bug#1079: GNUstep port segfaults making autoloads on x86_64

[Yavor Doganov]

Glenn Morris wrote:
> 
> I don't have either. gnustep-startup provides both, but uses ffcall
> by default. I tried to get it to use ffi instead, but it did not
> work out of the box and I haven't had time to look into it yet.

AFAICS ./InstallGNUstep --ffi=libffi should DTRT.
Failing that, ./configure --enable-libffi --disable-ffcall in the
gnustep-base tree should suffice (after installing libffi).  It is
enough to recompile only GNUstep Base, not the whole thing.

Maybe someone else with access on a x86_64 machine can try this.

> > can you compile this test program (it should never crash):
> 
> Never say never...

It is excellent that it crashes, this confirms it's a GNUstep bug.  It
looks like it is unable to pass the exception up to the call stack or
something like that.

Before reporting it to the GNUstep developers, it would be useful to:

1) Check if it still crashes with -base built against libffi.
2) Check if the problem persists in SVN trunk:
   svn co http://svn.gna.org/svn/gnustep/libs/base/trunk gnustep-base