Re: work for glibc 2.2.5-14
It's coming...building and testing today and tomorrow...
C
On Mon, 12 Aug 2002, Jack Howarth wrote:
> After re-reading all the messages in the thread...
>
> http://gcc.gnu.org/ml/gcc-patches/2002-08/subjects.html#00357
>
> I am more convinced that Jakub and Richard are right and no
> fixup to the version symboling is required. In particular, if
> you look at message...
>
> http://gcc.gnu.org/ml/gcc-patches/2002-08/msg00374.html
>
> Franz's objection there is no longer true with HJ's new binutils
> fix. This brings me to the following.
> It will be ABSOLUTELY ESSENTIAL that we upgrade binutils to
> 2.13.90.0.3 PLUS the patches submitted into the binutils-cvs today
> by HJ...
>
> http://sources.redhat.com/ml/binutils-cvs/2002-08/msg00077.html
> http://sources.redhat.com/ml/binutils-cvs/2002-08/msg00076.html
>
> If won't do this on ppc before glibc 2.2.4-14 is built and installed
> we will suffer the problem Franz complained about. That is
> with the libgcc-compat code for ppc pulled in from glibc-2-2-branch
> we end up with the error...
>
> /lib/libgcc_s.so.1(*IND*+0x0): multiple definition of `__divdi3@GLIBC_2.0'
> /lib/libc.so.6(.text+0x574): first defined here
>
> when linking after this new glibc is installed. This is a binutils
> bug fixed in the patches cited above. You don't need that
> binutils patch to build glibc with the libgcc-compat patch but
> you WILL need it to build anything against that glibc after it
> is installed. So we should get binutils squared away ASAP so
> building on voltaire isn't broken once glibc 2.2.5-14 gets installed.
> Jack
>
>
Reply to: