今天分析Android的Audio系统时,对mpAudioPolicy->get_input进行了分析,没想到这一句话的背后如此复杂,简直是一句话引出的血案啊!
分析结果如下:(关于排版:各个变量的调用关系的图在博客里屏幕太窄了,可以复制到notepad++中看)
调用mpAudioPolicy的get_input,它是mpAudioPolicyDev->create_audio_policy(mpAudioPolicyDev, &aps_ops, this,&mpAudioPolicy)中赋值的
定义:struct audio_policy_device *mpAudioPolicyDev;
定义:struct audio_policy *mpAudioPolicy;
而audio_policy_dev_open(module, &mpAudioPolicyDev);中将赋值给mpAudioPolicyDev
而audio_policy_dev_open定义如下
static inline int audio_policy_dev_open(const hw_module_t* module,
struct audio_policy_device** device)
{
return module->methods->open(module, AUDIO_POLICY_INTERFACE,
(hw_device_t**)device);
}
看来得找model的来源了
AudioPolicyInterface
|继
|承
AudioPolicyManagerBase
|继
|承
AudioPolicyManagerDefault编译成的audio_policy.default.so//hardware/libhardware_legacy/audio/AudioPolicyManagerDefault.cpp、
| //AudioPolicyManagerDefault.h
| // 而audio_policy.default.so又依赖于静态库libaudiopolicy_legacy.a(hardware/libhardware_legacy/audio/Android.mk
| //中:LOCAL_WHOLE_STATIC_LIBRARIES := libaudiopolicy_legacy)
| // libaudiopolicy_legacy.a由 AudioPolicyManagerBase.cpp、AudioPolicyCompatClient.cpp、audio_policy_hal.cpp生成
| // HAL_MODULE_INFO_SYM在audio_policy_hal.cpp中定义,下面分析audio_policy_hal.cpp中的内容
|
|
hw_get_module_by_class(id, NULL, module)//id=AUDIO_POLICY_HARDWARE_MODULE_ID,hw_get_module_by_class即会加载audio_policy的HAL的SO,
| //此从原生系统中找到了audio_policy.stub.so、audio_policy.default.so、audio_policy.msm8960.so
| // 根据使用getprop命令查看模拟器的系统属性可知,模拟器中将会加载audio_policy.default.so,
| //相关源码在hardware/libhardware_legacy/audio/目录下
|
|
hw_get_module(AUDIO_POLICY_HARDWARE_MODULE_ID, &module)//#define AUDIO_POLICY_HARDWARE_MODULE_ID "audio_policy"; 而module的类型为hw_module_t
|
|
module->methods->open(module, AUDIO_POLICY_INTERFACE,(hw_device_t**)device)//audio_policy_dev_open函数中,将hw_device_t类型转换成了
|//audio_policy_device返回给mpAudioPolicyDev
|
|
|
audio_policy_dev_open(module, &mpAudioPolicyDev)
|
|
mpAudioPolicyDev->
create_audio_policy(mpAudioPolicyDev, &aps_ops, this,&mpAudioPolicy)
|
|
mpAudioPolicy->get_input(...)
看audio_policy_hal.cpp中的:
static struct hw_module_methods_t legacy_ap_module_methods = {
open: legacy_ap_dev_open
};
struct legacy_ap_module HAL_MODULE_INFO_SYM = {
module: {
common: {
tag: HARDWARE_MODULE_TAG,
version_major: 1,
version_minor: 0,
id: AUDIO_POLICY_HARDWARE_MODULE_ID,
name: "LEGACY Audio Policy HAL",
author: "The Android Open Source Project",
methods: &legacy_ap_module_methods,
dso : NULL,
reserved : {0},
},
},
};
所以module->methods->open将调用到legacy_ap_dev_open函数
而mpAudioPolicyDev最终将成为legacy_ap_dev_open函数的最后一个参数hw_device_t** device
而由:
dev->device.create_audio_policy = create_legacy_ap;
*device = &dev->device.common;
知mpAudioPolicy成为create_legacy_ap的最后一个参数struct audio_policy **ap
而由:
lap->policy.get_input = ap_get_input;
*ap = &lap->policy;
知
mpAudioPolicy->get_input(...)最终将变成调用ap_get_input
所以:mpAudioPolicy->get_input将调用hardware/libhardware_legacy/audio/audio_policy_hal.cpp中的ap_get_input函数
下面分析ap_get_input函数它的内容如下:
struct legacy_audio_policy *lap = to_lap(pol);
return lap->apm->getInput((int) inputSource, sampling_rate, (int) format, channelMask,(AudioSystem::audio_in_acoustics)acoustics);
而lap->apm由create_legacy_ap中如下语句创造
lap->apm = createAudioPolicyManager(lap->service_client);
所以此处的apm即new AudioPolicyManagerDefault(lap->service_client)对象,而lap->service_client = new AudioPolicyCompatClient(aps_ops, service);
其中的参数是本cpp中调用create_audio_policy(mpAudioPolicyDev, &aps_ops, this,&mpAudioPolicy)中传入的aps_ops和this参数
AudioPolicyManagerBase由定义知它完全等价于AudioPolicyManagerBase
最终分析结论:mpAudioPolicy->get_input将最终调用hardware/libhardware_legacy/audio/AudioPolicyManagerBase.cpp中的getInput函数
而在AudioPolicyManagerBase中调用AudioPolicyCompatClient的openInput,在openInput中会根据audio_policy.conf文件的配置信息决定使用什么音频设备,
以此确定audio_module_handle_t的值,
然后又在openInput中又将通过本cpp传入的aps_ops回调本cpp中的aps_open_input_on_module
aps_open_input_on_module中调用AudioFlinger的openInput了
AudioFlinger的openInput会产生一个唯一整数作为audio_io_handle_t
AudioFlinger的openInput还会根据AudioPolicyManagerBase中选出的合适音频设备的audio_module_handle_t来在mAudioHwDevs中查找对应的设备的AudioHwDevice对象inHwDev
找出inHwDev设备后,调用改设备的HAL中的open_input_stream,即audio_hw.c中的adev_open_input_stream
在adev_open_input_stream中构造HAL层的自定义类型stream_in(如stub_stream_in),并在stream_in中植入自己定义的HAL层函数,然后把该stream_in返回AudioFlinger
AudioFlinger使用HAL返回的stream_in以及前面找出的设备的AudioHwDevice构造AF中的AudioStreamIn类型对象input,然后创建一个新的RecordThread线程,
并将该线程以前面传入的audio_io_handle_t做索引存入mRecordThreads中,返回audio_io_handle_t,即audio_io_handle_t是在AudioFlinger中找出对应线程的索引