1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-14 09:24:57 +00:00
serenity/Userland/Services/AudioServer/Mixer.h
kleines Filmröllchen b4fbd30b70 AudioServer+Userland: Decouple client sample rates from device rate
This change was a long time in the making ever since we obtained sample
rate awareness in the system. Now, each client has its own sample rate,
accessible via new IPC APIs, and the device sample rate is only
accessible via the management interface. AudioServer takes care of
resampling client streams into the device sample rate. Therefore, the
main improvement introduced with this commit is full responsiveness to
sample rate changes; all open audio programs will continue to play at
correct speed with the audio resampled to the new device rate.

The immediate benefits are manifold:
- Gets rid of the legacy hardware sample rate IPC message in the
  non-managing client
- Removes duplicate resampling and sample index rescaling code
  everywhere
- Avoids potential sample index scaling bugs in SoundPlayer (which have
  happened many times before) and fixes a sample index scaling bug in
  aplay
- Removes several FIXMEs
- Reduces amount of sample copying in all applications (especially
  Piano, where this is critical), improving performance
- Reduces number of resampling users, making future API changes (which
  will need to happen for correct resampling to be implemented) easier

I also threw in a simple race condition fix for Piano's audio player
loop.
2023-07-01 23:27:24 +01:00

179 lines
6.3 KiB
C++

/*
* Copyright (c) 2018-2020, Andreas Kling <kling@serenityos.org>
* Copyright (c) 2021-2022, kleines Filmröllchen <filmroellchen@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#pragma once
#include "ConnectionFromClient.h"
#include "FadingProperty.h"
#include <AK/Atomic.h>
#include <AK/Badge.h>
#include <AK/ByteBuffer.h>
#include <AK/Debug.h>
#include <AK/Queue.h>
#include <AK/RefCounted.h>
#include <AK/WeakPtr.h>
#include <LibAudio/Queue.h>
#include <LibAudio/Resampler.h>
#include <LibCore/ConfigFile.h>
#include <LibCore/File.h>
#include <LibCore/Timer.h>
#include <LibThreading/ConditionVariable.h>
#include <LibThreading/Mutex.h>
#include <LibThreading/Thread.h>
#include <sys/types.h>
namespace AudioServer {
// Headroom, i.e. fixed attenuation for all audio streams.
// This is to prevent clipping when two streams with low headroom (e.g. normalized & compressed) are playing.
constexpr double SAMPLE_HEADROOM = 0.95;
// The size of the buffer in samples that the hardware receives through write() calls to the audio device.
constexpr size_t HARDWARE_BUFFER_SIZE = 512;
// The hardware buffer size in bytes; there's two channels of 16-bit samples.
constexpr size_t HARDWARE_BUFFER_SIZE_BYTES = HARDWARE_BUFFER_SIZE * 2 * sizeof(i16);
class ConnectionFromClient;
class ClientAudioStream : public RefCounted<ClientAudioStream> {
public:
explicit ClientAudioStream(ConnectionFromClient&);
~ClientAudioStream() = default;
bool get_next_sample(Audio::Sample& sample, u32 audiodevice_sample_rate)
{
// Note: Even though we only check client state here, we will probably close the client much earlier.
if (!is_connected())
return false;
if (m_paused)
return false;
if (m_in_chunk_location >= m_current_audio_chunk.size()) {
auto result = m_buffer->dequeue();
if (result.is_error()) {
if (result.error() == Audio::AudioQueue::QueueStatus::Empty) {
dbgln_if(AUDIO_DEBUG, "Audio client {} can't keep up!", m_client->client_id());
}
return false;
}
// FIXME: Our resampler and the way we resample here are bad.
// Ideally, we should both do perfect band-corrected resampling,
// as well as carry resampling state over between buffers.
auto attempted_resample = Audio::ResampleHelper<Audio::Sample> {
m_sample_rate == 0 ? audiodevice_sample_rate : m_sample_rate, audiodevice_sample_rate
}
.try_resample(result.release_value());
if (attempted_resample.is_error())
return false;
// If the sample rate changes underneath us, we will still play the existing buffer unchanged until we're done.
// This is not a significant problem since the buffers are very small (~100 samples or less).
m_current_audio_chunk = attempted_resample.release_value();
m_in_chunk_location = 0;
}
sample = m_current_audio_chunk[m_in_chunk_location++];
return true;
}
bool is_connected() const { return m_client && m_client->is_open(); }
ConnectionFromClient* client() { return m_client.ptr(); }
void set_buffer(OwnPtr<Audio::AudioQueue> buffer) { m_buffer = move(buffer); }
void clear()
{
ErrorOr<Array<Audio::Sample, Audio::AUDIO_BUFFER_SIZE>, Audio::AudioQueue::QueueStatus> result = Audio::AudioQueue::QueueStatus::Invalid;
do {
result = m_buffer->dequeue();
} while (!result.is_error() || result.error() != Audio::AudioQueue::QueueStatus::Empty);
}
void set_paused(bool paused) { m_paused = paused; }
FadingProperty<double>& volume() { return m_volume; }
double volume() const { return m_volume; }
void set_volume(double const volume) { m_volume = volume; }
bool is_muted() const { return m_muted; }
void set_muted(bool muted) { m_muted = muted; }
u32 sample_rate() const { return m_sample_rate; }
void set_sample_rate(u32 sample_rate)
{
dbgln_if(AUDIO_DEBUG, "queue {} got sample rate {} Hz", m_client->client_id(), sample_rate);
m_sample_rate = sample_rate;
}
private:
OwnPtr<Audio::AudioQueue> m_buffer;
Vector<Audio::Sample> m_current_audio_chunk;
size_t m_in_chunk_location;
bool m_paused { true };
bool m_muted { false };
u32 m_sample_rate;
WeakPtr<ConnectionFromClient> m_client;
FadingProperty<double> m_volume { 1 };
};
class Mixer : public Core::Object {
C_OBJECT_ABSTRACT(Mixer)
public:
static ErrorOr<NonnullRefPtr<Mixer>> try_create(NonnullRefPtr<Core::ConfigFile> config)
{
// FIXME: Allow AudioServer to use other audio channels as well
auto device = TRY(Core::File::open("/dev/audio/0"sv, Core::File::OpenMode::Write));
return adopt_nonnull_ref_or_enomem(new (nothrow) Mixer(move(config), move(device)));
}
virtual ~Mixer() override = default;
NonnullRefPtr<ClientAudioStream> create_queue(ConnectionFromClient&);
// To the outside world, we pretend that the target volume is already reached, even though it may be still fading.
double main_volume() const { return m_main_volume.target(); }
void set_main_volume(double volume);
bool is_muted() const { return m_muted; }
void set_muted(bool);
int audiodevice_set_sample_rate(u32 sample_rate);
u32 audiodevice_get_sample_rate() const;
private:
Mixer(NonnullRefPtr<Core::ConfigFile> config, NonnullOwnPtr<Core::File> device);
void request_setting_sync();
Vector<NonnullRefPtr<ClientAudioStream>> m_pending_mixing;
Threading::Mutex m_pending_mutex;
Threading::ConditionVariable m_mixing_necessary { m_pending_mutex };
NonnullOwnPtr<Core::File> m_device;
mutable Optional<u32> m_cached_sample_rate {};
NonnullRefPtr<Threading::Thread> m_sound_thread;
bool m_muted { false };
FadingProperty<double> m_main_volume { 1 };
NonnullRefPtr<Core::ConfigFile> m_config;
RefPtr<Core::Timer> m_config_write_timer;
Array<u8, HARDWARE_BUFFER_SIZE_BYTES> m_stream_buffer;
Array<u8, HARDWARE_BUFFER_SIZE_BYTES> const m_zero_filled_buffer {};
void mix();
};
// Interval in ms when the server tries to save its configuration to disk.
constexpr unsigned AUDIO_CONFIG_WRITE_INTERVAL = 2000;
}