support 2nd player streaming from separate raw file
[teensytas.git] / main.c
1 /*
2  * TeensyTAS, TAS input player for MegaDrive
3  * Copyright (c) 2014 notaz
4  *
5  * Permission is hereby granted, free of charge, to any person obtaining
6  * a copy of this software and associated documentation files (the
7  * "Software"), to deal in the Software without restriction, including
8  * without limitation the rights to use, copy, modify, merge, publish,
9  * distribute, sublicense, and/or sell copies of the Software, and to
10  * permit persons to whom the Software is furnished to do so, subject to
11  * the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be
14  * included in all copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
19  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
20  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
21  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
23  * SOFTWARE.
24  */
25
26 #include <stdint.h>
27 #include <stdio.h>
28 #include <string.h>
29 #include "teensy3/core_pins.h"
30 #include "teensy3/usb_seremu.h"
31 #include "teensy3/usb_rawhid.h"
32 #include "pkts.h"
33
34 #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
35
36 #define noinline __attribute__((noinline))
37
38 // use power of 2
39 #define STREAM_BUF_SIZE 512
40 #define STREAM_BUF_MASK (512 - 1)
41
42 /* ?0SA 00DU, ?1CB RLDU */
43 #define STREAM_EL_SZ 2
44
45 static struct {
46         uint8_t stream_to[2][STREAM_BUF_SIZE][STREAM_EL_SZ];
47         uint8_t stream_from[STREAM_BUF_SIZE][STREAM_EL_SZ];
48         struct {
49                 union {
50                         uint8_t fixed_state[4];
51                         uint32_t fixed_state32;
52                 };
53                 union {
54                         uint8_t pending_state[4];
55                         uint32_t pending_state32;
56                 };
57         } pl[2];
58         uint32_t stream_enable_to:1;
59         uint32_t stream_enable_from:1;
60         uint32_t stream_started:1;
61         uint32_t stream_ended:1;
62         uint32_t inc_mode:2;
63         uint32_t use_pending:1;
64         uint32_t frame_cnt;
65         uint32_t edge_cnt;
66         struct {
67                 uint32_t i;
68                 uint32_t o;
69         } pos_to_p[2], pos_from;
70 } g;
71
72 ssize_t _write(int fd, const void *buf, size_t nbyte)
73 {
74         char tbuf[64];
75         int ret;
76
77         if (fd != 1 && fd != 2) {
78                 snprintf(tbuf, sizeof(tbuf), "write to fd %d\n", fd);
79                 usb_seremu_write(tbuf, strlen(tbuf));
80         }
81
82         ret = usb_seremu_write(buf, nbyte);
83         return ret < 0 ? ret : nbyte;
84 }
85
86 void yield(void)
87 {
88 }
89
90 static noinline void choose_isrs_idle(void);
91
92 /* player1 TH */
93 #define PL1_ISFR   PORTD_ISFR
94 #define PL1_TH()   ((CORE_PIN21_PINREG >> CORE_PIN21_BIT) & 1)
95
96 static void pl1th_isr_fixed(void)
97 {
98         uint32_t isfr, th;
99
100         isfr = PL1_ISFR;
101         PL1_ISFR = isfr;
102         th = PL1_TH();
103
104         GPIOD_PDOR = g.pl[0].fixed_state[th];
105         g.edge_cnt++;
106 }
107
108 static noinline void do_to_step_pl1(void)
109 {
110         g.frame_cnt++;
111
112         g.pos_to_p[0].o = (g.pos_to_p[0].o + 1) & STREAM_BUF_MASK;
113         if (g.pos_to_p[0].o == g.pos_to_p[0].i)
114                 // done
115                 choose_isrs_idle();
116 }
117
118 static void pl1th_isr_do_to_inc(void)
119 {
120         uint32_t isfr, th;
121
122         isfr = PL1_ISFR;
123         PL1_ISFR = isfr;
124         th = PL1_TH();
125
126         GPIOD_PDOR = g.stream_to[0][g.pos_to_p[0].o][th];
127         if (th)
128                 do_to_step_pl1();
129 }
130
131 static void pl1th_isr_do_to(void)
132 {
133         uint32_t isfr, th;
134
135         isfr = PL1_ISFR;
136         PL1_ISFR = isfr;
137         th = PL1_TH();
138
139         GPIOD_PDOR = g.stream_to[0][g.pos_to_p[0].o][th];
140         g.edge_cnt++;
141 }
142
143 /* player2 TH */
144 #define PL2_ISFR   PORTC_ISFR
145 #define PL2_TH()   ((CORE_PIN15_PINREG >> CORE_PIN15_BIT) & 1)
146 #define PL2_ADJ(x) ((x) | ((x) << 12))
147
148 static void pl2th_isr_fixed(void)
149 {
150         uint32_t isfr, th, v;
151
152         isfr = PL2_ISFR;
153         PL2_ISFR = isfr;
154         th = PL2_TH();
155
156         v = g.pl[1].fixed_state[th];
157         GPIOB_PDOR = PL2_ADJ(v);
158 }
159
160 static noinline void do_to_step_pl2(void)
161 {
162         g.pos_to_p[1].o = (g.pos_to_p[1].o + 1) & STREAM_BUF_MASK;
163         if (g.pos_to_p[1].o == g.pos_to_p[1].i)
164                 // done
165                 choose_isrs_idle();
166 }
167
168 static void pl2th_isr_do_to_inc(void)
169 {
170         uint32_t isfr, th, v;
171
172         isfr = PL2_ISFR;
173         PL2_ISFR = isfr;
174         th = PL2_TH();
175
176         v = g.stream_to[1][g.pos_to_p[1].o][th];
177         GPIOB_PDOR = PL2_ADJ(v);
178         if (th)
179                 do_to_step_pl2();
180 }
181
182 static void pl2th_isr_do_to_p1d(void)
183 {
184         uint32_t isfr, th, v;
185
186         isfr = PL2_ISFR;
187         PL2_ISFR = isfr;
188         th = PL2_TH();
189
190         v = g.stream_to[1][g.pos_to_p[0].o][th];
191         GPIOB_PDOR = PL2_ADJ(v);
192
193         g.pos_to_p[1].o = g.pos_to_p[0].o;
194 }
195
196 static void pl2th_isr_do_to_inc_pl1(void)
197 {
198         uint32_t isfr, th, v;
199
200         isfr = PL2_ISFR;
201         PL2_ISFR = isfr;
202         th = PL2_TH();
203
204         v = g.stream_to[1][g.pos_to_p[1].o][th];
205         GPIOB_PDOR = PL2_ADJ(v);
206         if (th) {
207                 do_to_step_pl1();
208                 g.pos_to_p[1].o = g.pos_to_p[0].o;
209         }
210 }
211
212 /* vsync handler */
213 #define VSYNC_ISFR PORTC_ISFR
214
215 static void vsync_isr_nop(void)
216 {
217         uint32_t isfr;
218
219         isfr = VSYNC_ISFR;
220         VSYNC_ISFR = isfr;
221 }
222
223 // /vsync starts at line 235/259 (ntsc/pal), just as vcounter jumps back
224 // we care when it comes out (/vsync goes high) after 3 lines at 238/262
225 static void vsync_isr_frameinc(void)
226 {
227         uint32_t isfr;
228
229         isfr = VSYNC_ISFR;
230         VSYNC_ISFR = isfr;
231
232         g.pos_to_p[0].o = (g.pos_to_p[0].o + 1) & STREAM_BUF_MASK;
233         g.pos_to_p[1].o = g.pos_to_p[0].o;
234
235         if (g.pos_to_p[0].o == g.pos_to_p[0].i)
236                 choose_isrs_idle();
237         g.frame_cnt++;
238 }
239
240 /* "recording" data */
241 static noinline void do_from_step(void)
242 {
243         uint32_t s;
244
245         // should hopefully give atomic fixed_state read..
246         s = g.pl[0].fixed_state32;
247         g.pl[0].fixed_state32 = g.pl[0].pending_state32;
248         g.stream_from[g.pos_from.i][0] = s;
249         g.stream_from[g.pos_from.i][1] = s >> 8;
250         g.pos_from.i = (g.pos_from.i + 1) & STREAM_BUF_MASK;
251 }
252
253 static void pl1th_isr_fixed_do_from(void)
254 {
255         uint32_t isfr, th;
256
257         isfr = PL1_ISFR;
258         PL1_ISFR = isfr;
259         th = PL1_TH();
260
261         GPIOD_PDOR = g.pl[0].fixed_state[th];
262         if (th)
263                 do_from_step();
264         g.edge_cnt++;
265 }
266
267 static void vsync_isr_frameinc_do_from(void)
268 {
269         uint32_t isfr;
270
271         isfr = VSYNC_ISFR;
272         VSYNC_ISFR = isfr;
273
274         do_from_step();
275         g.frame_cnt++;
276 }
277
278 /* * */
279 static void choose_isrs(void)
280 {
281         void (*pl1th_handler)(void) = pl1th_isr_fixed;
282         void (*pl2th_handler)(void) = pl2th_isr_fixed;
283         void (*vsync_handler)(void) = vsync_isr_nop;
284
285         if (g.stream_enable_to) {
286                 switch (g.inc_mode) {
287                 case INC_MODE_VSYNC:
288                         pl1th_handler = pl1th_isr_do_to;
289                         pl2th_handler = pl2th_isr_do_to_p1d;
290                         vsync_handler = vsync_isr_frameinc;
291                         break;
292                 case INC_MODE_SHARED_PL1:
293                         pl1th_handler = pl1th_isr_do_to_inc;
294                         pl2th_handler = pl2th_isr_do_to_p1d;
295                         break;
296                 case INC_MODE_SHARED_PL2:
297                         pl1th_handler = pl1th_isr_do_to;
298                         pl2th_handler = pl2th_isr_do_to_inc_pl1;
299                         break;
300                 case INC_MODE_SEPARATE:
301                         pl1th_handler = pl1th_isr_do_to_inc;
302                         pl2th_handler = pl2th_isr_do_to_inc;
303                         break;
304                 }
305         }
306         else if (g.stream_enable_from) {
307                 g.use_pending = 1;
308                 switch (g.inc_mode) {
309                 case INC_MODE_VSYNC:
310                         vsync_handler = vsync_isr_frameinc_do_from;
311                         break;
312                 case INC_MODE_SHARED_PL1:
313                         pl1th_handler = pl1th_isr_fixed_do_from;
314                         break;
315                 case INC_MODE_SHARED_PL2:
316                 case INC_MODE_SEPARATE:
317                         /* TODO */
318                         break;
319                 }
320         }
321
322         attachInterruptVector(IRQ_PORTD, pl1th_handler);
323         attachInterruptVector(IRQ_PORTC, pl2th_handler);
324         attachInterruptVector(IRQ_PORTA, vsync_handler);
325 }
326
327 static noinline void choose_isrs_idle(void)
328 {
329         attachInterruptVector(IRQ_PORTD, pl1th_isr_fixed);
330         attachInterruptVector(IRQ_PORTC, pl2th_isr_fixed);
331         attachInterruptVector(IRQ_PORTA, vsync_isr_nop);
332 }
333
334 static void udelay(uint32_t us)
335 {
336         uint32_t start = micros();
337
338         while ((micros() - start) < us) {
339                 asm volatile("nop; nop; nop; nop");
340                 yield();
341         }
342 }
343
344 static void do_start_seq(void)
345 {
346         uint32_t edge_cnt_last;
347         uint32_t edge_cnt;
348         uint32_t start, t1, t2;
349         int tout;
350
351         start = micros();
352         edge_cnt = g.edge_cnt;
353
354         /* magic value */
355         g.pl[0].fixed_state[0] =
356         g.pl[0].fixed_state[1] = 0x25;
357
358         for (tout = 10000; tout > 0; tout--) {
359                 edge_cnt_last = edge_cnt;
360                 udelay(100);
361                 edge_cnt = g.edge_cnt;
362
363                 if (edge_cnt != edge_cnt_last)
364                         continue;
365                 if (!PL1_TH())
366                         break;
367         }
368
369         g.pl[0].fixed_state[0] = 0x33;
370         g.pl[0].fixed_state[1] = 0x3f;
371         GPIOD_PDOR = 0x33;
372
373         t1 = micros();
374         if (tout == 0) {
375                 printf("start_seq timeout1, t=%u\n", t1 - start);
376                 return;
377         }
378
379         for (tout = 100000; tout > 0; tout--) {
380                 udelay(1);
381
382                 if (PL1_TH())
383                         break;
384         }
385
386         t2 = micros();
387         if (tout == 0) {
388                 printf("start_seq timeout2, t1=%u, t2=%u\n",
389                         t1 - start, t2 - t1);
390                 return;
391         }
392
393         //printf(" t1=%u, t2=%u\n", t1 - start, t2 - t1);
394
395         if (g.stream_started) {
396                 printf("got start_seq when already started\n");
397                 return;
398         }
399
400         if (!g.stream_enable_to && !g.stream_enable_from) {
401                 printf("got start_seq, without enable from USB\n");
402                 return;
403         }
404
405         if (g.stream_enable_to && g.pos_to_p[0].i == g.pos_to_p[0].o) {
406                 printf("got start_seq while stream_to is empty\n");
407                 return;
408         }
409
410         if (g.stream_enable_from && g.pos_from.i != g.pos_from.o) {
411                 printf("got start_seq while stream_from is not empty\n");
412                 return;
413         }
414
415         __disable_irq();
416         choose_isrs();
417         g.stream_started = 1;
418         __enable_irq();
419 }
420
421 // callers must disable IRQs
422 static void clear_state(void)
423 {
424         int i;
425
426         g.stream_enable_to = 0;
427         g.stream_enable_from = 0;
428         g.stream_started = 0;
429         g.stream_ended = 0;
430         g.inc_mode = INC_MODE_VSYNC;
431         g.use_pending = 0;
432         for (i = 0; i < ARRAY_SIZE(g.pos_to_p); i++)
433                 g.pos_to_p[i].i = g.pos_to_p[i].o = 0;
434         g.pos_from.i = g.pos_from.o = 0;
435         g.frame_cnt = 0;
436         memset(g.stream_to[1], 0x3f, sizeof(g.stream_to[1]));
437         choose_isrs_idle();
438 }
439
440 static int get_space_to(int p)
441 {
442         return STREAM_BUF_SIZE - ((g.pos_to_p[p].i - g.pos_to_p[p].o)
443                 & STREAM_BUF_MASK);
444 }
445
446 static int get_used_from(void)
447 {
448         return (g.pos_from.i - g.pos_from.o) & STREAM_BUF_MASK;
449 }
450
451 static void do_usb(void *buf)
452 {
453         struct tas_pkt *pkt = buf;
454         uint32_t pos_to_i, i, p;
455         int space;
456
457         switch (pkt->type) {
458         case PKT_FIXED_STATE:
459                 memcpy(&i, pkt->data, sizeof(i));
460                 if (g.use_pending)
461                         g.pl[0].pending_state32 = i;
462                 else
463                         g.pl[0].fixed_state32 = i;
464                 break;
465         case PKT_STREAM_ENABLE:
466                 __disable_irq();
467                 clear_state();
468                 /* wait for start from MD */
469                 g.stream_enable_to = pkt->enable.stream_to;
470                 g.stream_enable_from = pkt->enable.stream_from;
471                 g.inc_mode = pkt->enable.inc_mode;
472                 if (pkt->enable.no_start_seq) {
473                         GPIOD_PDOR = 0x3f;
474                         choose_isrs();
475                         g.stream_started = 1;
476                 }
477                 __enable_irq();
478                 break;
479         case PKT_STREAM_ABORT:
480                 __disable_irq();
481                 clear_state();
482                 __enable_irq();
483                 break;
484         case PKT_STREAM_END:
485                 g.stream_ended = 1;
486                 printf("end of stream\n");
487                 break;
488         case PKT_STREAM_DATA_TO_P1:
489         case PKT_STREAM_DATA_TO_P2:
490                 p = pkt->type == PKT_STREAM_DATA_TO_P1 ? 0 : 1;
491                 pos_to_i = g.pos_to_p[p].i;
492                 space = get_space_to(p);
493                 if (space <= pkt->size / STREAM_EL_SZ) {
494                         printf("got data pkt while space=%d\n", space);
495                         return;
496                 }
497                 for (i = 0; i < pkt->size / STREAM_EL_SZ; i++) {
498                         memcpy(&g.stream_to[p][pos_to_i++],
499                                pkt->data + i * STREAM_EL_SZ,
500                                STREAM_EL_SZ);
501                         pos_to_i &= STREAM_BUF_MASK;
502                 }
503                 g.pos_to_p[p].i = pos_to_i;
504                 break;
505         default:
506                 printf("got unknown pkt type: %04x\n", pkt->type);
507                 break;
508         }
509 }
510
511 static void check_get_data(int p)
512 {
513         struct tas_pkt pkt;
514         uint8_t buf[64];
515         int ret;
516
517         if (get_space_to(p) <= sizeof(pkt.data) / STREAM_EL_SZ)
518                 return;
519
520         if (g.pos_to_p[p].i == g.pos_to_p[p].o && g.frame_cnt != 0) {
521                 printf("underflow detected\n");
522                 g.stream_enable_to = 0;
523                 return;
524         }
525
526         pkt.type = PKT_STREAM_REQ;
527         pkt.req.frame = g.frame_cnt;
528         pkt.req.is_p2 = p;
529
530         ret = usb_rawhid_send(&pkt, 1000);
531         if (ret != sizeof(pkt)) {
532                 printf("send STREAM_REQ/%d: %d\n", p, ret);
533                 return;
534         }
535
536         ret = usb_rawhid_recv(buf, 1000);
537         if (ret != 64)
538                 printf("usb_rawhid_recv/s: %d\n", ret);
539         else
540                 do_usb(buf);
541 }
542
543 int main(void)
544 {
545         uint32_t led_time = 0;
546         uint32_t scheck_time = 0;
547         uint32_t edge_cnt_last;
548         uint32_t edge_cnt;
549         uint8_t buf[64];
550         int i, ret;
551
552         delay(1000); // wait for usb..
553
554         /* ?0SA 00DU, ?1CB RLDU */
555         for (i = 0; i < 2; i++) {
556                 g.pl[i].fixed_state[0] = 0x33;
557                 g.pl[i].fixed_state[1] = 0x3f;
558         }
559
560         printf("starting, rawhid: %d\n", usb_rawhid_available());
561
562         choose_isrs_idle();
563
564         // md pin   th tr tl  r  l  d  u vsync   th  tr  tl  r  l  d  u
565         //           7  9  6  4  3  2  1          7   9   6  4  3  2  1
566         // md bit*   6  5  4  3  2  1  0          6   5   4  3  2  1  0
567         // t bit    d6 d5 d4 d3 d2 d1 d0   a12   c0 b17 b16 b3 b2 b1 b0
568         // t pin    21 20  6  8  7 14  2     3   15   1   0 18 19 17 16
569         // * - note: tl/tr mixed in most docs
570
571         // player1
572         pinMode(21, INPUT);
573         attachInterrupt(21, pl1th_isr_fixed, CHANGE);
574         NVIC_SET_PRIORITY(IRQ_PORTD, 0);
575
576         pinMode( 2, OUTPUT);
577         pinMode(14, OUTPUT);
578         pinMode( 7, OUTPUT);
579         pinMode( 8, OUTPUT);
580         pinMode( 6, OUTPUT);
581         pinMode(20, OUTPUT);
582
583         // player2
584         pinMode(15, INPUT);
585         attachInterrupt(15, pl1th_isr_fixed, CHANGE);
586         NVIC_SET_PRIORITY(IRQ_PORTC, 0);
587
588         pinMode(16, OUTPUT);
589         pinMode(17, OUTPUT);
590         pinMode(19, OUTPUT);
591         pinMode(18, OUTPUT);
592         pinMode( 0, OUTPUT);
593         pinMode( 1, OUTPUT);
594
595         // vsync line
596         pinMode(3, INPUT);
597         attachInterrupt(3, vsync_isr_nop, RISING);
598         NVIC_SET_PRIORITY(IRQ_PORTA, 16);
599
600         // led
601         pinMode(13, OUTPUT);
602
603         // lower other priorities
604         SCB_SHPR1 = SCB_SHPR2 = SCB_SHPR3 = 0x10101010;
605
606         // CORE_PIN0_PORTSET CORE_PIN0_BITMASK PORTB_PCR16
607         printf("GPIOB PDDR, PDIR: %08x %08x\n", GPIOB_PDIR, GPIOB_PDDR);
608         printf("GPIOC PDDR, PDIR: %08x %08x\n", GPIOC_PDIR, GPIOC_PDDR);
609         printf("GPIOD PDDR, PDIR: %08x %08x\n", GPIOD_PDIR, GPIOD_PDDR);
610         printf("PORTB_PCR16: %08x\n", PORTB_PCR16);
611         printf("PORTC_PCR6:  %08x\n", PORTC_PCR6);
612         printf("PORTD_PCR0:  %08x\n", PORTD_PCR0);
613
614         asm("mrs %0, BASEPRI" : "=r"(ret));
615         printf("BASEPRI: %d, SHPR: %08x %08x %08x\n",
616                 ret, SCB_SHPR1, SCB_SHPR2, SCB_SHPR3);
617
618         edge_cnt_last = g.edge_cnt;
619
620         while (1) {
621                 struct tas_pkt pkt;
622                 uint32_t now;
623
624                 if (g.stream_enable_to && !g.stream_ended) {
625                         check_get_data(0);
626                         if (g.inc_mode == INC_MODE_SHARED_PL2
627                             || g.inc_mode == INC_MODE_SEPARATE)
628                                 check_get_data(1);
629                 }
630
631                 while (g.stream_enable_from && !g.stream_ended
632                   && get_used_from() >= sizeof(pkt.data) / STREAM_EL_SZ)
633                 {
634                         uint32_t o;
635                         int i;
636
637                         o = g.pos_from.o;
638                         for (i = 0; i < sizeof(pkt.data); i += STREAM_EL_SZ) {
639                                 memcpy(pkt.data + i, &g.stream_from[o++],
640                                         STREAM_EL_SZ);
641                                 o &= STREAM_BUF_MASK;
642                         }
643                         g.pos_from.o = o;
644
645                         pkt.type = PKT_STREAM_DATA_FROM;
646                         pkt.size = i;
647
648                         ret = usb_rawhid_send(&pkt, 1000);
649                         if (ret != sizeof(pkt)) {
650                                 printf("send DATA_FROM: %d\n", ret);
651                                 break;
652                         }
653                 }
654
655                 now = millis();
656
657                 // start condition check
658                 if (now - scheck_time > 1000) {
659                         edge_cnt = g.edge_cnt;
660                         //printf("e: %d th: %d\n", edge_cnt - edge_cnt_last,
661                         //      PL1_TH());
662                         if ((g.stream_enable_to || g.stream_enable_from)
663                             && !g.stream_started
664                             && edge_cnt - edge_cnt_last > 10000)
665                         {
666                                 do_start_seq();
667                                 edge_cnt = g.edge_cnt;
668                         }
669                         edge_cnt_last = edge_cnt;
670                         scheck_time = now;
671                 }
672
673                 // led?
674                 if (CORE_PIN13_PORTREG & CORE_PIN13_BITMASK) {
675                         if ((int)(now - led_time) > 10)
676                                 CORE_PIN13_PORTCLEAR = CORE_PIN13_BITMASK;
677                 }
678
679                 // something on rawhid?
680                 if (usb_rawhid_available() > 0)
681                 {
682                         ret = usb_rawhid_recv(buf, 20);
683                         if (ret == 64) {
684                                 led_time = millis();
685                                 CORE_PIN13_PORTSET = CORE_PIN13_BITMASK;
686
687                                 do_usb(buf);
688                         }
689                         else {
690                                 printf("usb_rawhid_recv: %d\n", ret);
691                         }
692                 }
693         }
694
695         return 0;
696 }