xpcom/threads/CPUUsageWatcher.cpp
author Henri Sivonen <hsivonen@hsivonen.fi>
Fri, 06 Jul 2018 10:44:43 +0300
changeset 489140 4ef0f163fdeb9afeddd87b37bfd987298c038542
parent 480058 167d3bf700d3d43b5c5e6802d090cf6cc0056a51
child 508163 6f3709b3878117466168c40affa7bca0b60cf75b
permissions -rw-r--r--
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9

/* -*- 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/. */

#include "mozilla/CPUUsageWatcher.h"

#include "prsystem.h"

#ifdef XP_MACOSX
#include <sys/resource.h>
#include <mach/clock.h>
#include <mach/mach_host.h>
#endif

namespace mozilla {

#ifdef CPU_USAGE_WATCHER_ACTIVE

// Even if the machine only has one processor, tolerate up to 50%
// external CPU usage.
static const float kTolerableExternalCPUUsageFloor = 0.5f;

struct CPUStats {
  // The average CPU usage time, which can be summed across all cores in the
  // system, or averaged between them. Whichever it is, it needs to be in the
  // same units as updateTime.
  uint64_t usageTime;
  // A monotonically increasing value in the same units as usageTime, which can
  // be used to determine the percentage of active vs idle time
  uint64_t updateTime;
};

#ifdef XP_MACOSX

static const uint64_t kMicrosecondsPerSecond = 1000000LL;
static const uint64_t kNanosecondsPerMicrosecond = 1000LL;
static const uint64_t kCPUCheckInterval = kMicrosecondsPerSecond / 2LL;

uint64_t GetMicroseconds(timeval time) {
    return ((uint64_t)time.tv_sec) * kMicrosecondsPerSecond +
           (uint64_t)time.tv_usec;
}

uint64_t GetMicroseconds(mach_timespec_t time) {
    return ((uint64_t)time.tv_sec) * kMicrosecondsPerSecond +
           ((uint64_t)time.tv_nsec) / kNanosecondsPerMicrosecond;
}

Result<CPUStats, CPUUsageWatcherError>
GetProcessCPUStats(int32_t numCPUs) {
  CPUStats result = {};
  rusage usage;
  int32_t rusageResult = getrusage(RUSAGE_SELF, &usage);
  if (rusageResult == -1) {
    return Err(GetProcessTimesError);
  }
  result.usageTime = GetMicroseconds(usage.ru_utime) + GetMicroseconds(usage.ru_stime);

  clock_serv_t realtimeClock;
  kern_return_t errorResult =
      host_get_clock_service(mach_host_self(), REALTIME_CLOCK, &realtimeClock);
  if (errorResult != KERN_SUCCESS) {
    return Err(GetProcessTimesError);
  }
  mach_timespec_t time;
  errorResult = clock_get_time(realtimeClock, &time);
  if (errorResult != KERN_SUCCESS) {
    return Err(GetProcessTimesError);
  }
  result.updateTime = GetMicroseconds(time);

  // getrusage will give us the sum of the values across all
  // of our cores. Divide by the number of CPUs to get an average.
  result.usageTime /= numCPUs;
  return result;
}

Result<CPUStats, CPUUsageWatcherError>
GetGlobalCPUStats() {
  CPUStats result = {};
  host_cpu_load_info_data_t loadInfo;
  mach_msg_type_number_t loadInfoCount = HOST_CPU_LOAD_INFO_COUNT;
  kern_return_t statsResult = host_statistics(mach_host_self(),
                                              HOST_CPU_LOAD_INFO,
                                              (host_info_t)&loadInfo,
                                              &loadInfoCount);
  if (statsResult != KERN_SUCCESS) {
    return Err(HostStatisticsError);
  }

  result.usageTime = loadInfo.cpu_ticks[CPU_STATE_USER] +
                     loadInfo.cpu_ticks[CPU_STATE_NICE] +
                     loadInfo.cpu_ticks[CPU_STATE_SYSTEM];
  result.updateTime = result.usageTime + loadInfo.cpu_ticks[CPU_STATE_IDLE];
  return result;
}

#endif // XP_MACOSX

#ifdef XP_WIN

// A FILETIME represents the number of 100-nanosecond ticks since 1/1/1601 UTC
static const uint64_t kFILETIMETicksPerSecond = 10000000;
static const uint64_t kCPUCheckInterval = kFILETIMETicksPerSecond / 2;

uint64_t
FiletimeToInteger(FILETIME filetime) {
  return ((uint64_t)filetime.dwLowDateTime) |
         (uint64_t)filetime.dwHighDateTime << 32;
}

Result<CPUStats, CPUUsageWatcherError> GetProcessCPUStats(int32_t numCPUs) {
  CPUStats result = {};
  FILETIME creationFiletime;
  FILETIME exitFiletime;
  FILETIME kernelFiletime;
  FILETIME userFiletime;
  bool success = GetProcessTimes(GetCurrentProcess(),
                                 &creationFiletime,
                                 &exitFiletime,
                                 &kernelFiletime,
                                 &userFiletime);
  if (!success) {
    return Err(GetProcessTimesError);
  }

  result.usageTime = FiletimeToInteger(kernelFiletime) +
                     FiletimeToInteger(userFiletime);

  FILETIME nowFiletime;
  GetSystemTimeAsFileTime(&nowFiletime);
  result.updateTime = FiletimeToInteger(nowFiletime);

  result.usageTime /= numCPUs;

  return result;
}

Result<CPUStats, CPUUsageWatcherError>
GetGlobalCPUStats() {
  CPUStats result = {};
  FILETIME idleFiletime;
  FILETIME kernelFiletime;
  FILETIME userFiletime;
  bool success = GetSystemTimes(&idleFiletime,
                                &kernelFiletime,
                                &userFiletime);

  if (!success) {
    return Err(GetSystemTimesError);
  }

  result.usageTime = FiletimeToInteger(kernelFiletime) +
                     FiletimeToInteger(userFiletime);
  result.updateTime = result.usageTime + FiletimeToInteger(idleFiletime);

  return result;
}

#endif // XP_WIN

Result<Ok, CPUUsageWatcherError>
CPUUsageWatcher::Init()
{
  mNumCPUs = PR_GetNumberOfProcessors();
  if (mNumCPUs <= 0) {
    mExternalUsageThreshold = 1.0f;
    return Err(GetNumberOfProcessorsError);
  }
  mExternalUsageThreshold = std::max(1.0f - 1.0f / (float)mNumCPUs,
                                     kTolerableExternalCPUUsageFloor);

  CPUStats processTimes;
  MOZ_TRY_VAR(processTimes, GetProcessCPUStats(mNumCPUs));
  mProcessUpdateTime = processTimes.updateTime;
  mProcessUsageTime = processTimes.usageTime;

  CPUStats globalTimes;
  MOZ_TRY_VAR(globalTimes, GetGlobalCPUStats());
  mGlobalUpdateTime = globalTimes.updateTime;
  mGlobalUsageTime = globalTimes.usageTime;

  mInitialized = true;

  CPUUsageWatcher* self = this;
  NS_DispatchToMainThread(
    NS_NewRunnableFunction("CPUUsageWatcher::Init",
                           [=]() { BackgroundHangMonitor::RegisterAnnotator(*self); }));

  return Ok();
}

void
CPUUsageWatcher::Uninit()
{
  if (mInitialized) {
    BackgroundHangMonitor::UnregisterAnnotator(*this);
  }
  mInitialized = false;
}

Result<Ok, CPUUsageWatcherError>
CPUUsageWatcher::CollectCPUUsage()
{
  if (!mInitialized) {
    return Ok();
  }

  mExternalUsageRatio = 0.0f;

  CPUStats processTimes;
  MOZ_TRY_VAR(processTimes, GetProcessCPUStats(mNumCPUs));
  CPUStats globalTimes;
  MOZ_TRY_VAR(globalTimes, GetGlobalCPUStats());

  uint64_t processUsageDelta = processTimes.usageTime - mProcessUsageTime;
  uint64_t processUpdateDelta = processTimes.updateTime - mProcessUpdateTime;
  float processUsageNormalized = processUsageDelta > 0 ?
                                (float)processUsageDelta / (float)processUpdateDelta :
                                0.0f;

  uint64_t globalUsageDelta = globalTimes.usageTime - mGlobalUsageTime;
  uint64_t globalUpdateDelta = globalTimes.updateTime - mGlobalUpdateTime;
  float globalUsageNormalized = globalUsageDelta > 0 ?
                                (float)globalUsageDelta / (float)globalUpdateDelta :
                                0.0f;

  mProcessUsageTime = processTimes.usageTime;
  mProcessUpdateTime = processTimes.updateTime;
  mGlobalUsageTime = globalTimes.usageTime;
  mGlobalUpdateTime = globalTimes.updateTime;

  mExternalUsageRatio = std::max(0.0f,
                                 globalUsageNormalized - processUsageNormalized);

  return Ok();
}

void
CPUUsageWatcher::AnnotateHang(BackgroundHangAnnotations& aAnnotations) {
  if (!mInitialized) {
    return;
  }

  if (mExternalUsageRatio > mExternalUsageThreshold) {
    aAnnotations.AddAnnotation(NS_LITERAL_STRING("ExternalCPUHigh"), true);
  }
}

#else // !CPU_USAGE_WATCHER_ACTIVE

Result<Ok, CPUUsageWatcherError>
CPUUsageWatcher::Init()
{
  return Ok();
}

void CPUUsageWatcher::Uninit() {}

Result<Ok, CPUUsageWatcherError>
CPUUsageWatcher::CollectCPUUsage()
{
  return Ok();
}

void CPUUsageWatcher::AnnotateHang(BackgroundHangAnnotations& aAnnotations) {}

#endif // CPU_USAGE_WATCHER_ACTIVE

} // namespace mozilla