ACE/TAO on RTEMS

Joel Sherrill joel.sherrill at OARcorp.com
Fri Mar 16 18:36:36 UTC 2001


Sergei Organov wrote:
> 
> "Rosimildo daSilva" <rdasilva at connecttel.com> writes:
> > From: Sergei Organov <osv at javad.ru>
> > To: rtems-users at oarcorp.com <rtems-users at oarcorp.com>
> > Date: Friday, March 16, 2001 11:07 AM
> > Subject: Re: ACE/TAO on RTEMS
> >
> >
> > >
> > >Are gcc flag '-ffunction-sections' and ld flag '--gc-sections' available
> > for
> > >your target? Do you use them?
> > >
> >
> >
> > I really like to hear if these makes difference.  I added them to a
> > makefile and I did not notice much difference. But, I did not tried
> > hard enough.  I started getting bizzar linking errors, and suddenly,
> > the order of the libraries started matter. I guess, I did something
> > wrong.
> 
> It really makes difference when you have a lot of routines in one object file
> only few of which are used. It effectively makes all your object files
> libraries. Don't know if it's the case with ACE/TAO.
> 
> I haven't got any troubles using these options after I've changed all
> '*(.text)' to '*(.text*)' in my linker command files.

I did this and it worked.  I don't know that I trust the resulting
executables.  Strangely, they are now all the same size. :)

bash$ sparc-rtems-size *Test
   text	   data	    bss	    dec	    hex	filename
 234768	   4544	  22796	 262108	  3ffdc	ARGV_Test
 234768	   4544	  22796	 262108	  3ffdc	Aio_Platform_Test
...

I think this is because the test is now actually not being linked
in since it is a C++ main. :)  

It certainly reduced the size though.

Now I am facing the standard problem that the tests have
a "main()" and I need to get to it.    Ralf/Eric .. have we
gotten to the point where we can move the init/fini code
and have main in user space?
 
> Sergei.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985



More information about the users mailing list