public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <richard.guenther@gmail.com>
To: Aldy Hernandez <aldyh@redhat.com>
Cc: GCC patches <gcc-patches@gcc.gnu.org>, Jakub Jelinek <jakub@redhat.com>
Subject: Re: [COMMITTED] Improve union of ranges containing NAN.
Date: Wed, 31 Aug 2022 08:22:27 +0200	[thread overview]
Message-ID: <CAFiYyc2wo_QndrUq71Yu0SOS=6HxX3eRFZ1uDSJBSGXdOA47zw@mail.gmail.com> (raw)
In-Reply-To: <20220830154301.1965174-1-aldyh@redhat.com>

On Tue, Aug 30, 2022 at 5:43 PM Aldy Hernandez via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> Previously [5,6] U NAN would just drop to VARYING.  With this patch,
> the resulting range becomes [5,6] with the NAN bit set to unknown.
>
> [I still have yet to decide what to do with intersections.  ISTM, the
> intersection of a known NAN with anything else should be a NAN, but it
> could also be undefined (the empty set).  I'll have to run some tests
> and see.  Currently, we drop to VARYING cause well... it's always safe
> to give up;-).]

Giving up in too many places will hide bugs in others for too long ...

> gcc/ChangeLog:
>
>         * value-range.cc (early_nan_resolve): Change comment.
>         (frange::union_): Handle union when one side is a NAN.
>         (range_tests_nan): Add tests for NAN union.
> ---
>  gcc/value-range.cc | 44 ++++++++++++++++++++++++++++++++++----------
>  1 file changed, 34 insertions(+), 10 deletions(-)
>
> diff --git a/gcc/value-range.cc b/gcc/value-range.cc
> index b6d6c62c06c..473139c6dbd 100644
> --- a/gcc/value-range.cc
> +++ b/gcc/value-range.cc
> @@ -403,7 +403,7 @@ early_nan_resolve (frange &r, const frange &other)
>    // There's nothing to do for both NANs.
>    if (r.get_nan ().yes_p () == other.get_nan ().yes_p ())
>      return false;
> -  // But only one NAN complicates things.
> +  // ?? Perhaps the intersection of a NAN and anything is a NAN ??.
>    r.set_varying (r.type ());
>    return true;
>  }
> @@ -420,10 +420,22 @@ frange::union_ (const vrange &v)
>        *this = r;
>        return true;
>      }
> -  // ?? We could do better here.  [5,6] U NAN should be [5,6] with the
> -  // NAN maybe bits set.  For now, this is conservatively correct.
> -  if (get_nan ().yes_p () || r.get_nan ().yes_p ())
> -    return early_nan_resolve (*this, r);
> +
> +  // If one side has a NAN, the union is just the other side plus the
> +  // NAN bit.
> +  if (get_nan ().yes_p ())
> +    {
> +      *this = r;
> +      // NOP if NAN already set.
> +      set_nan (fp_prop::VARYING);
> +      return true;
> +    }
> +  if (r.get_nan ().yes_p ())
> +    {
> +      // NOP if NAN already set.
> +      set_nan (fp_prop::VARYING);
> +      return true;
> +    }
>
>    bool changed = m_props.union_ (r.m_props);
>
> @@ -3520,6 +3532,7 @@ static void
>  range_tests_nan ()
>  {
>    frange r0, r1;
> +  REAL_VALUE_TYPE q, r;
>
>    // Equal ranges but with differing NAN bits are not equal.
>    r1 = frange_float ("10", "12");
> @@ -3537,13 +3550,24 @@ range_tests_nan ()
>    ASSERT_FALSE (r0 == r0);
>    ASSERT_TRUE (r0 != r0);
>
> -  // Make sure that combining NAN and INF doesn't give any crazy results.
> +  // [5,6] U NAN is [5,6] with an unknown NAN bit.
> +  r0 = frange_float ("5", "6");
> +  r0.set_nan (fp_prop::NO);
> +  r1 = frange_nan (float_type_node);
> +  r0.union_ (r1);
> +  real_from_string (&q, "5");
> +  real_from_string (&r, "6");
> +  ASSERT_TRUE (real_identical (&q, &r0.lower_bound ()));
> +  ASSERT_TRUE (real_identical (&r, &r0.upper_bound ()));
> +  ASSERT_TRUE (r0.get_nan ().varying_p ());
> +
> +  // NAN U NAN = NAN
>    r0 = frange_nan (float_type_node);
> -  ASSERT_TRUE (r0.get_nan ().yes_p ());
> -  r1 = frange_float ("+Inf", "+Inf");
> +  r1 = frange_nan (float_type_node);
>    r0.union_ (r1);
> -  // [INF, INF] U NAN = VARYING
> -  ASSERT_TRUE (r0.varying_p ());
> +  ASSERT_TRUE (real_isnan (&r0.lower_bound ()));
> +  ASSERT_TRUE (real_isnan (&r1.upper_bound ()));
> +  ASSERT_TRUE (r0.get_nan ().yes_p ());
>
>    // [INF, INF] ^ NAN = VARYING
>    r0 = frange_nan (float_type_node);
> --
> 2.37.1
>

  reply	other threads:[~2022-08-31  6:22 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-30 15:43 Aldy Hernandez
2022-08-31  6:22 ` Richard Biener [this message]
2022-08-31  6:25   ` Aldy Hernandez

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAFiYyc2wo_QndrUq71Yu0SOS=6HxX3eRFZ1uDSJBSGXdOA47zw@mail.gmail.com' \
    --to=richard.guenther@gmail.com \
    --cc=aldyh@redhat.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jakub@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).