libgcc fix approach

[Jack Howarth]

Ulrich,
    So on the flip side of that, you wouldn't object if they
wrapped any libgcc-compat stuff in glibc-2-2-branch so that it
doesn't build below gcc 3.1? I think that is the cleanest
approach for that branch since its not like we really want to
deal with the libgcc bug fix getting backported into gcc < 3.1
anyway. I ask because the current glibc-2-2-branch doesn't build
with gcc < 3.1 on ppc because of the duplicate symbols from
libgcc-compat and the simplest fix is just not to build that code
unless gcc > 3.1 is used. HJ seems hesitant to do that though,
wanting to not export these libgcc symbols from glibc-2-2-branch
at all. It doesn't seem like worth seem the bother to me.
                          Jack