Re: [patch] m68k: don't try to push bytes

[Graham Stott <grahams at redhat dot com>]

DJ Delorie wrote:
> 
> There's a comment in m68k.c:output_move_qimode that says:
> 
>   /* This is probably useless, since it loses for pushing a struct
>      of several bytes a byte at a time.  */
> 
> Well, the execute/931004-* tests fail because of that, so this tells
> gcc to avoid using byte pushes.  OK to apply?
> 
> 2001-02-28  DJ Delorie  <dj@redhat.com>
> 
>         * config/m68k/m68k.h (MOVE_BY_PIECES_P): Avoid pushing bytes,
>         since that doesn't work the way gcc wants on a generic m68k.
> 
> Index: config/m68k/m68k.h
> ===================================================================
> RCS file: /cvs/gcc/gcc/gcc/config/m68k/m68k.h,v
> retrieving revision 1.48
> diff -p -3 -r1.48 m68k.h
> *** m68k.h      2001/01/01 20:35:31     1.48
> --- m68k.h      2001/02/28 20:27:17
> *************** extern enum reg_class regno_reg_class[];
> *** 921,926 ****
> --- 921,931 ----
>      On the 5200 (coldfire), sp@- in a byte insn pushes just a byte.  */
>   #define PUSH_ROUNDING(BYTES) (TARGET_5200 ? BYTES : ((BYTES) + 1) & ~1)
> 
> + /* We want to avoid trying to push bytes. */
> + #define MOVE_BY_PIECES_P(SIZE, ALIGN) \
> +   (move_by_pieces_ninsns (SIZE, ALIGN) < MOVE_RATIO) \
> +     && ((SIZE >=16 && ALIGN >= 16) || TARGET_5200)
> +
Suggest you parenthesis the macro parameters to avoid surprises.

>   /* Offset of first parameter from the argument pointer register value.  */
>   #define FIRST_PARM_OFFSET(FNDECL) 8
> 

Graham