Feature request: make 0's uninitialized in PE section

Index > Compiler Internals > Feature request: make 0's uninitialized in PE section

Author

Thread

SFeLi

Joined: 03 Nov 2004
Posts: 138

SFeLi 21 May 2009, 16:59

What do you think about automatically converting to uninitialized data all data definitions at the end of the section, which produce zero bytes? It wouldn’t break anything in PE.

Code:

        format  pe gui 4.1
        entry   start
        include 'win32a.inc'

start:
        retn

section '.data' data readable writeable
foo dd 12345678
bar dd 0 ; Automatically make it bar dd ?
baz dd 0 ; This too.

Last edited by SFeLi on 21 May 2009, 17:03; edited 1 time in total

21 May 2009, 16:59

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20343
Location: In your JS exploiting you and your system

revolution 21 May 2009, 17:01

Why would we need to do this?

21 May 2009, 17:01

SFeLi

Joined: 03 Nov 2004
Posts: 138

SFeLi 21 May 2009, 17:09

Yesterday I saw a source that defined a huge list of data via macros and file directives. Resulting EXE had about 600 (!) kb of trailing zeroes in the .flat section. Why not to make PE formatter think about this?

21 May 2009, 17:09

LocoDelAssembly
Your code has a bug

Joined: 06 May 2005
Posts: 4624
Location: Argentina

LocoDelAssembly 21 May 2009, 17:11

Quote:

at the end of the section, which produce zero bytes? It wouldn’t break anything in PE.

Sure about this? I remember that some of my MASM programs were not working correctly when placed on ".data?" section variables and assumed them to be zero at start on Win98.

21 May 2009, 17:11

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20343
Location: In your JS exploiting you and your system

revolution 21 May 2009, 17:15

There are times where putting zeros is needed. With post-processing of files data may be inserted after the .exe is generated.

For the above file, just change your code to correctly generate the uninitialised data.

I don't think it is a good idea to make the assembler guess at what you wanted. What happens when the guess is wrong and you have to find a workaround to fix the assembler's "error".

21 May 2009, 17:15

SFeLi

Joined: 03 Nov 2004
Posts: 138

SFeLi 21 May 2009, 17:24

LocoDelAssembly, I’m using 98, almost always try to dd ? everything that needs to be 0 and never encountered such problems…
revolution, do you know a way to store dword ? at addr?
But ok, I was wrong.

21 May 2009, 17:24

LocoDelAssembly
Your code has a bug

Joined: 06 May 2005
Posts: 4624
Location: Argentina

LocoDelAssembly 21 May 2009, 17:40

Microsoft Portable Executable and Common Object File Format Specification - Revision 8.1 – February 15, 2008 wrote:

VirtualSize: The total size of the section when loaded into memory. If this value is greater than SizeOfRawData, the section is zero-padded. This field is valid only for executable images and should be set to zero for object files.

This don't back my experience very much... I suppose the specs haven't changed over the time about this so probably my memory is failing me.

21 May 2009, 17:40

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20343
Location: In your JS exploiting you and your system

revolution 21 May 2009, 18:01

SFeLi wrote:

do you know a way to store dword ? at addr?

Maybe not exactly what you are asking but this a way of finding a length of some non-zero data and then reserving it in the PE

Code:

virtual
  ...
  ; do your 'file' and whatever other stuff
  ...
  MyDataSize = $-$$
end virtual
rb MyDataSize

21 May 2009, 18:01

< 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