PulseAudio中插件module-loopback的latency延迟的计算方法

1， module-loopback播放时 ，pulse audio基本的插件连接情况如下图所示
 
   
  
 
  
2，module-loopback中latency的计算如下：
 
  
     a: source中的latency计算：
 
  
 
  
 pa_asyncmsgq_send(u->source_output->source->asyncmsgq, PA_MSGOBJECT(u->source_output), SOURCE_OUTPUT_MESSAGE_LATENCY_SNAPSHOT, NULL, 0, NULL);
 
  
 
  
 
  
其中SOURCE_OUTPUT_MESSAGE_LATENCY_SNAPSHOT的处理逻辑又如下：
 
  

 
  
而pa_source_get_latency_within_thread调用process_msg(PA_SOURCE_MESSAGE_GET_LATENCY).  
 
  
作为例子，module-null-source的计算如下所示：
 
  

 
  
 b: sink中的latency计算（与source基本类似)
 
  
 
  
pa_asyncmsgq_send(u->sink_input->sink->asyncmsgq, PA_MSGOBJECT(u->sink_input), SINK_INPUT_MESSAGE_LATENCY_SNAPSHOT, NULL, 0, NULL);
 
  
 
  

 SINK_INPUT_MESSAGE_LATENCY_SNAPSHOT处理如下： 
  

 其中的pa_sink_get_latency_within_thread调用process_msg(PA_SINK_MESSAGE_GET_LATENCY). 作为例子，module-null-sink的PA_SINK_MESSAGE_GET_LATENCY处理如下所示：
 
  

 
  
c: loopback插件本身中的latency的计算，其本身又分为3部分:
 
  
一个是source output中队列的缓存：
 
  
 
  
length = pa_memblockq_get_length(u->source_output->thread_info.delay_memblockq);
 
  
 
  
 
  
一个是sink input总队列的缓存（注意：此计算方式包括了loopback中队列的缓存)：
 
  
 
  
  length = pa_memblockq_get_length(u->sink_input->thread_info.render_memblockq);

  u->latency_snapshot.sink_input_buffer =
                pa_memblockq_get_length(u->memblockq) +
                (u->sink_input->thread_info.resampler ? pa_resampler_request(u->sink_input->thread_info.resampler, length) : length);

再有一个是计算收发平衡的点： 
  
 
  
 
  
buffer =
        u->latency_snapshot.sink_input_buffer +
        u->latency_snapshot.source_output_buffer;

    if (u->latency_snapshot.recv_counter <= u->latency_snapshot.send_counter)
        buffer += (size_t) (u->latency_snapshot.send_counter - u->latency_snapshot.recv_counter);
    else
        buffer += PA_CLIP_SUB(buffer, (size_t) (u->latency_snapshot.recv_counter - u->latency_snapshot.send_counter));

    buffer_latency = pa_bytes_to_usec(buffer, &u->sink_input->sample_spec);
 
  

 
  
如上， 总共的latency的计算，就是将a, b , c三部分相加即可。