public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r13-3453] Check HONOR_NANS instead of flag_finite_math_only in frange:verify_range.
@ 2022-10-24 10:16 Aldy Hernandez
0 siblings, 0 replies; only message in thread
From: Aldy Hernandez @ 2022-10-24 10:16 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:1e9d9ed095df3d064cf9d91d46f3e5426c2a05a7
commit r13-3453-g1e9d9ed095df3d064cf9d91d46f3e5426c2a05a7
Author: Aldy Hernandez <aldyh@redhat.com>
Date: Sun Oct 23 16:51:17 2022 +0200
Check HONOR_NANS instead of flag_finite_math_only in frange:verify_range.
[Jakub and other FP experts, would this be OK, or am I missing
something?]
Vax does not seem to have !flag_finite_math_only, but float_type_node
does not HONOR_NANS. The check in frange::verify_range dependend on
flag_finite_math_only, which is technically not correct since
frange::set_varying() checks HONOR_NANS instead of
flag_finite_math_only.
I'm actually getting tired of flag_finite_math_only and
!flag_finite_math_only discrepancies in the selftests (Vax and rx-elf
come to mind). I think we should just test both alternatives in the
selftests as in this patch.
We could also check flag_finite_math_only=0 with a float_type_node
that does not HONOR_NANs, but I have no idea how to twiddle
FLOAT_MODE_FORMAT temporarily, and that may be over thinking it.
PR tree-optimization/107365
gcc/ChangeLog:
* value-range.cc (frange::verify_range): Predicate NAN check in
VARYING range on HONOR_NANS instead of flag_finite_math_only.
(range_tests_floats): Same.
(range_tests_floats_various): New.
(range_tests): Call range_tests_floats_various.
Diff:
---
gcc/value-range.cc | 33 +++++++++++++++++++++++++--------
1 file changed, 25 insertions(+), 8 deletions(-)
diff --git a/gcc/value-range.cc b/gcc/value-range.cc
index d779e9819e2..d8ee6ec0d0f 100644
--- a/gcc/value-range.cc
+++ b/gcc/value-range.cc
@@ -720,13 +720,13 @@ frange::verify_range ()
gcc_checking_assert (!m_type);
return;
case VR_VARYING:
- if (flag_finite_math_only)
- gcc_checking_assert (!m_pos_nan && !m_neg_nan);
- else
- gcc_checking_assert (m_pos_nan && m_neg_nan);
gcc_checking_assert (m_type);
gcc_checking_assert (frange_val_is_min (m_min, m_type));
gcc_checking_assert (frange_val_is_max (m_max, m_type));
+ if (HONOR_NANS (m_type))
+ gcc_checking_assert (m_pos_nan && m_neg_nan);
+ else
+ gcc_checking_assert (!m_pos_nan && !m_neg_nan);
return;
case VR_RANGE:
gcc_checking_assert (m_type);
@@ -3957,10 +3957,9 @@ range_tests_floats ()
// A range of [-INF,+INF] is actually VARYING if no other properties
// are set.
r0 = frange_float ("-Inf", "+Inf");
- if (r0.maybe_isnan ())
- ASSERT_TRUE (r0.varying_p ());
+ ASSERT_TRUE (r0.varying_p ());
// ...unless it has some special property...
- if (!flag_finite_math_only)
+ if (HONOR_NANS (r0.type ()))
{
r0.clear_nan ();
ASSERT_FALSE (r0.varying_p ());
@@ -4041,6 +4040,24 @@ range_tests_floats ()
}
}
+// Run floating range tests for various combinations of NAN and INF
+// support.
+
+static void
+range_tests_floats_various ()
+{
+ int save_finite_math_only = flag_finite_math_only;
+
+ // Test -ffinite-math-only.
+ flag_finite_math_only = 1;
+ range_tests_floats ();
+ // Test -fno-finite-math-only.
+ flag_finite_math_only = 0;
+ range_tests_floats ();
+
+ flag_finite_math_only = save_finite_math_only;
+}
+
void
range_tests ()
{
@@ -4049,7 +4066,7 @@ range_tests ()
range_tests_int_range_max ();
range_tests_strict_enum ();
range_tests_nonzero_bits ();
- range_tests_floats ();
+ range_tests_floats_various ();
range_tests_misc ();
}
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2022-10-24 10:16 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-24 10:16 [gcc r13-3453] Check HONOR_NANS instead of flag_finite_math_only in frange:verify_range Aldy Hernandez
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).