Re: [PATCH] 3.2: Fixup symbol versioning mistake on powerpc-linux-gnu

[Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>]

On Montag, 12. August 2002 21:51, Richard Henderson wrote:
> On Mon, Aug 12, 2002 at 03:30:15PM -0400, Jakub Jelinek wrote:
> > > > 1. do nothing:
> > > > 	This leaves use with linktime version for the divmod functions in
> > > > both glibc and libgcc_s, which everybody seems to agree that this is
> > > > a bad idea.
> >
> > #1 seems like the preferred choice when the symbol versions are identical
> > IMNSHO.
>
> Exactly.  I've still not heard a *coherent* explanation
> of why any change is needed.  Only several incomplete
> explanations.

Uhm, what was incoherent about my last explanation?

But it general it's _extremely_ hard for me to find the right words on this 
issue, simply because I don't know them when it comes to binutils issues as a 
non-native speaker. Additionally I seem to wade in a maze of binutils bugs, 
which makes it hard for to tell the real culprits.

So, with the current binutils state I see the following facts:

1. exporting these symbols both from glibc and libgcc_s with linktime 
reference is a bug and causes undefined symbols when running gcc3.1 compiled 
binaries on a gcc2 compiled glibc (even if libgcc_s is distributed with the 
binary!)
2. if I demote the symbols in glibc to a runtime reference as Jakub suggested, 
gcc fails to build a correct libgcc_s

Now, since nobody is answering me with any hints, what else do you expect me 
to do?

Franz.