[libc][NFC] Move `Sign` type to separate header #85930

gchatelet · 2024-03-20T13:30:58Z

No description provided.

gchatelet · 2024-03-20T13:33:28Z

This also saves quite a few using statements.

llvmbot · 2024-03-20T13:34:56Z

@llvm/pr-subscribers-libc

Author: Guillaume Chatelet (gchatelet)

Changes

Patch is 33.99 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/85930.diff

54 Files Affected:

(modified) libc/src/__support/CMakeLists.txt (+8)
(modified) libc/src/__support/FPUtil/CMakeLists.txt (+1)
(modified) libc/src/__support/FPUtil/FPBits.h (+1-26)
(modified) libc/src/__support/FPUtil/fpbits_str.h (-1)
(added) libc/src/__support/sign.h (+40)
(modified) libc/src/__support/str_to_float.h (+1-3)
(modified) libc/src/math/generic/acosf.cpp (+1-1)
(modified) libc/src/math/generic/asinf.cpp (+1-1)
(modified) libc/src/math/generic/atanf.cpp (-1)
(modified) libc/src/math/generic/atanhf.cpp (+1-1)
(modified) libc/src/math/generic/cosf.cpp (+1-1)
(modified) libc/src/math/generic/coshf.cpp (+1-1)
(modified) libc/src/math/generic/exp.cpp (+1-1)
(modified) libc/src/math/generic/exp10.cpp (+1-1)
(modified) libc/src/math/generic/exp2.cpp (+1-1)
(modified) libc/src/math/generic/expm1.cpp (+2-2)
(modified) libc/src/math/generic/log.cpp (+2-2)
(modified) libc/src/math/generic/log10.cpp (+2-2)
(modified) libc/src/math/generic/log10f.cpp (+1-1)
(modified) libc/src/math/generic/log1p.cpp (+2-2)
(modified) libc/src/math/generic/log1pf.cpp (+1-1)
(modified) libc/src/math/generic/log2.cpp (+2-2)
(modified) libc/src/math/generic/log2f.cpp (+1-1)
(modified) libc/src/math/generic/log_range_reduction.h (-1)
(modified) libc/src/math/generic/logf.cpp (+1-1)
(modified) libc/src/math/generic/powf.cpp (+2-2)
(modified) libc/src/stdio/printf_core/float_dec_converter.h (+1-1)
(modified) libc/test/UnitTest/FPMatcher.h (+1-2)
(modified) libc/test/src/__support/FPUtil/CMakeLists.txt (+1)
(modified) libc/test/src/__support/FPUtil/dyadic_float_test.cpp (-1)
(modified) libc/test/src/__support/FPUtil/fpbits_test.cpp (+1-1)
(modified) libc/test/src/math/FDimTest.h (-1)
(modified) libc/test/src/math/FmaTest.h (-1)
(modified) libc/test/src/math/HypotTest.h (+1-1)
(modified) libc/test/src/math/ILogbTest.h (+1-1)
(modified) libc/test/src/math/LdExpTest.h (-1)
(modified) libc/test/src/math/NextAfterTest.h (-1)
(modified) libc/test/src/math/RIntTest.h (-1)
(modified) libc/test/src/math/RemQuoTest.h (-1)
(modified) libc/test/src/math/RoundToIntegerTest.h (-1)
(modified) libc/test/src/math/atanhf_test.cpp (+1-1)
(modified) libc/test/src/math/smoke/FDimTest.h (-1)
(modified) libc/test/src/math/smoke/FmaTest.h (-1)
(modified) libc/test/src/math/smoke/HypotTest.h (+1-1)
(modified) libc/test/src/math/smoke/ILogbTest.h (-1)
(modified) libc/test/src/math/smoke/LdExpTest.h (-1)
(modified) libc/test/src/math/smoke/NextAfterTest.h (-1)
(modified) libc/test/src/math/smoke/NextTowardTest.h (-1)
(modified) libc/test/src/math/smoke/RIntTest.h (-1)
(modified) libc/test/src/math/smoke/RemQuoTest.h (-1)
(modified) libc/test/src/math/smoke/RoundToIntegerTest.h (-1)
(modified) libc/test/src/math/smoke/atanhf_test.cpp (+1-1)
(modified) utils/bazel/llvm-project-overlay/libc/BUILD.bazel (+9)
(modified) utils/bazel/llvm-project-overlay/libc/test/src/__support/FPUtil/BUILD.bazel (+1)

diff --git a/libc/src/__support/CMakeLists.txt b/libc/src/__support/CMakeLists.txt
index 4c1f271e1df43a..7b1820d9bf3538 100644
--- a/libc/src/__support/CMakeLists.txt
+++ b/libc/src/__support/CMakeLists.txt
@@ -41,6 +41,14 @@ add_header_library(
     libc.src.__support.macros.config
 )
 
+add_header_library(
+  sign
+  HDRS
+    sign.h
+  DEPENDS
+    libc.src.__support.macros.attributes
+)
+
 add_header_library(
   error_or
   HDRS
diff --git a/libc/src/__support/FPUtil/CMakeLists.txt b/libc/src/__support/FPUtil/CMakeLists.txt
index f1c6fba22856dd..4ded70a675ea88 100644
--- a/libc/src/__support/FPUtil/CMakeLists.txt
+++ b/libc/src/__support/FPUtil/CMakeLists.txt
@@ -35,6 +35,7 @@ add_header_library(
     libc.src.__support.macros.attributes
     libc.src.__support.macros.properties.types
     libc.src.__support.math_extras
+    libc.src.__support.sign
     libc.src.__support.uint128
 )
 
diff --git a/libc/src/__support/FPUtil/FPBits.h b/libc/src/__support/FPUtil/FPBits.h
index b06b3f7b73959a..155bff2f558102 100644
--- a/libc/src/__support/FPUtil/FPBits.h
+++ b/libc/src/__support/FPUtil/FPBits.h
@@ -17,6 +17,7 @@
 #include "src/__support/macros/attributes.h" // LIBC_INLINE, LIBC_INLINE_VAR
 #include "src/__support/macros/properties/types.h" // LIBC_TYPES_HAS_FLOAT128
 #include "src/__support/math_extras.h"             // mask_trailing_ones
+#include "src/__support/sign.h"                    // Sign
 
 #include <stdint.h>
 
@@ -32,32 +33,6 @@ enum class FPType {
   X86_Binary80,
 };
 
-// A type to interact with floating point type signs.
-// This may be moved outside of 'fputil' if useful.
-struct Sign {
-  LIBC_INLINE constexpr bool is_pos() const { return !is_negative; }
-  LIBC_INLINE constexpr bool is_neg() const { return is_negative; }
-
-  LIBC_INLINE friend constexpr bool operator==(Sign a, Sign b) {
-    return a.is_negative == b.is_negative;
-  }
-  LIBC_INLINE friend constexpr bool operator!=(Sign a, Sign b) {
-    return !(a == b);
-  }
-
-  static const Sign POS;
-  static const Sign NEG;
-
-private:
-  LIBC_INLINE constexpr explicit Sign(bool is_negative)
-      : is_negative(is_negative) {}
-
-  bool is_negative;
-};
-
-LIBC_INLINE_VAR constexpr Sign Sign::NEG = Sign(true);
-LIBC_INLINE_VAR constexpr Sign Sign::POS = Sign(false);
-
 // The classes hierarchy is as follows:
 //
 //             ┌───────────────────┐
diff --git a/libc/src/__support/FPUtil/fpbits_str.h b/libc/src/__support/FPUtil/fpbits_str.h
index 212265bb9ad4a8..97689867da4de4 100644
--- a/libc/src/__support/FPUtil/fpbits_str.h
+++ b/libc/src/__support/FPUtil/fpbits_str.h
@@ -35,7 +35,6 @@ using ZeroPaddedHexFmt = IntegerToString<
 //    floating encoding.
 template <typename T> LIBC_INLINE cpp::string str(fputil::FPBits<T> x) {
   using StorageType = typename fputil::FPBits<T>::StorageType;
-  using Sign = fputil::Sign;
 
   if (x.is_nan())
     return "(NaN)";
diff --git a/libc/src/__support/sign.h b/libc/src/__support/sign.h
new file mode 100644
index 00000000000000..cba0ec7c0e8b0b
--- /dev/null
+++ b/libc/src/__support/sign.h
@@ -0,0 +1,40 @@
+//===-- A simple sign type --------------------------------------*- 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 LLVM_LIBC_SRC___SUPPORT_SIGN_H
+#define LLVM_LIBC_SRC___SUPPORT_SIGN_H
+
+#include "src/__support/macros/attributes.h" // LIBC_INLINE, LIBC_INLINE_VAR
+
+// A type to interact with signed arithmetic types.
+struct Sign {
+  LIBC_INLINE constexpr bool is_pos() const { return !is_negative; }
+  LIBC_INLINE constexpr bool is_neg() const { return is_negative; }
+
+  LIBC_INLINE friend constexpr bool operator==(Sign a, Sign b) {
+    return a.is_negative == b.is_negative;
+  }
+  
+  LIBC_INLINE friend constexpr bool operator!=(Sign a, Sign b) {
+    return !(a == b);
+  }
+
+  static const Sign POS;
+  static const Sign NEG;
+
+private:
+  LIBC_INLINE constexpr explicit Sign(bool is_negative)
+      : is_negative(is_negative) {}
+
+  bool is_negative;
+};
+
+LIBC_INLINE_VAR constexpr Sign Sign::NEG = Sign(true);
+LIBC_INLINE_VAR constexpr Sign Sign::POS = Sign(false);
+
+#endif // LLVM_LIBC_SRC___SUPPORT_SIGN_H
diff --git a/libc/src/__support/str_to_float.h b/libc/src/__support/str_to_float.h
index 2cf2cfb027243e..f622b7edaa8a72 100644
--- a/libc/src/__support/str_to_float.h
+++ b/libc/src/__support/str_to_float.h
@@ -513,7 +513,6 @@ clinger_fast_path(ExpandedFloat<T> init_num,
                   RoundDirection round = RoundDirection::Nearest) {
   using FPBits = typename fputil::FPBits<T>;
   using StorageType = typename FPBits::StorageType;
-  using Sign = fputil::Sign;
 
   StorageType mantissa = init_num.mantissa;
   int32_t exp10 = init_num.exponent;
@@ -1085,7 +1084,6 @@ template <class T>
 LIBC_INLINE StrToNumResult<T> strtofloatingpoint(const char *__restrict src) {
   using FPBits = typename fputil::FPBits<T>;
   using StorageType = typename FPBits::StorageType;
-  using Sign = fputil::Sign;
 
   FPBits result = FPBits();
   bool seen_digit = false;
@@ -1223,7 +1221,7 @@ template <class T> LIBC_INLINE StrToNumResult<T> strtonan(const char *arg) {
   if (arg[index] == '\0')
     nan_mantissa = nan_mantissa_from_ncharseq<T>(cpp::string_view(arg, index));
 
-  result = FPBits::quiet_nan(fputil::Sign::POS, nan_mantissa);
+  result = FPBits::quiet_nan(Sign::POS, nan_mantissa);
   return {result.get_val(), 0, error};
 }
 
diff --git a/libc/src/math/generic/acosf.cpp b/libc/src/math/generic/acosf.cpp
index 0c1fdbc6869302..e6e28d43ef61f5 100644
--- a/libc/src/math/generic/acosf.cpp
+++ b/libc/src/math/generic/acosf.cpp
@@ -38,7 +38,7 @@ static constexpr fputil::ExceptValues<float, N_EXCEPTS> ACOSF_EXCEPTS = {{
 
 LLVM_LIBC_FUNCTION(float, acosf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   uint32_t x_uint = xbits.uintval();
   uint32_t x_abs = xbits.uintval() & 0x7fff'ffffU;
diff --git a/libc/src/math/generic/asinf.cpp b/libc/src/math/generic/asinf.cpp
index 6e3a27238ac998..d9133333d2561a 100644
--- a/libc/src/math/generic/asinf.cpp
+++ b/libc/src/math/generic/asinf.cpp
@@ -44,7 +44,7 @@ static constexpr fputil::ExceptValues<float, N_EXCEPTS> ASINF_EXCEPTS_HI = {{
 
 LLVM_LIBC_FUNCTION(float, asinf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   uint32_t x_uint = xbits.uintval();
   uint32_t x_abs = xbits.uintval() & 0x7fff'ffffU;
diff --git a/libc/src/math/generic/atanf.cpp b/libc/src/math/generic/atanf.cpp
index 5f66ea52d0d7ae..4adda429cc041c 100644
--- a/libc/src/math/generic/atanf.cpp
+++ b/libc/src/math/generic/atanf.cpp
@@ -20,7 +20,6 @@ namespace LIBC_NAMESPACE {
 
 LLVM_LIBC_FUNCTION(float, atanf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
 
   constexpr double FINAL_SIGN[2] = {1.0, -1.0};
   constexpr double SIGNED_PI_OVER_2[2] = {0x1.921fb54442d18p0,
diff --git a/libc/src/math/generic/atanhf.cpp b/libc/src/math/generic/atanhf.cpp
index fe2c36494a72f8..97fd1b23360067 100644
--- a/libc/src/math/generic/atanhf.cpp
+++ b/libc/src/math/generic/atanhf.cpp
@@ -15,7 +15,7 @@ namespace LIBC_NAMESPACE {
 
 LLVM_LIBC_FUNCTION(float, atanhf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   Sign sign = xbits.sign();
   uint32_t x_abs = xbits.abs().uintval();
diff --git a/libc/src/math/generic/cosf.cpp b/libc/src/math/generic/cosf.cpp
index d59304933d60d8..180a44e947eaff 100644
--- a/libc/src/math/generic/cosf.cpp
+++ b/libc/src/math/generic/cosf.cpp
@@ -42,7 +42,7 @@ static constexpr fputil::ExceptValues<float, N_EXCEPTS> COSF_EXCEPTS{{
 
 LLVM_LIBC_FUNCTION(float, cosf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   xbits.set_sign(Sign::POS);
 
diff --git a/libc/src/math/generic/coshf.cpp b/libc/src/math/generic/coshf.cpp
index a618056a64dc8a..a8ea324c950528 100644
--- a/libc/src/math/generic/coshf.cpp
+++ b/libc/src/math/generic/coshf.cpp
@@ -17,7 +17,7 @@ namespace LIBC_NAMESPACE {
 
 LLVM_LIBC_FUNCTION(float, coshf, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   xbits.set_sign(Sign::POS);
   x = xbits.get_val();
diff --git a/libc/src/math/generic/exp.cpp b/libc/src/math/generic/exp.cpp
index 42a4491131a04e..3d060bcbd3be30 100644
--- a/libc/src/math/generic/exp.cpp
+++ b/libc/src/math/generic/exp.cpp
@@ -31,7 +31,7 @@ namespace LIBC_NAMESPACE {
 using fputil::DoubleDouble;
 using fputil::TripleDouble;
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 // log2(e)
diff --git a/libc/src/math/generic/exp10.cpp b/libc/src/math/generic/exp10.cpp
index 72ece669765688..a4ae41407112bd 100644
--- a/libc/src/math/generic/exp10.cpp
+++ b/libc/src/math/generic/exp10.cpp
@@ -31,7 +31,7 @@ namespace LIBC_NAMESPACE {
 using fputil::DoubleDouble;
 using fputil::TripleDouble;
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 // log2(10)
diff --git a/libc/src/math/generic/exp2.cpp b/libc/src/math/generic/exp2.cpp
index 83f545eb116bd3..1a2fa3feb83e54 100644
--- a/libc/src/math/generic/exp2.cpp
+++ b/libc/src/math/generic/exp2.cpp
@@ -31,7 +31,7 @@ namespace LIBC_NAMESPACE {
 using fputil::DoubleDouble;
 using fputil::TripleDouble;
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 // Error bounds:
diff --git a/libc/src/math/generic/expm1.cpp b/libc/src/math/generic/expm1.cpp
index 9f14a8c2068ec1..574c4b9aaf39f7 100644
--- a/libc/src/math/generic/expm1.cpp
+++ b/libc/src/math/generic/expm1.cpp
@@ -39,7 +39,7 @@ namespace LIBC_NAMESPACE {
 using fputil::DoubleDouble;
 using fputil::TripleDouble;
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 // log2(e)
@@ -276,7 +276,7 @@ double set_exceptional(double x) {
 
 LLVM_LIBC_FUNCTION(double, expm1, (double x)) {
   using FPBits = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
 
   bool x_is_neg = xbits.is_neg();
diff --git a/libc/src/math/generic/log.cpp b/libc/src/math/generic/log.cpp
index 339e0297560f79..6de0d90be80e11 100644
--- a/libc/src/math/generic/log.cpp
+++ b/libc/src/math/generic/log.cpp
@@ -24,7 +24,7 @@ namespace LIBC_NAMESPACE {
 
 // 128-bit precision dyadic floating point numbers.
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 namespace {
@@ -735,7 +735,7 @@ double log_accurate(int e_x, int index, double m_x) {
 
 LLVM_LIBC_FUNCTION(double, log, (double x)) {
   using FPBits_t = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   FPBits_t xbits(x);
   uint64_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/log10.cpp b/libc/src/math/generic/log10.cpp
index c690ca28704075..fb839c111e6a0f 100644
--- a/libc/src/math/generic/log10.cpp
+++ b/libc/src/math/generic/log10.cpp
@@ -24,7 +24,7 @@ namespace LIBC_NAMESPACE {
 
 // 128-bit precision dyadic floating point numbers.
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 namespace {
@@ -737,7 +737,7 @@ double log10_accurate(int e_x, int index, double m_x) {
 
 LLVM_LIBC_FUNCTION(double, log10, (double x)) {
   using FPBits_t = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   FPBits_t xbits(x);
   uint64_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/log10f.cpp b/libc/src/math/generic/log10f.cpp
index 0216bb2133f1bb..1b6979d4414a9e 100644
--- a/libc/src/math/generic/log10f.cpp
+++ b/libc/src/math/generic/log10f.cpp
@@ -106,7 +106,7 @@ LLVM_LIBC_FUNCTION(float, log10f, (float x)) {
   constexpr double LOG10_2 = 0x1.34413509f79ffp-2;
 
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   uint32_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/log1p.cpp b/libc/src/math/generic/log1p.cpp
index 26bb4d369278af..83bd753cde5da8 100644
--- a/libc/src/math/generic/log1p.cpp
+++ b/libc/src/math/generic/log1p.cpp
@@ -23,7 +23,7 @@ namespace LIBC_NAMESPACE {
 
 // 128-bit precision dyadic floating point numbers.
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 namespace {
@@ -877,7 +877,7 @@ LIBC_INLINE double log1p_accurate(int e_x, int index,
 
 LLVM_LIBC_FUNCTION(double, log1p, (double x)) {
   using FPBits_t = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   constexpr int EXP_BIAS = FPBits_t::EXP_BIAS;
   constexpr int FRACTION_LEN = FPBits_t::FRACTION_LEN;
   constexpr uint64_t FRACTION_MASK = FPBits_t::FRACTION_MASK;
diff --git a/libc/src/math/generic/log1pf.cpp b/libc/src/math/generic/log1pf.cpp
index 28426a88e64902..e3c7d95418b1fc 100644
--- a/libc/src/math/generic/log1pf.cpp
+++ b/libc/src/math/generic/log1pf.cpp
@@ -106,7 +106,7 @@ LLVM_LIBC_FUNCTION(float, log1pf, (float x)) {
     case 0xbf800000U: // x = -1.0
       fputil::set_errno_if_required(ERANGE);
       fputil::raise_except_if_required(FE_DIVBYZERO);
-      return FPBits::inf(fputil::Sign::NEG).get_val();
+      return FPBits::inf(Sign::NEG).get_val();
 #ifndef LIBC_TARGET_CPU_HAS_FMA
     case 0x4cc1c80bU: // x = 0x1.839016p+26f
       return fputil::round_result_slightly_down(0x1.26fc04p+4f);
diff --git a/libc/src/math/generic/log2.cpp b/libc/src/math/generic/log2.cpp
index 648850b80b046e..c68bc60e8468bb 100644
--- a/libc/src/math/generic/log2.cpp
+++ b/libc/src/math/generic/log2.cpp
@@ -24,7 +24,7 @@ namespace LIBC_NAMESPACE {
 
 // 128-bit precision dyadic floating point numbers.
 using Float128 = typename fputil::DyadicFloat<128>;
-using Sign = fputil::Sign;
+
 using LIBC_NAMESPACE::operator""_u128;
 
 namespace {
@@ -857,7 +857,7 @@ double log2_accurate(int e_x, int index, double m_x) {
 
 LLVM_LIBC_FUNCTION(double, log2, (double x)) {
   using FPBits_t = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   FPBits_t xbits(x);
   uint64_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/log2f.cpp b/libc/src/math/generic/log2f.cpp
index 8651316d282cab..c9f7b2121519df 100644
--- a/libc/src/math/generic/log2f.cpp
+++ b/libc/src/math/generic/log2f.cpp
@@ -55,7 +55,7 @@ namespace LIBC_NAMESPACE {
 
 LLVM_LIBC_FUNCTION(float, log2f, (float x)) {
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   uint32_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/log_range_reduction.h b/libc/src/math/generic/log_range_reduction.h
index 8c9b7d2eabebef..64c0fc3aa4f539 100644
--- a/libc/src/math/generic/log_range_reduction.h
+++ b/libc/src/math/generic/log_range_reduction.h
@@ -37,7 +37,6 @@ log_range_reduction(double m_x, const LogRR &log_table,
                     fputil::DyadicFloat<128> &sum) {
   using Float128 = typename fputil::DyadicFloat<128>;
   using MType = typename Float128::MantissaType;
-  using Sign = fputil::Sign;
 
   int64_t v = static_cast<int64_t>(m_x * 0x1.0p60); // ulp = 2^-60
 
diff --git a/libc/src/math/generic/logf.cpp b/libc/src/math/generic/logf.cpp
index 49d258ecc13342..5296ba6bc13cf6 100644
--- a/libc/src/math/generic/logf.cpp
+++ b/libc/src/math/generic/logf.cpp
@@ -54,7 +54,7 @@ namespace LIBC_NAMESPACE {
 LLVM_LIBC_FUNCTION(float, logf, (float x)) {
   constexpr double LOG_2 = 0x1.62e42fefa39efp-1;
   using FPBits = typename fputil::FPBits<float>;
-  using Sign = fputil::Sign;
+
   FPBits xbits(x);
   uint32_t x_u = xbits.uintval();
 
diff --git a/libc/src/math/generic/powf.cpp b/libc/src/math/generic/powf.cpp
index 2c666bab6d628f..0450ffd711fff1 100644
--- a/libc/src/math/generic/powf.cpp
+++ b/libc/src/math/generic/powf.cpp
@@ -424,7 +424,7 @@ LIBC_INLINE bool larger_exponent(double a, double b) {
 double powf_double_double(int idx_x, double dx, double y6, double lo6_hi,
                           const DoubleDouble &exp2_hi_mid) {
   using DoubleBits = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   // Perform a second range reduction step:
   //   idx2 = round(2^14 * (dx  + 2^-8)) = round ( dx * 2^14 + 2^6)
   //   dx2 = (1 + dx) * r2 - 1
@@ -513,7 +513,7 @@ double powf_double_double(int idx_x, double dx, double y6, double lo6_hi,
 LLVM_LIBC_FUNCTION(float, powf, (float x, float y)) {
   using FloatBits = typename fputil::FPBits<float>;
   using DoubleBits = typename fputil::FPBits<double>;
-  using Sign = fputil::Sign;
+
   FloatBits xbits(x), ybits(y);
 
   uint32_t x_u = xbits.uintval();
diff --git a/libc/src/stdio/printf_core/float_dec_converter.h b/libc/src/stdio/printf_core/float_dec_converter.h
index 5270fc9de037ac..c4e8aaa2f0e2e9 100644
--- a/libc/src/stdio/printf_core/float_dec_converter.h
+++ b/libc/src/stdio/printf_core/float_dec_converter.h
@@ -48,7 +48,7 @@ constexpr uint32_t MAX_BLOCK = 999999999;
 constexpr char DECIMAL_POINT = '.';
 
 LIBC_INLINE RoundDirection get_round_direction(int last_digit, bool truncated,
-                                               fputil::Sign sign) {
+                                               Sign sign) {
   switch (fputil::quick_get_round()) {
   case FE_TONEAREST:
     // Round to nearest, if it's exactly halfway then round to even.
diff --git a/libc/test/UnitTest/FPMatcher.h b/libc/test/UnitTest/FPMatcher.h
index 43000efa09a39c..ee618a623efe19 100644
--- a/libc/test/UnitTest/FPMatcher.h
+++ b/libc/test/UnitTest/FPMatcher.h
@@ -63,7 +63,6 @@ template <TestCond C, typename T> FPMatcher<T, C> getMatcher(T expectedValue) {
 template <typename T> struct FPTest : public Test {
   using FPBits = LIBC_NAMESPACE::fputil::FPBits<T>;
   using StorageType = typename FPBits::StorageType;
-  using Sign = LIBC_NAMESPACE::fputil::Sign;
   static constexpr StorageType STORAGE_MAX =
       LIBC_NAMESPACE::cpp::numeric_limits<StorageType>::max();
   static constexpr T zero = FPBits::zero(Sign::POS).get_val();
@@ -92,7 +91,7 @@ template <typename T> struct FPTest : public Test {
 #define DECLARE_SPECIAL_CONSTANTS(T)                                           \
   using FPBits = LIBC_NAMESPACE::fputil::FPBits<T>;                            \
   using StorageType = typename FPBits::StorageType;                            \
-  using Sign = LIBC_NAMESPACE::fputil::Sign;                                   \
+                                                                               \
   static constexpr StorageType STORAGE_MAX =                                   \
       LIBC_NAMESPACE::cpp::numeric_limits<StorageType>::max();                 \
   const T zero = FPBits::zero(Sign::POS).get_val();                            \
diff --git a/libc/test/src/__support/FPUtil/CMakeLists.txt b/libc/test/src/__support/FPUtil/CMakeLists.txt
index f1a027a514ba23..1cbeec0cc4eb08 100644
--- a/libc/test/src/__support/FPUtil/CMakeLists.txt
+++ b/libc/test/src/__support/FPUtil/CMakeLists.txt
@@ -24,6 +24,7 @@ add_libc_test(
     libc.src.__support.FPUtil.fp_bits
     libc.src.__support.FPUtil.fpbits_str
     libc.src.__support.integer_literals
+    libc.src.__support.sign
 )
 
 add_fp_unittest(
diff --git a/libc/test/src/__support/FPUtil/dyadic_float_test.cpp b/libc/test/src/__support/FPUtil/dyadic_float_test.cpp
index 625aa70973b9f1..5ee9aaad563827 100644
--- a/libc/test/src/__support/FPUtil/dyadic_float_test.cpp
+++ b/libc/test/src/__support/FPUtil/dyadic_float_test.cpp
@@ -15,7 +15,6 @@
 using Float128 = LIBC_NAMESPACE::fputil::DyadicFloat<128>;
 using Float192 = LIBC_NAMESPACE::fputil::DyadicFloat<192>;
 using Float256 = LIBC_NAMESPACE::fputil::DyadicFloat<256>;
-using Sign = LIBC_NAMESPACE::fputil::Sign;
 
 TEST(LlvmLibcDyadicFloatTest, BasicConversions) {
   Float128 x(Sign::POS, /*exponent*/ 0,
diff --git a/libc/test/src/__support/FPUtil/fpbits_test.cpp b/libc/test/src/__support/FPUtil/fpbits_test.cpp
index f5c27d4fc0302b..af20b1a0bdc7ef 100644
--- a/libc/test/src/__support/FPUtil/fpbits_tes...
[truncated]

github-actions · 2024-03-20T13:37:02Z

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

[libc][NFC] Move Sign type to separate header

493d972

gchatelet requested a review from rupprecht as a code owner March 20, 2024 13:30

gchatelet changed the title ~~[libc][NFC] Move Sign type to separate header~~ [libc][NFC] Move Sign type to separate header Mar 20, 2024

gchatelet requested a review from lntue March 20, 2024 13:33

llvmbot added libc bazel "Peripheral" support tier build system: utils/bazel labels Mar 20, 2024

Fix formatting

f5a3c5c

lntue approved these changes Mar 20, 2024

View reviewed changes

gchatelet removed the request for review from rupprecht March 20, 2024 14:10

gchatelet merged commit 2137894 into llvm:main Mar 20, 2024

gchatelet deleted the move_sign_to_separate_header branch March 20, 2024 14:10

chencha3 pushed a commit to chencha3/llvm-project that referenced this pull request Mar 23, 2024

[libc][NFC] Move Sign type to separate header (llvm#85930)

8ec5b1e

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[libc][NFC] Move `Sign` type to separate header #85930

[libc][NFC] Move `Sign` type to separate header #85930

Uh oh!

gchatelet commented Mar 20, 2024

Uh oh!

gchatelet commented Mar 20, 2024

Uh oh!

llvmbot commented Mar 20, 2024

Uh oh!

github-actions bot commented Mar 20, 2024 •

edited

Loading

Uh oh!

Uh oh!

[libc][NFC] Move Sign type to separate header #85930

[libc][NFC] Move Sign type to separate header #85930

Uh oh!

Conversation

gchatelet commented Mar 20, 2024

Uh oh!

gchatelet commented Mar 20, 2024

Uh oh!

llvmbot commented Mar 20, 2024

Uh oh!

github-actions bot commented Mar 20, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

[libc][NFC] Move `Sign` type to separate header #85930

[libc][NFC] Move `Sign` type to separate header #85930

github-actions bot commented Mar 20, 2024 •

edited

Loading