path: root/audio/out/ao_wasapi0.c

/*
 * This file is part of mpv.
 *
 * mpv is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 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 General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with mpv.  If not, see <http://www.gnu.org/licenses/>.
 */

#define COBJMACROS 1
#define _WIN32_WINNT 0x600

#include <stdlib.h>
#include <inttypes.h>
#include <process.h>
#include <initguid.h>
#include <audioclient.h>
#include <endpointvolume.h>
#include <mmdeviceapi.h>
#include <avrt.h>

#include "config.h"
#include "core/subopt-helper.h"
#include "core/m_option.h"
#include "core/m_config.h"
#include "audio/format.h"
#include "core/mp_msg.h"
#include "core/mp_ring.h"
#include "ao.h"

#ifndef BASE_SEARCH_PATH_ENABLE_SAFE_SEARCHMODE
#define BASE_SEARCH_PATH_ENABLE_SAFE_SEARCHMODE 0x00000001
#endif

#ifndef PKEY_Device_FriendlyName
DEFINE_PROPERTYKEY(PKEY_Device_FriendlyName,
                   0xa45c254e, 0xdf1c, 0x4efd, 0x80, 0x20,
                   0x67, 0xd1, 0x46, 0xa8, 0x50, 0xe0, 14);
#endif

#define RING_BUFFER_COUNT 64

/* 20 millisecond buffer? */
#define BUFFER_TIME 20000000.0
#define EXIT_ON_ERROR(hres)  \
              if (FAILED(hres)) { goto exit_label; }
#define SAFE_RELEASE(unk, release) \
              if ((unk) != NULL) { release; (unk) = NULL; }

/* Supposed to use __uuidof, but it is C++ only, declare our own */
static const GUID local_KSDATAFORMAT_SUBTYPE_PCM = {
    0x1, 0x0000, 0x0010, {0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}
};
static const GUID local_KSDATAFORMAT_SUBTYPE_IEEE_FLOAT = {
    0x3, 0x0000, 0x0010, {0x80, 0x00, 0x00, 0xaa, 0x00, 0x38, 0x9b, 0x71}
};

union WAVEFMT {
    WAVEFORMATEX *ex;
    WAVEFORMATEXTENSIBLE *extensible;
};

typedef struct wasapi0_state {
    HANDLE threadLoop;

    /* Init phase */
    int init_ret;
    HANDLE init_done;
    HANDLE fatal_error; /* signal to indicate unrecoverable error */
    int share_mode;

    /* Events */
    HANDLE hPause;

    /* Play */
    HANDLE hPlay;
    int is_playing;

    /* Reset */
    HANDLE hReset;

    /* uninit */
    HANDLE hUninit;
    LONG immed;

    /* volume control */
    HANDLE hGetvol, hSetvol, hDoneVol;
    DWORD vol_hw_support, status;
    float audio_volume;

    /* Prints, for in case line buffers are disabled */
    CRITICAL_SECTION print_lock;

    /* Buffers */
    struct mp_ring *ringbuff;
    size_t buffer_block_size; /* Size of each block in bytes */
    REFERENCE_TIME
        minRequestedDuration; /* minimum wasapi buffer block size, in 100-nanosecond units */
    REFERENCE_TIME
        defaultRequestedDuration; /* default wasapi default block size, in 100-nanosecond units */
    UINT32 bufferFrameCount; /* wasapi buffer block size, number of frames, frame size at format.nBlockAlign */

    /* WASAPI handles, owned by other thread */
    IMMDeviceEnumerator *pEnumerator;
    IMMDevice *pDevice;
    IAudioClient *pAudioClient;
    IAudioRenderClient *pRenderClient;
    IAudioEndpointVolume *pEndpointVolume;
    HANDLE hFeed; /* wasapi event */
    HANDLE hTask; /* AV thread */
    DWORD taskIndex; /* AV task ID */
    WAVEFORMATEXTENSIBLE format;

    int opt_exclusive;
    int opt_list;
    char *opt_device;

    /* We still need to support XP, don't use these functions directly, blob owned by main thread */
    struct {
        HMODULE hAvrt;
        HANDLE (WINAPI *pAvSetMmThreadCharacteristicsW)(LPCWSTR, LPDWORD);
        WINBOOL (WINAPI *pAvRevertMmThreadCharacteristics)(HANDLE);
    } VistaBlob;
} wasapi0_state;

static int fill_VistaBlob(wasapi0_state *state)
{
    if (!state)
        return 1;
    HMODULE hkernel32 = GetModuleHandleW(L"kernel32.dll");
    if (!hkernel32)
        return 1;
    WINBOOL (WINAPI *pSetDllDirectory)(LPCWSTR lpPathName) =
        (WINBOOL (WINAPI *)(LPCWSTR))GetProcAddress(hkernel32, "SetDllDirectoryW");
    WINBOOL (WINAPI *pSetSearchPathMode)(DWORD Flags) =
        (WINBOOL (WINAPI *)(DWORD))GetProcAddress(hkernel32, "SetSearchPathMode");
    if (pSetSearchPathMode)
        pSetDllDirectory(L"");  /* Attempt to use safe search paths */
    if (pSetSearchPathMode)
        pSetSearchPathMode(BASE_SEARCH_PATH_ENABLE_SAFE_SEARCHMODE);
    state->VistaBlob.hAvrt = LoadLibraryW(L"avrt.dll");
    if (!state->VistaBlob.hAvrt)
        goto exit_label;
    state->VistaBlob.pAvSetMmThreadCharacteristicsW =
        (HANDLE (WINAPI *)(LPCWSTR, LPDWORD))
            GetProcAddress(state->VistaBlob.hAvrt, "AvSetMmThreadCharacteristicsW");
    state->VistaBlob.pAvRevertMmThreadCharacteristics =
        (WINBOOL (WINAPI *)(HANDLE))
            GetProcAddress(state->VistaBlob.hAvrt, "AvRevertMmThreadCharacteristics");
    return 0;
exit_label:
    if (state->VistaBlob.hAvrt)
        FreeLibrary(state->VistaBlob.hAvrt);
    if (pSetSearchPathMode)
        pSetDllDirectory(NULL);
    return 1;
}

static const char *explain_err(const HRESULT hr)
{
    switch (hr) {
    case S_OK:
        return "S_OK";
    case AUDCLNT_E_NOT_INITIALIZED:
        return "AUDCLNT_E_NOT_INITIALIZED";
    case AUDCLNT_E_ALREADY_INITIALIZED:
        return "AUDCLNT_E_ALREADY_INITIALIZED";
    case AUDCLNT_E_WRONG_ENDPOINT_TYPE:
        return "AUDCLNT_E_WRONG_ENDPOINT_TYPE";
    case AUDCLNT_E_DEVICE_INVALIDATED:
        return "AUDCLNT_E_DEVICE_INVALIDATED";
    case AUDCLNT_E_NOT_STOPPED:
        return "AUDCLNT_E_NOT_STOPPED";
    case AUDCLNT_E_BUFFER_TOO_LARGE:
        return "AUDCLNT_E_BUFFER_TOO_LARGE";
    case AUDCLNT_E_OUT_OF_ORDER:
        return "AUDCLNT_E_OUT_OF_ORDER";
    case AUDCLNT_E_UNSUPPORTED_FORMAT:
        return "AUDCLNT_E_UNSUPPORTED_FORMAT";
    case AUDCLNT_E_INVALID_SIZE:
        return "AUDCLNT_E_INVALID_SIZE";
    case AUDCLNT_E_DEVICE_IN_USE:
        return "AUDCLNT_E_DEVICE_IN_USE";
    case AUDCLNT_E_BUFFER_OPERATION_PENDING:
        return "AUDCLNT_E_BUFFER_OPERATION_PENDING";
    case AUDCLNT_E_THREAD_NOT_REGISTERED:
        return "AUDCLNT_E_THREAD_NOT_REGISTERED";
    case AUDCLNT_E_EXCLUSIVE_MODE_NOT_ALLOWED:
        return "AUDCLNT_E_EXCLUSIVE_MODE_NOT_ALLOWED";
    case AUDCLNT_E_ENDPOINT_CREATE_FAILED:
        return "AUDCLNT_E_ENDPOINT_CREATE_FAILED";
    case AUDCLNT_E_SERVICE_NOT_RUNNING:
        return "AUDCLNT_E_SERVICE_NOT_RUNNING";
    case AUDCLNT_E_EVENTHANDLE_NOT_EXPECTED:
        return "AUDCLNT_E_EVENTHANDLE_NOT_EXPECTED";
    case AUDCLNT_E_EXCLUSIVE_MODE_ONLY:
        return "AUDCLNT_E_EXCLUSIVE_MODE_ONLY";
    case AUDCLNT_E_BUFDURATION_PERIOD_NOT_EQUAL:
        return "AUDCLNT_E_BUFDURATION_PERIOD_NOT_EQUAL";
    case AUDCLNT_E_EVENTHANDLE_NOT_SET:
        return "AUDCLNT_E_EVENTHANDLE_NOT_SET";
    case AUDCLNT_E_INCORRECT_BUFFER_SIZE:
        return "AUDCLNT_E_INCORRECT_BUFFER_SIZE";
    case AUDCLNT_E_BUFFER_SIZE_ERROR:
        return "AUDCLNT_E_BUFFER_SIZE_ERROR";
    case AUDCLNT_E_CPUUSAGE_EXCEEDED:
        return "AUDCLNT_E_CPUUSAGE_EXCEEDED";
    case AUDCLNT_E_BUFFER_ERROR:
        return "AUDCLNT_E_BUFFER_ERROR";
    case AUDCLNT_E_BUFFER_SIZE_NOT_ALIGNED:
        return "AUDCLNT_E_BUFFER_SIZE_NOT_ALIGNED";
    case AUDCLNT_E_INVALID_DEVICE_PERIOD:
        return "AUDCLNT_E_INVALID_DEVICE_PERIOD";
    case AUDCLNT_E_INVALID_STREAM_FLAG:
        return "AUDCLNT_E_INVALID_STREAM_FLAG";
    case AUDCLNT_E_ENDPOINT_OFFLOAD_NOT_CAPABLE:
        return "AUDCLNT_E_ENDPOINT_OFFLOAD_NOT_CAPABLE";
    case AUDCLNT_E_RESOURCES_INVALIDATED:
        return "AUDCLNT_E_RESOURCES_INVALIDATED";
    case AUDCLNT_S_BUFFER_EMPTY:
        return "AUDCLNT_S_BUFFER_EMPTY";
    case AUDCLNT_S_THREAD_ALREADY_REGISTERED:
        return "AUDCLNT_S_THREAD_ALREADY_REGISTERED";
    case AUDCLNT_S_POSITION_STALLED:
        return "AUDCLNT_S_POSITION_STALLED";
    default:
        return "<Unknown>";
    }
}

static void set_format(WAVEFORMATEXTENSIBLE *wformat, WORD bytepersample,
                       DWORD samplerate, WORD channels, DWORD chanmask)
{
    wformat->Format.wFormatTag = WAVE_FORMAT_EXTENSIBLE; /* Only PCM is supported */
    wformat->Format.nChannels = channels;
    wformat->Format.nSamplesPerSec = samplerate;
    wformat->Format.nAvgBytesPerSec = wformat->Format.nChannels *
                                      bytepersample *
                                      wformat->Format.nSamplesPerSec;
    wformat->Format.nBlockAlign = wformat->Format.nChannels * bytepersample;
    wformat->Format.wBitsPerSample = bytepersample * 8;
    wformat->Format.cbSize =
        22; /* must be at least 22 for WAVE_FORMAT_EXTENSIBLE */
    if (bytepersample == 4)
        wformat->SubFormat = local_KSDATAFORMAT_SUBTYPE_IEEE_FLOAT;
    else
        wformat->SubFormat = local_KSDATAFORMAT_SUBTYPE_PCM;
    wformat->Samples.wValidBitsPerSample = wformat->Format.wBitsPerSample;
    wformat->dwChannelMask = chanmask;
}

static int format_set_bits(int old_format, int bits, int fp) {
    int format = old_format;
    format &= (~AF_FORMAT_BITS_MASK) & (~AF_FORMAT_POINT_MASK) & (~AF_FORMAT_SIGN_MASK);
    format |= AF_FORMAT_SI;

    switch (bits) {
    case 32:
        format |= AF_FORMAT_32BIT;
        break;
    case 24:
        format |= AF_FORMAT_24BIT;
        break;
    case 16:
        format |= AF_FORMAT_16BIT;
        break;
    default:
        abort(); // (should be) unreachable
    }

    if (fp) {
        format |= AF_FORMAT_F;
    } else {
        format |= AF_FORMAT_I;
    }

    return format;
}

static int set_ao_format(struct wasapi0_state *state,
                         struct ao *const ao,
                         WAVEFORMATEXTENSIBLE wformat) {
    // .Data1 == 1 is PCM, .Data1 == 3 is IEEE_FLOAT
    int format = format_set_bits(ao->format,
        wformat.Format.wBitsPerSample, wformat.SubFormat.Data1 == 3);

    if (wformat.SubFormat.Data1 != 1 && wformat.SubFormat.Data1 != 3) {
        mp_msg(MSGT_AO, MSGL_ERR, "ao-wasapi: unknown SubFormat %d\n",
            wformat.SubFormat.Data1);
        return 0;
    }

    ao->samplerate = wformat.Format.nSamplesPerSec;
    ao->bps = wformat.Format.nAvgBytesPerSec;
    ao->format = format;

    if (ao->channels.num != wformat.Format.nChannels) {
        mp_chmap_from_channels(&ao->channels, wformat.Format.nChannels);
    }

    state->format = wformat;
    return 1;
}

static int try_format(struct wasapi0_state *state,
                      struct ao *const ao,
                      int bits, int samplerate,
                      const struct mp_chmap channels)
{
    WAVEFORMATEXTENSIBLE wformat;
    set_format(&wformat, bits / 8, samplerate, channels.num, mp_chmap_to_waveext(&channels));

    int af_format = format_set_bits(ao->format, bits, bits == 32);

    EnterCriticalSection(&state->print_lock);
    mp_msg(MSGT_AO, MSGL_V, "ao-wasapi: trying %dch %s @ %dhz\n",
           channels.num, af_fmt2str_short(af_format), samplerate);
    LeaveCriticalSection(&state->print_lock);

    union WAVEFMT u;
    u.extensible = &wformat;

    WAVEFORMATEX *closestMatch;
    HRESULT hr = IAudioClient_IsFormatSupported(state->pAudioClient,
                                                state->share_mode,
                                                u.ex, &closestMatch);

    if (closestMatch) {
        if (closestMatch->wFormatTag == WAVE_FORMAT_EXTENSIBLE) {
            u.ex = closestMatch;
            wformat = *u.extensible;
        } else {
            wformat.Format = *closestMatch;
        }

        CoTaskMemFree(closestMatch);
    }

    if (hr == S_FALSE) {
        if (set_ao_format(state, ao, wformat)) {
            EnterCriticalSection(&state->print_lock);
            mp_msg(MSGT_AO, MSGL_V, "ao-wasapi: accepted as %dch %s @ %dhz\n",
                ao->channels.num, af_fmt2str_short(ao->format), ao->samplerate);
            LeaveCriticalSection(&state->print_lock);

            return 1;
        }
    } if (hr == S_OK || (!state->opt_exclusive && hr == AUDCLNT_E_UNSUPPORTED_FORMAT)) {
        // AUDCLNT_E_UNSUPPORTED_FORMAT here means "works in shared, doesn't in exclusive"
        if (set_ao_format(state, ao, wformat)) {
            EnterCriticalSection(&state->print_lock);
            mp_msg(MSGT_AO, MSGL_V, "ao-wasapi: %dch %s @ %dhz accepted\n",
                   ao->channels.num, af_fmt2str_short(af_format), samplerate);
            LeaveCriticalSection(&state->print_lock);
            return 1;
        }
    }
    return 0;
}

static int try_mix_format(struct wasapi0_state *state,
                          struct ao *const ao)
{
    WAVEFORMATEX *deviceFormat = NULL;
    WAVEFORMATEX *closestMatch = NULL;
    int ret = 0;

    HRESULT hr = IAudioClient_GetMixFormat(state->pAudioClient, &deviceFormat);
    EXIT_ON_ERROR(hr)

    union WAVEFMT u;
    u.ex = deviceFormat;
    WAVEFORMATEXTENSIBLE wformat = *u.extensible;

    ret = try_format(state, ao, wformat.Format.wBitsPerSample,
                     wformat.Format.nSamplesPerSec, ao->channels);

exit_label:
    SAFE_RELEASE(deviceFormat, CoTaskMemFree(deviceFormat));
    SAFE_RELEASE(closestMatch, CoTaskMemFree(closestMatch));
    return ret;
}

static int find_formats(struct ao *const ao)
{
    struct wasapi0_state *state = (struct wasapi0_state *)ao->priv;

    /* See if the format works as-is */
    int bits = af_fmt2bits(ao->format);
    /* don't try 8bits -- there are various 8bit modes other than PCM (*-law et al);
       let's just stick to PCM or float here. */
    if (bits == 8) {
        bits = 16;
    } else if (try_format(state, ao, bits, ao->samplerate, ao->channels)) {
        return 0;
    }
    if (!state->opt_exclusive) {
        /* shared mode, we can use the system default mix format. */
        if (try_mix_format(state, ao)) {
            return 0;
        }

        EnterCriticalSection(&state->print_lock);
        mp_msg(MSGT_AO, MSGL_WARN, "ao-wasapi: couldn't use default mix format!\n");
        LeaveCriticalSection(&state->print_lock);
    }

    /* Exclusive mode, we have to guess. */

    /* as far as testing shows, only PCM 16/24LE (44100Hz - 192kHz) is supported
     * Tested on Realtek High Definition Audio, (Realtek Semiconductor Corp. 6.0.1.6312)
     * Drivers dated 2/18/2011
     */

    /* try float first for non-16bit audio */
    if (bits != 16) {
        bits = 32;
    }

    int start_bits = bits;
    while (1) { // not infinite -- returns at bottom
        for (; bits > 8; bits -= 8) {
            int samplerate = ao->samplerate;
            if (try_format(state, ao, bits, samplerate, ao->channels)) {
                return 0;
            }

            // make samplerate fit in [44100 192000]
            // we check for samplerate > 96k so that we can upsample instead of downsampling later
            if (samplerate < 44100 || samplerate > 96000) {
                if (samplerate < 44100)
                    samplerate = 44100;
                if (samplerate > 96000)
                    samplerate = 192000;

                if (try_format(state, ao, bits, samplerate, ao->channels)) {
                    return 0;
                }
            }

            // try bounding to 96kHz
            if (samplerate > 48000) {
                samplerate = 96000;
                if (try_format(state, ao, bits, samplerate, ao->channels)) {
                    return 0;
                }
            }

            // try bounding to 48kHz
            if (samplerate > 44100) {
                samplerate = 48000;
                if (try_format(state, ao, bits, samplerate, ao->channels)) {
                    return 0;
                }
            }

            /* How bad is this? try 44100hz, but only on 16bit */
            if (bits == 16 && samplerate != 44100) {
                samplerate = 44100;

                if (try_format(state, ao, bits, samplerate, ao->channels)) {
                    return 0;
                }
            }
        }
        if (ao->channels.num > 6) {
            /* Maybe this is 5.1 hardware with no support for more. */
            bits = start_bits;
            mp_chmap_from_channels(&ao->channels, 6);
        } else if (ao->channels.num != 2) {
            /* Poor quality hardware? Try stereo mode, go through the list again. */
            bits = start_bits;
            mp_chmap_from_channels(&ao->channels, 2);
        } else {
            EnterCriticalSection(&state->print_lock);
            mp_msg(MSGT_AO, MSGL_ERR, "ao-wasapi: couldn't find acceptable audio format!\n");
            LeaveCriticalSection(&state->print_lock);
            return -1;
        }
    }
}

static int fix_format(struct wasapi0_state *state)
{
    HRESULT hr;
    double offset = 0.5;

    /* cargo cult code to negotiate buffer block size, affected by hardware/drivers combinations,
       gradually grow it to 10s, by 0.5s, consider failure if it still doesn't work
     */
    hr = IAudioClient_GetDevicePeriod(state->pAudioClient,
                                      &state->defaultRequestedDuration,
                                      &state->minRequestedDuration);
reinit:
    hr = IAudioClient_Initialize(state->pAudioClient,
                                 state->share_mode,
                                 AUDCLNT_STREAMFLAGS_EVENTCALLBACK,
                                 state->defaultRequestedDuration,
                                 state->defaultRequestedDuration,
                                 &(state->format.Format),
                                 NULL);
    /* something about buffer sizes on Win7, fixme might loop forever */
    if (hr == AUDCLNT_E_BUFFER_SIZE_NOT_ALIGNED) {
        EnterCriticalSection(&state->print_lock);
        mp_msg(
            MSGT_AO, MSGL_V,
            "ao-wasapi: IAudioClient::Initialize negotiation failed with %s, used %lld * 100ns\n",
            explain_err(hr), state->defaultRequestedDuration);
        LeaveCriticalSection(&state->print_lock);
        if (offset > 10.0)
            goto exit_label;                /* is 10 enough to break out of the loop?*/
        IAudioClient_GetBufferSize(state->pAudioClient, &state->bufferFrameCount);
        state->defaultRequestedDuration =
            (REFERENCE_TIME)((BUFFER_TIME / state->format.Format.nSamplesPerSec *
                              state->bufferFrameCount) + offset);
        offset += 0.5;
        IAudioClient_Release(state->pAudioClient);
        state->pAudioClient = NULL;
        hr = IMMDeviceActivator_Activate(state->pDevice,
                                         &IID_IAudioClient, CLSCTX_ALL,
                                         NULL, (void **)&state->pAudioClient);
        goto reinit;
    }
    EXIT_ON_ERROR(hr)
    hr = IAudioClient_GetService(