that's all.

Code:

print_str_bw:
   push      ebp
   mov       ebp,esp
   pushad
   label     .param dword at ebp + 8
   mov       esi,[.param]
   call      .update_print_pos
   mov       ah,7
   .next_char:
      lodsb
      cmp       al,0ah
      je        .lf
      cmp       al,0dh
      je        .cr
      cmp       al,0
      je        .done
      stosw
      cmp       [col],80
      jae       .eol
      add       [col],1
      jmp       .next_char
   .cr:
      add       esi,1
   .lf:
   .eol:
      mov       [col],0
      cmp       [row],23
      jae       .eos
      add       [row],1
      call      .update_print_pos
      jmp       .next_char
      .eos:
         call      scroll_down
         call      .update_print_pos
         jmp       .next_char
   .update_print_pos:
      push      eax edx
      call      update_pos
      mov       edi,[vid_mem]
      mov       eax,[pos]
      imul      eax,eax,2
      add       edi,eax
      pop       edx eax
      ret
   .done:
   call         update_cursor
   popad
   mov          esp,ebp
   pop          ebp
   ret

I went ahead and used imul in .update_print_pos. If you see any mistakes or unnecessary instructions, please do tell.

27 Apr 2010, 05:30

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20526
Location: In your JS exploiting you and your system

revolution 27 Apr 2010, 05:34

'imul eax,eax,2' == 'add eax,eax' == 'shl eax,1' == 'lea eax,[eax*2]' == 'lea eax,[eax+eax]'

27 Apr 2010, 05:34

Tyler

Joined: 19 Nov 2009
Posts: 1216
Location: NC, USA

Tyler 27 Apr 2010, 05:42

Okay, you got me... Smile

There's also this were I have to multiply by 80, forgot about it.

Code:

update_pos:
   pushad
   xor       eax,eax
   mov       al,[row]             ; Get new row
   mov       dl,80
   mul       dl                   ; row*80
   xor       edx,edx
   mov       dl,[col]
   add       eax,edx              ; row*80+col
   mov       [pos],eax            ; pos=row*80+col
   popad
   ret

27 Apr 2010, 05:42

Goto page 1, 2 Next

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum