This is the mail archive of the
mailing list for the glibc project.
Re: libm.so vs libgcj.so
On Sun, May 18, 2003 at 05:48:52PM -0400, Jack Howarth wrote:
> HJ and Jakub,
> Does the fact that glibc >= 2.3.2 no longer considers non-weak
> and weak symbols in shared libs to be different imply that the
> warning (and perhaps the actual test) in prelink should be changed
> from "undefined non-weak symbols" to just "undefined symbols"?
> I found that confusing myself when I was looking for the missing
> linkage for some shared libs and discovered that the symbols in
> question provided by libm.so were in fact weak rather than non-weak.
Nope. That's a different thing.
prelink warning talks about SHN_UND symbols other than STB_WEAK.
It is ok if SHN_UND STB_WEAK symbols cannot be resolved, they just resolve
to 0 in that case.
The ld.so change in glibc 2.3.2 was about STB_WEAK vs. != STB_WEAK
symbol definitions (ie. other than SHN_UND) which are now treated equally.