// // This file is part of Dire Wolf, an amateur radio packet TNC. // // Copyright (C) 2014, 2015 John Langner, WB2OSZ // // This program is free software: you can redistribute it and/or modify // it under the terms of the GNU General Public License as published by // the Free Software Foundation, either version 2 of the License, or // (at your option) any later version. // // This program is distributed in the hope that it will be useful, // but WITHOUT ANY WARRANTY; without even the implied warranty of // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the // GNU General Public License for more details. // // You should have received a copy of the GNU General Public License // along with this program. If not, see . // /*------------------------------------------------------------------ * * Module: recv.c * * Purpose: Process audio input for receiving. * * This is for all platforms. * * * Description: In earlier versions, we supported a single audio device * and the main program looped around processing the * audio samples. The structure looked like this: * * main in direwolf.c: * * audio_init() * various other *_init() * * loop forever: * s = demod_get_sample. * multi_modem_process_sample(s) * * * When a packet is succesfully decoded, somebody calls * app_process_rec_frame, also in direwolf.c * * * Starting in version 1.2, we support multiple audio * devices at the same time. We now have a separate * thread for each audio device. Decoded frames are * sent to a single queue for serial processing. * * The new flow looks like this: * * main in direwolf.c: * * audio_init() * various other *_init() * recv_init() * recv_process() -- does not return * * * recv_init() This starts up a separate thread * for each audio device. * Each thread reads audio samples and * passes them to multi_modem_process_sample. * * The difference is that app_process_rec_frame * is no longer called directly. Instead * the frame is appended to a queue with dlq_append. * * Received frames can now be processed one at * a time and we don't need to worry about later * processing being reentrant. * * recv_process() This simply waits for something to show up * in the dlq queue and calls app_process_rec_frame * for each. * *---------------------------------------------------------------*/ #include #include #include #include #include //#include //#include //#include #include #ifdef __FreeBSD__ #include #endif #include "direwolf.h" #include "audio.h" #include "demod.h" #include "multi_modem.h" #include "textcolor.h" #include "dlq.h" #include "recv.h" #include "dtmf.h" #include "aprs_tt.h" #if __WIN32__ static unsigned __stdcall recv_adev_thread (void *arg); #else static void * recv_adev_thread (void *arg); #endif static struct audio_s *save_pa; /* Keep pointer to audio configuration */ /* for later use. */ /*------------------------------------------------------------------ * * Name: recv_init * * Purpose: Start up a thread for each audio device. * * * Inputs: pa - Address of structure of type audio_s. * * * Returns: None. * * Errors: Exit if error. * No point in going on if we can't get audio. * *----------------------------------------------------------------*/ void recv_init (struct audio_s *pa) { #if __WIN32__ HANDLE xmit_th[MAX_ADEVS]; #else pthread_t xmit_tid[MAX_ADEVS]; #endif int a; save_pa = pa; for (a=0; aadev[a].defined) { #if DEBUG text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_init: start up thread, a=%d\n", a); #endif #if __WIN32__ xmit_th[a] = (HANDLE)_beginthreadex (NULL, 0, recv_adev_thread, (void*)(long)a, 0, NULL); if (xmit_th[a] == NULL) { text_color_set(DW_COLOR_ERROR); dw_printf ("FATAL: Could not create audio receive thread for device %d.\n", a); exit(1); } #else int e; e = pthread_create (&xmit_tid[a], NULL, recv_adev_thread, (void *)(long)a); if (e != 0) { text_color_set(DW_COLOR_ERROR); dw_printf ("FATAL: Could not create audio receive thread for device %d.\n", a); exit(1); } #endif } #if DEBUG text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_init: all done\n"); #endif } } /* end recv_init */ /* Try using "hot" attribute for all functions */ /* which are used for each audio sample. */ /* Compiler & linker might gather */ /* them together to improve memory cache performance. */ /* Or maybe it won't make any difference. */ __attribute__((hot)) #if __WIN32__ static unsigned __stdcall recv_adev_thread (void *arg) #else static void * recv_adev_thread (void *arg) #endif { int a = (int)(long)arg; // audio device number. int eof; /* This audio device can have one (mono) or two (stereo) channels. */ /* Find number of the first channel. */ int first_chan = ADEVFIRSTCHAN(a); int num_chan = save_pa->adev[a].num_channels; #if DEBUG text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_adev_thread is now running for a=%d\n", a); #endif /* * Get sound samples and decode them. */ eof = 0; while ( ! eof) { int audio_sample; int c; char tt; for (c=0; c= 256 * 256) eof = 1; multi_modem_process_sample(first_chan + c, audio_sample); /* Originally, the DTMF decoder was always active. */ /* It took very little CPU time and the thinking was that an */ /* attached application might be interested in this even when */ /* the APRStt gateway was not being used. */ /* Unfortunately it resulted in too many false detections of */ /* touch tones when hearing other types of digital communications */ /* on HF. Starting in version 1.0, the DTMF decoder is active */ /* only when the APRStt gateway is configured. */ /* The test below allows us to listen to only a single channel for */ /* for touch tone sequences. The DTMF decoder and the accumulation */ /* of digits into a sequence maintain separate data for each channel. */ /* We should be able to accept touch tone sequences concurrently on */ /* all channels. The only issue is when a complete sequence is */ /* sent to aprs_tt_sequence which doesn't have separate data for each */ /* channel. This shouldn't be a problem unless we have multiple */ /* sequences arriving at the same instant. */ if (save_pa->achan[first_chan + c].dtmf_decode != DTMF_DECODE_OFF) { tt = dtmf_sample (first_chan + c, audio_sample/16384.); if (tt != ' ') { aprs_tt_button (first_chan + c, tt); } } } /* When a complete frame is accumulated, */ /* dlq_append, is called. */ /* recv_process, below, drains the queue. */ } // What should we do now? // Seimply terminate the application? // Try to re-init the audio device a couple times before giving up? text_color_set(DW_COLOR_ERROR); dw_printf ("Terminating after audio input failure.\n"); exit (1); } void recv_process (void) { int ok; dlq_type_t type; int chan; int subchan; packet_t pp; alevel_t alevel; retry_t retries; char spectrum[MAX_SUBCHANS+1]; while (1) { dlq_wait_while_empty (); #if DEBUG text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_process: woke up\n"); #endif ok = dlq_remove (&type, &chan, &subchan, &pp, &alevel, &retries, spectrum, sizeof(spectrum)); #if DEBUG text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_process: dlq_remove() returned ok=%d, type=%d, chan=%d, pp=%p\n", ok, (int)type, chan, pp); #endif if (ok) { app_process_rec_packet (chan, subchan, pp, alevel, retries, spectrum); } #if DEBUG else { text_color_set(DW_COLOR_DEBUG); dw_printf ("recv_process: spurious wakeup. (Temp debugging message - not a problem if only occasional.)\n"); } #endif } } /* end recv_process */ /* end recv.c */