mirror of
https://github.com/mozilla/gecko-dev.git
synced 2024-11-23 21:01:08 +00:00
003f31a5b7
Differential Revision: https://phabricator.services.mozilla.com/D182676
620 lines
22 KiB
C++
620 lines
22 KiB
C++
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
|
|
/* This Source Code Form is subject to the terms of the Mozilla Public
|
|
* License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
|
|
|
|
#ifndef mozilla_TimeStamp_h
|
|
#define mozilla_TimeStamp_h
|
|
|
|
#include "mozilla/Assertions.h"
|
|
#include "mozilla/Attributes.h"
|
|
#include "mozilla/FloatingPoint.h"
|
|
#include "mozilla/Types.h"
|
|
#include <algorithm> // for std::min, std::max
|
|
#include <ostream>
|
|
#include <stdint.h>
|
|
#include <type_traits>
|
|
|
|
namespace IPC {
|
|
template <typename T>
|
|
struct ParamTraits;
|
|
} // namespace IPC
|
|
|
|
#ifdef XP_WIN
|
|
// defines TimeStampValue as a complex value keeping both
|
|
// GetTickCount and QueryPerformanceCounter values
|
|
# include "TimeStamp_windows.h"
|
|
|
|
# include "mozilla/Maybe.h" // For TimeStamp::RawQueryPerformanceCounterValue
|
|
#endif
|
|
|
|
namespace mozilla {
|
|
|
|
#ifndef XP_WIN
|
|
typedef uint64_t TimeStampValue;
|
|
#endif
|
|
|
|
class TimeStamp;
|
|
class TimeStampTests;
|
|
|
|
/**
|
|
* Platform-specific implementation details of BaseTimeDuration.
|
|
*/
|
|
class BaseTimeDurationPlatformUtils {
|
|
public:
|
|
static MFBT_API double ToSeconds(int64_t aTicks);
|
|
static MFBT_API double ToSecondsSigDigits(int64_t aTicks);
|
|
static MFBT_API int64_t TicksFromMilliseconds(double aMilliseconds);
|
|
static MFBT_API int64_t ResolutionInTicks();
|
|
};
|
|
|
|
/**
|
|
* Instances of this class represent the length of an interval of time.
|
|
* Negative durations are allowed, meaning the end is before the start.
|
|
*
|
|
* Internally the duration is stored as a int64_t in units of
|
|
* PR_TicksPerSecond() when building with NSPR interval timers, or a
|
|
* system-dependent unit when building with system clocks. The
|
|
* system-dependent unit must be constant, otherwise the semantics of
|
|
* this class would be broken.
|
|
*
|
|
* The ValueCalculator template parameter determines how arithmetic
|
|
* operations are performed on the integer count of ticks (mValue).
|
|
*/
|
|
template <typename ValueCalculator>
|
|
class BaseTimeDuration {
|
|
public:
|
|
// The default duration is 0.
|
|
constexpr BaseTimeDuration() : mValue(0) {}
|
|
// Allow construction using '0' as the initial value, for readability,
|
|
// but no other numbers (so we don't have any implicit unit conversions).
|
|
struct _SomethingVeryRandomHere;
|
|
MOZ_IMPLICIT BaseTimeDuration(_SomethingVeryRandomHere* aZero) : mValue(0) {
|
|
MOZ_ASSERT(!aZero, "Who's playing funny games here?");
|
|
}
|
|
// Default copy-constructor and assignment are OK
|
|
|
|
// Converting copy-constructor and assignment operator
|
|
template <typename E>
|
|
explicit BaseTimeDuration(const BaseTimeDuration<E>& aOther)
|
|
: mValue(aOther.mValue) {}
|
|
|
|
template <typename E>
|
|
BaseTimeDuration& operator=(const BaseTimeDuration<E>& aOther) {
|
|
mValue = aOther.mValue;
|
|
return *this;
|
|
}
|
|
|
|
double ToSeconds() const {
|
|
if (mValue == INT64_MAX) {
|
|
return PositiveInfinity<double>();
|
|
}
|
|
if (mValue == INT64_MIN) {
|
|
return NegativeInfinity<double>();
|
|
}
|
|
return BaseTimeDurationPlatformUtils::ToSeconds(mValue);
|
|
}
|
|
// Return a duration value that includes digits of time we think to
|
|
// be significant. This method should be used when displaying a
|
|
// time to humans.
|
|
double ToSecondsSigDigits() const {
|
|
if (mValue == INT64_MAX) {
|
|
return PositiveInfinity<double>();
|
|
}
|
|
if (mValue == INT64_MIN) {
|
|
return NegativeInfinity<double>();
|
|
}
|
|
return BaseTimeDurationPlatformUtils::ToSecondsSigDigits(mValue);
|
|
}
|
|
double ToMilliseconds() const { return ToSeconds() * 1000.0; }
|
|
double ToMicroseconds() const { return ToMilliseconds() * 1000.0; }
|
|
|
|
// Using a double here is safe enough; with 53 bits we can represent
|
|
// durations up to over 280,000 years exactly. If the units of
|
|
// mValue do not allow us to represent durations of that length,
|
|
// long durations are clamped to the max/min representable value
|
|
// instead of overflowing.
|
|
static inline BaseTimeDuration FromSeconds(double aSeconds) {
|
|
return FromMilliseconds(aSeconds * 1000.0);
|
|
}
|
|
static BaseTimeDuration FromMilliseconds(double aMilliseconds) {
|
|
if (aMilliseconds == PositiveInfinity<double>()) {
|
|
return Forever();
|
|
}
|
|
if (aMilliseconds == NegativeInfinity<double>()) {
|
|
return FromTicks(INT64_MIN);
|
|
}
|
|
return FromTicks(
|
|
BaseTimeDurationPlatformUtils::TicksFromMilliseconds(aMilliseconds));
|
|
}
|
|
static inline BaseTimeDuration FromMicroseconds(double aMicroseconds) {
|
|
return FromMilliseconds(aMicroseconds / 1000.0);
|
|
}
|
|
|
|
static constexpr BaseTimeDuration Zero() { return BaseTimeDuration(); }
|
|
static constexpr BaseTimeDuration Forever() { return FromTicks(INT64_MAX); }
|
|
|
|
BaseTimeDuration operator+(const BaseTimeDuration& aOther) const {
|
|
return FromTicks(ValueCalculator::Add(mValue, aOther.mValue));
|
|
}
|
|
BaseTimeDuration operator-(const BaseTimeDuration& aOther) const {
|
|
return FromTicks(ValueCalculator::Subtract(mValue, aOther.mValue));
|
|
}
|
|
BaseTimeDuration& operator+=(const BaseTimeDuration& aOther) {
|
|
mValue = ValueCalculator::Add(mValue, aOther.mValue);
|
|
return *this;
|
|
}
|
|
BaseTimeDuration& operator-=(const BaseTimeDuration& aOther) {
|
|
mValue = ValueCalculator::Subtract(mValue, aOther.mValue);
|
|
return *this;
|
|
}
|
|
BaseTimeDuration operator-() const {
|
|
// We don't just use FromTicks(ValueCalculator::Subtract(0, mValue))
|
|
// since that won't give the correct result for -TimeDuration::Forever().
|
|
int64_t ticks;
|
|
if (mValue == INT64_MAX) {
|
|
ticks = INT64_MIN;
|
|
} else if (mValue == INT64_MIN) {
|
|
ticks = INT64_MAX;
|
|
} else {
|
|
ticks = -mValue;
|
|
}
|
|
|
|
return FromTicks(ticks);
|
|
}
|
|
|
|
static BaseTimeDuration Max(const BaseTimeDuration& aA,
|
|
const BaseTimeDuration& aB) {
|
|
return FromTicks(std::max(aA.mValue, aB.mValue));
|
|
}
|
|
static BaseTimeDuration Min(const BaseTimeDuration& aA,
|
|
const BaseTimeDuration& aB) {
|
|
return FromTicks(std::min(aA.mValue, aB.mValue));
|
|
}
|
|
|
|
#if defined(DEBUG)
|
|
int64_t GetValue() const { return mValue; }
|
|
#endif
|
|
|
|
private:
|
|
// Block double multiplier (slower, imprecise if long duration) - Bug 853398.
|
|
// If required, use MultDouble explicitly and with care.
|
|
BaseTimeDuration operator*(const double aMultiplier) const = delete;
|
|
|
|
// Block double divisor (for the same reason, and because dividing by
|
|
// fractional values would otherwise invoke the int64_t variant, and rounding
|
|
// the passed argument can then cause divide-by-zero) - Bug 1147491.
|
|
BaseTimeDuration operator/(const double aDivisor) const = delete;
|
|
|
|
public:
|
|
BaseTimeDuration MultDouble(double aMultiplier) const {
|
|
return FromTicks(ValueCalculator::Multiply(mValue, aMultiplier));
|
|
}
|
|
BaseTimeDuration operator*(const int32_t aMultiplier) const {
|
|
return FromTicks(ValueCalculator::Multiply(mValue, aMultiplier));
|
|
}
|
|
BaseTimeDuration operator*(const uint32_t aMultiplier) const {
|
|
return FromTicks(ValueCalculator::Multiply(mValue, aMultiplier));
|
|
}
|
|
BaseTimeDuration operator*(const int64_t aMultiplier) const {
|
|
return FromTicks(ValueCalculator::Multiply(mValue, aMultiplier));
|
|
}
|
|
BaseTimeDuration operator*(const uint64_t aMultiplier) const {
|
|
if (aMultiplier > INT64_MAX) {
|
|
return Forever();
|
|
}
|
|
return FromTicks(ValueCalculator::Multiply(mValue, aMultiplier));
|
|
}
|
|
BaseTimeDuration operator/(const int64_t aDivisor) const {
|
|
MOZ_ASSERT(aDivisor != 0, "Division by zero");
|
|
return FromTicks(ValueCalculator::Divide(mValue, aDivisor));
|
|
}
|
|
double operator/(const BaseTimeDuration& aOther) const {
|
|
MOZ_ASSERT(aOther.mValue != 0, "Division by zero");
|
|
return ValueCalculator::DivideDouble(mValue, aOther.mValue);
|
|
}
|
|
BaseTimeDuration operator%(const BaseTimeDuration& aOther) const {
|
|
MOZ_ASSERT(aOther.mValue != 0, "Division by zero");
|
|
return FromTicks(ValueCalculator::Modulo(mValue, aOther.mValue));
|
|
}
|
|
|
|
template <typename E>
|
|
bool operator<(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue < aOther.mValue;
|
|
}
|
|
template <typename E>
|
|
bool operator<=(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue <= aOther.mValue;
|
|
}
|
|
template <typename E>
|
|
bool operator>=(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue >= aOther.mValue;
|
|
}
|
|
template <typename E>
|
|
bool operator>(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue > aOther.mValue;
|
|
}
|
|
template <typename E>
|
|
bool operator==(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue == aOther.mValue;
|
|
}
|
|
template <typename E>
|
|
bool operator!=(const BaseTimeDuration<E>& aOther) const {
|
|
return mValue != aOther.mValue;
|
|
}
|
|
bool IsZero() const { return mValue == 0; }
|
|
explicit operator bool() const { return mValue != 0; }
|
|
|
|
friend std::ostream& operator<<(std::ostream& aStream,
|
|
const BaseTimeDuration& aDuration) {
|
|
return aStream << aDuration.ToMilliseconds() << " ms";
|
|
}
|
|
|
|
// Return a best guess at the system's current timing resolution,
|
|
// which might be variable. BaseTimeDurations below this order of
|
|
// magnitude are meaningless, and those at the same order of
|
|
// magnitude or just above are suspect.
|
|
static BaseTimeDuration Resolution() {
|
|
return FromTicks(BaseTimeDurationPlatformUtils::ResolutionInTicks());
|
|
}
|
|
|
|
// We could define additional operators here:
|
|
// -- convert to/from other time units
|
|
// -- scale duration by a float
|
|
// but let's do that on demand.
|
|
// Comparing durations for equality will only lead to bugs on
|
|
// platforms with high-resolution timers.
|
|
|
|
private:
|
|
friend class TimeStamp;
|
|
friend struct IPC::ParamTraits<mozilla::BaseTimeDuration<ValueCalculator>>;
|
|
template <typename>
|
|
friend class BaseTimeDuration;
|
|
|
|
static constexpr BaseTimeDuration FromTicks(int64_t aTicks) {
|
|
BaseTimeDuration t;
|
|
t.mValue = aTicks;
|
|
return t;
|
|
}
|
|
|
|
static BaseTimeDuration FromTicks(double aTicks) {
|
|
// NOTE: this MUST be a >= test, because int64_t(double(INT64_MAX))
|
|
// overflows and gives INT64_MIN.
|
|
if (aTicks >= double(INT64_MAX)) {
|
|
return FromTicks(INT64_MAX);
|
|
}
|
|
|
|
// This MUST be a <= test.
|
|
if (aTicks <= double(INT64_MIN)) {
|
|
return FromTicks(INT64_MIN);
|
|
}
|
|
|
|
return FromTicks(int64_t(aTicks));
|
|
}
|
|
|
|
// Duration, result is implementation-specific difference of two TimeStamps
|
|
int64_t mValue;
|
|
};
|
|
|
|
/**
|
|
* Perform arithmetic operations on the value of a BaseTimeDuration without
|
|
* doing strict checks on the range of values.
|
|
*/
|
|
class TimeDurationValueCalculator {
|
|
public:
|
|
static int64_t Add(int64_t aA, int64_t aB) { return aA + aB; }
|
|
static int64_t Subtract(int64_t aA, int64_t aB) { return aA - aB; }
|
|
|
|
template <typename T>
|
|
static int64_t Multiply(int64_t aA, T aB) {
|
|
static_assert(std::is_integral_v<T>,
|
|
"Using integer multiplication routine with non-integer type."
|
|
" Further specialization required");
|
|
return aA * static_cast<int64_t>(aB);
|
|
}
|
|
|
|
static int64_t Divide(int64_t aA, int64_t aB) { return aA / aB; }
|
|
static double DivideDouble(int64_t aA, int64_t aB) {
|
|
return static_cast<double>(aA) / aB;
|
|
}
|
|
static int64_t Modulo(int64_t aA, int64_t aB) { return aA % aB; }
|
|
};
|
|
|
|
template <>
|
|
inline int64_t TimeDurationValueCalculator::Multiply<double>(int64_t aA,
|
|
double aB) {
|
|
return static_cast<int64_t>(aA * aB);
|
|
}
|
|
|
|
/**
|
|
* Specialization of BaseTimeDuration that uses TimeDurationValueCalculator for
|
|
* arithmetic on the mValue member.
|
|
*
|
|
* Use this class for time durations that are *not* expected to hold values of
|
|
* Forever (or the negative equivalent) or when such time duration are *not*
|
|
* expected to be used in arithmetic operations.
|
|
*/
|
|
typedef BaseTimeDuration<TimeDurationValueCalculator> TimeDuration;
|
|
|
|
/**
|
|
* Instances of this class represent moments in time, or a special
|
|
* "null" moment. We do not use the non-monotonic system clock or
|
|
* local time, since they can be reset, causing apparent backward
|
|
* travel in time, which can confuse algorithms. Instead we measure
|
|
* elapsed time according to the system. This time can never go
|
|
* backwards (i.e. it never wraps around, at least not in less than
|
|
* five million years of system elapsed time). It might not advance
|
|
* while the system is sleeping. If TimeStamp::SetNow() is not called
|
|
* at all for hours or days, we might not notice the passage of some
|
|
* of that time.
|
|
*
|
|
* We deliberately do not expose a way to convert TimeStamps to some
|
|
* particular unit. All you can do is compute a difference between two
|
|
* TimeStamps to get a TimeDuration. You can also add a TimeDuration
|
|
* to a TimeStamp to get a new TimeStamp. You can't do something
|
|
* meaningless like add two TimeStamps.
|
|
*
|
|
* Internally this is implemented as either a wrapper around
|
|
* - high-resolution, monotonic, system clocks if they exist on this
|
|
* platform
|
|
* - PRIntervalTime otherwise. We detect wraparounds of
|
|
* PRIntervalTime and work around them.
|
|
*
|
|
* This class is similar to C++11's time_point, however it is
|
|
* explicitly nullable and provides an IsNull() method. time_point
|
|
* is initialized to the clock's epoch and provides a
|
|
* time_since_epoch() method that functions similiarly. i.e.
|
|
* t.IsNull() is equivalent to t.time_since_epoch() ==
|
|
* decltype(t)::duration::zero();
|
|
*
|
|
* Note that, since TimeStamp objects are small, prefer to pass them by value
|
|
* unless there is a specific reason not to do so.
|
|
*/
|
|
#if defined(XP_WIN)
|
|
// If this static_assert fails then possibly the warning comment below is no
|
|
// longer valid and should be removed.
|
|
static_assert(sizeof(TimeStampValue) > 8);
|
|
#endif
|
|
/*
|
|
* WARNING: On Windows, each TimeStamp is represented internally by two
|
|
* different raw values (one from GTC and one from QPC) and which value gets
|
|
* used for a given operation depends on whether both operands have QPC values
|
|
* or not. This duality of values can lead to some surprising results when
|
|
* mixing TimeStamps with and without QPC values, such as comparisons being
|
|
* non-transitive (ie, a > b > c might not imply a > c). See bug 1829983 for
|
|
* more details/an example.
|
|
*/
|
|
class TimeStamp {
|
|
public:
|
|
/**
|
|
* Initialize to the "null" moment
|
|
*/
|
|
constexpr TimeStamp() : mValue(0) {}
|
|
// Default copy-constructor and assignment are OK
|
|
|
|
/**
|
|
* The system timestamps are the same as the TimeStamp
|
|
* retrieved by mozilla::TimeStamp. Since we need this for
|
|
* vsync timestamps, we enable the creation of mozilla::TimeStamps
|
|
* on platforms that support vsync aligned refresh drivers / compositors
|
|
* Verified true as of Jan 31, 2015: B2G and OS X
|
|
* False on Windows 7
|
|
* Android's event time uses CLOCK_MONOTONIC via SystemClock.uptimeMilles.
|
|
* So it is same value of TimeStamp posix implementation.
|
|
* Wayland/GTK event time also uses CLOCK_MONOTONIC on Weston/Mutter
|
|
* compositors.
|
|
* UNTESTED ON OTHER PLATFORMS
|
|
*/
|
|
#if defined(XP_DARWIN) || defined(MOZ_WIDGET_ANDROID) || defined(MOZ_WIDGET_GTK)
|
|
static TimeStamp FromSystemTime(int64_t aSystemTime) {
|
|
static_assert(sizeof(aSystemTime) == sizeof(TimeStampValue),
|
|
"System timestamp should be same units as TimeStampValue");
|
|
return TimeStamp(aSystemTime);
|
|
}
|
|
#endif
|
|
|
|
/**
|
|
* Return true if this is the "null" moment
|
|
*/
|
|
constexpr bool IsNull() const { return mValue == 0; }
|
|
|
|
/**
|
|
* Return true if this is not the "null" moment, may be used in tests, e.g.:
|
|
* |if (timestamp) { ... }|
|
|
*/
|
|
explicit operator bool() const { return mValue != 0; }
|
|
|
|
/**
|
|
* Return a timestamp reflecting the current elapsed system time. This
|
|
* is monotonically increasing (i.e., does not decrease) over the
|
|
* lifetime of this process' XPCOM session.
|
|
*
|
|
* Now() is trying to ensure the best possible precision on each platform,
|
|
* at least one millisecond.
|
|
*
|
|
* NowLoRes() has been introduced to workaround performance problems of
|
|
* QueryPerformanceCounter on the Windows platform. NowLoRes() is giving
|
|
* lower precision, usually 15.6 ms, but with very good performance benefit.
|
|
* Use it for measurements of longer times, like >200ms timeouts.
|
|
*/
|
|
static TimeStamp Now() { return Now(true); }
|
|
static TimeStamp NowLoRes() { return Now(false); }
|
|
|
|
/**
|
|
* Return a timestamp representing the time when the current process was
|
|
* created which will be comparable with other timestamps taken with this
|
|
* class.
|
|
*
|
|
* @returns A timestamp representing the time when the process was created
|
|
*/
|
|
static MFBT_API TimeStamp ProcessCreation();
|
|
|
|
/**
|
|
* Return the very first timestamp that was taken. This can be used instead
|
|
* of TimeStamp::ProcessCreation() by code that might not allow running the
|
|
* complex logic required to compute the real process creation. This will
|
|
* necessarily have been recorded sometimes after TimeStamp::ProcessCreation()
|
|
* or at best should be equal to it.
|
|
*
|
|
* @returns The first tiemstamp that was taken by this process
|
|
*/
|
|
static MFBT_API TimeStamp FirstTimeStamp();
|
|
|
|
/**
|
|
* Records a process restart. After this call ProcessCreation() will return
|
|
* the time when the browser was restarted instead of the actual time when
|
|
* the process was created.
|
|
*/
|
|
static MFBT_API void RecordProcessRestart();
|
|
|
|
#ifdef XP_LINUX
|
|
uint64_t RawClockMonotonicNanosecondsSinceBoot() {
|
|
return static_cast<uint64_t>(mValue);
|
|
}
|
|
#endif
|
|
|
|
#ifdef XP_MACOSX
|
|
uint64_t RawMachAbsoluteTimeValue() { return static_cast<uint64_t>(mValue); }
|
|
#endif
|
|
|
|
#ifdef XP_WIN
|
|
Maybe<uint64_t> RawQueryPerformanceCounterValue() {
|
|
// mQPC is stored in `mt` i.e. QueryPerformanceCounter * 1000
|
|
// so divide out the 1000
|
|
return mValue.mHasQPC ? Some(mValue.mQPC / 1000ULL) : Nothing();
|
|
}
|
|
#endif
|
|
|
|
/**
|
|
* Compute the difference between two timestamps. Both must be non-null.
|
|
*/
|
|
TimeDuration operator-(const TimeStamp& aOther) const {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
MOZ_ASSERT(!aOther.IsNull(), "Cannot compute with aOther null value");
|
|
static_assert(-INT64_MAX > INT64_MIN, "int64_t sanity check");
|
|
int64_t ticks = int64_t(mValue - aOther.mValue);
|
|
// Check for overflow.
|
|
if (mValue > aOther.mValue) {
|
|
if (ticks < 0) {
|
|
ticks = INT64_MAX;
|
|
}
|
|
} else {
|
|
if (ticks > 0) {
|
|
ticks = INT64_MIN;
|
|
}
|
|
}
|
|
return TimeDuration::FromTicks(ticks);
|
|
}
|
|
|
|
TimeStamp operator+(const TimeDuration& aOther) const {
|
|
TimeStamp result = *this;
|
|
result += aOther;
|
|
return result;
|
|
}
|
|
TimeStamp operator-(const TimeDuration& aOther) const {
|
|
TimeStamp result = *this;
|
|
result -= aOther;
|
|
return result;
|
|
}
|
|
TimeStamp& operator+=(const TimeDuration& aOther) {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
TimeStampValue value = mValue + aOther.mValue;
|
|
// Check for underflow.
|
|
// (We don't check for overflow because it's not obvious what the error
|
|
// behavior should be in that case.)
|
|
if (aOther.mValue < 0 && value > mValue) {
|
|
value = 0;
|
|
}
|
|
mValue = value;
|
|
return *this;
|
|
}
|
|
TimeStamp& operator-=(const TimeDuration& aOther) {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
TimeStampValue value = mValue - aOther.mValue;
|
|
// Check for underflow.
|
|
// (We don't check for overflow because it's not obvious what the error
|
|
// behavior should be in that case.)
|
|
if (aOther.mValue > 0 && value > mValue) {
|
|
value = 0;
|
|
}
|
|
mValue = value;
|
|
return *this;
|
|
}
|
|
|
|
constexpr bool operator<(const TimeStamp& aOther) const {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
MOZ_ASSERT(!aOther.IsNull(), "Cannot compute with aOther null value");
|
|
return mValue < aOther.mValue;
|
|
}
|
|
constexpr bool operator<=(const TimeStamp& aOther) const {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
MOZ_ASSERT(!aOther.IsNull(), "Cannot compute with aOther null value");
|
|
return mValue <= aOther.mValue;
|
|
}
|
|
constexpr bool operator>=(const TimeStamp& aOther) const {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
MOZ_ASSERT(!aOther.IsNull(), "Cannot compute with aOther null value");
|
|
return mValue >= aOther.mValue;
|
|
}
|
|
constexpr bool operator>(const TimeStamp& aOther) const {
|
|
MOZ_ASSERT(!IsNull(), "Cannot compute with a null value");
|
|
MOZ_ASSERT(!aOther.IsNull(), "Cannot compute with aOther null value");
|
|
return mValue > aOther.mValue;
|
|
}
|
|
bool operator==(const TimeStamp& aOther) const {
|
|
return IsNull() ? aOther.IsNull()
|
|
: !aOther.IsNull() && mValue == aOther.mValue;
|
|
}
|
|
bool operator!=(const TimeStamp& aOther) const { return !(*this == aOther); }
|
|
|
|
// Comparing TimeStamps for equality should be discouraged. Adding
|
|
// two TimeStamps, or scaling TimeStamps, is nonsense and must never
|
|
// be allowed.
|
|
|
|
static MFBT_API void Startup();
|
|
static MFBT_API void Shutdown();
|
|
|
|
#if defined(DEBUG)
|
|
TimeStampValue GetValue() const { return mValue; }
|
|
#endif
|
|
|
|
private:
|
|
friend struct IPC::ParamTraits<mozilla::TimeStamp>;
|
|
friend struct TimeStampInitialization;
|
|
friend class TimeStampTests;
|
|
|
|
constexpr MOZ_IMPLICIT TimeStamp(TimeStampValue aValue) : mValue(aValue) {}
|
|
|
|
static MFBT_API TimeStamp Now(bool aHighResolution);
|
|
|
|
/**
|
|
* Computes the uptime of the current process in microseconds. The result
|
|
* is platform-dependent and needs to be checked against existing timestamps
|
|
* for consistency.
|
|
*
|
|
* @returns The number of microseconds since the calling process was started
|
|
* or 0 if an error was encountered while computing the uptime
|
|
*/
|
|
static MFBT_API uint64_t ComputeProcessUptime();
|
|
|
|
/**
|
|
* When built with PRIntervalTime, a value of 0 means this instance
|
|
* is "null". Otherwise, the low 32 bits represent a PRIntervalTime,
|
|
* and the high 32 bits represent a counter of the number of
|
|
* rollovers of PRIntervalTime that we've seen. This counter starts
|
|
* at 1 to avoid a real time colliding with the "null" value.
|
|
*
|
|
* PR_INTERVAL_MAX is set at 100,000 ticks per second. So the minimum
|
|
* time to wrap around is about 2^64/100000 seconds, i.e. about
|
|
* 5,849,424 years.
|
|
*
|
|
* When using a system clock, a value is system dependent.
|
|
*/
|
|
TimeStampValue mValue;
|
|
};
|
|
|
|
} // namespace mozilla
|
|
|
|
#endif /* mozilla_TimeStamp_h */
|