| // Copyright 2013 The Chromium Authors | 
 | // Use of this source code is governed by a BSD-style license that can be | 
 | // found in the LICENSE file. | 
 |  | 
 | // This file defines some bit utilities. | 
 |  | 
 | #ifndef BASE_BITS_H_ | 
 | #define BASE_BITS_H_ | 
 |  | 
 | #include <limits.h> | 
 | #include <stddef.h> | 
 | #include <stdint.h> | 
 |  | 
 | #include <type_traits> | 
 |  | 
 | #include "base/check.h" | 
 | #include "base/compiler_specific.h" | 
 | #include "build/build_config.h" | 
 |  | 
 | namespace base { | 
 | namespace bits { | 
 |  | 
 | // Returns true iff |value| is a power of 2. | 
 | // | 
 | // TODO(pkasting): When C++20 is available, replace with std::has_single_bit(). | 
 | template <typename T, typename = std::enable_if_t<std::is_integral<T>::value>> | 
 | constexpr bool IsPowerOfTwo(T value) { | 
 |   // From "Hacker's Delight": Section 2.1 Manipulating Rightmost Bits. | 
 |   // | 
 |   // Only positive integers with a single bit set are powers of two. If only one | 
 |   // bit is set in x (e.g. 0b00000100000000) then |x-1| will have that bit set | 
 |   // to zero and all bits to its right set to 1 (e.g. 0b00000011111111). Hence | 
 |   // |x & (x-1)| is 0 iff x is a power of two. | 
 |   return value > 0 && (value & (value - 1)) == 0; | 
 | } | 
 |  | 
 | // Round down |size| to a multiple of alignment, which must be a power of two. | 
 | template <typename T, typename = std::enable_if_t<std::is_integral_v<T>>> | 
 | constexpr T AlignDown(T size, T alignment) { | 
 |   DCHECK(IsPowerOfTwo(alignment)); | 
 |   return size & ~(alignment - 1); | 
 | } | 
 |  | 
 | // Move |ptr| back to the previous multiple of alignment, which must be a power | 
 | // of two. Defined for types where sizeof(T) is one byte. | 
 | template <typename T, typename = typename std::enable_if<sizeof(T) == 1>::type> | 
 | inline T* AlignDown(T* ptr, uintptr_t alignment) { | 
 |   return reinterpret_cast<T*>( | 
 |       AlignDown(reinterpret_cast<uintptr_t>(ptr), alignment)); | 
 | } | 
 |  | 
 | // Round up |size| to a multiple of alignment, which must be a power of two. | 
 | template <typename T, typename = std::enable_if_t<std::is_integral_v<T>>> | 
 | constexpr T AlignUp(T size, T alignment) { | 
 |   DCHECK(IsPowerOfTwo(alignment)); | 
 |   return (size + alignment - 1) & ~(alignment - 1); | 
 | } | 
 |  | 
 | // Advance |ptr| to the next multiple of alignment, which must be a power of | 
 | // two. Defined for types where sizeof(T) is one byte. | 
 | template <typename T, typename = typename std::enable_if<sizeof(T) == 1>::type> | 
 | inline T* AlignUp(T* ptr, uintptr_t alignment) { | 
 |   return reinterpret_cast<T*>( | 
 |       AlignUp(reinterpret_cast<uintptr_t>(ptr), alignment)); | 
 | } | 
 |  | 
 | // CountLeadingZeroBits(value) returns the number of zero bits following the | 
 | // most significant 1 bit in |value| if |value| is non-zero, otherwise it | 
 | // returns {sizeof(T) * 8}. | 
 | // Example: 00100010 -> 2 | 
 | // | 
 | // CountTrailingZeroBits(value) returns the number of zero bits preceding the | 
 | // least significant 1 bit in |value| if |value| is non-zero, otherwise it | 
 | // returns {sizeof(T) * 8}. | 
 | // Example: 00100010 -> 1 | 
 | // | 
 | // C does not have an operator to do this, but fortunately the various | 
 | // compilers have built-ins that map to fast underlying processor instructions. | 
 | // | 
 | // TODO(pkasting): When C++20 is available, replace with std::countl_zero() and | 
 | // similar. | 
 |  | 
 | // __builtin_clz has undefined behaviour for an input of 0, even though there's | 
 | // clearly a return value that makes sense, and even though some processor clz | 
 | // instructions have defined behaviour for 0. We could drop to raw __asm__ to | 
 | // do better, but we'll avoid doing that unless we see proof that we need to. | 
 | template <typename T, int bits = sizeof(T) * 8> | 
 | ALWAYS_INLINE constexpr | 
 |     typename std::enable_if<std::is_unsigned<T>::value && sizeof(T) <= 8, | 
 |                             int>::type | 
 |     CountLeadingZeroBits(T value) { | 
 |   static_assert(bits > 0, "invalid instantiation"); | 
 |   return LIKELY(value) | 
 |              ? bits == 64 | 
 |                    ? __builtin_clzll(static_cast<uint64_t>(value)) | 
 |                    : __builtin_clz(static_cast<uint32_t>(value)) - (32 - bits) | 
 |              : bits; | 
 | } | 
 |  | 
 | template <typename T, int bits = sizeof(T) * 8> | 
 | ALWAYS_INLINE constexpr | 
 |     typename std::enable_if<std::is_unsigned<T>::value && sizeof(T) <= 8, | 
 |                             int>::type | 
 |     CountTrailingZeroBits(T value) { | 
 |   return LIKELY(value) ? bits == 64 | 
 |                              ? __builtin_ctzll(static_cast<uint64_t>(value)) | 
 |                              : __builtin_ctz(static_cast<uint32_t>(value)) | 
 |                        : bits; | 
 | } | 
 |  | 
 | // Returns the integer i such as 2^i <= n < 2^(i+1). | 
 | // | 
 | // There is a common `BitLength` function, which returns the number of bits | 
 | // required to represent a value. Rather than implement that function, | 
 | // use `Log2Floor` and add 1 to the result. | 
 | // | 
 | // TODO(pkasting): When C++20 is available, replace with std::bit_xxx(). | 
 | constexpr int Log2Floor(uint32_t n) { | 
 |   return 31 - CountLeadingZeroBits(n); | 
 | } | 
 |  | 
 | // Returns the integer i such as 2^(i-1) < n <= 2^i. | 
 | constexpr int Log2Ceiling(uint32_t n) { | 
 |   // When n == 0, we want the function to return -1. | 
 |   // When n == 0, (n - 1) will underflow to 0xFFFFFFFF, which is | 
 |   // why the statement below starts with (n ? 32 : -1). | 
 |   return (n ? 32 : -1) - CountLeadingZeroBits(n - 1); | 
 | } | 
 |  | 
 | // Returns a value of type T with a single bit set in the left-most position. | 
 | // Can be used instead of manually shifting a 1 to the left. | 
 | template <typename T> | 
 | constexpr T LeftmostBit() { | 
 |   static_assert(std::is_integral<T>::value, | 
 |                 "This function can only be used with integral types."); | 
 |   T one(1u); | 
 |   return one << ((CHAR_BIT * sizeof(T) - 1)); | 
 | } | 
 |  | 
 | }  // namespace bits | 
 | }  // namespace base | 
 |  | 
 | #endif  // BASE_BITS_H_ |