RE: Object files know where they're built?!?

RE: Object files know where they're built?!?

[Bruce Korb]

Hi Nathanael,

Try messing with libelf.  It makes it pretty trivial to
extract certain sections (e.g. "text" and "data"), and
ignore others (that may have date stamps, source file
names, etc.)

Cheers,
	Bruce

Re: Object files know where they're built?!?

[Alexandre Oliva]

On Mar  1, 2003, Nathanael Nerode <neroden@twcny.rr.com> wrote:

> Do the object files (on i686-pc-linux-gnu) somehow encode their build 
> location?

Yes.  This is necessary for the debugger to find source files when
they're referenced using relative pathnames.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer

Re: Object files know where they're built?!?

[Nathanael Nerode]

Richard Earnshaw said:
>Not necessarily, but they do include the location of source and include 
>files (so that debuggers can find things) -- and most files will 
>include 
>config.h, insn-flags.h, etc which are put in the build directory.
>
>R.

Right.  Stripping debugging information allows me to get a clean 
compare (for purposes of checking that my relocation doesn't change 
anything).  Thanks.

--Nathanael

Re: Object files know where they're built?!?

[Richard Earnshaw]

> Joseph Myers said:
> >dependencies, between all stages, to one invocation of make, but this
> >can't work because each stage needs to be built separately, with the 
> >files
> >in the same location (not in the stage subdirectory), for the compare
> >stage to succeed.
> 
> I ran into this when I tried to use a different scheme for building 
> stage1.  I wanted to compare the files produced by my new scheme with 
> the files produced by the original.  My new scheme built stage1 in a 
> different directory (stage1-gcc).  Every object file differed at byte 33 
> or 34.
> 
> I couldn't figure out why.
> 
> Do the object files (on i686-pc-linux-gnu) somehow encode their build 
> location?  This would be stupid, obviously, but if it's true it's true. 
> :-)  Is there a way to compare the object files without regard to 
> spurious differences like this?

Not necessarily, but they do include the location of source and include 
files (so that debuggers can find things) -- and most files will include 
config.h, insn-flags.h, etc which are put in the build directory.

R.

Object files know where they're built?!?

[Nathanael Nerode]

Joseph Myers said:
>dependencies, between all stages, to one invocation of make, but this
>can't work because each stage needs to be built separately, with the 
>files
>in the same location (not in the stage subdirectory), for the compare
>stage to succeed.

I ran into this when I tried to use a different scheme for building 
stage1.  I wanted to compare the files produced by my new scheme with 
the files produced by the original.  My new scheme built stage1 in a 
different directory (stage1-gcc).  Every object file differed at byte 33 
or 34.

I couldn't figure out why.

Do the object files (on i686-pc-linux-gnu) somehow encode their build 
location?  This would be stupid, obviously, but if it's true it's true. 
:-)  Is there a way to compare the object files without regard to 
spurious differences like this?

--Nathanael