[libc-commits] [libc] [llvm] [reland][libc][NFC] Refactor FPBits and remove LongDoubleBits specialization (PR #78447)
via libc-commits
libc-commits at lists.llvm.org
Wed Jan 17 05:46:22 PST 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-libc
Author: Guillaume Chatelet (gchatelet)
<details>
<summary>Changes</summary>
- [reland][libc][NFC] Refactor FPBits and remove LongDoubleBits specialization
- Use opaque structs instead of enum so that is works when StorageType is BigInt
---
Patch is 46.05 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/78447.diff
8 Files Affected:
- (modified) libc/src/__support/FPUtil/FPBits.h (+384-92)
- (modified) libc/src/__support/FPUtil/generic/sqrt_80_bit_long_double.h (+1-1)
- (removed) libc/src/__support/FPUtil/x86_64/LongDoubleBits.h (-179)
- (modified) libc/src/__support/FPUtil/x86_64/NextAfterLongDouble.h (+4-4)
- (modified) libc/test/src/__support/FPUtil/fpbits_test.cpp (+213)
- (modified) libc/test/utils/FPUtil/x86_long_double_test.cpp (+6-6)
- (modified) utils/bazel/llvm-project-overlay/libc/BUILD.bazel (-1)
- (added) utils/bazel/llvm-project-overlay/libc/test/src/__support/FPUtil/BUILD.bazel (+42)
``````````diff
diff --git a/libc/src/__support/FPUtil/FPBits.h b/libc/src/__support/FPUtil/FPBits.h
index 93e32ba7cc9415b..7456913a01ec380 100644
--- a/libc/src/__support/FPUtil/FPBits.h
+++ b/libc/src/__support/FPUtil/FPBits.h
@@ -31,6 +31,40 @@ enum class FPType {
X86_Binary80,
};
+// The classes hierarchy is as follows:
+//
+// ┌───────────────────┐
+// │ FPLayout<FPType> │
+// └─────────▲─────────┘
+// │
+// ┌─────────┴─────────┐
+// │ FPRepBase<FPType> │
+// └─────────▲─────────┘
+// │
+// ┌────────────┴─────────────┐
+// │ │
+// ┌────────┴──────┐ ┌─────────────┴──────────────┐
+// │ FPRep<FPType> │ │ FPRep<FPType::X86_Binary80 │
+// └────────▲──────┘ └─────────────▲──────────────┘
+// │ │
+// └────────────┬─────────────┘
+// │
+// ┌─────┴─────┐
+// │ FPBits<T> │
+// └───────────┘
+//
+// - 'FPLayout' defines only a few constants, namely the 'StorageType' and the
+// length of the sign, the exponent and significand parts.
+// - 'FPRepBase' builds more constants on top of those from 'FPLayout' like
+// exponent bias, shifts and masks. It also defines tools to assemble or test
+// these parts.
+// - 'FPRep' defines functions to interact with the floating point
+// representation. The default implementation is the one for 'IEEE754', a
+// specialization is provided for X86 Extended Precision that has a different
+// encoding.
+// - 'FPBits' is templated on the platform floating point types. Contrary to
+// 'FPRep' that is platform agnostic 'FPBits' is architecture dependent.
+
namespace internal {
// Defines the layout (sign, exponent, significand) of a floating point type in
@@ -132,11 +166,124 @@ struct FPRepBase : public internal::FPLayout<fp_type> {
static_assert((SIG_MASK & EXP_MASK & SIGN_MASK) == 0, "masks disjoint");
static_assert((SIG_MASK | EXP_MASK | SIGN_MASK) == FP_MASK, "masks cover");
-private:
+protected:
LIBC_INLINE static constexpr StorageType bit_at(int position) {
return StorageType(1) << position;
}
+ template <typename T> struct Opaque {
+ using value_type = T;
+ LIBC_INLINE constexpr explicit Opaque(T value) : value(value) {}
+ LIBC_INLINE constexpr Opaque(const Opaque &value) = default;
+
+ LIBC_INLINE constexpr explicit operator T() const { return value; }
+
+ private:
+ T value;
+ };
+
+ // An opaque type to store a floating point exponent.
+ // We define special values but it is valid to create arbitrary values as long
+ // as they are in the range [MIN, MAX].
+ struct Exponent : public Opaque<int32_t> {
+ using UP = Opaque<int32_t>;
+ using UP::UP;
+ LIBC_INLINE
+ static constexpr auto MIN() { return Exponent{1 - EXP_BIAS}; }
+ LIBC_INLINE static constexpr auto ZERO() { return Exponent{0}; }
+ LIBC_INLINE static constexpr auto MAX() { return Exponent{EXP_BIAS}; }
+ };
+
+ // An opaque type to store a floating point biased exponent.
+ // We define special values but it is valid to create arbitrary values as long
+ // as they are in the range [BITS_ALL_ZEROES, BITS_ALL_ONES].
+ // Values greater than BITS_ALL_ONES are truncated.
+ struct BiasedExponent : public Opaque<uint32_t> {
+ using UP = Opaque<uint32_t>;
+ using UP::UP;
+
+ LIBC_INLINE constexpr BiasedExponent(Exponent exp)
+ : UP(static_cast<int32_t>(exp) + EXP_BIAS) {}
+ // The exponent value for denormal numbers.
+ LIBC_INLINE static constexpr auto BITS_ALL_ZEROES() {
+ return BiasedExponent{uint32_t(0)};
+ }
+ // The exponent value for infinity.
+ LIBC_INLINE static constexpr auto BITS_ALL_ONES() {
+ return BiasedExponent{uint32_t(2 * EXP_BIAS + 1)};
+ }
+ };
+
+ // An opaque type to store a floating point significand.
+ // We define special values but it is valid to create arbitrary values as long
+ // as they are in the range [BITS_ALL_ZEROES, BITS_ALL_ONES].
+ // Note that the semantics of the Significand are implementation dependent.
+ // Values greater than BITS_ALL_ONES are truncated.
+ struct Significand : public Opaque<StorageType> {
+ using UP = Opaque<StorageType>;
+ using UP::UP;
+
+ LIBC_INLINE static constexpr auto ZERO() {
+ return Significand{StorageType(0)};
+ }
+ LIBC_INLINE static constexpr auto LSB() {
+ return Significand{StorageType(1)};
+ }
+ LIBC_INLINE static constexpr auto MSB() {
+ return Significand{StorageType(bit_at(SIG_LEN - 1))};
+ }
+ // Aliases
+ LIBC_INLINE static constexpr auto BITS_ALL_ZEROES() { return ZERO(); }
+ LIBC_INLINE static constexpr auto BITS_ALL_ONES() {
+ return Significand{SIG_MASK};
+ }
+ };
+
+ template <typename To, typename T>
+ LIBC_INLINE static constexpr To as(Opaque<T> opaque) {
+ return To(static_cast<T>(opaque));
+ }
+
+ LIBC_INLINE friend constexpr Significand operator|(const Significand a,
+ const Significand b) {
+ return Significand{StorageType(as<StorageType>(a) | as<StorageType>(b))};
+ }
+ LIBC_INLINE friend constexpr Significand operator^(const Significand a,
+ const Significand b) {
+ return Significand{StorageType(as<StorageType>(a) ^ as<StorageType>(b))};
+ }
+ LIBC_INLINE friend constexpr Significand operator>>(const Significand a,
+ int shift) {
+ return Significand{StorageType(as<StorageType>(a) >> shift)};
+ }
+
+ LIBC_INLINE static constexpr StorageType encode(BiasedExponent exp) {
+ return (as<StorageType>(exp) << SIG_LEN) & EXP_MASK;
+ }
+
+ LIBC_INLINE static constexpr StorageType encode(Significand value) {
+ return as<StorageType>(value) & SIG_MASK;
+ }
+
+ LIBC_INLINE static constexpr StorageType encode(BiasedExponent exp,
+ Significand sig) {
+ return encode(exp) | encode(sig);
+ }
+
+ LIBC_INLINE static constexpr StorageType encode(bool sign, BiasedExponent exp,
+ Significand sig) {
+ if (sign)
+ return SIGN_MASK | encode(exp, sig);
+ return encode(exp, sig);
+ }
+
+ LIBC_INLINE constexpr StorageType exp_bits() const { return bits & EXP_MASK; }
+ LIBC_INLINE constexpr StorageType sig_bits() const { return bits & SIG_MASK; }
+ LIBC_INLINE constexpr StorageType exp_sig_bits() const {
+ return bits & EXP_SIG_MASK;
+ }
+
+private:
// Merge bits from 'a' and 'b' values according to 'mask'.
// Use 'a' bits when corresponding 'mask' bits are zeroes and 'b' bits when
// corresponding bits are ones.
@@ -155,20 +302,6 @@ struct FPRepBase : public internal::FPLayout<fp_type> {
LIBC_INLINE_VAR static constexpr StorageType FRACTION_MASK =
mask_trailing_ones<StorageType, FRACTION_LEN>();
- // If a number x is a NAN, then it is a quiet NAN if:
- // QUIET_NAN_MASK & bits(x) != 0
- LIBC_INLINE_VAR static constexpr StorageType QUIET_NAN_MASK =
- fp_type == FPType::X86_Binary80
- ? bit_at(SIG_LEN - 1) | bit_at(SIG_LEN - 2) // 0b1100...
- : bit_at(SIG_LEN - 1); // 0b1000...
-
- // Mask to generate a default signaling NAN. Any NAN that is not
- // a quiet NAN is considered a signaling NAN.
- LIBC_INLINE_VAR static constexpr StorageType DEFAULT_SIGNALING_NAN =
- fp_type == FPType::X86_Binary80
- ? bit_at(SIG_LEN - 1) | bit_at(SIG_LEN - 3) // 0b1010...
- : bit_at(SIG_LEN - 2); // 0b0100...
-
// The floating point number representation as an unsigned integer.
StorageType bits = 0;
@@ -220,11 +353,14 @@ struct FPRepBase : public internal::FPLayout<fp_type> {
}
LIBC_INLINE constexpr StorageType uintval() const { return bits & FP_MASK; }
+ LIBC_INLINE constexpr void set_uintval(StorageType value) {
+ bits = (value & FP_MASK);
+ }
LIBC_INLINE constexpr bool is_zero() const {
return (bits & EXP_SIG_MASK) == 0;
}
-};
+}; // namespace fputil
namespace internal {
@@ -241,6 +377,214 @@ template <FPType fp_type> struct FPRep : public FPRepBase<fp_type> {
using UP::FRACTION_LEN;
using UP::FRACTION_MASK;
using UP::MANTISSA_PRECISION;
+
+protected:
+ using typename UP::BiasedExponent;
+ using typename UP::Exponent;
+ using typename UP::Significand;
+ using UP::encode;
+ using UP::exp_bits;
+ using UP::exp_sig_bits;
+ using UP::sig_bits;
+
+public:
+ LIBC_INLINE constexpr bool is_nan() const {
+ return exp_sig_bits() >
+ encode(BiasedExponent::BITS_ALL_ONES(), Significand::ZERO());
+ }
+ LIBC_INLINE constexpr bool is_quiet_nan() const {
+ return exp_sig_bits() >=
+ encode(BiasedExponent::BITS_ALL_ONES(), Significand::MSB());
+ }
+ LIBC_INLINE constexpr bool is_signaling_nan() const {
+ return is_nan() && !is_quiet_nan();
+ }
+ LIBC_INLINE constexpr bool is_inf() const {
+ return exp_sig_bits() ==
+ encode(BiasedExponent::BITS_ALL_ONES(), Significand::ZERO());
+ }
+ LIBC_INLINE constexpr bool is_zero() const {
+ return exp_sig_bits() ==
+ encode(BiasedExponent::BITS_ALL_ZEROES(), Significand::ZERO());
+ }
+ LIBC_INLINE constexpr bool is_finite() const {
+ return exp_bits() != encode(BiasedExponent::BITS_ALL_ONES());
+ }
+ LIBC_INLINE
+ constexpr bool is_subnormal() const {
+ return exp_bits() == encode(BiasedExponent::BITS_ALL_ZEROES());
+ }
+ LIBC_INLINE constexpr bool is_normal() const {
+ return is_finite() && !is_subnormal();
+ }
+
+ LIBC_INLINE static constexpr StorageType zero(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(), Significand::ZERO());
+ }
+ LIBC_INLINE static constexpr StorageType one(bool sign = false) {
+ return encode(sign, Exponent::ZERO(), Significand::ZERO());
+ }
+ LIBC_INLINE static constexpr StorageType min_subnormal(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(), Significand::LSB());
+ }
+ LIBC_INLINE static constexpr StorageType max_subnormal(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(),
+ Significand::BITS_ALL_ONES());
+ }
+ LIBC_INLINE static constexpr StorageType min_normal(bool sign = false) {
+ return encode(sign, Exponent::MIN(), Significand::ZERO());
+ }
+ LIBC_INLINE static constexpr StorageType max_normal(bool sign = false) {
+ return encode(sign, Exponent::MAX(), Significand::BITS_ALL_ONES());
+ }
+ LIBC_INLINE static constexpr StorageType inf(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(), Significand::ZERO());
+ }
+ LIBC_INLINE static constexpr StorageType build_nan(bool sign = false,
+ StorageType v = 0) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(),
+ (v ? Significand{v} : (Significand::MSB() >> 1)));
+ }
+ LIBC_INLINE static constexpr StorageType build_quiet_nan(bool sign = false,
+ StorageType v = 0) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(),
+ Significand::MSB() | Significand{v});
+ }
+
+ // The function return mantissa with the implicit bit set iff the current
+ // value is a valid normal number.
+ LIBC_INLINE constexpr StorageType get_explicit_mantissa() {
+ if (is_subnormal())
+ return sig_bits();
+ return (StorageType(1) << UP::SIG_LEN) | sig_bits();
+ }
+};
+
+// Specialization for the X86 Extended Precision type.
+template <>
+struct FPRep<FPType::X86_Binary80> : public FPRepBase<FPType::X86_Binary80> {
+ using UP = FPRepBase<FPType::X86_Binary80>;
+ using typename UP::StorageType;
+ using UP::FRACTION_LEN;
+ using UP::FRACTION_MASK;
+ using UP::MANTISSA_PRECISION;
+
+protected:
+ using typename UP::BiasedExponent;
+ using typename UP::Significand;
+ using UP::encode;
+
+public:
+ // 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");
+
+ LIBC_INLINE constexpr bool is_nan() const {
+ // Most encoding forms from the table found in
+ // https://en.wikipedia.org/wiki/Extended_precision#x86_extended_precision_format
+ // are interpreted as NaN.
+ // More precisely :
+ // - Pseudo-Infinity
+ // - Pseudo Not a Number
+ // - Signalling Not a Number
+ // - Floating-point Indefinite
+ // - Quiet Not a Number
+ // - Unnormal
+ // This can be reduced to the following logic:
+ if (exp_bits() == encode(BiasedExponent::BITS_ALL_ONES()))
+ return !is_inf();
+ if (exp_bits() != encode(BiasedExponent::BITS_ALL_ZEROES()))
+ return (sig_bits() & encode(Significand::MSB())) == 0;
+ return false;
+ }
+ LIBC_INLINE constexpr bool is_quiet_nan() const {
+ return exp_sig_bits() >=
+ encode(BiasedExponent::BITS_ALL_ONES(),
+ Significand::MSB() | (Significand::MSB() >> 1));
+ }
+ LIBC_INLINE constexpr bool is_signaling_nan() const {
+ return is_nan() && !is_quiet_nan();
+ }
+ LIBC_INLINE constexpr bool is_inf() const {
+ return exp_sig_bits() ==
+ encode(BiasedExponent::BITS_ALL_ONES(), Significand::MSB());
+ }
+ LIBC_INLINE constexpr bool is_zero() const {
+ return exp_sig_bits() ==
+ encode(BiasedExponent::BITS_ALL_ZEROES(), Significand::ZERO());
+ }
+ LIBC_INLINE constexpr bool is_finite() const {
+ return !is_inf() && !is_nan();
+ }
+ LIBC_INLINE
+ constexpr bool is_subnormal() const {
+ return exp_sig_bits() >
+ encode(BiasedExponent::BITS_ALL_ZEROES(), Significand::ZERO());
+ }
+ LIBC_INLINE constexpr bool is_normal() const {
+ const auto exp = exp_bits();
+ if (exp == encode(BiasedExponent::BITS_ALL_ZEROES()) ||
+ exp == encode(BiasedExponent::BITS_ALL_ONES()))
+ return false;
+ return get_implicit_bit();
+ }
+
+ LIBC_INLINE static constexpr StorageType zero(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(), Significand::ZERO());
+ }
+ LIBC_INLINE static constexpr StorageType one(bool sign = false) {
+ return encode(sign, Exponent::ZERO(), Significand::MSB());
+ }
+ LIBC_INLINE static constexpr StorageType min_subnormal(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(), Significand::LSB());
+ }
+ LIBC_INLINE static constexpr StorageType max_subnormal(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ZEROES(),
+ Significand::BITS_ALL_ONES() ^ Significand::MSB());
+ }
+ LIBC_INLINE static constexpr StorageType min_normal(bool sign = false) {
+ return encode(sign, Exponent::MIN(), Significand::MSB());
+ }
+ LIBC_INLINE static constexpr StorageType max_normal(bool sign = false) {
+ return encode(sign, Exponent::MAX(), Significand::BITS_ALL_ONES());
+ }
+ LIBC_INLINE static constexpr StorageType inf(bool sign = false) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(), Significand::MSB());
+ }
+ LIBC_INLINE static constexpr StorageType build_nan(bool sign = false,
+ StorageType v = 0) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(),
+ Significand::MSB() |
+ (v ? Significand{v} : (Significand::MSB() >> 2)));
+ }
+ LIBC_INLINE static constexpr StorageType build_quiet_nan(bool sign = false,
+ StorageType v = 0) {
+ return encode(sign, BiasedExponent::BITS_ALL_ONES(),
+ Significand::MSB() | (Significand::MSB() >> 1) |
+ Significand{v});
+ }
+
+ LIBC_INLINE constexpr StorageType get_explicit_mantissa() const {
+ return sig_bits();
+ }
+
+ // The following functions are specific to FPRep<FPType::X86_Binary80>.
+ // TODO: Remove if possible.
+ LIBC_INLINE constexpr bool get_implicit_bit() const {
+ return static_cast<bool>(bits & EXPLICIT_BIT_MASK);
+ }
+
+ LIBC_INLINE constexpr void set_implicit_bit(bool implicitVal) {
+ if (get_implicit_bit() != implicitVal)
+ bits ^= EXPLICIT_BIT_MASK;
+ }
};
} // namespace internal
@@ -276,47 +620,29 @@ template <typename T> LIBC_INLINE static constexpr FPType get_fp_type() {
static_assert(cpp::always_false<UnqualT>, "Unsupported type");
}
-// A generic class to represent single precision, double precision, and quad
-// precision IEEE 754 floating point formats.
+// A generic class to represent floating point formats.
// On most platforms, the 'float' type corresponds to single precision floating
// point numbers, the 'double' type corresponds to double precision floating
// point numers, and the 'long double' type corresponds to the quad precision
// floating numbers. On x86 platforms however, the 'long double' type maps to
-// an x87 floating point format. This format is an IEEE 754 extension format.
-// It is handled as an explicit specialization of this class.
+// an x87 floating point format.
template <typename T> struct FPBits : public internal::FPRep<get_fp_type<T>()> {
static_assert(cpp::is_floating_point_v<T>,
"FPBits instantiated with invalid type.");
using UP = internal::FPRep<get_fp_type<T>()>;
-
-private:
- using UP::EXP_SIG_MASK;
- using UP::QUIET_NAN_MASK;
- using UP::SIG_LEN;
- using UP::SIG_MASK;
-
-public:
+ using Rep = UP;
using StorageType = typename UP::StorageType;
+
using UP::bits;
- using UP::EXP_BIAS;
using UP::EXP_LEN;
- using UP::EXP_MASK;
- using UP::EXP_MASK_SHIFT;
- using UP::FRACTION_LEN;
- using UP::FRACTION_MASK;
- using UP::SIGN_MASK;
- using UP::TOTAL_LEN;
using UP::UP;
- using UP::get_biased_exponent;
- using UP::is_zero;
// Constants.
static constexpr int MAX_BIASED_EXPONENT = (1 << EXP_LEN) - 1;
- static constexpr StorageType MIN_SUBNORMAL = StorageType(1);
- static constexpr StorageType MAX_SUBNORMAL = FRACTION_MASK;
- static constexpr StorageType MIN_NORMAL = (StorageType(1) << FRACTION_LEN);
- static constexpr StorageType MAX_NORMAL =
- (StorageType(MAX_BIASED_EXPONENT - 1) << SIG_LEN) | SIG_MASK;
+ static constexpr StorageType MIN_NORMAL = UP::min_normal(false);
+ static constexpr StorageType MAX_NORMAL = UP::max_normal(false);
+ static constexpr StorageType MIN_SUBNORMAL = UP::min_subnormal(false);
+ static constexpr StorageType MAX_SUBNORMAL = UP::max_subnormal(false);
// Constructors.
LIBC_INLINE constexpr FPBits() = default;
@@ -338,88 +664,56 @@ template <typename T> struct FPBits : public internal::FPRep<get_fp_type<T>()> {
LIBC_INLINE constexpr explicit operator T() const { return get_val(); }
- // The function return mantissa with the implicit bit set iff the current
- // value is a valid normal number.
- LIBC_INLINE constexpr StorageType get_explicit_mantissa() {
- return ((get_biased_exponent() > 0 && !is_inf_or_nan())
- ? (FRACTION_MASK + 1)
- : 0) |
- (FRACTION_MASK & bits);
- }
-
- LIBC_INLINE constexpr bool is_inf() const {
- return (bits & EXP_SIG_MASK) == EXP_MASK;
- }
-
- LIBC_INLINE constexpr bool is_nan() const {
- return (bits & EXP_SIG_MASK) > EXP_MASK;
- }
-
- LIBC_INLINE constexpr bool is_quiet_nan() const {
- return (bits & EXP_SIG_MASK) >= (EXP_MASK | QUIET_NAN_MASK);
- }
-
- LIBC_INLINE constexpr bool is_inf_or_nan() const {
- return (bits & EXP_MASK) == EXP_MASK;
- }
+ LIBC_INLINE constexpr bool is_inf_or_nan() const { return !UP::is_finite(); }
LIBC_INLINE constexpr FPBits abs() const {
- return FPBits(bits & EXP_SIG_MASK);
+ return FPBits(bits & UP::EXP_SIG_MASK);
}
// Methods...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/78447
More information about the libc-commits
mailing list