forked from mikebrady/shairport-sync
-
Notifications
You must be signed in to change notification settings - Fork 0
/
player.c
1457 lines (1279 loc) · 54.6 KB
/
player.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Slave-clocked ALAC stream player. This file is part of Shairport.
* Copyright (c) James Laird 2011, 2013
* All rights reserved.
*
* Modifications for audio synchronisation
* and related work, copyright (c) Mike Brady 2014
* All rights reserved.
*
* 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.
*/
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/types.h>
#include <pthread.h>
#include <math.h>
#include <sys/stat.h>
#include <signal.h>
#include <sys/syslog.h>
#include <assert.h>
#include <fcntl.h>
#include <stdlib.h>
#include <errno.h>
#include <limits.h>
#include "config.h"
#ifdef HAVE_LIBPOLARSSL
#include <polarssl/aes.h>
#include <polarssl/havege.h>
#endif
#ifdef HAVE_LIBSSL
#include <openssl/aes.h>
#endif
#ifdef HAVE_LIBSOXR
#include <soxr.h>
#endif
#include "common.h"
#include "player.h"
#include "rtp.h"
#include "rtsp.h"
#include "alac.h"
// parameters from the source
static unsigned char *aesiv;
#ifdef HAVE_LIBSSL
static AES_KEY aes;
#endif
static int sampling_rate, frame_size;
#define FRAME_BYTES(frame_size) (4 * frame_size)
// maximal resampling shift - conservative
#define OUTFRAME_BYTES(frame_size) (4 * (frame_size + 3))
#ifdef HAVE_LIBPOLARSSL
static aes_context dctx;
#endif
static pthread_t player_thread;
static int please_stop;
static int encrypted; // Normally the audio is encrypted, but it may not be
static int connection_state_to_output; // if true, then play incoming stuff; if false drop everything
static alac_file *decoder_info;
// debug variables
static int late_packet_message_sent;
static uint64_t packet_count = 0;
static int32_t last_seqno_read;
// interthread variables
static int fix_volume = 0x10000;
static pthread_mutex_t vol_mutex = PTHREAD_MUTEX_INITIALIZER;
// default buffer size
// needs to be a power of 2 because of the way BUFIDX(seqno) works
#define BUFFER_FRAMES 512
#define MAX_PACKET 2048
// DAC buffer occupancy stuff
#define DAC_BUFFER_QUEUE_MINIMUM_LENGTH 5000
typedef struct audio_buffer_entry { // decoded audio packets
int ready;
uint32_t timestamp;
seq_t sequence_number;
signed short *data;
} abuf_t;
static abuf_t audio_buffer[BUFFER_FRAMES];
#define BUFIDX(seqno) ((seq_t)(seqno) % BUFFER_FRAMES)
// mutex-protected variables
static seq_t ab_read, ab_write;
static int ab_buffering = 1, ab_synced = 0;
static uint32_t first_packet_timestamp = 0;
static int flush_requested = 0;
static uint32_t flush_rtp_timestamp;
static uint64_t time_of_last_audio_packet;
static int shutdown_requested;
// mutexes and condition variables
static pthread_mutex_t ab_mutex = PTHREAD_MUTEX_INITIALIZER;
static pthread_mutex_t flush_mutex = PTHREAD_MUTEX_INITIALIZER;
static pthread_cond_t flowcontrol;
static int64_t first_packet_time_to_play, time_since_play_started; // nanoseconds
static audio_parameters audio_information;
// stats
static uint64_t missing_packets, late_packets, too_late_packets, resend_requests;
static void ab_resync(void) {
int i;
for (i = 0; i < BUFFER_FRAMES; i++) {
audio_buffer[i].ready = 0;
audio_buffer[i].sequence_number = 0;
}
ab_synced = 0;
last_seqno_read = -1;
ab_buffering = 1;
}
// the sequence number is a 16-bit unsigned number which wraps pretty often
// to work out if one seqno is 'after' another therefore depends whether wrap has occurred
// this function works out the actual ordinate of the seqno, i.e. the distance up from
// the zeroth element, at ab_read, taking due account of wrap.
static inline seq_t SUCCESSOR(seq_t x) {
uint32_t p = x & 0xffff;
p += 1;
p = p & 0xffff;
return p;
}
static inline seq_t PREDECESSOR(seq_t x) {
uint32_t p = (x & 0xffff) + 0x10000;
p -= 1;
p = p & 0xffff;
return p;
}
// anything with ORDINATE in it must be proctected by the ab_mutex
static inline int32_t ORDINATE(seq_t x) {
int32_t p = x & 0xffff;
int32_t q = ab_read & 0x0ffff;
int32_t t = (p + 0x10000 - q) & 0xffff;
// we definitely will get a positive number in t at this point, but it might be a
// positive alias of a negative number, i.e. x might actually be "before" ab_read
// So, if the result is greater than 32767, we will assume its an
// alias and subtract 65536 from it
if (t >= 32767) {
// debug(1,"OOB: %u, ab_r: %u, ab_w: %u",x,ab_read,ab_write);
t -= 65536;
}
return t;
}
// wrapped number between two seq_t.
int32_t seq_diff(seq_t a, seq_t b) {
int32_t diff = ORDINATE(b) - ORDINATE(a);
return diff;
}
// the sequence numbers will wrap pretty often.
// this returns true if the second arg is after the first
static inline int seq_order(seq_t a, seq_t b) {
int32_t d = ORDINATE(b) - ORDINATE(a);
return d > 0;
}
static inline seq_t seq_sum(seq_t a, seq_t b) {
uint32_t p = a & 0xffff;
uint32_t q = b & 0x0ffff;
uint32_t r = (a + b) & 0xffff;
return r;
}
// now for 32-bit wrapping in timestamps
// this returns true if the second arg is strictly after the first
// on the assumption that the gap between them is never greater than (2^31)-1
// Represent a and b in 64 bits
static inline int seq32_order(uint32_t a, uint32_t b) {
if (a == b)
return 0;
int64_t A = a & 0xffffffff;
int64_t B = b & 0xffffffff;
int64_t C = B - A;
// if bit 31 is set, it means either b is before (i.e. less than) a or
// b is (2^31)-1 ahead of a.
// If we assume the gap between b and a should never reach 2 billion, then
// bit 31 == 0 means b is strictly after a
return (C & 0x80000000) == 0;
}
static void alac_decode(short *dest, uint8_t *buf, int len) {
unsigned char packet[MAX_PACKET];
unsigned char packetp[MAX_PACKET];
assert(len <= MAX_PACKET);
int outsize;
if (encrypted) {
unsigned char iv[16];
int aeslen = len & ~0xf;
memcpy(iv, aesiv, sizeof(iv));
#ifdef HAVE_LIBPOLARSSL
aes_crypt_cbc(&dctx, AES_DECRYPT, aeslen, iv, buf, packet);
#endif
#ifdef HAVE_LIBSSL
AES_cbc_encrypt(buf, packet, aeslen, &aes, iv, AES_DECRYPT);
#endif
memcpy(packet + aeslen, buf + aeslen, len - aeslen);
alac_decode_frame(decoder_info, packet, dest, &outsize);
} else {
alac_decode_frame(decoder_info, buf, dest, &outsize);
}
if (outsize!=FRAME_BYTES(frame_size)) {
if(outsize<FRAME_BYTES(frame_size)) {
debug(1,"Output from alac_decode is smaller than expected. Encrypted = %d.",encrypted);
} else {
debug(1,"OUtput from alac_decode larger than expected -- truncated, but buffer overflow possible! Encrypted = %d.",encrypted);
}
}
}
static int init_decoder(int32_t fmtp[12]) {
alac_file *alac;
frame_size = fmtp[1]; // stereo samples
sampling_rate = fmtp[11];
int sample_size = fmtp[3];
if (sample_size != 16)
die("only 16-bit samples supported!");
alac = alac_create(sample_size, 2);
if (!alac)
return 1;
decoder_info = alac;
alac->setinfo_max_samples_per_frame = frame_size;
alac->setinfo_7a = fmtp[2];
alac->setinfo_sample_size = sample_size;
alac->setinfo_rice_historymult = fmtp[4];
alac->setinfo_rice_initialhistory = fmtp[5];
alac->setinfo_rice_kmodifier = fmtp[6];
alac->setinfo_7f = fmtp[7];
alac->setinfo_80 = fmtp[8];
alac->setinfo_82 = fmtp[9];
alac->setinfo_86 = fmtp[10];
alac->setinfo_8a_rate = fmtp[11];
alac_allocate_buffers(alac);
return 0;
}
static void free_decoder(void) { alac_free(decoder_info); }
static void init_buffer(void) {
int i;
for (i = 0; i < BUFFER_FRAMES; i++)
audio_buffer[i].data = malloc(OUTFRAME_BYTES(frame_size));
ab_resync();
}
static void free_buffer(void) {
int i;
for (i = 0; i < BUFFER_FRAMES; i++)
free(audio_buffer[i].data);
}
void player_put_packet(seq_t seqno, uint32_t timestamp, uint8_t *data, int len) {
pthread_mutex_lock(&ab_mutex);
packet_count++;
time_of_last_audio_packet = get_absolute_time_in_fp();
if (connection_state_to_output) { // if we are supposed to be processing these packets
// if (flush_rtp_timestamp != 0)
// debug(1,"Flush_rtp_timestamp is %u",flush_rtp_timestamp);
if ((flush_rtp_timestamp != 0) &&
((timestamp == flush_rtp_timestamp) || seq32_order(timestamp, flush_rtp_timestamp))) {
debug(2, "Dropping flushed packet in player_put_packet, seqno %u, timestamp %u, flushing to "
"timestamp: %u.",
seqno, timestamp, flush_rtp_timestamp);
} else {
if ((flush_rtp_timestamp != 0x0) &&
(!seq32_order(timestamp,
flush_rtp_timestamp))) // if we have gone past the flush boundary time
flush_rtp_timestamp = 0x0;
abuf_t *abuf = 0;
if (!ab_synced) {
debug(2, "syncing to seqno %u.", seqno);
ab_write = seqno;
ab_read = seqno;
ab_synced = 1;
}
if (ab_write == seqno) { // expected packet
abuf = audio_buffer + BUFIDX(seqno);
ab_write = SUCCESSOR(seqno);
} else if (seq_order(ab_write, seqno)) { // newer than expected
// if (ORDINATE(seqno)>(BUFFER_FRAMES*7)/8)
// debug(1,"An interval of %u frames has opened, with ab_read: %u, ab_write: %u and seqno:
// %u.",seq_diff(ab_read,seqno),ab_read,ab_write,seqno);
int32_t gap = seq_diff(ab_write, PREDECESSOR(seqno)) + 1;
if (gap <= 0)
debug(1, "Unexpected gap size: %d.", gap);
int i;
for (i = 0; i < gap; i++) {
abuf = audio_buffer + BUFIDX(seq_sum(ab_write, i));
abuf->ready = 0; // to be sure, to be sure
abuf->timestamp = 0;
abuf->sequence_number = 0;
}
// debug(1,"N %d s %u.",seq_diff(ab_write,PREDECESSOR(seqno))+1,ab_write);
abuf = audio_buffer + BUFIDX(seqno);
rtp_request_resend(ab_write, gap);
resend_requests++;
ab_write = SUCCESSOR(seqno);
} else if (seq_order(ab_read, seqno)) { // late but not yet played
late_packets++;
abuf = audio_buffer + BUFIDX(seqno);
} else { // too late.
too_late_packets++;
/*
if (!late_packet_message_sent) {
debug(1, "too-late packet received: %u; ab_read: %u; ab_write: %u.", seqno, ab_read,
ab_write);
late_packet_message_sent=1;
}
*/
}
// pthread_mutex_unlock(&ab_mutex);
if (abuf) {
alac_decode(abuf->data, data, len);
abuf->ready = 1;
abuf->timestamp = timestamp;
abuf->sequence_number = seqno;
}
// pthread_mutex_lock(&ab_mutex);
}
int rc = pthread_cond_signal(&flowcontrol);
if (rc)
debug(1, "Error signalling flowcontrol.");
}
pthread_mutex_unlock(&ab_mutex);
}
int32_t rand_in_range(int32_t exclusive_range_limit) {
static uint32_t lcg_prev = 12345;
// returns a pseudo random integer in the range 0 to (exclusive_range_limit-1) inclusive
int64_t sp = lcg_prev;
int64_t rl = exclusive_range_limit;
lcg_prev = lcg_prev * 69069 + 3; // crappy psrg
sp = sp*rl; // 64 bit calculation. INtersting part if above the 32 rightmost bits;
return sp >> 32;
}
static inline short dithered_vol(short sample) {
long out;
out = (long)sample * fix_volume;
if (fix_volume < 0x10000) {
// add a TPDF dither -- see http:https://www.users.qwest.net/%7Evolt42/cadenzarecording/DitherExplained.pdf
// and the discussion around https://www.hydrogenaud.io/forums/index.php?showtopic=16963&st=25
// I think, for a 32 --> 16 bits, the range of
// random numbers needs to be from -2^16 to 2^16, i.e. from -65536 to 65536 inclusive, not from -32768 to +32767
// See the original paper at http:https://www.ece.rochester.edu/courses/ECE472/resources/Papers/Lipshitz_1992.pdf
// by Lipshitz, Wannamaker and Vanderkooy, 1992.
long tpdf = rand_in_range(65536+1) - rand_in_range(65536+1);
// Check there's no clipping -- if there is,
if (tpdf>=0) {
if (LONG_MAX-tpdf>=out)
out += tpdf;
else
out = LONG_MAX;
} else {
if (LONG_MIN-tpdf<=out)
out += tpdf;
else
out = LONG_MIN;
}
}
return out >> 16;
}
// get the next frame, when available. return 0 if underrun/stream reset.
static abuf_t *buffer_get_frame(void) {
int16_t buf_fill;
uint64_t local_time_now;
// struct timespec tn;
abuf_t *abuf = 0;
int i;
abuf_t *curframe;
pthread_mutex_lock(&ab_mutex);
int wait;
int32_t dac_delay = 0;
do {
// get the time
local_time_now = get_absolute_time_in_fp();
// if config.timeout (default 120) seconds have elapsed since the last audio packet was
// received, then we should stop.
// config.timeout of zero means don't check..., but iTunes may be confused by a long gap
// followed by a resumption...
if ((time_of_last_audio_packet != 0) && (shutdown_requested == 0) &&
(config.dont_check_timeout == 0)) {
uint64_t ct = config.timeout; // go from int to 64-bit int
if ((local_time_now > time_of_last_audio_packet) &&
(local_time_now - time_of_last_audio_packet >= ct << 32)) {
debug(1, "As Yeats almost said, \"Too long a silence / can make a stone of the heart\"");
rtsp_request_shutdown_stream();
shutdown_requested = 1;
}
}
int rco = get_requested_connection_state_to_output();
if (connection_state_to_output != rco) {
connection_state_to_output = rco;
// change happening
if (connection_state_to_output == 0) { // going off
pthread_mutex_lock(&flush_mutex);
flush_requested = 1;
pthread_mutex_unlock(&flush_mutex);
}
}
pthread_mutex_lock(&flush_mutex);
if (flush_requested == 1) {
if (config.output->flush)
config.output->flush();
ab_resync();
first_packet_timestamp = 0;
first_packet_time_to_play = 0;
time_since_play_started = 0;
flush_requested = 0;
}
pthread_mutex_unlock(&flush_mutex);
uint32_t flush_limit = 0;
if (ab_synced) {
do {
curframe = audio_buffer + BUFIDX(ab_read);
if ((ab_read!=ab_write) && (curframe->ready)) { // it could be synced and empty, under exceptional circumstances, with the frame unused, thus apparently ready
if (curframe->sequence_number != ab_read) {
// some kind of sync problem has occurred.
if (BUFIDX(curframe->sequence_number) == BUFIDX(ab_read)) {
// it looks like some kind of aliasing has happened
if (seq_order(ab_read, curframe->sequence_number)) {
ab_read = curframe->sequence_number;
debug(1, "Aliasing of buffer index -- reset.");
}
} else {
debug(1, "Inconsistent sequence numbers detected");
}
}
if ((flush_rtp_timestamp != 0) &&
((curframe->timestamp == flush_rtp_timestamp) ||
seq32_order(curframe->timestamp, flush_rtp_timestamp))) {
debug(1, "Dropping flushed packet seqno %u, timestamp %u", curframe->sequence_number,
curframe->timestamp);
curframe->ready = 0;
flush_limit++;
ab_read = SUCCESSOR(ab_read);
}
if ((flush_rtp_timestamp != 0) &&
(!seq32_order(curframe->timestamp,
flush_rtp_timestamp))) // if we have gone past the flush boundary time
flush_rtp_timestamp = 0;
}
} while ((flush_rtp_timestamp != 0) && (flush_limit <= 8820) && (curframe->ready == 0));
if (flush_limit == 8820) {
debug(1, "Flush hit the 8820 frame limit!");
flush_limit = 0;
}
curframe = audio_buffer + BUFIDX(ab_read);
if (curframe->ready) {
if (ab_buffering) { // if we are getting packets but not yet forwarding them to the player
int have_sent_prefiller_silence; // set true when we have send some silent frames to the DAC
uint32_t reference_timestamp;
uint64_t reference_timestamp_time,remote_reference_timestamp_time;
get_reference_timestamp_stuff(&reference_timestamp, &reference_timestamp_time, &remote_reference_timestamp_time);
if (first_packet_timestamp == 0) { // if this is the very first packet
// debug(1,"First frame seen, time %u, with %d
// frames...",curframe->timestamp,seq_diff(ab_read, ab_write));
if (reference_timestamp) { // if we have a reference time
// debug(1,"First frame seen with timestamp...");
first_packet_timestamp = curframe->timestamp; // we will keep buffering until we are
// supposed to start playing this
have_sent_prefiller_silence = 0;
// Here, calculate when we should start playing. We need to know when to allow the
// packets to be sent to the player.
// We will send packets of silence from now until that time and then we will send the
// first packet, which will be followed by the subsequent packets.
// we will get a fix every second or so, which will be stored as a pair consisting of
// the time when the packet with a particular timestamp should be played, neglecting
// latencies, etc.
// It probably won't be the timestamp of our first packet, however, so we might have
// to do some calculations.
// To calculate when the first packet will be played, we figure out the exact time the
// packet should be played according to its timestamp and the reference time.
// We then need to add the desired latency, typically 88200 frames.
// Then we need to offset this by the backend latency offset. For example, if we knew
// that the audio back end has a latency of 100 ms, we would
// ask for the first packet to be emitted 100 ms earlier than it should, i.e. -4410
// frames, so that when it got through the audio back end,
// if would be in sync. To do this, we would give it a latency offset of -100 ms, i.e.
// -4410 frames.
int64_t delta = ((int64_t)first_packet_timestamp - (int64_t)reference_timestamp);
first_packet_time_to_play =
reference_timestamp_time +
((delta + (int64_t)config.latency + (int64_t)config.audio_backend_latency_offset)
<< 32) /
44100;
if (local_time_now >= first_packet_time_to_play) {
debug(
1,
"First packet is late! It should have played before now. Flushing 0.1 seconds");
player_flush(first_packet_timestamp + 4410);
}
}
}
if (first_packet_time_to_play != 0) {
// recalculate first_packet_time_to_play -- the latency might change
int64_t delta = ((int64_t)first_packet_timestamp - (int64_t)reference_timestamp);
first_packet_time_to_play =
reference_timestamp_time +
((delta + (int64_t)config.latency + (int64_t)config.audio_backend_latency_offset)
<< 32) /
44100;
uint32_t filler_size = frame_size;
uint32_t max_dac_delay = 4410;
filler_size = 4410; // 0.1 second -- the maximum we'll add to the DAC
if (local_time_now >= first_packet_time_to_play) {
// we've gone past the time...
// debug(1,"Run past the exact start time by %llu frames, with time now of %llx, fpttp
// of %llx and dac_delay of %d and %d packets;
// flush.",(((tn-first_packet_time_to_play)*44100)>>32)+dac_delay,tn,first_packet_time_to_play,dac_delay,seq_diff(ab_read,
// ab_write));
if (config.output->flush)
config.output->flush();
ab_resync();
first_packet_timestamp = 0;
first_packet_time_to_play = 0;
time_since_play_started = 0;
} else {
if ((config.output->delay) && (have_sent_prefiller_silence != 0)) {
dac_delay = config.output->delay();
if (dac_delay == -1) {
debug(1, "Error getting dac_delay in buffer_get_frame.");
dac_delay = 0;
}
} else
dac_delay = 0;
uint64_t gross_frame_gap =
((first_packet_time_to_play - local_time_now) * 44100) >> 32;
int64_t exact_frame_gap = gross_frame_gap - dac_delay;
if (exact_frame_gap <= 0) {
// we've gone past the time...
// debug(1,"Run a bit past the exact start time by %lld frames, with time now of
// %llx, fpttp of %llx and dac_delay of %d and %d packets;
// flush.",-exact_frame_gap,tn,first_packet_time_to_play,dac_delay,seq_diff(ab_read,
// ab_write));
if (config.output->flush)
config.output->flush();
ab_resync();
first_packet_timestamp = 0;
first_packet_time_to_play = 0;
} else {
uint32_t fs = filler_size;
if (fs > (max_dac_delay - dac_delay))
fs = max_dac_delay - dac_delay;
if ((exact_frame_gap <= fs) || (exact_frame_gap <= frame_size * 2)) {
fs = exact_frame_gap;
// debug(1,"Exact frame gap is %llu; play %d frames of silence. Dac_delay is %d,
// with %d packets, ab_read is %04x, ab_write is
// %04x.",exact_frame_gap,fs,dac_delay,seq_diff(ab_read,
// ab_write),ab_read,ab_write);
ab_buffering = 0;
}
signed short *silence;
silence = malloc(FRAME_BYTES(fs));
memset(silence, 0, FRAME_BYTES(fs));
// debug(1,"Exact frame gap is %llu; play %d frames of silence. Dac_delay is %d,
// with %d packets.",exact_frame_gap,fs,dac_delay,seq_diff(ab_read, ab_write));
config.output->play(silence, fs);
free(silence);
have_sent_prefiller_silence = 1;
if (ab_buffering == 0) {
uint64_t reference_timestamp_time; // don't need this...
get_reference_timestamp_stuff(&play_segment_reference_frame, &reference_timestamp_time, &play_segment_reference_frame_remote_time);
#ifdef CONFIG_METADATA
send_ssnc_metadata('prsm', NULL, 0, 0); // "resume", but don't wait if the queue is locked
#endif
}
}
}
}
}
}
}
// Here, we work out whether to release a packet or wait
// We release a buffer when the time is right.
// To work out when the time is right, we need to take account of (1) the actual time the packet
// should be released,
// (2) the latency requested, (3) the audio backend latency offset and (4) the desired length of
// the audio backend's buffer
// The time is right if the current time is later or the same as
// The packet time + (latency + latency offset - backend_buffer_length).
// Note: the last three items are expressed in frames and must be converted to time.
int do_wait = 1;
if ((ab_synced) && (curframe) && (curframe->ready) && (curframe->timestamp)) {
uint32_t reference_timestamp;
uint64_t reference_timestamp_time,remote_reference_timestamp_time;
get_reference_timestamp_stuff(&reference_timestamp, &reference_timestamp_time, &remote_reference_timestamp_time);
if (reference_timestamp) { // if we have a reference time
uint32_t packet_timestamp = curframe->timestamp;
int64_t delta = ((int64_t)packet_timestamp - (int64_t)reference_timestamp);
int64_t offset = (int64_t)config.latency + config.audio_backend_latency_offset -
(int64_t)config.audio_backend_buffer_desired_length;
int64_t net_offset = delta + offset;
int64_t time_to_play = reference_timestamp_time;
int64_t net_offset_fp_sec;
if (net_offset >= 0) {
net_offset_fp_sec = (net_offset << 32) / 44100;
time_to_play += net_offset_fp_sec; // using the latency requested...
// debug(2,"Net Offset: %lld, adjusted: %lld.",net_offset,net_offset_fp_sec);
} else {
net_offset_fp_sec = ((-net_offset) << 32) / 44100;
time_to_play -= net_offset_fp_sec;
// debug(2,"Net Offset: %lld, adjusted: -%lld.",net_offset,net_offset_fp_sec);
}
if (local_time_now >= time_to_play) {
do_wait = 0;
}
}
}
wait = (ab_buffering || (do_wait != 0) || (!ab_synced)) && (!please_stop);
if (wait) {
uint64_t time_to_wait_for_wakeup_fp =
((uint64_t)1 << 32) / 44100; // this is time period of one frame
time_to_wait_for_wakeup_fp *= 4 * 352; // four full 352-frame packets
time_to_wait_for_wakeup_fp /= 3; // four thirds of a packet time
#ifdef COMPILE_FOR_LINUX_AND_FREEBSD_AND_CYGWIN
uint64_t time_of_wakeup_fp = local_time_now + time_to_wait_for_wakeup_fp;
uint64_t sec = time_of_wakeup_fp >> 32;
uint64_t nsec = ((time_of_wakeup_fp & 0xffffffff) * 1000000000) >> 32;
struct timespec time_of_wakeup;
time_of_wakeup.tv_sec = sec;
time_of_wakeup.tv_nsec = nsec;
pthread_cond_timedwait(&flowcontrol, &ab_mutex, &time_of_wakeup);
// int rc = pthread_cond_timedwait(&flowcontrol,&ab_mutex,&time_of_wakeup);
// if (rc!=0)
// debug(1,"pthread_cond_timedwait returned error code %d.",rc);
#endif
#ifdef COMPILE_FOR_OSX
uint64_t sec = time_to_wait_for_wakeup_fp >> 32;
;
uint64_t nsec = ((time_to_wait_for_wakeup_fp & 0xffffffff) * 1000000000) >> 32;
struct timespec time_to_wait;
time_to_wait.tv_sec = sec;
time_to_wait.tv_nsec = nsec;
pthread_cond_timedwait_relative_np(&flowcontrol, &ab_mutex, &time_to_wait);
#endif
}
} while (wait);
if (please_stop) {
pthread_mutex_unlock(&ab_mutex);
return 0;
}
seq_t read = ab_read;
// check if t+8, t+16, t+32, t+64, t+128, ... (buffer_start_fill / 2)
// packets have arrived... last-chance resend
if (!ab_buffering) {
for (i = 8; i < (seq_diff(ab_read, ab_write) / 2); i = (i * 2)) {
seq_t next = seq_sum(ab_read, i);
abuf = audio_buffer + BUFIDX(next);
if (!abuf->ready) {
rtp_request_resend(next, 1);
// debug(1,"Resend %u.",next);
resend_requests++;
}
}
}
if (!curframe->ready) {
// debug(1, " %d. Supplying a silent frame.", read);
missing_packets++;
memset(curframe->data, 0, FRAME_BYTES(frame_size));
curframe->timestamp = 0;
}
curframe->ready = 0;
ab_read = SUCCESSOR(ab_read);
pthread_mutex_unlock(&ab_mutex);
return curframe;
}
static inline short shortmean(short a, short b) {
long al = (long)a;
long bl = (long)b;
long longmean = (al + bl) / 2;
short r = (short)longmean;
if (r != longmean)
debug(1, "Error calculating average of two shorts");
return r;
}
// stuff: 1 means add 1; 0 means do nothing; -1 means remove 1
static int stuff_buffer_basic(short *inptr, short *outptr, int stuff) {
if ((stuff > 1) || (stuff < -1)) {
debug(1, "Stuff argument to stuff_buffer must be from -1 to +1.");
return frame_size;
}
int i;
int stuffsamp = frame_size;
if (stuff)
// stuffsamp = rand() % (frame_size - 1);
stuffsamp =
(rand() % (frame_size - 2)) + 1; // ensure there's always a sample before and after the item
pthread_mutex_lock(&vol_mutex);
for (i = 0; i < stuffsamp; i++) { // the whole frame, if no stuffing
*outptr++ = dithered_vol(*inptr++);
*outptr++ = dithered_vol(*inptr++);
};
if (stuff) {
if (stuff == 1) {
// debug(3, "+++++++++");
// interpolate one sample
//*outptr++ = dithered_vol(((long)inptr[-2] + (long)inptr[0]) >> 1);
//*outptr++ = dithered_vol(((long)inptr[-1] + (long)inptr[1]) >> 1);
*outptr++ = dithered_vol(shortmean(inptr[-2], inptr[0]));
*outptr++ = dithered_vol(shortmean(inptr[-1], inptr[1]));
} else if (stuff == -1) {
// debug(3, "---------");
inptr++;
inptr++;
}
for (i = stuffsamp; i < frame_size + stuff; i++) {
*outptr++ = dithered_vol(*inptr++);
*outptr++ = dithered_vol(*inptr++);
}
}
pthread_mutex_unlock(&vol_mutex);
return frame_size + stuff;
}
#ifdef HAVE_LIBSOXR
// stuff: 1 means add 1; 0 means do nothing; -1 means remove 1
static int stuff_buffer_soxr(short *inptr, short *outptr, int stuff) {
if ((stuff > 1) || (stuff < -1)) {
debug(1, "Stuff argument to sox_stuff_buffer must be from -1 to +1.");
return frame_size;
}
int i;
short *ip, *op;
ip = inptr;
op = outptr;
if (stuff) {
// debug(1,"Stuff %d.",stuff);
soxr_io_spec_t io_spec;
io_spec.itype = SOXR_INT16_I;
io_spec.otype = SOXR_INT16_I;
io_spec.scale = 1.0; // this seems to crash if not = 1.0
io_spec.e = NULL;
io_spec.flags = 0;
size_t odone;
soxr_error_t error = soxr_oneshot(frame_size, frame_size + stuff, 2, /* Rates and # of chans. */
inptr, frame_size, NULL, /* Input. */
outptr, frame_size + stuff, &odone, /* Output. */
&io_spec, /* Input, output and transfer spec. */
NULL, NULL); /* Default configuration.*/
if (error)
die("soxr error: %s\n", "error: %s\n", soxr_strerror(error));
if (odone > frame_size + 1)
die("odone = %d!\n", odone);
const int gpm = 5;
// keep the first (dpm) samples, to mitigate the Gibbs phenomenon
for (i = 0; i < gpm; i++) {
*op++ = *ip++;
*op++ = *ip++;
}
// keep the last (dpm) samples, to mitigate the Gibbs phenomenon
op = outptr + (frame_size + stuff - gpm) * sizeof(short);
ip = inptr + (frame_size - gpm) * sizeof(short);
for (i = 0; i < gpm; i++) {
*op++ = *ip++;
*op++ = *ip++;
}
// finally, adjust the volume, if necessary
if (fix_volume != 65536.0) {
// pthread_mutex_lock(&vol_mutex);
op = outptr;
for (i = 0; i < frame_size + stuff; i++) {
*op = dithered_vol(*op);
op++;
*op = dithered_vol(*op);
op++;
};
// pthread_mutex_unlock(&vol_mutex);
}
} else { // the whole frame, if no stuffing
// pthread_mutex_lock(&vol_mutex);
for (i = 0; i < frame_size; i++) {
*op++ = dithered_vol(*ip++);
*op++ = dithered_vol(*ip++);
};
// pthread_mutex_unlock(&vol_mutex);
}
return frame_size + stuff;
}
#endif
typedef struct stats { // statistics for running averages
int64_t sync_error, correction, drift;
} stats_t;
static void *player_thread_func(void *arg) {
session_corrections = 0;
play_segment_reference_frame = 0; // zero signals that we are not in a play segment
// check that there are enough buffers to accommodate the desired latency and the latency offset
int maximum_latency = config.latency+config.audio_backend_latency_offset;
if ((maximum_latency+(352-1))/352 + 10 > BUFFER_FRAMES)
die("Not enough buffers available for a total latency of %d frames. A maximum of %d 352-frame packets may be accommodated.",maximum_latency,BUFFER_FRAMES);
connection_state_to_output = get_requested_connection_state_to_output();
// this is about half a minute
#define trend_interval 3758
stats_t statistics[trend_interval];
int number_of_statistics, oldest_statistic, newest_statistic;
int at_least_one_frame_seen = 0;
int64_t tsum_of_sync_errors, tsum_of_corrections, tsum_of_insertions_and_deletions,
tsum_of_drifts;
int64_t previous_sync_error, previous_correction;
int64_t minimum_dac_queue_size = 1000000;
int32_t minimum_buffer_occupancy = BUFFER_FRAMES;
int32_t maximum_buffer_occupancy = 0;
buffer_occupancy = 0;
int play_samples;
int64_t current_delay;
int play_number = 0;
time_of_last_audio_packet = 0;
shutdown_requested = 0;
number_of_statistics = oldest_statistic = newest_statistic = 0;
tsum_of_sync_errors = tsum_of_corrections = tsum_of_insertions_and_deletions = tsum_of_drifts = 0;
const int print_interval = trend_interval; // don't ask...
// I think it's useful to keep this prime to prevent it from falling into a pattern with some
// other process.
char rnstate[256];
initstate(time(NULL), rnstate, 256);
signed short *inbuf, *outbuf, *silence;
outbuf = malloc(OUTFRAME_BYTES(frame_size));
silence = malloc(OUTFRAME_BYTES(frame_size));
memset(silence, 0, OUTFRAME_BYTES(frame_size));
late_packet_message_sent = 0;
first_packet_timestamp = 0;
missing_packets = late_packets = too_late_packets = resend_requests = 0;
flush_rtp_timestamp = 0; // it seems this number has a special significance -- it seems to be used
// as a null operand, so we'll use it like that too
int sync_error_out_of_bounds = 0; // number of times in a row that there's been a serious sync error
uint64_t tens_of_seconds = 0;
while (!please_stop) {
abuf_t *inframe = buffer_get_frame();
if (inframe) {
inbuf = inframe->data;
if (inbuf) {
play_number++;
// if it's a supplied silent frame, let us know...
if (inframe->timestamp == 0) {
// debug(1,"Player has a supplied silent frame.");
last_seqno_read =
(SUCCESSOR(last_seqno_read) & 0xffff); // manage the packet out of sequence minder
config.output->play(inbuf, frame_size);
} else {
// We have a frame of data. We need to see if we want to add or remove a frame from it to
// keep in sync.
// So we calculate the timing error for the first frame in the DAC.
// If it's ahead of time, we add one audio frame to this frame to delay a subsequent frame
// If it's late, we remove an audio frame from this frame to bring a subsequent frame
// forward in time
at_least_one_frame_seen = 1;
uint32_t reference_timestamp;
uint64_t reference_timestamp_time,remote_reference_timestamp_time;
get_reference_timestamp_stuff(&reference_timestamp, &reference_timestamp_time, &remote_reference_timestamp_time);
int64_t rt, nt;
rt = reference_timestamp;
nt = inframe->timestamp;
uint64_t local_time_now = get_absolute_time_in_fp();
// struct timespec tn;
// clock_gettime(CLOCK_MONOTONIC,&tn);
// uint64_t
// local_time_now=((uint64_t)tn.tv_sec<<32)+((uint64_t)tn.tv_nsec<<32)/1000000000;
int64_t td_in_frames;
int64_t td = local_time_now - reference_timestamp_time;
// debug(1,"td is %lld.",td);
if (td >= 0) {
td_in_frames = (td * 44100) >> 32;
} else {
td_in_frames = -((-td * 44100) >> 32);
}
// This is the timing error for the next audio frame in the DAC, if applicable
int64_t sync_error = 0;
int amount_to_stuff = 0;
// check sequencing
if (last_seqno_read == -1)
last_seqno_read = inframe->sequence_number;
else {
last_seqno_read = (SUCCESSOR(last_seqno_read) & 0xffff);