[patch] Fix PR42621, "Computed gotos on AMD 800% slower"

[Richard Guenther]

On Thu, Jan 7, 2010 at 6:58 PM, Steven Bosscher <stevenb.gcc@gmail.com> wrote:
> Hello,
>
> Attached patch fixes PR42621 by going back to the state of things from
> before GCC 4.4. Unfactoring computed gotos based on whether we are
> optimizing a basic block for speed or size doesn't really make no
> sense, especially if we have to work with a guessed profile.
>
> Bootstrapped+tested on {ia64,x86_64}-unknown-linux-gnu. OK?

Ok.

Thanks,
Richard.

> Ciao!
> Steven
>
>
>
>
>        PR rtl-optimization/42621
>        * bb-reorder.c (gate_duplicated_computed_gotos): Only run if not
>        optimizing for size.
>        (duplicate_computed_gotos): Remove now-redundant check.
>
> Index: bb-reorder.c
> ===================================================================
> --- bb-reorder.c        (revision 155697)
> +++ bb-reorder.c        (working copy)
> @@ -1981,7 +1981,9 @@ gate_duplicate_computed_gotos (void)
>  {
>   if (targetm.cannot_modify_jumps_p ())
>     return false;
> -  return (optimize > 0 && flag_expensive_optimizations);
> +  return (optimize > 0
> +         && flag_expensive_optimizations
> +         && ! optimize_function_for_size_p (cfun));
>  }
>
>
> @@ -2072,9 +2074,6 @@ duplicate_computed_gotos (void)
>          || single_pred_p (single_succ (bb)))
>        continue;
>
> -      if (!optimize_bb_for_size_p (bb))
> -       continue;
> -
>       /* The successor block has to be a duplication candidate.  */
>       if (!bitmap_bit_p (candidates, single_succ (bb)->index))
>        continue;
>