Android Framework中Thread类
Thread类是Android为线程操作而做的一个封装。代码在Thread.cpp中,其中还封装了一些与线程同步相关的类。
Thread类
Thread类的构造函数中的有一个canCallJava
Thread.cpp
/system/core/libutils/Threads.cpp
/* * Copyright (C) 2007 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ // #define LOG_NDEBUG 0 #define LOG_TAG "libutils.threads" #include <assert.h> #include <errno.h> #include <memory.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> #if !defined(_WIN32) # include <pthread.h> # include <sched.h> # include <sys/resource.h> #else # include <windows.h> # include <stdint.h> # include <process.h> # define HAVE_CREATETHREAD // Cygwin, vs. HAVE__BEGINTHREADEX for MinGW #endif #if defined(__linux__) #include <sys/prctl.h> #endif #include <utils/threads.h> #include <utils/Log.h> #include <cutils/sched_policy.h> #ifdef HAVE_ANDROID_OS # define __android_unused #else # define __android_unused __attribute__((__unused__)) #endif /* * =========================================================================== * Thread wrappers * =========================================================================== */ using namespace android; // ---------------------------------------------------------------------------- #if !defined(_WIN32) // ---------------------------------------------------------------------------- /* * Create and run a new thread. * * We create it "detached", so it cleans up after itself. */ typedef void* (*android_pthread_entry)(void*); struct thread_data_t { thread_func_t entryFunction; void* userData; int priority; char * threadName; // we use this trampoline when we need to set the priority with // nice/setpriority, and name with prctl. static int trampoline(const thread_data_t* t) { thread_func_t f = t->entryFunction; void* u = t->userData; int prio = t->priority; char * name = t->threadName; delete t; setpriority(PRIO_PROCESS, 0, prio); if (prio >= ANDROID_PRIORITY_BACKGROUND) { set_sched_policy(0, SP_BACKGROUND); } else { set_sched_policy(0, SP_FOREGROUND); } if (name) { androidSetThreadName(name); free(name); } return f(u); } }; void androidSetThreadName(const char* name) { #if defined(__linux__) // Mac OS doesn't have this, and we build libutil for the host too int hasAt = 0; int hasDot = 0; const char *s = name; while (*s) { if (*s == '.') hasDot = 1; else if (*s == '@') hasAt = 1; s++; } int len = s - name; if (len < 15 || hasAt || !hasDot) { s = name; } else { s = name + len - 15; } prctl(PR_SET_NAME, (unsigned long) s, 0, 0, 0); #endif } int androidCreateRawThreadEtc(android_thread_func_t entryFunction, void *userData, const char* threadName __android_unused, int32_t threadPriority, size_t threadStackSize, android_thread_id_t *threadId) { pthread_attr_t attr; pthread_attr_init(&attr); pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); #ifdef HAVE_ANDROID_OS /* valgrind is rejecting RT-priority create reqs */ if (threadPriority != PRIORITY_DEFAULT || threadName != NULL) { // Now that the pthread_t has a method to find the associated // android_thread_id_t (pid) from pthread_t, it would be possible to avoid // this trampoline in some cases as the parent could set the properties // for the child. However, there would be a race condition because the // child becomes ready immediately, and it doesn't work for the name. // prctl(PR_SET_NAME) only works for self; prctl(PR_SET_THREAD_NAME) was // proposed but not yet accepted. thread_data_t* t = new thread_data_t; t->priority = threadPriority; t->threadName = threadName ? strdup(threadName) : NULL; t->entryFunction = entryFunction; t->userData = userData; entryFunction = (android_thread_func_t)&thread_data_t::trampoline; userData = t; } #endif if (threadStackSize) { pthread_attr_setstacksize(&attr, threadStackSize); } errno = 0; pthread_t thread; int result = pthread_create(&thread, &attr, (android_pthread_entry)entryFunction, userData); pthread_attr_destroy(&attr); if (result != 0) { ALOGE("androidCreateRawThreadEtc failed (entry=%p, res=%d, errno=%d)\n" "(android threadPriority=%d)", entryFunction, result, errno, threadPriority); return 0; } // Note that *threadID is directly available to the parent only, as it is // assigned after the child starts. Use memory barrier / lock if the child // or other threads also need access. if (threadId != NULL) { *threadId = (android_thread_id_t)thread; // XXX: this is not portable } return 1; } #ifdef HAVE_ANDROID_OS static pthread_t android_thread_id_t_to_pthread(android_thread_id_t thread) { return (pthread_t) thread; } #endif android_thread_id_t androidGetThreadId() { return (android_thread_id_t)pthread_self(); } // ---------------------------------------------------------------------------- #else // !defined(_WIN32) // ---------------------------------------------------------------------------- /* * Trampoline to make us __stdcall-compliant. * * We're expected to delete "vDetails" when we're done. */ struct threadDetails { int (*func)(void*); void* arg; }; static __stdcall unsigned int threadIntermediary(void* vDetails) { struct threadDetails* pDetails = (struct threadDetails*) vDetails; int result; result = (*(pDetails->func))(pDetails->arg); delete pDetails; ALOG(LOG_VERBOSE, "thread", "thread exiting\n"); return (unsigned int) result; } /* * Create and run a new thread. */ static bool doCreateThread(android_thread_func_t fn, void* arg, android_thread_id_t *id) { HANDLE hThread; struct threadDetails* pDetails = new threadDetails; // must be on heap unsigned int thrdaddr; pDetails->func = fn; pDetails->arg = arg; #if defined(HAVE__BEGINTHREADEX) hThread = (HANDLE) _beginthreadex(NULL, 0, threadIntermediary, pDetails, 0, &thrdaddr); if (hThread == 0) #elif defined(HAVE_CREATETHREAD) hThread = CreateThread(NULL, 0, (LPTHREAD_START_ROUTINE) threadIntermediary, (void*) pDetails, 0, (DWORD*) &thrdaddr); if (hThread == NULL) #endif { ALOG(LOG_WARN, "thread", "WARNING: thread create failed\n"); return false; } #if defined(HAVE_CREATETHREAD) /* close the management handle */ CloseHandle(hThread); #endif if (id != NULL) { *id = (android_thread_id_t)thrdaddr; } return true; } int androidCreateRawThreadEtc(android_thread_func_t fn, void *userData, const char* /*threadName*/, int32_t /*threadPriority*/, size_t /*threadStackSize*/, android_thread_id_t *threadId) { return doCreateThread( fn, userData, threadId); } android_thread_id_t androidGetThreadId() { return (android_thread_id_t)GetCurrentThreadId(); } // ---------------------------------------------------------------------------- #endif // !defined(_WIN32) // ---------------------------------------------------------------------------- int androidCreateThread(android_thread_func_t fn, void* arg) { return createThreadEtc(fn, arg); } int androidCreateThreadGetID(android_thread_func_t fn, void *arg, android_thread_id_t *id) { return createThreadEtc(fn, arg, "android:unnamed_thread", PRIORITY_DEFAULT, 0, id); } static android_create_thread_fn gCreateThreadFn = androidCreateRawThreadEtc; int androidCreateThreadEtc(android_thread_func_t entryFunction, void *userData, const char* threadName, int32_t threadPriority, size_t threadStackSize, android_thread_id_t *threadId) { return gCreateThreadFn(entryFunction, userData, threadName, threadPriority, threadStackSize, threadId); } void androidSetCreateThreadFunc(android_create_thread_fn func) { gCreateThreadFn = func; } #ifdef HAVE_ANDROID_OS int androidSetThreadPriority(pid_t tid, int pri) { int rc = 0; #if !defined(_WIN32) int lasterr = 0; if (pri >= ANDROID_PRIORITY_BACKGROUND) { rc = set_sched_policy(tid, SP_BACKGROUND); } else if (getpriority(PRIO_PROCESS, tid) >= ANDROID_PRIORITY_BACKGROUND) { rc = set_sched_policy(tid, SP_FOREGROUND); } if (rc) { lasterr = errno; } if (setpriority(PRIO_PROCESS, tid, pri) < 0) { rc = INVALID_OPERATION; } else { errno = lasterr; } #endif return rc; } int androidGetThreadPriority(pid_t tid) { #if !defined(_WIN32) return getpriority(PRIO_PROCESS, tid); #else return ANDROID_PRIORITY_NORMAL; #endif } #endif namespace android { /* * =========================================================================== * Mutex class * =========================================================================== */ #if !defined(_WIN32) // implemented as inlines in threads.h #else Mutex::Mutex() { HANDLE hMutex; assert(sizeof(hMutex) == sizeof(mState)); hMutex = CreateMutex(NULL, FALSE, NULL); mState = (void*) hMutex; } Mutex::Mutex(const char* name) { // XXX: name not used for now HANDLE hMutex; assert(sizeof(hMutex) == sizeof(mState)); hMutex = CreateMutex(NULL, FALSE, NULL); mState = (void*) hMutex; } Mutex::Mutex(int type, const char* name) { // XXX: type and name not used for now HANDLE hMutex; assert(sizeof(hMutex) == sizeof(mState)); hMutex = CreateMutex(NULL, FALSE, NULL); mState = (void*) hMutex; } Mutex::~Mutex() { CloseHandle((HANDLE) mState); } status_t Mutex::lock() { DWORD dwWaitResult; dwWaitResult = WaitForSingleObject((HANDLE) mState, INFINITE); return dwWaitResult != WAIT_OBJECT_0 ? -1 : NO_ERROR; } void Mutex::unlock() { if (!ReleaseMutex((HANDLE) mState)) ALOG(LOG_WARN, "thread", "WARNING: bad result from unlocking mutex\n"); } status_t Mutex::tryLock() { DWORD dwWaitResult; dwWaitResult = WaitForSingleObject((HANDLE) mState, 0); if (dwWaitResult != WAIT_OBJECT_0 && dwWaitResult != WAIT_TIMEOUT) ALOG(LOG_WARN, "thread", "WARNING: bad result from try-locking mutex\n"); return (dwWaitResult == WAIT_OBJECT_0) ? 0 : -1; } #endif // !defined(_WIN32) /* * =========================================================================== * Condition class * =========================================================================== */ #if !defined(_WIN32) // implemented as inlines in threads.h #else /* * Windows doesn't have a condition variable solution. It's possible * to create one, but it's easy to get it wrong. For a discussion, and * the origin of this implementation, see: * * http://www.cs.wustl.edu/~schmidt/win32-cv-1.html * * The implementation shown on the page does NOT follow POSIX semantics. * As an optimization they require acquiring the external mutex before * calling signal() and broadcast(), whereas POSIX only requires grabbing * it before calling wait(). The implementation here has been un-optimized * to have the correct behavior. */ typedef struct WinCondition { // Number of waiting threads. int waitersCount; // Serialize access to waitersCount. CRITICAL_SECTION waitersCountLock; // Semaphore used to queue up threads waiting for the condition to // become signaled. HANDLE sema; // An auto-reset event used by the broadcast/signal thread to wait // for all the waiting thread(s) to wake up and be released from // the semaphore. HANDLE waitersDone; // This mutex wouldn't be necessary if we required that the caller // lock the external mutex before calling signal() and broadcast(). // I'm trying to mimic pthread semantics though. HANDLE internalMutex; // Keeps track of whether we were broadcasting or signaling. This // allows us to optimize the code if we're just signaling. bool wasBroadcast; status_t wait(WinCondition* condState, HANDLE hMutex, nsecs_t* abstime) { // Increment the wait count, avoiding race conditions. EnterCriticalSection(&condState->waitersCountLock); condState->waitersCount++; //printf("+++ wait: incr waitersCount to %d (tid=%ld)\n", // condState->waitersCount, getThreadId()); LeaveCriticalSection(&condState->waitersCountLock); DWORD timeout = INFINITE; if (abstime) { nsecs_t reltime = *abstime - systemTime(); if (reltime < 0) reltime = 0; timeout = reltime/1000000; } // Atomically release the external mutex and wait on the semaphore. DWORD res = SignalObjectAndWait(hMutex, condState->sema, timeout, FALSE); //printf("+++ wait: awake (tid=%ld)\n", getThreadId()); // Reacquire lock to avoid race conditions. EnterCriticalSection(&condState->waitersCountLock); // No longer waiting. condState->waitersCount--; // Check to see if we're the last waiter after a broadcast. bool lastWaiter = (condState->wasBroadcast && condState->waitersCount == 0); //printf("+++ wait: lastWaiter=%d (wasBc=%d wc=%d)\n", // lastWaiter, condState->wasBroadcast, condState->waitersCount); LeaveCriticalSection(&condState->waitersCountLock); // If we're the last waiter thread during this particular broadcast // then signal broadcast() that we're all awake. It'll drop the // internal mutex. if (lastWaiter) { // Atomically signal the "waitersDone" event and wait until we // can acquire the internal mutex. We want to do this in one step // because it ensures that everybody is in the mutex FIFO before // any thread has a chance to run. Without it, another thread // could wake up, do work, and hop back in ahead of us. SignalObjectAndWait(condState->waitersDone, condState->internalMutex, INFINITE, FALSE); } else { // Grab the internal mutex. WaitForSingleObject(condState->internalMutex, INFINITE); } // Release the internal and grab the external. ReleaseMutex(condState->internalMutex); WaitForSingleObject(hMutex, INFINITE); return res == WAIT_OBJECT_0 ? NO_ERROR : -1; } } WinCondition; /* * Constructor. Set up the WinCondition stuff. */ Condition::Condition() { WinCondition* condState = new WinCondition; condState->waitersCount = 0; condState->wasBroadcast = false; // semaphore: no security, initial value of 0 condState->sema = CreateSemaphore(NULL, 0, 0x7fffffff, NULL); InitializeCriticalSection(&condState->waitersCountLock); // auto-reset event, not signaled initially condState->waitersDone = CreateEvent(NULL, FALSE, FALSE, NULL); // used so we don't have to lock external mutex on signal/broadcast condState->internalMutex = CreateMutex(NULL, FALSE, NULL); mState = condState; } /* * Destructor. Free Windows resources as well as our allocated storage. */ Condition::~Condition() { WinCondition* condState = (WinCondition*) mState; if (condState != NULL) { CloseHandle(condState->sema); CloseHandle(condState->waitersDone); delete condState; } } status_t Condition::wait(Mutex& mutex) { WinCondition* condState = (WinCondition*) mState; HANDLE hMutex = (HANDLE) mutex.mState; return ((WinCondition*)mState)->wait(condState, hMutex, NULL); } status_t Condition::waitRelative(Mutex& mutex, nsecs_t reltime) { WinCondition* condState = (WinCondition*) mState; HANDLE hMutex = (HANDLE) mutex.mState; nsecs_t absTime = systemTime()+reltime; return ((WinCondition*)mState)->wait(condState, hMutex, &absTime); } /* * Signal the condition variable, allowing one thread to continue. */ void Condition::signal() { WinCondition* condState = (WinCondition*) mState; // Lock the internal mutex. This ensures that we don't clash with // broadcast(). WaitForSingleObject(condState->internalMutex, INFINITE); EnterCriticalSection(&condState->waitersCountLock); bool haveWaiters = (condState->waitersCount > 0); LeaveCriticalSection(&condState->waitersCountLock); // If no waiters, then this is a no-op. Otherwise, knock the semaphore // down a notch. if (haveWaiters) ReleaseSemaphore(condState->sema, 1, 0); // Release internal mutex. ReleaseMutex(condState->internalMutex); } /* * Signal the condition variable, allowing all threads to continue. * * First we have to wake up all threads waiting on the semaphore, then * we wait until all of the threads have actually been woken before * releasing the internal mutex. This ensures that all threads are woken. */ void Condition::broadcast() { WinCondition* condState = (WinCondition*) mState; // Lock the internal mutex. This keeps the guys we're waking up // from getting too far. WaitForSingleObject(condState->internalMutex, INFINITE); EnterCriticalSection(&condState->waitersCountLock); bool haveWaiters = false; if (condState->waitersCount > 0) { haveWaiters = true; condState->wasBroadcast = true; } if (haveWaiters) { // Wake up all the waiters. ReleaseSemaphore(condState->sema, condState->waitersCount, 0); LeaveCriticalSection(&condState->waitersCountLock); // Wait for all awakened threads to acquire the counting semaphore. // The last guy who was waiting sets this. WaitForSingleObject(condState->waitersDone, INFINITE); // Reset wasBroadcast. (No crit section needed because nobody // else can wake up to poke at it.) condState->wasBroadcast = 0; } else { // nothing to do LeaveCriticalSection(&condState->waitersCountLock); } // Release internal mutex. ReleaseMutex(condState->internalMutex); } #endif // !defined(_WIN32) // ---------------------------------------------------------------------------- /* * This is our thread object! */ Thread::Thread(bool canCallJava) : mCanCallJava(canCallJava), mThread(thread_id_t(-1)), mLock("Thread::mLock"), mStatus(NO_ERROR), mExitPending(false), mRunning(false) #ifdef HAVE_ANDROID_OS , mTid(-1) #endif { } Thread::~Thread() { } status_t Thread::readyToRun() { return NO_ERROR; } status_t Thread::run(const char* name, int32_t priority, size_t stack) { Mutex::Autolock _l(mLock); if (mRunning) { // thread already started return INVALID_OPERATION; } // reset status and exitPending to their default value, so we can // try again after an error happened (either below, or in readyToRun()) mStatus = NO_ERROR; mExitPending = false; mThread = thread_id_t(-1); // hold a strong reference on ourself mHoldSelf = this; mRunning = true; bool res; if (mCanCallJava) { res = createThreadEtc(_threadLoop, this, name, priority, stack, &mThread); } else { res = androidCreateRawThreadEtc(_threadLoop, this, name, priority, stack, &mThread); } if (res == false) { mStatus = UNKNOWN_ERROR; // something happened! mRunning = false; mThread = thread_id_t(-1); mHoldSelf.clear(); // "this" may have gone away after this. return UNKNOWN_ERROR; } // Do not refer to mStatus here: The thread is already running (may, in fact // already have exited with a valid mStatus result). The NO_ERROR indication // here merely indicates successfully starting the thread and does not // imply successful termination/execution. return NO_ERROR; // Exiting scope of mLock is a memory barrier and allows new thread to run } int Thread::_threadLoop(void* user) { Thread* const self = static_cast<Thread*>(user); sp<Thread> strong(self->mHoldSelf); wp<Thread> weak(strong); self->mHoldSelf.clear(); #ifdef HAVE_ANDROID_OS // this is very useful for debugging with gdb self->mTid = gettid(); #endif bool first = true; do { bool result; if (first) { first = false; self->mStatus = self->readyToRun(); result = (self->mStatus == NO_ERROR); if (result && !self->exitPending()) { // Binder threads (and maybe others) rely on threadLoop // running at least once after a successful ::readyToRun() // (unless, of course, the thread has already been asked to exit // at that point). // This is because threads are essentially used like this: // (new ThreadSubclass())->run(); // The caller therefore does not retain a strong reference to // the thread and the thread would simply disappear after the // successful ::readyToRun() call instead of entering the // threadLoop at least once. result = self->threadLoop(); } } else { result = self->threadLoop(); } // establish a scope for mLock { Mutex::Autolock _l(self->mLock); if (result == false || self->mExitPending) { self->mExitPending = true; self->mRunning = false; // clear thread ID so that requestExitAndWait() does not exit if // called by a new thread using the same thread ID as this one. self->mThread = thread_id_t(-1); // note that interested observers blocked in requestExitAndWait are // awoken by broadcast, but blocked on mLock until break exits scope self->mThreadExitedCondition.broadcast(); break; } } // Release our strong reference, to let a chance to the thread // to die a peaceful death. strong.clear(); // And immediately, re-acquire a strong reference for the next loop strong = weak.promote(); } while(strong != 0); return 0; } void Thread::requestExit() { Mutex::Autolock _l(mLock); mExitPending = true; } status_t Thread::requestExitAndWait() { Mutex::Autolock _l(mLock); if (mThread == getThreadId()) { ALOGW( "Thread (this=%p): don't call waitForExit() from this " "Thread object's thread. It's a guaranteed deadlock!", this); return WOULD_BLOCK; } mExitPending = true; while (mRunning == true) { mThreadExitedCondition.wait(mLock); } // This next line is probably not needed any more, but is being left for // historical reference. Note that each interested party will clear flag. mExitPending = false; return mStatus; } status_t Thread::join() { Mutex::Autolock _l(mLock); if (mThread == getThreadId()) { ALOGW( "Thread (this=%p): don't call join() from this " "Thread object's thread. It's a guaranteed deadlock!", this); return WOULD_BLOCK; } while (mRunning == true) { mThreadExitedCondition.wait(mLock); } return mStatus; } bool Thread::isRunning() const { Mutex::Autolock _l(mLock); return mRunning; } #ifdef HAVE_ANDROID_OS pid_t Thread::getTid() const { // mTid is not defined until the child initializes it, and the caller may need it earlier Mutex::Autolock _l(mLock); pid_t tid; if (mRunning) { pthread_t pthread = android_thread_id_t_to_pthread(mThread); tid = pthread_gettid_np(pthread); } else { ALOGW("Thread (this=%p): getTid() is undefined before run()", this); tid = -1; } return tid; } #endif bool Thread::exitPending() const { Mutex::Autolock _l(mLock); return mExitPending; } }; // namespace android
http://androidxref.com/6.0.0_r1/xref/system/core/libutils/Threads.cpp
status_t Thread::run(const char* name, int32_tpriority, size_t stack) { Mutex::Autolock_l(mLock); .... //如果mCanCallJava为真,则调用createThreadEtc函数,线程函数是_threadLoop。 //_threadLoop是Thread.cpp中定义的一个函数。 if(mCanCallJava) { res = createThreadEtc(_threadLoop,this, name, priority, stack,&mThread); } else{ res = androidCreateRawThreadEtc(_threadLoop, this, name, priority, stack,&mThread); }
上面的mCanCallJava将线程创建函数的逻辑分为两个分支,虽传入的参数都有_threadLoop,但调用的函数却不同。先直接看mCanCallJava为true的这个分支
Thread.h::createThreadEtc()
inline bool createThreadEtc(thread_func_tentryFunction, void *userData, const char*threadName = "android:unnamed_thread", int32_tthreadPriority = PRIORITY_DEFAULT, size_tthreadStackSize = 0, thread_id_t*threadId = 0) { return androidCreateThreadEtc(entryFunction, userData, threadName, threadPriority, threadStackSize,threadId) ? true : false; }
它调用的是androidCreateThreadEtc函数
// gCreateThreadFn是函数指针,初始化时和mCanCallJava为false时使用的是同一个 //线程创建函数。 static android_create_thread_fn gCreateThreadFn= androidCreateRawThreadEtc; int androidCreateThreadEtc(android_thread_func_tentryFunction, void*userData,const char* threadName, int32_tthreadPriority,size_t threadStackSize, android_thread_id_t*threadId) { return gCreateThreadFn(entryFunction, userData, threadName, threadPriority,threadStackSize, threadId); }
androidCreateThreadEtc方法最终会调用CreateThreadFn方法,初始化时和mCanCallJava为false时使用的是同一个
线程创建函数,所以我们要看一下到底什么地方会修改这个mCanCallJava的值。答案就在AndroidRuntime调用startReg的地方,就有可能修改这个函数指针
AndroidRuntime.cpp
/*static*/ int AndroidRuntime::startReg(JNIEnv*env) { //这里会修改函数指针为javaCreateThreadEtc androidSetCreateThreadFunc((android_create_thread_fn)javaCreateThreadEtc); return 0; }
所以,如果mCanCallJava为true,则将调用javaCreateThreadEtc。
AndroidRuntime.cpp
int AndroidRuntime::javaCreateThreadEtc( android_thread_func_tentryFunction, void* userData, const char*threadName, int32_tthreadPriority, size_t threadStackSize, android_thread_id_t* threadId) { void**args = (void**) malloc(3 * sizeof(void*)); intresult; args[0] = (void*) entryFunction; args[1] = userData; args[2] = (void*) strdup(threadName); //调用的还是androidCreateRawThreadEtc,但线程函数却换成了javaThreadShell。 result= androidCreateRawThreadEtc(AndroidRuntime::javaThreadShell, args, threadName, threadPriority,threadStackSize, threadId); return result; }
AndroidRuntime.cpp
http://androidxref.com/6.0.0_r1/xref/frameworks/base/core/jni/AndroidRuntime.cpp
int AndroidRuntime::javaThreadShell(void* args){ ...... intresult; //把这个线程attach到JNI环境中,这样这个线程就可以调用JNI的函数了 if(javaAttachThread(name, &env) != JNI_OK) return -1; //调用实际的线程函数干活 result = (*(android_thread_func_t)start)(userData); //从JNI环境中detach出来。 javaDetachThread(); free(name); returnresult; }
到这里,终于明白了mCanCallJava为true的目的:
1.在调用你的线程函数之前会attach到 JNI环境中,这样,你的线程函数就可以无忧无虑地使用JNI函数了。
2.线程函数退出后,它会从JNI环境中detach,释放一些资源。
进程退出前,dalvik虚拟机会检查是否有attach了,但是最后未detach的线程如果有,则会直接abort,这显然是不好的。
_threadLoop
还记得上面的代码
if(mCanCallJava) { res = createThreadEtc(_threadLoop,this, name, priority, stack,&mThread); } else{ res = androidCreateRawThreadEtc(_threadLoop, this, name, priority, stack,&mThread); }
尽管根据mCanCallJava不同会调用不同的函数,但是都是传入了_threadLoop,所以我们有必要分析这个方法。
int Thread::_threadLoop(void* user) { Thread* const self = static_cast<Thread*>(user); sp<Thread> strong(self->mHoldSelf); wp<Thread> weak(strong); self->mHoldSelf.clear(); #if HAVE_ANDROID_OS self->mTid = gettid(); #endif boolfirst = true; do { bool result; if(first) { first = false; //self代表继承Thread类的对象,第一次进来将调用readyToRun,看看是否准备好 self->mStatus = self->readyToRun(); result = (self->mStatus == NO_ERROR); if (result && !self->mExitPending) { result = self->threadLoop(); } }else { /* 调用子类实现的threadLoop函数,注意这段代码运行在一个do-while循环中。 这表示即使我们的threadLoop返回了,线程也不一定会退出。 */ result = self->threadLoop(); } /* 线程退出的条件: 1)result 为false。这表明,如果子类在threadLoop中返回false,线程就可以 退出。这属于主动退出的情况,是threadLoop自己不想继续干活了,所以返回false。千万别写错threadLoop的返回值。 2)mExitPending为true,这个变量可由Thread类的requestExit函数设置,这种 情况属于被动退出,因为由外界强制设置了退出条件。 */ if(result == false || self->mExitPending) { self->mExitPending = true; self->mLock.lock(); self->mRunning = false; self->mThreadExitedCondition.broadcast(); self->mLock.unlock(); break; } strong.clear(); strong = weak.promote(); }while(strong != 0); return 0; }
_threadLoop运行在一个循环中,它的返回值可以决定是否退出线程。
常用同步类
互斥类——Mutex
Mutex是互斥类,用于多线程访问同一个资源的时候,保证一次只能有一个线程能访问该资源。例如想象你在飞机上如厕,这时卫生间的信息牌上显示“有人”,你必须等里边的人出来后才可进去。这就是互斥的含义。
Thread.h::Mutex的声明和实现
inline Mutex::Mutex(int type, const char* name){ if(type == SHARED) { //type如果是SHARED,则表明这个Mutex支持跨进程的线程同步 //在Audio系统和Surface系统中会经常见到这种用法 pthread_mutexattr_t attr; pthread_mutexattr_init(&attr); pthread_mutexattr_setpshared(&attr, PTHREAD_PROCESS_SHARED); pthread_mutex_init(&mMutex, &attr); pthread_mutexattr_destroy(&attr); } else { pthread_mutex_init(&mMutex, NULL); } } inline Mutex::~Mutex() { pthread_mutex_destroy(&mMutex); } inline status_t Mutex::lock() { return-pthread_mutex_lock(&mMutex); } inline void Mutex::unlock() { pthread_mutex_unlock(&mMutex); } inline status_t Mutex::tryLock() { return-pthread_mutex_trylock(&mMutex); }
关于Mutex的使用,除了初始化外,最重要的是lock和unlock函数的使用,它们的用法如下:
要想独占卫生间,必须先调用Mutex的lock函数。这样,这个区域就被锁住了。如果这块区域之前已被别人锁住,lock函数则会等待,直到可以进入这块区域为止。系统保证一次只有一个线程能lock成功。
· 当你“方便”完毕,记得调用Mutex的unlock以释放互斥区域。这样,其他人的lock才可以成功返回。
· 另外,Mutex还提供了一个trylock函数,该函数只是尝试去锁住该区域,使用者需要根据trylock的返回值判断是否成功锁住了该区域。
AutoLock介绍
AutoLock类是定义在Mutex内部的一个类,Mutex的使用如下
· 显示调用Mutex的lock。
· 在某个时候要记住调用该Mutex的unlock。以上这些操作都必须一一对应,否则会出现“死锁”!充分利用了C++的构造和析构函数,可以达到不忘了释放锁的目的。
Thread.h Mutex::Autolock声明和实现
classAutolock { public: //构造的时候调用lock inline Autolock(Mutex& mutex) : mLock(mutex) { mLock.lock(); } inline Autolock(Mutex* mutex) : mLock(*mutex) { mLock.lock(); } //析构的时候调用unlock inline ~Autolock() { mLock.unlock(); } private: Mutex& mLock; };
AutoLock的用法很简单:
· 先定义一个Mutex,如 Mutex xlock;
· 在使用xlock的地方,定义一个AutoLock,如 AutoLock autoLock(xlock)。
由于C++对象的构造和析构函数都是自动被调用的,所以在AutoLock的生命周期内,xlock的lock和unlock也就自动被调用了,这样就省去了重复书写unlock的麻烦,而且lock和unlock的调用肯定是一一对应的,这样就绝对不会出错。
条件类——Condition
· 线程A做初始化工作,而其他线程比如线程B、C必须等到初始化工作完后才能工作,即线程B、C在等待一个条件,我们称B、C为等待者。
· 当线程A完成初始化工作时,会触发这个条件,那么等待者B、C就会被唤醒。触发这个条件的A就是触发者。
Thread.h::Condition的声明和实现
class Condition { public: enum { PRIVATE = 0, SHARED = 1 }; Condition(); Condition(int type);//如果type是SHARED,表示支持跨进程的条件同步 ~Condition(); //线程B和C等待事件,wait这个名字是不是很形象呢? status_t wait(Mutex& mutex); //线程B和C的超时等待,B和C可以指定等待时间,当超过这个时间,条件却还不满足,则退出等待 status_t waitRelative(Mutex& mutex, nsecs_t reltime); //触发者A用来通知条件已经满足,但是B和C只有一个会被唤醒 voidsignal(); //触发者A用来通知条件已经满足,所有等待者都会被唤醒 voidbroadcast(); private: #if defined(HAVE_PTHREADS) pthread_cond_t mCond; #else void* mState; #endif }
声明很简单,定义也很简单
inline Condition::Condition() { pthread_cond_init(&mCond, NULL); } inline Condition::Condition(int type) { if(type == SHARED) {//设置跨进程的同步支持 pthread_condattr_t attr; pthread_condattr_init(&attr); pthread_condattr_setpshared(&attr, PTHREAD_PROCESS_SHARED); pthread_cond_init(&mCond, &attr); pthread_condattr_destroy(&attr); } else{ pthread_cond_init(&mCond, NULL); } } inline Condition::~Condition() { pthread_cond_destroy(&mCond); } inline status_t Condition::wait(Mutex&mutex) { return-pthread_cond_wait(&mCond, &mutex.mMutex); } inline status_tCondition::waitRelative(Mutex& mutex, nsecs_t reltime) { #if defined(HAVE_PTHREAD_COND_TIMEDWAIT_RELATIVE) structtimespec ts; ts.tv_sec = reltime/1000000000; ts.tv_nsec = reltime%1000000000; return-pthread_cond_timedwait_relative_np(&mCond, &mutex.mMutex, &ts); ...... //有些系统没有实现POSIX的相关函数,所以不同系统需要调用不同的函数 #endif } inline void Condition::signal() { pthread_cond_signal(&mCond); } inline void Condition::broadcast() { pthread_cond_broadcast(&mCond); }
可以看出,Condition的实现全是凭借调用了Raw API的pthread_cond_xxx函数。这里要重点说明的是,Condition类必须配合Mutex来使用。上面代码中,不论是wait、waitRelative、signal还是broadcast的调用,都放在一个Mutex的lock和unlock范围中,尤其是wait和waitRelative函数的调用,这是强制性的。
Condition类和Mutex类使用的例子,在Thread类的requestExitAndWait中就可以体现
Thread.cpp
status_t Thread::requestExitAndWait() { ...... requestExit();//设置退出变量mExitPending为true Mutex::Autolock_l(mLock);//使用Autolock,mLock被锁住 while(mRunning == true) { /* 条件变量的等待,这里为什么要通过while循环来反复检测mRunning? 因为某些时候即使条件类没有被触发,wait也会返回。 */ mThreadExitedCondition.wait(mLock); } mExitPending = false; //退出前,局部变量Mutex::Autolock _l的析构会被调用,unlock也就会被自动调用。 returnmStatus; }
Thread.cpp
int Thread::_threadLoop(void* user) { Thread* const self =static_cast<Thread*>(user); sp<Thread> strong(self->mHoldSelf); wp<Thread> weak(strong); self->mHoldSelf.clear(); do { ...... result= self->threadLoop();//调用子类的threadLoop函数 ...... //如果mExitPending为true,则退出 if(result == false || self->mExitPending) { self->mExitPending = true; //退出前触发条件变量,唤醒等待者 self->mLock.lock();//lock锁住 //mRunning的修改位于锁的保护中。 self->mRunning = false; self->mThreadExitedCondition.broadcast(); self->mLock.unlock();//释放锁 break;//退出循环,此后该线程函数会退出 } ...... }while(strong != 0); return0; }
原子操作函数
所谓原子操作,就是该操作绝不会在执行完毕前被任何其他任务或事件打断,也就说,原子操作是最小的执行单位
static int g_flag = 0; //全局变量g_flag static Mutex lock ;//全局的锁 //线程1执行thread1 void thread1() { //g_flag递减,每次操作前锁住 lock.lock(); g_flag--; lock.unlock(); } //线程2中执行thread2函数 void thread2() { lock.lock(); g_flag++; //线程2对g_flag进行递增操作,每次操作前要取得锁 lock.unlock(); }
为什么需要Mutex来帮忙呢?因为g_flags++或者g_flags—操作都不是原子操作。从汇编指令的角度看,C/C++中的一条语句对应了数条汇编指令。以g_flags++操作为例,它生成的汇编指令可能就是以下三条:
· 从内存中取数据到寄存器。
· 对寄存器中的数据进行递增操作,结果还在寄存器中。
· 寄存器的结果写回内存。
这三条汇编指令,如果按正常的顺序连续执行,是没有问题的,但在多线程时就不能保证了。例如,线程1在执行第一条指令后,线程2由于调度的原因,抢先在线程1之前连续执行完了三条指令。这样,线程1继续执行指令时,它所使用的值就不是线程2更新后的值,而是之前的旧值。再对这个值进行操作便没有意义了。
在一般情况下,处理这种问题可以使用Mutex来加锁保护,但Mutex的使用比它所要保护的内容还复杂,例如,锁的使用将导致从用户态转入内核态,有较大的浪费。那么,有没有简便些的办法让这些加、减等操作不被中断呢?
Android提供了相关的原子操作函数。这里,有必要介绍一下各个函数的作用。
Atomic.h
注意该文件位置在system/core/include/cutils目录中
//原子赋值操作,结果是*addr=value void android_atomic_write(int32_t value,volatile int32_t* addr); //下面所有函数的返回值都是操作前的旧值 //原子加1和原子减1 int32_t android_atomic_inc(volatile int32_t*addr); int32_t android_atomic_dec(volatile int32_t*addr); //原子加法操作,value为被加数 int32_t android_atomic_add(int32_t value,volatile int32_t* addr); //原子“与”和“或”操作 int32_t android_atomic_and(int32_t value,volatile int32_t* addr); int32_t android_atomic_or(int32_t value,volatile int32_t* addr); /* 条件交换的原子操作。只有在oldValue等于*addr时,才会把newValue赋值给*addr 这个函数的返回值须特别注意。返回值非零,表示没有进行赋值操作。返回值为零,表示 进行了原子操作。 */ int android_atomic_cmpxchg(int32_t oldvalue,int32_t newvalue, volatile int32_t*addr);