Revert "[libc][NFC] Refactor FPBits and remove LongDoubleBits special…

…ization (#78192)" (#78329)

This reverts commit fdbf255.

Causes build breakage on 32b arm (see reports:
#78192).

These are reproducible for the 32b arm baremetal target on x86 hosts as
well.

libc/src/__support/FPUtil/generic/sqrt_80_bit_long_double.h

@@ -131,7 +131,7 @@ LIBC_INLINE long double sqrt(long double x) {
     out.set_implicit_bit(1);
     out.set_mantissa((y & (ONE - 1)));
 
-    return out.get_val();
+    return out;
   }
 }
 #endif // LIBC_LONG_DOUBLE_IS_X86_FLOAT80

libc/src/__support/FPUtil/x86_64/LongDoubleBits.h

@@ -0,0 +1,179 @@
+//===-- Bit representation of x86 long double numbers -----------*- 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_FPUTIL_X86_64_LONGDOUBLEBITS_H
+#define LLVM_LIBC_SRC___SUPPORT_FPUTIL_X86_64_LONGDOUBLEBITS_H
+
+#include "src/__support/CPP/bit.h"
+#include "src/__support/UInt128.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/attributes.h" // LIBC_INLINE
+#include "src/__support/macros/properties/architectures.h"
+
+#if !defined(LIBC_TARGET_ARCH_IS_X86)
+#error "Invalid include"
+#endif
+
+#include "src/__support/FPUtil/FPBits.h"
+
+#include <stdint.h>
+
+namespace LIBC_NAMESPACE {
+namespace fputil {
+
+template <>
+struct FPBits<long double> : public internal::FPRep<FPType::X86_Binary80> {
+  using UP = internal::FPRep<FPType::X86_Binary80>;
+  using StorageType = typename UP::StorageType;
+
+private:
+  using UP::bits;
+  using UP::EXP_SIG_MASK;
+  using UP::QUIET_NAN_MASK;
+
+public:
+  // Constants.
+  static constexpr int MAX_BIASED_EXPONENT = (1 << EXP_LEN) - 1;
+  // The x86 80 bit float represents the leading digit of the mantissa
+  // explicitly. This is the mask for that bit.
+  static constexpr StorageType EXPLICIT_BIT_MASK = StorageType(1)
+                                                   << FRACTION_LEN;
+  // The X80 significand is made of an explicit bit and the fractional part.
+  static_assert((EXPLICIT_BIT_MASK & FRACTION_MASK) == 0,
+                "the explicit bit and the fractional part should not overlap");
+  static_assert((EXPLICIT_BIT_MASK | FRACTION_MASK) == SIG_MASK,
+                "the explicit bit and the fractional part should cover the "
+                "whole significand");
+  static constexpr StorageType MIN_SUBNORMAL = StorageType(1);
+  // Subnormal numbers include the implicit bit in x86 long double formats.
+  static constexpr StorageType MAX_SUBNORMAL = FRACTION_MASK;
+  static constexpr StorageType MIN_NORMAL =
+      (StorageType(1) << SIG_LEN) | EXPLICIT_BIT_MASK;
+  static constexpr StorageType MAX_NORMAL =
+      (StorageType(MAX_BIASED_EXPONENT - 1) << SIG_LEN) | SIG_MASK;
+
+  // Constructors.
+  LIBC_INLINE constexpr FPBits() = default;
+
+  template <typename XType> LIBC_INLINE constexpr explicit FPBits(XType x) {
+    using Unqual = typename cpp::remove_cv_t<XType>;
+    if constexpr (cpp::is_same_v<Unqual, long double>) {
+      bits = cpp::bit_cast<StorageType>(x);
+    } else if constexpr (cpp::is_same_v<Unqual, StorageType>) {
+      bits = x;
+    } else {
+      // We don't want accidental type promotions/conversions, so we require
+      // exact type match.
+      static_assert(cpp::always_false<XType>);
+    }
+  }
+
+  // Floating-point conversions.
+  LIBC_INLINE constexpr long double get_val() const {
+    return cpp::bit_cast<long double>(bits);
+  }
+
+  LIBC_INLINE constexpr operator long double() const {
+    return cpp::bit_cast<long double>(bits);
+  }
+
+  LIBC_INLINE constexpr StorageType get_explicit_mantissa() const {
+    return bits & SIG_MASK;
+  }
+
+  LIBC_INLINE constexpr bool get_implicit_bit() const {
+    return bits & EXPLICIT_BIT_MASK;
+  }
+
+  LIBC_INLINE constexpr void set_implicit_bit(bool implicitVal) {
+    if (get_implicit_bit() != implicitVal)
+      bits ^= EXPLICIT_BIT_MASK;
+  }
+
+  LIBC_INLINE constexpr bool is_inf() const {
+    return get_biased_exponent() == MAX_BIASED_EXPONENT &&
+           get_mantissa() == 0 && get_implicit_bit() == 1;
+  }
+
+  LIBC_INLINE constexpr bool is_nan() const {
+    if (get_biased_exponent() == MAX_BIASED_EXPONENT) {
+      return (get_implicit_bit() == 0) || get_mantissa() != 0;
+    } else if (get_biased_exponent() != 0) {
+      return get_implicit_bit() == 0;
+    }
+    return false;
+  }
+
+  LIBC_INLINE constexpr bool is_inf_or_nan() const {
+    return (get_biased_exponent() == MAX_BIASED_EXPONENT) ||
+           (get_biased_exponent() != 0 && get_implicit_bit() == 0);
+  }
+
+  LIBC_INLINE constexpr bool is_quiet_nan() const {
+    return (bits & EXP_SIG_MASK) >= (EXP_MASK | QUIET_NAN_MASK);
+  }
+
+  // Methods below this are used by tests.
+
+  LIBC_INLINE static constexpr long double zero(bool sign = false) {
+    StorageType rep = (sign ? SIGN_MASK : StorageType(0)) // sign
+                      | 0                                 // exponent
+                      | 0                                 // explicit bit
+                      | 0;                                // mantissa
+    return FPBits(rep).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double neg_zero() { return zero(true); }
+
+  LIBC_INLINE static constexpr long double inf(bool sign = false) {
+    StorageType rep = (sign ? SIGN_MASK : StorageType(0)) // sign
+                      | EXP_MASK                          // exponent
+                      | EXPLICIT_BIT_MASK                 // explicit bit
+                      | 0;                                // mantissa
+    return FPBits(rep).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double neg_inf() { return inf(true); }
+
+  LIBC_INLINE static constexpr long double min_normal() {
+    return FPBits(MIN_NORMAL).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double max_normal() {
+    return FPBits(MAX_NORMAL).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double min_denormal() {
+    return FPBits(MIN_SUBNORMAL).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double max_denormal() {
+    return FPBits(MAX_SUBNORMAL).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double build_nan(StorageType v) {
+    StorageType rep = 0                      // sign
+                      | EXP_MASK             // exponent
+                      | EXPLICIT_BIT_MASK    // explicit bit
+                      | (v & FRACTION_MASK); // mantissa
+    return FPBits(rep).get_val();
+  }
+
+  LIBC_INLINE static constexpr long double build_quiet_nan(StorageType v) {
+    return build_nan(QUIET_NAN_MASK | v);
+  }
+};
+
+static_assert(
+    sizeof(FPBits<long double>) == sizeof(long double),
+    "Internal long double representation does not match the machine format.");
+
+} // namespace fputil
+} // namespace LIBC_NAMESPACE
+
+#endif // LLVM_LIBC_SRC___SUPPORT_FPUTIL_X86_64_LONGDOUBLEBITS_H

libc/src/__support/FPUtil/x86_64/NextAfterLongDouble.h

@@ -61,7 +61,7 @@ LIBC_INLINE long double nextafter(long double from, long double to) {
         from_bits.set_biased_exponent(from_bits.get_biased_exponent() + 1);
         if (from_bits.is_inf())
           raise_except_if_required(FE_OVERFLOW | FE_INEXACT);
-        return from_bits.get_val();
+        return from_bits;
       } else {
         ++int_val;
       }
@@ -75,7 +75,7 @@ LIBC_INLINE long double nextafter(long double from, long double to) {
         // from == 0 is handled separately so decrementing the exponent will not
         // lead to underflow.
         from_bits.set_biased_exponent(from_bits.get_biased_exponent() - 1);
-        return from_bits.get_val();
+        return from_bits;
       } else {
         --int_val;
       }
@@ -94,7 +94,7 @@ LIBC_INLINE long double nextafter(long double from, long double to) {
         // from == 0 is handled separately so decrementing the exponent will not
         // lead to underflow.
         from_bits.set_biased_exponent(from_bits.get_biased_exponent() - 1);
-        return from_bits.get_val();
+        return from_bits;
       } else {
         --int_val;
       }
@@ -109,7 +109,7 @@ LIBC_INLINE long double nextafter(long double from, long double to) {
         from_bits.set_biased_exponent(from_bits.get_biased_exponent() + 1);
         if (from_bits.is_inf())
           raise_except_if_required(FE_OVERFLOW | FE_INEXACT);
-        return from_bits.get_val();
+        return from_bits;
       } else {
         ++int_val;
       }