Re: [PATCH] gimple-range-op: Use widen method even for sqrt reverse op [PR126534]

On 2026-08-07 17:30, Jakub Jelinek wrote:
> Hi!
> 
> For glibc the target hook returns 0 for sqrt precision (i.e. 0.5ulp
> precise), but we didn't apply the +-0.5ulp widening when doing
> reverse op and so still came up with a wrong range, even with 0.5ulp
> precision certain values around 25.0 result in sqrt (x) being 5.0.
> Even if the hook returns non-zero, I'd say more reasonable handling
> of it is that it is say worst case 1ulp (or 2ulps etc.) from the
> correct result (i.e. half ulp precise), rather than just 1ulp from
> mathematically exact result.
> 
> So, the following patch uses the newly added widen method to add
> this 0.5ulp before we feed it into frange_arithmetic.
> 
> Ok for trunk if it passes bootstrap/regtest?
> 
> 2026-08-07  Jakub Jelinek  <[email protected]>
> 
> 	PR tree-optimization/126534
> 	* gimple-range-op.cc (cfn_sqrt::op1_range): Widen lb or ub
> 	by further 0.5ulp or 1ulp before squaring it.
> 
> 	* gcc.dg/pr126534.c: New test.
> 
> --- gcc/gimple-range-op.cc.jj	2026-03-27 10:17:14.167330201 +0100
> +++ gcc/gimple-range-op.cc	2026-08-07 17:11:25.720918034 +0200
> @@ -573,7 +573,10 @@ public:
>  	  frange_nextafter (TYPE_MODE (type), lb, dconstninf);
>  	if (real_less (&dconst0, &lb))
>  	  {
> -	    REAL_VALUE_TYPE op = lb;
> +	    frange wlhs (type, lb, dconstinf);
> +	    wlhs.clear_nan ();
> +	    wlhs.widen (type);
> +	    REAL_VALUE_TYPE op = wlhs.lower_bound ();
>  	    frange_arithmetic (MULT_EXPR, type, lb, op, op, dconstninf);
>  	  }
>  	else
> @@ -587,7 +590,10 @@ public:
>  	  frange_nextafter (TYPE_MODE (type), ub, dconstinf);
>  	if (real_isfinite (&ub))
>  	  {
> -	    REAL_VALUE_TYPE op = ub;
> +	    frange wlhs (type, dconstninf, ub);
> +	    wlhs.clear_nan ();
> +	    wlhs.widen (type);
> +	    REAL_VALUE_TYPE op = wlhs.upper_bound ();
>  	    frange_arithmetic (MULT_EXPR, type, ub, op, op, dconstinf);
>  	  }
>  	else
> --- gcc/testsuite/gcc.dg/pr126534.c.jj	2026-08-07 17:16:40.246010859 
> +0200
> +++ gcc/testsuite/gcc.dg/pr126534.c	2026-08-07 17:16:24.296208995 +0200
> @@ -0,0 +1,45 @@
> +/* PR tree-optimization/126534 */
> +/* { dg-do run } */
> +/* { dg-options "-O2" } */
> +
> +[[gnu::noipa]] int
> +foo (double x)
> +{
> +  double y = __builtin_sqrt (x);
> +  if (y <= 5.0)
> +    {
> +      if (x > 25.0)
> +	return 1;
> +      return 2;
> +    }
> +  return 3;
> +}
> +
> +[[gnu::noipa]] int
> +bar (double x)
> +{
> +  double y = __builtin_sqrt (x);
> +  if (y >= 5.0)
> +    {
> +      if (x < 25.0)
> +	return 1;
> +      return 2;
> +    }
> +  return 3;
> +}
> +
> +int
> +main ()
> +{
> +#if __DBL_MANT_DIG__ == 53 && __DBL_MAX_EXP__ == 1024 &&
> __FLT_EVAL_METHOD__ == 0
> +  volatile double x = 0x1.9000000000001p+4;
> +  volatile double y = 0x1.8ffffffffffffp+4;
> +  if (__builtin_sqrt (x) == 5.0 && __builtin_sqrt (y) == 5.0)
> +    {
> +      if (foo (0x1.9000000000001p+4) != 1)    /* 25.000000000000004 */
> +	__builtin_abort ();
> +      if (bar (0x1.8ffffffffffffp+4) != 1)    /* 24.999999999999996 */
> +	__builtin_abort ();
> +    }
> +#endif
> +}
> 
> 	Jakub

LGTM