This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: other/4878: gcc 3.0.2 output crashes HP-UX linker
- From: law at redhat dot com
- To: "John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>
- Cc: porten at trolltech dot com (Harri Porten), gcc-bugs at gcc dot gnu dot org
- Date: Wed, 09 Jan 2002 16:57:30 -0700
- Subject: Re: other/4878: gcc 3.0.2 output crashes HP-UX linker
- Reply-to: law at redhat dot com
> > Unlike the example program my project doesn't seem to have any of these
> > obvious defects.
> >
> > > non-virtual thunk to W::foo()(data)
> >
> > But dozens of those. Might be related to #4135. No idea.
>
> Yup, there is a missing `.IMPORT' for this symbol in the assembler
> output. It should in fact be a CODE symbol. I believe that that
> ld dumps core because a P fixup is being used with a DATA symbol
> (the default when no .IMPORT statement is provided).
>
> There are also missing imports for the typeinfo symbols but these
> don't hurt unless you use the HP assembler. The following patch
> was posted a long time ago but never reviewed. It fixes the problem
> for the typeinfo symbols.
>
>
> 2001-08-29 John David Anglin <dave@hiauly1.hia.nrc.ca>
>
> * cp-tree.h (tinfo_node_p): Add prototype.
> * decl2.c (import_export_vtable): Use external linkage for tinfo types.
> * rtti.c (tinfo_node_p): Add new function to test for tinfo node.
You might re-try the testcase using the current development sources -- Jason
just checked in a number of patches to address thunk linkage problems on HPs.
jeff