summaryrefslogtreecommitdiff
path: root/libartbase/base/bit_utils.h
diff options
context:
space:
mode:
author Vladimir Marko <vmarko@google.com> 2022-04-14 10:59:53 +0000
committer Treehugger Robot <treehugger-gerrit@google.com> 2022-04-14 17:25:01 +0000
commit495311c865d1f4e08d71a2dd2a319b14c5f0409a (patch)
tree1bc6084161dc0e6885753c5851ce9fa7e7986fd8 /libartbase/base/bit_utils.h
parent998c2655397d281c8abcc4151290d3f33b7c604f (diff)
Modernize libartbase/ usage of <type_traits>...
... and other standard library templates. Use `std::*_t<...>` and `std::*_v<...>` instead of `std::*<...>::type` and `std::*<...>::value`. Remove some unnecessary "typename" qualifiers and second arguments to `static_assert()`. Test: Rely on TreeHugger. Change-Id: I4dc8e6fd905c60f652d3ac74bdf3d275b98daee8
Diffstat (limited to 'libartbase/base/bit_utils.h')
-rw-r--r--libartbase/base/bit_utils.h68
1 files changed, 34 insertions, 34 deletions
diff --git a/libartbase/base/bit_utils.h b/libartbase/base/bit_utils.h
index 58cc78ccca..ffaffc3a30 100644
--- a/libartbase/base/bit_utils.h
+++ b/libartbase/base/bit_utils.h
@@ -30,8 +30,8 @@ namespace art {
// Like sizeof, but count how many bits a type takes. Pass type explicitly.
template <typename T>
constexpr size_t BitSizeOf() {
- static_assert(std::is_integral<T>::value, "T must be integral");
- using unsigned_type = typename std::make_unsigned<T>::type;
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ using unsigned_type = std::make_unsigned_t<T>;
static_assert(sizeof(T) == sizeof(unsigned_type), "Unexpected type size mismatch!");
static_assert(std::numeric_limits<unsigned_type>::radix == 2, "Unexpected radix!");
return std::numeric_limits<unsigned_type>::digits;
@@ -45,8 +45,8 @@ constexpr size_t BitSizeOf(T /*x*/) {
template<typename T>
constexpr int CLZ(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- static_assert(std::is_unsigned<T>::value, "T must be unsigned");
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ static_assert(std::is_unsigned_v<T>, "T must be unsigned");
static_assert(std::numeric_limits<T>::radix == 2, "Unexpected radix!");
static_assert(sizeof(T) == sizeof(uint64_t) || sizeof(T) <= sizeof(uint32_t),
"Unsupported sizeof(T)");
@@ -60,14 +60,14 @@ constexpr int CLZ(T x) {
// Similar to CLZ except that on zero input it returns bitwidth and supports signed integers.
template<typename T>
constexpr int JAVASTYLE_CLZ(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- using unsigned_type = typename std::make_unsigned<T>::type;
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ using unsigned_type = std::make_unsigned_t<T>;
return (x == 0) ? BitSizeOf<T>() : CLZ(static_cast<unsigned_type>(x));
}
template<typename T>
constexpr int CTZ(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
+ static_assert(std::is_integral_v<T>, "T must be integral");
// It is not unreasonable to ask for trailing zeros in a negative number. As such, do not check
// that T is an unsigned type.
static_assert(sizeof(T) == sizeof(uint64_t) || sizeof(T) <= sizeof(uint32_t),
@@ -79,8 +79,8 @@ constexpr int CTZ(T x) {
// Similar to CTZ except that on zero input it returns bitwidth and supports signed integers.
template<typename T>
constexpr int JAVASTYLE_CTZ(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- using unsigned_type = typename std::make_unsigned<T>::type;
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ using unsigned_type = std::make_unsigned_t<T>;
return (x == 0) ? BitSizeOf<T>() : CTZ(static_cast<unsigned_type>(x));
}
@@ -105,8 +105,8 @@ constexpr T BSWAP(T x) {
// Find the bit position of the most significant bit (0-based), or -1 if there were no bits set.
template <typename T>
constexpr ssize_t MostSignificantBit(T value) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- static_assert(std::is_unsigned<T>::value, "T must be unsigned");
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ static_assert(std::is_unsigned_v<T>, "T must be unsigned");
static_assert(std::numeric_limits<T>::radix == 2, "Unexpected radix!");
return (value == 0) ? -1 : std::numeric_limits<T>::digits - 1 - CLZ(value);
}
@@ -114,8 +114,8 @@ constexpr ssize_t MostSignificantBit(T value) {
// Find the bit position of the least significant bit (0-based), or -1 if there were no bits set.
template <typename T>
constexpr ssize_t LeastSignificantBit(T value) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- static_assert(std::is_unsigned<T>::value, "T must be unsigned");
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ static_assert(std::is_unsigned_v<T>, "T must be unsigned");
return (value == 0) ? -1 : CTZ(value);
}
@@ -127,8 +127,8 @@ constexpr size_t MinimumBitsToStore(T value) {
template <typename T>
constexpr T RoundUpToPowerOfTwo(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- static_assert(std::is_unsigned<T>::value, "T must be unsigned");
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ static_assert(std::is_unsigned_v<T>, "T must be unsigned");
// NOTE: Undefined if x > (1 << (std::numeric_limits<T>::digits - 1)).
return (x < 2u) ? x : static_cast<T>(1u) << (std::numeric_limits<T>::digits - CLZ(x - 1u));
}
@@ -136,21 +136,21 @@ constexpr T RoundUpToPowerOfTwo(T x) {
// Return highest possible N - a power of two - such that val >= N.
template <typename T>
constexpr T TruncToPowerOfTwo(T val) {
- static_assert(std::is_integral<T>::value, "T must be integral");
- static_assert(std::is_unsigned<T>::value, "T must be unsigned");
+ static_assert(std::is_integral_v<T>, "T must be integral");
+ static_assert(std::is_unsigned_v<T>, "T must be unsigned");
return (val != 0) ? static_cast<T>(1u) << (BitSizeOf<T>() - CLZ(val) - 1u) : 0;
}
template<typename T>
constexpr bool IsPowerOfTwo(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
+ static_assert(std::is_integral_v<T>, "T must be integral");
// TODO: assert unsigned. There is currently many uses with signed values.
return (x & (x - 1)) == 0;
}
template<typename T>
constexpr int WhichPowerOf2(T x) {
- static_assert(std::is_integral<T>::value, "T must be integral");
+ static_assert(std::is_integral_v<T>, "T must be integral");
// TODO: assert unsigned. There is currently many uses with signed values.
DCHECK((x != 0) && IsPowerOfTwo(x));
return CTZ(x);
@@ -168,10 +168,10 @@ constexpr T RoundDown(T x, typename Identity<T>::type n) {
}
template<typename T>
-constexpr T RoundUp(T x, typename std::remove_reference<T>::type n) WARN_UNUSED;
+constexpr T RoundUp(T x, std::remove_reference_t<T> n) WARN_UNUSED;
template<typename T>
-constexpr T RoundUp(T x, typename std::remove_reference<T>::type n) {
+constexpr T RoundUp(T x, std::remove_reference_t<T> n) {
return RoundDown(x + n - 1, n);
}
@@ -265,7 +265,7 @@ template <size_t kBits, typename T>
constexpr bool IsInt(T value) {
static_assert(kBits > 0, "kBits cannot be zero.");
static_assert(kBits <= BitSizeOf<T>(), "kBits must be <= max.");
- static_assert(std::is_signed<T>::value, "Needs a signed type.");
+ static_assert(std::is_signed_v<T>, "Needs a signed type.");
// Corner case for "use all bits." Can't use the limits, as they would overflow, but it is
// trivially true.
return (kBits == BitSizeOf<T>()) ?
@@ -277,12 +277,12 @@ template <size_t kBits, typename T>
constexpr bool IsUint(T value) {
static_assert(kBits > 0, "kBits cannot be zero.");
static_assert(kBits <= BitSizeOf<T>(), "kBits must be <= max.");
- static_assert(std::is_integral<T>::value, "Needs an integral type.");
+ static_assert(std::is_integral_v<T>, "Needs an integral type.");
// Corner case for "use all bits." Can't use the limits, as they would overflow, but it is
// trivially true.
// NOTE: To avoid triggering assertion in GetIntLimit(kBits+1) if kBits+1==BitSizeOf<T>(),
// use GetIntLimit(kBits)*2u. The unsigned arithmetic works well for us if it overflows.
- using unsigned_type = typename std::make_unsigned<T>::type;
+ using unsigned_type = std::make_unsigned_t<T>;
return (0 <= value) &&
(kBits == BitSizeOf<T>() ||
(static_cast<unsigned_type>(value) <= GetIntLimit<unsigned_type>(kBits) * 2u - 1u));
@@ -291,8 +291,8 @@ constexpr bool IsUint(T value) {
template <size_t kBits, typename T>
constexpr bool IsAbsoluteUint(T value) {
static_assert(kBits <= BitSizeOf<T>(), "kBits must be <= max.");
- static_assert(std::is_integral<T>::value, "Needs an integral type.");
- using unsigned_type = typename std::make_unsigned<T>::type;
+ static_assert(std::is_integral_v<T>, "Needs an integral type.");
+ using unsigned_type = std::make_unsigned_t<T>;
return (kBits == BitSizeOf<T>())
? true
: IsUint<kBits>(value < 0
@@ -303,23 +303,23 @@ constexpr bool IsAbsoluteUint(T value) {
// Generate maximum/minimum values for signed/unsigned n-bit integers
template <typename T>
constexpr T MaxInt(size_t bits) {
- DCHECK(std::is_unsigned<T>::value || bits > 0u) << "bits cannot be zero for signed.";
+ DCHECK(std::is_unsigned_v<T> || bits > 0u) << "bits cannot be zero for signed.";
DCHECK_LE(bits, BitSizeOf<T>());
- using unsigned_type = typename std::make_unsigned<T>::type;
+ using unsigned_type = std::make_unsigned_t<T>;
return bits == BitSizeOf<T>()
? std::numeric_limits<T>::max()
- : std::is_signed<T>::value
+ : std::is_signed_v<T>
? ((bits == 1u) ? 0 : static_cast<T>(MaxInt<unsigned_type>(bits - 1)))
: static_cast<T>(UINT64_C(1) << bits) - static_cast<T>(1);
}
template <typename T>
constexpr T MinInt(size_t bits) {
- DCHECK(std::is_unsigned<T>::value || bits > 0) << "bits cannot be zero for signed.";
+ DCHECK(std::is_unsigned_v<T> || bits > 0) << "bits cannot be zero for signed.";
DCHECK_LE(bits, BitSizeOf<T>());
return bits == BitSizeOf<T>()
? std::numeric_limits<T>::min()
- : std::is_signed<T>::value
+ : std::is_signed_v<T>
? ((bits == 1u) ? -1 : static_cast<T>(-1) - MaxInt<T>(bits))
: static_cast<T>(0);
}
@@ -334,7 +334,7 @@ inline static kind LowestOneBitValue(kind opnd) {
// Returns value with bit set in hightest one-bit position or 0 if 0. (java.lang.X.highestOneBit).
template <typename T>
inline static T HighestOneBitValue(T opnd) {
- using unsigned_type = typename std::make_unsigned<T>::type;
+ using unsigned_type = std::make_unsigned_t<T>;
T res;
if (opnd == 0) {
res = 0;
@@ -351,7 +351,7 @@ inline static T Rot(T opnd, int distance) {
int mask = BitSizeOf<T>() - 1;
int unsigned_right_shift = left ? (-distance & mask) : (distance & mask);
int signed_left_shift = left ? (distance & mask) : (-distance & mask);
- using unsigned_type = typename std::make_unsigned<T>::type;
+ using unsigned_type = std::make_unsigned_t<T>;
return (static_cast<unsigned_type>(opnd) >> unsigned_right_shift) | (opnd << signed_left_shift);
}
@@ -485,7 +485,7 @@ inline static constexpr T BitFieldExtract(T value, size_t lsb, size_t width) {
const T bitfield_unsigned =
static_cast<T>((val >> lsb) & MaskLeastSignificant<T>(width));
- if (std::is_signed<T>::value) {
+ if (std::is_signed_v<T>) {
// Perform sign extension
if (width == 0) { // Avoid underflow.
return static_cast<T>(0);