Fix PR debug/24444: incorrect frame base in IA64 prologues

[Alexandre Oliva]

On Feb  5, 2006, Daniel Jacobowitz <drow@false.org> wrote:

> I don't think you need a different mechanism; but it might be wise
> to use an ia64-specific builtin (since this is ia64-specific code and
> you know that it will use ia64-specific unwind data).

That particular piece of code doesn't care what unwind data it's going
to use.  What it wants is the caller's stack pointer.  Or something at
a fixed offset from it.  

> Make it clear that you're not dealing with the DWARF instance of CFA
> at all.

But we are.  The DWARF instance of the CFA is *exactly* the caller's
stack pointer.


That said, Richard seems to be of the opinion that we should avoid an
ABI change, or even something that's not quite an ABI change but might
be construed as such, until we're 100% sure it is called for; perhaps
he's no longer sure the change was correct, after all :-(

So I suggest Jakub to check the patch in as soon as his testing is
done, because I'm going to be away for most of tomorrow.

-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
Secretary for FSF Latin America        http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}