This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] 3.2: Fixup symbol versioning mistake on powerpc-linux-gnu


On Mittwoch, 7. August 2002 00:01, Richard Henderson wrote:
> On Tue, Aug 06, 2002 at 11:41:56PM +0200, Franz Sirl wrote:
> > Huh? "just to shift"? I don't understand, I've already seen bugs
> > caused by the current setup...
>
> Then I still don't understand.  If either libc.so or libgcc_s.so
> exports the link-time version, and both export the runtime version,
> then the symbol should be resolvable from either library.

As explained in an earlier message, if you deliver your c++ app together with 
libgcc_s and use it on on a gcc-2.95 compiled glibc, the app may not run 
because of missing symbol versions. That happened with the OpenOffice/mozilla 
bundle cause still a lot of shared libs are wrongly compiled with -lc -lgcc, 
either hardcoded or because libtool does so (kde3 for example).

That wouldn't happen if the gcc version would be easily distinguashable from 
the glibc version and glibc doesn't export them as a linktime reference.

Maybe I should ask a question myself, what is the reason for this strange 
setup on all linux platforms that don't override it? Why is 
config/libgcc-glibc.ver done this way?

Franz.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]