This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
RE: Linking libgcj, and Java security
- From: "Boehm, Hans" <hans_boehm at hp dot com>
- To: "'Andrew Haley'" <aph at redhat dot com>, Jeff Sturm <jsturm at one-point dot com>
- Cc: java at gcc dot gnu dot org
- Date: Tue, 3 Dec 2002 12:13:01 -0800
- Subject: RE: Linking libgcj, and Java security
> -----Original Message-----
> From: Andrew Haley [mailto:aph@redhat.com]
>
> Jeff Sturm writes:
> > On Tue, 3 Dec 2002, Andrew Haley wrote:
> > > Before I submit a patch, is there any good reason _not_ to use
> > > -Bsymbolic?
> >
> > Back when Bryce and I tried it, it didn't quite work.
> IIRC the GC relies
> > on global symbol relocations to resolve DATASTART,
> DATAEND. There may be
> > other problems. There should be some discussion in the archives.
>
> Okay. I would have thought that the best bet was to fix
> whatever issues
> prevent -Bsymbolic from being used. Hans?
Unfortunately, I neither recall issues related to this, nor can I find them in the archives. The collector needs __data_start resolved to the start of the main (not libgcj) data segment, and analogously for _end. If -Bsymbolic causes it to refer to something else, then we have a problem. Under Linux, this can probably be worked around with /proc/self/maps, though that's not ideal. There may be other solutions.
> I confess that linker magic is not my strongest subject.
>
> Andrew.
>
Unfortunately, it isn't mine either.
Hans