From: User for realtime audio Subject: Re: Application using ALSA-driver es-1968 for Maestro-2e: Bug or my fault? Date: Mon, 17 Jun 2002 15:08:43 +0200 Sender: alsa-devel-admin@lists.sourceforge.net Message-ID: <3D0DDF5B.8384EA17@ind.rwth-aachen.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Return-path: Errors-To: alsa-devel-admin@lists.sourceforge.net List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: To: alsa-devel@lists.sourceforge.net List-Id: alsa-devel@alsa-project.org Hi again, I have added some more information for my previous message: The write_loop function is the modified version from the pcm-example: (Buffer "samples_input" has been allocated using: samples_input = (signed short*)malloc(buffersize*amountChannels_input * snd_pcm_format_width(handle)); with buffersize = 64, amountChannels_input = 2, snd_pcm_format_width(handle)=16;) // SOURCECODE: Function write_loop #define DATATYPE short int #define DUMP_DATA int realTimeIO::write_loop() { int err; int avail_in; int chans; void** bufs_input = (void**)malloc(sizeof(void*)*amountChannels_input); void** bufs_output = (void**)malloc(sizeof(void*)*amountChannels_output); #ifdef DUMP_DATA // Set data to 1 for buffersize=64 to check whether readn copies data to buffer (quick hack for testing) for(int i=0; i < 128; i++) samples_input[i] = -10000; #endif for(int i = 0; i < amountChannels_input; i++) { bufs_input[i] = (((DATATYPE*)samples_input)) + (buffersize*i); } for(int i = 0; i < amountChannels_output; i++) bufs_output[i] = (((DATATYPE*)samples_output)) + (buffersize*i); while (1) { avail_in = snd_pcm_avail_update(pcmHandle_input); if(avail_in >= buffersize) { if(accessModeInput == SND_PCM_ACCESS_RW_INTERLEAVED) { err = snd_pcm_readi(pcmHandle_input, samples_input, buffersize); } else { #ifdef DUMP_DATA cnt ++; if(cnt > 100) cout << "Buffersize: Reading " << buffersize << " elements!" << endl; #endif err = snd_pcm_readn(pcmHandle_input, bufs_input, buffersize); } #ifdef DUMP_DATA if(cnt > 100) { cout << "DATA DUMP FOR DEBUGGING:" << endl; cout << "Buffersize:" << buffersize << endl; cout << "Amount channels input:" << amountChannels_input << endl; cout << "Samples available:" << avail_in << endl; // Dump data for buffersize=64 (quick hack for testing) for(int i = 0; i < amountChannels_input * 64; i++) cout << "Val[" << i << "]=" << samples_input[i] << endl; exit(0); } #endif } if(accessModeOutput == SND_PCM_ACCESS_RW_INTERLEAVED) { err = snd_pcm_writei(pcmHandle_output, samples_input, buffersize); } else { err = snd_pcm_writen(pcmHandle_output, bufs_input, buffersize); } if (err < 0) { if (xrun_recovery(pcmHandle_input, err) < 0) { printf("Write error: %s\n", snd_strerror(err)); exit(EXIT_FAILURE); } } } } // END FUNCTION write_loop Of course I get non interleaved and interleaved modes mixed up but the problem is only about the capture anyway! The output from the execution of the application is the following: ================================== Success in opening devices! Success in determining PCM devices! Device for input is:ESS Maestro Subdevice for input is:subdevice #0 Stream: Direction:1 Device for output is:ESS Maestro Subdevice for input is:subdevice #0 Stream: Direction:0 ================================================== Hardware parameters have been prepared! For input device: Formats: Supported:SND_PCM_FORMAT_S16 Access types: Supported:MMAP_NONINTERLEAVED Supported:RW_NONINTERLEAVED Channels: ============================== Min:1 Max:2 Sampling rate: ============================== Min:4000 Max:48000 For output device: Formats: Supported:SND_PCM_FORMAT_U8 Supported:SND_PCM_FORMAT_S16 Access types: Supported:MMAP_INTERLEAVED Supported:RW_INTERLEAVED Channels: ============================== Min:1 Max:2 Sampling rate: ============================== Min:4000 Max:48000 Buffersize:64 Bufferlength:128 Modified Buffersize:64 Setting up 2 channels for input.! Setting up 2 channels for output.! Bits input:16 Bits output:16 Output and input have been linked together! Playback device is hw:0,0 Stream parameters are 16000Hz, S16_LE, 2 amountChannels_output !!!Scheduler set to Round Robin with priority FAILED!!!99 Buffersize: Reading 64 elements! DATA DUMP FOR DEBUGGING: Buffersize:64 Amount channels input:2 Samples available:68 Val[0]=-4 Val[1]=-1 Val[2]=0 Val[3]=-4 Val[4]=-5 Val[5]=0 Val[6]=0 Val[7]=0 Val[8]=-1 Val[9]=0 Val[10]=-1 Val[11]=-2 Val[12]=-2 Val[13]=-1 Val[14]=0 Val[15]=-1 Val[16]=-2 Val[17]=0 Val[18]=-5 Val[19]=-3 Val[20]=-1 Val[21]=0 Val[22]=0 Val[23]=-3 Val[24]=-1 Val[25]=-3 Val[26]=-3 Val[27]=0 Val[28]=0 Val[29]=1 Val[30]=2 Val[31]=0 Val[32]=-10000 Val[33]=-10000 Val[34]=-10000 Val[35]=-10000 Val[36]=-10000 Val[37]=-10000 Val[38]=-10000 Val[39]=-10000 Val[40]=-10000 Val[41]=-10000 Val[42]=-10000 Val[43]=-10000 Val[44]=-10000 Val[45]=-10000 Val[46]=-10000 Val[47]=-10000 Val[48]=-10000 Val[49]=-10000 Val[50]=-10000 Val[51]=-10000 Val[52]=-10000 Val[53]=-10000 Val[54]=-10000 Val[55]=-10000 Val[56]=-10000 Val[57]=-10000 Val[58]=-10000 Val[59]=-10000 Val[60]=-10000 Val[61]=-10000 Val[62]=-10000 Val[63]=-10000 Val[64]=-4 Val[65]=-1 Val[66]=0 Val[67]=-4 Val[68]=-5 Val[69]=0 Val[70]=0 Val[71]=0 Val[72]=-1 Val[73]=0 Val[74]=-1 Val[75]=-2 Val[76]=-2 Val[77]=-1 Val[78]=0 Val[79]=-1 Val[80]=-2 Val[81]=0 Val[82]=-5 Val[83]=-3 Val[84]=-1 Val[85]=0 Val[86]=0 Val[87]=-3 Val[88]=-1 Val[89]=-3 Val[90]=-3 Val[91]=0 Val[92]=0 Val[93]=1 Val[94]=2 Val[95]=0 Val[96]=-10000 Val[97]=-10000 Val[98]=-10000 Val[99]=-10000 Val[100]=-10000 Val[101]=-10000 Val[102]=-10000 Val[103]=-10000 Val[104]=-10000 Val[105]=-10000 Val[106]=-10000 Val[107]=-10000 Val[108]=-10000 Val[109]=-10000 Val[110]=-10000 Val[111]=-10000 Val[112]=-10000 Val[113]=-10000 Val[114]=-10000 Val[115]=-10000 Val[116]=-10000 Val[117]=-10000 Val[118]=-10000 Val[119]=-10000 Val[120]=-10000 Val[121]=-10000 Val[122]=-10000 Val[123]=-10000 Val[124]=-10000 Val[125]=-10000 Val[126]=-10000 Val[127]=-10000 The first lines are only for verbose mode, however at the end the data dump files are printed. Everything seems fine, there are 68 samples available and 64 are requested to be read but also it can be seen (therefore the initialization with -10000) that the values from 32 to 63 and those from 96 to 127 are NOT set to new values by a copy process. I have run the application without data dump and printed all information in /proc/asound/card0/pcm0c/sub0/* while the application was running: access: RW_NONINTERLEAVED format: S16_LE subformat: STD channels: 2 rate: 16000 (16000/1) period_size: 64 buffer_size: 128 tick_time: 10000 card: 0 device: 0 subdevice: 0 stream: CAPTURE id: ESS Maestro name: ESS Maestro subname: subdevice #0 class: 0 subclass: 0 subdevices_count: 1 subdevices_avail: 0 state: RUNNING trigger_time: 1024319237.293920 tstamp : 1024319251.312477 delay : 54 avail : 54 avail_max : 142 ----- hw_ptr : 224310 appl_ptr : 224256 tstamp_mode: NONE period_step: 1 sleep_min: 0 avail_min: 64 xfer_align: 1 start_threshold: 128 stop_threshold: 128 silence_threshold: 0 silence_size: 0 boundary: 1073741824 I hope this gives a better impression about the used application. I may send the complete sourcecode of my application to anyone who intends to test it with his own device. Again thanks for any help in advance! Hauke > FROM: User for realtime audio > DATE: 06/17/2002 02:41:20 > SUBJECT: [Alsa-devel] Application using ALSA-driver es-1968 for Maestro-2e: Bug or my fault? > Hi, > I have a question to all those who have experience with a Maestro-2e > soundcard and ALSA-applications: > I have written a realtime-capture plus playback application for realtime > signal processing algorithms. It is simply > a modification of the PCM-example. > The software did work with my old notebook including an es-1668 > soundcard but with the new es-1968 device none of the possible transfer > modes function propperly. > The problem is that only parts of the buffers seem to be read correctly > for capture, for example I have the IO-read access to a 16 bit capture > device, periodsize is 64, buffersize 128 and two channels input. > The card only support non interleaved input (although only interleaved > output for playback!). > Furthermore I have one buffer for sampleinput: > signed short inputBuffer[128]; > (2 channels, 64 samples for call of snd_pcm_readn(..)) > In order to use non interleaved read-function I have created the > void**-field: > void** bufsInput[2]; > for each channel and assigned the sample-buffer to each channel: > bufsInput[0] = inputBuffer; > bufsInput[1] = &inputBuffer[64]; > In my callback-loop for audio read process I check that there are more > than 64 samples available > If so I call the read function: > err=snd_pcm_readn(pcmDevice_capture, bufsInput, 64); > After this call err is equal 64 so everything is fine. > When dumping the received data some strange values are stored: > - sampleBuffer[0] .. sampleBuffer[15] seem to be the result from > channel1 which is still correct. > - sampleBuffer[16]..sampleBuffer[31] seem to be the result from the > other channel which is not correct anymore > (I can distinguish the channels as I am using different signals on the > two channels) > - sampleBuffer[32].. sampleBuffer[63] have not been copied. The values > in sampleBuffer BEFORE the snd_pcm_readn-call still remain unchanged! > - From sampleBuffer[64] .. sampleBuffer[127] the values from > sampleBuffer[0] .. sampleBuffer[63] are repeated. > I would expect that values from 0 to 63 contain channel1 input and > values from 64 to 127 contain other channel input.. > Did I misunderstand the ALSA-API, what am I doing wrong or is there a > bug in the ALSA-driver/library for this specific application? > aplay and the pcm example do function as expected but arecord seems to > generate some clicking noise either.... > System: I am using a Dell Lattitude CPH-500 notebook, SUSE linux, kernel > 2.4.16 with the lowlatency patches + rtc-patch found in the > alsa-beta0.12 package. The current alsa-version however is the current > release candidate (rc1). I have also tried to use the > unpatched 2.4.18 kernel which did not lead to different results. > I d appreceate any kind of hint or other experiences with this device, > thanks in advance > Hauke _______________________________________________________________ Sponsored by: ThinkGeek at http://www.ThinkGeek.com/