blob: ae053a94a2a65f2adf50ff5c0e0eb6cd6c534a38 [file] [log] [blame]
* Copyright (C) 2009 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
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* See the License for the specific language governing permissions and
* limitations under the License.
#include <cutils/misc.h>
#include <cutils/config_utils.h>
#include <cutils/compiler.h>
#include <utils/String8.h>
#include <utils/Vector.h>
#include <utils/SortedVector.h>
#include <binder/BinderService.h>
#include <system/audio.h>
#include <system/audio_policy.h>
#include <hardware/audio_policy.h>
#include <media/IAudioPolicyService.h>
#include <media/ToneGenerator.h>
#include <media/AudioEffect.h>
namespace android {
// ----------------------------------------------------------------------------
class AudioPolicyService :
public BinderService<AudioPolicyService>,
public BnAudioPolicyService,
// public AudioPolicyClientInterface,
public IBinder::DeathRecipient
friend class BinderService<AudioPolicyService>;
// for BinderService
static const char *getServiceName() ANDROID_API { return "media.audio_policy"; }
virtual status_t dump(int fd, const Vector<String16>& args);
// BnAudioPolicyService (see AudioPolicyInterface for method descriptions)
virtual status_t setDeviceConnectionState(audio_devices_t device,
audio_policy_dev_state_t state,
const char *device_address);
virtual audio_policy_dev_state_t getDeviceConnectionState(
audio_devices_t device,
const char *device_address);
virtual status_t setPhoneState(audio_mode_t state);
virtual status_t setForceUse(audio_policy_force_use_t usage, audio_policy_forced_cfg_t config);
virtual audio_policy_forced_cfg_t getForceUse(audio_policy_force_use_t usage);
virtual audio_io_handle_t getOutput(audio_stream_type_t stream,
uint32_t samplingRate = 0,
audio_format_t format = AUDIO_FORMAT_DEFAULT,
audio_channel_mask_t channelMask = 0,
audio_output_flags_t flags =
const audio_offload_info_t *offloadInfo = NULL);
virtual status_t startOutput(audio_io_handle_t output,
audio_stream_type_t stream,
int session = 0);
virtual status_t stopOutput(audio_io_handle_t output,
audio_stream_type_t stream,
int session = 0);
virtual void releaseOutput(audio_io_handle_t output);
virtual audio_io_handle_t getInput(audio_source_t inputSource,
uint32_t samplingRate = 0,
audio_format_t format = AUDIO_FORMAT_DEFAULT,
audio_channel_mask_t channelMask = 0,
int audioSession = 0);
virtual status_t startInput(audio_io_handle_t input);
virtual status_t stopInput(audio_io_handle_t input);
virtual void releaseInput(audio_io_handle_t input);
virtual status_t initStreamVolume(audio_stream_type_t stream,
int indexMin,
int indexMax);
virtual status_t setStreamVolumeIndex(audio_stream_type_t stream,
int index,
audio_devices_t device);
virtual status_t getStreamVolumeIndex(audio_stream_type_t stream,
int *index,
audio_devices_t device);
virtual uint32_t getStrategyForStream(audio_stream_type_t stream);
virtual audio_devices_t getDevicesForStream(audio_stream_type_t stream);
virtual audio_io_handle_t getOutputForEffect(const effect_descriptor_t *desc);
virtual status_t registerEffect(const effect_descriptor_t *desc,
audio_io_handle_t io,
uint32_t strategy,
int session,
int id);
virtual status_t unregisterEffect(int id);
virtual status_t setEffectEnabled(int id, bool enabled);
virtual bool isStreamActive(audio_stream_type_t stream, uint32_t inPastMs = 0) const;
virtual bool isStreamActiveRemotely(audio_stream_type_t stream, uint32_t inPastMs = 0) const;
virtual bool isSourceActive(audio_source_t source) const;
virtual status_t queryDefaultPreProcessing(int audioSession,
effect_descriptor_t *descriptors,
uint32_t *count);
virtual status_t onTransact(
uint32_t code,
const Parcel& data,
Parcel* reply,
uint32_t flags);
// IBinder::DeathRecipient
virtual void binderDied(const wp<IBinder>& who);
// Helpers for the struct audio_policy_service_ops implementation.
// This is used by the audio policy manager for certain operations that
// are implemented by the policy service.
virtual void setParameters(audio_io_handle_t ioHandle,
const char *keyValuePairs,
int delayMs);
virtual status_t setStreamVolume(audio_stream_type_t stream,
float volume,
audio_io_handle_t output,
int delayMs = 0);
virtual status_t startTone(audio_policy_tone_t tone, audio_stream_type_t stream);
virtual status_t stopTone();
virtual status_t setVoiceVolume(float volume, int delayMs = 0);
virtual bool isOffloadSupported(const audio_offload_info_t &config);
status_t doStopOutput(audio_io_handle_t output,
audio_stream_type_t stream,
int session = 0);
void doReleaseOutput(audio_io_handle_t output);
AudioPolicyService() ANDROID_API;
virtual ~AudioPolicyService();
status_t dumpInternals(int fd);
// Thread used for tone playback and to send audio config commands to audio flinger
// For tone playback, using a separate thread is necessary to avoid deadlock with mLock because
// startTone() and stopTone() are normally called with mLock locked and requesting a tone start
// or stop will cause calls to AudioPolicyService and an attempt to lock mLock.
// For audio config commands, it is necessary because audio flinger requires that the calling
// process (user) has permission to modify audio settings.
class AudioCommandThread : public Thread {
class AudioCommand;
// commands for tone AudioCommand
enum {
AudioCommandThread (String8 name, const wp<AudioPolicyService>& service);
virtual ~AudioCommandThread();
status_t dump(int fd);
// Thread virtuals
virtual void onFirstRef();
virtual bool threadLoop();
void exit();
void startToneCommand(ToneGenerator::tone_type type,
audio_stream_type_t stream);
void stopToneCommand();
status_t volumeCommand(audio_stream_type_t stream, float volume,
audio_io_handle_t output, int delayMs = 0);
status_t parametersCommand(audio_io_handle_t ioHandle,
const char *keyValuePairs, int delayMs = 0);
status_t voiceVolumeCommand(float volume, int delayMs = 0);
void stopOutputCommand(audio_io_handle_t output,
audio_stream_type_t stream,
int session);
void releaseOutputCommand(audio_io_handle_t output);
void insertCommand_l(AudioCommand *command, int delayMs = 0);
// descriptor for requested tone playback event
class AudioCommand {
: mCommand(-1) {}
void dump(char* buffer, size_t size);
int mCommand; // START_TONE, STOP_TONE ...
nsecs_t mTime; // time stamp
Condition mCond; // condition for status return
status_t mStatus; // command status
bool mWaitStatus; // true if caller is waiting for status
void *mParam; // command parameter (ToneData, VolumeData, ParametersData)
class ToneData {
ToneGenerator::tone_type mType; // tone type (START_TONE only)
audio_stream_type_t mStream; // stream type (START_TONE only)
class VolumeData {
audio_stream_type_t mStream;
float mVolume;
audio_io_handle_t mIO;
class ParametersData {
audio_io_handle_t mIO;
String8 mKeyValuePairs;
class VoiceVolumeData {
float mVolume;
class StopOutputData {
audio_io_handle_t mIO;
audio_stream_type_t mStream;
int mSession;
class ReleaseOutputData {
audio_io_handle_t mIO;
Mutex mLock;
Condition mWaitWorkCV;
Vector <AudioCommand *> mAudioCommands; // list of pending commands
ToneGenerator *mpToneGenerator; // the tone generator
AudioCommand mLastCommand; // last processed command (used by dump)
String8 mName; // string used by wake lock fo delayed commands
wp<AudioPolicyService> mService;
class EffectDesc {
EffectDesc(const char *name, const effect_uuid_t& uuid) :
mUuid(uuid) { }
EffectDesc(const EffectDesc& orig) :
mUuid(orig.mUuid) {
// deep copy mParams
for (size_t k = 0; k < orig.mParams.size(); k++) {
effect_param_t *origParam = orig.mParams[k];
// psize and vsize are rounded up to an int boundary for allocation
size_t origSize = sizeof(effect_param_t) +
((origParam->psize + 3) & ~3) +
((origParam->vsize + 3) & ~3);
effect_param_t *dupParam = (effect_param_t *) malloc(origSize);
memcpy(dupParam, origParam, origSize);
// This works because the param buffer allocation is also done by
// multiples of 4 bytes originally. In theory we should memcpy only
// the actual param size, that is without rounding vsize.
/*virtual*/ ~EffectDesc() {
for (size_t k = 0; k < mParams.size(); k++) {
char *mName;
effect_uuid_t mUuid;
Vector <effect_param_t *> mParams;
class InputSourceDesc {
InputSourceDesc() {}
/*virtual*/ ~InputSourceDesc() {
for (size_t j = 0; j < mEffects.size(); j++) {
delete mEffects[j];
Vector <EffectDesc *> mEffects;
class InputDesc {
InputDesc(int session) : mSessionId(session) {}
/*virtual*/ ~InputDesc() {}
const int mSessionId;
Vector< sp<AudioEffect> >mEffects;
static const char * const kInputSourceNames[AUDIO_SOURCE_CNT -1];
void setPreProcessorEnabled(const InputDesc *inputDesc, bool enabled);
status_t loadPreProcessorConfig(const char *path);
status_t loadEffects(cnode *root, Vector <EffectDesc *>& effects);
EffectDesc *loadEffect(cnode *root);
status_t loadInputSources(cnode *root, const Vector <EffectDesc *>& effects);
audio_source_t inputSourceNameToEnum(const char *name);
InputSourceDesc *loadInputSource(cnode *root, const Vector <EffectDesc *>& effects);
void loadEffectParameters(cnode *root, Vector <effect_param_t *>& params);
effect_param_t *loadEffectParameter(cnode *root);
size_t readParamValue(cnode *node,
char *param,
size_t *curSize,
size_t *totSize);
size_t growParamSize(char *param,
size_t size,
size_t *curSize,
size_t *totSize);
// Internal dump utilities.
status_t dumpPermissionDenial(int fd);
mutable Mutex mLock; // prevents concurrent access to AudioPolicy manager functions changing
// device connection state or routing
sp<AudioCommandThread> mAudioCommandThread; // audio commands thread
sp<AudioCommandThread> mTonePlaybackThread; // tone playback thread
sp<AudioCommandThread> mOutputCommandThread; // process stop and release output
struct audio_policy_device *mpAudioPolicyDev;
struct audio_policy *mpAudioPolicy;
KeyedVector< audio_source_t, InputSourceDesc* > mInputSources;
KeyedVector< audio_io_handle_t, InputDesc* > mInputs;
}; // namespace android