[patch RFA] Fix sh64 build failure after dataflow merge

[Seongbae Park (박성배, 朴成培)]

On 6/17/07, Kaz Kojima <kkojima@rr.iij4u.or.jp> wrote:
> The trunk fails to build on sh64-* after dataflow merge.
> I applied the patch
>
>   http://gcc.gnu.org/ml/gcc-patches/2007-05/msg01775.html
>
> on dataflow branch, but clearly I've tested it wrongly for sh64-*.
> Sorry for my silly mistake.  The attached one liner fixes the order
> of arguments to validate_replace_rtx in bt-load.c:move_btr_def.
> It's tested with bootstrap and "make -k check" on i686-pc-linux-gnu
> and fixes the build failure for sh64-unknown-linux-gnu.
>
> Regards,
>         kaz
> --
> :ADDPATCH rtl-optimization:
>
> 2007-06-17  Kaz Kojima  <kkojima@gcc.gnu.org>
>
>         * bt-load.c (move_btr_def): Fix the order of arguments
>         to validate_replace_rtx.
>
> --- ORIG/trunk/gcc/bt-load.c    Tue Jun 12 09:34:43 2007
> +++ LOCAL/trunk/gcc/bt-load.c   Sun Jun 17 11:47:50 2007
> @@ -1228,7 +1228,7 @@ move_btr_def (basic_block new_def_bb, in
>         replacement_rtx = btr_rtx;
>        else
>         replacement_rtx = gen_rtx_REG (GET_MODE (user->use), btr);
> -      validate_replace_rtx (user->insn, user->use, replacement_rtx);
> +      validate_replace_rtx (user->use, replacement_rtx, user->insn);
>        user->use = replacement_rtx;
>      }
>  }

I'm not sure I can approve this patch,
but this is definitely ok from my point of view
 (methinks this qualifies as obvious but others may beg to differ).
-- 
#pragma ident "Seongbae Park, compiler, http://seongbae.blogspot.com"