[flang-commits] [PATCH] D114027: [flang] Fix off-by-one results from folding MAXEXPONENT and MINEXPONENT

[Peter Klausler via Phabricator via flang-commits]

klausler created this revision.
klausler added a reviewer: mleair.
klausler added a project: Flang.
Herald added a subscriber: jdoerfert.
klausler requested review of this revision.

Every one was too low by one.


https://reviews.llvm.org/D114027

Files:
  flang/include/flang/Evaluate/real.h
  flang/test/Evaluate/folding07.f90


Index: flang/test/Evaluate/folding07.f90
===================================================================
--- flang/test/Evaluate/folding07.f90
+++ flang/test/Evaluate/folding07.f90
@@ -204,12 +204,12 @@
     max8 = maxexponent(0._8), &
     max10 = maxexponent(0._10), &
     max16 = maxexponent(0._16)
-  logical, parameter :: test_max2 = max2 == 15
-  logical, parameter :: test_max3 = max3 == 127
-  logical, parameter :: test_max4 = max4 == 127
-  logical, parameter :: test_max8 = max8 == 1023
-  logical, parameter :: test_max10 = max10 == 16383
-  logical, parameter :: test_max16 = max16 == 16383
+  logical, parameter :: test_max2 = max2 == 16
+  logical, parameter :: test_max3 = max3 == 128
+  logical, parameter :: test_max4 = max4 == 128
+  logical, parameter :: test_max8 = max8 == 1024
+  logical, parameter :: test_max10 = max10 == 16384
+  logical, parameter :: test_max16 = max16 == 16384
 
   integer, parameter :: &
     min2 = minexponent(0._2), &
@@ -218,12 +218,12 @@
     min8 = minexponent(0._8), &
     min10 = minexponent(0._10), &
     min16 = minexponent(0._16)
-  logical, parameter :: test_min2 = min2 == -14
-  logical, parameter :: test_min3 = min3 == -126
-  logical, parameter :: test_min4 = min4 == -126
-  logical, parameter :: test_min8 = min8 == -1022
-  logical, parameter :: test_min10 = min10 == -16382
-  logical, parameter :: test_min16 = min16 == -16382
+  logical, parameter :: test_min2 = min2 == -13
+  logical, parameter :: test_min3 = min3 == -125
+  logical, parameter :: test_min4 = min4 == -125
+  logical, parameter :: test_min8 = min8 == -1021
+  logical, parameter :: test_min10 = min10 == -16381
+  logical, parameter :: test_min16 = min16 == -16381
 
   integer, parameter :: &
     irange1 = range(0_1), &
Index: flang/include/flang/Evaluate/real.h
===================================================================
--- flang/include/flang/Evaluate/real.h
+++ flang/include/flang/Evaluate/real.h
@@ -150,8 +150,8 @@
   static constexpr int DIGITS{binaryPrecision};
   static constexpr int PRECISION{Details::decimalPrecision};
   static constexpr int RANGE{Details::decimalRange};
-  static constexpr int MAXEXPONENT{maxExponent - 1 - exponentBias};
-  static constexpr int MINEXPONENT{1 - exponentBias};
+  static constexpr int MAXEXPONENT{maxExponent - exponentBias};
+  static constexpr int MINEXPONENT{2 - exponentBias};
 
   constexpr Real FlushSubnormalToZero() const {
     if (IsSubnormal()) {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D114027.387750.patch
Type: text/x-patch
Size: 2466 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/flang-commits/attachments/20211116/6b5bcab9/attachment.bin>