首页 » 技术分享 » snd_pcm_ops

snd_pcm_ops

 

origin: https://blog.csdn.net/azloong/article/details/6146378

之前写过一个音频驱动CODEC分析,当时忽略了阐述最基本的概念。要了解一个东西,首先要明白它是什么它起到什么作用,然后才会更好对它的工作流程更好的分析。所以这里提一下:

CODEC :音频芯片的控制,比如静音、打开(关闭)ADC(DAC)、设置ADC(DAC)的增益、耳机模式的检测等操作。

I2S   :数字音频接口,用于CPU和Codec之间的数字音频流raw data的传输。每当有playback或record操作时,snd_soc_dai_ops.prepare()会被调用,启动I2S总线。

PCM   :我不知道为什么会取这个模块名,它其实是定义DMA操作的,用于将音频数据通过DMA传到I2S控制器的FIFO中。

音频数据流向:

  | copy_from_user |          | DMA |               | I2S/PCM/AC97 |
RAM--------------->dma buffer--------> I2S tx FIFO -----------------> CODEC -> SPK/Headset

PCM模块初始化

    struct snd_soc_platform s3c_soc_platform = {
           .name         = "s3c-pcm-audio",
           .pcm_ops      = &s3c_pcm_ops,
           .pcm_new      = s3c_pcm_new,
           .pcm_free     = s3c_pcm_free_dma_buffers,
           .suspend      = s3c_pcm_suspend,
           .resume       = s3c_pcm_resume,
    };

调用snd_soc_register_platform()向ALSA core注册一个snd_soc_platform结构体。

成员pcm_new需要调用dma_alloc_writecombine()给DMA分配一块write-combining的内存空间,并把这块缓冲区的相关信息保存到substream->dma_buffer中,相当于构造函数。pcm_free则相反。这些成员函数都还算简单,看看代码即可以理解其流程。

snd_pcm_ops

接着我们看一下snd_pcm_ops结构体,该结构体的操作函数集的实现是本模块的主体。

    struct snd_pcm_ops {
           int (*open)(struct snd_pcm_substream *substream);
           int (*close)(struct snd_pcm_substream *substream);
           int (*ioctl)(struct snd_pcm_substream * substream,
                       unsigned int cmd, void *arg);
           int (*hw_params)(struct snd_pcm_substream *substream,
                          struct snd_pcm_hw_params *params);
           int (*hw_free)(struct snd_pcm_substream *substream);
           int (*prepare)(struct snd_pcm_substream *substream);
           int (*trigger)(struct snd_pcm_substream *substream, int cmd);
           snd_pcm_uframes_t (*pointer)(struct snd_pcm_substream *substream);
           int (*copy)(struct snd_pcm_substream *substream, int channel,
                      snd_pcm_uframes_t pos,
                      void __user *buf, snd_pcm_uframes_t count);
           int (*silence)(struct snd_pcm_substream *substream, int channel,
                         snd_pcm_uframes_t pos, snd_pcm_uframes_t count);
           struct page *(*page)(struct snd_pcm_substream *substream,
                              unsigned long offset);
           int (*mmap)(struct snd_pcm_substream *substream, struct vm_area_struct *vma);
           int (*ack)(struct snd_pcm_substream *substream);
    };

我们主要实现open、close、hw_params、hw_free、prepare和trigger接口。

open

open函数为PCM模块设定支持的传输模式、数据格式、通道数、period等参数,并为playback/capture stream分配相应的DMA通道。其一般实现如下:

    static int s3c_pcm_open(struct snd_pcm_substream *substream)
    {
           struct snd_soc_pcm_runtime *rtd = substream->private_data;
           struct snd_soc_dai *cpu_dai = rtd->dai->cpu_dai;
           struct snd_pcm_runtime *runtime = substream->runtime;
           struct audio_stream_a *s = runtime->private_data;
           int ret;
     
           if (!cpu_dai->active) {
                  audio_dma_request(&s[0], audio_dma_callback); //为playback stream分配DMA
                  audio_dma_request(&s[1], audio_dma_callback); //为capture stream分配DMA
           }
           
           //设定runtime硬件参数
           snd_soc_set_runtime_hwparams(substream, &s3c_pcm_hardware);
     
           /* Ensure that buffer size is a multiple of period size */
           ret = snd_pcm_hw_constraint_integer(runtime,
                                 SNDRV_PCM_HW_PARAM_PERIODS);
     
           return ret;
    }

其中硬件参数要根据芯片的数据手册来定义,如:

    static const struct snd_pcm_hardware s3c_pcm_hardware = {
           .info            = SNDRV_PCM_INFO_INTERLEAVED |
                                    SNDRV_PCM_INFO_BLOCK_TRANSFER |
                                    SNDRV_PCM_INFO_MMAP |
                                    SNDRV_PCM_INFO_MMAP_VALID |
                                    SNDRV_PCM_INFO_PAUSE |
                                    SNDRV_PCM_INFO_RESUME,
           .formats         = SNDRV_PCM_FMTBIT_S16_LE |
                                    SNDRV_PCM_FMTBIT_U16_LE |
                                    SNDRV_PCM_FMTBIT_U8 |
                                    SNDRV_PCM_FMTBIT_S8,
           .channels_min     = 2,
           .channels_max     = 2,
           .buffer_bytes_max = 128*1024,
           .period_bytes_min = PAGE_SIZE,
           .period_bytes_max = PAGE_SIZE*2,
           .periods_min      = 2,
           .periods_max      = 128,
           .fifo_size        = 32,
    };

关于peroid的概念有这样的描述:The “period” is a term that corresponds to a fragment in the OSS world. The period defines the size at which a PCM interrupt is generated. peroid的概念很重要,建议去alsa官网找相关详细说明了解一下。

上层ALSA lib可以通过接口来获得这些参数的,如snd_pcm_hw_params_get_buffer_size_max()来取得buffer_bytes_max。

关于DMA的中断处理

另外留意open函数中的audio_dma_request(&s[0], audio_dma_callback);中的audio_dma_callback,这是dma的中断函数,这里以callback的形式存在,其实到dma的底层还是这样的形式:static irqreturn_t dma_irq_handler(int irq, void *dev_id),在DMA中断处理dma_irq_handler()中调用callback。这些跟具体硬件平台的DMA实现相关,如果没有类似的机制,那么还是要在pcm模块中实现这个中断。

    /*
     *  This is called when dma IRQ occurs at the end of each transmited block
     */
    static void audio_dma_callback(void *data)
    {
           struct audio_stream_a *s = data;    
     
           /*
            * If we are getting a callback for an active stream then we inform
            * the PCM middle layer we've finished a period
            */
           if (s->active)
                  snd_pcm_period_elapsed(s->stream);
     
           spin_lock(&s->dma_lock);
           if (s->periods > 0)
                  s->periods--;    
     
           audio_process_dma(s); //dma启动
           spin_unlock(&s->dma_lock);
    }

hw_params

hw_params函数为substream(每打开一个playback或capture,ALSA core均产生相应的一个substream)设定DMA的源(目的)地址,以及DMA缓冲区的大小。

    static int s3c_pcm_hw_params(struct snd_pcm_substream *substream,
                               struct snd_pcm_hw_params *params)
    {
           struct snd_pcm_runtime *runtime = substream->runtime;
           int err = 0;
     
           snd_pcm_set_runtime_buffer(substream, &substream->dma_buffer);
           runtime->dma_bytes = params_buffer_bytes(params);
           return err;
    }

hw_free是hw_params的相反操作,调用snd_pcm_set_runtime_buffer(substream, NULL)即可。

注:代码中的dma_buffer是DMA缓冲区,它通过4个字段定义:dma_area、dma_addr、dma_bytes和dma_private。其中dma_area是缓冲区逻辑地址,dma_addr是缓冲区的物理地址,dma_bytes是缓冲区的大小,dma_private是ALSA的DMA管理用到的。dma_buffer是在pcm_new()中初始化的;当然也可以把分配dma缓冲区的工作放到这部分来实现,但考虑到减少碎片,故还是在pcm_new中以最大size(即buffer_bytes_max)来分配。

prepare

当pcm“准备好了”调用该函数。在这里根据channels、buffer_bytes等来设定DMA传输参数,跟具体硬件平台相关。注:每次调用snd_pcm_prepare()的时候均会调用prepare函数。

trigger

当pcm开始、停止、暂停的时候都会调用trigger函数。

    static int s3c_pcm_trigger(struct snd_pcm_substream *substream, int cmd)
    {
           struct runtime_data *prtd = substream->runtime->private_data;
           int ret = 0;
     
           spin_lock(&prtd->lock);
     
           switch (cmd) {
           case SNDRV_PCM_TRIGGER_START:
           case SNDRV_PCM_TRIGGER_RESUME:
           case SNDRV_PCM_TRIGGER_PAUSE_RELEASE:
                  prtd->state |= ST_RUNNING;
                  dma_ctrl(prtd->params->channel, DMAOP_START); //DMA开启
                  break;
     
           case SNDRV_PCM_TRIGGER_STOP:
           case SNDRV_PCM_TRIGGER_SUSPEND:
           case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
                  prtd->state &= ~ST_RUNNING;
                  dma_ctrl(prtd->params->channel, DMAOP_STOP); //DMA停止
                  break;
     
           default:
                  ret = -EINVAL;
                  break;
           }
     
           spin_unlock(&prtd->lock);
     
           return ret;
    }

Trigger函数里面的操作应该是原子的,不要在调用这些操作时进入睡眠,trigger函数应尽量小,甚至仅仅是触发DMA。

pointer

static snd_pcm_uframes_t s3c_pcm_pointer(struct snd_pcm_substream *substream)

PCM中间层通过调用这个函数来获取缓冲区的位置。一般情况下,在中断函数中调用snd_pcm_period_elapsed()或在pcm中间层更新buffer的时候调用它。然后pcm中间层会更新指针位置和计算缓冲区可用空间,唤醒那些在等待的线程。这个函数也是原子的。

snd_pcm_runtime

我们会留意到ops各成员函数均需要取得一个snd_pcm_runtime结构体指针,这个指针可以通过substream->runtime来获得。snd_pcm_runtime是pcm运行时的信息。当打开一个pcm子流时,pcm运行时实例就会分配给这个子流。它拥有很多多种信息:hw_params和sw_params配置拷贝,缓冲区指针,mmap记录,自旋锁等。snd_pcm_runtime对于驱动程序操作集函数是只读的,仅pcm中间层可以改变或更新这些信息。

 
---------------------  
作者:sepnic  
来源:CSDN  
原文:https://blog.csdn.net/azloong/article/details/6146378  
版权声明:本文为博主原创文章,转载请附上博文链接!

转载自原文链接, 如需删除请联系管理员。

原文链接:snd_pcm_ops,转载请注明来源!

0