On Wed, 2009-12-09 at 15:31 +0100, Vincent R. wrote:
> > It reminds me what was reported the first time someone talked about
> issue
> > between cegcc and Windows Mobile 6.1.
> > In this initial report there was something about the size and under some
> > value, it was working fine.
> >
>
> Here is extract of the original post :
>
> "I tried to modify binutils/bfd/peXXigen.c to have .idata without write
> attribute and my previous test case is loading successfully. But if I
> compile with all the CRT stuff (a DLL with DllMainCRTStartup), the
> application crash at load. When looking at the crash, the problem is
> that only the first 64kB is loaded and DllMainCRTStartup try to call a
> function after those 64kB, causing a page exception. Compiling without
> the CRT stuff load all the code. I'm searching why there is such a
> difference."
>
> Now I don't know if it's the same thing you are talking about.
Thanks for reminding me of these, I had almost forgotten. I'll do more
tests on this.
What I find odd is :
- those mails all talk about a change happening when SizeOfImage>10000
- a number of websites talk about 64KB boundaries (which is the same
number, but maybe not the same variable)
- I found a change when text segment size hit 0x1000 which is 4K
> please try to upx your binaries and check if it still doesn't work.
> Just curious to know the result.
Running upx on lib2.dll does make it work on the emulator. Or at least,
testapi can now load it. Haven't tried more, this is not really a useful
DLL, I'm afraid.
The new DLL causes objdump to generate an error message. It also shows
only 3 sections, not sure how reliable this is though (haven't looked at
a hex dump) :
Sections:
Idx Name Size VMA LMA File off Algn
0 UPX0 00010000 00011000 00011000 00000000 2**2
ALLOC, CODE
1 UPX1 00001e00 00021000 00021000 00000200 2**2
CONTENTS, ALLOC, LOAD, CODE, DATA
2 UPX2 00000200 00023000 00023000 00002000 2**2
CONTENTS, ALLOC, LOAD, READONLY, DATA
arm-mingw32ce-objdump: lib2.dll: File truncated
One of my ideas is that the mere number of sections of some type (CODE,
LOAD, DATA ?) might be a differentiator. This is in line with your
original assessment that .edata and .idata have to be absorbed
into .data or so.
Some more strangeness :
pavilion: {13} arm-mingw32ce-size orig/lib2.dll lib2.dll
text data bss dec hex filename
4933 500 0 5433 1539 orig/lib2.dll
73728 0 0 73728 12000 lib2.dll
pavilion: {14} ls -l orig/lib2.dll lib2.dll
-rwxrwxr-x 1 danny danny 19254 2009-12-11 17:41 lib2.dll
-rwxrwxr-x 1 danny danny 33078 2009-12-11 17:30 orig/lib2.dll
the file size goes down but the section sizes don't. But this output may
suffer from the same problem as objdump - may not be reliable info.
Danny
--
Danny Backx ; danny.backx - at - scarlet.be ; http://danny.backx.info
------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
_______________________________________________
Cegcc-devel mailing list
Cegcc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cegcc-devel
