gecko/xpcom/threads/nsThreadPool.cpp

445 lines
11 KiB
C++
Raw Normal View History

/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
2012-05-21 04:12:37 -07:00
/* 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 "nsIClassInfoImpl.h"
#include "nsThreadPool.h"
#include "nsThreadManager.h"
#include "nsThread.h"
#include "nsMemory.h"
#include "nsAutoPtr.h"
#include "prinrval.h"
#include "mozilla/Logging.h"
#include "nsThreadSyncDispatch.h"
using namespace mozilla;
static PRLogModuleInfo*
GetThreadPoolLog()
{
static PRLogModuleInfo* sLog;
if (!sLog) {
sLog = PR_NewLogModule("nsThreadPool");
}
return sLog;
}
#ifdef LOG
#undef LOG
#endif
#define LOG(args) MOZ_LOG(GetThreadPoolLog(), mozilla::LogLevel::Debug, args)
// DESIGN:
// o Allocate anonymous threads.
// o Use nsThreadPool::Run as the main routine for each thread.
// o Each thread waits on the event queue's monitor, checking for
// pending events and rescheduling itself as an idle thread.
#define DEFAULT_THREAD_LIMIT 4
#define DEFAULT_IDLE_THREAD_LIMIT 1
#define DEFAULT_IDLE_THREAD_TIMEOUT PR_SecondsToInterval(60)
NS_IMPL_ADDREF(nsThreadPool)
NS_IMPL_RELEASE(nsThreadPool)
NS_IMPL_CLASSINFO(nsThreadPool, nullptr, nsIClassInfo::THREADSAFE,
NS_THREADPOOL_CID)
NS_IMPL_QUERY_INTERFACE_CI(nsThreadPool, nsIThreadPool, nsIEventTarget,
nsIRunnable)
NS_IMPL_CI_INTERFACE_GETTER(nsThreadPool, nsIThreadPool, nsIEventTarget)
nsThreadPool::nsThreadPool()
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
: mMutex("[nsThreadPool.mMutex]")
, mEvents(mMutex)
, mThreadLimit(DEFAULT_THREAD_LIMIT)
, mIdleThreadLimit(DEFAULT_IDLE_THREAD_LIMIT)
, mIdleThreadTimeout(DEFAULT_IDLE_THREAD_TIMEOUT)
, mIdleCount(0)
, mStackSize(nsIThreadManager::DEFAULT_STACK_SIZE)
, mShutdown(false)
{
LOG(("THRD-P(%p) constructor!!!\n", this));
}
nsThreadPool::~nsThreadPool()
{
// Threads keep a reference to the nsThreadPool until they return from Run()
// after removing themselves from mThreads.
MOZ_ASSERT(mThreads.IsEmpty());
}
nsresult
nsThreadPool::PutEvent(nsIRunnable* aEvent)
{
nsCOMPtr<nsIRunnable> event(aEvent);
return PutEvent(event.forget());
}
nsresult
nsThreadPool::PutEvent(already_AddRefed<nsIRunnable>&& aEvent)
{
// Avoid spawning a new thread while holding the event queue lock...
bool spawnThread = false;
uint32_t stackSize = 0;
{
MutexAutoLock lock(mMutex);
if (NS_WARN_IF(mShutdown)) {
return NS_ERROR_NOT_AVAILABLE;
}
LOG(("THRD-P(%p) put [%d %d %d]\n", this, mIdleCount, mThreads.Count(),
mThreadLimit));
MOZ_ASSERT(mIdleCount <= (uint32_t)mThreads.Count(), "oops");
// Make sure we have a thread to service this event.
if (mThreads.Count() < (int32_t)mThreadLimit &&
// Spawn a new thread if we don't have enough idle threads to serve
// pending events immediately.
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.Count(lock) >= mIdleCount) {
spawnThread = true;
}
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.PutEvent(Move(aEvent), lock);
stackSize = mStackSize;
}
LOG(("THRD-P(%p) put [spawn=%d]\n", this, spawnThread));
if (!spawnThread) {
return NS_OK;
}
nsCOMPtr<nsIThread> thread;
nsThreadManager::get()->NewThread(0,
stackSize,
getter_AddRefs(thread));
if (NS_WARN_IF(!thread)) {
return NS_ERROR_UNEXPECTED;
}
bool killThread = false;
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
if (mThreads.Count() < (int32_t)mThreadLimit) {
mThreads.AppendObject(thread);
} else {
killThread = true; // okay, we don't need this thread anymore
}
}
LOG(("THRD-P(%p) put [%p kill=%d]\n", this, thread.get(), killThread));
if (killThread) {
// We never dispatched any events to the thread, so we can shut it down
// asynchronously without worrying about anything.
MOZ_ALWAYS_TRUE(NS_SUCCEEDED(thread->AsyncShutdown()));
} else {
thread->Dispatch(this, NS_DISPATCH_NORMAL);
}
return NS_OK;
}
void
nsThreadPool::ShutdownThread(nsIThread* aThread)
{
LOG(("THRD-P(%p) shutdown async [%p]\n", this, aThread));
// This method is responsible for calling Shutdown on |aThread|. This must be
// done from some other thread, so we use the main thread of the application.
MOZ_ASSERT(!NS_IsMainThread(), "wrong thread");
nsCOMPtr<nsIRunnable> r = NS_NewRunnableMethod(aThread, &nsIThread::Shutdown);
NS_DispatchToMainThread(r);
}
NS_IMETHODIMP
nsThreadPool::Run()
{
mThreadNaming.SetThreadPoolName(mName);
LOG(("THRD-P(%p) enter %s\n", this, mName.BeginReading()));
nsCOMPtr<nsIThread> current;
nsThreadManager::get()->GetCurrentThread(getter_AddRefs(current));
bool shutdownThreadOnExit = false;
bool exitThread = false;
bool wasIdle = false;
PRIntervalTime idleSince;
nsCOMPtr<nsIThreadPoolListener> listener;
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
listener = mListener;
}
if (listener) {
listener->OnThreadCreated();
}
do {
nsCOMPtr<nsIRunnable> event;
{
MutexAutoLock lock(mMutex);
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
if (!mEvents.GetPendingEvent(getter_AddRefs(event), lock)) {
PRIntervalTime now = PR_IntervalNow();
PRIntervalTime timeout = PR_MillisecondsToInterval(mIdleThreadTimeout);
// If we are shutting down, then don't keep any idle threads
if (mShutdown) {
exitThread = true;
} else {
if (wasIdle) {
// if too many idle threads or idle for too long, then bail.
if (mIdleCount > mIdleThreadLimit || (now - idleSince) >= timeout) {
exitThread = true;
}
} else {
// if would be too many idle threads...
if (mIdleCount == mIdleThreadLimit) {
exitThread = true;
} else {
++mIdleCount;
idleSince = now;
wasIdle = true;
}
}
}
if (exitThread) {
if (wasIdle) {
--mIdleCount;
}
shutdownThreadOnExit = mThreads.RemoveObject(current);
} else {
PRIntervalTime delta = timeout - (now - idleSince);
LOG(("THRD-P(%p) %s waiting [%d]\n", this, mName.BeginReading(), delta));
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.Wait(delta);
LOG(("THRD-P(%p) done waiting\n", this));
}
} else if (wasIdle) {
wasIdle = false;
--mIdleCount;
}
}
if (event) {
LOG(("THRD-P(%p) %s running [%p]\n", this, mName.BeginReading(), event.get()));
event->Run();
}
} while (!exitThread);
if (listener) {
listener->OnThreadShuttingDown();
}
if (shutdownThreadOnExit) {
ShutdownThread(current);
}
LOG(("THRD-P(%p) leave\n", this));
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::DispatchFromScript(nsIRunnable* aEvent, uint32_t aFlags)
{
nsCOMPtr<nsIRunnable> event(aEvent);
return Dispatch(event.forget(), aFlags);
}
NS_IMETHODIMP
nsThreadPool::Dispatch(already_AddRefed<nsIRunnable>&& aEvent, uint32_t aFlags)
{
LOG(("THRD-P(%p) dispatch [%p %x]\n", this, /* XXX aEvent*/ nullptr, aFlags));
if (NS_WARN_IF(mShutdown)) {
return NS_ERROR_NOT_AVAILABLE;
}
if (aFlags & DISPATCH_SYNC) {
nsCOMPtr<nsIThread> thread;
nsThreadManager::get()->GetCurrentThread(getter_AddRefs(thread));
if (NS_WARN_IF(!thread)) {
return NS_ERROR_NOT_AVAILABLE;
}
nsRefPtr<nsThreadSyncDispatch> wrapper =
new nsThreadSyncDispatch(thread, Move(aEvent));
PutEvent(wrapper);
while (wrapper->IsPending()) {
NS_ProcessNextEvent(thread);
}
} else {
NS_ASSERTION(aFlags == NS_DISPATCH_NORMAL, "unexpected dispatch flags");
PutEvent(Move(aEvent));
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::IsOnCurrentThread(bool* aResult)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
if (NS_WARN_IF(mShutdown)) {
return NS_ERROR_NOT_AVAILABLE;
}
nsIThread* thread = NS_GetCurrentThread();
for (uint32_t i = 0; i < static_cast<uint32_t>(mThreads.Count()); ++i) {
if (mThreads[i] == thread) {
*aResult = true;
return NS_OK;
}
}
*aResult = false;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::Shutdown()
{
nsCOMArray<nsIThread> threads;
nsCOMPtr<nsIThreadPoolListener> listener;
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
mShutdown = true;
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.NotifyAll();
threads.AppendObjects(mThreads);
mThreads.Clear();
// Swap in a null listener so that we release the listener at the end of
// this method. The listener will be kept alive as long as the other threads
// that were created when it was set.
mListener.swap(listener);
}
// It's important that we shutdown the threads while outside the event queue
// monitor. Otherwise, we could end up dead-locking.
for (int32_t i = 0; i < threads.Count(); ++i) {
threads[i]->Shutdown();
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::GetThreadLimit(uint32_t* aValue)
{
*aValue = mThreadLimit;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetThreadLimit(uint32_t aValue)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
LOG(("THRD-P(%p) thread limit [%u]\n", this, aValue));
mThreadLimit = aValue;
if (mIdleThreadLimit > mThreadLimit) {
mIdleThreadLimit = mThreadLimit;
}
if (static_cast<uint32_t>(mThreads.Count()) > mThreadLimit) {
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.NotifyAll(); // wake up threads so they observe this change
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::GetIdleThreadLimit(uint32_t* aValue)
{
*aValue = mIdleThreadLimit;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetIdleThreadLimit(uint32_t aValue)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
LOG(("THRD-P(%p) idle thread limit [%u]\n", this, aValue));
mIdleThreadLimit = aValue;
if (mIdleThreadLimit > mThreadLimit) {
mIdleThreadLimit = mThreadLimit;
}
// Do we need to kill some idle threads?
if (mIdleCount > mIdleThreadLimit) {
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.NotifyAll(); // wake up threads so they observe this change
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::GetIdleThreadTimeout(uint32_t* aValue)
{
*aValue = mIdleThreadTimeout;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetIdleThreadTimeout(uint32_t aValue)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
uint32_t oldTimeout = mIdleThreadTimeout;
mIdleThreadTimeout = aValue;
// Do we need to notify any idle threads that their sleep time has shortened?
if (mIdleThreadTimeout < oldTimeout && mIdleCount > 0) {
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
mEvents.NotifyAll(); // wake up threads so they observe this change
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::GetThreadStackSize(uint32_t* aValue)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
*aValue = mStackSize;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetThreadStackSize(uint32_t aValue)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
mStackSize = aValue;
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::GetListener(nsIThreadPoolListener** aListener)
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
NS_IF_ADDREF(*aListener = mListener);
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetListener(nsIThreadPoolListener* aListener)
{
nsCOMPtr<nsIThreadPoolListener> swappedListener(aListener);
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
mListener.swap(swappedListener);
}
return NS_OK;
}
NS_IMETHODIMP
nsThreadPool::SetName(const nsACString& aName)
{
{
Bug 1202497 - part 7 - make nsEventQueue use external locking; r=gerald We want to ensure that nsThread's use of nsEventQueue uses locking done in nsThread instead of nsEventQueue, for efficiency's sake: we only need to lock once in nsThread, rather than the current situation of locking in nsThread and additionally in nsEventQueue. With the current structure of nsEventQueue, that would mean that nsThread should be using a Monitor internally, rather than a Mutex. Which would be well and good, except that DOM workers use nsThread's mutex to protect their own, internal CondVar. Switching nsThread to use a Monitor would mean that either: - DOM workers drop their internal CondVar in favor of nsThread's Monitor-owned CondVar. This change seems unlikely to work out well, because now the Monitor-owned CondVar is performing double duty: tracking availability of events in nsThread's event queue and additionally whatever DOM workers were using a CondVar for. Having a single CondVar track two things in such a fashion is for Experts Only. - DOM workers grow their own Mutex to protect their own CondVar. Adding a mutex like this would change locking in subtle ways and seems unlikely to lead to success. Using a Monitor in nsThread is therefore untenable, and we would like to retain the current Mutex that lives in nsThread. Therefore, we need to have nsEventQueue manage its own condition variable and push the required (Mutex) locking to the client of nsEventQueue. This scheme also seems more fitting: external clients merely need synchronized access to the event queue; the details of managing notifications about events in the event queue should be left up to the event queue itself. Doing so also forces us to merge nsEventQueueBase and nsEventQueue: there's no way to have nsEventQueueBase require an externally-defined Mutex and then have nsEventQueue subclass nsEventQueueBase and provide its own Mutex to the superclass. C++ initialization rules (and the way things like CondVar are constructed) simply forbid it. But that's OK, because we want a world where nsEventQueue is externally locked anyway, so there's no reason to have separate classes here. One casualty of this work is removing ChaosMode support from nsEventQueue. nsEventQueue had support to delay placing events into the queue, theoretically giving other threads the chance to put events there first. Unfortunately, since the thread would have been holding a lock (as is evident from the MutexAutoLock& parameter required), sleeping in PutEvent accomplishes nothing but delaying the thread from getting useful work done. We should support this, but it's complicated to figure out how to reasonably support this right now. A wrinkle in this overall pleasant refactoring is that nsThreadPool's threads wait for limited amounts of time for new events to be placed in the event queue, so that they can shut themselves down if no new events are appearing. Setting limits on the number of threads also needs to be able to wake up all threads, so threads can shut themselves down if necessary. Unfortunately, with the transition to nsEventQueue managing its own condition variable, there's no way for nsThreadPool to perform these functions, since there's no Monitor to wait on. Therefore, we add a private API for accessing the condition variable and performing the tasks nsThreadPool needs. Prior to all the previous patches, placing items in an nsThread's event queue required three lock/unlock pairs: one for nsThread's Mutex, one to enter nsEventQueue's ReentrantMonitor, and one to exit nsEventQueue's ReentrantMonitor. The upshot of all this work is that we now only require one lock/unlock pair in nsThread itself, as things should be.
2015-09-20 02:13:09 -07:00
MutexAutoLock lock(mMutex);
if (mThreads.Count()) {
return NS_ERROR_NOT_AVAILABLE;
}
}
mName = aName;
return NS_OK;
}