[libc++][numeric] P0543R3: Saturation arithmetic #77967

H-G-Hristov · 2024-01-12T19:25:39Z

https://eel.is/c++draft/numeric.sat

Additional references:

Division: https://eel.is/c++draft/expr.mul#4
Arithmetic conversions: https://eel.is/c++draft/expr.arith.conv#1
Clang builtins: https://clang.llvm.org/docs/LanguageExtensions.html#builtin-functions

Depends on: #78086

github-actions · 2024-01-12T19:27:58Z

✅ With the latest revision this PR passed the C/C++ code formatter.

Implements: https://wg21.link/P0543R3 - https://eel.is/c++draft/numeric.sat Additional notes: - Division: https://eel.is/c++draft/expr.mul#4 - Arithmetic conversions: https://eel.is/c++draft/expr.arith.conv#1.5 - Clang builtins: https://clang.llvm.org/docs/LanguageExtensions.html#builtin-functions

llvmbot · 2024-01-12T22:46:03Z

@llvm/pr-subscribers-libcxx

Author: Hristo Hristov (H-G-Hristov)

Changes

Implements: https://wg21.link/P0543R3

https://eel.is/c++draft/numeric.sat

Additional references:

Division: https://eel.is/c++draft/expr.mul#4
Arithmetic conversions: https://eel.is/c++draft/expr.arith.conv#1
Clang builtins: https://clang.llvm.org/docs/LanguageExtensions.html#builtin-functions

Patch is 39.88 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/77967.diff

23 Files Affected:

(modified) libcxx/docs/FeatureTestMacroTable.rst (+1-1)
(modified) libcxx/docs/ReleaseNotes/18.rst (+4-3)
(modified) libcxx/docs/Status/Cxx2cPapers.csv (+1-1)
(modified) libcxx/include/CMakeLists.txt (+1)
(added) libcxx/include/__numeric/saturation_arithmetic.h (+119)
(modified) libcxx/include/module.modulemap.in (+1)
(modified) libcxx/include/numeric (+13)
(modified) libcxx/include/version (+1-1)
(modified) libcxx/modules/std/numeric.inc (+10)
(modified) libcxx/test/std/language.support/support.limits/support.limits.general/numeric.version.compile.pass.cpp (+5-11)
(modified) libcxx/test/std/language.support/support.limits/support.limits.general/version.version.compile.pass.cpp (+5-11)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp (+98)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.verify.cpp (+39)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.assert.pass.cpp (+53)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp (+98)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.verify.cpp (+39)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/mul_sat.pass.cpp (+105)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/mul_sat.verify.cpp (+39)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_cast.pass.cpp (+109)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_cast.verify.cpp (+44)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/sub_sat.pass.cpp (+93)
(added) libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/sub_sat.verify.cpp (+39)
(modified) libcxx/utils/generate_feature_test_macro_components.py (+1-4)

diff --git a/libcxx/docs/FeatureTestMacroTable.rst b/libcxx/docs/FeatureTestMacroTable.rst
index 893a3b13ca06e0..9dd9c0c023bc8a 100644
--- a/libcxx/docs/FeatureTestMacroTable.rst
+++ b/libcxx/docs/FeatureTestMacroTable.rst
@@ -432,7 +432,7 @@ Status
     --------------------------------------------------- -----------------
     ``__cpp_lib_rcu``                                   *unimplemented*
     --------------------------------------------------- -----------------
-    ``__cpp_lib_saturation_arithmetic``                 *unimplemented*
+    ``__cpp_lib_saturation_arithmetic``                 ``202311L``
     --------------------------------------------------- -----------------
     ``__cpp_lib_smart_ptr_owner_equality``              *unimplemented*
     --------------------------------------------------- -----------------
diff --git a/libcxx/docs/ReleaseNotes/18.rst b/libcxx/docs/ReleaseNotes/18.rst
index 6de7d07e454d34..877e387d9280fd 100644
--- a/libcxx/docs/ReleaseNotes/18.rst
+++ b/libcxx/docs/ReleaseNotes/18.rst
@@ -54,12 +54,13 @@ Implemented Papers
 - P2905R2 - Runtime format strings
 - P2918R2 - Runtime format strings II
 - P2871R3 - Remove Deprecated Unicode Conversion Facets from C++26
-- P2870R3 - Remove basic_string::reserve()
+- P2870R3 - Remove ``basic_string::reserve()``
 - P2909R4 - Fix formatting of code units as integers (Dude, where’s my ``char``?)
-- P2821R5 - span.at()
-- P0521R0 - Proposed Resolution for CA 14 (shared_ptr use_count/unique)
+- P2821R5 - ``span.at()``
+- P0521R0 - Proposed Resolution for CA 14 (``shared_ptr`` ``use_count/unique``)
 - P1759R6 - Native handles and file streams
 - P2517R1 - Add a conditional ``noexcept`` specification to ``std::apply``
+- P0543R3 - Saturation arithmetic
 
 
 Improvements and New Features
diff --git a/libcxx/docs/Status/Cxx2cPapers.csv b/libcxx/docs/Status/Cxx2cPapers.csv
index 5701717f39766c..b38b3028863fee 100644
--- a/libcxx/docs/Status/Cxx2cPapers.csv
+++ b/libcxx/docs/Status/Cxx2cPapers.csv
@@ -27,7 +27,7 @@
 "`P2714R1 <https://wg21.link/P2714R1>`__","LWG","Bind front and back to NTTP callables","Varna June 2023","","",""
 "`P2630R4 <https://wg21.link/P2630R4>`__","LWG","``submdspan``","Varna June 2023","","",""
 "","","","","","",""
-"`P0543R3 <https://wg21.link/P0543R3>`__","LWG","Saturation arithmetic","Kona November 2023","","",""
+"`P0543R3 <https://wg21.link/P0543R3>`__","LWG","Saturation arithmetic","Kona November 2023","|Complete|","18.0",""
 "`P2407R5 <https://wg21.link/P2407R5>`__","LWG","Freestanding Library: Partial Classes","Kona November 2023","","",""
 "`P2546R5 <https://wg21.link/P2546R5>`__","LWG","Debugging Support","Kona November 2023","","",""
 "`P2905R2 <https://wg21.link/P2905R2>`__","LWG","Runtime format strings","Kona November 2023","|Complete|","18.0","|format| |DR|"
diff --git a/libcxx/include/CMakeLists.txt b/libcxx/include/CMakeLists.txt
index 0fe3ab44d2466e..c4d8a9f092de14 100644
--- a/libcxx/include/CMakeLists.txt
+++ b/libcxx/include/CMakeLists.txt
@@ -569,6 +569,7 @@ set(files
   __numeric/pstl_reduce.h
   __numeric/pstl_transform_reduce.h
   __numeric/reduce.h
+  __numeric/saturation_arithmetic.h
   __numeric/transform_exclusive_scan.h
   __numeric/transform_inclusive_scan.h
   __numeric/transform_reduce.h
diff --git a/libcxx/include/__numeric/saturation_arithmetic.h b/libcxx/include/__numeric/saturation_arithmetic.h
new file mode 100644
index 00000000000000..6d13ec5b3fd5bc
--- /dev/null
+++ b/libcxx/include/__numeric/saturation_arithmetic.h
@@ -0,0 +1,119 @@
+// -*- C++ -*-
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP___NUMERIC_SATURATION_ARITHMETIC_H
+#define _LIBCPP___NUMERIC_SATURATION_ARITHMETIC_H
+
+#include <__concepts/arithmetic.h>
+#include <__config>
+#include <__type_traits/decay.h>
+#include <__utility/cmp.h>
+#include <limits>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#  pragma GCC system_header
+#endif
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+#if _LIBCPP_STD_VER >= 26
+
+template <typename _Tp>
+concept __libcpp_standard_integer = __libcpp_unsigned_integer<decay_t<_Tp>> || __libcpp_signed_integer<decay_t<_Tp>>;
+
+template <__libcpp_standard_integer _Tp>
+_LIBCPP_HIDE_FROM_ABI constexpr _Tp add_sat(_Tp __x, _Tp __y) noexcept {
+  if (_Tp __sum; !__builtin_add_overflow(__x, __y, &__sum))
+    return __sum;
+  // Handle overflow
+  if constexpr (__libcpp_unsigned_integer<_Tp>) {
+    return std::numeric_limits<_Tp>::max();
+  } else {
+    // Signed addition overflow
+    if (__x > 0)
+      // Overflows if (x > 0 && y > 0)
+      return std::numeric_limits<_Tp>::max();
+    else
+      // Overflows if  (x < 0 && y < 0)
+      return std::numeric_limits<_Tp>::min();
+  }
+}
+
+template <__libcpp_standard_integer _Tp>
+_LIBCPP_HIDE_FROM_ABI constexpr _Tp sub_sat(_Tp __x, _Tp __y) noexcept {
+  if (_Tp __sub; !__builtin_sub_overflow(__x, __y, &__sub))
+    return __sub;
+  // Handle overflow
+  if constexpr (__libcpp_unsigned_integer<_Tp>) {
+    // Overflows if (x < y)
+    return std::numeric_limits<_Tp>::min();
+  } else {
+    // Signed subtration overflow
+    if (__x > 0)
+      // Overflows if (x > 0 && y < 0)
+      return std::numeric_limits<_Tp>::max();
+    else
+      // Overflows if (x < 0 && y > 0)
+      return std::numeric_limits<_Tp>::min();
+  }
+}
+
+template <__libcpp_standard_integer _Tp>
+_LIBCPP_HIDE_FROM_ABI constexpr _Tp mul_sat(_Tp __x, _Tp __y) noexcept {
+  if (_Tp __mul; !__builtin_mul_overflow(__x, __y, &__mul))
+    return __mul;
+  // Handle overflow
+  if constexpr (__libcpp_unsigned_integer<_Tp>) {
+    return std::numeric_limits<_Tp>::max();
+  } else {
+    // Signed multiplication overflow
+    if (__x > 0) {
+      if (__y > 0)
+        // Overflows if (x > 0 && y > 0)
+        return std::numeric_limits<_Tp>::max();
+      // Overflows if (x > 0 && y < 0)
+      return std::numeric_limits<_Tp>::min();
+    }
+    if (__y > 0)
+      // Overflows if (x < 0 && y > 0)
+      return std::numeric_limits<_Tp>::min();
+    // Overflows if (x < 0 && y < 0)
+    return std::numeric_limits<_Tp>::max();
+  }
+}
+
+template <__libcpp_standard_integer _Tp>
+_LIBCPP_HIDE_FROM_ABI constexpr _Tp div_sat(_Tp __x, _Tp __y) noexcept {
+  _LIBCPP_ASSERT_UNCATEGORIZED(__y != 0, "Division by 0 is undefined");
+  if constexpr (__libcpp_unsigned_integer<_Tp>) {
+    return __x / __y;
+  } else {
+    // Handle signed division overflow
+    if (__x == std::numeric_limits<_Tp>::min() && __y == _Tp{-1})
+      return std::numeric_limits<_Tp>::max();
+    return __x / __y;
+  }
+}
+
+template <__libcpp_standard_integer _Rp, __libcpp_standard_integer _Tp>
+_LIBCPP_HIDE_FROM_ABI constexpr _Rp saturate_cast(_Tp __x) noexcept {
+  // Handle overflow
+  if (std::cmp_less_equal(__x, std::numeric_limits<_Rp>::min()))
+    return std::numeric_limits<_Rp>::min();
+  if (std::cmp_greater_equal(__x, std::numeric_limits<_Rp>::max()))
+    return std::numeric_limits<_Rp>::max();
+  // No overflow
+  return static_cast<_Rp>(__x);
+}
+
+#endif // _LIBCPP_STD_VER >= 26
+
+_LIBCPP_END_NAMESPACE_STD
+
+#endif // _LIBCPP___NUMERIC_SATURATION_ARITHMETIC_H
diff --git a/libcxx/include/module.modulemap.in b/libcxx/include/module.modulemap.in
index d10670d4faaffc..194a74a1e07b14 100644
--- a/libcxx/include/module.modulemap.in
+++ b/libcxx/include/module.modulemap.in
@@ -1580,6 +1580,7 @@ module std_private_numeric_pstl_transform_reduce    [system] {
   export *
 }
 module std_private_numeric_reduce                   [system] { header "__numeric/reduce.h" }
+module std_private_numeric_saturation_arithmetic    [system] { header "__numeric/saturation_arithmetic.h" }
 module std_private_numeric_transform_exclusive_scan [system] { header "__numeric/transform_exclusive_scan.h" }
 module std_private_numeric_transform_inclusive_scan [system] { header "__numeric/transform_inclusive_scan.h" }
 module std_private_numeric_transform_reduce         [system] { header "__numeric/transform_reduce.h" }
diff --git a/libcxx/include/numeric b/libcxx/include/numeric
index d09d0a81fcc03a..0fe7115f1c666e 100644
--- a/libcxx/include/numeric
+++ b/libcxx/include/numeric
@@ -140,6 +140,18 @@ template<class T>
 template<class T>
     constexpr T* midpoint(T* a, T* b);        // C++20
 
+// [numeric.sat], saturation arithmetic
+template<class T>
+constexpr T add_sat(T x, T y) noexcept;                     // freestanding, Since C++26
+template<class T>
+constexpr T sub_sat(T x, T y) noexcept;                     // freestanding, Since C++26
+template<class T>
+constexpr T mul_sat(T x, T y) noexcept;                     // freestanding, Since C++26
+template<class T>
+constexpr T div_sat(T x, T y) noexcept;                     // freestanding, Since C++26
+template<class T, class U>
+constexpr T saturate_cast(U x) noexcept;                    // freestanding, Since C++26
+
 }  // std
 
 */
@@ -160,6 +172,7 @@ template<class T>
 #include <__numeric/pstl_reduce.h>
 #include <__numeric/pstl_transform_reduce.h>
 #include <__numeric/reduce.h>
+#include <__numeric/saturation_arithmetic.h>
 #include <__numeric/transform_exclusive_scan.h>
 #include <__numeric/transform_inclusive_scan.h>
 #include <__numeric/transform_reduce.h>
diff --git a/libcxx/include/version b/libcxx/include/version
index c96647894dce63..b8d0c4e6f3c898 100644
--- a/libcxx/include/version
+++ b/libcxx/include/version
@@ -504,7 +504,7 @@ __cpp_lib_within_lifetime                               202306L <type_traits>
 // # define __cpp_lib_out_ptr                              202311L
 # define __cpp_lib_ratio                                202306L
 // # define __cpp_lib_rcu                                  202306L
-// # define __cpp_lib_saturation_arithmetic                202311L
+# define __cpp_lib_saturation_arithmetic                202311L
 // # define __cpp_lib_smart_ptr_owner_equality             202306L
 # define __cpp_lib_span_at                              202311L
 // # define __cpp_lib_span_initializer_list                202311L
diff --git a/libcxx/modules/std/numeric.inc b/libcxx/modules/std/numeric.inc
index d2b7688d4e5f10..3bc7b231681584 100644
--- a/libcxx/modules/std/numeric.inc
+++ b/libcxx/modules/std/numeric.inc
@@ -54,4 +54,14 @@ export namespace std {
 
   // [numeric.ops.midpoint], midpoint
   using std::midpoint;
+
+#if _LIBCPP_STD_VER >= 26
+  // [numeric.sat], saturation arithmetic
+  using std::add_sat;
+  using std::div_sat;
+  using std::mul_sat;
+  using std::saturate_cast;
+  using std::sub_sat;
+#endif
+
 } // namespace std
diff --git a/libcxx/test/std/language.support/support.limits/support.limits.general/numeric.version.compile.pass.cpp b/libcxx/test/std/language.support/support.limits/support.limits.general/numeric.version.compile.pass.cpp
index b510eefc69a5d3..d132b7c7b9c4f5 100644
--- a/libcxx/test/std/language.support/support.limits/support.limits.general/numeric.version.compile.pass.cpp
+++ b/libcxx/test/std/language.support/support.limits/support.limits.general/numeric.version.compile.pass.cpp
@@ -263,17 +263,11 @@
 #   endif
 # endif
 
-# if !defined(_LIBCPP_VERSION)
-#   ifndef __cpp_lib_saturation_arithmetic
-#     error "__cpp_lib_saturation_arithmetic should be defined in c++26"
-#   endif
-#   if __cpp_lib_saturation_arithmetic != 202311L
-#     error "__cpp_lib_saturation_arithmetic should have the value 202311L in c++26"
-#   endif
-# else // _LIBCPP_VERSION
-#   ifdef __cpp_lib_saturation_arithmetic
-#     error "__cpp_lib_saturation_arithmetic should not be defined because it is unimplemented in libc++!"
-#   endif
+# ifndef __cpp_lib_saturation_arithmetic
+#   error "__cpp_lib_saturation_arithmetic should be defined in c++26"
+# endif
+# if __cpp_lib_saturation_arithmetic != 202311L
+#   error "__cpp_lib_saturation_arithmetic should have the value 202311L in c++26"
 # endif
 
 #endif // TEST_STD_VER > 23
diff --git a/libcxx/test/std/language.support/support.limits/support.limits.general/version.version.compile.pass.cpp b/libcxx/test/std/language.support/support.limits/support.limits.general/version.version.compile.pass.cpp
index d5a0839b30f824..edb36da28bbe15 100644
--- a/libcxx/test/std/language.support/support.limits/support.limits.general/version.version.compile.pass.cpp
+++ b/libcxx/test/std/language.support/support.limits/support.limits.general/version.version.compile.pass.cpp
@@ -7167,17 +7167,11 @@
 #   error "__cpp_lib_sample should have the value 201603L in c++26"
 # endif
 
-# if !defined(_LIBCPP_VERSION)
-#   ifndef __cpp_lib_saturation_arithmetic
-#     error "__cpp_lib_saturation_arithmetic should be defined in c++26"
-#   endif
-#   if __cpp_lib_saturation_arithmetic != 202311L
-#     error "__cpp_lib_saturation_arithmetic should have the value 202311L in c++26"
-#   endif
-# else // _LIBCPP_VERSION
-#   ifdef __cpp_lib_saturation_arithmetic
-#     error "__cpp_lib_saturation_arithmetic should not be defined because it is unimplemented in libc++!"
-#   endif
+# ifndef __cpp_lib_saturation_arithmetic
+#   error "__cpp_lib_saturation_arithmetic should be defined in c++26"
+# endif
+# if __cpp_lib_saturation_arithmetic != 202311L
+#   error "__cpp_lib_saturation_arithmetic should have the value 202311L in c++26"
 # endif
 
 # ifndef __cpp_lib_scoped_lock
diff --git a/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp
new file mode 100644
index 00000000000000..ccd3594fd3a652
--- /dev/null
+++ b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp
@@ -0,0 +1,98 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11, c++14, c++17, c++20, c++23
+
+// <numeric>
+
+// template<class T>
+// constexpr T add_sat(T x, T y) noexcept;                     // freestanding
+
+#include <cassert>
+#include <concepts>
+#include <limits>
+#include <numeric>
+
+template <typename IntegerT>
+constexpr bool test_signed() {
+  constexpr auto minVal = std::numeric_limits<IntegerT>::min();
+  constexpr auto maxVal = std::numeric_limits<IntegerT>::max();
+
+  static_assert(noexcept(std::div_sat(minVal, maxVal)));
+
+  // No saturation
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(IntegerT{3}, IntegerT{4});
+    assert(sum == IntegerT{7});
+  }
+
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(IntegerT{-3}, IntegerT{4});
+    assert(sum == IntegerT{1});
+  }
+
+  // Saturation - max - both arguments positive
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(maxVal, IntegerT{4});
+    assert(sum == maxVal);
+  }
+
+  // Saturation - min - both arguments negative
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(minVal, IntegerT{-4});
+    assert(sum == minVal);
+  }
+
+  return true;
+}
+
+template <typename IntegerT>
+constexpr bool test_unsigned() {
+  constexpr auto minVal = std::numeric_limits<IntegerT>::min();
+  constexpr auto maxVal = std::numeric_limits<IntegerT>::max();
+
+  static_assert(noexcept(std::div_sat(minVal, maxVal)));
+
+  // No Saturation
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(IntegerT{3}, IntegerT{4});
+    assert(sum == IntegerT{7});
+  }
+
+  // Saturation - max only
+  {
+    std::same_as<IntegerT> decltype(auto) sum = std::add_sat(maxVal, IntegerT{4});
+    assert(sum == maxVal);
+  }
+
+  return true;
+}
+
+constexpr bool test() {
+  // Signed
+  test_signed<signed char>();
+  test_signed<short int>();
+  test_signed<int>();
+  test_signed<long int>();
+  test_signed<long long int>();
+  // Unsigned
+  test_unsigned<unsigned char>();
+  test_unsigned<unsigned short int>();
+  test_unsigned<unsigned int>();
+  test_unsigned<unsigned long int>();
+  test_unsigned<unsigned long long int>();
+
+  return true;
+}
+
+int main(int, char**) {
+  test();
+  static_assert(test());
+
+  return 0;
+}
diff --git a/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.verify.cpp b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.verify.cpp
new file mode 100644
index 00000000000000..b05b30abea6fdb
--- /dev/null
+++ b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.verify.cpp
@@ -0,0 +1,39 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11, c++14, c++17, c++20, c++23
+
+// <numeric>
+
+// template<class T>
+// constexpr T add_sat(T x, T y) noexcept;                     // freestanding
+
+#include <cstdint>
+#include <numeric>
+
+#include "test_macros.h"
+
+template <typename IntegerT>
+constexpr void test_constraint() {
+  // expected-error-re@*:* 0-2 {{constant expression evaluates to {{.*}} which cannot be narrowed to type {{.*}}}}
+  // expected-error@*:* 0-3 {{no matching function for call to 'add_sat'}}
+  // expected-error@*:* 0-2 {{expected unqualified-id}}
+  [[maybe_unused]] auto sum = std::add_sat(IntegerT{3}, IntegerT{4});
+}
+
+constexpr bool test() {
+  test_constraint<bool>();
+  test_constraint<char>();
+#ifndef TEST_HAS_NO_WIDE_CHARACTERS
+  test_constraint<wchar_t>();
+#endif
+  test_constraint<std::char16_t>();
+  test_constraint<std::char32_t>();
+
+  return true;
+}
diff --git a/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.assert.pass.cpp b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.assert.pass.cpp
new file mode 100644
index 00000000000000..98616d2c1c427c
--- /dev/null
+++ b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.assert.pass.cpp
@@ -0,0 +1,53 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11, c++14, c++17, c++20, c++23
+
+// REQUIRES: has-unix-headers
+// REQUIRES: libcpp-hardening-mode={{extensive|debug}}
+// XFAIL: availability-verbose_abort-missing
+
+// <numeric>
+
+// template<class T>
+// constexpr T div_sat(T x, T y) noexcept;                     // freestanding
+
+#include <cassert>
+#include <concepts>
+#include <limits>
+#include <numeric>
+
+#include "check_assertion.h"
+
+template <typename IntegerT>
+constexpr void test() {
+  TEST_LIBCPP_ASSERT_FAILURE((void)std::div_sat(IntegerT{3}, IntegerT{0}), "Division by 0 is undefined");
+}
+
+constexpr bool test() {
+  // signed
+  test<signed char>();
+  test<short int>();
+  test<int>();
+  test<long int>();
+  test<long long int>();
+  // unsigned
+  test<unsigned char>();
+  test<unsigned short int>();
+  test<unsigned int>();
+  test<unsigned long int>();
+  test<unsigned long long int>();
+
+  return true;
+}
+
+int main(int, char**) {
+  assert(test());
+
+  return 0;
+}
diff --git a/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp
new file mode 100644
index 00000000000000..d2891f77c564fa
--- /dev/null
+++ b/libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp
@@ -0,0 +1,98 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--------------------------------------------------...
[truncated]

libcxx/include/__numeric/saturation_arithmetic.h

mordante

Thanks for working on this. I'm happy to accept the patch without 128-bit integral support and add that later.

I have not looked closely at the tests, I'd like a bit more test cases before looking at all of them.

libcxx/docs/ReleaseNotes/18.rst

libcxx/include/__numeric/saturation_arithmetic.h

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.verify.cpp

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.assert.pass.cpp

libcxx/include/__numeric/saturation_arithmetic.h

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/mul_sat.verify.cpp

libcxx/include/__numeric/saturation_arithmetic.h

H-G-Hristov · 2024-01-14T07:44:07Z

Thank you for the review!

libcxx/include/__numeric/saturation_arithmetic.h

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp

H-G-Hristov · 2024-01-21T12:23:35Z

I did some re-imagining of the tests, there is a some redundancy but it is cleared that nothing was missed.
I also used the Clang 18's "Placeholder variables with no name" to make the test a bit cleaner, so I disabled the unsupported compilers.
The tests are formatted in tabular form manually to make them easier to read.

If the above is unacceptable, it can be fixed easily. For now it makes it easier to read and review.

mordante

I did some re-imagining of the tests, there is a some redundancy but it is cleared that nothing was missed. The tests are formatted in tabular form manually to make them easier to read.

Redundant code in tests when if improves readability is not too bad. I really like how the new testing code looks. It was a lot easier to follow. Thanks a lot!

I also used the Clang 18's "Placeholder variables with no name" to make the test a bit cleaner, so I disabled the unsupported compilers.
If the above is unacceptable, it can be fixed easily. For now it makes it easier to read and review.

IMO we should not disable tests for compilers, only for the reason to be able to use newer language features in the test. Note if the header requires language features I'm fine to disable that for a compiler. In this case we can easily allow clang-17 with a minor change as pointed out in the review.

The patch is getting close to be ready.

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/add_sat.pass.cpp

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_cast.pass.cpp

H-G-Hristov · 2024-01-21T19:02:28Z

~~There is a problem with the test. Working on it.~~

mordante

LGTM modulo one typo.

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_cast.pass.cpp

…cast.pass.cpp Co-authored-by: Mark de Wever <zar-rpg@xs4all.nl>

vvereschaka · 2024-01-23T00:03:25Z

@Zingam ,
these changes break the aarch64 cross builder with the failed test llvm-libc++-static.cfg.in::mul_sat.pass.cpp

https://lab.llvm.org/buildbot/#/builders/119/builds/16778

# .---command stderr------------
# | ld.lld: error: undefined symbol: __muloti4
# | >>> referenced by mul_sat.pass.cpp
# | >>>               C:\Users\buildbot\AppData\Local\Temp\lit-tmp-vspdi658\mul_sat-857336.o:(__int128 std::__2::mul_sat[abi:ne180000]<__int128>(__int128, __int128))
# | clang++: error: linker command failed with exit code 1 (use -v to see invocation)
# `-----------------------------

vvereschaka · 2024-01-23T17:07:14Z

@H-G-Hristov, @Zingam any updates?

vvereschaka · 2024-01-24T05:11:58Z

Looks like I see the problem. The test configuration must be updated on the builder.
Sorry for the false alarm.

Zingam added 4 commits January 12, 2024 23:26

Cleanup

2c939cc

Fixed "code formatter" error

f5db95e

Cleanup

db3e72e

H-G-Hristov force-pushed the hgh/libcxx/P0543R3-Saturation-arithmetic branch from cecbc2e to db3e72e Compare January 12, 2024 21:28

H-G-Hristov marked this pull request as ready for review January 12, 2024 22:45

H-G-Hristov requested a review from a team as a code owner January 12, 2024 22:45

llvmbot added the libc++ libc++ C++ Standard Library. Not GNU libstdc++. Not libc++abi. label Jan 12, 2024

frederick-vs-ja reviewed Jan 13, 2024

View reviewed changes

libcxx/include/__numeric/saturation_arithmetic.h Outdated Show resolved Hide resolved

mordante self-assigned this Jan 13, 2024

mordante requested changes Jan 13, 2024

View reviewed changes

philnik777 reviewed Jan 13, 2024

View reviewed changes

libcxx/include/__numeric/saturation_arithmetic.h Show resolved Hide resolved

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/mul_sat.verify.cpp Outdated Show resolved Hide resolved

libcxx/include/__numeric/saturation_arithmetic.h Show resolved Hide resolved

Zingam added 5 commits January 13, 2024 20:23

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

880ca3a

WIP: prepare to address comments

30a1593

WIP: Cleaned up tests

2793acb

WIP: Test constexpr

540cfec

WIP: Optimized saturate_cast

83fff64

H-G-Hristov mentioned this pull request Jan 14, 2024

[libc++][concepts] Implements concept helper __libcpp_integer #78086

Merged

Zingam and others added 7 commits January 14, 2024 10:11

Renamed test

2cab21a

WIP: add_sat - edge case tests

68cbc9f

WIP: Added extended integers types to tests

e284255

WIP: Added check_constexpr.h

fb43978

WIP: Fixed code formatter error

39837d6

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

e8e21ce

WIP: Addtional tests sat_sub

3fbec01

philnik777 reviewed Jan 14, 2024

View reviewed changes

libcxx/include/__numeric/saturation_arithmetic.h Show resolved Hide resolved

libcxx/include/__numeric/saturation_arithmetic.h Show resolved Hide resolved

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/div_sat.pass.cpp Outdated Show resolved Hide resolved

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

ea68a59

Zingam and others added 3 commits January 21, 2024 13:42

Updated test div_sat

82ef906

Updated test sub_sat

5d1af8c

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

fc15282

H-G-Hristov requested a review from mordante January 21, 2024 12:27

H-G-Hristov marked this pull request as ready for review January 21, 2024 12:31

Minor tweaks to comments

efb7355

mordante requested changes Jan 21, 2024

View reviewed changes

Zingam added 5 commits January 21, 2024 17:57

Re-introduced [[maybe_unused]] and re-enabled the tests

75c08b7

Updated saturate_cast.pass

8ce388c

Cleanup

51166ff

Final update and cleanup

43a9b27

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

eaa0168

H-G-Hristov requested a review from mordante January 21, 2024 17:41

Zingam added 3 commits January 21, 2024 20:06

DEBUG

b06e016

Debug 32-bit

71a10a0

Try again

71c1cb5

H-G-Hristov marked this pull request as draft January 21, 2024 19:02

Zingam and others added 2 commits January 21, 2024 21:09

Try to fix CI

0066189

Merge branch 'main' into hgh/libcxx/P0543R3-Saturation-arithmetic

618d6e4

H-G-Hristov marked this pull request as ready for review January 21, 2024 19:32

mordante approved these changes Jan 21, 2024

View reviewed changes

libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_cast.pass.cpp Outdated Show resolved Hide resolved

Update libcxx/test/std/numerics/numeric.ops/numeric.ops.sat/saturate_…

fc9af61

…cast.pass.cpp Co-authored-by: Mark de Wever <zar-rpg@xs4all.nl>

Zingam merged commit 03c19e9 into llvm:main Jan 22, 2024
44 checks passed

H-G-Hristov deleted the hgh/libcxx/P0543R3-Saturation-arithmetic branch January 23, 2024 20:50

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[libc++][numeric] P0543R3: Saturation arithmetic #77967

[libc++][numeric] P0543R3: Saturation arithmetic #77967

H-G-Hristov commented Jan 12, 2024 •

edited

github-actions bot commented Jan 12, 2024 •

edited

llvmbot commented Jan 12, 2024

mordante left a comment

H-G-Hristov commented Jan 14, 2024

H-G-Hristov commented Jan 21, 2024

mordante left a comment

H-G-Hristov commented Jan 21, 2024 •

edited

mordante left a comment

vvereschaka commented Jan 23, 2024

vvereschaka commented Jan 23, 2024

vvereschaka commented Jan 24, 2024

[libc++][numeric] P0543R3: Saturation arithmetic #77967

[libc++][numeric] P0543R3: Saturation arithmetic #77967

Conversation

H-G-Hristov commented Jan 12, 2024 • edited

github-actions bot commented Jan 12, 2024 • edited

llvmbot commented Jan 12, 2024

mordante left a comment

Choose a reason for hiding this comment

H-G-Hristov commented Jan 14, 2024

H-G-Hristov commented Jan 21, 2024

mordante left a comment

Choose a reason for hiding this comment

H-G-Hristov commented Jan 21, 2024 • edited

mordante left a comment

Choose a reason for hiding this comment

vvereschaka commented Jan 23, 2024

vvereschaka commented Jan 23, 2024

vvereschaka commented Jan 24, 2024

H-G-Hristov commented Jan 12, 2024 •

edited

github-actions bot commented Jan 12, 2024 •

edited

H-G-Hristov commented Jan 21, 2024 •

edited