正文
Android 4.4 Kitkat 音频实现及简要分析
提示：扫一扫查出行【扫一扫了解最新限行尾号】
复制提示
版权声明：本文为博主原创文章，未经博主同意不得转载。https://blog.csdn.net/jingxia2008/article/details/26701899
在 Android 4.4 上实现录放音背景Android 自 ICS 開始，音频系统就有了非常大的变化，先是抛弃了 alsalib。然后是採用了 AIO。各级框架上，都有了自己的特色，与 Linux 的音频应用渐行渐远，形成了自己独特的音频管理和音频配置功能。总的来说改进还是非常大，至少在用户体验上已经大大的超越了之前的版本号。
我们就从 4.4 的音频实现上来分析当中的一些变化和实现机制。
要求首先是硬件功能正常，这个不表。 Linux 支持 alsa 驱动。生成 alsa 子系统。最好是能够在 buildroot 等其它文件系统上事先測试音频的播放和录制。HAL音频的 HAL 简单实现，參考 device/asus/grouper/audio ， 假设没有太复杂的音频配置，基本上能够在这个代码基础上略微改动，主要是一些播放和录制的參数。这个 HAL 已经实现了通用的接口，而且调用的也是标准的 tinyalsa 的接口，移植性非常高。我们这里使用的 wm8904，功能不多。直接使用就可以。Android 的配置主要是4个文件 audio_policy.conf media_profiles.xml media_codecs.xml mixer_paths.xml 參考 asus ，不必大改。基本照抄，全然能够直接使用，开源万岁。做好文件系统。这个时候系统应该就不使用 default 的 stub 音频 HAL ， 而是用我们加入的 audio HAL 了。 可是是否能发声，这个多半还是不能。调试audio 系统调用了 libtinyalsa libaudioutils libaudioroute 几个底层库。这几个移植了一些通用的 alsa 设备打开配置功能，可是详细平台却并不一定都能正常运行，主要是这些库实现都非常easy，没有考虑全面，你的硬件细节可能刚好被他们忽略了。相同以我们的 wm8904 来说，我们不支持 time stamp ，可是 tinyalsa 是默认打开的必须将其关掉。
disable tstamp for wm8904. Change-Id: Ia22aa6ed39ede6214657487344d0488be93e5544diff --git a/pcm.c b/pcm.c
index 4501777..94cf6ee 100644
--- a/pcm.c
+++ b/pcm.c
@@ -691,7 +691,7 @@ struct pcm *pcm_open(unsigned int card, unsigned int device, memset(&sparams, 0, sizeof(sparams));
- sparams.tstamp_mode = SNDRV_PCM_TSTAMP_ENABLE;
+ sparams.tstamp_mode = SNDRV_PCM_TSTAMP_NONE;
sparams.period_step = 1; if (!config->start_threshold) {
详细哪些參数不正确，或者不合适，就须要 Android 驱动project师依据硬件设计和芯片手冊，逐个查证。配置到一个音频系统工作的最佳状态。
那么用户体验才干最好。
Android 音频系统分析下面分析基于 4.4.2audio HAL
tinyalsa 与 audioroute
Android 音频系统基于 Linux 的 ALSA 驱动， tinyalsa 在 alsa 的驱动基础上封装音频接口。提供给 audio HAL。 audio HAL 提供接口给 Android audioflinger 等 framework。HAL 须要实现 audio 硬件的打开与关闭（这里是 android 觉得的硬件）。static inline int audio_hw_device_open(const struct hw_module_t* module,
struct audio_hw_device** device)
{
return module->methods->open(module, AUDIO_HARDWARE_INTERFACE,
(struct hw_device_t**)device);
}static inline int audio_hw_device_close(struct audio_hw_device* device)
{
return device->common.close(&device->common);
}
须要实现 in 和 out 的 数据流
struct audio_stream_out struct audio_stream_in in 主要有 read 方法用于读取音频数据。 out 主要有 write 方法，写入数据到设备，分别实现录音和放音。详见： hardware/libhardware/include/hardware/audio.h当中的方法又是调用的 tinyalsa 的接口，关于 pcm 的操作：/* Open and close a stream */
struct pcm *pcm_open(unsigned int card, unsigned int device,
unsigned int flags, struct pcm_config *config);
int pcm_close(struct pcm *pcm);
int pcm_is_ready(struct pcm *pcm);/* Obtain the parameters for a PCM */
struct pcm_params *pcm_params_get(unsigned int card, unsigned int device,
unsigned int flags);
void pcm_params_free(struct pcm_params *pcm_params);
unsigned int pcm_params_get_min(struct pcm_params *pcm_params,
enum pcm_param param);
unsigned int pcm_params_get_max(struct pcm_params *pcm_params,
enum pcm_param param);/* Set and get config */
int pcm_get_config(struct pcm *pcm, struct pcm_config *config);
int pcm_set_config(struct pcm *pcm, struct pcm_config *config);/* Returns a human readable reason for the last error */
const char *pcm_get_error(struct pcm *pcm);/* Returns the sample size in bits for a PCM format.
* As with ALSA formats, this is the storage size for the format, whereas the
* format represents the number of significant bits. For example,
* PCM_FORMAT_S24_LE uses 32 bits of storage.
*/
unsigned int pcm_format_to_bits(enum pcm_format format);/* Returns the buffer size (int frames) that should be used for pcm_write. */
unsigned int pcm_get_buffer_size(struct pcm *pcm);
unsigned int pcm_frames_to_bytes(struct pcm *pcm, unsigned int frames);
unsigned int pcm_bytes_to_frames(struct pcm *pcm, unsigned int bytes);/* Returns the pcm latency in ms */
unsigned int pcm_get_latency(struct pcm *pcm);/* Returns available frames in pcm buffer and corresponding time stamp.
* The clock is CLOCK_MONOTONIC if flag PCM_MONOTONIC was specified in pcm_open,
* otherwise the clock is CLOCK_REALTIME.
* For an input stream, frames available are frames ready for the
* application to read.
* For an output stream, frames available are the number of empty frames available
* for the application to write.
*/
int pcm_get_htimestamp(struct pcm *pcm, unsigned int *avail,
struct timespec *tstamp);/* Write data to the fifo.
* Will start playback on the first write or on a write that
* occurs after a fifo underrun.
*/
int pcm_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_read(struct pcm *pcm, void *data, unsigned int count);/*
* mmap() support.
*/
int pcm_mmap_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_mmap_read(struct pcm *pcm, void *data, unsigned int count);
int pcm_mmap_begin(struct pcm *pcm, void **areas, unsigned int *offset,
unsigned int *frames);
int pcm_mmap_commit(struct pcm *pcm, unsigned int offset, unsigned int frames);/* Start and stop a PCM channel that doesn't transfer data */
int pcm_start(struct pcm *pcm);
int pcm_stop(struct pcm *pcm);/* Interrupt driven API */
int pcm_wait(struct pcm *pcm, int timeout);/* Change avail_min after the stream has been opened with no need to stop the stream.
* Only accepted if opened with PCM_MMAP and PCM_NOIRQ flags
*/
int pcm_set_avail_min(struct pcm *pcm, int avail_min);
值得一提的是 HAL 如今不包括 route 的操作， audio route 交给了 libaudioroute.so ， 它也是调用 tinyalsa 的接口，并包括一个 xml 解析器。解析 mixer_paths.xml 里面的 route 配置数据。这样系统就能够对 alsa 的 pcm 和 mixer 进行操作了，理论上应该能够放音了。使用 tinyalsa 提供的工具能够进行測试，当然无法測试 HAL 的接口。tinycap tinymix tinypcminfo tinyplaytinyplay 能够播放 wav 格式的纯音频数据。 tinymix 能够查看和配置音频路径：
root@sama5d3:/ # tinymix
Mixer name: 'wm8904 @ SAMA5D3EK'
Number of controls: 41
ctl type num name value
0 INT 1 EQ1 Volume 12
1 INT 1 EQ2 Volume 12
2 INT 1 EQ3 Volume 12
3 INT 1 EQ4 Volume 12
4 INT 1 EQ5 Volume 12
5 INT 2 Digital Capture Volume 96 96
6 ENUM 1 Left Caputure Mode Single-Ended
7 ENUM 1 Right Capture Mode Single-Ended
8 INT 2 Capture Volume 5 5
9 BOOL 2 Capture Switch Off Off
10 BOOL 1 High Pass Filter Switch On
11 ENUM 1 High Pass Filter Mode Hi-fi
12 BOOL 1 ADC 128x OSR Switch On
13 INT 1 Digital Playback Boost Volume 0
14 INT 2 Digital Playback Volume 96 96
15 INT 2 Headphone Volume 45 45
16 BOOL 2 Headphone Switch On On
17 BOOL 2 Headphone ZC Switch On On
18 INT 2 Line Output Volume 57 57
19 BOOL 2 Line Output Switch On On
20 BOOL 2 Line Output ZC Switch On On
21 BOOL 1 EQ Switch Off
22 BOOL 1 DRC Switch Off
23 ENUM 1 DRC Path ADC
24 BOOL 1 DAC OSRx2 Switch Off
25 BOOL 1 DAC Deemphasis Switch Off
26 INT 2 Digital Sidetone Volume 0 0
27 ENUM 1 LINER Mux DAC
28 ENUM 1 LINEL Mux DAC
29 ENUM 1 HPR Mux DAC
30 ENUM 1 HPL Mux DAC
31 ENUM 1 Right Sidetone None
32 ENUM 1 Left Sidetone None
33 ENUM 1 DACR Mux Right
34 ENUM 1 DACL Mux Left
35 ENUM 1 AIFOUTR Mux Right
36 ENUM 1 AIFOUTL Mux Left
37 ENUM 1 Right Capture Inverting Mux IN1R
38 ENUM 1 Right Capture Mux IN2R
39 ENUM 1 Left Capture Inverting Mux IN1L
40 ENUM 1 Left Capture Mux IN2L
audioflinger
audioflinger 是 audio 音频server。它会载入 audio hal ，并处理 audio 应用发出音频请求。这个分析的有非常多，參考下面：http://blog.csdn.net/DroidPhone/article/details/5941344
Android 自 ICS 開始，音频系统就有了非常大的变化，先是抛弃了 alsalib。然后是採用了 AIO。各级框架上，都有了自己的特色，与 Linux 的音频应用渐行渐远，形成了自己独特的音频管理和音频配置功能。
总的来说改进还是非常大，至少在用户体验上已经大大的超越了之前的版本号。
我们就从 4.4 的音频实现上来分析当中的一些变化和实现机制。
要求首先是硬件功能正常，这个不表。 Linux 支持 alsa 驱动。生成 alsa 子系统。最好是能够在 buildroot 等其它文件系统上事先測试音频的播放和录制。HAL音频的 HAL 简单实现，參考 device/asus/grouper/audio ， 假设没有太复杂的音频配置，基本上能够在这个代码基础上略微改动，主要是一些播放和录制的參数。这个 HAL 已经实现了通用的接口，而且调用的也是标准的 tinyalsa 的接口，移植性非常高。我们这里使用的 wm8904，功能不多。直接使用就可以。Android 的配置主要是4个文件 audio_policy.conf media_profiles.xml media_codecs.xml mixer_paths.xml 參考 asus ，不必大改。基本照抄，全然能够直接使用，开源万岁。做好文件系统。这个时候系统应该就不使用 default 的 stub 音频 HAL ， 而是用我们加入的 audio HAL 了。 可是是否能发声，这个多半还是不能。调试audio 系统调用了 libtinyalsa libaudioutils libaudioroute 几个底层库。这几个移植了一些通用的 alsa 设备打开配置功能，可是详细平台却并不一定都能正常运行，主要是这些库实现都非常easy，没有考虑全面，你的硬件细节可能刚好被他们忽略了。相同以我们的 wm8904 来说，我们不支持 time stamp ，可是 tinyalsa 是默认打开的必须将其关掉。
disable tstamp for wm8904. Change-Id: Ia22aa6ed39ede6214657487344d0488be93e5544diff --git a/pcm.c b/pcm.c
index 4501777..94cf6ee 100644
--- a/pcm.c
+++ b/pcm.c
@@ -691,7 +691,7 @@ struct pcm *pcm_open(unsigned int card, unsigned int device, memset(&sparams, 0, sizeof(sparams));
- sparams.tstamp_mode = SNDRV_PCM_TSTAMP_ENABLE;
+ sparams.tstamp_mode = SNDRV_PCM_TSTAMP_NONE;
sparams.period_step = 1; if (!config->start_threshold) {
详细哪些參数不正确，或者不合适，就须要 Android 驱动project师依据硬件设计和芯片手冊，逐个查证。配置到一个音频系统工作的最佳状态。
那么用户体验才干最好。
Android 音频系统分析下面分析基于 4.4.2audio HAL
tinyalsa 与 audioroute
Android 音频系统基于 Linux 的 ALSA 驱动， tinyalsa 在 alsa 的驱动基础上封装音频接口。提供给 audio HAL。 audio HAL 提供接口给 Android audioflinger 等 framework。HAL 须要实现 audio 硬件的打开与关闭（这里是 android 觉得的硬件）。static inline int audio_hw_device_open(const struct hw_module_t* module,
struct audio_hw_device** device)
{
return module->methods->open(module, AUDIO_HARDWARE_INTERFACE,
(struct hw_device_t**)device);
}static inline int audio_hw_device_close(struct audio_hw_device* device)
{
return device->common.close(&device->common);
}
须要实现 in 和 out 的 数据流
struct audio_stream_out struct audio_stream_in in 主要有 read 方法用于读取音频数据。 out 主要有 write 方法，写入数据到设备，分别实现录音和放音。详见： hardware/libhardware/include/hardware/audio.h当中的方法又是调用的 tinyalsa 的接口，关于 pcm 的操作：/* Open and close a stream */
struct pcm *pcm_open(unsigned int card, unsigned int device,
unsigned int flags, struct pcm_config *config);
int pcm_close(struct pcm *pcm);
int pcm_is_ready(struct pcm *pcm);/* Obtain the parameters for a PCM */
struct pcm_params *pcm_params_get(unsigned int card, unsigned int device,
unsigned int flags);
void pcm_params_free(struct pcm_params *pcm_params);
unsigned int pcm_params_get_min(struct pcm_params *pcm_params,
enum pcm_param param);
unsigned int pcm_params_get_max(struct pcm_params *pcm_params,
enum pcm_param param);/* Set and get config */
int pcm_get_config(struct pcm *pcm, struct pcm_config *config);
int pcm_set_config(struct pcm *pcm, struct pcm_config *config);/* Returns a human readable reason for the last error */
const char *pcm_get_error(struct pcm *pcm);/* Returns the sample size in bits for a PCM format.
* As with ALSA formats, this is the storage size for the format, whereas the
* format represents the number of significant bits. For example,
* PCM_FORMAT_S24_LE uses 32 bits of storage.
*/
unsigned int pcm_format_to_bits(enum pcm_format format);/* Returns the buffer size (int frames) that should be used for pcm_write. */
unsigned int pcm_get_buffer_size(struct pcm *pcm);
unsigned int pcm_frames_to_bytes(struct pcm *pcm, unsigned int frames);
unsigned int pcm_bytes_to_frames(struct pcm *pcm, unsigned int bytes);/* Returns the pcm latency in ms */
unsigned int pcm_get_latency(struct pcm *pcm);/* Returns available frames in pcm buffer and corresponding time stamp.
* The clock is CLOCK_MONOTONIC if flag PCM_MONOTONIC was specified in pcm_open,
* otherwise the clock is CLOCK_REALTIME.
* For an input stream, frames available are frames ready for the
* application to read.
* For an output stream, frames available are the number of empty frames available
* for the application to write.
*/
int pcm_get_htimestamp(struct pcm *pcm, unsigned int *avail,
struct timespec *tstamp);/* Write data to the fifo.
* Will start playback on the first write or on a write that
* occurs after a fifo underrun.
*/
int pcm_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_read(struct pcm *pcm, void *data, unsigned int count);/*
* mmap() support.
*/
int pcm_mmap_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_mmap_read(struct pcm *pcm, void *data, unsigned int count);
int pcm_mmap_begin(struct pcm *pcm, void **areas, unsigned int *offset,
unsigned int *frames);
int pcm_mmap_commit(struct pcm *pcm, unsigned int offset, unsigned int frames);/* Start and stop a PCM channel that doesn't transfer data */
int pcm_start(struct pcm *pcm);
int pcm_stop(struct pcm *pcm);/* Interrupt driven API */
int pcm_wait(struct pcm *pcm, int timeout);/* Change avail_min after the stream has been opened with no need to stop the stream.
* Only accepted if opened with PCM_MMAP and PCM_NOIRQ flags
*/
int pcm_set_avail_min(struct pcm *pcm, int avail_min);
值得一提的是 HAL 如今不包括 route 的操作， audio route 交给了 libaudioroute.so ， 它也是调用 tinyalsa 的接口，并包括一个 xml 解析器。解析 mixer_paths.xml 里面的 route 配置数据。这样系统就能够对 alsa 的 pcm 和 mixer 进行操作了，理论上应该能够放音了。使用 tinyalsa 提供的工具能够进行測试，当然无法測试 HAL 的接口。tinycap tinymix tinypcminfo tinyplaytinyplay 能够播放 wav 格式的纯音频数据。 tinymix 能够查看和配置音频路径：
root@sama5d3:/ # tinymix
Mixer name: 'wm8904 @ SAMA5D3EK'
Number of controls: 41
ctl type num name value
0 INT 1 EQ1 Volume 12
1 INT 1 EQ2 Volume 12
2 INT 1 EQ3 Volume 12
3 INT 1 EQ4 Volume 12
4 INT 1 EQ5 Volume 12
5 INT 2 Digital Capture Volume 96 96
6 ENUM 1 Left Caputure Mode Single-Ended
7 ENUM 1 Right Capture Mode Single-Ended
8 INT 2 Capture Volume 5 5
9 BOOL 2 Capture Switch Off Off
10 BOOL 1 High Pass Filter Switch On
11 ENUM 1 High Pass Filter Mode Hi-fi
12 BOOL 1 ADC 128x OSR Switch On
13 INT 1 Digital Playback Boost Volume 0
14 INT 2 Digital Playback Volume 96 96
15 INT 2 Headphone Volume 45 45
16 BOOL 2 Headphone Switch On On
17 BOOL 2 Headphone ZC Switch On On
18 INT 2 Line Output Volume 57 57
19 BOOL 2 Line Output Switch On On
20 BOOL 2 Line Output ZC Switch On On
21 BOOL 1 EQ Switch Off
22 BOOL 1 DRC Switch Off
23 ENUM 1 DRC Path ADC
24 BOOL 1 DAC OSRx2 Switch Off
25 BOOL 1 DAC Deemphasis Switch Off
26 INT 2 Digital Sidetone Volume 0 0
27 ENUM 1 LINER Mux DAC
28 ENUM 1 LINEL Mux DAC
29 ENUM 1 HPR Mux DAC
30 ENUM 1 HPL Mux DAC
31 ENUM 1 Right Sidetone None
32 ENUM 1 Left Sidetone None
33 ENUM 1 DACR Mux Right
34 ENUM 1 DACL Mux Left
35 ENUM 1 AIFOUTR Mux Right
36 ENUM 1 AIFOUTL Mux Left
37 ENUM 1 Right Capture Inverting Mux IN1R
38 ENUM 1 Right Capture Mux IN2R
39 ENUM 1 Left Capture Inverting Mux IN1L
40 ENUM 1 Left Capture Mux IN2L
audioflinger
audioflinger 是 audio 音频server。它会载入 audio hal ，并处理 audio 应用发出音频请求。这个分析的有非常多，參考下面：http://blog.csdn.net/DroidPhone/article/details/5941344
音频的 HAL 简单实现，參考 device/asus/grouper/audio ， 假设没有太复杂的音频配置，基本上能够在这个代码基础上略微改动，主要是一些播放和录制的參数。这个 HAL 已经实现了通用的接口，而且调用的也是标准的 tinyalsa 的接口，移植性非常高。我们这里使用的 wm8904，功能不多。直接使用就可以。
Android 的配置主要是4个文件 audio_policy.conf media_profiles.xml media_codecs.xml mixer_paths.xml 參考 asus ，不必大改。基本照抄，全然能够直接使用，开源万岁。做好文件系统。这个时候系统应该就不使用 default 的 stub 音频 HAL ， 而是用我们加入的 audio HAL 了。 可是是否能发声，这个多半还是不能。调试audio 系统调用了 libtinyalsa libaudioutils libaudioroute 几个底层库。这几个移植了一些通用的 alsa 设备打开配置功能，可是详细平台却并不一定都能正常运行，主要是这些库实现都非常easy，没有考虑全面，你的硬件细节可能刚好被他们忽略了。相同以我们的 wm8904 来说，我们不支持 time stamp ，可是 tinyalsa 是默认打开的必须将其关掉。
disable tstamp for wm8904. Change-Id: Ia22aa6ed39ede6214657487344d0488be93e5544diff --git a/pcm.c b/pcm.c
index 4501777..94cf6ee 100644
--- a/pcm.c
+++ b/pcm.c
@@ -691,7 +691,7 @@ struct pcm *pcm_open(unsigned int card, unsigned int device, memset(&sparams, 0, sizeof(sparams));
- sparams.tstamp_mode = SNDRV_PCM_TSTAMP_ENABLE;
+ sparams.tstamp_mode = SNDRV_PCM_TSTAMP_NONE;
sparams.period_step = 1; if (!config->start_threshold) {
详细哪些參数不正确，或者不合适，就须要 Android 驱动project师依据硬件设计和芯片手冊，逐个查证。配置到一个音频系统工作的最佳状态。
那么用户体验才干最好。
Android 音频系统分析下面分析基于 4.4.2audio HAL
tinyalsa 与 audioroute
Android 音频系统基于 Linux 的 ALSA 驱动， tinyalsa 在 alsa 的驱动基础上封装音频接口。提供给 audio HAL。 audio HAL 提供接口给 Android audioflinger 等 framework。HAL 须要实现 audio 硬件的打开与关闭（这里是 android 觉得的硬件）。static inline int audio_hw_device_open(const struct hw_module_t* module,
struct audio_hw_device** device)
{
return module->methods->open(module, AUDIO_HARDWARE_INTERFACE,
(struct hw_device_t**)device);
}static inline int audio_hw_device_close(struct audio_hw_device* device)
{
return device->common.close(&device->common);
}
须要实现 in 和 out 的 数据流
struct audio_stream_out struct audio_stream_in in 主要有 read 方法用于读取音频数据。 out 主要有 write 方法，写入数据到设备，分别实现录音和放音。详见： hardware/libhardware/include/hardware/audio.h当中的方法又是调用的 tinyalsa 的接口，关于 pcm 的操作：/* Open and close a stream */
struct pcm *pcm_open(unsigned int card, unsigned int device,
unsigned int flags, struct pcm_config *config);
int pcm_close(struct pcm *pcm);
int pcm_is_ready(struct pcm *pcm);/* Obtain the parameters for a PCM */
struct pcm_params *pcm_params_get(unsigned int card, unsigned int device,
unsigned int flags);
void pcm_params_free(struct pcm_params *pcm_params);
unsigned int pcm_params_get_min(struct pcm_params *pcm_params,
enum pcm_param param);
unsigned int pcm_params_get_max(struct pcm_params *pcm_params,
enum pcm_param param);/* Set and get config */
int pcm_get_config(struct pcm *pcm, struct pcm_config *config);
int pcm_set_config(struct pcm *pcm, struct pcm_config *config);/* Returns a human readable reason for the last error */
const char *pcm_get_error(struct pcm *pcm);/* Returns the sample size in bits for a PCM format.
* As with ALSA formats, this is the storage size for the format, whereas the
* format represents the number of significant bits. For example,
* PCM_FORMAT_S24_LE uses 32 bits of storage.
*/
unsigned int pcm_format_to_bits(enum pcm_format format);/* Returns the buffer size (int frames) that should be used for pcm_write. */
unsigned int pcm_get_buffer_size(struct pcm *pcm);
unsigned int pcm_frames_to_bytes(struct pcm *pcm, unsigned int frames);
unsigned int pcm_bytes_to_frames(struct pcm *pcm, unsigned int bytes);/* Returns the pcm latency in ms */
unsigned int pcm_get_latency(struct pcm *pcm);/* Returns available frames in pcm buffer and corresponding time stamp.
* The clock is CLOCK_MONOTONIC if flag PCM_MONOTONIC was specified in pcm_open,
* otherwise the clock is CLOCK_REALTIME.
* For an input stream, frames available are frames ready for the
* application to read.
* For an output stream, frames available are the number of empty frames available
* for the application to write.
*/
int pcm_get_htimestamp(struct pcm *pcm, unsigned int *avail,
struct timespec *tstamp);/* Write data to the fifo.
* Will start playback on the first write or on a write that
* occurs after a fifo underrun.
*/
int pcm_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_read(struct pcm *pcm, void *data, unsigned int count);/*
* mmap() support.
*/
int pcm_mmap_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_mmap_read(struct pcm *pcm, void *data, unsigned int count);
int pcm_mmap_begin(struct pcm *pcm, void **areas, unsigned int *offset,
unsigned int *frames);
int pcm_mmap_commit(struct pcm *pcm, unsigned int offset, unsigned int frames);/* Start and stop a PCM channel that doesn't transfer data */
int pcm_start(struct pcm *pcm);
int pcm_stop(struct pcm *pcm);/* Interrupt driven API */
int pcm_wait(struct pcm *pcm, int timeout);/* Change avail_min after the stream has been opened with no need to stop the stream.
* Only accepted if opened with PCM_MMAP and PCM_NOIRQ flags
*/
int pcm_set_avail_min(struct pcm *pcm, int avail_min);
值得一提的是 HAL 如今不包括 route 的操作， audio route 交给了 libaudioroute.so ， 它也是调用 tinyalsa 的接口，并包括一个 xml 解析器。解析 mixer_paths.xml 里面的 route 配置数据。这样系统就能够对 alsa 的 pcm 和 mixer 进行操作了，理论上应该能够放音了。使用 tinyalsa 提供的工具能够进行測试，当然无法測试 HAL 的接口。tinycap tinymix tinypcminfo tinyplaytinyplay 能够播放 wav 格式的纯音频数据。 tinymix 能够查看和配置音频路径：
root@sama5d3:/ # tinymix
Mixer name: 'wm8904 @ SAMA5D3EK'
Number of controls: 41
ctl type num name value
0 INT 1 EQ1 Volume 12
1 INT 1 EQ2 Volume 12
2 INT 1 EQ3 Volume 12
3 INT 1 EQ4 Volume 12
4 INT 1 EQ5 Volume 12
5 INT 2 Digital Capture Volume 96 96
6 ENUM 1 Left Caputure Mode Single-Ended
7 ENUM 1 Right Capture Mode Single-Ended
8 INT 2 Capture Volume 5 5
9 BOOL 2 Capture Switch Off Off
10 BOOL 1 High Pass Filter Switch On
11 ENUM 1 High Pass Filter Mode Hi-fi
12 BOOL 1 ADC 128x OSR Switch On
13 INT 1 Digital Playback Boost Volume 0
14 INT 2 Digital Playback Volume 96 96
15 INT 2 Headphone Volume 45 45
16 BOOL 2 Headphone Switch On On
17 BOOL 2 Headphone ZC Switch On On
18 INT 2 Line Output Volume 57 57
19 BOOL 2 Line Output Switch On On
20 BOOL 2 Line Output ZC Switch On On
21 BOOL 1 EQ Switch Off
22 BOOL 1 DRC Switch Off
23 ENUM 1 DRC Path ADC
24 BOOL 1 DAC OSRx2 Switch Off
25 BOOL 1 DAC Deemphasis Switch Off
26 INT 2 Digital Sidetone Volume 0 0
27 ENUM 1 LINER Mux DAC
28 ENUM 1 LINEL Mux DAC
29 ENUM 1 HPR Mux DAC
30 ENUM 1 HPL Mux DAC
31 ENUM 1 Right Sidetone None
32 ENUM 1 Left Sidetone None
33 ENUM 1 DACR Mux Right
34 ENUM 1 DACL Mux Left
35 ENUM 1 AIFOUTR Mux Right
36 ENUM 1 AIFOUTL Mux Left
37 ENUM 1 Right Capture Inverting Mux IN1R
38 ENUM 1 Right Capture Mux IN2R
39 ENUM 1 Left Capture Inverting Mux IN1L
40 ENUM 1 Left Capture Mux IN2L
audioflinger
audioflinger 是 audio 音频server。它会载入 audio hal ，并处理 audio 应用发出音频请求。这个分析的有非常多，參考下面：http://blog.csdn.net/DroidPhone/article/details/5941344
audio 系统调用了 libtinyalsa libaudioutils libaudioroute 几个底层库。
这几个移植了一些通用的 alsa 设备打开配置功能，可是详细平台却并不一定都能正常运行，主要是这些库实现都非常easy，没有考虑全面，你的硬件细节可能刚好被他们忽略了。相同以我们的 wm8904 来说，我们不支持 time stamp ，可是 tinyalsa 是默认打开的必须将其关掉。
disable tstamp for wm8904. Change-Id: Ia22aa6ed39ede6214657487344d0488be93e5544diff --git a/pcm.c b/pcm.c
index 4501777..94cf6ee 100644
--- a/pcm.c
+++ b/pcm.c
@@ -691,7 +691,7 @@ struct pcm *pcm_open(unsigned int card, unsigned int device, memset(&sparams, 0, sizeof(sparams));
- sparams.tstamp_mode = SNDRV_PCM_TSTAMP_ENABLE;
+ sparams.tstamp_mode = SNDRV_PCM_TSTAMP_NONE;
sparams.period_step = 1; if (!config->start_threshold) {
详细哪些參数不正确，或者不合适，就须要 Android 驱动project师依据硬件设计和芯片手冊，逐个查证。配置到一个音频系统工作的最佳状态。
那么用户体验才干最好。
Android 音频系统分析下面分析基于 4.4.2audio HAL
tinyalsa 与 audioroute
Android 音频系统基于 Linux 的 ALSA 驱动， tinyalsa 在 alsa 的驱动基础上封装音频接口。提供给 audio HAL。 audio HAL 提供接口给 Android audioflinger 等 framework。HAL 须要实现 audio 硬件的打开与关闭（这里是 android 觉得的硬件）。static inline int audio_hw_device_open(const struct hw_module_t* module,
struct audio_hw_device** device)
{
return module->methods->open(module, AUDIO_HARDWARE_INTERFACE,
(struct hw_device_t**)device);
}static inline int audio_hw_device_close(struct audio_hw_device* device)
{
return device->common.close(&device->common);
}
须要实现 in 和 out 的 数据流
struct audio_stream_out struct audio_stream_in in 主要有 read 方法用于读取音频数据。 out 主要有 write 方法，写入数据到设备，分别实现录音和放音。详见： hardware/libhardware/include/hardware/audio.h当中的方法又是调用的 tinyalsa 的接口，关于 pcm 的操作：/* Open and close a stream */
struct pcm *pcm_open(unsigned int card, unsigned int device,
unsigned int flags, struct pcm_config *config);
int pcm_close(struct pcm *pcm);
int pcm_is_ready(struct pcm *pcm);/* Obtain the parameters for a PCM */
struct pcm_params *pcm_params_get(unsigned int card, unsigned int device,
unsigned int flags);
void pcm_params_free(struct pcm_params *pcm_params);
unsigned int pcm_params_get_min(struct pcm_params *pcm_params,
enum pcm_param param);
unsigned int pcm_params_get_max(struct pcm_params *pcm_params,
enum pcm_param param);/* Set and get config */
int pcm_get_config(struct pcm *pcm, struct pcm_config *config);
int pcm_set_config(struct pcm *pcm, struct pcm_config *config);/* Returns a human readable reason for the last error */
const char *pcm_get_error(struct pcm *pcm);/* Returns the sample size in bits for a PCM format.
* As with ALSA formats, this is the storage size for the format, whereas the
* format represents the number of significant bits. For example,
* PCM_FORMAT_S24_LE uses 32 bits of storage.
*/
unsigned int pcm_format_to_bits(enum pcm_format format);/* Returns the buffer size (int frames) that should be used for pcm_write. */
unsigned int pcm_get_buffer_size(struct pcm *pcm);
unsigned int pcm_frames_to_bytes(struct pcm *pcm, unsigned int frames);
unsigned int pcm_bytes_to_frames(struct pcm *pcm, unsigned int bytes);/* Returns the pcm latency in ms */
unsigned int pcm_get_latency(struct pcm *pcm);/* Returns available frames in pcm buffer and corresponding time stamp.
* The clock is CLOCK_MONOTONIC if flag PCM_MONOTONIC was specified in pcm_open,
* otherwise the clock is CLOCK_REALTIME.
* For an input stream, frames available are frames ready for the
* application to read.
* For an output stream, frames available are the number of empty frames available
* for the application to write.
*/
int pcm_get_htimestamp(struct pcm *pcm, unsigned int *avail,
struct timespec *tstamp);/* Write data to the fifo.
* Will start playback on the first write or on a write that
* occurs after a fifo underrun.
*/
int pcm_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_read(struct pcm *pcm, void *data, unsigned int count);/*
* mmap() support.
*/
int pcm_mmap_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_mmap_read(struct pcm *pcm, void *data, unsigned int count);
int pcm_mmap_begin(struct pcm *pcm, void **areas, unsigned int *offset,
unsigned int *frames);
int pcm_mmap_commit(struct pcm *pcm, unsigned int offset, unsigned int frames);/* Start and stop a PCM channel that doesn't transfer data */
int pcm_start(struct pcm *pcm);
int pcm_stop(struct pcm *pcm);/* Interrupt driven API */
int pcm_wait(struct pcm *pcm, int timeout);/* Change avail_min after the stream has been opened with no need to stop the stream.
* Only accepted if opened with PCM_MMAP and PCM_NOIRQ flags
*/
int pcm_set_avail_min(struct pcm *pcm, int avail_min);
值得一提的是 HAL 如今不包括 route 的操作， audio route 交给了 libaudioroute.so ， 它也是调用 tinyalsa 的接口，并包括一个 xml 解析器。解析 mixer_paths.xml 里面的 route 配置数据。这样系统就能够对 alsa 的 pcm 和 mixer 进行操作了，理论上应该能够放音了。使用 tinyalsa 提供的工具能够进行測试，当然无法測试 HAL 的接口。tinycap tinymix tinypcminfo tinyplaytinyplay 能够播放 wav 格式的纯音频数据。 tinymix 能够查看和配置音频路径：
root@sama5d3:/ # tinymix
Mixer name: 'wm8904 @ SAMA5D3EK'
Number of controls: 41
ctl type num name value
0 INT 1 EQ1 Volume 12
1 INT 1 EQ2 Volume 12
2 INT 1 EQ3 Volume 12
3 INT 1 EQ4 Volume 12
4 INT 1 EQ5 Volume 12
5 INT 2 Digital Capture Volume 96 96
6 ENUM 1 Left Caputure Mode Single-Ended
7 ENUM 1 Right Capture Mode Single-Ended
8 INT 2 Capture Volume 5 5
9 BOOL 2 Capture Switch Off Off
10 BOOL 1 High Pass Filter Switch On
11 ENUM 1 High Pass Filter Mode Hi-fi
12 BOOL 1 ADC 128x OSR Switch On
13 INT 1 Digital Playback Boost Volume 0
14 INT 2 Digital Playback Volume 96 96
15 INT 2 Headphone Volume 45 45
16 BOOL 2 Headphone Switch On On
17 BOOL 2 Headphone ZC Switch On On
18 INT 2 Line Output Volume 57 57
19 BOOL 2 Line Output Switch On On
20 BOOL 2 Line Output ZC Switch On On
21 BOOL 1 EQ Switch Off
22 BOOL 1 DRC Switch Off
23 ENUM 1 DRC Path ADC
24 BOOL 1 DAC OSRx2 Switch Off
25 BOOL 1 DAC Deemphasis Switch Off
26 INT 2 Digital Sidetone Volume 0 0
27 ENUM 1 LINER Mux DAC
28 ENUM 1 LINEL Mux DAC
29 ENUM 1 HPR Mux DAC
30 ENUM 1 HPL Mux DAC
31 ENUM 1 Right Sidetone None
32 ENUM 1 Left Sidetone None
33 ENUM 1 DACR Mux Right
34 ENUM 1 DACL Mux Left
35 ENUM 1 AIFOUTR Mux Right
36 ENUM 1 AIFOUTL Mux Left
37 ENUM 1 Right Capture Inverting Mux IN1R
38 ENUM 1 Right Capture Mux IN2R
39 ENUM 1 Left Capture Inverting Mux IN1L
40 ENUM 1 Left Capture Mux IN2L
audioflinger
audioflinger 是 audio 音频server。它会载入 audio hal ，并处理 audio 应用发出音频请求。这个分析的有非常多，參考下面：http://blog.csdn.net/DroidPhone/article/details/5941344
static inline int audio_hw_device_open(const struct hw_module_t* module,
struct audio_hw_device** device)
{
return module->methods->open(module, AUDIO_HARDWARE_INTERFACE,
(struct hw_device_t**)device);
}static inline int audio_hw_device_close(struct audio_hw_device* device)
{
return device->common.close(&device->common);
}
须要实现 in 和 out 的 数据流
/* Open and close a stream */
struct pcm *pcm_open(unsigned int card, unsigned int device,
unsigned int flags, struct pcm_config *config);
int pcm_close(struct pcm *pcm);
int pcm_is_ready(struct pcm *pcm);/* Obtain the parameters for a PCM */
struct pcm_params *pcm_params_get(unsigned int card, unsigned int device,
unsigned int flags);
void pcm_params_free(struct pcm_params *pcm_params);
unsigned int pcm_params_get_min(struct pcm_params *pcm_params,
enum pcm_param param);
unsigned int pcm_params_get_max(struct pcm_params *pcm_params,
enum pcm_param param);/* Set and get config */
int pcm_get_config(struct pcm *pcm, struct pcm_config *config);
int pcm_set_config(struct pcm *pcm, struct pcm_config *config);/* Returns a human readable reason for the last error */
const char *pcm_get_error(struct pcm *pcm);/* Returns the sample size in bits for a PCM format.
* As with ALSA formats, this is the storage size for the format, whereas the
* format represents the number of significant bits. For example,
* PCM_FORMAT_S24_LE uses 32 bits of storage.
*/
unsigned int pcm_format_to_bits(enum pcm_format format);/* Returns the buffer size (int frames) that should be used for pcm_write. */
unsigned int pcm_get_buffer_size(struct pcm *pcm);
unsigned int pcm_frames_to_bytes(struct pcm *pcm, unsigned int frames);
unsigned int pcm_bytes_to_frames(struct pcm *pcm, unsigned int bytes);/* Returns the pcm latency in ms */
unsigned int pcm_get_latency(struct pcm *pcm);/* Returns available frames in pcm buffer and corresponding time stamp.
* The clock is CLOCK_MONOTONIC if flag PCM_MONOTONIC was specified in pcm_open,
* otherwise the clock is CLOCK_REALTIME.
* For an input stream, frames available are frames ready for the
* application to read.
* For an output stream, frames available are the number of empty frames available
* for the application to write.
*/
int pcm_get_htimestamp(struct pcm *pcm, unsigned int *avail,
struct timespec *tstamp);/* Write data to the fifo.
* Will start playback on the first write or on a write that
* occurs after a fifo underrun.
*/
int pcm_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_read(struct pcm *pcm, void *data, unsigned int count);/*
* mmap() support.
*/
int pcm_mmap_write(struct pcm *pcm, const void *data, unsigned int count);
int pcm_mmap_read(struct pcm *pcm, void *data, unsigned int count);
int pcm_mmap_begin(struct pcm *pcm, void **areas, unsigned int *offset,
unsigned int *frames);
int pcm_mmap_commit(struct pcm *pcm, unsigned int offset, unsigned int frames);/* Start and stop a PCM channel that doesn't transfer data */
int pcm_start(struct pcm *pcm);
int pcm_stop(struct pcm *pcm);/* Interrupt driven API */
int pcm_wait(struct pcm *pcm, int timeout);/* Change avail_min after the stream has been opened with no need to stop the stream.
* Only accepted if opened with PCM_MMAP and PCM_NOIRQ flags
*/
int pcm_set_avail_min(struct pcm *pcm, int avail_min);
tinymix 能够查看和配置音频路径：
root@sama5d3:/ # tinymix
Mixer name: 'wm8904 @ SAMA5D3EK'
Number of controls: 41
ctl type num name value
0 INT 1 EQ1 Volume 12
1 INT 1 EQ2 Volume 12
2 INT 1 EQ3 Volume 12
3 INT 1 EQ4 Volume 12
4 INT 1 EQ5 Volume 12
5 INT 2 Digital Capture Volume 96 96
6 ENUM 1 Left Caputure Mode Single-Ended
7 ENUM 1 Right Capture Mode Single-Ended
8 INT 2 Capture Volume 5 5
9 BOOL 2 Capture Switch Off Off
10 BOOL 1 High Pass Filter Switch On
11 ENUM 1 High Pass Filter Mode Hi-fi
12 BOOL 1 ADC 128x OSR Switch On
13 INT 1 Digital Playback Boost Volume 0
14 INT 2 Digital Playback Volume 96 96
15 INT 2 Headphone Volume 45 45
16 BOOL 2 Headphone Switch On On
17 BOOL 2 Headphone ZC Switch On On
18 INT 2 Line Output Volume 57 57
19 BOOL 2 Line Output Switch On On
20 BOOL 2 Line Output ZC Switch On On
21 BOOL 1 EQ Switch Off
22 BOOL 1 DRC Switch Off
23 ENUM 1 DRC Path ADC
24 BOOL 1 DAC OSRx2 Switch Off
25 BOOL 1 DAC Deemphasis Switch Off
26 INT 2 Digital Sidetone Volume 0 0
27 ENUM 1 LINER Mux DAC
28 ENUM 1 LINEL Mux DAC
29 ENUM 1 HPR Mux DAC
30 ENUM 1 HPL Mux DAC
31 ENUM 1 Right Sidetone None
32 ENUM 1 Left Sidetone None
33 ENUM 1 DACR Mux Right
34 ENUM 1 DACL Mux Left
35 ENUM 1 AIFOUTR Mux Right
36 ENUM 1 AIFOUTL Mux Left
37 ENUM 1 Right Capture Inverting Mux IN1R
38 ENUM 1 Right Capture Mux IN2R
39 ENUM 1 Left Capture Inverting Mux IN1L
40 ENUM 1 Left Capture Mux IN2L
