mpv/audio/out/internal.h

248 lines
10 KiB
C
Raw Normal View History

/*
* This file is part of mpv.
*
* mpv is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* mpv is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with mpv. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MP_AO_INTERNAL_H_
#define MP_AO_INTERNAL_H_
#include <stdbool.h>
#include <pthread.h>
#include "osdep/atomic.h"
#include "audio/out/ao.h"
/* global data used by ao.c and ao drivers */
struct ao {
int samplerate;
struct mp_chmap channels;
int format; // one of AF_FORMAT_...
int bps; // bytes per second (per plane)
int sstride; // size of a sample on each plane
// (format_size*num_channels/num_planes)
int num_planes;
bool probing; // if true, don't fail loudly on init
bool untimed; // don't assume realtime playback
int device_buffer; // device buffer in samples (guessed by
// common init code if not set by driver)
const struct ao_driver *api; // entrypoints to the wrapper (push.c/pull.c)
const struct ao_driver *driver;
void *priv;
struct mpv_global *global;
struct encode_lavc_context *encode_lavc_ctx;
void (*wakeup_cb)(void *ctx);
void *wakeup_ctx;
struct mp_log *log; // Using e.g. "[ao/coreaudio]" as prefix
int init_flags; // AO_INIT_* flags
bool stream_silence; // if audio inactive, just play silence
// Set by the driver on init. This is typically the period size, and the
// smallest unit the driver will accept in one piece (although if
// AOPLAY_FINAL_CHUNK is set, the driver must accept everything).
// This value is in complete samples (i.e. 1 for stereo means 1 sample
// for both channels each).
// Used for push based API only.
int period_size;
// The device as selected by the user, usually using ao_device_desc.name
// from an entry from the list returned by driver->list_devices. If the
// default device should be used, this is set to NULL.
char *device;
// Application name to report to the audio API.
char *client_name;
// Used during init: if init fails, redirect to this ao
char *redirect;
// Internal events (use ao_request_reload(), ao_hotplug_event())
atomic_uint events_;
// Float gain multiplicator
mp_atomic_float gain;
audio: add audio softvol processing to AO This does what af_volume used to do. Since we couldn't relicense it, just rewrite it. Since we don't have a new filter mechanism yet, and the libavfilter is too inconvenient, do applying the volume gain in ao.c directly. This is done before handling the audio data to the driver. Since push.c runs a separate thread, and pull.c is called asynchronously from the audio driver's thread, the volume value needs to be synchronized. There's no existing central mutex, so do some shit with atomics. Since there's no atomic_float type predefined (which is at least needed when using the legacy wrapper), do some nonsense about reinterpret casting the float value to an int for the purpose of atomic access. Not sure if using memcpy() is undefined behavior, but for now I don't care. The advantage of not using a filter is lower complexity (no filter auto insertion), and lower latency (gain processing is done after our internal audio buffer of at least 200ms). Disavdantages include inability to use native volume control _before_ other filters with custom filter chains, and the need to add new processing for each new sample type. Since this doesn't reuse any of the old GPL code, nor does indirectly rely on it, volume and replaygain handling now works in LGPL mode. How to process the gain is inspired by libavfilter's af_volume (LGPL). In particular, we use exactly the same rounding, and we quantize processing for integer sample types by 256 steps. Some of libavfilter's copyright may or may not apply, but I think not, and it's the same license anyway.
2017-11-29 20:30:10 +00:00
int buffer;
double def_buffer;
void *api_priv;
};
extern const struct ao_driver ao_api_push;
extern const struct ao_driver ao_api_pull;
/* Note:
*
* In general, there are two types of audio drivers:
* a) push based (the user queues data that should be played)
* b) pull callback based (the audio API calls a callback to get audio)
*
* The ao.c code can handle both. It basically implements two audio paths
* and provides a uniform API for them. If ao_driver->play is NULL, it assumes
* that the driver uses a callback based audio API, otherwise push based.
*
* Requirements:
* a) ->play is called to queue audio. push.c creates a thread to regularly
* refill audio device buffers with ->play, but all driver functions are
* always called under an exclusive lock.
* Mandatory:
* init
* uninit
* reset
* get_space
* play
* get_delay
* pause
* resume
audio/out/pull: remove race conditions There were subtle and minor race conditions in the pull.c code, and AOs using it (jack, portaudio, sdl, wasapi). Attempt to remove these. There was at least a race condition in the ao_reset() implementation: mp_ring_reset() was called concurrently to the audio callback. While the ringbuffer uses atomics to allow concurrent access, the reset function wasn't concurrency-safe (and can't easily be made to). Fix this by stopping the audio callback before doing a reset. After that, we can do anything without needing synchronization. The callback is resumed when resuming playback at a later point. Don't call driver->pause, and make driver->resume and driver->reset start/stop the audio callback. In the initial state, the audio callback must be disabled. JackAudio of course is different. Maybe there is no way to suspend the audio callback without "disconnecting" it (what jack_deactivate() would do), so I'm not trying my luck, and implemented a really bad hack doing active waiting until we get the audio callback into a state where it won't interfere. Once the callback goes from AO_STATE_WAIT to NONE, we can be sure that the callback doesn't access the ringbuffer or anything else anymore. Since both sched_yield() and pthread_yield() apparently are not always available, use mp_sleep_us(1) to avoid burning CPU during active waiting. The ao_jack.c change also removes a race condition: apparently we didn't initialize _all_ ao fields before starting the audio callback. In ao_wasapi.c, I'm not sure whether reset really waits for the audio callback to return. Kovensky says it's not guaranteed, so disable the reset callback - for now the behavior of ao_wasapi.c is like with ao_jack.c, and active waiting is used to deal with the audio callback.
2014-05-29 00:24:17 +00:00
* Optional:
* control
* drain
* wait
* wakeup
* b) ->play must be NULL. ->resume must be provided, and should make the
* audio API start calling the audio callback. Your audio callback should
* in turn call ao_read_data() to get audio data. Most functions are
* optional and will be emulated if missing (e.g. pausing is emulated as
* silence). ->get_delay and ->get_space are never called.
* Mandatory:
* init
* uninit
audio/out/pull: remove race conditions There were subtle and minor race conditions in the pull.c code, and AOs using it (jack, portaudio, sdl, wasapi). Attempt to remove these. There was at least a race condition in the ao_reset() implementation: mp_ring_reset() was called concurrently to the audio callback. While the ringbuffer uses atomics to allow concurrent access, the reset function wasn't concurrency-safe (and can't easily be made to). Fix this by stopping the audio callback before doing a reset. After that, we can do anything without needing synchronization. The callback is resumed when resuming playback at a later point. Don't call driver->pause, and make driver->resume and driver->reset start/stop the audio callback. In the initial state, the audio callback must be disabled. JackAudio of course is different. Maybe there is no way to suspend the audio callback without "disconnecting" it (what jack_deactivate() would do), so I'm not trying my luck, and implemented a really bad hack doing active waiting until we get the audio callback into a state where it won't interfere. Once the callback goes from AO_STATE_WAIT to NONE, we can be sure that the callback doesn't access the ringbuffer or anything else anymore. Since both sched_yield() and pthread_yield() apparently are not always available, use mp_sleep_us(1) to avoid burning CPU during active waiting. The ao_jack.c change also removes a race condition: apparently we didn't initialize _all_ ao fields before starting the audio callback. In ao_wasapi.c, I'm not sure whether reset really waits for the audio callback to return. Kovensky says it's not guaranteed, so disable the reset callback - for now the behavior of ao_wasapi.c is like with ao_jack.c, and active waiting is used to deal with the audio callback.
2014-05-29 00:24:17 +00:00
* resume (starts the audio callback)
* Also, the following optional callbacks can be provided:
* reset (stops the audio callback, resume() restarts it)
* control
*/
struct ao_driver {
// If true, use with encoding only.
bool encode;
// Name used for --ao.
const char *name;
// Description shown with --ao=help.
const char *description;
// This requires waiting for a AO_EVENT_INITIAL_UNBLOCK event before the
// first play() call is done. Encode mode uses this, and push mode
// respects it automatically (don't use with pull mode).
bool initially_blocked;
// Whether underruns are strictly _always_ reported via ao_underrun_event().
// Do not set this to true if underruns may be missed in some way. If the
// AO can't guarantee to play silence after underruns, it may be better not
// to set this.
player: consider audio buffer if AO driver does not report underruns AOs can report audio underruns, but only ao_alsa and ao_sdl (???) currently do so. If the AO was marked as not reporting it, the cache state was used to determine whether playback was interrupted due to slow input. This caused problems in some cases, such as video with very low video frame rate: when a new frame is displayed, a new frame has to be decoded, and since there it's so much further into the file (long frame durations), the cache gets into an underrun state for a short moment, even though both audio and video are playing fine. Enlarging the audio buffer didn't help. Fix this by making all AOs report underruns. If the AO driver does not report underruns, fall back to using the buffer state. pull.c behavior is slightly changed. Pull AOs are normally intended to be used by pseudo-realtime audio APIs that fetch an audio buffer from the API user via callback. I think it makes no sense to consider a buffer underflow not an underrun in any situation, since we return silence to the reader. (OK, maybe the reader could check the return value? But let's not go there as long as there's no implementation.) Remove the flag from ao_sdl.c, since it just worked via the generic mechanism. Make the redundant underrun message verbose only. push.c seems to log a redundant underflow message when resuming (because somehow ao_play_data() is called when there's still no new data in the buffer). But since ao_alsa does its own underrun reporting, and I only use ao_alsa, I don't really care. Also in all my tests, there seemed to be a rather high delay until the underflow was logged (with audio only). I have no idea why this happened and didn't try to debug this, but there's probably something wrong somewhere. This commit may cause random regressions. See: #7440
2020-02-13 00:28:59 +00:00
// If not set, the generic buffer code will report an underrun if the buffer
// becomes empty.
bool reports_underruns;
// Init the device using ao->format/ao->channels/ao->samplerate. If the
// device doesn't accept these parameters, you can attempt to negotiate
// fallback parameters, and set the ao format fields accordingly.
int (*init)(struct ao *ao);
// Optional. See ao_control() etc. in ao.c
int (*control)(struct ao *ao, enum aocontrol cmd, void *arg);
void (*uninit)(struct ao *ao);
// push based: see ao_reset()
// pull based: stop the audio callback
void (*reset)(struct ao *ao);
// push based: see ao_pause()
void (*pause)(struct ao *ao);
// push based: see ao_resume()
// pull based: start the audio callback
void (*resume)(struct ao *ao);
// push based: see ao_play()
int (*get_space)(struct ao *ao);
// push based: see ao_play()
int (*play)(struct ao *ao, void **data, int samples, int flags);
// push based: see ao_get_delay()
double (*get_delay)(struct ao *ao);
// push based: block until all queued audio is played (optional)
void (*drain)(struct ao *ao);
// Optional. Return true if audio has stopped in any way.
bool (*get_eof)(struct ao *ao);
// Wait until the audio buffer needs to be refilled. The lock is the
// internal mutex usually protecting the internal AO state (and used to
// protect driver calls), and must be temporarily unlocked while waiting.
// ->wakeup will be called (with lock held) if the wait should be canceled.
// Returns 0 on success, -1 on error.
// Optional; if this is not provided, generic code using audio timing is
// used to estimate when the AO needs to be refilled.
// Warning: it's only called if the feed thread truly needs to know when
// the audio thread takes data again. Often, it will just copy
// the complete soft-buffer to the AO, and then wait for the
// decoder instead. Don't do necessary work in this callback.
int (*wait)(struct ao *ao, pthread_mutex_t *lock);
// In combination with wait(). Lock may or may not be held.
void (*wakeup)(struct ao *ao);
// Return the list of devices currently available in the system. Use
// ao_device_list_add() to add entries. The selected device will be set as
// ao->device (using ao_device_desc.name).
// Warning: the ao struct passed is not initialized with ao_driver->init().
// Instead, hotplug_init/hotplug_uninit is called. If these
// callbacks are not set, no driver initialization call is done
// on the ao struct.
void (*list_devs)(struct ao *ao, struct ao_device_list *list);
// If set, these are called before/after ao_driver->list_devs is called.
// It is also assumed that the driver can do hotplugging - which means
// it is expected to call ao_hotplug_event(ao) whenever the system's
// audio device list changes. The player will then call list_devs() again.
int (*hotplug_init)(struct ao *ao);
void (*hotplug_uninit)(struct ao *ao);
// For option parsing (see vo.h)
int priv_size;
const void *priv_defaults;
const struct m_option *options;
const char *options_prefix;
const struct m_sub_options *global_opts;
};
// These functions can be called by AOs.
int ao_play_silence(struct ao *ao, int samples);
int ao_read_data(struct ao *ao, void **data, int samples, int64_t out_time_us);
struct pollfd;
int ao_wait_poll(struct ao *ao, struct pollfd *fds, int num_fds,
pthread_mutex_t *lock);
void ao_wakeup_poll(struct ao *ao);
bool ao_underrun_event(struct ao *ao);
bool ao_chmap_sel_adjust(struct ao *ao, const struct mp_chmap_sel *s,
struct mp_chmap *map);
bool ao_chmap_sel_adjust2(struct ao *ao, const struct mp_chmap_sel *s,
struct mp_chmap *map, bool safe_multichannel);
bool ao_chmap_sel_get_def(struct ao *ao, const struct mp_chmap_sel *s,
struct mp_chmap *map, int num);
// Add a deep copy of e to the list.
// Call from ao_driver->list_devs callback only.
void ao_device_list_add(struct ao_device_list *list, struct ao *ao,
struct ao_device_desc *e);
audio: add audio softvol processing to AO This does what af_volume used to do. Since we couldn't relicense it, just rewrite it. Since we don't have a new filter mechanism yet, and the libavfilter is too inconvenient, do applying the volume gain in ao.c directly. This is done before handling the audio data to the driver. Since push.c runs a separate thread, and pull.c is called asynchronously from the audio driver's thread, the volume value needs to be synchronized. There's no existing central mutex, so do some shit with atomics. Since there's no atomic_float type predefined (which is at least needed when using the legacy wrapper), do some nonsense about reinterpret casting the float value to an int for the purpose of atomic access. Not sure if using memcpy() is undefined behavior, but for now I don't care. The advantage of not using a filter is lower complexity (no filter auto insertion), and lower latency (gain processing is done after our internal audio buffer of at least 200ms). Disavdantages include inability to use native volume control _before_ other filters with custom filter chains, and the need to add new processing for each new sample type. Since this doesn't reuse any of the old GPL code, nor does indirectly rely on it, volume and replaygain handling now works in LGPL mode. How to process the gain is inspired by libavfilter's af_volume (LGPL). In particular, we use exactly the same rounding, and we quantize processing for integer sample types by 256 steps. Some of libavfilter's copyright may or may not apply, but I think not, and it's the same license anyway.
2017-11-29 20:30:10 +00:00
void ao_post_process_data(struct ao *ao, void **data, int num_samples);
struct ao_convert_fmt {
int src_fmt; // source AF_FORMAT_*
int channels; // number of channels
int dst_bits; // total target data sample size
int pad_msb; // padding in the MSB (i.e. required shifting)
int pad_lsb; // padding in LSB (required 0 bits) (ignored)
};
bool ao_can_convert_inplace(struct ao_convert_fmt *fmt);
bool ao_need_conversion(struct ao_convert_fmt *fmt);
void ao_convert_inplace(struct ao_convert_fmt *fmt, void **data, int num_samples);
int ao_read_data_converted(struct ao *ao, struct ao_convert_fmt *fmt,
void **data, int samples, int64_t out_time_us);
#endif