[mips] Change the definition of macro SHIFT_COUNT_TRUNCATED

Richard Sandiford rdsandiford@googlemail.com
Thu Sep 2 18:32:00 GMT 2010


Mingjie Xing <mingjie.xing@gmail.com> writes:
>         * config/mips/mips.h (SHIFT_COUNT_TRUNCATED): Change the definition.
>         * config/mips/mips.c (mips_shift_truncation_mask): Implement
>           TARGET_SHIFT_TRUNCATION_MASK.

I think a more canonical version would be:

	* config/mips/mips.h (SHIFT_COUNT_TRUNCATED): Change the definition.
	* config/mips/mips.c (mips_shift_truncation_mask): New function.
	(TARGET_SHIFT_TRUNCATION_MASK): Define.

> Index: config/mips/mips.c
> ===================================================================
> --- config/mips/mips.c	(revision 163495)
> +++ config/mips/mips.c	(working copy)
> @@ -16335,6 +16335,20 @@ void mips_function_profiler (FILE *file)
>      fprintf (file, "\tmove\t%s,%s\n", reg_names[STATIC_CHAIN_REGNUM],
>  	     reg_names[2]);
>  }
> +
> +/* Implement TARGET_SHIFT_TRUNCATION_MASK. We want to keep the current
> +   behaviour of TARGET_SHIFT_TRUNCATION_MASK for non-vector modes even
> +   when TARGET_LOONGSON_2EF is true. */

s/current/default/, since we're writing this for future readers.
Coding style says there should be two spaces after each sentence, so:

/* Implement TARGET_SHIFT_TRUNCATION_MASK.  We want to keep the default
   behaviour of TARGET_SHIFT_TRUNCATION_MASK for non-vector modes even
   when TARGET_LOONGSON_2EF is true.  */

>  /* Define this to be nonzero if shift instructions ignore all but the low-order
>     few bits.  */
> -#define SHIFT_COUNT_TRUNCATED 1
> +#define SHIFT_COUNT_TRUNCATED (TARGET_LOONGSON_2EF ? 0 : 1)

(!TARGET_LOONGSON_2EF)

Please also document why we're doing this.  Let's do that by removing
the current "Define this..." comment -- it's useless boilerplate --
and replacing it with:

/* Standard MIPS integer shifts truncate the shift amount to the
   width of the shifted operand.  However, Loongson vector shifts
   do not truncate the shift amount at all.  */

OK with those changes, thanks.

Richard



More information about the Gcc-patches mailing list