On Mon, Apr 01, 2002 at 12:28:40PM -0500, John David Anglin wrote:
> > > I added $$dyncall to milli64.S and modified t-linux to use milli64.S.
> > > This fixes the arith-rand.c fail and there are no other regressions,
> > > so I think we can use milli64.S on both the 32-bit and 64-bit ports.
> >
> > It looks like the appropriate #ifdefs are present in milli64.S so
> > that it can be used as-is for 32-bit (so long as the "pa64" CPP
> > symbol is properly defined).
>
> We have the appropriate #ifdefs and defines to handle the different
> return conventions. My only real concern was some corner case arising
> from the difference in register size. If Alan doesn't recall anything
> in the next day or so, I will apply the patch to the gcc trunk.
No, I don't recall any such corner cases, but then I don't think
I ever looked into the code in any detail. It was more a case of
doing robotic edits to suit the gcc build machinery.
-- Alan Modra IBM OzLabs - Linux Technology Centre