This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: /usr/lib/dld.sl: Unresolved symbol: <FF>libstdc++.sl.3 (plabel)
- To: law at redhat dot com
- Subject: Re: /usr/lib/dld.sl: Unresolved symbol: <FF>libstdc++.sl.3 (plabel)
- From: "John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>
- Date: Sat, 10 Mar 2001 11:35:28 -0500 (EST)
- Cc: gcc-bugs at gcc dot gnu dot org
> In message <200103072104.QAA02089@hiauly1.hia.nrc.ca>you write:
> > With yesterdays 3.0 cvs source, many of the g++ tests fail to link:
> >
> > ...
> > /usr/lib/dld.sl: Unresolved symbol: <FF>libstdc++.sl.3 (plabel) from /xxx/
> > gnu/gcc-3.0/objdir/hppa1.1-hp-hpux10.20//libstdc++-v3/src/.libs/libstdc++.s
> > l.3
> > FAIL: g++.abi/cxa_vec.C Execution test
> >
> > HP nm dumps core when I try to list the library symbols. GNU nm lists
> > them but I can't see the problem. Ideas?
> It's related to weak symbols. I know that much :-)
HP nm still dumps core even with weak turned off. However, HP `nm -p'
seems to work ok. GNU nm is confused about symbol types and possibly
its scope, printing `?' for most symbols in objects generated by g++.
I think it is the `-ffunction-sections' and `-fdata-sections' options
used in building libstdc++ that is confusing GNU nm. This is probably
also the cause of HP nm dumping core without `-p'. My guess is this
is also giving the linker problems.
Dave
--
J. David Anglin dave.anglin@nrc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6605)