This is the mail archive of the
libstdc++@sourceware.cygnus.com
mailing list for the libstdc++ project.
Re: ctore_copy_dtor (was Re: string find)
- To: Phil Edwards <pedwards at disaster dot jaj dot com>
- Subject: Re: ctore_copy_dtor (was Re: string find)
- From: Branko Čibej <branko dot cibej at hermes dot si>
- Date: Mon, 26 Jun 2000 21:11:20 +0200
- CC: anthony at anthonyw dot cjb dot net, bkoz at redhat dot com, libstdc++ at sourceware dot cygnus dot com
- Organization: HERMES SoftLab
- References: <200006261747.NAA20638@disaster.jaj.com>
Phil Edwards wrote:
>
> > > > Sorry, 21_strings ctor_copy_dtor still fails on sparc-solaris-2.7 with
> > CVS
> > > > gcc,
> > > > CVS libstdc++ and binutils 2.9.1
> > >
> > > Hmm. Anyway you can try to track this down? this has been plaguing the
> > > solaris builds for a long time.
>
> After this patch, the shared build works, but the static build still fails
> under solaris 8 with binutils 000502 snapshot and CVS everything else.
> Getting a report from the coredump is interesting:
>
> % gdb ctor_copy_dtor.st-exe ctor_copy_dtor.st-exe.core
> GNU gdb 5.0
> [blah blah blah]
>
> warning: core file may not match specified executable file.
> [huh? and why not?]
Well, a bus error is a pretty horible thing to do ...
> Core was generated by `/stuff/pme/3test/sparc-sun-solaris2.8/libstdc++-v3/testsuite/ctor_copy_dtor.st-'.
> Program terminated with signal 10, Bus Error.
> #0 0x3ed90 in std::exception type_info function ()
> at /stuff/pme/3test/gcc/include/exception:403
> Line number 403 out of range; /stuff/pme/3test/gcc/include/exception has 65 lines.
> (gdb) quit
> %
>
> When I have time tomorrow, I'll look over my build environment again.
> I expect this to be a failure local to my systems somehow, but you never
> know.
I can second this. I'v Solaris 2.6, binutils 2.10 prerelese (2000-05-02),
and the following statically-linked tests fail, all of them with a bus
error:
21_strings/append.cc
21_strings/ctor_copy_dtor.cc
21_strings/element_access.cc
21_strings/insert.cc
21_strings/substr.cc
23_containers/bitset_ctor.cc
27_io/ios_members.cc
From what I've seen, they _all_ happen after an exception is thrown.
Perhaps the unwind data are getting stripped from the static library?
There was some discussion about that on one of the lists, IIRC.
Brane
--
Branko Čibej <branko.cibej@hermes.si>
HERMES SoftLab, Litijska 51, 1000 Ljubljana, Slovenia
voice: (+386 1) 586 53 49 fax: (+386 1) 586 52 70