[Bug ld/2750] ld has problem with -shared, and GDB, SUN solaris 10 X86 (amd)

[chong_guan_tan at yahoo dot com]

------- Additional Comments From chong_guan_tan at yahoo dot com  2006-08-08 
23:10 -------
Subject: Re:  ld has problem with -shared, and GDB, SUN solaris 10 X86 (amd)

nick,
your description is correct.  
I have built a version of medusa using sun ld and g++,
called it medusa sun, and one using g++ 4.0.2 and the
latest binutil, called it medusa.

readelf -h medusa > header.dump
readelf -h medusa.sun > header.sun
diff header.dump header.sun 
11c11
<   Entry point address:               0x804cb00
---
>   Entry point address:               0x80d4ec0
13c13
<   Start of section headers:          17001156 (bytes
into file)
---
>   Start of section headers:          20910516 (bytes
into file)
19,20c19,20
<   Number of section headers:         37
<   Section header string table index: 34
---
>   Number of section headers:         4313
>   Section header string table index: 4312

radelf -S   :: a ton of diff
readelf -s  :: a ton of diff



there are a lot of diff almost in all category.  If
you don't mind, I will email you the 2 image.

tan


--- nickc at redhat dot com
<address@hidden> wrote:

> 
> ------- Additional Comments From nickc at redhat dot
> com  2006-08-08 09:54 -------
> Hi Tan,
> 
>   Sorry for taking so long to get back to this
> issue.
> 
>   I am afraid I will not be able to visit you in
> Santa Clara since I am
> currently living in the UK...  So, perhaps you can
> do some investigation for me ?
> 
>   If I understand correctly the problem is that if
> you link your "medusa"
> application using SUN's linker you get an executable
> that is able to dynamically
> load the "my.so" library but which cannot be
> debugged by GDB because GDB
> complains about corrupted debug info.  On the other
> hand if you link "medusa"
> using the binutils linker you can then debug the
> executable with GDB(*), but it
> is no longer able to dynamically load "my.so".  Is
> this correct ?
> 
> (*) Apart from not being able to break into routines
> in header files which I
> think is a completely separate issue.
> 
>   If this is right, then what we need to do is to
> find out how the two different
> versions of the executable differ.  Can you examine
> them using the "readelf"
> program ?  Does that show any obvious differences in
> the symbol table, program
> headers or section headers ?  (Also does readelf
> give you any more information
> about the corrupt debug info ?)
> 
> Cheers
>   Nick
> 
> 
> -- 
> 
> 
> http://sourceware.org/bugzilla/show_bug.cgi?id=2750
> 
> ------- You are receiving this mail because: -------
> You reported the bug, or are watching the reporter.
> 
> 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2750

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.