gecko-dev/xpcom/threads/nsThread.h

285 строки
8.0 KiB
C
Исходник Обычный вид История

/* -*- 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 15:12:37 +04:00
/* 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/. */
1999-04-02 13:20:44 +04:00
#ifndef nsThread_h__
#define nsThread_h__
#include "mozilla/Mutex.h"
#include "nsIIdlePeriod.h"
#include "nsIThreadInternal.h"
#include "nsISupportsPriority.h"
#include "nsEventQueue.h"
#include "nsThreadUtils.h"
#include "nsString.h"
#include "nsTObserverArray.h"
#include "mozilla/Attributes.h"
#include "mozilla/NotNull.h"
#include "nsAutoPtr.h"
#include "mozilla/AlreadyAddRefed.h"
#include "mozilla/UniquePtr.h"
1999-04-02 13:20:44 +04:00
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
2015-08-11 16:10:46 +03:00
namespace mozilla {
class CycleCollectedJSContext;
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
2015-08-11 16:10:46 +03:00
}
using mozilla::NotNull;
// A native thread
class nsThread
: public nsIThreadInternal
, public nsISupportsPriority
1999-04-02 13:20:44 +04:00
{
public:
NS_DECL_THREADSAFE_ISUPPORTS
NS_DECL_NSIEVENTTARGET
NS_DECL_NSITHREAD
NS_DECL_NSITHREADINTERNAL
NS_DECL_NSISUPPORTSPRIORITY
using nsIEventTarget::Dispatch;
1999-04-02 13:20:44 +04:00
enum MainThreadFlag
{
MAIN_THREAD,
NOT_MAIN_THREAD
};
nsThread(MainThreadFlag aMainThread, uint32_t aStackSize);
1999-04-02 13:20:44 +04:00
// Initialize this as a wrapper for a new PRThread, and optionally give it a name.
nsresult Init(const nsACString& aName = NS_LITERAL_CSTRING(""));
// Initialize this as a wrapper for the current PRThread.
nsresult InitCurrentThread();
// The PRThread corresponding to this thread.
PRThread* GetPRThread()
{
return mThread;
}
1999-04-02 13:20:44 +04:00
// If this flag is true, then the nsThread was created using
// nsIThreadManager::NewThread.
bool ShutdownRequired()
{
return mShutdownRequired;
}
// Clear the observer list.
void ClearObservers()
{
mEventObservers.Clear();
}
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
2015-08-11 16:10:46 +03:00
void
SetScriptObserver(mozilla::CycleCollectedJSContext* aScriptObserver);
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
2015-08-11 16:10:46 +03:00
uint32_t
RecursionDepth() const;
void ShutdownComplete(NotNull<struct nsThreadShutdownContext*> aContext);
void WaitForAllAsynchronousShutdowns();
#ifdef MOZ_CRASHREPORTER
enum class ShouldSaveMemoryReport
{
kMaybeReport,
kForceReport
};
static bool SaveMemoryReportNearOOM(ShouldSaveMemoryReport aShouldSave);
#endif
private:
void DoMainThreadSpecificProcessing(bool aReallyWait);
void GetIdleEvent(nsIRunnable** aEvent, mozilla::MutexAutoLock& aProofOfLock);
void GetEvent(bool aWait, nsIRunnable** aEvent,
mozilla::MutexAutoLock& aProofOfLock);
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
2015-08-11 16:10:46 +03:00
protected:
class nsChainedEventQueue;
class nsNestedEventTarget;
friend class nsNestedEventTarget;
friend class nsThreadShutdownEvent;
virtual ~nsThread();
bool ShuttingDown()
{
return mShutdownContext != nullptr;
}
static void ThreadFunc(void* aArg);
// Helper
already_AddRefed<nsIThreadObserver> GetObserver()
{
nsIThreadObserver* obs;
nsThread::GetObserver(&obs);
return already_AddRefed<nsIThreadObserver>(obs);
}
// Wrappers for event queue methods:
nsresult PutEvent(nsIRunnable* aEvent, nsNestedEventTarget* aTarget);
nsresult PutEvent(already_AddRefed<nsIRunnable> aEvent,
nsNestedEventTarget* aTarget);
nsresult DispatchInternal(already_AddRefed<nsIRunnable> aEvent,
uint32_t aFlags, nsNestedEventTarget* aTarget);
struct nsThreadShutdownContext* ShutdownInternal(bool aSync);
// Wrapper for nsEventQueue that supports chaining.
class nsChainedEventQueue
{
public:
explicit nsChainedEventQueue(mozilla::Mutex& aLock)
: mNext(nullptr)
, mEventsAvailable(aLock, "[nsChainedEventQueue.mEventsAvailable]")
, mProcessSecondaryQueueRunnable(false)
{
mNormalQueue =
mozilla::MakeUnique<nsEventQueue>(mEventsAvailable,
nsEventQueue::eSharedCondVarQueue);
// Both queues need to use the same CondVar!
mSecondaryQueue =
mozilla::MakeUnique<nsEventQueue>(mEventsAvailable,
nsEventQueue::eSharedCondVarQueue);
}
bool GetEvent(bool aMayWait, nsIRunnable** aEvent,
mozilla::MutexAutoLock& aProofOfLock);
void PutEvent(nsIRunnable* aEvent, mozilla::MutexAutoLock& aProofOfLock)
{
RefPtr<nsIRunnable> event(aEvent);
PutEvent(event.forget(), aProofOfLock);
}
void PutEvent(already_AddRefed<nsIRunnable> aEvent,
mozilla::MutexAutoLock& aProofOfLock)
{
RefPtr<nsIRunnable> event(aEvent);
nsCOMPtr<nsIRunnablePriority> runnablePrio =
do_QueryInterface(event);
uint32_t prio = nsIRunnablePriority::PRIORITY_NORMAL;
if (runnablePrio) {
runnablePrio->GetPriority(&prio);
}
MOZ_ASSERT(prio == nsIRunnablePriority::PRIORITY_NORMAL ||
prio == nsIRunnablePriority::PRIORITY_HIGH);
if (prio == nsIRunnablePriority::PRIORITY_NORMAL) {
mNormalQueue->PutEvent(event.forget(), aProofOfLock);
} else {
mSecondaryQueue->PutEvent(event.forget(), aProofOfLock);
}
}
bool HasPendingEvent(mozilla::MutexAutoLock& aProofOfLock)
{
return mNormalQueue->HasPendingEvent(aProofOfLock) ||
mSecondaryQueue->HasPendingEvent(aProofOfLock);
}
nsChainedEventQueue* mNext;
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsNestedEventTarget> mEventTarget;
private:
mozilla::CondVar mEventsAvailable;
mozilla::UniquePtr<nsEventQueue> mNormalQueue;
mozilla::UniquePtr<nsEventQueue> mSecondaryQueue;
// Try to process one high priority runnable after each normal
// priority runnable. This gives the processing model HTML spec has for
// 'Update the rendering' in the case only vsync messages are in the
// secondary queue and prevents starving the normal queue.
bool mProcessSecondaryQueueRunnable;
};
class nsNestedEventTarget final : public nsIEventTarget
{
public:
NS_DECL_THREADSAFE_ISUPPORTS
NS_DECL_NSIEVENTTARGET
nsNestedEventTarget(NotNull<nsThread*> aThread,
NotNull<nsChainedEventQueue*> aQueue)
: mThread(aThread)
, mQueue(aQueue)
{
}
NotNull<RefPtr<nsThread>> mThread;
// This is protected by mThread->mLock.
nsChainedEventQueue* mQueue;
private:
~nsNestedEventTarget()
{
}
};
// This lock protects access to mObserver, mEvents, mIdleEvents,
// mIdlePeriod and mEventsAreDoomed. All of those fields are only
// modified on the thread itself (never from another thread). This
// means that we can avoid holding the lock while using mObserver
// and mEvents on the thread itself. When calling PutEvent on
// mEvents, we have to hold the lock to synchronize with
// PopEventQueue.
mozilla::Mutex mLock;
nsCOMPtr<nsIThreadObserver> mObserver;
mozilla::CycleCollectedJSContext* mScriptObserver;
// Only accessed on the target thread.
nsAutoTObserverArray<NotNull<nsCOMPtr<nsIThreadObserver>>, 2> mEventObservers;
NotNull<nsChainedEventQueue*> mEvents; // never null
nsChainedEventQueue mEventsRoot;
// mIdlePeriod keeps track of the current idle period. If at any
// time the main event queue is empty, calling
// mIdlePeriod->GetIdlePeriodHint() will give an estimate of when
// the current idle period will end.
nsCOMPtr<nsIIdlePeriod> mIdlePeriod;
mozilla::CondVar mIdleEventsAvailable;
nsEventQueue mIdleEvents;
int32_t mPriority;
PRThread* mThread;
uint32_t mNestedEventLoopDepth;
uint32_t mStackSize;
// The shutdown context for ourselves.
struct nsThreadShutdownContext* mShutdownContext;
// The shutdown contexts for any other threads we've asked to shut down.
nsTArray<nsAutoPtr<struct nsThreadShutdownContext>> mRequestedShutdownContexts;
bool mShutdownRequired;
// Set to true when events posted to this thread will never run.
bool mEventsAreDoomed;
MainThreadFlag mIsMainThread;
// Set to true if this thread creates a JSRuntime.
bool mCanInvokeJS;
1999-04-02 13:20:44 +04:00
};
#if defined(XP_UNIX) && !defined(ANDROID) && !defined(DEBUG) && HAVE_UALARM \
&& defined(_GNU_SOURCE)
# define MOZ_CANARY
extern int sCanaryOutputFD;
#endif
#endif // nsThread_h__