/* * $Id: pa_win_wmme.c,v 1.1 2007-08-18 23:49:33 millerpuckette Exp $ * pa_win_wmme.c * Implementation of PortAudio for Windows MultiMedia Extensions (WMME) * * PortAudio Portable Real-Time Audio Library * Latest Version at: http://www.portaudio.com * * Authors: Ross Bencina and Phil Burk * Copyright (c) 1999-2000 Ross Bencina and Phil Burk * * Permission is hereby granted, free of charge, to any person obtaining * a copy of this software and associated documentation files * (the "Software"), to deal in the Software without restriction, * including without limitation the rights to use, copy, modify, merge, * publish, distribute, sublicense, and/or sell copies of the Software, * and to permit persons to whom the Software is furnished to do so, * subject to the following conditions: * * The above copyright notice and this permission notice shall be * included in all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ /* * The text above constitutes the entire PortAudio license; however, * the PortAudio community also makes the following non-binding requests: * * Any person wishing to distribute modifications to the Software is * requested to send the modifications to the original developer so that * they can be incorporated into the canonical version. It is also * requested that these non-binding requests be included along with the * license above. */ /* Modification History: PLB = Phil Burk JM = Julien Maillard RDB = Ross Bencina PLB20010402 - sDevicePtrs now allocates based on sizeof(pointer) PLB20010413 - check for excessive numbers of channels PLB20010422 - apply Mike Berry's changes for CodeWarrior on PC including conditional inclusion of memory.h, and explicit typecasting on memory allocation PLB20010802 - use GlobalAlloc for sDevicesPtr instead of PaHost_AllocFastMemory PLB20010816 - pass process instead of thread to SetPriorityClass() PLB20010927 - use number of frames instead of real-time for CPULoad calculation. JM20020118 - prevent hung thread when buffers underflow. PLB20020321 - detect Win XP versus NT, 9x; fix DBUG typo; removed init of CurrentCount RDB20020411 - various renaming cleanups, factored streamData alloc and cpu usage init RDB20020417 - stopped counting WAVE_MAPPER when there were no real devices refactoring, renaming and fixed a few edge case bugs RDB20020531 - converted to V19 framework ** NOTE maintanance history is now stored in CVS ** */ /** @file @ingroup hostaip_src @todo Fix buffer catch up code, can sometimes get stuck (perhaps fixed now, needs to be reviewed and tested.) @todo implement paInputUnderflow, paOutputOverflow streamCallback statusFlags, paNeverDropInput. @todo BUG: PA_MME_SET_LAST_WAVEIN/OUT_ERROR is used in functions which may be called asynchronously from the callback thread. this is bad. @todo implement inputBufferAdcTime in callback thread @todo review/fix error recovery and cleanup in marked functions @todo implement timeInfo for stream priming @todo handle the case where the callback returns paAbort or paComplete during stream priming. @todo review input overflow and output underflow handling in ReadStream and WriteStream Non-critical stuff for the future: @todo Investigate supporting host buffer formats > 16 bits @todo define UNICODE and _UNICODE in the project settings and see what breaks */ /* How it works: For both callback and blocking read/write streams we open the MME devices in CALLBACK_EVENT mode. In this mode, MME signals an Event object whenever it has finished with a buffer (either filled it for input, or played it for output). Where necessary we block waiting for Event objects using WaitMultipleObjects(). When implementing a PA callback stream, we set up a high priority thread which waits on the MME buffer Events and drains/fills the buffers when they are ready. When implementing a PA blocking read/write stream, we simply wait on these Events (when necessary) inside the ReadStream() and WriteStream() functions. */ #include #include #include #include #include #ifndef UNDER_CE #include #endif #include /* PLB20010422 - "memory.h" doesn't work on CodeWarrior for PC. Thanks Mike Berry for the mod. */ #ifndef __MWERKS__ #include #include #endif /* __MWERKS__ */ #include "portaudio.h" #include "pa_trace.h" #include "pa_util.h" #include "pa_allocation.h" #include "pa_hostapi.h" #include "pa_stream.h" #include "pa_cpuload.h" #include "pa_process.h" #include "pa_debugprint.h" #include "pa_win_wmme.h" #if (defined(UNDER_CE)) #pragma comment(lib, "Coredll.lib") #elif (defined(WIN32) && (defined(_MSC_VER) && (_MSC_VER >= 1200))) /* MSC version 6 and above */ #pragma comment(lib, "winmm.lib") #endif /* provided in newer platform sdks */ #ifndef DWORD_PTR #define DWORD_PTR DWORD #endif /************************************************* Constants ********/ #define PA_MME_USE_HIGH_DEFAULT_LATENCY_ (0) /* For debugging glitches. */ #if PA_MME_USE_HIGH_DEFAULT_LATENCY_ #define PA_MME_WIN_9X_DEFAULT_LATENCY_ (0.4) #define PA_MME_MIN_HOST_OUTPUT_BUFFER_COUNT_ (4) #define PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_ (4) #define PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_HALF_DUPLEX_ (4) #define PA_MME_MIN_HOST_BUFFER_FRAMES_WHEN_UNSPECIFIED_ (16) #define PA_MME_MAX_HOST_BUFFER_SECS_ (0.3) /* Do not exceed unless user buffer exceeds */ #define PA_MME_MAX_HOST_BUFFER_BYTES_ (32 * 1024) /* Has precedence over PA_MME_MAX_HOST_BUFFER_SECS_, some drivers are known to crash with buffer sizes > 32k */ #else #define PA_MME_WIN_9X_DEFAULT_LATENCY_ (0.2) #define PA_MME_MIN_HOST_OUTPUT_BUFFER_COUNT_ (2) #define PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_ (3) #define PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_HALF_DUPLEX_ (2) #define PA_MME_MIN_HOST_BUFFER_FRAMES_WHEN_UNSPECIFIED_ (16) #define PA_MME_MAX_HOST_BUFFER_SECS_ (0.1) /* Do not exceed unless user buffer exceeds */ #define PA_MME_MAX_HOST_BUFFER_BYTES_ (32 * 1024) /* Has precedence over PA_MME_MAX_HOST_BUFFER_SECS_, some drivers are known to crash with buffer sizes > 32k */ #endif /* Use higher latency for NT because it is even worse at real-time operation than Win9x. */ #define PA_MME_WIN_NT_DEFAULT_LATENCY_ (PA_MME_WIN_9X_DEFAULT_LATENCY_ * 2) #define PA_MME_WIN_WDM_DEFAULT_LATENCY_ (PA_MME_WIN_9X_DEFAULT_LATENCY_) #define PA_MME_MIN_TIMEOUT_MSEC_ (1000) static const char constInputMapperSuffix_[] = " - Input"; static const char constOutputMapperSuffix_[] = " - Output"; /********************************************************************/ typedef struct PaWinMmeStream PaWinMmeStream; /* forward declaration */ /* prototypes for functions declared in this file */ #ifdef __cplusplus extern "C" { #endif /* __cplusplus */ PaError PaWinMme_Initialize( PaUtilHostApiRepresentation **hostApi, PaHostApiIndex index ); #ifdef __cplusplus } #endif /* __cplusplus */ static void Terminate( struct PaUtilHostApiRepresentation *hostApi ); static PaError OpenStream( struct PaUtilHostApiRepresentation *hostApi, PaStream** stream, const PaStreamParameters *inputParameters, const PaStreamParameters *outputParameters, double sampleRate, unsigned long framesPerBuffer, PaStreamFlags streamFlags, PaStreamCallback *streamCallback, void *userData ); static PaError IsFormatSupported( struct PaUtilHostApiRepresentation *hostApi, const PaStreamParameters *inputParameters, const PaStreamParameters *outputParameters, double sampleRate ); static PaError CloseStream( PaStream* stream ); static PaError StartStream( PaStream *stream ); static PaError StopStream( PaStream *stream ); static PaError AbortStream( PaStream *stream ); static PaError IsStreamStopped( PaStream *s ); static PaError IsStreamActive( PaStream *stream ); static PaTime GetStreamTime( PaStream *stream ); static double GetStreamCpuLoad( PaStream* stream ); static PaError ReadStream( PaStream* stream, void *buffer, unsigned long frames ); static PaError WriteStream( PaStream* stream, const void *buffer, unsigned long frames ); static signed long GetStreamReadAvailable( PaStream* stream ); static signed long GetStreamWriteAvailable( PaStream* stream ); /* macros for setting last host error information */ #ifdef UNICODE #define PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ) \ { \ wchar_t mmeErrorTextWide[ MAXERRORLENGTH ]; \ char mmeErrorText[ MAXERRORLENGTH ]; \ waveInGetErrorText( mmresult, mmeErrorTextWide, MAXERRORLENGTH ); \ WideCharToMultiByte( CP_ACP, WC_COMPOSITECHECK | WC_DEFAULTCHAR,\ mmeErrorTextWide, -1, mmeErrorText, MAXERRORLENGTH, NULL, NULL ); \ PaUtil_SetLastHostErrorInfo( paMME, mmresult, mmeErrorText ); \ } #define PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ) \ { \ wchar_t mmeErrorTextWide[ MAXERRORLENGTH ]; \ char mmeErrorText[ MAXERRORLENGTH ]; \ waveOutGetErrorText( mmresult, mmeErrorTextWide, MAXERRORLENGTH ); \ WideCharToMultiByte( CP_ACP, WC_COMPOSITECHECK | WC_DEFAULTCHAR,\ mmeErrorTextWide, -1, mmeErrorText, MAXERRORLENGTH, NULL, NULL ); \ PaUtil_SetLastHostErrorInfo( paMME, mmresult, mmeErrorText ); \ } #else /* !UNICODE */ #define PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ) \ { \ char mmeErrorText[ MAXERRORLENGTH ]; \ waveInGetErrorText( mmresult, mmeErrorText, MAXERRORLENGTH ); \ PaUtil_SetLastHostErrorInfo( paMME, mmresult, mmeErrorText ); \ } #define PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ) \ { \ char mmeErrorText[ MAXERRORLENGTH ]; \ waveOutGetErrorText( mmresult, mmeErrorText, MAXERRORLENGTH ); \ PaUtil_SetLastHostErrorInfo( paMME, mmresult, mmeErrorText ); \ } #endif /* UNICODE */ static void PaMme_SetLastSystemError( DWORD errorCode ) { char *lpMsgBuf; FormatMessage( FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM, NULL, errorCode, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPTSTR) &lpMsgBuf, 0, NULL ); PaUtil_SetLastHostErrorInfo( paMME, errorCode, lpMsgBuf ); LocalFree( lpMsgBuf ); } #define PA_MME_SET_LAST_SYSTEM_ERROR( errorCode ) \ PaMme_SetLastSystemError( errorCode ) /* PaError returning wrappers for some commonly used win32 functions note that we allow passing a null ptr to have no effect. */ static PaError CreateEventWithPaError( HANDLE *handle, LPSECURITY_ATTRIBUTES lpEventAttributes, BOOL bManualReset, BOOL bInitialState, LPCTSTR lpName ) { PaError result = paNoError; *handle = NULL; *handle = CreateEvent( lpEventAttributes, bManualReset, bInitialState, lpName ); if( *handle == NULL ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_SYSTEM_ERROR( GetLastError() ); } return result; } static PaError ResetEventWithPaError( HANDLE handle ) { PaError result = paNoError; if( handle ) { if( ResetEvent( handle ) == 0 ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_SYSTEM_ERROR( GetLastError() ); } } return result; } static PaError CloseHandleWithPaError( HANDLE handle ) { PaError result = paNoError; if( handle ) { if( CloseHandle( handle ) == 0 ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_SYSTEM_ERROR( GetLastError() ); } } return result; } /* PaWinMmeHostApiRepresentation - host api datastructure specific to this implementation */ typedef struct { PaUtilHostApiRepresentation inheritedHostApiRep; PaUtilStreamInterface callbackStreamInterface; PaUtilStreamInterface blockingStreamInterface; PaUtilAllocationGroup *allocations; int inputDeviceCount, outputDeviceCount; /** winMmeDeviceIds is an array of WinMme device ids. fields in the range [0, inputDeviceCount) are input device ids, and [inputDeviceCount, inputDeviceCount + outputDeviceCount) are output device ids. */ UINT *winMmeDeviceIds; } PaWinMmeHostApiRepresentation; typedef struct { PaDeviceInfo inheritedDeviceInfo; DWORD dwFormats; /**<< standard formats bitmask from the WAVEINCAPS and WAVEOUTCAPS structures */ } PaWinMmeDeviceInfo; /************************************************************************* * Returns recommended device ID. * On the PC, the recommended device can be specified by the user by * setting an environment variable. For example, to use device #1. * * set PA_RECOMMENDED_OUTPUT_DEVICE=1 * * The user should first determine the available device ID by using * the supplied application "pa_devs". */ #define PA_ENV_BUF_SIZE_ (32) #define PA_REC_IN_DEV_ENV_NAME_ ("PA_RECOMMENDED_INPUT_DEVICE") #define PA_REC_OUT_DEV_ENV_NAME_ ("PA_RECOMMENDED_OUTPUT_DEVICE") static PaDeviceIndex GetEnvDefaultDeviceID( char *envName ) { PaDeviceIndex recommendedIndex = paNoDevice; DWORD hresult; char envbuf[PA_ENV_BUF_SIZE_]; #ifndef WIN32_PLATFORM_PSPC /* no GetEnvironmentVariable on PocketPC */ /* Let user determine default device by setting environment variable. */ hresult = GetEnvironmentVariable( envName, envbuf, PA_ENV_BUF_SIZE_ ); if( (hresult > 0) && (hresult < PA_ENV_BUF_SIZE_) ) { recommendedIndex = atoi( envbuf ); } #endif return recommendedIndex; } static void InitializeDefaultDeviceIdsFromEnv( PaWinMmeHostApiRepresentation *hostApi ) { PaDeviceIndex device; /* input */ device = GetEnvDefaultDeviceID( PA_REC_IN_DEV_ENV_NAME_ ); if( device != paNoDevice && ( device >= 0 && device < hostApi->inheritedHostApiRep.info.deviceCount ) && hostApi->inheritedHostApiRep.deviceInfos[ device ]->maxInputChannels > 0 ) { hostApi->inheritedHostApiRep.info.defaultInputDevice = device; } /* output */ device = GetEnvDefaultDeviceID( PA_REC_OUT_DEV_ENV_NAME_ ); if( device != paNoDevice && ( device >= 0 && device < hostApi->inheritedHostApiRep.info.deviceCount ) && hostApi->inheritedHostApiRep.deviceInfos[ device ]->maxOutputChannels > 0 ) { hostApi->inheritedHostApiRep.info.defaultOutputDevice = device; } } /** Convert external PA ID to a windows multimedia device ID */ static UINT LocalDeviceIndexToWinMmeDeviceId( PaWinMmeHostApiRepresentation *hostApi, PaDeviceIndex device ) { assert( device >= 0 && device < hostApi->inputDeviceCount + hostApi->outputDeviceCount ); return hostApi->winMmeDeviceIds[ device ]; } static PaError QueryInputWaveFormatEx( int deviceId, WAVEFORMATEX *waveFormatEx ) { MMRESULT mmresult; switch( mmresult = waveInOpen( NULL, deviceId, waveFormatEx, 0, 0, WAVE_FORMAT_QUERY ) ) { case MMSYSERR_NOERROR: return paNoError; case MMSYSERR_ALLOCATED: /* Specified resource is already allocated. */ return paDeviceUnavailable; case MMSYSERR_NODRIVER: /* No device driver is present. */ return paDeviceUnavailable; case MMSYSERR_NOMEM: /* Unable to allocate or lock memory. */ return paInsufficientMemory; case WAVERR_BADFORMAT: /* Attempted to open with an unsupported waveform-audio format. */ return paSampleFormatNotSupported; case MMSYSERR_BADDEVICEID: /* Specified device identifier is out of range. */ /* falls through */ default: PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); return paUnanticipatedHostError; } } static PaError QueryOutputWaveFormatEx( int deviceId, WAVEFORMATEX *waveFormatEx ) { MMRESULT mmresult; switch( mmresult = waveOutOpen( NULL, deviceId, waveFormatEx, 0, 0, WAVE_FORMAT_QUERY ) ) { case MMSYSERR_NOERROR: return paNoError; case MMSYSERR_ALLOCATED: /* Specified resource is already allocated. */ return paDeviceUnavailable; case MMSYSERR_NODRIVER: /* No device driver is present. */ return paDeviceUnavailable; case MMSYSERR_NOMEM: /* Unable to allocate or lock memory. */ return paInsufficientMemory; case WAVERR_BADFORMAT: /* Attempted to open with an unsupported waveform-audio format. */ return paSampleFormatNotSupported; case MMSYSERR_BADDEVICEID: /* Specified device identifier is out of range. */ /* falls through */ default: PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); return paUnanticipatedHostError; } } static PaError QueryFormatSupported( PaDeviceInfo *deviceInfo, PaError (*waveFormatExQueryFunction)(int, WAVEFORMATEX*), int winMmeDeviceId, int channels, double sampleRate ) { PaWinMmeDeviceInfo *winMmeDeviceInfo = (PaWinMmeDeviceInfo*)deviceInfo; WAVEFORMATEX waveFormatEx; if( sampleRate == 11025.0 && ( (channels == 1 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_1M16)) || (channels == 2 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_1S16)) ) ){ return paNoError; } if( sampleRate == 22050.0 && ( (channels == 1 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_2M16)) || (channels == 2 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_2S16)) ) ){ return paNoError; } if( sampleRate == 44100.0 && ( (channels == 1 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_4M16)) || (channels == 2 && (winMmeDeviceInfo->dwFormats & WAVE_FORMAT_4S16)) ) ){ return paNoError; } waveFormatEx.wFormatTag = WAVE_FORMAT_PCM; waveFormatEx.nChannels = (WORD)channels; waveFormatEx.nSamplesPerSec = (DWORD)sampleRate; waveFormatEx.nAvgBytesPerSec = waveFormatEx.nSamplesPerSec * channels * sizeof(short); waveFormatEx.nBlockAlign = (WORD)(channels * sizeof(short)); waveFormatEx.wBitsPerSample = 16; waveFormatEx.cbSize = 0; return waveFormatExQueryFunction( winMmeDeviceId, &waveFormatEx ); } #define PA_DEFAULTSAMPLERATESEARCHORDER_COUNT_ (13) /* must match array length below */ static double defaultSampleRateSearchOrder_[] = { 44100.0, 48000.0, 32000.0, 24000.0, 22050.0, 88200.0, 96000.0, 192000.0, 16000.0, 12000.0, 11025.0, 9600.0, 8000.0 }; static void DetectDefaultSampleRate( PaWinMmeDeviceInfo *winMmeDeviceInfo, int winMmeDeviceId, PaError (*waveFormatExQueryFunction)(int, WAVEFORMATEX*), int maxChannels ) { PaDeviceInfo *deviceInfo = &winMmeDeviceInfo->inheritedDeviceInfo; int i; deviceInfo->defaultSampleRate = 0.; for( i=0; i < PA_DEFAULTSAMPLERATESEARCHORDER_COUNT_; ++i ) { double sampleRate = defaultSampleRateSearchOrder_[ i ]; PaError paerror = QueryFormatSupported( deviceInfo, waveFormatExQueryFunction, winMmeDeviceId, maxChannels, sampleRate ); if( paerror == paNoError ) { deviceInfo->defaultSampleRate = sampleRate; break; } } } static PaError InitializeInputDeviceInfo( PaWinMmeHostApiRepresentation *winMmeHostApi, PaWinMmeDeviceInfo *winMmeDeviceInfo, UINT winMmeInputDeviceId, int *success ) { PaError result = paNoError; char *deviceName; /* non-const ptr */ MMRESULT mmresult; WAVEINCAPS wic; PaDeviceInfo *deviceInfo = &winMmeDeviceInfo->inheritedDeviceInfo; *success = 0; mmresult = waveInGetDevCaps( winMmeInputDeviceId, &wic, sizeof( WAVEINCAPS ) ); if( mmresult == MMSYSERR_NOMEM ) { result = paInsufficientMemory; goto error; } else if( mmresult != MMSYSERR_NOERROR ) { /* instead of returning paUnanticipatedHostError we return paNoError, but leave success set as 0. This allows Pa_Initialize to just ignore this device, without failing the entire initialisation process. */ return paNoError; } if( winMmeInputDeviceId == WAVE_MAPPER ) { /* Append I/O suffix to WAVE_MAPPER device. */ deviceName = (char *)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, strlen( wic.szPname ) + 1 + sizeof(constInputMapperSuffix_) ); if( !deviceName ) { result = paInsufficientMemory; goto error; } strcpy( deviceName, wic.szPname ); strcat( deviceName, constInputMapperSuffix_ ); } else { deviceName = (char*)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, strlen( wic.szPname ) + 1 ); if( !deviceName ) { result = paInsufficientMemory; goto error; } strcpy( deviceName, wic.szPname ); } deviceInfo->name = deviceName; deviceInfo->maxInputChannels = wic.wChannels; /* Sometimes a device can return a rediculously large number of channels. * This happened with an SBLive card on a Windows ME box. * If that happens, then force it to 2 channels. PLB20010413 */ if( (deviceInfo->maxInputChannels < 1) || (deviceInfo->maxInputChannels > 256) ) { PA_DEBUG(("Pa_GetDeviceInfo: Num input channels reported as %d! Changed to 2.\n", deviceInfo->maxInputChannels )); deviceInfo->maxInputChannels = 2; } winMmeDeviceInfo->dwFormats = wic.dwFormats; DetectDefaultSampleRate( winMmeDeviceInfo, winMmeInputDeviceId, QueryInputWaveFormatEx, deviceInfo->maxInputChannels ); *success = 1; error: return result; } static PaError InitializeOutputDeviceInfo( PaWinMmeHostApiRepresentation *winMmeHostApi, PaWinMmeDeviceInfo *winMmeDeviceInfo, UINT winMmeOutputDeviceId, int *success ) { PaError result = paNoError; char *deviceName; /* non-const ptr */ MMRESULT mmresult; WAVEOUTCAPS woc; PaDeviceInfo *deviceInfo = &winMmeDeviceInfo->inheritedDeviceInfo; *success = 0; mmresult = waveOutGetDevCaps( winMmeOutputDeviceId, &woc, sizeof( WAVEOUTCAPS ) ); if( mmresult == MMSYSERR_NOMEM ) { result = paInsufficientMemory; goto error; } else if( mmresult != MMSYSERR_NOERROR ) { /* instead of returning paUnanticipatedHostError we return paNoError, but leave success set as 0. This allows Pa_Initialize to just ignore this device, without failing the entire initialisation process. */ return paNoError; } if( winMmeOutputDeviceId == WAVE_MAPPER ) { /* Append I/O suffix to WAVE_MAPPER device. */ deviceName = (char *)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, strlen( woc.szPname ) + 1 + sizeof(constOutputMapperSuffix_) ); if( !deviceName ) { result = paInsufficientMemory; goto error; } strcpy( deviceName, woc.szPname ); strcat( deviceName, constOutputMapperSuffix_ ); } else { deviceName = (char*)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, strlen( woc.szPname ) + 1 ); if( !deviceName ) { result = paInsufficientMemory; goto error; } strcpy( deviceName, woc.szPname ); } deviceInfo->name = deviceName; deviceInfo->maxOutputChannels = woc.wChannels; /* Sometimes a device can return a rediculously large number of channels. * This happened with an SBLive card on a Windows ME box. * It also happens on Win XP! */ if( (deviceInfo->maxOutputChannels < 1) || (deviceInfo->maxOutputChannels > 256) ) { PA_DEBUG(("Pa_GetDeviceInfo: Num output channels reported as %d! Changed to 2.\n", deviceInfo->maxOutputChannels )); deviceInfo->maxOutputChannels = 2; } winMmeDeviceInfo->dwFormats = woc.dwFormats; DetectDefaultSampleRate( winMmeDeviceInfo, winMmeOutputDeviceId, QueryOutputWaveFormatEx, deviceInfo->maxOutputChannels ); *success = 1; error: return result; } static void GetDefaultLatencies( PaTime *defaultLowLatency, PaTime *defaultHighLatency ) { OSVERSIONINFO osvi; osvi.dwOSVersionInfoSize = sizeof( osvi ); GetVersionEx( &osvi ); /* Check for NT */ if( (osvi.dwMajorVersion == 4) && (osvi.dwPlatformId == 2) ) { *defaultLowLatency = PA_MME_WIN_NT_DEFAULT_LATENCY_; } else if(osvi.dwMajorVersion >= 5) { *defaultLowLatency = PA_MME_WIN_WDM_DEFAULT_LATENCY_; } else { *defaultLowLatency = PA_MME_WIN_9X_DEFAULT_LATENCY_; } *defaultHighLatency = *defaultLowLatency * 2; } PaError PaWinMme_Initialize( PaUtilHostApiRepresentation **hostApi, PaHostApiIndex hostApiIndex ) { PaError result = paNoError; int i; PaWinMmeHostApiRepresentation *winMmeHostApi; int inputDeviceCount, outputDeviceCount, maximumPossibleDeviceCount; PaWinMmeDeviceInfo *deviceInfoArray; int deviceInfoInitializationSucceeded; PaTime defaultLowLatency, defaultHighLatency; winMmeHostApi = (PaWinMmeHostApiRepresentation*)PaUtil_AllocateMemory( sizeof(PaWinMmeHostApiRepresentation) ); if( !winMmeHostApi ) { result = paInsufficientMemory; goto error; } winMmeHostApi->allocations = PaUtil_CreateAllocationGroup(); if( !winMmeHostApi->allocations ) { result = paInsufficientMemory; goto error; } *hostApi = &winMmeHostApi->inheritedHostApiRep; (*hostApi)->info.structVersion = 1; (*hostApi)->info.type = paMME; (*hostApi)->info.name = "MME"; /* initialise device counts and default devices under the assumption that there are no devices. These values are incremented below if and when devices are successfully initialized. */ (*hostApi)->info.deviceCount = 0; (*hostApi)->info.defaultInputDevice = paNoDevice; (*hostApi)->info.defaultOutputDevice = paNoDevice; winMmeHostApi->inputDeviceCount = 0; winMmeHostApi->outputDeviceCount = 0; maximumPossibleDeviceCount = 0; inputDeviceCount = waveInGetNumDevs(); if( inputDeviceCount > 0 ) maximumPossibleDeviceCount += inputDeviceCount + 1; /* assume there is a WAVE_MAPPER */ outputDeviceCount = waveOutGetNumDevs(); if( outputDeviceCount > 0 ) maximumPossibleDeviceCount += outputDeviceCount + 1; /* assume there is a WAVE_MAPPER */ if( maximumPossibleDeviceCount > 0 ){ (*hostApi)->deviceInfos = (PaDeviceInfo**)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, sizeof(PaDeviceInfo*) * maximumPossibleDeviceCount ); if( !(*hostApi)->deviceInfos ) { result = paInsufficientMemory; goto error; } /* allocate all device info structs in a contiguous block */ deviceInfoArray = (PaWinMmeDeviceInfo*)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, sizeof(PaWinMmeDeviceInfo) * maximumPossibleDeviceCount ); if( !deviceInfoArray ) { result = paInsufficientMemory; goto error; } winMmeHostApi->winMmeDeviceIds = (UINT*)PaUtil_GroupAllocateMemory( winMmeHostApi->allocations, sizeof(int) * maximumPossibleDeviceCount ); if( !winMmeHostApi->winMmeDeviceIds ) { result = paInsufficientMemory; goto error; } GetDefaultLatencies( &defaultLowLatency, &defaultHighLatency ); if( inputDeviceCount > 0 ){ /* -1 is the WAVE_MAPPER */ for( i = -1; i < inputDeviceCount; ++i ){ UINT winMmeDeviceId = (UINT)((i==-1) ? WAVE_MAPPER : i); PaWinMmeDeviceInfo *wmmeDeviceInfo = &deviceInfoArray[ (*hostApi)->info.deviceCount ]; PaDeviceInfo *deviceInfo = &wmmeDeviceInfo->inheritedDeviceInfo; deviceInfo->structVersion = 2; deviceInfo->hostApi = hostApiIndex; deviceInfo->maxInputChannels = 0; deviceInfo->maxOutputChannels = 0; deviceInfo->defaultLowInputLatency = defaultLowLatency; deviceInfo->defaultLowOutputLatency = defaultLowLatency; deviceInfo->defaultHighInputLatency = defaultHighLatency; deviceInfo->defaultHighOutputLatency = defaultHighLatency; result = InitializeInputDeviceInfo( winMmeHostApi, wmmeDeviceInfo, winMmeDeviceId, &deviceInfoInitializationSucceeded ); if( result != paNoError ) goto error; if( deviceInfoInitializationSucceeded ){ if( (*hostApi)->info.defaultInputDevice == paNoDevice ) (*hostApi)->info.defaultInputDevice = (*hostApi)->info.deviceCount; winMmeHostApi->winMmeDeviceIds[ (*hostApi)->info.deviceCount ] = winMmeDeviceId; (*hostApi)->deviceInfos[ (*hostApi)->info.deviceCount ] = deviceInfo; winMmeHostApi->inputDeviceCount++; (*hostApi)->info.deviceCount++; } } } if( outputDeviceCount > 0 ){ /* -1 is the WAVE_MAPPER */ for( i = -1; i < outputDeviceCount; ++i ){ UINT winMmeDeviceId = (UINT)((i==-1) ? WAVE_MAPPER : i); PaWinMmeDeviceInfo *wmmeDeviceInfo = &deviceInfoArray[ (*hostApi)->info.deviceCount ]; PaDeviceInfo *deviceInfo = &wmmeDeviceInfo->inheritedDeviceInfo; deviceInfo->structVersion = 2; deviceInfo->hostApi = hostApiIndex; deviceInfo->maxInputChannels = 0; deviceInfo->maxOutputChannels = 0; deviceInfo->defaultLowInputLatency = defaultLowLatency; deviceInfo->defaultLowOutputLatency = defaultLowLatency; deviceInfo->defaultHighInputLatency = defaultHighLatency; deviceInfo->defaultHighOutputLatency = defaultHighLatency; result = InitializeOutputDeviceInfo( winMmeHostApi, wmmeDeviceInfo, winMmeDeviceId, &deviceInfoInitializationSucceeded ); if( result != paNoError ) goto error; if( deviceInfoInitializationSucceeded ){ if( (*hostApi)->info.defaultOutputDevice == paNoDevice ) (*hostApi)->info.defaultOutputDevice = (*hostApi)->info.deviceCount; winMmeHostApi->winMmeDeviceIds[ (*hostApi)->info.deviceCount ] = winMmeDeviceId; (*hostApi)->deviceInfos[ (*hostApi)->info.deviceCount ] = deviceInfo; winMmeHostApi->outputDeviceCount++; (*hostApi)->info.deviceCount++; } } } } InitializeDefaultDeviceIdsFromEnv( winMmeHostApi ); (*hostApi)->Terminate = Terminate; (*hostApi)->OpenStream = OpenStream; (*hostApi)->IsFormatSupported = IsFormatSupported; PaUtil_InitializeStreamInterface( &winMmeHostApi->callbackStreamInterface, CloseStream, StartStream, StopStream, AbortStream, IsStreamStopped, IsStreamActive, GetStreamTime, GetStreamCpuLoad, PaUtil_DummyRead, PaUtil_DummyWrite, PaUtil_DummyGetReadAvailable, PaUtil_DummyGetWriteAvailable ); PaUtil_InitializeStreamInterface( &winMmeHostApi->blockingStreamInterface, CloseStream, StartStream, StopStream, AbortStream, IsStreamStopped, IsStreamActive, GetStreamTime, PaUtil_DummyGetCpuLoad, ReadStream, WriteStream, GetStreamReadAvailable, GetStreamWriteAvailable ); return result; error: if( winMmeHostApi ) { if( winMmeHostApi->allocations ) { PaUtil_FreeAllAllocations( winMmeHostApi->allocations ); PaUtil_DestroyAllocationGroup( winMmeHostApi->allocations ); } PaUtil_FreeMemory( winMmeHostApi ); } return result; } static void Terminate( struct PaUtilHostApiRepresentation *hostApi ) { PaWinMmeHostApiRepresentation *winMmeHostApi = (PaWinMmeHostApiRepresentation*)hostApi; if( winMmeHostApi->allocations ) { PaUtil_FreeAllAllocations( winMmeHostApi->allocations ); PaUtil_DestroyAllocationGroup( winMmeHostApi->allocations ); } PaUtil_FreeMemory( winMmeHostApi ); } static PaError IsFormatSupported( struct PaUtilHostApiRepresentation *hostApi, const PaStreamParameters *inputParameters, const PaStreamParameters *outputParameters, double sampleRate ) { PaWinMmeHostApiRepresentation *winMmeHostApi = (PaWinMmeHostApiRepresentation*)hostApi; PaDeviceInfo *inputDeviceInfo, *outputDeviceInfo; int inputChannelCount, outputChannelCount; int inputMultipleDeviceChannelCount, outputMultipleDeviceChannelCount; PaSampleFormat inputSampleFormat, outputSampleFormat; PaWinMmeStreamInfo *inputStreamInfo, *outputStreamInfo; UINT winMmeInputDeviceId, winMmeOutputDeviceId; unsigned int i; PaError paerror; /* The calls to QueryFormatSupported below are intended to detect invalid sample rates. If we assume that the channel count and format are OK, then the only thing that could fail is the sample rate. This isn't strictly true, but I can't think of a better way to test that the sample rate is valid. */ if( inputParameters ) { inputChannelCount = inputParameters->channelCount; inputSampleFormat = inputParameters->sampleFormat; inputStreamInfo = inputParameters->hostApiSpecificStreamInfo; /* all standard sample formats are supported by the buffer adapter, this implementation doesn't support any custom sample formats */ if( inputSampleFormat & paCustomFormat ) return paSampleFormatNotSupported; if( inputParameters->device == paUseHostApiSpecificDeviceSpecification && inputStreamInfo && (inputStreamInfo->flags & paWinMmeUseMultipleDevices) ) { inputMultipleDeviceChannelCount = 0; for( i=0; i< inputStreamInfo->deviceCount; ++i ) { inputMultipleDeviceChannelCount += inputStreamInfo->devices[i].channelCount; inputDeviceInfo = hostApi->deviceInfos[ inputStreamInfo->devices[i].device ]; /* check that input device can support inputChannelCount */ if( inputStreamInfo->devices[i].channelCount <= 0 || inputStreamInfo->devices[i].channelCount > inputDeviceInfo->maxInputChannels ) return paInvalidChannelCount; /* test for valid sample rate, see comment above */ winMmeInputDeviceId = LocalDeviceIndexToWinMmeDeviceId( winMmeHostApi, inputStreamInfo->devices[i].device ); paerror = QueryFormatSupported( inputDeviceInfo, QueryInputWaveFormatEx, winMmeInputDeviceId, inputStreamInfo->devices[i].channelCount, sampleRate ); if( paerror != paNoError ) return paInvalidSampleRate; } if( inputMultipleDeviceChannelCount != inputChannelCount ) return paIncompatibleHostApiSpecificStreamInfo; } else { if( inputStreamInfo && (inputStreamInfo->flags & paWinMmeUseMultipleDevices) ) return paIncompatibleHostApiSpecificStreamInfo; /* paUseHostApiSpecificDeviceSpecification was not supplied as the input device */ inputDeviceInfo = hostApi->deviceInfos[ inputParameters->device ]; /* check that input device can support inputChannelCount */ if( inputChannelCount > inputDeviceInfo->maxInputChannels ) return paInvalidChannelCount; /* test for valid sample rate, see comment above */ winMmeInputDeviceId = LocalDeviceIndexToWinMmeDeviceId( winMmeHostApi, inputParameters->device ); paerror = QueryFormatSupported( inputDeviceInfo, QueryInputWaveFormatEx, winMmeInputDeviceId, inputChannelCount, sampleRate ); if( paerror != paNoError ) return paInvalidSampleRate; } } if( outputParameters ) { outputChannelCount = outputParameters->channelCount; outputSampleFormat = outputParameters->sampleFormat; outputStreamInfo = outputParameters->hostApiSpecificStreamInfo; /* all standard sample formats are supported by the buffer adapter, this implementation doesn't support any custom sample formats */ if( outputSampleFormat & paCustomFormat ) return paSampleFormatNotSupported; if( outputParameters->device == paUseHostApiSpecificDeviceSpecification && outputStreamInfo && (outputStreamInfo->flags & paWinMmeUseMultipleDevices) ) { outputMultipleDeviceChannelCount = 0; for( i=0; i< outputStreamInfo->deviceCount; ++i ) { outputMultipleDeviceChannelCount += outputStreamInfo->devices[i].channelCount; outputDeviceInfo = hostApi->deviceInfos[ outputStreamInfo->devices[i].device ]; /* check that output device can support outputChannelCount */ if( outputStreamInfo->devices[i].channelCount <= 0 || outputStreamInfo->devices[i].channelCount > outputDeviceInfo->maxOutputChannels ) return paInvalidChannelCount; /* test for valid sample rate, see comment above */ winMmeOutputDeviceId = LocalDeviceIndexToWinMmeDeviceId( winMmeHostApi, outputStreamInfo->devices[i].device ); paerror = QueryFormatSupported( outputDeviceInfo, QueryOutputWaveFormatEx, winMmeOutputDeviceId, outputStreamInfo->devices[i].channelCount, sampleRate ); if( paerror != paNoError ) return paInvalidSampleRate; } if( outputMultipleDeviceChannelCount != outputChannelCount ) return paIncompatibleHostApiSpecificStreamInfo; } else { if( outputStreamInfo && (outputStreamInfo->flags & paWinMmeUseMultipleDevices) ) return paIncompatibleHostApiSpecificStreamInfo; /* paUseHostApiSpecificDeviceSpecification was not supplied as the output device */ outputDeviceInfo = hostApi->deviceInfos[ outputParameters->device ]; /* check that output device can support outputChannelCount */ if( outputChannelCount > outputDeviceInfo->maxOutputChannels ) return paInvalidChannelCount; /* test for valid sample rate, see comment above */ winMmeOutputDeviceId = LocalDeviceIndexToWinMmeDeviceId( winMmeHostApi, outputParameters->device ); paerror = QueryFormatSupported( outputDeviceInfo, QueryOutputWaveFormatEx, winMmeOutputDeviceId, outputChannelCount, sampleRate ); if( paerror != paNoError ) return paInvalidSampleRate; } } /* - if a full duplex stream is requested, check that the combination of input and output parameters is supported - check that the device supports sampleRate for mme all we can do is test that the input and output devices support the requested sample rate and number of channels. we cannot test for full duplex compatibility. */ return paFormatIsSupported; } static void SelectBufferSizeAndCount( unsigned long baseBufferSize, unsigned long requestedLatency, unsigned long baseBufferCount, unsigned long minimumBufferCount, unsigned long maximumBufferSize, unsigned long *hostBufferSize, unsigned long *hostBufferCount ) { unsigned long sizeMultiplier, bufferCount, latency; unsigned long nextLatency, nextBufferSize; int baseBufferSizeIsPowerOfTwo; sizeMultiplier = 1; bufferCount = baseBufferCount; /* count-1 below because latency is always determined by one less than the total number of buffers. */ latency = (baseBufferSize * sizeMultiplier) * (bufferCount-1); if( latency > requestedLatency ) { /* reduce number of buffers without falling below suggested latency */ nextLatency = (baseBufferSize * sizeMultiplier) * (bufferCount-2); while( bufferCount > minimumBufferCount && nextLatency >= requestedLatency ) { --bufferCount; nextLatency = (baseBufferSize * sizeMultiplier) * (bufferCount-2); } }else if( latency < requestedLatency ){ baseBufferSizeIsPowerOfTwo = (! (baseBufferSize & (baseBufferSize - 1))); if( baseBufferSizeIsPowerOfTwo ){ /* double size of buffers without exceeding requestedLatency */ nextBufferSize = (baseBufferSize * (sizeMultiplier*2)); nextLatency = nextBufferSize * (bufferCount-1); while( nextBufferSize <= maximumBufferSize && nextLatency < requestedLatency ) { sizeMultiplier *= 2; nextBufferSize = (baseBufferSize * (sizeMultiplier*2)); nextLatency = nextBufferSize * (bufferCount-1); } }else{ /* increase size of buffers upto first excess of requestedLatency */ nextBufferSize = (baseBufferSize * (sizeMultiplier+1)); nextLatency = nextBufferSize * (bufferCount-1); while( nextBufferSize <= maximumBufferSize && nextLatency < requestedLatency ) { ++sizeMultiplier; nextBufferSize = (baseBufferSize * (sizeMultiplier+1)); nextLatency = nextBufferSize * (bufferCount-1); } if( nextLatency < requestedLatency ) ++sizeMultiplier; } /* increase number of buffers until requestedLatency is reached */ latency = (baseBufferSize * sizeMultiplier) * (bufferCount-1); while( latency < requestedLatency ) { ++bufferCount; latency = (baseBufferSize * sizeMultiplier) * (bufferCount-1); } } *hostBufferSize = baseBufferSize * sizeMultiplier; *hostBufferCount = bufferCount; } static void ReselectBufferCount( unsigned long bufferSize, unsigned long requestedLatency, unsigned long baseBufferCount, unsigned long minimumBufferCount, unsigned long *hostBufferCount ) { unsigned long bufferCount, latency; unsigned long nextLatency; bufferCount = baseBufferCount; /* count-1 below because latency is always determined by one less than the total number of buffers. */ latency = bufferSize * (bufferCount-1); if( latency > requestedLatency ) { /* reduce number of buffers without falling below suggested latency */ nextLatency = bufferSize * (bufferCount-2); while( bufferCount > minimumBufferCount && nextLatency >= requestedLatency ) { --bufferCount; nextLatency = bufferSize * (bufferCount-2); } }else if( latency < requestedLatency ){ /* increase number of buffers until requestedLatency is reached */ latency = bufferSize * (bufferCount-1); while( latency < requestedLatency ) { ++bufferCount; latency = bufferSize * (bufferCount-1); } } *hostBufferCount = bufferCount; } /* CalculateBufferSettings() fills the framesPerHostInputBuffer, hostInputBufferCount, framesPerHostOutputBuffer and hostOutputBufferCount parameters based on the values of the other parameters. */ static PaError CalculateBufferSettings( unsigned long *framesPerHostInputBuffer, unsigned long *hostInputBufferCount, unsigned long *framesPerHostOutputBuffer, unsigned long *hostOutputBufferCount, int inputChannelCount, PaSampleFormat hostInputSampleFormat, PaTime suggestedInputLatency, PaWinMmeStreamInfo *inputStreamInfo, int outputChannelCount, PaSampleFormat hostOutputSampleFormat, PaTime suggestedOutputLatency, PaWinMmeStreamInfo *outputStreamInfo, double sampleRate, unsigned long framesPerBuffer ) { PaError result = paNoError; int effectiveInputChannelCount, effectiveOutputChannelCount; int hostInputFrameSize = 0; unsigned int i; if( inputChannelCount > 0 ) { int hostInputSampleSize = Pa_GetSampleSize( hostInputSampleFormat ); if( hostInputSampleSize < 0 ) { result = hostInputSampleSize; goto error; } if( inputStreamInfo && ( inputStreamInfo->flags & paWinMmeUseMultipleDevices ) ) { /* set effectiveInputChannelCount to the largest number of channels on any one device. */ effectiveInputChannelCount = 0; for( i=0; i< inputStreamInfo->deviceCount; ++i ) { if( inputStreamInfo->devices[i].channelCount > effectiveInputChannelCount ) effectiveInputChannelCount = inputStreamInfo->devices[i].channelCount; } } else { effectiveInputChannelCount = inputChannelCount; } hostInputFrameSize = hostInputSampleSize * effectiveInputChannelCount; if( inputStreamInfo && ( inputStreamInfo->flags & paWinMmeUseLowLevelLatencyParameters ) ) { if( inputStreamInfo->bufferCount <= 0 || inputStreamInfo->framesPerBuffer <= 0 ) { result = paIncompatibleHostApiSpecificStreamInfo; goto error; } *framesPerHostInputBuffer = inputStreamInfo->framesPerBuffer; *hostInputBufferCount = inputStreamInfo->bufferCount; } else { unsigned long hostBufferSizeBytes, hostBufferCount; unsigned long minimumBufferCount = (outputChannelCount > 0) ? PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_ : PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_HALF_DUPLEX_; unsigned long maximumBufferSize = (long) ((PA_MME_MAX_HOST_BUFFER_SECS_ * sampleRate) * hostInputFrameSize); if( maximumBufferSize > PA_MME_MAX_HOST_BUFFER_BYTES_ ) maximumBufferSize = PA_MME_MAX_HOST_BUFFER_BYTES_; /* compute the following in bytes, then convert back to frames */ SelectBufferSizeAndCount( ((framesPerBuffer == paFramesPerBufferUnspecified) ? PA_MME_MIN_HOST_BUFFER_FRAMES_WHEN_UNSPECIFIED_ : framesPerBuffer ) * hostInputFrameSize, /* baseBufferSize */ ((unsigned long)(suggestedInputLatency * sampleRate)) * hostInputFrameSize, /* suggestedLatency */ 4, /* baseBufferCount */ minimumBufferCount, maximumBufferSize, &hostBufferSizeBytes, &hostBufferCount ); *framesPerHostInputBuffer = hostBufferSizeBytes / hostInputFrameSize; *hostInputBufferCount = hostBufferCount; } } else { *framesPerHostInputBuffer = 0; *hostInputBufferCount = 0; } if( outputChannelCount > 0 ) { if( outputStreamInfo && ( outputStreamInfo->flags & paWinMmeUseLowLevelLatencyParameters ) ) { if( outputStreamInfo->bufferCount <= 0 || outputStreamInfo->framesPerBuffer <= 0 ) { result = paIncompatibleHostApiSpecificStreamInfo; goto error; } *framesPerHostOutputBuffer = outputStreamInfo->framesPerBuffer; *hostOutputBufferCount = outputStreamInfo->bufferCount; if( inputChannelCount > 0 ) /* full duplex */ { if( *framesPerHostInputBuffer != *framesPerHostOutputBuffer ) { if( inputStreamInfo && ( inputStreamInfo->flags & paWinMmeUseLowLevelLatencyParameters ) ) { /* a custom StreamInfo was used for specifying both input and output buffer sizes, the larger buffer size must be a multiple of the smaller buffer size */ if( *framesPerHostInputBuffer < *framesPerHostOutputBuffer ) { if( *framesPerHostOutputBuffer % *framesPerHostInputBuffer != 0 ) { result = paIncompatibleHostApiSpecificStreamInfo; goto error; } } else { assert( *framesPerHostInputBuffer > *framesPerHostOutputBuffer ); if( *framesPerHostInputBuffer % *framesPerHostOutputBuffer != 0 ) { result = paIncompatibleHostApiSpecificStreamInfo; goto error; } } } else { /* a custom StreamInfo was not used for specifying the input buffer size, so use the output buffer size, and approximately the same latency. */ *framesPerHostInputBuffer = *framesPerHostOutputBuffer; *hostInputBufferCount = (((unsigned long)(suggestedInputLatency * sampleRate)) / *framesPerHostInputBuffer) + 1; if( *hostInputBufferCount < PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_ ) *hostInputBufferCount = PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_; } } } } else { unsigned long hostBufferSizeBytes, hostBufferCount; unsigned long minimumBufferCount = PA_MME_MIN_HOST_OUTPUT_BUFFER_COUNT_; unsigned long maximumBufferSize; int hostOutputFrameSize; int hostOutputSampleSize; hostOutputSampleSize = Pa_GetSampleSize( hostOutputSampleFormat ); if( hostOutputSampleSize < 0 ) { result = hostOutputSampleSize; goto error; } if( outputStreamInfo && ( outputStreamInfo->flags & paWinMmeUseMultipleDevices ) ) { /* set effectiveOutputChannelCount to the largest number of channels on any one device. */ effectiveOutputChannelCount = 0; for( i=0; i< outputStreamInfo->deviceCount; ++i ) { if( outputStreamInfo->devices[i].channelCount > effectiveOutputChannelCount ) effectiveOutputChannelCount = outputStreamInfo->devices[i].channelCount; } } else { effectiveOutputChannelCount = outputChannelCount; } hostOutputFrameSize = hostOutputSampleSize * effectiveOutputChannelCount; maximumBufferSize = (long) ((PA_MME_MAX_HOST_BUFFER_SECS_ * sampleRate) * hostOutputFrameSize); if( maximumBufferSize > PA_MME_MAX_HOST_BUFFER_BYTES_ ) maximumBufferSize = PA_MME_MAX_HOST_BUFFER_BYTES_; /* compute the following in bytes, then convert back to frames */ SelectBufferSizeAndCount( ((framesPerBuffer == paFramesPerBufferUnspecified) ? PA_MME_MIN_HOST_BUFFER_FRAMES_WHEN_UNSPECIFIED_ : framesPerBuffer ) * hostOutputFrameSize, /* baseBufferSize */ ((unsigned long)(suggestedOutputLatency * sampleRate)) * hostOutputFrameSize, /* suggestedLatency */ 4, /* baseBufferCount */ minimumBufferCount, maximumBufferSize, &hostBufferSizeBytes, &hostBufferCount ); *framesPerHostOutputBuffer = hostBufferSizeBytes / hostOutputFrameSize; *hostOutputBufferCount = hostBufferCount; if( inputChannelCount > 0 ) { /* ensure that both input and output buffer sizes are the same. if they don't match at this stage, choose the smallest one and use that for input and output */ if( *framesPerHostOutputBuffer != *framesPerHostInputBuffer ) { if( framesPerHostInputBuffer < framesPerHostOutputBuffer ) { unsigned long framesPerHostBuffer = *framesPerHostInputBuffer; minimumBufferCount = PA_MME_MIN_HOST_OUTPUT_BUFFER_COUNT_; ReselectBufferCount( framesPerHostBuffer * hostOutputFrameSize, /* bufferSize */ ((unsigned long)(suggestedOutputLatency * sampleRate)) * hostOutputFrameSize, /* suggestedLatency */ 4, /* baseBufferCount */ minimumBufferCount, &hostBufferCount ); *framesPerHostOutputBuffer = framesPerHostBuffer; *hostOutputBufferCount = hostBufferCount; } else { unsigned long framesPerHostBuffer = *framesPerHostOutputBuffer; minimumBufferCount = PA_MME_MIN_HOST_INPUT_BUFFER_COUNT_FULL_DUPLEX_; ReselectBufferCount( framesPerHostBuffer * hostInputFrameSize, /* bufferSize */ ((unsigned long)(suggestedInputLatency * sampleRate)) * hostInputFrameSize, /* suggestedLatency */ 4, /* baseBufferCount */ minimumBufferCount, &hostBufferCount ); *framesPerHostInputBuffer = framesPerHostBuffer; *hostInputBufferCount = hostBufferCount; } } } } } else { *framesPerHostOutputBuffer = 0; *hostOutputBufferCount = 0; } error: return result; } typedef struct { HANDLE bufferEvent; void *waveHandles; unsigned int deviceCount; /* unsigned int channelCount; */ WAVEHDR **waveHeaders; /* waveHeaders[device][buffer] */ unsigned int bufferCount; unsigned int currentBufferIndex; unsigned int framesPerBuffer; unsigned int framesUsedInCurrentBuffer; }PaWinMmeSingleDirectionHandlesAndBuffers; /* prototypes for functions operating on PaWinMmeSingleDirectionHandlesAndBuffers */ static void InitializeSingleDirectionHandlesAndBuffers( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers ); static PaError InitializeWaveHandles( PaWinMmeHostApiRepresentation *winMmeHostApi, PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, unsigned long bytesPerHostSample, double sampleRate, PaWinMmeDeviceAndChannelCount *devices, unsigned int deviceCount, int isInput ); static PaError TerminateWaveHandles( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, int isInput, int currentlyProcessingAnError ); static PaError InitializeWaveHeaders( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, unsigned long hostBufferCount, PaSampleFormat hostSampleFormat, unsigned long framesPerHostBuffer, PaWinMmeDeviceAndChannelCount *devices, int isInput ); static void TerminateWaveHeaders( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, int isInput ); static void InitializeSingleDirectionHandlesAndBuffers( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers ) { handlesAndBuffers->bufferEvent = 0; handlesAndBuffers->waveHandles = 0; handlesAndBuffers->deviceCount = 0; handlesAndBuffers->waveHeaders = 0; handlesAndBuffers->bufferCount = 0; } static PaError InitializeWaveHandles( PaWinMmeHostApiRepresentation *winMmeHostApi, PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, unsigned long bytesPerHostSample, double sampleRate, PaWinMmeDeviceAndChannelCount *devices, unsigned int deviceCount, int isInput ) { PaError result; MMRESULT mmresult; unsigned long bytesPerFrame; WAVEFORMATEX wfx; signed int i; /* for error cleanup we expect that InitializeSingleDirectionHandlesAndBuffers() has already been called to zero some fields */ result = CreateEventWithPaError( &handlesAndBuffers->bufferEvent, NULL, FALSE, FALSE, NULL ); if( result != paNoError ) goto error; if( isInput ) handlesAndBuffers->waveHandles = (void*)PaUtil_AllocateMemory( sizeof(HWAVEIN) * deviceCount ); else handlesAndBuffers->waveHandles = (void*)PaUtil_AllocateMemory( sizeof(HWAVEOUT) * deviceCount ); if( !handlesAndBuffers->waveHandles ) { result = paInsufficientMemory; goto error; } handlesAndBuffers->deviceCount = deviceCount; for( i = 0; i < (signed int)deviceCount; ++i ) { if( isInput ) ((HWAVEIN*)handlesAndBuffers->waveHandles)[i] = 0; else ((HWAVEOUT*)handlesAndBuffers->waveHandles)[i] = 0; } wfx.wFormatTag = WAVE_FORMAT_PCM; wfx.nSamplesPerSec = (DWORD) sampleRate; wfx.cbSize = 0; for( i = 0; i < (signed int)deviceCount; ++i ) { UINT winMmeDeviceId; winMmeDeviceId = LocalDeviceIndexToWinMmeDeviceId( winMmeHostApi, devices[i].device ); wfx.nChannels = (WORD)devices[i].channelCount; bytesPerFrame = wfx.nChannels * bytesPerHostSample; wfx.nAvgBytesPerSec = (DWORD)(bytesPerFrame * sampleRate); wfx.nBlockAlign = (WORD)bytesPerFrame; wfx.wBitsPerSample = (WORD)((bytesPerFrame/wfx.nChannels) * 8); /* REVIEW: consider not firing an event for input when a full duplex stream is being used. this would probably depend on the neverDropInput flag. */ if( isInput ) mmresult = waveInOpen( &((HWAVEIN*)handlesAndBuffers->waveHandles)[i], winMmeDeviceId, &wfx, (DWORD_PTR)handlesAndBuffers->bufferEvent, (DWORD_PTR)0, CALLBACK_EVENT ); else mmresult = waveOutOpen( &((HWAVEOUT*)handlesAndBuffers->waveHandles)[i], winMmeDeviceId, &wfx, (DWORD_PTR)handlesAndBuffers->bufferEvent, (DWORD_PTR)0, CALLBACK_EVENT ); if( mmresult != MMSYSERR_NOERROR ) { switch( mmresult ) { case MMSYSERR_ALLOCATED: /* Specified resource is already allocated. */ result = paDeviceUnavailable; break; case MMSYSERR_NODRIVER: /* No device driver is present. */ result = paDeviceUnavailable; break; case MMSYSERR_NOMEM: /* Unable to allocate or lock memory. */ result = paInsufficientMemory; break; case MMSYSERR_BADDEVICEID: /* Specified device identifier is out of range. */ /* falls through */ case WAVERR_BADFORMAT: /* Attempted to open with an unsupported waveform-audio format. */ /* falls through */ default: result = paUnanticipatedHostError; if( isInput ) { PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); } else { PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); } } goto error; } } return result; error: TerminateWaveHandles( handlesAndBuffers, isInput, 1 /* currentlyProcessingAnError */ ); return result; } static PaError TerminateWaveHandles( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, int isInput, int currentlyProcessingAnError ) { PaError result = paNoError; MMRESULT mmresult; signed int i; if( handlesAndBuffers->waveHandles ) { for( i = handlesAndBuffers->deviceCount-1; i >= 0; --i ) { if( isInput ) { if( ((HWAVEIN*)handlesAndBuffers->waveHandles)[i] ) mmresult = waveInClose( ((HWAVEIN*)handlesAndBuffers->waveHandles)[i] ); else mmresult = MMSYSERR_NOERROR; } else { if( ((HWAVEOUT*)handlesAndBuffers->waveHandles)[i] ) mmresult = waveOutClose( ((HWAVEOUT*)handlesAndBuffers->waveHandles)[i] ); else mmresult = MMSYSERR_NOERROR; } if( mmresult != MMSYSERR_NOERROR && !currentlyProcessingAnError ) /* don't update the error state if we're already processing an error */ { result = paUnanticipatedHostError; if( isInput ) { PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); } else { PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); } /* note that we don't break here, we try to continue closing devices */ } } PaUtil_FreeMemory( handlesAndBuffers->waveHandles ); handlesAndBuffers->waveHandles = 0; } if( handlesAndBuffers->bufferEvent ) { result = CloseHandleWithPaError( handlesAndBuffers->bufferEvent ); handlesAndBuffers->bufferEvent = 0; } return result; } static PaError InitializeWaveHeaders( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, unsigned long hostBufferCount, PaSampleFormat hostSampleFormat, unsigned long framesPerHostBuffer, PaWinMmeDeviceAndChannelCount *devices, int isInput ) { PaError result = paNoError; MMRESULT mmresult; WAVEHDR *deviceWaveHeaders; signed int i, j; /* for error cleanup we expect that InitializeSingleDirectionHandlesAndBuffers() has already been called to zero some fields */ /* allocate an array of pointers to arrays of wave headers, one array of wave headers per device */ handlesAndBuffers->waveHeaders = (WAVEHDR**)PaUtil_AllocateMemory( sizeof(WAVEHDR*) * handlesAndBuffers->deviceCount ); if( !handlesAndBuffers->waveHeaders ) { result = paInsufficientMemory; goto error; } for( i = 0; i < (signed int)handlesAndBuffers->deviceCount; ++i ) handlesAndBuffers->waveHeaders[i] = 0; handlesAndBuffers->bufferCount = hostBufferCount; for( i = 0; i < (signed int)handlesAndBuffers->deviceCount; ++i ) { int bufferBytes = Pa_GetSampleSize( hostSampleFormat ) * framesPerHostBuffer * devices[i].channelCount; if( bufferBytes < 0 ) { result = paInternalError; goto error; } /* Allocate an array of wave headers for device i */ deviceWaveHeaders = (WAVEHDR *) PaUtil_AllocateMemory( sizeof(WAVEHDR)*hostBufferCount ); if( !deviceWaveHeaders ) { result = paInsufficientMemory; goto error; } for( j=0; j < (signed int)hostBufferCount; ++j ) deviceWaveHeaders[j].lpData = 0; handlesAndBuffers->waveHeaders[i] = deviceWaveHeaders; /* Allocate a buffer for each wave header */ for( j=0; j < (signed int)hostBufferCount; ++j ) { deviceWaveHeaders[j].lpData = (char *)PaUtil_AllocateMemory( bufferBytes ); if( !deviceWaveHeaders[j].lpData ) { result = paInsufficientMemory; goto error; } deviceWaveHeaders[j].dwBufferLength = bufferBytes; deviceWaveHeaders[j].dwUser = 0xFFFFFFFF; /* indicates that *PrepareHeader() has not yet been called, for error clean up code */ if( isInput ) { mmresult = waveInPrepareHeader( ((HWAVEIN*)handlesAndBuffers->waveHandles)[i], &deviceWaveHeaders[j], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); goto error; } } else /* output */ { mmresult = waveOutPrepareHeader( ((HWAVEOUT*)handlesAndBuffers->waveHandles)[i], &deviceWaveHeaders[j], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); goto error; } } deviceWaveHeaders[j].dwUser = devices[i].channelCount; } } return result; error: TerminateWaveHeaders( handlesAndBuffers, isInput ); return result; } static void TerminateWaveHeaders( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers, int isInput ) { signed int i, j; WAVEHDR *deviceWaveHeaders; if( handlesAndBuffers->waveHeaders ) { for( i = handlesAndBuffers->deviceCount-1; i >= 0 ; --i ) { deviceWaveHeaders = handlesAndBuffers->waveHeaders[i]; /* wave headers for device i */ if( deviceWaveHeaders ) { for( j = handlesAndBuffers->bufferCount-1; j >= 0; --j ) { if( deviceWaveHeaders[j].lpData ) { if( deviceWaveHeaders[j].dwUser != 0xFFFFFFFF ) { if( isInput ) waveInUnprepareHeader( ((HWAVEIN*)handlesAndBuffers->waveHandles)[i], &deviceWaveHeaders[j], sizeof(WAVEHDR) ); else waveOutUnprepareHeader( ((HWAVEOUT*)handlesAndBuffers->waveHandles)[i], &deviceWaveHeaders[j], sizeof(WAVEHDR) ); } PaUtil_FreeMemory( deviceWaveHeaders[j].lpData ); } } PaUtil_FreeMemory( deviceWaveHeaders ); } } PaUtil_FreeMemory( handlesAndBuffers->waveHeaders ); handlesAndBuffers->waveHeaders = 0; } } /* PaWinMmeStream - a stream data structure specifically for this implementation */ /* note that struct PaWinMmeStream is typedeffed to PaWinMmeStream above. */ struct PaWinMmeStream { PaUtilStreamRepresentation streamRepresentation; PaUtilCpuLoadMeasurer cpuLoadMeasurer; PaUtilBufferProcessor bufferProcessor; int primeStreamUsingCallback; PaWinMmeSingleDirectionHandlesAndBuffers input; PaWinMmeSingleDirectionHandlesAndBuffers output; /* Processing thread management -------------- */ HANDLE abortEvent; HANDLE processingThread; DWORD processingThreadId; char throttleProcessingThreadOnOverload; /* 0 -> don't throtte, non-0 -> throttle */ int processingThreadPriority; int highThreadPriority; int throttledThreadPriority; unsigned long throttledSleepMsecs; int isStopped; volatile int isActive; volatile int stopProcessing; /* stop thread once existing buffers have been returned */ volatile int abortProcessing; /* stop thread immediately */ DWORD allBuffersDurationMs; /* used to calculate timeouts */ }; /* updates deviceCount if PaWinMmeUseMultipleDevices is used */ static PaError ValidateWinMmeSpecificStreamInfo( const PaStreamParameters *streamParameters, const PaWinMmeStreamInfo *streamInfo, char *throttleProcessingThreadOnOverload, unsigned long *deviceCount ) { if( streamInfo ) { if( streamInfo->size != sizeof( PaWinMmeStreamInfo ) || streamInfo->version != 1 ) { return paIncompatibleHostApiSpecificStreamInfo; } if( streamInfo->flags & paWinMmeDontThrottleOverloadedProcessingThread ) *throttleProcessingThreadOnOverload = 0; if( streamInfo->flags & paWinMmeUseMultipleDevices ) { if( streamParameters->device != paUseHostApiSpecificDeviceSpecification ) return paInvalidDevice; *deviceCount = streamInfo->deviceCount; } } return paNoError; } static PaError RetrieveDevicesFromStreamParameters( struct PaUtilHostApiRepresentation *hostApi, const PaStreamParameters *streamParameters, const PaWinMmeStreamInfo *streamInfo, PaWinMmeDeviceAndChannelCount *devices, unsigned long deviceCount ) { PaError result = paNoError; unsigned int i; int totalChannelCount; PaDeviceIndex hostApiDevice; if( streamInfo && streamInfo->flags & paWinMmeUseMultipleDevices ) { totalChannelCount = 0; for( i=0; i < deviceCount; ++i ) { /* validate that the device number is within range */ result = PaUtil_DeviceIndexToHostApiDeviceIndex( &hostApiDevice, streamInfo->devices[i].device, hostApi ); if( result != paNoError ) return result; devices[i].device = hostApiDevice; devices[i].channelCount = streamInfo->devices[i].channelCount; totalChannelCount += devices[i].channelCount; } if( totalChannelCount != streamParameters->channelCount ) { /* channelCount must match total channels specified by multiple devices */ return paInvalidChannelCount; /* REVIEW use of this error code */ } } else { devices[0].device = streamParameters->device; devices[0].channelCount = streamParameters->channelCount; } return result; } static PaError ValidateInputChannelCounts( struct PaUtilHostApiRepresentation *hostApi, PaWinMmeDeviceAndChannelCount *devices, unsigned long deviceCount ) { unsigned int i; for( i=0; i < deviceCount; ++i ) { if( devices[i].channelCount < 1 || devices[i].channelCount > hostApi->deviceInfos[ devices[i].device ]->maxInputChannels ) return paInvalidChannelCount; } return paNoError; } static PaError ValidateOutputChannelCounts( struct PaUtilHostApiRepresentation *hostApi, PaWinMmeDeviceAndChannelCount *devices, unsigned long deviceCount ) { unsigned int i; for( i=0; i < deviceCount; ++i ) { if( devices[i].channelCount < 1 || devices[i].channelCount > hostApi->deviceInfos[ devices[i].device ]->maxOutputChannels ) return paInvalidChannelCount; } return paNoError; } /* the following macros are intended to improve the readability of the following code */ #define PA_IS_INPUT_STREAM_( stream ) ( stream ->input.waveHandles ) #define PA_IS_OUTPUT_STREAM_( stream ) ( stream ->output.waveHandles ) #define PA_IS_FULL_DUPLEX_STREAM_( stream ) ( stream ->input.waveHandles && stream ->output.waveHandles ) #define PA_IS_HALF_DUPLEX_STREAM_( stream ) ( !(stream ->input.waveHandles && stream ->output.waveHandles) ) static PaError OpenStream( struct PaUtilHostApiRepresentation *hostApi, PaStream** s, const PaStreamParameters *inputParameters, const PaStreamParameters *outputParameters, double sampleRate, unsigned long framesPerBuffer, PaStreamFlags streamFlags, PaStreamCallback *streamCallback, void *userData ) { PaError result; PaWinMmeHostApiRepresentation *winMmeHostApi = (PaWinMmeHostApiRepresentation*)hostApi; PaWinMmeStream *stream = 0; int bufferProcessorIsInitialized = 0; int streamRepresentationIsInitialized = 0; PaSampleFormat hostInputSampleFormat, hostOutputSampleFormat; int inputChannelCount, outputChannelCount; PaSampleFormat inputSampleFormat, outputSampleFormat; double suggestedInputLatency, suggestedOutputLatency; PaWinMmeStreamInfo *inputStreamInfo, *outputStreamInfo; unsigned long framesPerHostInputBuffer; unsigned long hostInputBufferCount; unsigned long framesPerHostOutputBuffer; unsigned long hostOutputBufferCount; unsigned long framesPerBufferProcessorCall; PaWinMmeDeviceAndChannelCount *inputDevices = 0; /* contains all devices and channel counts as local host api ids, even when PaWinMmeUseMultipleDevices is not used */ unsigned long inputDeviceCount = 0; PaWinMmeDeviceAndChannelCount *outputDevices = 0; unsigned long outputDeviceCount = 0; /* contains all devices and channel counts as local host api ids, even when PaWinMmeUseMultipleDevices is not used */ char throttleProcessingThreadOnOverload = 1; if( inputParameters ) { inputChannelCount = inputParameters->channelCount; inputSampleFormat = inputParameters->sampleFormat; suggestedInputLatency = inputParameters->suggestedLatency; inputDeviceCount = 1; /* validate input hostApiSpecificStreamInfo */ inputStreamInfo = (PaWinMmeStreamInfo*)inputParameters->hostApiSpecificStreamInfo; result = ValidateWinMmeSpecificStreamInfo( inputParameters, inputStreamInfo, &throttleProcessingThreadOnOverload, &inputDeviceCount ); if( result != paNoError ) return result; inputDevices = (PaWinMmeDeviceAndChannelCount*)alloca( sizeof(PaWinMmeDeviceAndChannelCount) * inputDeviceCount ); if( !inputDevices ) return paInsufficientMemory; result = RetrieveDevicesFromStreamParameters( hostApi, inputParameters, inputStreamInfo, inputDevices, inputDeviceCount ); if( result != paNoError ) return result; result = ValidateInputChannelCounts( hostApi, inputDevices, inputDeviceCount ); if( result != paNoError ) return result; hostInputSampleFormat = PaUtil_SelectClosestAvailableFormat( paInt16 /* native formats */, inputSampleFormat ); } else { inputChannelCount = 0; inputSampleFormat = 0; suggestedInputLatency = 0.; inputStreamInfo = 0; hostInputSampleFormat = 0; } if( outputParameters ) { outputChannelCount = outputParameters->channelCount; outputSampleFormat = outputParameters->sampleFormat; suggestedOutputLatency = outputParameters->suggestedLatency; outputDeviceCount = 1; /* validate output hostApiSpecificStreamInfo */ outputStreamInfo = (PaWinMmeStreamInfo*)outputParameters->hostApiSpecificStreamInfo; result = ValidateWinMmeSpecificStreamInfo( outputParameters, outputStreamInfo, &throttleProcessingThreadOnOverload, &outputDeviceCount ); if( result != paNoError ) return result; outputDevices = (PaWinMmeDeviceAndChannelCount*)alloca( sizeof(PaWinMmeDeviceAndChannelCount) * outputDeviceCount ); if( !outputDevices ) return paInsufficientMemory; result = RetrieveDevicesFromStreamParameters( hostApi, outputParameters, outputStreamInfo, outputDevices, outputDeviceCount ); if( result != paNoError ) return result; result = ValidateOutputChannelCounts( hostApi, outputDevices, outputDeviceCount ); if( result != paNoError ) return result; hostOutputSampleFormat = PaUtil_SelectClosestAvailableFormat( paInt16 /* native formats */, outputSampleFormat ); } else { outputChannelCount = 0; outputSampleFormat = 0; outputStreamInfo = 0; hostOutputSampleFormat = 0; suggestedOutputLatency = 0.; } /* IMPLEMENT ME: - alter sampleRate to a close allowable rate if possible / necessary */ /* validate platform specific flags */ if( (streamFlags & paPlatformSpecificFlags) != 0 ) return paInvalidFlag; /* unexpected platform specific flag */ result = CalculateBufferSettings( &framesPerHostInputBuffer, &hostInputBufferCount, &framesPerHostOutputBuffer, &hostOutputBufferCount, inputChannelCount, hostInputSampleFormat, suggestedInputLatency, inputStreamInfo, outputChannelCount, hostOutputSampleFormat, suggestedOutputLatency, outputStreamInfo, sampleRate, framesPerBuffer ); if( result != paNoError ) goto error; stream = (PaWinMmeStream*)PaUtil_AllocateMemory( sizeof(PaWinMmeStream) ); if( !stream ) { result = paInsufficientMemory; goto error; } InitializeSingleDirectionHandlesAndBuffers( &stream->input ); InitializeSingleDirectionHandlesAndBuffers( &stream->output ); stream->abortEvent = 0; stream->processingThread = 0; stream->throttleProcessingThreadOnOverload = throttleProcessingThreadOnOverload; PaUtil_InitializeStreamRepresentation( &stream->streamRepresentation, ( (streamCallback) ? &winMmeHostApi->callbackStreamInterface : &winMmeHostApi->blockingStreamInterface ), streamCallback, userData ); streamRepresentationIsInitialized = 1; PaUtil_InitializeCpuLoadMeasurer( &stream->cpuLoadMeasurer, sampleRate ); if( inputParameters && outputParameters ) /* full duplex */ { if( framesPerHostInputBuffer < framesPerHostOutputBuffer ) { assert( (framesPerHostOutputBuffer % framesPerHostInputBuffer) == 0 ); /* CalculateBufferSettings() should guarantee this condition */ framesPerBufferProcessorCall = framesPerHostInputBuffer; } else { assert( (framesPerHostInputBuffer % framesPerHostOutputBuffer) == 0 ); /* CalculateBufferSettings() should guarantee this condition */ framesPerBufferProcessorCall = framesPerHostOutputBuffer; } } else if( inputParameters ) { framesPerBufferProcessorCall = framesPerHostInputBuffer; } else if( outputParameters ) { framesPerBufferProcessorCall = framesPerHostOutputBuffer; } stream->input.framesPerBuffer = framesPerHostInputBuffer; stream->output.framesPerBuffer = framesPerHostOutputBuffer; result = PaUtil_InitializeBufferProcessor( &stream->bufferProcessor, inputChannelCount, inputSampleFormat, hostInputSampleFormat, outputChannelCount, outputSampleFormat, hostOutputSampleFormat, sampleRate, streamFlags, framesPerBuffer, framesPerBufferProcessorCall, paUtilFixedHostBufferSize, streamCallback, userData ); if( result != paNoError ) goto error; bufferProcessorIsInitialized = 1; stream->streamRepresentation.streamInfo.inputLatency = (double)(PaUtil_GetBufferProcessorInputLatency(&stream->bufferProcessor) +(framesPerHostInputBuffer * (hostInputBufferCount-1))) / sampleRate; stream->streamRepresentation.streamInfo.outputLatency = (double)(PaUtil_GetBufferProcessorOutputLatency(&stream->bufferProcessor) +(framesPerHostOutputBuffer * (hostOutputBufferCount-1))) / sampleRate; stream->streamRepresentation.streamInfo.sampleRate = sampleRate; stream->primeStreamUsingCallback = ( (streamFlags&paPrimeOutputBuffersUsingStreamCallback) && streamCallback ) ? 1 : 0; /* time to sleep when throttling due to >100% cpu usage. -a quater of a buffer's duration */ stream->throttledSleepMsecs = (unsigned long)(stream->bufferProcessor.framesPerHostBuffer * stream->bufferProcessor.samplePeriod * .25 * 1000); stream->isStopped = 1; stream->isActive = 0; /* for maximum compatibility with multi-device multichannel drivers, we first open all devices, then we prepare all buffers, finally we start all devices ( in StartStream() ). teardown in reverse order. */ if( inputParameters ) { result = InitializeWaveHandles( winMmeHostApi, &stream->input, stream->bufferProcessor.bytesPerHostInputSample, sampleRate, inputDevices, inputDeviceCount, 1 /* isInput */ ); if( result != paNoError ) goto error; } if( outputParameters ) { result = InitializeWaveHandles( winMmeHostApi, &stream->output, stream->bufferProcessor.bytesPerHostOutputSample, sampleRate, outputDevices, outputDeviceCount, 0 /* isInput */ ); if( result != paNoError ) goto error; } if( inputParameters ) { result = InitializeWaveHeaders( &stream->input, hostInputBufferCount, hostInputSampleFormat, framesPerHostInputBuffer, inputDevices, 1 /* isInput */ ); if( result != paNoError ) goto error; } if( outputParameters ) { result = InitializeWaveHeaders( &stream->output, hostOutputBufferCount, hostOutputSampleFormat, framesPerHostOutputBuffer, outputDevices, 0 /* not isInput */ ); if( result != paNoError ) goto error; stream->allBuffersDurationMs = (DWORD) (1000.0 * (framesPerHostOutputBuffer * stream->output.bufferCount) / sampleRate); } else { stream->allBuffersDurationMs = (DWORD) (1000.0 * (framesPerHostInputBuffer * stream->input.bufferCount) / sampleRate); } if( streamCallback ) { /* abort event is only needed for callback streams */ result = CreateEventWithPaError( &stream->abortEvent, NULL, TRUE, FALSE, NULL ); if( result != paNoError ) goto error; } *s = (PaStream*)stream; return result; error: if( stream ) { if( stream->abortEvent ) CloseHandle( stream->abortEvent ); TerminateWaveHeaders( &stream->output, 0 /* not isInput */ ); TerminateWaveHeaders( &stream->input, 1 /* isInput */ ); TerminateWaveHandles( &stream->output, 0 /* not isInput */, 1 /* currentlyProcessingAnError */ ); TerminateWaveHandles( &stream->input, 1 /* isInput */, 1 /* currentlyProcessingAnError */ ); if( bufferProcessorIsInitialized ) PaUtil_TerminateBufferProcessor( &stream->bufferProcessor ); if( streamRepresentationIsInitialized ) PaUtil_TerminateStreamRepresentation( &stream->streamRepresentation ); PaUtil_FreeMemory( stream ); } return result; } /* return non-zero if all current buffers are done */ static int BuffersAreDone( WAVEHDR **waveHeaders, unsigned int deviceCount, int bufferIndex ) { unsigned int i; for( i=0; i < deviceCount; ++i ) { if( !(waveHeaders[i][ bufferIndex ].dwFlags & WHDR_DONE) ) { return 0; } } return 1; } static int CurrentInputBuffersAreDone( PaWinMmeStream *stream ) { return BuffersAreDone( stream->input.waveHeaders, stream->input.deviceCount, stream->input.currentBufferIndex ); } static int CurrentOutputBuffersAreDone( PaWinMmeStream *stream ) { return BuffersAreDone( stream->output.waveHeaders, stream->output.deviceCount, stream->output.currentBufferIndex ); } /* return non-zero if any buffers are queued */ static int NoBuffersAreQueued( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers ) { unsigned int i, j; if( handlesAndBuffers->waveHandles ) { for( i=0; i < handlesAndBuffers->bufferCount; ++i ) { for( j=0; j < handlesAndBuffers->deviceCount; ++j ) { if( !( handlesAndBuffers->waveHeaders[ j ][ i ].dwFlags & WHDR_DONE) ) { return 0; } } } } return 1; } #define PA_CIRCULAR_INCREMENT_( current, max )\ ( (((current) + 1) >= (max)) ? (0) : (current+1) ) #define PA_CIRCULAR_DECREMENT_( current, max )\ ( ((current) == 0) ? ((max)-1) : (current-1) ) static signed long GetAvailableFrames( PaWinMmeSingleDirectionHandlesAndBuffers *handlesAndBuffers ) { signed long result = 0; unsigned int i; if( BuffersAreDone( handlesAndBuffers->waveHeaders, handlesAndBuffers->deviceCount, handlesAndBuffers->currentBufferIndex ) ) { /* we could calculate the following in O(1) if we kept track of the last done buffer */ result = handlesAndBuffers->framesPerBuffer - handlesAndBuffers->framesUsedInCurrentBuffer; i = PA_CIRCULAR_INCREMENT_( handlesAndBuffers->currentBufferIndex, handlesAndBuffers->bufferCount ); while( i != handlesAndBuffers->currentBufferIndex ) { if( BuffersAreDone( handlesAndBuffers->waveHeaders, handlesAndBuffers->deviceCount, i ) ) { result += handlesAndBuffers->framesPerBuffer; i = PA_CIRCULAR_INCREMENT_( i, handlesAndBuffers->bufferCount ); } else break; } } return result; } static PaError AdvanceToNextInputBuffer( PaWinMmeStream *stream ) { PaError result = paNoError; MMRESULT mmresult; unsigned int i; for( i=0; i < stream->input.deviceCount; ++i ) { mmresult = waveInAddBuffer( ((HWAVEIN*)stream->input.waveHandles)[i], &stream->input.waveHeaders[i][ stream->input.currentBufferIndex ], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); } } stream->input.currentBufferIndex = PA_CIRCULAR_INCREMENT_( stream->input.currentBufferIndex, stream->input.bufferCount ); stream->input.framesUsedInCurrentBuffer = 0; return result; } static PaError AdvanceToNextOutputBuffer( PaWinMmeStream *stream ) { PaError result = paNoError; MMRESULT mmresult; unsigned int i; for( i=0; i < stream->output.deviceCount; ++i ) { mmresult = waveOutWrite( ((HWAVEOUT*)stream->output.waveHandles)[i], &stream->output.waveHeaders[i][ stream->output.currentBufferIndex ], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); } } stream->output.currentBufferIndex = PA_CIRCULAR_INCREMENT_( stream->output.currentBufferIndex, stream->output.bufferCount ); stream->output.framesUsedInCurrentBuffer = 0; return result; } /* requeue all but the most recent input with the driver. Used for catching up after a total input buffer underrun */ static PaError CatchUpInputBuffers( PaWinMmeStream *stream ) { PaError result = paNoError; unsigned int i; for( i=0; i < stream->input.bufferCount - 1; ++i ) { result = AdvanceToNextInputBuffer( stream ); if( result != paNoError ) break; } return result; } /* take the most recent output and duplicate it to all other output buffers and requeue them. Used for catching up after a total output buffer underrun. */ static PaError CatchUpOutputBuffers( PaWinMmeStream *stream ) { PaError result = paNoError; unsigned int i, j; unsigned int previousBufferIndex = PA_CIRCULAR_DECREMENT_( stream->output.currentBufferIndex, stream->output.bufferCount ); for( i=0; i < stream->output.bufferCount - 1; ++i ) { for( j=0; j < stream->output.deviceCount; ++j ) { if( stream->output.waveHeaders[j][ stream->output.currentBufferIndex ].lpData != stream->output.waveHeaders[j][ previousBufferIndex ].lpData ) { CopyMemory( stream->output.waveHeaders[j][ stream->output.currentBufferIndex ].lpData, stream->output.waveHeaders[j][ previousBufferIndex ].lpData, stream->output.waveHeaders[j][ stream->output.currentBufferIndex ].dwBufferLength ); } } result = AdvanceToNextOutputBuffer( stream ); if( result != paNoError ) break; } return result; } static DWORD WINAPI ProcessingThreadProc( void *pArg ) { PaWinMmeStream *stream = (PaWinMmeStream *)pArg; HANDLE events[3]; int eventCount = 0; DWORD result = paNoError; DWORD waitResult; DWORD timeout = (unsigned long)(stream->allBuffersDurationMs * 0.5); int hostBuffersAvailable; signed int hostInputBufferIndex, hostOutputBufferIndex; PaStreamCallbackFlags statusFlags; int callbackResult; int done = 0; unsigned int channel, i; unsigned long framesProcessed; /* prepare event array for call to WaitForMultipleObjects() */ if( stream->input.bufferEvent ) events[eventCount++] = stream->input.bufferEvent; if( stream->output.bufferEvent ) events[eventCount++] = stream->output.bufferEvent; events[eventCount++] = stream->abortEvent; statusFlags = 0; /** @todo support paInputUnderflow, paOutputOverflow and paNeverDropInput */ /* loop until something causes us to stop */ do{ /* wait for MME to signal that a buffer is available, or for the PA abort event to be signaled. When this indicates that one or more buffers are available NoBuffersAreQueued() and Current*BuffersAreDone are used below to poll for additional done buffers. NoBuffersAreQueued() will fail to identify an underrun/overflow if the driver doesn't mark all done buffers prior to signalling the event. Some drivers do this (eg RME Digi96, and others don't eg VIA PC 97 input). This isn't a huge problem, it just means that we won't always be able to detect underflow/overflow. */ waitResult = WaitForMultipleObjects( eventCount, events, FALSE /* wait all = FALSE */, timeout ); if( waitResult == WAIT_FAILED ) { result = paUnanticipatedHostError; /** @todo FIXME/REVIEW: can't return host error info from an asyncronous thread */ done = 1; } else if( waitResult == WAIT_TIMEOUT ) { /* if a timeout is encountered, continue */ } if( stream->abortProcessing ) { /* Pa_AbortStream() has been called, stop processing immediately */ done = 1; } else if( stream->stopProcessing ) { /* Pa_StopStream() has been called or the user callback returned non-zero, processing will continue until all output buffers are marked as done. The stream will stop immediately if it is input-only. */ if( PA_IS_OUTPUT_STREAM_(stream) ) { if( NoBuffersAreQueued( &stream->output ) ) done = 1; /* Will cause thread to return. */ } else { /* input only stream */ done = 1; /* Will cause thread to return. */ } } else { hostBuffersAvailable = 1; /* process all available host buffers */ do { hostInputBufferIndex = -1; hostOutputBufferIndex = -1; if( PA_IS_INPUT_STREAM_(stream) ) { if( CurrentInputBuffersAreDone( stream ) ) { if( NoBuffersAreQueued( &stream->input ) ) { /** @todo if all of the other buffers are also ready then we discard all but the most recent. This is an input buffer overflow. FIXME: these buffers should be passed to the callback in a paNeverDropInput stream. note that it is also possible for an input overflow to happen while the callback is processing a buffer. that is handled further down. */ result = CatchUpInputBuffers( stream ); if( result != paNoError ) done = 1; statusFlags |= paInputOverflow; } hostInputBufferIndex = stream->input.currentBufferIndex; } } if( PA_IS_OUTPUT_STREAM_(stream) ) { if( CurrentOutputBuffersAreDone( stream ) ) { /* ok, we have an output buffer */ if( NoBuffersAreQueued( &stream->output ) ) { /* if all of the other buffers are also ready, catch up by copying the most recently generated buffer into all but one of the output buffers. note that this catch up code only handles the case where all buffers have been played out due to this thread not having woken up at all. a more common case occurs when this thread is woken up, processes one buffer, but takes too long, and as a result all the other buffers have become un-queued. that case is handled further down. */ result = CatchUpOutputBuffers( stream ); if( result != paNoError ) done = 1; statusFlags |= paOutputUnderflow; } hostOutputBufferIndex = stream->output.currentBufferIndex; } } if( (PA_IS_FULL_DUPLEX_STREAM_(stream) && hostInputBufferIndex != -1 && hostOutputBufferIndex != -1) || (PA_IS_HALF_DUPLEX_STREAM_(stream) && ( hostInputBufferIndex != -1 || hostOutputBufferIndex != -1 ) ) ) { PaStreamCallbackTimeInfo timeInfo = {0,0,0}; /** @todo implement inputBufferAdcTime */ if( PA_IS_OUTPUT_STREAM_(stream) ) { /* set timeInfo.currentTime and calculate timeInfo.outputBufferDacTime from the current wave out position */ MMTIME mmtime; double timeBeforeGetPosition, timeAfterGetPosition; double time; long framesInBufferRing; long writePosition; long playbackPosition; HWAVEOUT firstWaveOutDevice = ((HWAVEOUT*)stream->output.waveHandles)[0]; mmtime.wType = TIME_SAMPLES; timeBeforeGetPosition = PaUtil_GetTime(); waveOutGetPosition( firstWaveOutDevice, &mmtime, sizeof(MMTIME) ); timeAfterGetPosition = PaUtil_GetTime(); timeInfo.currentTime = timeAfterGetPosition; /* approximate time at which wave out position was measured as half way between timeBeforeGetPosition and timeAfterGetPosition */ time = timeBeforeGetPosition + (timeAfterGetPosition - timeBeforeGetPosition) * .5; framesInBufferRing = stream->output.bufferCount * stream->bufferProcessor.framesPerHostBuffer; playbackPosition = mmtime.u.sample % framesInBufferRing; writePosition = stream->output.currentBufferIndex * stream->bufferProcessor.framesPerHostBuffer + stream->output.framesUsedInCurrentBuffer; if( playbackPosition >= writePosition ){ timeInfo.outputBufferDacTime = time + ((double)( writePosition + (framesInBufferRing - playbackPosition) ) * stream->bufferProcessor.samplePeriod ); }else{ timeInfo.outputBufferDacTime = time + ((double)( writePosition - playbackPosition ) * stream->bufferProcessor.samplePeriod ); } } PaUtil_BeginCpuLoadMeasurement( &stream->cpuLoadMeasurer ); PaUtil_BeginBufferProcessing( &stream->bufferProcessor, &timeInfo, statusFlags ); /* reset status flags once they have been passed to the buffer processor */ statusFlags = 0; if( PA_IS_INPUT_STREAM_(stream) ) { PaUtil_SetInputFrameCount( &stream->bufferProcessor, 0 /* default to host buffer size */ ); channel = 0; for( i=0; iinput.deviceCount; ++i ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->input.waveHeaders[i][ hostInputBufferIndex ].dwUser; PaUtil_SetInterleavedInputChannels( &stream->bufferProcessor, channel, stream->input.waveHeaders[i][ hostInputBufferIndex ].lpData + stream->input.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostInputSample, channelCount ); channel += channelCount; } } if( PA_IS_OUTPUT_STREAM_(stream) ) { PaUtil_SetOutputFrameCount( &stream->bufferProcessor, 0 /* default to host buffer size */ ); channel = 0; for( i=0; ioutput.deviceCount; ++i ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->output.waveHeaders[i][ hostOutputBufferIndex ].dwUser; PaUtil_SetInterleavedOutputChannels( &stream->bufferProcessor, channel, stream->output.waveHeaders[i][ hostOutputBufferIndex ].lpData + stream->output.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostOutputSample, channelCount ); channel += channelCount; } } callbackResult = paContinue; framesProcessed = PaUtil_EndBufferProcessing( &stream->bufferProcessor, &callbackResult ); stream->input.framesUsedInCurrentBuffer += framesProcessed; stream->output.framesUsedInCurrentBuffer += framesProcessed; PaUtil_EndCpuLoadMeasurement( &stream->cpuLoadMeasurer, framesProcessed ); if( callbackResult == paContinue ) { /* nothing special to do */ } else if( callbackResult == paAbort ) { stream->abortProcessing = 1; done = 1; /** @todo FIXME: should probably reset the output device immediately once the callback returns paAbort */ result = paNoError; } else { /* User callback has asked us to stop with paComplete or other non-zero value */ stream->stopProcessing = 1; /* stop once currently queued audio has finished */ result = paNoError; } if( PA_IS_INPUT_STREAM_(stream) && stream->stopProcessing == 0 && stream->abortProcessing == 0 && stream->input.framesUsedInCurrentBuffer == stream->input.framesPerBuffer ) { if( NoBuffersAreQueued( &stream->input ) ) { /** @todo need to handle PaNeverDropInput here where necessary */ result = CatchUpInputBuffers( stream ); if( result != paNoError ) done = 1; statusFlags |= paInputOverflow; } result = AdvanceToNextInputBuffer( stream ); if( result != paNoError ) done = 1; } if( PA_IS_OUTPUT_STREAM_(stream) && !stream->abortProcessing ) { if( stream->stopProcessing && stream->output.framesUsedInCurrentBuffer < stream->output.framesPerBuffer ) { /* zero remaining samples in output output buffer and flush */ stream->output.framesUsedInCurrentBuffer += PaUtil_ZeroOutput( &stream->bufferProcessor, stream->output.framesPerBuffer - stream->output.framesUsedInCurrentBuffer ); /* we send the entire buffer to the output devices, but we could just send a partial buffer, rather than zeroing the unused samples. */ } if( stream->output.framesUsedInCurrentBuffer == stream->output.framesPerBuffer ) { /* check for underflow before enquing the just-generated buffer, but recover from underflow after enquing it. This ensures that the most recent audio segment is repeated */ int outputUnderflow = NoBuffersAreQueued( &stream->output ); result = AdvanceToNextOutputBuffer( stream ); if( result != paNoError ) done = 1; if( outputUnderflow && !done && !stream->stopProcessing ) { /* Recover from underflow in the case where the underflow occured while processing the buffer we just finished */ result = CatchUpOutputBuffers( stream ); if( result != paNoError ) done = 1; statusFlags |= paOutputUnderflow; } } } if( stream->throttleProcessingThreadOnOverload != 0 ) { if( stream->stopProcessing || stream->abortProcessing ) { if( stream->processingThreadPriority != stream->highThreadPriority ) { SetThreadPriority( stream->processingThread, stream->highThreadPriority ); stream->processingThreadPriority = stream->highThreadPriority; } } else if( PaUtil_GetCpuLoad( &stream->cpuLoadMeasurer ) > 1. ) { if( stream->processingThreadPriority != stream->throttledThreadPriority ) { SetThreadPriority( stream->processingThread, stream->throttledThreadPriority ); stream->processingThreadPriority = stream->throttledThreadPriority; } /* sleep to give other processes a go */ Sleep( stream->throttledSleepMsecs ); } else { if( stream->processingThreadPriority != stream->highThreadPriority ) { SetThreadPriority( stream->processingThread, stream->highThreadPriority ); stream->processingThreadPriority = stream->highThreadPriority; } } } } else { hostBuffersAvailable = 0; } } while( hostBuffersAvailable && stream->stopProcessing == 0 && stream->abortProcessing == 0 && !done ); } } while( !done ); stream->isActive = 0; if( stream->streamRepresentation.streamFinishedCallback != 0 ) stream->streamRepresentation.streamFinishedCallback( stream->streamRepresentation.userData ); PaUtil_ResetCpuLoadMeasurer( &stream->cpuLoadMeasurer ); return result; } /* When CloseStream() is called, the multi-api layer ensures that the stream has already been stopped or aborted. */ static PaError CloseStream( PaStream* s ) { PaError result; PaWinMmeStream *stream = (PaWinMmeStream*)s; result = CloseHandleWithPaError( stream->abortEvent ); if( result != paNoError ) goto error; TerminateWaveHeaders( &stream->output, 0 /* not isInput */ ); TerminateWaveHeaders( &stream->input, 1 /* isInput */ ); TerminateWaveHandles( &stream->output, 0 /* not isInput */, 0 /* not currentlyProcessingAnError */ ); TerminateWaveHandles( &stream->input, 1 /* isInput */, 0 /* not currentlyProcessingAnError */ ); PaUtil_TerminateBufferProcessor( &stream->bufferProcessor ); PaUtil_TerminateStreamRepresentation( &stream->streamRepresentation ); PaUtil_FreeMemory( stream ); error: /** @todo REVIEW: what is the best way to clean up a stream if an error is detected? */ return result; } static PaError StartStream( PaStream *s ) { PaError result; PaWinMmeStream *stream = (PaWinMmeStream*)s; MMRESULT mmresult; unsigned int i, j; int callbackResult; unsigned int channel; unsigned long framesProcessed; PaStreamCallbackTimeInfo timeInfo = {0,0,0}; /** @todo implement this for stream priming */ PaUtil_ResetBufferProcessor( &stream->bufferProcessor ); if( PA_IS_INPUT_STREAM_(stream) ) { for( i=0; iinput.bufferCount; ++i ) { for( j=0; jinput.deviceCount; ++j ) { mmresult = waveInAddBuffer( ((HWAVEIN*)stream->input.waveHandles)[j], &stream->input.waveHeaders[j][i], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); goto error; } } } stream->input.currentBufferIndex = 0; stream->input.framesUsedInCurrentBuffer = 0; } if( PA_IS_OUTPUT_STREAM_(stream) ) { for( i=0; ioutput.deviceCount; ++i ) { if( (mmresult = waveOutPause( ((HWAVEOUT*)stream->output.waveHandles)[i] )) != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); goto error; } } for( i=0; ioutput.bufferCount; ++i ) { if( stream->primeStreamUsingCallback ) { stream->output.framesUsedInCurrentBuffer = 0; do{ PaUtil_BeginBufferProcessing( &stream->bufferProcessor, &timeInfo, paPrimingOutput | ((stream->input.bufferCount > 0 ) ? paInputUnderflow : 0)); if( stream->input.bufferCount > 0 ) PaUtil_SetNoInput( &stream->bufferProcessor ); PaUtil_SetOutputFrameCount( &stream->bufferProcessor, 0 /* default to host buffer size */ ); channel = 0; for( j=0; joutput.deviceCount; ++j ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->output.waveHeaders[j][i].dwUser; PaUtil_SetInterleavedOutputChannels( &stream->bufferProcessor, channel, stream->output.waveHeaders[j][i].lpData + stream->output.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostOutputSample, channelCount ); /* we have stored the number of channels in the buffer in dwUser */ channel += channelCount; } callbackResult = paContinue; framesProcessed = PaUtil_EndBufferProcessing( &stream->bufferProcessor, &callbackResult ); stream->output.framesUsedInCurrentBuffer += framesProcessed; if( callbackResult != paContinue ) { /** @todo fix this, what do we do if callback result is non-zero during stream priming? for complete: play out primed waveHeaders as usual for abort: clean up immediately. */ } }while( stream->output.framesUsedInCurrentBuffer != stream->output.framesPerBuffer ); } else { for( j=0; joutput.deviceCount; ++j ) { ZeroMemory( stream->output.waveHeaders[j][i].lpData, stream->output.waveHeaders[j][i].dwBufferLength ); } } /* we queue all channels of a single buffer frame (accross all devices, because some multidevice multichannel drivers work better this way */ for( j=0; joutput.deviceCount; ++j ) { mmresult = waveOutWrite( ((HWAVEOUT*)stream->output.waveHandles)[j], &stream->output.waveHeaders[j][i], sizeof(WAVEHDR) ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); goto error; } } } stream->output.currentBufferIndex = 0; stream->output.framesUsedInCurrentBuffer = 0; } stream->isStopped = 0; stream->isActive = 1; stream->stopProcessing = 0; stream->abortProcessing = 0; result = ResetEventWithPaError( stream->input.bufferEvent ); if( result != paNoError ) goto error; result = ResetEventWithPaError( stream->output.bufferEvent ); if( result != paNoError ) goto error; if( stream->streamRepresentation.streamCallback ) { /* callback stream */ result = ResetEventWithPaError( stream->abortEvent ); if( result != paNoError ) goto error; /* Create thread that waits for audio buffers to be ready for processing. */ stream->processingThread = CreateThread( 0, 0, ProcessingThreadProc, stream, 0, &stream->processingThreadId ); if( !stream->processingThread ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_SYSTEM_ERROR( GetLastError() ); goto error; } /** @todo could have mme specific stream parameters to allow the user to set the callback thread priorities */ stream->highThreadPriority = THREAD_PRIORITY_TIME_CRITICAL; stream->throttledThreadPriority = THREAD_PRIORITY_NORMAL; if( !SetThreadPriority( stream->processingThread, stream->highThreadPriority ) ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_SYSTEM_ERROR( GetLastError() ); goto error; } stream->processingThreadPriority = stream->highThreadPriority; } else { /* blocking read/write stream */ } if( PA_IS_INPUT_STREAM_(stream) ) { for( i=0; i < stream->input.deviceCount; ++i ) { mmresult = waveInStart( ((HWAVEIN*)stream->input.waveHandles)[i] ); PA_DEBUG(("Pa_StartStream: waveInStart returned = 0x%X.\n", mmresult)); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); goto error; } } } if( PA_IS_OUTPUT_STREAM_(stream) ) { for( i=0; i < stream->output.deviceCount; ++i ) { if( (mmresult = waveOutRestart( ((HWAVEOUT*)stream->output.waveHandles)[i] )) != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); goto error; } } } return result; error: /** @todo FIXME: implement recovery as best we can This should involve rolling back to a state as-if this function had never been called */ return result; } static PaError StopStream( PaStream *s ) { PaError result = paNoError; PaWinMmeStream *stream = (PaWinMmeStream*)s; int timeout; DWORD waitResult; MMRESULT mmresult; signed int hostOutputBufferIndex; unsigned int channel, waitCount, i; /** @todo REVIEW: the error checking in this function needs review. the basic idea is to return from this function in a known state - for example there is no point avoiding calling waveInReset just because the thread times out. */ if( stream->processingThread ) { /* callback stream */ /* Tell processing thread to stop generating more data and to let current data play out. */ stream->stopProcessing = 1; /* Calculate timeOut longer than longest time it could take to return all buffers. */ timeout = (int)(stream->allBuffersDurationMs * 1.5); if( timeout < PA_MME_MIN_TIMEOUT_MSEC_ ) timeout = PA_MME_MIN_TIMEOUT_MSEC_; PA_DEBUG(("WinMME StopStream: waiting for background thread.\n")); waitResult = WaitForSingleObject( stream->processingThread, timeout ); if( waitResult == WAIT_TIMEOUT ) { /* try to abort */ stream->abortProcessing = 1; SetEvent( stream->abortEvent ); waitResult = WaitForSingleObject( stream->processingThread, timeout ); if( waitResult == WAIT_TIMEOUT ) { PA_DEBUG(("WinMME StopStream: timed out while waiting for background thread to finish.\n")); result = paTimedOut; } } CloseHandle( stream->processingThread ); stream->processingThread = NULL; } else { /* blocking read / write stream */ if( PA_IS_OUTPUT_STREAM_(stream) ) { if( stream->output.framesUsedInCurrentBuffer > 0 ) { /* there are still unqueued frames in the current buffer, so flush them */ hostOutputBufferIndex = stream->output.currentBufferIndex; PaUtil_SetOutputFrameCount( &stream->bufferProcessor, stream->output.framesPerBuffer - stream->output.framesUsedInCurrentBuffer ); channel = 0; for( i=0; ioutput.deviceCount; ++i ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->output.waveHeaders[i][ hostOutputBufferIndex ].dwUser; PaUtil_SetInterleavedOutputChannels( &stream->bufferProcessor, channel, stream->output.waveHeaders[i][ hostOutputBufferIndex ].lpData + stream->output.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostOutputSample, channelCount ); channel += channelCount; } PaUtil_ZeroOutput( &stream->bufferProcessor, stream->output.framesPerBuffer - stream->output.framesUsedInCurrentBuffer ); /* we send the entire buffer to the output devices, but we could just send a partial buffer, rather than zeroing the unused samples. */ AdvanceToNextOutputBuffer( stream ); } timeout = (stream->allBuffersDurationMs / stream->output.bufferCount) + 1; if( timeout < PA_MME_MIN_TIMEOUT_MSEC_ ) timeout = PA_MME_MIN_TIMEOUT_MSEC_; waitCount = 0; while( !NoBuffersAreQueued( &stream->output ) && waitCount <= stream->output.bufferCount ) { /* wait for MME to signal that a buffer is available */ waitResult = WaitForSingleObject( stream->output.bufferEvent, timeout ); if( waitResult == WAIT_FAILED ) { break; } else if( waitResult == WAIT_TIMEOUT ) { /* keep waiting */ } ++waitCount; } } } if( PA_IS_OUTPUT_STREAM_(stream) ) { for( i =0; i < stream->output.deviceCount; ++i ) { mmresult = waveOutReset( ((HWAVEOUT*)stream->output.waveHandles)[i] ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); } } } if( PA_IS_INPUT_STREAM_(stream) ) { for( i=0; i < stream->input.deviceCount; ++i ) { mmresult = waveInReset( ((HWAVEIN*)stream->input.waveHandles)[i] ); if( mmresult != MMSYSERR_NOERROR ) { result = paUnanticipatedHostError; PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); } } } stream->isStopped = 1; stream->isActive = 0; return result; } static PaError AbortStream( PaStream *s ) { PaError result = paNoError; PaWinMmeStream *stream = (PaWinMmeStream*)s; int timeout; DWORD waitResult; MMRESULT mmresult; unsigned int i; /** @todo REVIEW: the error checking in this function needs review. the basic idea is to return from this function in a known state - for example there is no point avoiding calling waveInReset just because the thread times out. */ if( stream->processingThread ) { /* callback stream */ /* Tell processing thread to abort immediately */ stream->abortProcessing = 1; SetEvent( stream->abortEvent ); } if( PA_IS_OUTPUT_STREAM_(stream) ) { for( i =0; i < stream->output.deviceCount; ++i ) { mmresult = waveOutReset( ((HWAVEOUT*)stream->output.waveHandles)[i] ); if( mmresult != MMSYSERR_NOERROR ) { PA_MME_SET_LAST_WAVEOUT_ERROR( mmresult ); return paUnanticipatedHostError; } } } if( PA_IS_INPUT_STREAM_(stream) ) { for( i=0; i < stream->input.deviceCount; ++i ) { mmresult = waveInReset( ((HWAVEIN*)stream->input.waveHandles)[i] ); if( mmresult != MMSYSERR_NOERROR ) { PA_MME_SET_LAST_WAVEIN_ERROR( mmresult ); return paUnanticipatedHostError; } } } if( stream->processingThread ) { /* callback stream */ PA_DEBUG(("WinMME AbortStream: waiting for background thread.\n")); /* Calculate timeOut longer than longest time it could take to return all buffers. */ timeout = (int)(stream->allBuffersDurationMs * 1.5); if( timeout < PA_MME_MIN_TIMEOUT_MSEC_ ) timeout = PA_MME_MIN_TIMEOUT_MSEC_; waitResult = WaitForSingleObject( stream->processingThread, timeout ); if( waitResult == WAIT_TIMEOUT ) { PA_DEBUG(("WinMME AbortStream: timed out while waiting for background thread to finish.\n")); return paTimedOut; } CloseHandle( stream->processingThread ); stream->processingThread = NULL; } stream->isStopped = 1; stream->isActive = 0; return result; } static PaError IsStreamStopped( PaStream *s ) { PaWinMmeStream *stream = (PaWinMmeStream*)s; return stream->isStopped; } static PaError IsStreamActive( PaStream *s ) { PaWinMmeStream *stream = (PaWinMmeStream*)s; return stream->isActive; } static PaTime GetStreamTime( PaStream *s ) { (void) s; /* unused parameter */ return PaUtil_GetTime(); } static double GetStreamCpuLoad( PaStream* s ) { PaWinMmeStream *stream = (PaWinMmeStream*)s; return PaUtil_GetCpuLoad( &stream->cpuLoadMeasurer ); } /* As separate stream interfaces are used for blocking and callback streams, the following functions can be guaranteed to only be called for blocking streams. */ static PaError ReadStream( PaStream* s, void *buffer, unsigned long frames ) { PaError result = paNoError; PaWinMmeStream *stream = (PaWinMmeStream*)s; void *userBuffer; unsigned long framesRead = 0; unsigned long framesProcessed; signed int hostInputBufferIndex; DWORD waitResult; DWORD timeout = (unsigned long)(stream->allBuffersDurationMs * 0.5); unsigned int channel, i; if( PA_IS_INPUT_STREAM_(stream) ) { /* make a local copy of the user buffer pointer(s). this is necessary because PaUtil_CopyInput() advances these pointers every time it is called. */ if( stream->bufferProcessor.userInputIsInterleaved ) { userBuffer = buffer; } else { userBuffer = alloca( sizeof(void*) * stream->bufferProcessor.inputChannelCount ); if( !userBuffer ) return paInsufficientMemory; for( i = 0; ibufferProcessor.inputChannelCount; ++i ) ((void**)userBuffer)[i] = ((void**)buffer)[i]; } do{ if( CurrentInputBuffersAreDone( stream ) ) { if( NoBuffersAreQueued( &stream->input ) ) { /** @todo REVIEW: consider what to do if the input overflows. do we requeue all of the buffers? should we be running a thread to make sure they are always queued? */ result = paInputOverflowed; } hostInputBufferIndex = stream->input.currentBufferIndex; PaUtil_SetInputFrameCount( &stream->bufferProcessor, stream->input.framesPerBuffer - stream->input.framesUsedInCurrentBuffer ); channel = 0; for( i=0; iinput.deviceCount; ++i ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->input.waveHeaders[i][ hostInputBufferIndex ].dwUser; PaUtil_SetInterleavedInputChannels( &stream->bufferProcessor, channel, stream->input.waveHeaders[i][ hostInputBufferIndex ].lpData + stream->input.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostInputSample, channelCount ); channel += channelCount; } framesProcessed = PaUtil_CopyInput( &stream->bufferProcessor, &userBuffer, frames - framesRead ); stream->input.framesUsedInCurrentBuffer += framesProcessed; if( stream->input.framesUsedInCurrentBuffer == stream->input.framesPerBuffer ) { result = AdvanceToNextInputBuffer( stream ); if( result != paNoError ) break; } framesRead += framesProcessed; }else{ /* wait for MME to signal that a buffer is available */ waitResult = WaitForSingleObject( stream->input.bufferEvent, timeout ); if( waitResult == WAIT_FAILED ) { result = paUnanticipatedHostError; break; } else if( waitResult == WAIT_TIMEOUT ) { /* if a timeout is encountered, continue, perhaps we should give up eventually */ } } }while( framesRead < frames ); } else { result = paCanNotReadFromAnOutputOnlyStream; } return result; } static PaError WriteStream( PaStream* s, const void *buffer, unsigned long frames ) { PaError result = paNoError; PaWinMmeStream *stream = (PaWinMmeStream*)s; const void *userBuffer; unsigned long framesWritten = 0; unsigned long framesProcessed; signed int hostOutputBufferIndex; DWORD waitResult; DWORD timeout = (unsigned long)(stream->allBuffersDurationMs * 0.5); unsigned int channel, i; if( PA_IS_OUTPUT_STREAM_(stream) ) { /* make a local copy of the user buffer pointer(s). this is necessary because PaUtil_CopyOutput() advances these pointers every time it is called. */ if( stream->bufferProcessor.userOutputIsInterleaved ) { userBuffer = buffer; } else { userBuffer = alloca( sizeof(void*) * stream->bufferProcessor.outputChannelCount ); if( !userBuffer ) return paInsufficientMemory; for( i = 0; ibufferProcessor.outputChannelCount; ++i ) ((const void**)userBuffer)[i] = ((const void**)buffer)[i]; } do{ if( CurrentOutputBuffersAreDone( stream ) ) { if( NoBuffersAreQueued( &stream->output ) ) { /** @todo REVIEW: consider what to do if the output underflows. do we requeue all the existing buffers with zeros? should we run a separate thread to keep the buffers enqueued at all times? */ result = paOutputUnderflowed; } hostOutputBufferIndex = stream->output.currentBufferIndex; PaUtil_SetOutputFrameCount( &stream->bufferProcessor, stream->output.framesPerBuffer - stream->output.framesUsedInCurrentBuffer ); channel = 0; for( i=0; ioutput.deviceCount; ++i ) { /* we have stored the number of channels in the buffer in dwUser */ int channelCount = stream->output.waveHeaders[i][ hostOutputBufferIndex ].dwUser; PaUtil_SetInterleavedOutputChannels( &stream->bufferProcessor, channel, stream->output.waveHeaders[i][ hostOutputBufferIndex ].lpData + stream->output.framesUsedInCurrentBuffer * channelCount * stream->bufferProcessor.bytesPerHostOutputSample, channelCount ); channel += channelCount; } framesProcessed = PaUtil_CopyOutput( &stream->bufferProcessor, &userBuffer, frames - framesWritten ); stream->output.framesUsedInCurrentBuffer += framesProcessed; if( stream->output.framesUsedInCurrentBuffer == stream->output.framesPerBuffer ) { result = AdvanceToNextOutputBuffer( stream ); if( result != paNoError ) break; } framesWritten += framesProcessed; } else { /* wait for MME to signal that a buffer is available */ waitResult = WaitForSingleObject( stream->output.bufferEvent, timeout ); if( waitResult == WAIT_FAILED ) { result = paUnanticipatedHostError; break; } else if( waitResult == WAIT_TIMEOUT ) { /* if a timeout is encountered, continue, perhaps we should give up eventually */ } } }while( framesWritten < frames ); } else { result = paCanNotWriteToAnInputOnlyStream; } return result; } static signed long GetStreamReadAvailable( PaStream* s ) { PaWinMmeStream *stream = (PaWinMmeStream*)s; if( PA_IS_INPUT_STREAM_(stream) ) return GetAvailableFrames( &stream->input ); else return paCanNotReadFromAnOutputOnlyStream; } static signed long GetStreamWriteAvailable( PaStream* s ) { PaWinMmeStream *stream = (PaWinMmeStream*)s; if( PA_IS_OUTPUT_STREAM_(stream) ) return GetAvailableFrames( &stream->output ); else return paCanNotWriteToAnInputOnlyStream; } /* NOTE: the following functions are MME-stream specific, and are called directly by client code. We need to check for many more error conditions here because we don't have the benefit of pa_front.c's parameter checking. */ static PaError GetWinMMEStreamPointer( PaWinMmeStream **stream, PaStream *s ) { PaError result; PaUtilHostApiRepresentation *hostApi; PaWinMmeHostApiRepresentation *winMmeHostApi; result = PaUtil_ValidateStreamPointer( s ); if( result != paNoError ) return result; result = PaUtil_GetHostApiRepresentation( &hostApi, paMME ); if( result != paNoError ) return result; winMmeHostApi = (PaWinMmeHostApiRepresentation*)hostApi; /* note, the following would be easier if there was a generic way of testing that a stream belongs to a specific host API */ if( PA_STREAM_REP( s )->streamInterface == &winMmeHostApi->callbackStreamInterface || PA_STREAM_REP( s )->streamInterface == &winMmeHostApi->blockingStreamInterface ) { /* s is a WinMME stream */ *stream = (PaWinMmeStream *)s; return paNoError; } else { return paIncompatibleStreamHostApi; } } int PaWinMME_GetStreamInputHandleCount( PaStream* s ) { PaWinMmeStream *stream; PaError result = GetWinMMEStreamPointer( &stream, s ); if( result == paNoError ) return (PA_IS_INPUT_STREAM_(stream)) ? stream->input.deviceCount : 0; else return result; } HWAVEIN PaWinMME_GetStreamInputHandle( PaStream* s, int handleIndex ) { PaWinMmeStream *stream; PaError result = GetWinMMEStreamPointer( &stream, s ); if( result == paNoError && PA_IS_INPUT_STREAM_(stream) && handleIndex >= 0 && (unsigned int)handleIndex < stream->input.deviceCount ) return ((HWAVEIN*)stream->input.waveHandles)[handleIndex]; else return 0; } int PaWinMME_GetStreamOutputHandleCount( PaStream* s) { PaWinMmeStream *stream; PaError result = GetWinMMEStreamPointer( &stream, s ); if( result == paNoError ) return (PA_IS_OUTPUT_STREAM_(stream)) ? stream->output.deviceCount : 0; else return result; } HWAVEOUT PaWinMME_GetStreamOutputHandle( PaStream* s, int handleIndex ) { PaWinMmeStream *stream; PaError result = GetWinMMEStreamPointer( &stream, s ); if( result == paNoError && PA_IS_OUTPUT_STREAM_(stream) && handleIndex >= 0 && (unsigned int)handleIndex < stream->output.deviceCount ) return ((HWAVEOUT*)stream->output.waveHandles)[handleIndex]; else return 0; }