/* * Copyright 2015 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #ifndef ANDROID_MODULO_H #define ANDROID_MODULO_H namespace android { // Modulo class is used for intentionally wrapping variables such as // counters and timers. // // It may also be used for variables whose computation depends on the // associativity of addition or subtraction. // // Features: // 1) Modulo checks type sizes before performing operations to ensure // that the wrap points match. This is critical for safe modular arithmetic. // 2) Modulo returns Modulo types from arithmetic operations, thereby // avoiding unintentional use in a non-modular computation. A Modulo // type is converted to its base non-Modulo type through the value() function. // 3) Modulo separates out overflowable types from non-overflowable types. // A signed overflow is technically undefined in C and C++. // Modulo types do not participate in sanitization. // 4) Modulo comparisons are based on signed differences to account for wrap; // this is not the same as the direct comparison of values. // 5) Safe use of binary arithmetic operations relies on conversions of // signed operands to unsigned operands (which are modular arithmetic safe). // Conversions which are implementation-defined are assumed to use 2's complement // representation. (See A, B, C, D from the ISO/IEC FDIS 14882 // Information technology — Programming languages — C++). // // A: ISO/IEC 14882:2011(E) p84 section 4.7 Integral conversions // (2) If the destination type is unsigned, the resulting value is the least unsigned // integer congruent to the source integer (modulo 2^n where n is the number of bits // used to represent the unsigned type). [ Note: In a two’s complement representation, // this conversion is conceptual and there is no change in the bit pattern (if there // is no truncation). — end note ] // (3) If the destination type is signed, the value is unchanged if it can be represented // in the destination type (and bit-field width); otherwise, the value is // implementation-defined. // // B: ISO/IEC 14882:2011(E) p88 section 5 Expressions // (9) Many binary operators that expect operands of arithmetic or enumeration type // cause conversions and yield result types in a similar way. The purpose is to // yield a common type, which is also the type of the result. This pattern is called // the usual arithmetic conversions, which are defined as follows: // [...] // Otherwise, if both operands have signed integer types or both have unsigned // integer types, the operand with the type of lesser integer conversion rank shall be // converted to the type of the operand with greater rank. // — Otherwise, if the operand that has unsigned integer type has rank greater than // or equal to the rank of the type of the other operand, the operand with signed // integer type shall be converted to the type of the operand with unsigned integer type. // // C: ISO/IEC 14882:2011(E) p86 section 4.13 Integer conversion rank // [...] The rank of long long int shall be greater than the rank of long int, // which shall be greater than the rank of int, which shall be greater than the // rank of short int, which shall be greater than the rank of signed char. // — The rank of any unsigned integer type shall equal the rank of the corresponding // signed integer type. // // D: ISO/IEC 14882:2011(E) p75 section 3.9.1 Fundamental types // [...] Unsigned integers, declared unsigned, shall obey the laws of arithmetic modulo // 2^n where n is the number of bits in the value representation of that particular // size of integer. // // Note: // Other libraries do exist for safe integer operations which can detect the // possibility of overflow (SafeInt from MS and safe-iop in android). // Signed safe computation is also possible from the art header safe_math.h. template <typename T> class Modulo { T mValue; public: typedef typename std::make_signed<T>::type signedT; typedef typename std::make_unsigned<T>::type unsignedT; Modulo() { } // intentionally uninitialized data Modulo(const T &value) { mValue = value; } const T & value() const { return mValue; } // not assignable signedT signedValue() const { return mValue; } unsignedT unsignedValue() const { return mValue; } void getValue(T *value) const { *value = mValue; } // more type safe than value() // modular operations valid only if size of T <= size of S. template <typename S> __attribute__((no_sanitize("integer"))) Modulo<T> operator +=(const Modulo<S> &other) { static_assert(sizeof(T) <= sizeof(S), "argument size mismatch"); mValue += other.unsignedValue(); return *this; } template <typename S> __attribute__((no_sanitize("integer"))) Modulo<T> operator -=(const Modulo<S> &other) { static_assert(sizeof(T) <= sizeof(S), "argument size mismatch"); mValue -= other.unsignedValue(); return *this; } // modular operations resulting in a value valid only at the smaller of the two // Modulo base type sizes, but we only allow equal sizes to avoid confusion. template <typename S> __attribute__((no_sanitize("integer"))) const Modulo<T> operator +(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return Modulo<T>(mValue + other.unsignedValue()); } template <typename S> __attribute__((no_sanitize("integer"))) const Modulo<T> operator -(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return Modulo<T>(mValue - other.unsignedValue()); } // modular operations that should be checked only at the smaller of // the two type sizes, but we only allow equal sizes to avoid confusion. // // Caution: These relational and comparison operations are not equivalent to // the base type operations. template <typename S> __attribute__((no_sanitize("integer"))) bool operator >(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return static_cast<signedT>(mValue - other.unsignedValue()) > 0; } template <typename S> __attribute__((no_sanitize("integer"))) bool operator >=(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return static_cast<signedT>(mValue - other.unsignedValue()) >= 0; } template <typename S> __attribute__((no_sanitize("integer"))) bool operator ==(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return static_cast<signedT>(mValue - other.unsignedValue()) == 0; } template <typename S> __attribute__((no_sanitize("integer"))) bool operator <=(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return static_cast<signedT>(mValue - other.unsignedValue()) <= 0; } template <typename S> __attribute__((no_sanitize("integer"))) bool operator <(const Modulo<S> &other) const { static_assert(sizeof(T) == sizeof(S), "argument size mismatch"); return static_cast<signedT>(mValue - other.unsignedValue()) < 0; } // modular operations with a non-Modulo type allowed with wrapping // because there should be no confusion as to the meaning. template <typename S> __attribute__((no_sanitize("integer"))) Modulo<T> operator +=(const S &other) { mValue += unsignedT(other); return *this; } template <typename S> __attribute__((no_sanitize("integer"))) Modulo<T> operator -=(const S &other) { mValue -= unsignedT(other); return *this; } // modular operations with a non-Modulo type allowed with wrapping, // but we restrict this only when size of T is greater than or equal to // the size of S to avoid confusion with the nature of overflow. // // Use of this follows left-associative style. // // Note: a Modulo type may be promoted by using "differences" off of // a larger sized type, but we do not automate this. template <typename S> __attribute__((no_sanitize("integer"))) const Modulo<T> operator +(const S &other) const { static_assert(sizeof(T) >= sizeof(S), "argument size mismatch"); return Modulo<T>(mValue + unsignedT(other)); } template <typename S> __attribute__((no_sanitize("integer"))) const Modulo<T> operator -(const S &other) const { static_assert(sizeof(T) >= sizeof(S), "argument size mismatch"); return Modulo<T>(mValue - unsignedT(other)); } // multiply is intentionally omitted, but it is a common operator in // modular arithmetic. // shift operations are intentionally omitted, but perhaps useful. // For example, left-shifting a negative number is undefined in C++11. }; } // namespace android #endif /* ANDROID_MODULO_H */