Patch: libffi on IA64 HP-UX

[Steve Ellcey]

> From: Richard Henderson <rth@redhat.com>
>
> Perhaps better as
> 
>    	/* Retrieve closure pointer and real gp.  */
> #ifdef _ILP32
> 	addp4	out0 = 0, gp
> 	addp4	gp = 16, gp
> #else
>    	mov	out0 = gp
>    	add	gp = 16, gp
> #endif

Yes, that would be better.

> I'm not familiar enough with the ilp32 conventions ot know for sure,
> but I do wonder if either (1) gp ought to already be pointer-extended
> or (2) out0 need not be extended for the "ffi_closure *closure" argument.

I don't know the answer to this either but since your suggested change
means we either have two addp4's or an add and a move and both sequences
take the same amount of time there doesn't seem to be any reason not to
do the addp4's.  If the user already extended gp and/or didn't need out0
to be extended no harm is done.

I retested with this change and everthing still passes.

Steve Ellcey
sje@cup.hp.com