[PATCH] -fpic problems on PPC, a different approach

[Geoff Keating]

> From: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
> Date: Wed, 9 Jun 1999 18:57:10 +0200
> Cc: law@cygnus.com, Geoff Keating <geoffk@ozemail.com.au>,
>         egcs-patches@egcs.cygnus.com
> 
> >> +/*
> >>  #define PREFERRED_RELOAD_CLASS(X,CLASS)			\
> >>    ((GET_CODE (X) == CONST_DOUBLE			\
> >>      && GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT)	\
> >>     ? NO_REGS : (CLASS))
> >> +*/
> >> +#define PREFERRED_RELOAD_CLASS(X,CLASS)                \
> >> +  (CONSTANT_P (X)                                      \
> >> +   && ((CLASS) == FLOAT_REGS                           \
> >> +       || (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
> >> +          && (HOST_FLOAT_FORMAT != IEEE_FLOAT_FORMAT   \
> >> +              || HOST_BITS_PER_INT != BITS_PER_WORD))) \
> >> +   ? NO_REGS : (CLASS))
> >
> >Another stray change?
> 
> Hmm, I overlooked this one too, but maybe someone should comment on the
> proposed change. It's a change from Geoff that is independent of his old reload
> patch and he wrote back then that he adapted some code from sparc.h. But AFAIK
> nobody ever commented on this.

Yes.  I think the change is correct, but it really has nothing to do
with -fpic.  I needed it because my change turns off some of reload,
which stresses PREFERRED_RELOAD_CLASS more than usual.

-- 
Geoffrey Keating <geoffk@ozemail.com.au>