Disassembly question.

Index > Main > Disassembly question.

Author

Thread

Teehee

Joined: 05 Aug 2009
Posts: 570
Location: Brazil

Teehee 22 Jan 2010, 20:38

I was doing some tests in Ollydbg and i have a question.

The instruction ADD EAX, 0 is translated to 83 C0 00.
And the instruction ADD AX, 0 is translated to 66 83 C0 00

My question is: why that '66'? what it mean? (how that things work?)

and i was looking at http://ref.x86asm.net/coder32.html but I didn't find a reference to it.

_________________
Sorry if bad english.

22 Jan 2010, 20:38

baldr

Joined: 19 Mar 2008
Posts: 1651

baldr 22 Jan 2010, 20:56

Teehee,

Didn't you miss 66 in pf column?

Operand-size override prefix tells that the following instruction uses non-default (for code segment) operand size. I.e. in 32-bit code segment 66 83 C0 00 is decoded as add ax, 0, in 16-bit — as add eax, 0.

You may also refer to subchapter "2.1.1 Instruction Prefixes" of Intel SDM vol. 2A for detailed explanation of prefixes (67 address-size override prefix is important too), or subchapter "1.2 Instruction Prefixes" of AMD64 APM vol. 3 (even more details given).

22 Jan 2010, 20:56

Teehee

Joined: 05 Aug 2009
Posts: 570
Location: Brazil

Teehee 22 Jan 2010, 21:07

baldr wrote:

Didn't you miss 66 in pf column?

there is no 66 prefix in that collum Sad

There is in ADDPD instruction.

Thanks baldr, i will take a look at Intel SDM prefix topic.

_________________
Sorry if bad english.

22 Jan 2010, 21:07

baldr

Joined: 19 Mar 2008
Posts: 1651

baldr 22 Jan 2010, 21:23

Teehee,

Follow the "63" link in one-byte opcodes index and voilà!

22 Jan 2010, 21:23

Teehee

Joined: 05 Aug 2009
Posts: 570
Location: Brazil

Teehee 22 Jan 2010, 21:25

oh.. i see Confused

hehe [I was thinking it would be show in ADD instruction row]

22 Jan 2010, 21:25

MazeGen

Joined: 06 Oct 2003
Posts: 977
Location: Czechoslovakia

MazeGen 23 Jan 2010, 20:53

Teehee, I can't add another definition with 0x66 prefix to every primary opcode that depends on it, it would make the reference so huuuuge Smile

23 Jan 2010, 20:53

Borsuc

Joined: 29 Dec 2005
Posts: 2465
Location: Bucharest, Romania

Borsuc 24 Jan 2010, 20:45

In 32-bit mode, 0x66 simply switches from 32-bit to 16-bit operands on the instructions.

24 Jan 2010, 20:45

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20299
Location: In your JS exploiting you and your system

revolution 24 Jan 2010, 23:45

Borsuc wrote:

In 32-bit mode, 0x66 simply switches from 32-bit to 16-bit operands on the instructions.

Not true for SSE instructions.

24 Jan 2010, 23:45

edfed

Joined: 20 Feb 2006
Posts: 4330
Location: Now

edfed 25 Jan 2010, 02:44

in 16bit mode, 66 make it 32 bit
in 32bit mode, it is the oposite.

something like a xor inside CPU internal mode switch.

25 Jan 2010, 02:44

Tyler

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

Tyler 25 Jan 2010, 11:04

revolution wrote:

Not true for SSE instructions.

What does it do to SSE instructions?

25 Jan 2010, 11:04

Fanael

Joined: 03 Jul 2009
Posts: 168

Fanael 25 Jan 2010, 11:44

Tyler wrote:

revolution wrote:
Not true for SSE instructions.

What does it do to SSE instructions?

For instance, MOVAPS changes to MOVAPD.

25 Jan 2010, 11:44

hopcode

Joined: 04 Mar 2008
Posts: 563
Location: Germany

hopcode 25 Jan 2010, 11:52

Teehee wrote:

...tests in Ollydbg...

using use32/use16 to test it in your test app, for example

Code:

use16
 add eax,0
 add ax,0
use32
 add eax,0
 add ax,0
;...

I have drawn a simple skema to opcode functionalities orienting
here (for a better organizing purpouse)
http://board.flatassembler.net/topic.php?p=108824#108824

Cheers,
hopcode

25 Jan 2010, 11:52

Borsuc

Joined: 29 Dec 2005
Posts: 2465
Location: Bucharest, Romania

Borsuc 25 Jan 2010, 20:07

revolution wrote:

Borsuc wrote:
In 32-bit mode, 0x66 simply switches from 32-bit to 16-bit operands on the instructions.
Not true for SSE instructions.

But SSE instructions don't have 32-bit operands... I meant it switches from instructions that have 32-bit operands to 16-bit operands

_________________
Previously known as The_Grey_Beast

25 Jan 2010, 20:07

asmmsa

Joined: 06 Feb 2010
Posts: 45

asmmsa 09 Feb 2010, 20:51

0x66 is a prefix.
it normally swap default operand size. when first opcode byte is 0x0F meaning of 0x66 is changed.

09 Feb 2010, 20:51

baldr

Joined: 19 Mar 2008
Posts: 1651

baldr 10 Feb 2010, 00:45

asmmsa,

PS suffix changed to PD and likewise? How this is different from 32/16 (for 0Fs, literally)? Wink

10 Feb 2010, 00:45

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20299
Location: In your JS exploiting you and your system

revolution 10 Feb 2010, 01:44

baldr wrote:

PS suffix changed to PD and likewise? How this is different from 32/16 (for 0Fs, literally)?

SQRTPS XMM and SQRTPD XMM: Both without 0x66 byte.
ROUNDPS XMM and ROUNDPD XMM: Both have 0x66 byte.
PSHUFB MM and PSHUFB XMM: Only XMM version has 0x66 byte.

This is why Intel explicitly state that 0x66 is not a prefix byte when used with SIMD encoding.

10 Feb 2010, 01:44

< 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