在上一篇文章《（三）Audio子系統之AudioRecord.startRecording》中已經介紹了AudioRecord如何開始錄制音頻，接下來，繼續分析AudioRecord方法中的read的實現
函數原型：
public int read(byte[] audioData, int offsetInBytes, int sizeInBytes)
作用：
從音頻硬件錄制緩沖區讀取數據，直接復制到指定緩沖區。 如果audioBuffer不是直接的緩沖區，此方法總是返回0
參數：
audioData：寫入的音頻錄制數據
offsetInBytes：audioData的起始偏移值，單位byte
sizeInBytes：讀取的最大字節數
返回值：
讀入緩沖區的總byte數，如果對象屬性沒有初始化，則返回ERROR_INVALID_OPERATION，如果參數不能解析成有效的數據或索引，則返回ERROR_BAD_VALUE。 讀取的總byte數不會超過sizeInBytes
接下來進入系統分析具體實現
frameworks\base\media\java\android\media\AudioRecord.java
public int read(byte[] audioData, int offsetInBytes, int sizeInBytes) {
if (mState != STATE_INITIALIZED) {
return ERROR_INVALID_OPERATION;
}
if ( (audioData == null) || (offsetInBytes < 0 ) || (sizeInBytes < 0)
|| (offsetInBytes + sizeInBytes < 0) // detect integer overflow
|| (offsetInBytes + sizeInBytes > audioData.length)) {
return ERROR_BAD_VALUE;
}
return native_read_in_byte_array(audioData, offsetInBytes, sizeInBytes);
}
這里我們只分析獲取byte[]類型的數據
frameworks\base\core\jni\android_media_AudioRecord.cpp
static jint android_media_AudioRecord_readInByteArray(JNIEnv *env, jobject thiz,
jbyteArray javaAudioData,
jint offsetInBytes, jint sizeInBytes) {
jbyte* recordBuff = NULL;
// get the audio recorder from which we'll read new audio samples
sp<AudioRecord> lpRecorder = getAudioRecord(env, thiz);
if (lpRecorder == NULL) {
ALOGE("Unable to retrieve AudioRecord object, can't record");
return 0;
}
if (!javaAudioData) {
ALOGE("Invalid Java array to store recorded audio, can't record");
return 0;
}
recordBuff = (jbyte *)env->GetByteArrayElements(javaAudioData, NULL);
if (recordBuff == NULL) {
ALOGE("Error retrieving destination for recorded audio data, can't record");
return 0;
}
// read the new audio data from the native AudioRecord object
ssize_t recorderBuffSize = lpRecorder->frameCount()*lpRecorder->frameSize();
ssize_t readSize = lpRecorder->read(recordBuff + offsetInBytes,
sizeInBytes > (jint)recorderBuffSize ?
(jint)recorderBuffSize : sizeInBytes );
env->ReleaseByteArrayElements(javaAudioData, recordBuff, 0);
if (readSize < 0) {
readSize = (jint)AUDIO_JAVA_INVALID_OPERATION;
}
return (jint) readSize;
}
這里根據AudioRecord.cpp中的mFrameCount*mFrameSize
重新計算出Audio緩沖區大小，並與傳過來的size比較，選擇較小的那個數，然后再調用lpRecorder->read函數，需要注意的是mFrameCount這個值在調用openRecord_l函數的時候更新過，我在實際測試的時候發現mFrameCount重新計算過后為3072，而之前是2048，所以這里傳過去的buffSize依然是4092
frameworks\av\media\libmedia\AudioRecord.cpp
ssize_t AudioRecord::read(void* buffer, size_t userSize)
{
if (mTransfer != TRANSFER_SYNC) {
return INVALID_OPERATION;
}
if (ssize_t(userSize) < 0 || (buffer == NULL && userSize != 0)) {
// sanity-check. user is most-likely passing an error code, and it would
// make the return value ambiguous (actualSize vs error).
ALOGE("AudioRecord::read(buffer=%p, size=%zu (%zu)", buffer, userSize, userSize);
return BAD_VALUE;
}
ssize_t read = 0;
Buffer audioBuffer;
while (userSize >= mFrameSize) {
audioBuffer.frameCount = userSize / mFrameSize;
status_t err = obtainBuffer(&audioBuffer, &ClientProxy::kForever);
if (err < 0) {
if (read > 0) {
break;
}
return ssize_t(err);
}
size_t bytesRead = audioBuffer.size;
memcpy(buffer, audioBuffer.i8, bytesRead);
buffer = ((char *) buffer) + bytesRead;
userSize -= bytesRead;
read += bytesRead;
releaseBuffer(&audioBuffer);
}
return read;
}
這個mFrameSize是通過channelCount*采樣精度所占字節計算得出的，所以每次通過obtainBuffer獲取共享內存中的數據，然后通過memcpy把數據拷貝到應用層的buffer中，直到把整個userSize都拷貝到buffer中為止。
這里就詳細分析下obtainBuffer函數
status_t AudioRecord::obtainBuffer(Buffer* audioBuffer, const struct timespec *requested,
struct timespec *elapsed, size_t *nonContig)
{
// previous and new IAudioRecord sequence numbers are used to detect track re-creation
uint32_t oldSequence = 0;
uint32_t newSequence;
Proxy::Buffer buffer;
status_t status = NO_ERROR;
static const int32_t kMaxTries = 5;
int32_t tryCounter = kMaxTries;
do {
// obtainBuffer() is called with mutex unlocked, so keep extra references to these fields to
// keep them from going away if another thread re-creates the track during obtainBuffer()
sp<AudioRecordClientProxy> proxy;
sp<IMemory> iMem;
sp<IMemory> bufferMem;
{
// start of lock scope
AutoMutex lock(mLock);
newSequence = mSequence;
// did previous obtainBuffer() fail due to media server death or voluntary invalidation?
if (status == DEAD_OBJECT) {
// re-create track, unless someone else has already done so
if (newSequence == oldSequence) {
status = restoreRecord_l("obtainBuffer");
if (status != NO_ERROR) {
buffer.mFrameCount = 0;
buffer.mRaw = NULL;
buffer.mNonContig = 0;
break;
}
}
}
oldSequence = newSequence;
// Keep the extra references
proxy = mProxy;
iMem = mCblkMemory;
bufferMem = mBufferMemory;
// Non-blocking if track is stopped
if (!mActive) {
requested = &ClientProxy::kNonBlocking;
}
} // end of lock scope
buffer.mFrameCount = audioBuffer->frameCount;
// FIXME starts the requested timeout and elapsed over from scratch
status = proxy->obtainBuffer(&buffer, requested, elapsed);
} while ((status == DEAD_OBJECT) && (tryCounter-- > 0));
audioBuffer->frameCount = buffer.mFrameCount;
audioBuffer->size = buffer.mFrameCount * mFrameSize;
audioBuffer->raw = buffer.mRaw;
if (nonContig != NULL) {
*nonContig = buffer.mNonContig;
}
return status;
}
在這個函數中的主要工作如下：
1.獲取AudioRecordClientProxy代理，mCblkMemory與mBufferMemory，他們是在構建AudioRecord對象的時候，通過AF端獲取的，即RecordThread::RecordTrack->getCblk()與RecordThread::RecordTrack->getBuffers()
2.在start中，AudioRecordThread.resume之前，mActive已經標記為true了，所以requested還是&ClientProxy::kForever；
3.調用proxy->obtainBuffer繼續獲取數據；
3.更新audioBuffer的frameCount，size以及pcm數據；
這里繼續分析第3步：ClientProxy::obtainBuffer
frameworks\av\media\libmedia\AudioTrackShared.cpp
status_t ClientProxy::obtainBuffer(Buffer* buffer, const struct timespec *requested,
struct timespec *elapsed)
{
LOG_ALWAYS_FATAL_IF(buffer == NULL || buffer->mFrameCount == 0);
struct timespec total; // total elapsed time spent waiting
total.tv_sec = 0;
total.tv_nsec = 0;
bool measure = elapsed != NULL; // whether to measure total elapsed time spent waiting
status_t status;
enum {
TIMEOUT_ZERO, // requested == NULL || *requested == 0
TIMEOUT_INFINITE, // *requested == infinity
TIMEOUT_FINITE, // 0 < *requested < infinity
TIMEOUT_CONTINUE, // additional chances after TIMEOUT_FINITE
} timeout;
if (requested == NULL) {
timeout = TIMEOUT_ZERO;
} else if (requested->tv_sec == 0 && requested->tv_nsec == 0) {
timeout = TIMEOUT_ZERO;
} else if (requested->tv_sec == INT_MAX) {
timeout = TIMEOUT_INFINITE;
} else {
timeout = TIMEOUT_FINITE;
if (requested->tv_sec > 0 || requested->tv_nsec >= MEASURE_NS) {
measure = true;
}
}
struct timespec before;
bool beforeIsValid = false;
audio_track_cblk_t* cblk = mCblk;
bool ignoreInitialPendingInterrupt = true;
// check for shared memory corruption
if (mIsShutdown) {
status = NO_INIT;
goto end;
}
for (;;) {
int32_t flags = android_atomic_and(~CBLK_INTERRUPT, &cblk->mFlags);
// check for track invalidation by server, or server death detection
if (flags & CBLK_INVALID) {
ALOGV("Track invalidated");
status = DEAD_OBJECT;
goto end;
}
// check for obtainBuffer interrupted by client
if (!ignoreInitialPendingInterrupt && (flags & CBLK_INTERRUPT)) {
ALOGV("obtainBuffer() interrupted by client");
status = -EINTR;
goto end;
}
ignoreInitialPendingInterrupt = false;
// compute number of frames available to write (AudioTrack) or read (AudioRecord)
int32_t front;
int32_t rear;
if (mIsOut) {
// The barrier following the read of mFront is probably redundant.
// We're about to perform a conditional branch based on 'filled',
// which will force the processor to observe the read of mFront
// prior to allowing data writes starting at mRaw.
// However, the processor may support speculative execution,
// and be unable to undo speculative writes into shared memory.
// The barrier will prevent such speculative execution.
front = android_atomic_acquire_load(&cblk->u.mStreaming.mFront);
rear = cblk->u.mStreaming.mRear;
} else {
// On the other hand, this barrier is required.
rear = android_atomic_acquire_load(&cblk->u.mStreaming.mRear);
front = cblk->u.mStreaming.mFront;
}
ssize_t filled = rear - front;
// pipe should not be overfull
if (!(0 <= filled && (size_t) filled <= mFrameCount)) {
if (mIsOut) {
ALOGE("Shared memory control block is corrupt (filled=%zd, mFrameCount=%zu); "
"shutting down", filled, mFrameCount);
mIsShutdown = true;
status = NO_INIT;
goto end;
}
// for input, sync up on overrun
filled = 0;
cblk->u.mStreaming.mFront = rear;
(void) android_atomic_or(CBLK_OVERRUN, &cblk->mFlags);
}
// don't allow filling pipe beyond the nominal size
size_t avail = mIsOut ? mFrameCount - filled : filled;
if (avail > 0) {
// 'avail' may be non-contiguous, so return only the first contiguous chunk
size_t part1;
if (mIsOut) {
rear &= mFrameCountP2 - 1;
part1 = mFrameCountP2 - rear;
} else {
front &= mFrameCountP2 - 1;
part1 = mFrameCountP2 - front;
}
if (part1 > avail) {
part1 = avail;
}
if (part1 > buffer->mFrameCount) {
part1 = buffer->mFrameCount;
}
buffer->mFrameCount = part1;
buffer->mRaw = part1 > 0 ?
&((char *) mBuffers)[(mIsOut ? rear : front) * mFrameSize] : NULL;
buffer->mNonContig = avail - part1;
mUnreleased = part1;
status = NO_ERROR;
break;
}
struct timespec remaining;
const struct timespec *ts;
switch (timeout) {
case TIMEOUT_ZERO:
status = WOULD_BLOCK;
goto end;
case TIMEOUT_INFINITE:
ts = NULL;
break;
case TIMEOUT_FINITE:
timeout = TIMEOUT_CONTINUE;
if (MAX_SEC == 0) {
ts = requested;
break;
}
// fall through
case TIMEOUT_CONTINUE:
// FIXME we do not retry if requested < 10ms? needs documentation on this state machine
if (!measure || requested->tv_sec < total.tv_sec ||
(requested->tv_sec == total.tv_sec && requested->tv_nsec <= total.tv_nsec)) {
status = TIMED_OUT;
goto end;
}
remaining.tv_sec = requested->tv_sec - total.tv_sec;
if ((remaining.tv_nsec = requested->tv_nsec - total.tv_nsec) < 0) {
remaining.tv_nsec += 1000000000;
remaining.tv_sec++;
}
if (0 < MAX_SEC && MAX_SEC < remaining.tv_sec) {
remaining.tv_sec = MAX_SEC;
remaining.tv_nsec = 0;
}
ts = &remaining;
break;
default:
LOG_ALWAYS_FATAL("obtainBuffer() timeout=%d", timeout);
ts = NULL;
break;
}
int32_t old = android_atomic_and(~CBLK_FUTEX_WAKE, &cblk->mFutex);
if (!(old & CBLK_FUTEX_WAKE)) {
if (measure && !beforeIsValid) {
clock_gettime(CLOCK_MONOTONIC, &before);
beforeIsValid = true;
}
errno = 0;
(void) syscall(__NR_futex, &cblk->mFutex,
mClientInServer ? FUTEX_WAIT_PRIVATE : FUTEX_WAIT, old & ~CBLK_FUTEX_WAKE, ts);
// update total elapsed time spent waiting
if (measure) {
struct timespec after;
clock_gettime(CLOCK_MONOTONIC, &after);
total.tv_sec += after.tv_sec - before.tv_sec;
long deltaNs = after.tv_nsec - before.tv_nsec;
if (deltaNs < 0) {
deltaNs += 1000000000;
total.tv_sec--;
}
if ((total.tv_nsec += deltaNs) >= 1000000000) {
total.tv_nsec -= 1000000000;
total.tv_sec++;
}
before = after;
beforeIsValid = true;
}
switch (errno) {
case 0: // normal wakeup by server, or by binderDied()
case EWOULDBLOCK: // benign race condition with server
case EINTR: // wait was interrupted by signal or other spurious wakeup
case ETIMEDOUT: // time-out expired
// FIXME these error/non-0 status are being dropped
break;
default:
status = errno;
ALOGE("%s unexpected error %s", __func__, strerror(status));
goto end;
}
}
}
end:
if (status != NO_ERROR) {
buffer->mFrameCount = 0;
buffer->mRaw = NULL;
buffer->mNonContig = 0;
mUnreleased = 0;
}
if (elapsed != NULL) {
*elapsed = total;
}
if (requested == NULL) {
requested = &kNonBlocking;
}
if (measure) {
ALOGV("requested %ld.%03ld elapsed %ld.%03ld",
requested->tv_sec, requested->tv_nsec / 1000000,
total.tv_sec, total.tv_nsec / 1000000);
}
return status;
}
這個函數的主要工作如下：
1.從ClientProxy::kForever的定義可知，這個tv_sec是INT_MAX，所以timeout為TIMEOUT_INFINITE；
2.從cblk中獲取到rear以及front，之前分析過了，這兩個指針是在RecordThread線程中維護的，他一邊在讀取pcm數據，一直在更新最新數據指針的位置；
3.如果發現獲取到的數據小於mFrameCount，或者沒有獲取到數據，那么也就是表示應用讀的太快了，其實應該說RecordThread讀的太慢了導致，這時候也需要設置為OVERRUN，則更新cblk的指針，重置filled為0；
4.如果獲取到了數據，則把錄音數據放到mRaw中，把獲取到的數據大小放到mFrameCount中。
總結：
到這里，整個獲取pcm數據的流程就結束了，到這里我們應該能理解整個Audio系統中對AudioBuffer的管理策略了，即通過RecordThread線程把數據從硬件層讀取到IMemory中，然后應用層在去IMemory中去讀取。
由於作者內功有限，若文章中存在錯誤或不足的地方，還請給位大佬指出，不勝感激！