Checker and Binutils 2.12

Alan Modra amodra@bigpond.net.au
Mon Mar 25 03:20:00 GMT 2002


On Fri, Mar 22, 2002 at 06:00:32PM -0500, Allen Barnett wrote:
> On Friday 22 March 2002 04:29 pm, Christian Jönsson wrote:
> > On Fri, Mar 22, 2002 at 04:25:41PM -0500, Allen Barnett wrote:
> > > I used the command:
> > >
> > > cvs -d :pserver:anoncvs@subversions.gnu.org:/cvsroot/gcc -z 9 co
> > > -rgcc-2_95_3 gcc
> >
> > what about
> >
> > cvs -d :pserver:anoncvs@subversions.gnu.org:/cvsroot/gcc -z 9 co
> > -rgcc-2_95-branch
> >
> > that is the 2.95 branch instead of the 2.95.3 one...?
> 
> OK. This netted me version 2.95.4 20020322 (prerelease). However, I'm still 
> getting the linker error. I attached the verbose output from the compilation. 
> What else can I look at?

Hmm, I see now why I successfully linked your testcase: I was linking
against dynamic libs, you're linked against static libs.  Why this
makes a difference, I haven't investigated, nor looked at why tinfo.o
is being pulled in from libgcc.  One reason the link is never going
to work is that the "chkr." prefix in your ti.o stops name mangling
from working.  I'm afraid I can't help any further.

$ nm -C ti.o 
00000000 ? __FRAME_BEGIN__
00000008 C type_info type_info node
         U void type_info node
00000000 V type_info virtual table
00000000 T chkr._._9type_info
00000000 W type_info::chkr.(char const *)
         U chkr.__builtin_delete
00000038 T chkr.__rtti_user
00000000 W chkr.__tf9type_info
00000040 T chkr.__tfv
00000000 W type_info::chkr.name(void) const
00000000 t gcc2_compiled.
$ nm -C ti_nopref.o 
00000000 T type_info::~type_info(void)
00000000 W type_info::type_info(char const *)
00000000 ? __FRAME_BEGIN__
         U __builtin_delete
00000038 T __rtti_user
00000000 W type_info type_info function
00000040 T void type_info function
00000008 C type_info type_info node
         U void type_info node
00000000 V type_info virtual table
00000000 t gcc2_compiled.
00000000 W type_info::name(void) const

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre



More information about the Binutils mailing list