Revision control
Copy as Markdown
Other Tools
// Copyright © 2017-2018 Mozilla Foundation
//
// This program is made available under an ISC-style license. See the
// accompanying file LICENSE for details.
use backend::cork_state::CorkState;
use backend::*;
use cubeb_backend::{
ffi, log_enabled, ChannelLayout, DeviceId, DeviceRef, Error, InputProcessingParams, Result,
SampleFormat, StreamOps, StreamParamsRef, StreamPrefs,
};
use pulse::{self, CVolumeExt, ChannelMapExt, SampleSpecExt, StreamLatency, USecExt};
use pulse_ffi::*;
use ringbuf::RingBuffer;
use std::ffi::{CStr, CString};
use std::os::raw::{c_long, c_void};
use std::slice;
use std::sync::atomic::{AtomicPtr, AtomicUsize, Ordering};
use std::{mem, ptr};
use self::LinearInputBuffer::*;
use self::RingBufferConsumer::*;
use self::RingBufferProducer::*;
const PULSE_NO_GAIN: f32 = -1.0;
/// Iterator interface to `ChannelLayout`.
///
/// Iterates each channel in the set represented by `ChannelLayout`.
struct ChannelLayoutIter {
/// The layout set being iterated
layout: ChannelLayout,
/// The next flag to test
index: u8,
}
fn channel_layout_iter(layout: ChannelLayout) -> ChannelLayoutIter {
let index = 0;
ChannelLayoutIter { layout, index }
}
impl Iterator for ChannelLayoutIter {
type Item = ChannelLayout;
fn next(&mut self) -> Option<Self::Item> {
while !self.layout.is_empty() {
let test = Self::Item::from_bits_truncate(1 << self.index);
self.index += 1;
if self.layout.contains(test) {
self.layout.remove(test);
return Some(test);
}
}
None
}
}
fn cubeb_channel_to_pa_channel(channel: ffi::cubeb_channel) -> pa_channel_position_t {
match channel {
ffi::CHANNEL_FRONT_LEFT => PA_CHANNEL_POSITION_FRONT_LEFT,
ffi::CHANNEL_FRONT_RIGHT => PA_CHANNEL_POSITION_FRONT_RIGHT,
ffi::CHANNEL_FRONT_CENTER => PA_CHANNEL_POSITION_FRONT_CENTER,
ffi::CHANNEL_LOW_FREQUENCY => PA_CHANNEL_POSITION_LFE,
ffi::CHANNEL_BACK_LEFT => PA_CHANNEL_POSITION_REAR_LEFT,
ffi::CHANNEL_BACK_RIGHT => PA_CHANNEL_POSITION_REAR_RIGHT,
ffi::CHANNEL_FRONT_LEFT_OF_CENTER => PA_CHANNEL_POSITION_FRONT_LEFT_OF_CENTER,
ffi::CHANNEL_FRONT_RIGHT_OF_CENTER => PA_CHANNEL_POSITION_FRONT_RIGHT_OF_CENTER,
ffi::CHANNEL_BACK_CENTER => PA_CHANNEL_POSITION_REAR_CENTER,
ffi::CHANNEL_SIDE_LEFT => PA_CHANNEL_POSITION_SIDE_LEFT,
ffi::CHANNEL_SIDE_RIGHT => PA_CHANNEL_POSITION_SIDE_RIGHT,
ffi::CHANNEL_TOP_CENTER => PA_CHANNEL_POSITION_TOP_CENTER,
ffi::CHANNEL_TOP_FRONT_LEFT => PA_CHANNEL_POSITION_TOP_FRONT_LEFT,
ffi::CHANNEL_TOP_FRONT_CENTER => PA_CHANNEL_POSITION_TOP_FRONT_CENTER,
ffi::CHANNEL_TOP_FRONT_RIGHT => PA_CHANNEL_POSITION_TOP_FRONT_RIGHT,
ffi::CHANNEL_TOP_BACK_LEFT => PA_CHANNEL_POSITION_TOP_REAR_LEFT,
ffi::CHANNEL_TOP_BACK_CENTER => PA_CHANNEL_POSITION_TOP_REAR_CENTER,
ffi::CHANNEL_TOP_BACK_RIGHT => PA_CHANNEL_POSITION_TOP_REAR_RIGHT,
_ => PA_CHANNEL_POSITION_INVALID,
}
}
fn layout_to_channel_map(layout: ChannelLayout) -> pulse::ChannelMap {
assert_ne!(layout, ChannelLayout::UNDEFINED);
let mut cm = pulse::ChannelMap::init();
for (i, channel) in channel_layout_iter(layout).enumerate() {
cm.map[i] = cubeb_channel_to_pa_channel(channel.into());
}
cm.channels = layout.num_channels() as _;
// Special case single channel center mapping as mono.
if cm.channels == 1 && cm.map[0] == PA_CHANNEL_POSITION_FRONT_CENTER {
cm.map[0] = PA_CHANNEL_POSITION_MONO;
}
cm
}
fn default_layout_for_channels(ch: u32) -> ChannelLayout {
match ch {
1 => ChannelLayout::MONO,
2 => ChannelLayout::STEREO,
3 => ChannelLayout::_3F,
4 => ChannelLayout::QUAD,
5 => ChannelLayout::_3F2,
6 => ChannelLayout::_3F_LFE | ChannelLayout::SIDE_LEFT | ChannelLayout::SIDE_RIGHT,
7 => ChannelLayout::_3F3R_LFE,
8 => ChannelLayout::_3F4_LFE,
_ => panic!("channel must be between 1 to 8."),
}
}
pub struct Device(ffi::cubeb_device);
impl Drop for Device {
fn drop(&mut self) {
unsafe {
if !self.0.input_name.is_null() {
let _ = CString::from_raw(self.0.input_name as *mut _);
}
if !self.0.output_name.is_null() {
let _ = CString::from_raw(self.0.output_name as *mut _);
}
}
}
}
enum RingBufferConsumer {
IntegerRingBufferConsumer(ringbuf::Consumer<i16>),
FloatRingBufferConsumer(ringbuf::Consumer<f32>),
}
enum RingBufferProducer {
IntegerRingBufferProducer(ringbuf::Producer<i16>),
FloatRingBufferProducer(ringbuf::Producer<f32>),
}
enum LinearInputBuffer {
IntegerLinearInputBuffer(Vec<i16>),
FloatLinearInputBuffer(Vec<f32>),
}
struct BufferManager {
consumer: RingBufferConsumer,
producer: RingBufferProducer,
linear_input_buffer: LinearInputBuffer,
}
impl BufferManager {
// When opening a duplex stream, the sample-spec are guaranteed to match. It's ok to have
// either the input or output sample-spec here.
fn new(input_buffer_size: usize, sample_spec: &pulse::SampleSpec) -> BufferManager {
if sample_spec.format == PA_SAMPLE_S16BE || sample_spec.format == PA_SAMPLE_S16LE {
let ring = RingBuffer::<i16>::new(input_buffer_size);
let (prod, cons) = ring.split();
BufferManager {
producer: IntegerRingBufferProducer(prod),
consumer: IntegerRingBufferConsumer(cons),
linear_input_buffer: IntegerLinearInputBuffer(Vec::<i16>::with_capacity(
input_buffer_size,
)),
}
} else {
let ring = RingBuffer::<f32>::new(input_buffer_size);
let (prod, cons) = ring.split();
BufferManager {
producer: FloatRingBufferProducer(prod),
consumer: FloatRingBufferConsumer(cons),
linear_input_buffer: FloatLinearInputBuffer(Vec::<f32>::with_capacity(
input_buffer_size,
)),
}
}
}
fn push_input_data(&mut self, input_data: *const c_void, read_samples: usize) {
match &mut self.producer {
RingBufferProducer::FloatRingBufferProducer(p) => {
let input_data =
unsafe { slice::from_raw_parts::<f32>(input_data as *const f32, read_samples) };
// we don't do anything in particular if we can't push everything
p.push_slice(input_data);
}
RingBufferProducer::IntegerRingBufferProducer(p) => {
let input_data =
unsafe { slice::from_raw_parts::<i16>(input_data as *const i16, read_samples) };
p.push_slice(input_data);
}
}
}
fn pull_input_data(&mut self, input_data: *mut c_void, needed_samples: usize) {
match &mut self.consumer {
IntegerRingBufferConsumer(p) => {
let input: &mut [i16] = unsafe {
slice::from_raw_parts_mut::<i16>(input_data as *mut i16, needed_samples)
};
let read = p.pop_slice(input);
if read < needed_samples {
for i in 0..(needed_samples - read) {
input[read + i] = 0;
}
}
}
FloatRingBufferConsumer(p) => {
let input: &mut [f32] = unsafe {
slice::from_raw_parts_mut::<f32>(input_data as *mut f32, needed_samples)
};
let read = p.pop_slice(input);
if read < needed_samples {
for i in 0..(needed_samples - read) {
input[read + i] = 0.;
}
}
}
}
}
fn get_linear_input_data(&mut self, nsamples: usize) -> *const c_void {
let p = match &mut self.linear_input_buffer {
LinearInputBuffer::IntegerLinearInputBuffer(b) => {
b.resize(nsamples, 0);
b.as_mut_ptr() as *mut c_void
}
LinearInputBuffer::FloatLinearInputBuffer(b) => {
b.resize(nsamples, 0.);
b.as_mut_ptr() as *mut c_void
}
};
self.pull_input_data(p, nsamples);
p
}
pub fn trim(&mut self, final_size: usize) {
match &self.linear_input_buffer {
LinearInputBuffer::IntegerLinearInputBuffer(b) => {
let length = b.len();
assert!(final_size <= length);
let nframes_to_pop = length - final_size;
self.get_linear_input_data(nframes_to_pop);
}
LinearInputBuffer::FloatLinearInputBuffer(b) => {
let length = b.len();
assert!(final_size <= length);
let nframes_to_pop = length - final_size;
self.get_linear_input_data(nframes_to_pop);
}
}
}
pub fn available_samples(&mut self) -> usize {
match &self.linear_input_buffer {
LinearInputBuffer::IntegerLinearInputBuffer(b) => b.len(),
LinearInputBuffer::FloatLinearInputBuffer(b) => b.len(),
}
}
}
impl std::fmt::Debug for BufferManager {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(f, "")
}
}
#[repr(C)]
#[derive(Debug)]
pub struct PulseStream<'ctx> {
context: &'ctx PulseContext,
user_ptr: *mut c_void,
output_stream: Option<pulse::Stream>,
input_stream: Option<pulse::Stream>,
data_callback: ffi::cubeb_data_callback,
state_callback: ffi::cubeb_state_callback,
drain_timer: AtomicPtr<pa_time_event>,
output_sample_spec: pulse::SampleSpec,
input_sample_spec: pulse::SampleSpec,
// output frames count excluding pre-buffering
output_frame_count: AtomicUsize,
shutdown: bool,
volume: f32,
state: ffi::cubeb_state,
input_buffer_manager: Option<BufferManager>,
}
impl<'ctx> PulseStream<'ctx> {
#[allow(clippy::too_many_arguments)]
pub fn new(
context: &'ctx PulseContext,
stream_name: Option<&CStr>,
input_device: DeviceId,
input_stream_params: Option<&StreamParamsRef>,
output_device: DeviceId,
output_stream_params: Option<&StreamParamsRef>,
latency_frames: u32,
data_callback: ffi::cubeb_data_callback,
state_callback: ffi::cubeb_state_callback,
user_ptr: *mut c_void,
) -> Result<Box<Self>> {
fn check_error(s: &pulse::Stream, u: *mut c_void) {
let stm = unsafe { &mut *(u as *mut PulseStream) };
if !s.get_state().is_good() {
cubeb_alog!("Calling error callback");
stm.state_change_callback(ffi::CUBEB_STATE_ERROR);
}
stm.context.mainloop.signal();
}
fn read_data(s: &pulse::Stream, nbytes: usize, u: *mut c_void) {
fn read_from_input(
s: &pulse::Stream,
buffer: *mut *const c_void,
size: *mut usize,
) -> i32 {
let readable_size = s.readable_size().map(|s| s as i32).unwrap_or(-1);
if readable_size > 0 && unsafe { s.peek(buffer, size).is_err() } {
cubeb_logv!("Error while peeking the input stream");
return -1;
}
readable_size
}
cubeb_alogv!("Input callback buffer size {}", nbytes);
let stm = unsafe { &mut *(u as *mut PulseStream) };
if stm.shutdown {
return;
}
let mut read_data: *const c_void = ptr::null();
let mut read_size: usize = 0;
while read_from_input(s, &mut read_data, &mut read_size) > 0 {
/* read_data can be NULL in case of a hole. */
if !read_data.is_null() {
let in_frame_size = stm.input_sample_spec.frame_size();
let read_frames = read_size / in_frame_size;
let read_samples = read_size / stm.input_sample_spec.sample_size();
if stm.output_stream.is_some() {
// duplex stream: push the input data to the ring buffer.
stm.input_buffer_manager
.as_mut()
.unwrap()
.push_input_data(read_data, read_samples);
} else {
// input/capture only operation. Call callback directly
let got = unsafe {
stm.data_callback.unwrap()(
stm as *mut _ as *mut _,
stm.user_ptr,
read_data,
ptr::null_mut(),
read_frames as c_long,
)
};
if got < 0 || got as usize != read_frames {
let _ = s.cancel_write();
stm.shutdown = true;
if got < 0 {
unsafe {
stm.state_callback.unwrap()(
stm as *mut _ as *mut _,
stm.user_ptr,
ffi::CUBEB_STATE_ERROR,
);
}
}
break;
}
}
}
if read_size > 0 {
let _ = s.drop();
}
if stm.shutdown {
return;
}
}
}
fn write_data(_: &pulse::Stream, nbytes: usize, u: *mut c_void) {
cubeb_alogv!("Output callback to be written buffer size {}", nbytes);
let stm = unsafe { &mut *(u as *mut PulseStream) };
if stm.shutdown || stm.state != ffi::CUBEB_STATE_STARTED {
return;
}
let nframes = nbytes / stm.output_sample_spec.frame_size();
let first_callback = stm.output_frame_count.fetch_add(nframes, Ordering::SeqCst) == 0;
if stm.input_stream.is_some() {
let nsamples_input = nframes * stm.input_sample_spec.channels as usize;
let input_buffer_manager = stm.input_buffer_manager.as_mut().unwrap();
if first_callback {
let buffered_input_frames = input_buffer_manager.available_samples()
/ stm.input_sample_spec.channels as usize;
if buffered_input_frames > nframes {
// Trim the buffer to ensure minimal roundtrip latency
let popped_frames = buffered_input_frames - nframes;
input_buffer_manager
.trim(nframes * stm.input_sample_spec.channels as usize);
cubeb_alog!("Dropping {} frames in input buffer.", popped_frames);
}
}
let p = input_buffer_manager.get_linear_input_data(nsamples_input);
stm.trigger_user_callback(p, nbytes);
} else {
// Output/playback only operation.
// Write directly to output
debug_assert!(stm.output_stream.is_some());
stm.trigger_user_callback(ptr::null(), nbytes);
}
}
let mut stm = Box::new(PulseStream {
context,
output_stream: None,
input_stream: None,
data_callback,
state_callback,
user_ptr,
drain_timer: AtomicPtr::new(ptr::null_mut()),
output_sample_spec: pulse::SampleSpec::default(),
input_sample_spec: pulse::SampleSpec::default(),
output_frame_count: AtomicUsize::new(0),
shutdown: false,
volume: PULSE_NO_GAIN,
state: ffi::CUBEB_STATE_ERROR,
input_buffer_manager: None,
});
if let Some(ref context) = stm.context.context {
stm.context.mainloop.lock();
// Setup output stream
if let Some(stream_params) = output_stream_params {
match PulseStream::stream_init(context, stream_params, stream_name) {
Ok(s) => {
stm.output_sample_spec = *s.get_sample_spec();
s.set_state_callback(check_error, stm.as_mut() as *mut _ as *mut _);
s.set_write_callback(write_data, stm.as_mut() as *mut _ as *mut _);
let buffer_size_bytes =
latency_frames * stm.output_sample_spec.frame_size() as u32;
let battr = pa_buffer_attr {
maxlength: u32::MAX,
prebuf: u32::MAX,
fragsize: u32::MAX,
tlength: buffer_size_bytes * 2,
minreq: buffer_size_bytes / 4,
};
let device_name = super::try_cstr_from(output_device as *const _);
let mut stream_flags = pulse::StreamFlags::AUTO_TIMING_UPDATE
| pulse::StreamFlags::INTERPOLATE_TIMING
| pulse::StreamFlags::START_CORKED
| pulse::StreamFlags::ADJUST_LATENCY;
if device_name.is_some()
|| stream_params
.prefs()
.contains(StreamPrefs::DISABLE_DEVICE_SWITCHING)
{
stream_flags |= pulse::StreamFlags::DONT_MOVE;
}
let _ = s.connect_playback(device_name, &battr, stream_flags, None, None);
stm.output_stream = Some(s);
}
Err(e) => {
cubeb_log!("Output stream initialization error");
stm.context.mainloop.unlock();
stm.destroy();
return Err(e);
}
}
}
// Set up input stream
if let Some(stream_params) = input_stream_params {
match PulseStream::stream_init(context, stream_params, stream_name) {
Ok(s) => {
stm.input_sample_spec = *s.get_sample_spec();
s.set_state_callback(check_error, stm.as_mut() as *mut _ as *mut _);
s.set_read_callback(read_data, stm.as_mut() as *mut _ as *mut _);
let buffer_size_bytes =
latency_frames * stm.input_sample_spec.frame_size() as u32;
let battr = pa_buffer_attr {
maxlength: u32::MAX,
prebuf: u32::MAX,
fragsize: buffer_size_bytes,
tlength: buffer_size_bytes,
minreq: buffer_size_bytes,
};
let device_name = super::try_cstr_from(input_device as *const _);
let mut stream_flags = pulse::StreamFlags::AUTO_TIMING_UPDATE
| pulse::StreamFlags::INTERPOLATE_TIMING
| pulse::StreamFlags::START_CORKED
| pulse::StreamFlags::ADJUST_LATENCY;
if device_name.is_some()
|| stream_params
.prefs()
.contains(StreamPrefs::DISABLE_DEVICE_SWITCHING)
{
stream_flags |= pulse::StreamFlags::DONT_MOVE;
}
let _ = s.connect_record(device_name, &battr, stream_flags);
stm.input_stream = Some(s);
}
Err(e) => {
cubeb_log!("Input stream initialization error");
stm.context.mainloop.unlock();
stm.destroy();
return Err(e);
}
}
}
// Duplex, set up the ringbuffer
if input_stream_params.is_some() && output_stream_params.is_some() {
// A bit more room in case of output underrun.
let buffer_size_bytes =
2 * latency_frames * stm.input_sample_spec.frame_size() as u32;
stm.input_buffer_manager = Some(BufferManager::new(
buffer_size_bytes as usize,
&stm.input_sample_spec,
))
}
let r = if stm.wait_until_ready() {
/* force a timing update now, otherwise timing info does not become valid
until some point after initialization has completed. */
stm.update_timing_info()
} else {
false
};
stm.context.mainloop.unlock();
if !r {
stm.destroy();
cubeb_log!("Error while waiting for the stream to be ready");
return Err(Error::error());
}
// TODO:
if log_enabled() {
if let Some(ref output_stream) = stm.output_stream {
let output_att = output_stream.get_buffer_attr();
cubeb_log!(
"Output buffer attributes maxlength {}, tlength {}, \
prebuf {}, minreq {}, fragsize {}",
output_att.maxlength,
output_att.tlength,
output_att.prebuf,
output_att.minreq,
output_att.fragsize
);
}
if let Some(ref input_stream) = stm.input_stream {
let input_att = input_stream.get_buffer_attr();
cubeb_log!(
"Input buffer attributes maxlength {}, tlength {}, \
prebuf {}, minreq {}, fragsize {}",
input_att.maxlength,
input_att.tlength,
input_att.prebuf,
input_att.minreq,
input_att.fragsize
);
}
}
}
Ok(stm)
}
fn destroy(&mut self) {
self.cork(CorkState::cork());
self.context.mainloop.lock();
{
if let Some(stm) = self.output_stream.take() {
let drain_timer = self.drain_timer.load(Ordering::Acquire);
if !drain_timer.is_null() {
/* there's no pa_rttime_free, so use this instead. */
self.context.mainloop.get_api().time_free(drain_timer);
}
stm.clear_state_callback();
stm.clear_write_callback();
let _ = stm.disconnect();
stm.unref();
}
if let Some(stm) = self.input_stream.take() {
stm.clear_state_callback();
stm.clear_read_callback();
let _ = stm.disconnect();
stm.unref();
}
}
self.context.mainloop.unlock();
}
}
impl<'ctx> Drop for PulseStream<'ctx> {
fn drop(&mut self) {
self.destroy();
}
}
impl<'ctx> StreamOps for PulseStream<'ctx> {
fn start(&mut self) -> Result<()> {
fn output_preroll(_: &pulse::MainloopApi, u: *mut c_void) {
let stm = unsafe { &mut *(u as *mut PulseStream) };
if !stm.shutdown {
let size = stm
.output_stream
.as_ref()
.map_or(0, |s| s.writable_size().unwrap_or(0));
stm.trigger_user_callback(std::ptr::null(), size);
}
}
self.shutdown = false;
self.cork(CorkState::uncork() | CorkState::notify());
if self.output_stream.is_some() {
/* When doing output-only or duplex, we need to manually call user cb once in order to
* make things roll. This is done via a defer event in order to execute it from PA
* server thread. */
self.context.mainloop.lock();
self.context
.mainloop
.get_api()
.once(output_preroll, self as *const _ as *mut _);
self.context.mainloop.unlock();
}
Ok(())
}
fn stop(&mut self) -> Result<()> {
{
self.context.mainloop.lock();
self.shutdown = true;
// If draining is taking place wait to finish
cubeb_log!("Stream stop: waiting for drain");
while !self.drain_timer.load(Ordering::Acquire).is_null() {
self.context.mainloop.wait();
}
cubeb_log!("Stream stop: waited for drain");
self.context.mainloop.unlock();
}
self.cork(CorkState::cork() | CorkState::notify());
Ok(())
}
fn position(&mut self) -> Result<u64> {
let in_thread = self.context.mainloop.in_thread();
if !in_thread {
self.context.mainloop.lock();
}
if self.output_stream.is_none() {
cubeb_log!("Calling position() on an input-only stream");
return Err(Error::error());
}
let stm = self.output_stream.as_ref().unwrap();
let r = match stm.get_time() {
Ok(r_usec) => {
let bytes = USecExt::to_bytes(r_usec, &self.output_sample_spec);
Ok((bytes / self.output_sample_spec.frame_size()) as u64)
}
Err(_) => {
cubeb_log!("Error: stm.get_time failed");
Err(Error::error())
}
};
if !in_thread {
self.context.mainloop.unlock();
}
r
}
fn latency(&mut self) -> Result<u32> {
match self.output_stream {
None => {
cubeb_log!("Error: calling latency() on an input-only stream");
Err(Error::error())
}
Some(ref stm) => match stm.get_latency() {
Ok(StreamLatency::Positive(r_usec)) => {
let latency = (r_usec * pa_usec_t::from(self.output_sample_spec.rate)
/ PA_USEC_PER_SEC) as u32;
Ok(latency)
}
Ok(_) => {
panic!("Can not handle negative latency values.");
}
Err(_) => {
cubeb_log!("Error: get_latency() failed for an output stream");
Err(Error::error())
}
},
}
}
fn input_latency(&mut self) -> Result<u32> {
match self.input_stream {
None => {
cubeb_log!("Error: calling input_latency() on an output-only stream");
Err(Error::error())
}
Some(ref stm) => match stm.get_latency() {
Ok(StreamLatency::Positive(w_usec)) => {
let latency = (w_usec * pa_usec_t::from(self.input_sample_spec.rate)
/ PA_USEC_PER_SEC) as u32;
Ok(latency)
}
// Input stream can be negative only if it is attached to a
// monitor source device
Ok(StreamLatency::Negative(_)) => Ok(0),
Err(_) => {
cubeb_log!("Error: stm.get_latency() failed for an input stream");
Err(Error::error())
}
},
}
}
fn set_volume(&mut self, volume: f32) -> Result<()> {
match self.output_stream {
None => {
cubeb_log!("Error: can't set volume on an input-only stream");
Err(Error::error())
}
Some(ref stm) => {
if let Some(ref context) = self.context.context {
self.context.mainloop.lock();
let mut cvol: pa_cvolume = Default::default();
/* if the pulse daemon is configured to use flat
* volumes, apply our own gain instead of changing
* the input volume on the sink. */
let flags = {
match self.context.default_sink_info {
Some(ref info) => info.flags,
_ => pulse::SinkFlags::empty(),
}
};
if flags.contains(pulse::SinkFlags::FLAT_VOLUME) {
self.volume = volume;
} else {
let channels = stm.get_sample_spec().channels;
let vol = pulse::sw_volume_from_linear(f64::from(volume));
cvol.set(u32::from(channels), vol);
let index = stm.get_index();
let context_ptr = self.context as *const _ as *mut _;
if let Ok(o) = context.set_sink_input_volume(
index,
&cvol,
context_success,
context_ptr,
) {
self.context.operation_wait(stm, &o);
}
}
self.context.mainloop.unlock();
Ok(())
} else {
cubeb_log!("Error: set_volume: no context?");
Err(Error::error())
}
}
}
}
fn set_name(&mut self, name: &CStr) -> Result<()> {
match self.output_stream {
None => {
cubeb_log!("Error: can't set the name on a input-only stream.");
Err(Error::error())
}
Some(ref stm) => {
self.context.mainloop.lock();
if let Ok(o) = stm.set_name(name, stream_success, self as *const _ as *mut _) {
self.context.operation_wait(stm, &o);
}
self.context.mainloop.unlock();
Ok(())
}
}
}
fn current_device(&mut self) -> Result<&DeviceRef> {
if self.context.version_0_9_8 {
let mut dev: Box<ffi::cubeb_device> = Box::new(unsafe { mem::zeroed() });
if let Some(ref stm) = self.input_stream {
dev.input_name = match stm.get_device_name() {
Ok(name) => name.to_owned().into_raw(),
Err(_) => {
cubeb_log!("Error: couldn't get the input stream's device name");
return Err(Error::error());
}
}
}
if let Some(ref stm) = self.output_stream {
dev.output_name = match stm.get_device_name() {
Ok(name) => name.to_owned().into_raw(),
Err(_) => {
cubeb_log!("Error: couldn't get the output stream's device name");
return Err(Error::error());
}
}
}
Ok(unsafe { DeviceRef::from_ptr(Box::into_raw(dev) as *mut _) })
} else {
cubeb_log!("Error: PulseAudio context too old");
Err(not_supported())
}
}
fn set_input_mute(&mut self, _mute: bool) -> Result<()> {
Err(not_supported())
}
fn set_input_processing_params(&mut self, _params: InputProcessingParams) -> Result<()> {
Err(not_supported())
}
fn device_destroy(&mut self, device: &DeviceRef) -> Result<()> {
if device.as_ptr().is_null() {
cubeb_log!("Error: can't destroy null device");
Err(Error::error())
} else {
unsafe {
let _: Box<Device> = Box::from_raw(device.as_ptr() as *mut _);
}
Ok(())
}
}
fn register_device_changed_callback(
&mut self,
_: ffi::cubeb_device_changed_callback,
) -> Result<()> {
cubeb_log!("Error: register_device_change_callback unimplemented");
Err(Error::error())
}
}
impl<'ctx> PulseStream<'ctx> {
fn stream_init(
context: &pulse::Context,
stream_params: &StreamParamsRef,
stream_name: Option<&CStr>,
) -> Result<pulse::Stream> {
if stream_params.prefs() == StreamPrefs::LOOPBACK {
cubeb_log!("Error: StreamPref::LOOPBACK unimplemented");
return Err(not_supported());
}
fn to_pulse_format(format: SampleFormat) -> pulse::SampleFormat {
match format {
SampleFormat::S16LE => pulse::SampleFormat::Signed16LE,
SampleFormat::S16BE => pulse::SampleFormat::Signed16BE,
SampleFormat::Float32LE => pulse::SampleFormat::Float32LE,
SampleFormat::Float32BE => pulse::SampleFormat::Float32BE,
_ => pulse::SampleFormat::Invalid,
}
}
let fmt = to_pulse_format(stream_params.format());
if fmt == pulse::SampleFormat::Invalid {
cubeb_log!("Error: invalid sample format");
return Err(invalid_format());
}
let ss = pulse::SampleSpec {
channels: stream_params.channels() as u8,
format: fmt.into(),
rate: stream_params.rate(),
};
let cm: Option<pa_channel_map> = match stream_params.layout() {
ChannelLayout::UNDEFINED => {
if stream_params.channels() <= 8
&& pulse::ChannelMap::init_auto(
stream_params.channels(),
PA_CHANNEL_MAP_DEFAULT,
)
.is_none()
{
cubeb_log!("Layout undefined and PulseAudio's default layout has not been configured, guess one.");
Some(layout_to_channel_map(default_layout_for_channels(
stream_params.channels(),
)))
} else {
cubeb_log!("Layout undefined, PulseAudio will use its default.");
None
}
}
_ => Some(layout_to_channel_map(stream_params.layout())),
};
let stream = pulse::Stream::new(context, stream_name.unwrap(), &ss, cm.as_ref());
match stream {
None => {
cubeb_log!("Error: pulse::Stream::new failure");
Err(Error::error())
}
Some(stm) => Ok(stm),
}
}
pub fn cork_stream(&self, stream: Option<&pulse::Stream>, state: CorkState) {
if let Some(stm) = stream {
if let Ok(o) = stm.cork(
state.is_cork() as i32,
stream_success,
self as *const _ as *mut _,
) {
self.context.operation_wait(stream, &o);
}
}
}
fn cork(&mut self, state: CorkState) {
{
self.context.mainloop.lock();
self.cork_stream(self.output_stream.as_ref(), state);
self.cork_stream(self.input_stream.as_ref(), state);
self.context.mainloop.unlock()
}
if state.is_notify() {
self.state_change_callback(if state.is_cork() {
ffi::CUBEB_STATE_STOPPED
} else {
ffi::CUBEB_STATE_STARTED
});
}
}
fn update_timing_info(&self) -> bool {
let mut r = false;
if let Some(ref stm) = self.output_stream {
if let Ok(o) = stm.update_timing_info(stream_success, self as *const _ as *mut _) {
r = self.context.operation_wait(stm, &o);
}
if !r {
return r;
}
}
if let Some(ref stm) = self.input_stream {
if let Ok(o) = stm.update_timing_info(stream_success, self as *const _ as *mut _) {
r = self.context.operation_wait(stm, &o);
}
}
r
}
pub fn state_change_callback(&mut self, s: ffi::cubeb_state) {
self.state = s;
unsafe {
(self.state_callback.unwrap())(
self as *mut PulseStream as *mut ffi::cubeb_stream,
self.user_ptr,
s,
)
};
}
fn wait_until_ready(&self) -> bool {
fn wait_until_io_stream_ready(
stm: &pulse::Stream,
mainloop: &pulse::ThreadedMainloop,
) -> bool {
if mainloop.is_null() {
return false;
}
loop {
let state = stm.get_state();
if !state.is_good() {
return false;
}
if state == pulse::StreamState::Ready {
break;
}
mainloop.wait();
}
true
}
if let Some(ref stm) = self.output_stream {
if !wait_until_io_stream_ready(stm, &self.context.mainloop) {
return false;
}
}
if let Some(ref stm) = self.input_stream {
if !wait_until_io_stream_ready(stm, &self.context.mainloop) {
return false;
}
}
true
}
#[allow(clippy::cognitive_complexity)]
fn trigger_user_callback(&mut self, input_data: *const c_void, nbytes: usize) {
fn drained_cb(
a: &pulse::MainloopApi,
e: *mut pa_time_event,
_tv: &pulse::TimeVal,
u: *mut c_void,
) {
cubeb_logv!("Drain finished callback.");
let stm = unsafe { &mut *(u as *mut PulseStream) };
let drain_timer = stm.drain_timer.load(Ordering::Acquire);
debug_assert_eq!(drain_timer, e);
stm.state_change_callback(ffi::CUBEB_STATE_DRAINED);
/* there's no pa_rttime_free, so use this instead. */
a.time_free(drain_timer);
stm.drain_timer.store(ptr::null_mut(), Ordering::Release);
stm.context.mainloop.signal();
}
if let Some(ref stm) = self.output_stream {
let frame_size = self.output_sample_spec.frame_size();
debug_assert_eq!(nbytes % frame_size, 0);
let mut towrite = nbytes;
let mut read_offset = 0usize;
while towrite > 0 {
match stm.begin_write(towrite) {
Err(e) => {
cubeb_logv!("Error: failure to write data");
panic!("Failed to write data: {}", e);
}
Ok((buffer, size)) => {
debug_assert!(size > 0);
debug_assert_eq!(size % frame_size, 0);
cubeb_logv!(
"Trigger user callback with output buffer size={}, read_offset={}",
size,
read_offset
);
let read_ptr = unsafe { (input_data as *const u8).add(read_offset) };
#[allow(clippy::unnecessary_cast)]
let mut got = unsafe {
self.data_callback.unwrap()(
self as *const _ as *mut _,
self.user_ptr,
read_ptr as *const _ as *mut _,
buffer,
(size / frame_size) as c_long,
) as i64
};
if got < 0 {
let _ = stm.cancel_write();
self.shutdown = true;
unsafe {
self.state_callback.unwrap()(
self as *const _ as *mut _,
self.user_ptr,
ffi::CUBEB_STATE_ERROR,
);
}
return;
}
// If more iterations move offset of read buffer
if !input_data.is_null() {
let in_frame_size = self.input_sample_spec.frame_size();
read_offset += (size / frame_size) * in_frame_size;
}
if self.volume != PULSE_NO_GAIN {
let samples = (self.output_sample_spec.channels as usize * size
/ frame_size) as isize;
if self.output_sample_spec.format == PA_SAMPLE_S16BE
|| self.output_sample_spec.format == PA_SAMPLE_S16LE
{
let b = buffer as *mut i16;
for i in 0..samples {
unsafe { *b.offset(i) *= self.volume as i16 };
}
} else {
let b = buffer as *mut f32;
for i in 0..samples {
unsafe { *b.offset(i) *= self.volume };
}
}
}
let should_drain = (got as usize) < size / frame_size;
if should_drain && self.output_frame_count.load(Ordering::SeqCst) == 0 {
// Draining during preroll, ensure `prebuf` frames are written so
// the stream starts. If not, pad with a bit of silence.
let prebuf_size_bytes = stm.get_buffer_attr().prebuf as usize;
let got_bytes = got as usize * frame_size;
if prebuf_size_bytes > got_bytes {
let padding_bytes = prebuf_size_bytes - got_bytes;
if padding_bytes + got_bytes <= size {
// A slice that starts after the data provided by the callback,
// with just enough room to provide a final buffer big enough.
let padding_buf: &mut [u8] = unsafe {
slice::from_raw_parts_mut::<u8>(
buffer.add(got_bytes) as *mut u8,
padding_bytes,
)
};
padding_buf.fill(0);
got += (padding_bytes / frame_size) as i64;
}
} else {
cubeb_logv!(
"Not enough room to pad up to prebuf when prebuffering."
)
}
}
let r = stm.write(
buffer,
got as usize * frame_size,
0,
pulse::SeekMode::Relative,
);
if should_drain {
cubeb_logv!("Draining {} < {}", got, size / frame_size);
let latency = match stm.get_latency() {
Ok(StreamLatency::Positive(l)) => l,
Ok(_) => {
panic!("Can not handle negative latency values.");
}
Err(e) => {
debug_assert_eq!(
e,
pulse::ErrorCode::from_error_code(PA_ERR_NODATA)
);
/* this needs a better guess. */
100 * PA_USEC_PER_MSEC
}
};
/* pa_stream_drain is useless, see PA bug# 866. this is a workaround. */
/* arbitrary safety margin: double the current latency. */
debug_assert!(self.drain_timer.load(Ordering::Acquire).is_null());
let stream_ptr = self as *const _ as *mut _;
if let Some(ref context) = self.context.context {
self.drain_timer.store(
context.rttime_new(
pulse::rtclock_now() + 2 * latency,
drained_cb,
stream_ptr,
),
Ordering::Release,
);
}
self.shutdown = true;
return;
}
debug_assert!(r.is_ok());
towrite -= size;
}
}
}
debug_assert_eq!(towrite, 0);
}
}
}
fn stream_success(_: &pulse::Stream, success: i32, u: *mut c_void) {
let stm = unsafe { &*(u as *mut PulseStream) };
if success != 1 {
cubeb_log!("stream_success ignored failure: {}", success);
}
stm.context.mainloop.signal();
}
fn context_success(_: &pulse::Context, success: i32, u: *mut c_void) {
let ctx = unsafe { &*(u as *mut PulseContext) };
if success != 1 {
cubeb_log!("context_success ignored failure: {}", success);
}
ctx.mainloop.signal();
}
fn invalid_format() -> Error {
Error::from_raw(ffi::CUBEB_ERROR_INVALID_FORMAT)
}
fn not_supported() -> Error {
Error::from_raw(ffi::CUBEB_ERROR_NOT_SUPPORTED)
}
#[cfg(all(test, not(feature = "pulse-dlopen")))]
mod test {
use super::layout_to_channel_map;
use cubeb_backend::ChannelLayout;
use pulse_ffi::*;
macro_rules! channel_tests {
{$($name: ident, $layout: ident => [ $($channels: ident),* ]),+} => {
$(
#[test]
fn $name() {
let layout = ChannelLayout::$layout;
let mut iter = super::channel_layout_iter(layout);
$(
assert_eq!(Some(ChannelLayout::$channels), iter.next());
)*
assert_eq!(None, iter.next());
}
)*
}
}
channel_tests! {
channels_unknown, UNDEFINED => [ ],
channels_mono, MONO => [
FRONT_CENTER
],
channels_mono_lfe, MONO_LFE => [
FRONT_CENTER,
LOW_FREQUENCY
],
channels_stereo, STEREO => [
FRONT_LEFT,
FRONT_RIGHT
],
channels_stereo_lfe, STEREO_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
LOW_FREQUENCY
],
channels_3f, _3F => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER
],
channels_3f_lfe, _3F_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY
],
channels_2f1, _2F1 => [
FRONT_LEFT,
FRONT_RIGHT,
BACK_CENTER
],
channels_2f1_lfe, _2F1_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
LOW_FREQUENCY,
BACK_CENTER
],
channels_3f1, _3F1 => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
BACK_CENTER
],
channels_3f1_lfe, _3F1_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY,
BACK_CENTER
],
channels_2f2, _2F2 => [
FRONT_LEFT,
FRONT_RIGHT,
SIDE_LEFT,
SIDE_RIGHT
],
channels_2f2_lfe, _2F2_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
LOW_FREQUENCY,
SIDE_LEFT,
SIDE_RIGHT
],
channels_quad, QUAD => [
FRONT_LEFT,
FRONT_RIGHT,
BACK_LEFT,
BACK_RIGHT
],
channels_quad_lfe, QUAD_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
LOW_FREQUENCY,
BACK_LEFT,
BACK_RIGHT
],
channels_3f2, _3F2 => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
SIDE_LEFT,
SIDE_RIGHT
],
channels_3f2_lfe, _3F2_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY,
SIDE_LEFT,
SIDE_RIGHT
],
channels_3f2_back, _3F2_BACK => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
BACK_LEFT,
BACK_RIGHT
],
channels_3f2_lfe_back, _3F2_LFE_BACK => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY,
BACK_LEFT,
BACK_RIGHT
],
channels_3f3r_lfe, _3F3R_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY,
BACK_CENTER,
SIDE_LEFT,
SIDE_RIGHT
],
channels_3f4_lfe, _3F4_LFE => [
FRONT_LEFT,
FRONT_RIGHT,
FRONT_CENTER,
LOW_FREQUENCY,
BACK_LEFT,
BACK_RIGHT,
SIDE_LEFT,
SIDE_RIGHT
]
}
#[test]
fn mono_channels_enumerate() {
let layout = ChannelLayout::MONO;
let mut iter = super::channel_layout_iter(layout).enumerate();
assert_eq!(Some((0, ChannelLayout::FRONT_CENTER)), iter.next());
assert_eq!(None, iter.next());
}
#[test]
fn stereo_channels_enumerate() {
let layout = ChannelLayout::STEREO;
let mut iter = super::channel_layout_iter(layout).enumerate();
assert_eq!(Some((0, ChannelLayout::FRONT_LEFT)), iter.next());
assert_eq!(Some((1, ChannelLayout::FRONT_RIGHT)), iter.next());
assert_eq!(None, iter.next());
}
#[test]
fn quad_channels_enumerate() {
let layout = ChannelLayout::QUAD;
let mut iter = super::channel_layout_iter(layout).enumerate();
assert_eq!(Some((0, ChannelLayout::FRONT_LEFT)), iter.next());
assert_eq!(Some((1, ChannelLayout::FRONT_RIGHT)), iter.next());
assert_eq!(Some((2, ChannelLayout::BACK_LEFT)), iter.next());
assert_eq!(Some((3, ChannelLayout::BACK_RIGHT)), iter.next());
assert_eq!(None, iter.next());
}
macro_rules! map_channel_tests {
{$($name: ident, $layout: ident => [ $($channels: ident),* ]),+} => {
$(
#[test]
fn $name() {
let map = layout_to_channel_map(ChannelLayout::$layout);
assert_eq!(map.channels, map_channel_tests!(__COUNT__ $($channels)*));
map_channel_tests!(__EACH__ map, 0, $($channels)*);
}
)*
};
(__COUNT__) => (0u8);
(__COUNT__ $x:ident $($xs: ident)*) => (1u8 + map_channel_tests!(__COUNT__ $($xs)*));
(__EACH__ $map:expr, $i:expr, ) => {};
(__EACH__ $map:expr, $i:expr, $x:ident $($xs: ident)*) => {
assert_eq!($map.map[$i], $x);
map_channel_tests!(__EACH__ $map, $i+1, $($xs)* );
};
}
map_channel_tests! {
map_channel_mono, MONO => [
PA_CHANNEL_POSITION_MONO
],
map_channel_mono_lfe, MONO_LFE => [
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE
],
map_channel_stereo, STEREO => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT
],
map_channel_stereo_lfe, STEREO_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_LFE
],
map_channel_3f, _3F => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER
],
map_channel_3f_lfe, _3F_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE
],
map_channel_2f1, _2F1 => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_REAR_CENTER
],
map_channel_2f1_lfe, _2F1_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_CENTER
],
map_channel_3f1, _3F1 => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_REAR_CENTER
],
map_channel_3f1_lfe, _3F1_LFE =>[
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_CENTER
],
map_channel_2f2, _2F2 => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
],
map_channel_2f2_lfe, _2F2_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
],
map_channel_quad, QUAD => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_REAR_LEFT,
PA_CHANNEL_POSITION_REAR_RIGHT
],
map_channel_quad_lfe, QUAD_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_LEFT,
PA_CHANNEL_POSITION_REAR_RIGHT
],
map_channel_3f2, _3F2 => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
],
map_channel_3f2_lfe, _3F2_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
],
map_channel_3f2_back, _3F2_BACK => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_REAR_LEFT,
PA_CHANNEL_POSITION_REAR_RIGHT
],
map_channel_3f2_lfe_back, _3F2_LFE_BACK => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_LEFT,
PA_CHANNEL_POSITION_REAR_RIGHT
],
map_channel_3f3r_lfe, _3F3R_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_CENTER,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
],
map_channel_3f4_lfe, _3F4_LFE => [
PA_CHANNEL_POSITION_FRONT_LEFT,
PA_CHANNEL_POSITION_FRONT_RIGHT,
PA_CHANNEL_POSITION_FRONT_CENTER,
PA_CHANNEL_POSITION_LFE,
PA_CHANNEL_POSITION_REAR_LEFT,
PA_CHANNEL_POSITION_REAR_RIGHT,
PA_CHANNEL_POSITION_SIDE_LEFT,
PA_CHANNEL_POSITION_SIDE_RIGHT
]
}
}