GCC bug affecting the PPL on sparc64

[Roberto Bagnara <bagnara@cs.unipr.it>]

Hi there,

we are going through the list of architectures supported by Debian,
looking for problems affecting the Parma Polyhedra Library.  We need
advice concerning this bug (we have just added it to GCC Bugzilla):

    http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37661

Apparently something is wrong with the default -msoft-quad-float.
Should we:

1) do nothing, because (say) Debian will always ship binaries compiled
   with optimizations turned on;
2) compile with -mhard-quad-float on the sparc64 (would make sense
   only if SPARC implementations that don't have hardware support
   for the quad-word floating point instructions are not targeted
   by Debian);
3) apply a (yet to be identified) workaround;
4) disable support for long doubles on sparc64;
5) do something else.

Thanks for any suggestion.
All the best,

   Roberto

--
Prof. Roberto Bagnara
Computer Science Group
Department of Mathematics, University of Parma, Italy
http://www.cs.unipr.it/~bagnara/
mailto:bagnara@cs.unipr.it