Re: libgcc-compat patch and gcc 2.95.4 on ppclinux

["H. J. Lu" <hjl at lucon dot org>]

On Thu, Jul 11, 2002 at 10:30:24PM -0400, Jack Howarth wrote:
>      Has anyone else tried to build the current glibc-2-2-branch
> with gcc 2.95.4 on ppclinux? On my debian ppc sid machine, I am
> finding the the patches contained in glibc-2-2-branch for 
> the libgcc-compat support on the powerpc build and work fine under
> gcc 3.1.x. However, my attempts at building the same glibc 2.2.5
> under gcc 2.95.4 have failed due to duplicate symbols for the
> same symbols declared in sysdeps/powerpc/libgcc-compat.c. Isn't this
> bound to happen? My understanding was that gcc 2.95.4 was flawed
> and would link in these same symbols from it libgcc. Thus if
> you build the libgcc-compat support in glibc-2-2-branch with
> gcc 2.95.4 wouldn't these symbols be defined twice when libc6.so
> is linked? Once for the copies incorrectly linked in by gcc 2.95.4
> and once for the internal declarations in sysdeps/powerpc/libgcc-compat.c.
> Don't we need some wrappers for this libgcc-compat support that only
> allows it to be built if the gcc used is < 3.1? Thanks in advance
> for any insights on this.
>                             Jack
> ps I do understand that libgcc-compat doesn't export these symbols
> for linking. However isn't there a still a problem in terms of 
> symbols internal to libc6.so if libgcc brings in the same ones
> as libgcc-compat has? I bring this up because I assume the same
> patches exist in the trunk as well which would imply that the 
> trunk will only build on ppclinux under gcc >= 3.1 as well without
> running into this duplicate symbols. 

Please try this patch. The problem is gcc 2.95.4 won't make those
libgcc functions hidden. We have to help it.


H.J.

Attachment: glibc-ppc.patch
Description: Text document