Android-SharedPreferences源码学习与最佳实践 / 憋错料

最近有个任务是要做应用启动时间优化，然后记录系统启动的各个步骤所占用的时间，发现有一个方法是操作SharedPreferences的，里面仅仅是读了2个key，然后更新一下值，然后再写回去，耗时竟然在500ms以上（应用初次安装的时候），感到非常吃惊。以前只是隐约的知道SharedPreferences是跟硬盘上的一个xml文件对应的，具体的实现还真没研究过，下面我们就来看看SharedPreferences到底是个什么玩意，为什么效率会这么低？

SharedPreferences是存放在ContextImpl里面的，所以先看写ContextImpl这个类：

ContextImpl.java（https://github.com/android/platform_frameworks_base/blob/master/core/java/android/app/ContextImpl.java）：

/**

* Map from package name, to preference name, to cached preferences.

*/

private static ArrayMap<string, arraymap<string,="" sharedpreferencesimpl="">> sSharedPrefs;//在内存的一份缓存

@Override

public SharedPreferences getSharedPreferences(String name, int mode) {

SharedPreferencesImpl sp;

synchronized (ContextImpl.class) {//同步的

if (sSharedPrefs == null) {

sSharedPrefs = new ArrayMap<string, arraymap<string,="" sharedpreferencesimpl="">>();

}

final String packageName = getPackageName();

ArrayMap<string, sharedpreferencesimpl=""> packagePrefs = sSharedPrefs.get(packageName);

if (packagePrefs == null) {

packagePrefs = new ArrayMap<string, sharedpreferencesimpl="">();

sSharedPrefs.put(packageName, packagePrefs);

}

// At least one application in the world actually passes in a null

// name. This happened to work because when we generated the file name

// we would stringify it to "null.xml". Nice.

if (mPackageInfo.getApplicationInfo().targetSdkVersion <

Build.VERSION_CODES.KITKAT) {

if (name == null) {

name = "null";

}

sp = packagePrefs.get(name);

if (sp == null) {

File prefsFile = getSharedPrefsFile(name);//这里是找到文件

sp = new SharedPreferencesImpl(prefsFile, mode);//在这里会做初始化，从硬盘加载数据

packagePrefs.put(name, sp);//缓存起来

return sp;

}

if ((mode & Context.MODE_MULTI_PROCESS) != 0 ||

getApplicationInfo().targetSdkVersion < android.os.Build.VERSION_CODES.HONEYCOMB) {

// If somebody else (some other process) changed the prefs

// file behind our back, we reload it. This has been the

// historical (if undocumented) behavior.

sp.startReloadIfChangedUnexpectedly();

}

return sp;

}</string,></string,></string,></string,>

getSharedPreferences()做的事情很简单，一目了然，我们重点看下SharedPreferencesImpl.java这个类：
SharedPreferencesImpl.java（https://github.com/android/platform_frameworks_base/blob/master/core/java/android/app/SharedPreferencesImpl.java）
首先是构造函数：

SharedPreferencesImpl(File file, int mode) {

mFile = file;//这个是硬盘上的文件

mBackupFile = makeBackupFile(file);//这个是备份文件，当mFile出现crash的时候，会使用mBackupFile来替换

mMode = mode;//这个是打开方式

mLoaded = false;//这个是一个标志位，文件是否加载完成，因为文件的加载是一个异步的过程

mMap = null;//保存数据用

startLoadFromDisk();//开始从硬盘异步加载

}

//还两个很重要的成员：

private int mDiskWritesInFlight = 0; //有多少批次没有commit到disk的写操作，每个批次可能会对应多个k-v

private final Object mWritingToDiskLock = new Object();//写硬盘文件时候加锁

//从硬盘加载

private void startLoadFromDisk() {

synchronized (this) {//先把状态置为未加载

mLoaded = false;

}

new Thread("SharedPreferencesImpl-load") {//开了一个线程，异步加载

public void run() {

synchronized (SharedPreferencesImpl.this) {

loadFromDiskLocked();//由SharedPreferencesImpl.this锁保护

}

}.start();

}

//从硬盘加载

private void loadFromDiskLocked() {

if (mLoaded) {//如果已经加载，直接退出

return;

}

if (mBackupFile.exists()) {//如果存在备份文件，优先使用备份文件

mFile.delete();

mBackupFile.renameTo(mFile);

}

// Debugging

if (mFile.exists() && !mFile.canRead()) {

Log.w(TAG, "Attempt to read preferences file " + mFile + " without permission");

}

Map map = null;

StructStat stat = null;

try {

stat = Libcore.os.stat(mFile.getPath());

if (mFile.canRead()) {

BufferedInputStream str = null;

try {

str = new BufferedInputStream(new FileInputStream(mFile), 16*1024);//从硬盘把数据读出来

map = XmlUtils.readMapXml(str);//做xml解析

} catch (XmlPullParserException e) {

Log.w(TAG, "getSharedPreferences", e);

} catch (FileNotFoundException e) {

Log.w(TAG, "getSharedPreferences", e);

} catch (IOException e) {

Log.w(TAG, "getSharedPreferences", e);

} finally {

IoUtils.closeQuietly(str);

}

} catch (ErrnoException e) {

}

mLoaded = true;//设置标志位，已经加载完成

if (map != null) {

mMap = map; //保存到mMap

mStatTimestamp = stat.st_mtime;//记录文件的时间戳

mStatSize = stat.st_size;//记录文件的大小

} else {

mMap = new HashMap<string, object="">();

}

notifyAll();//唤醒等待线程

}

</string,>

然后我们随便看一个读请求：

public int getInt(String key, int defValue) {

synchronized (this) {//还是得首先获取this锁

awaitLoadedLocked(); //这一步完成以后，说明肯定已经加载完了

Integer v = (Integer)mMap.get(key);//直接从内存读取

return v != null ? v : defValue;

}

//等待数据加载完成

private void awaitLoadedLocked() {

if (!mLoaded) { //如果还没加载

// Raise an explicit StrictMode onReadFromDisk for this

// thread, since the real read will be in a different

// thread and otherwise ignored by StrictMode.

BlockGuard.getThreadPolicy().onReadFromDisk();//从硬盘加载

}

while (!mLoaded) {//这要是没加载完

try {

wait();//等

} catch (InterruptedException unused) {

}

看一下写操作，写是通过Editor来做的：

public Editor edit() {

// TODO: remove the need to call awaitLoadedLocked() when

// requesting an editor. will require some work on the

// Editor, but then we should be able to do:

//

// context.getSharedPreferences(..).edit().putString(..).apply()

//

// ... all without blocking.

//注释很有意思，获取edit的时候，可以把这个同步去掉，但是如果去掉就需要在Editor上做一些工作（？？？）。

//但是，好处是context.getSharedPreferences(..).edit().putString(..).apply()整个过程都不阻塞

synchronized (this) {//还是先等待加载完成

awaitLoadedLocked();

}

return new EditorImpl();//返回一个EditorImpl，它是一个内部类

}

public final class EditorImpl implements Editor {

//写操作暂时会把数据放在这里面

private final Map<string, object=""> mModified = Maps.newHashMap();//由this锁保护

//是否要清空所有的preferences

private boolean mClear = false;

public Editor putInt(String key, int value) {

synchronized (this) {//首先获取this锁

mModified.put(key, value);//并不是直接修改mMap，而是放到mModified里面

return this;

}

</string,>

看一下commit：

public boolean commit() {

MemoryCommitResult mcr = commitToMemory(); //首先提交到内存

SharedPreferencesImpl.this.enqueueDiskWrite(mcr, null /* sync write on this thread okay */);//然后提交到硬盘

try {

mcr.writtenToDiskLatch.await();//等待写硬盘完成

} catch (InterruptedException e) {

return false;

}

notifyListeners(mcr);

return mcr.writeToDiskResult;

}

commitToMemory()这个方法主要是用来更新内存缓存的mMap：

// Returns true if any changes were made

private MemoryCommitResult commitToMemory() {

MemoryCommitResult mcr = new MemoryCommitResult();

synchronized (SharedPreferencesImpl.this) { //加SharedPreferencesImpl锁，写内存的时候不允许读

// We optimistically don‘t make a deep copy until a memory commit comes in when we‘re already writing to disk.

if (mDiskWritesInFlight > 0) {//如果存在没有提交的写， mDiskWritesInFlight是SharedPreferences的成员变量

// We can‘t modify our mMap as a currently in-flight write owns it. Clone it before modifying it.

// noinspection unchecked

mMap = new HashMap<string, object="">(mMap);//clone一个mMap，没明白！

}

mcr.mapToWriteToDisk = mMap;

mDiskWritesInFlight++;//批次数目加1

boolean hasListeners = mListeners.size() > 0;

if (hasListeners) {

mcr.keysModified = new ArrayList<string>();

mcr.listeners = new HashSet<onsharedpreferencechangelistener>(mListeners.keySet());

}

synchronized (this) {//对当前的Editor加锁

if (mClear) {//只有当调用了clear()才会把这个值置为true

if (!mMap.isEmpty()) {//如果mMap不是空

mcr.changesMade = true;

mMap.clear();//清空mMap。mMap里面存的是整个的Preferences

}

mClear = false;

}

for (Map.Entry<string, object=""> e : mModified.entrySet()) {//遍历所有要commit的entry

String k = e.getKey();

Object v = e.getValue();

if (v == this) { // magic value for a removal mutation

if (!mMap.containsKey(k)) {

continue;

}

mMap.remove(k);

} else {

boolean isSame = false;

if (mMap.containsKey(k)) {

Object existingValue = mMap.get(k);

if (existingValue != null && existingValue.equals(v)) {

continue;

}

mMap.put(k, v);//这里是往里面放，因为最外层有对SharedPreferencesImpl.this加锁，写是没问题的

}

mcr.changesMade = true;

if (hasListeners) {

mcr.keysModified.add(k);

}

mModified.clear();//清空editor

}

return mcr;

}</string,></onsharedpreferencechangelistener></string></string,>

//这是随后的写硬盘

private void enqueueDiskWrite(final MemoryCommitResult mcr, final Runnable postWriteRunnable) {

final Runnable writeToDiskRunnable = new Runnable() {

public void run() {

synchronized (mWritingToDiskLock) {

writeToFile(mcr);

}

synchronized (SharedPreferencesImpl.this) {

mDiskWritesInFlight--;

}

if (postWriteRunnable != null) {

postWriteRunnable.run();

}

};

final boolean isFromSyncCommit = (postWriteRunnable == null);//如果是commit，postWriteRunnable是null

// Typical #commit() path with fewer allocations, doing a write on

// the current thread.

if (isFromSyncCommit) {//如果是调用的commit

boolean wasEmpty = false;

synchronized (SharedPreferencesImpl.this) {

wasEmpty = mDiskWritesInFlight == 1;//如果只有一个批次等待写入

}

if (wasEmpty) {

writeToDiskRunnable.run();//不用另起线程，直接在当前线程执行，很nice的优化！

return;

}

//如果不是调用的commit，会走下面的分支

//如或有多个批次等待写入，另起线程来写，从方法名可以看出来也是串行的写，写文件本来就应该串行！

QueuedWork.singleThreadExecutor().execute(writeToDiskRunnable);

}

看下writeToDiskRunnable都干了些什么：

final Runnable writeToDiskRunnable = new Runnable() {//这是工作在另一个线程

public void run() {

synchronized (mWritingToDiskLock) {//mWritingToDiskLock是SharedPreferencesImpl的成员变量，保证单线程写文件，

//不能用this锁是因为editor上可能会存在多个commit或者apply

//也不能用SharedPreferences锁，因为会阻塞读，不错！

writeToFile(mcr);//写到文件

}

synchronized (SharedPreferencesImpl.this) {

mDiskWritesInFlight--;//批次减1

}

if (postWriteRunnable != null) {

postWriteRunnable.run();//这个是写完以后的回调

}

};

下面是真正要写硬盘了：

// Note: must hold mWritingToDiskLock

private void writeToFile(MemoryCommitResult mcr) {

// Rename the current file so it may be used as a backup during the next read

if (mFile.exists()) {

if (!mcr.changesMade) {//如果没有修改，直接返回

// If the file already exists, but no changes were

// made to the underlying map, it‘s wasteful to

// re-write the file. Return as if we wrote it

// out.

mcr.setDiskWriteResult(true);

return;

}

if (!mBackupFile.exists()) {//先备份

if (!mFile.renameTo(mBackupFile)) {

Log.e(TAG, "Couldn‘t rename file " + mFile

+ " to backup file " + mBackupFile);

mcr.setDiskWriteResult(false);

return;

}

} else {//删除重建

mFile.delete();

}

// Attempt to write the file, delete the backup and return true as atomically as

// possible. If any exception occurs, delete the new file; next time we will restore

// from the backup.

try {

FileOutputStream str = createFileOutputStream(mFile);

if (str == null) {

mcr.setDiskWriteResult(false);

return;

}

XmlUtils.writeMapXml(mcr.mapToWriteToDisk, str);

FileUtils.sync(str);//强制写到硬盘

str.close();

ContextImpl.setFilePermissionsFromMode(mFile.getPath(), mMode, 0);

try {

final StructStat stat = Libcore.os.stat(mFile.getPath());

synchronized (this) {

mStatTimestamp = stat.st_mtime;//更新文件时间戳

mStatSize = stat.st_size;//更新文件大小

}

} catch (ErrnoException e) {

// Do nothing

}

// Writing was successful, delete the backup file if there is one.

mBackupFile.delete();

mcr.setDiskWriteResult(true);

return;

} catch (XmlPullParserException e) {

Log.w(TAG, "writeToFile: Got exception:", e);

} catch (IOException e) {

Log.w(TAG, "writeToFile: Got exception:", e);

}

// Clean up an unsuccessfully written file

if (mFile.exists()) {

if (!mFile.delete()) {

Log.e(TAG, "Couldn‘t clean up partially-written file " + mFile);

}

mcr.setDiskWriteResult(false);

}

public static boolean sync(FileOutputStream stream) {

try {

if (stream != null) {

stream.getFD().sync();//强制写硬盘

}

return true;

} catch (IOException e) {

}

return false;

}

这里面还有一个跟commit长得很像的方法叫apply():

public void apply() {

final MemoryCommitResult mcr = commitToMemory();//首先也是提交到内存

final Runnable awaitCommit = new Runnable() {

public void run() {

try {

mcr.writtenToDiskLatch.await();//等待写入到硬盘

} catch (InterruptedException ignored) {

}

};

QueuedWork.add(awaitCommit);

Runnable postWriteRunnable = new Runnable() {

public void run() {

awaitCommit.run();

QueuedWork.remove(awaitCommit);

}

};

SharedPreferencesImpl.this.enqueueDiskWrite(mcr, postWriteRunnable);//这个地方传递的postWriteRunnable不再是null

// Okay to notify the listeners before it‘s hit disk

// because the listeners should always get the same

// SharedPreferences instance back, which has the

// changes reflected in memory.

notifyListeners(mcr);

}

我们已经看过enqueueDiskWrite()这个方法了，因为参数postWriteRunnable不是null，最终会执行：
QueuedWork.singleThreadExecutor().execute(writeToDiskRunnable);
这是在单独的线程上做写硬盘的操作，写完以后会回调postWriteRunnable，等待写硬盘完成！

从上面的代码可以得出以下结论：
（1）SharedPreferences在第一次加载的时候，会从硬盘异步的读文件，然后会在内存做缓存。
（2）SharedPreferences的读都是读的内存缓存。
（3）如果是commmit()写，是先把数据更新到内存，然后同步到硬盘，整个过程是在同一个线程中同步来做的。
（4）如果是apply()写,首先也是写到内存，但是会另起一个线程异步的来写硬盘。因为我们在读的时候，是直接从内存读取的，因此，用apply()而不是commit()会提高性能。
（5）如果有多个key要写入，不要每次都commit或者apply，因为这里面会存在很多的加锁操作，更高效的使用方式是这样：editor.putInt("","").putString("","").putBoolean("","").apply();并且所有的putXXX()的结尾都会返回this，方便链式编程。
（6）这里面有三级的锁：SharedPreferences，Editor， mWritingToDiskLock。
mWritingToDiskLock是对应硬盘上的文件，Editor是保护mModified的，SharedPreferences是保护mMap的。
参考：
http://stackoverflow.com/questions/19148282/read-speed-of-sharedpreferences
http://stackoverflow.com/questions/12567077/is-sharedpreferences-access-time-consuming

原文：http://www.2cto.com/kf/201312/268547.html

时间： 2024-10-07 22:42:13

Android-SharedPreferences源码学习与最佳实践

Android-SharedPreferences源码学习与最佳实践的相关文章

Android SharedPreferences源码分析.md

Android系统源码学习步骤

[Android阅读代码]android-async-http源码学习一

多个Android项目源码-覆盖方方面面值得学习

Android事件分发详解(三)——ViewGroup的dispatchTouchEvent()源码学习

[Android FrameWork 6.0源码学习] View的重绘过程之WindowManager的addView方法

【流媒体开发】VLC Media Player - Android 平台源码编译与二次开发详解 (提供详细800M下载好的编译源码及eclipse可调试播放器源码下载)

2016年最牛逼的分类Android项目源码免费一次性打包下载！

最新app源码下载：200款优秀Android项目源码