Crash on fasm 1.71.17 and 1.71.19

Index > Compiler Internals > Crash on fasm 1.71.17 and 1.71.19

Author

Thread

profkid13

Joined: 21 Aug 2003
Posts: 111

profkid13 26 Feb 2014, 12:41

Hi,

I managed to crash fasm with an intermediate version of my code.
(I'm converting this driver from MS COFF to PE format and refactoring some things along the way)

Halfway the assembly process, fasm literally crashes.
I tried fasmw 1.71.17 and 1.71.19, and cli version fasm 1.71.17.
All this on a windows 7 64-bit platform.

It's very likely there are still a bunch of bugs and inconsistencies in the code, but this doesnt mean fasm is allowed to crash, is it? Smile

EDIT: original source code removed

Last edited by profkid13 on 01 Mar 2014, 13:33; edited 1 time in total

26 Feb 2014, 12:41

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20623
Location: In your JS exploiting you and your system

revolution 26 Feb 2014, 12:48

profkid13 wrote:

It's very likely there are still a bunch of bugs and inconsistencies in the code, but this doesnt mean fasm is allowed to crash, is it?

No. It is not supposed to crash.

I have confirmed the bug and stickied the thread for Tomasz's attention.

26 Feb 2014, 12:48

tthsqe

Joined: 20 May 2009
Posts: 767

tthsqe 26 Feb 2014, 12:58

yup - definitely chrashes.
It is on line 1634 of formats.inc:

Code:

      store_fixup:
        add     dword [ebx],2      <- access violation here
        mov     ah,[esi+1]
        and     ah,0Fh
        mov     al,[esi+4]
        shl     al,4
        or      ah,al
        mov     al,[esi]
        stos    word [edi]
        add     esi,5
        loop    make_fixup

26 Feb 2014, 12:58

Tomasz Grysztar

Joined: 16 Jun 2003
Posts: 8391
Location: Kraków, Poland

Tomasz Grysztar 27 Feb 2014, 12:32

I have reduced the reproducing code into this one:

Code:

format PE

dd $

data fixups
end data

dd $

There was a bug in the backward patching of the fixups. I have corrected it for the 1.71.20 release.

27 Feb 2014, 12:32

l_inc

Joined: 23 Oct 2009
Posts: 881

l_inc 27 Feb 2014, 13:40

Tomasz Grysztar
Is "code cannot be generated" known behavior for the following code?

Code:

format PE DLL native

It seems to be somehow related to the non-configurable section alignment attribute (remember the discussion? Smile

), but I actually don't see any significant difference compared to this compilable code:

Code:

format PE DLL native
section '.flat' code data readable writable executable

_________________
Faith is a superposition of knowledge and fallacy

27 Feb 2014, 13:40

profkid13

Joined: 21 Aug 2003
Posts: 111

profkid13 01 Mar 2014, 13:33

Thanks for the quick help and fix.

01 Mar 2014, 13:33

< 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