src/gba/gba-bios.c (view raw)
1#include "gba-bios.h"
2
3#include "gba.h"
4#include "gba-io.h"
5#include "gba-memory.h"
6
7#include <math.h>
8#include <stdlib.h>
9
10static void _unLz77(struct GBAMemory* memory, uint32_t source, uint8_t* dest);
11static void _unRl(struct GBAMemory* memory, uint32_t source, uint8_t* dest);
12
13static void _RegisterRamReset(struct GBA* gba) {
14 uint32_t registers = gba->cpu.gprs[0];
15 (void)(registers);
16 GBALog(gba, GBA_LOG_STUB, "RegisterRamReset unimplemented");
17}
18
19static void _CpuSet(struct GBA* gba) {
20 uint32_t source = gba->cpu.gprs[0];
21 uint32_t dest = gba->cpu.gprs[1];
22 uint32_t mode = gba->cpu.gprs[2];
23 int count = mode & 0x000FFFFF;
24 int fill = mode & 0x01000000;
25 int wordsize = (mode & 0x04000000) ? 4 : 2;
26 int i;
27 if (fill) {
28 if (wordsize == 4) {
29 source &= 0xFFFFFFFC;
30 dest &= 0xFFFFFFFC;
31 int32_t word = gba->memory.d.load32(&gba->memory.d, source, &gba->cpu.cycles);
32 for (i = 0; i < count; ++i) {
33 gba->memory.d.store32(&gba->memory.d, dest + (i << 2), word, &gba->cpu.cycles);
34 gba->board.d.processEvents(&gba->board.d);
35 }
36 } else {
37 source &= 0xFFFFFFFE;
38 dest &= 0xFFFFFFFE;
39 uint16_t word = gba->memory.d.load16(&gba->memory.d, source, &gba->cpu.cycles);
40 for (i = 0; i < count; ++i) {
41 gba->memory.d.store16(&gba->memory.d, dest + (i << 1), word, &gba->cpu.cycles);
42 gba->board.d.processEvents(&gba->board.d);
43 }
44 }
45 } else {
46 if (wordsize == 4) {
47 source &= 0xFFFFFFFC;
48 dest &= 0xFFFFFFFC;
49 for (i = 0; i < count; ++i) {
50 int32_t word = gba->memory.d.load32(&gba->memory.d, source + (i << 2), &gba->cpu.cycles);
51 gba->memory.d.store32(&gba->memory.d, dest + (i << 2), word, &gba->cpu.cycles);
52 gba->board.d.processEvents(&gba->board.d);
53 }
54 } else {
55 source &= 0xFFFFFFFE;
56 dest &= 0xFFFFFFFE;
57 for (i = 0; i < count; ++i) {
58 uint16_t word = gba->memory.d.load16(&gba->memory.d, source + (i << 1), &gba->cpu.cycles);
59 gba->memory.d.store16(&gba->memory.d, dest + (i << 1), word, &gba->cpu.cycles);
60 gba->board.d.processEvents(&gba->board.d);
61 }
62 }
63 }
64}
65
66static void _FastCpuSet(struct GBA* gba) {
67 uint32_t source = gba->cpu.gprs[0] & 0xFFFFFFFC;
68 uint32_t dest = gba->cpu.gprs[1] & 0xFFFFFFFC;
69 uint32_t mode = gba->cpu.gprs[2];
70 int count = mode & 0x000FFFFF;
71 int storeCycles = gba->memory.d.waitMultiple(&gba->memory.d, dest, 4);
72 count = ((count + 7) >> 3) << 3;
73 int i;
74 if (mode & 0x01000000) {
75 int32_t word = gba->memory.d.load32(&gba->memory.d, source, &gba->cpu.cycles);
76 for (i = 0; i < count; i += 4) {
77 gba->memory.d.store32(&gba->memory.d, dest + ((i + 0) << 2), word, 0);
78 gba->memory.d.store32(&gba->memory.d, dest + ((i + 1) << 2), word, 0);
79 gba->memory.d.store32(&gba->memory.d, dest + ((i + 2) << 2), word, 0);
80 gba->memory.d.store32(&gba->memory.d, dest + ((i + 3) << 2), word, 0);
81 gba->cpu.cycles += storeCycles;
82 gba->board.d.processEvents(&gba->board.d);
83 }
84 } else {
85 int loadCycles = gba->memory.d.waitMultiple(&gba->memory.d, source, 4);
86 for (i = 0; i < count; i += 4) {
87 int32_t word0 = gba->memory.d.load32(&gba->memory.d, source + ((i + 0) << 2), 0);
88 int32_t word1 = gba->memory.d.load32(&gba->memory.d, source + ((i + 1) << 2), 0);
89 int32_t word2 = gba->memory.d.load32(&gba->memory.d, source + ((i + 2) << 2), 0);
90 int32_t word3 = gba->memory.d.load32(&gba->memory.d, source + ((i + 3) << 2), 0);
91 gba->cpu.cycles += loadCycles;
92 gba->board.d.processEvents(&gba->board.d);
93 gba->memory.d.store32(&gba->memory.d, dest + ((i + 0) << 2), word0, 0);
94 gba->memory.d.store32(&gba->memory.d, dest + ((i + 1) << 2), word1, 0);
95 gba->memory.d.store32(&gba->memory.d, dest + ((i + 2) << 2), word2, 0);
96 gba->memory.d.store32(&gba->memory.d, dest + ((i + 3) << 2), word3, 0);
97 gba->cpu.cycles += storeCycles;
98 gba->board.d.processEvents(&gba->board.d);
99 }
100 }
101}
102
103static void _BgAffineSet(struct GBA* gba) {
104 int i = gba->cpu.gprs[2];
105 float ox, oy;
106 float cx, cy;
107 float sx, sy;
108 float theta;
109 int offset = gba->cpu.gprs[0];
110 int destination = gba->cpu.gprs[1];
111 int diff = gba->cpu.gprs[3];
112 (void)(diff); // Are we supposed to use this?
113 float a, b, c, d;
114 float rx, ry;
115 while (i--) {
116 // [ sx 0 0 ] [ cos(theta) -sin(theta) 0 ] [ 1 0 cx - ox ] [ A B rx ]
117 // [ 0 sy 0 ] * [ sin(theta) cos(theta) 0 ] * [ 0 1 cy - oy ] = [ C D ry ]
118 // [ 0 0 1 ] [ 0 0 1 ] [ 0 0 1 ] [ 0 0 1 ]
119 ox = gba->memory.d.load32(&gba->memory.d, offset, 0) / 256.f;
120 oy = gba->memory.d.load32(&gba->memory.d, offset + 4, 0) / 256.f;
121 cx = gba->memory.d.load16(&gba->memory.d, offset + 8, 0);
122 cy = gba->memory.d.load16(&gba->memory.d, offset + 10, 0);
123 sx = gba->memory.d.load16(&gba->memory.d, offset + 12, 0) / 256.f;
124 sy = gba->memory.d.load16(&gba->memory.d, offset + 14, 0) / 256.f;
125 theta = (gba->memory.d.loadU16(&gba->memory.d, offset + 16, 0) >> 8) / 128.f * M_PI;
126 offset += 20;
127 // Rotation
128 a = d = cosf(theta);
129 b = c = sinf(theta);
130 // Scale
131 a *= sx;
132 b *= -sx;
133 c *= sy;
134 d *= sy;
135 // Translate
136 rx = ox - (a * cx + b * cy);
137 ry = oy - (c * cx + d * cy);
138 gba->memory.d.store16(&gba->memory.d, destination, a * 256, 0);
139 gba->memory.d.store16(&gba->memory.d, destination + 2, b * 256, 0);
140 gba->memory.d.store16(&gba->memory.d, destination + 4, c * 256, 0);
141 gba->memory.d.store16(&gba->memory.d, destination + 6, d * 256, 0);
142 gba->memory.d.store32(&gba->memory.d, destination + 8, rx * 256, 0);
143 gba->memory.d.store32(&gba->memory.d, destination + 12, ry * 256, 0);
144 destination += 16;
145 }
146}
147
148static void _ObjAffineSet(struct GBA* gba) {
149 int i = gba->cpu.gprs[2];
150 float sx, sy;
151 float theta;
152 int offset = gba->cpu.gprs[0];
153 int destination = gba->cpu.gprs[1];
154 int diff = gba->cpu.gprs[3];
155 float a, b, c, d;
156 while (i--) {
157 // [ sx 0 ] [ cos(theta) -sin(theta) ] [ A B ]
158 // [ 0 sy ] * [ sin(theta) cos(theta) ] = [ C D ]
159 sx = gba->memory.d.load16(&gba->memory.d, offset, 0) / 256.f;
160 sy = gba->memory.d.load16(&gba->memory.d, offset + 2, 0) / 256.f;
161 theta = (gba->memory.d.loadU16(&gba->memory.d, offset + 4, 0) >> 8) / 128.f * M_PI;
162 offset += 6;
163 // Rotation
164 a = d = cosf(theta);
165 b = c = sinf(theta);
166 // Scale
167 a *= sx;
168 b *= -sx;
169 c *= sy;
170 d *= sy;
171 gba->memory.d.store16(&gba->memory.d, destination, a * 256, 0);
172 gba->memory.d.store16(&gba->memory.d, destination + diff, b * 256, 0);
173 gba->memory.d.store16(&gba->memory.d, destination + diff * 2, c * 256, 0);
174 gba->memory.d.store16(&gba->memory.d, destination + diff * 3, d * 256, 0);
175 destination += diff * 4;
176 }
177}
178
179static void _MidiKey2Freq(struct GBA* gba) {
180 uint32_t key = gba->memory.d.load32(&gba->memory.d, gba->cpu.gprs[0] + 4, 0);
181 gba->cpu.gprs[0] = key / powf(2, (180.f - gba->cpu.gprs[1] - gba->cpu.gprs[2] / 256.f) / 12.f);
182}
183
184void GBASwi16(struct ARMBoard* board, int immediate) {
185 struct GBA* gba = ((struct GBABoard*) board)->p;
186 if (gba->memory.fullBios) {
187 ARMRaiseSWI(&gba->cpu);
188 return;
189 }
190 switch (immediate) {
191 case 0x1:
192 _RegisterRamReset(gba);
193 break;
194 case 0x2:
195 GBAHalt(gba);
196 break;
197 case 0x05:
198 // VBlankIntrWait
199 gba->cpu.gprs[0] = 1;
200 gba->cpu.gprs[1] = 1;
201 // Fall through:
202 case 0x04:
203 // IntrWait
204 gba->memory.io[REG_IME >> 1] = 1;
205 if (!gba->cpu.gprs[0] && gba->memory.io[REG_IF >> 1] & gba->cpu.gprs[1]) {
206 break;
207 }
208 gba->memory.io[REG_IF >> 1] = 0;
209 ARMRaiseSWI(&gba->cpu);
210 break;
211 case 0x6:
212 {
213 div_t result = div(gba->cpu.gprs[0], gba->cpu.gprs[1]);
214 gba->cpu.gprs[0] = result.quot;
215 gba->cpu.gprs[1] = result.rem;
216 gba->cpu.gprs[3] = abs(result.quot);
217 }
218 break;
219 case 0x7:
220 {
221 div_t result = div(gba->cpu.gprs[1], gba->cpu.gprs[0]);
222 gba->cpu.gprs[0] = result.quot;
223 gba->cpu.gprs[1] = result.rem;
224 gba->cpu.gprs[3] = abs(result.quot);
225 }
226 break;
227 case 0x8:
228 gba->cpu.gprs[0] = sqrt(gba->cpu.gprs[0]);
229 break;
230 case 0xA:
231 gba->cpu.gprs[0] = atan2f(gba->cpu.gprs[1] / 16384.f, gba->cpu.gprs[0] / 16384.f) / (2 * M_PI) * 0x10000;
232 break;
233 case 0xB:
234 _CpuSet(gba);
235 break;
236 case 0xC:
237 _FastCpuSet(gba);
238 break;
239 case 0xE:
240 _BgAffineSet(gba);
241 break;
242 case 0xF:
243 _ObjAffineSet(gba);
244 break;
245 case 0x11:
246 case 0x12:
247 switch (gba->cpu.gprs[1] >> BASE_OFFSET) {
248 case REGION_WORKING_RAM:
249 _unLz77(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->memory.wram)[(gba->cpu.gprs[1] & (SIZE_WORKING_RAM - 1))]);
250 break;
251 case REGION_WORKING_IRAM:
252 _unLz77(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->memory.iwram)[(gba->cpu.gprs[1] & (SIZE_WORKING_IRAM - 1))]);
253 break;
254 case REGION_VRAM:
255 _unLz77(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->video.renderer->vram)[(gba->cpu.gprs[1] & 0x0001FFFF)]);
256 break;
257 default:
258 GBALog(gba, GBA_LOG_WARN, "Bad LZ77 destination");
259 break;
260 }
261 break;
262 case 0x14:
263 case 0x15:
264 switch (gba->cpu.gprs[1] >> BASE_OFFSET) {
265 case REGION_WORKING_RAM:
266 _unRl(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->memory.wram)[(gba->cpu.gprs[1] & (SIZE_WORKING_RAM - 1))]);
267 break;
268 case REGION_WORKING_IRAM:
269 _unRl(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->memory.iwram)[(gba->cpu.gprs[1] & (SIZE_WORKING_IRAM - 1))]);
270 break;
271 case REGION_VRAM:
272 _unRl(&gba->memory, gba->cpu.gprs[0], &((uint8_t*) gba->video.renderer->vram)[(gba->cpu.gprs[1] & 0x0001FFFF)]);
273 break;
274 default:
275 GBALog(gba, GBA_LOG_WARN, "Bad RL destination");
276 break;
277 }
278 break;
279 case 0x1F:
280 _MidiKey2Freq(gba);
281 break;
282 default:
283 GBALog(gba, GBA_LOG_STUB, "Stub software interrupt: %02x", immediate);
284 }
285}
286
287void GBASwi32(struct ARMBoard* board, int immediate) {
288 GBASwi16(board, immediate >> 16);
289}
290
291static void _unLz77(struct GBAMemory* memory, uint32_t source, uint8_t* dest) {
292 int remaining = (memory->d.load32(&memory->d, source, 0) & 0xFFFFFF00) >> 8;
293 // We assume the signature byte (0x10) is correct
294 int blockheader;
295 uint32_t sPointer = source + 4;
296 uint8_t* dPointer = dest;
297 int blocksRemaining = 0;
298 int block;
299 uint8_t* disp;
300 int bytes;
301 while (remaining > 0) {
302 if (blocksRemaining) {
303 if (blockheader & 0x80) {
304 // Compressed
305 block = memory->d.loadU8(&memory->d, sPointer, 0) | (memory->d.loadU8(&memory->d, sPointer + 1, 0) << 8);
306 sPointer += 2;
307 disp = dPointer - (((block & 0x000F) << 8) | ((block & 0xFF00) >> 8)) - 1;
308 bytes = ((block & 0x00F0) >> 4) + 3;
309 while (bytes-- && remaining) {
310 --remaining;
311 *dPointer = *disp;
312 ++disp;
313 ++dPointer;
314 }
315 } else {
316 // Uncompressed
317 *dPointer = memory->d.loadU8(&memory->d, sPointer++, 0);
318 ++dPointer;
319 --remaining;
320 }
321 blockheader <<= 1;
322 --blocksRemaining;
323 } else {
324 blockheader = memory->d.loadU8(&memory->d, sPointer++, 0);
325 blocksRemaining = 8;
326 }
327 }
328}
329
330static void _unRl(struct GBAMemory* memory, uint32_t source, uint8_t* dest) {
331 source = source & 0xFFFFFFFC;
332 int remaining = (memory->d.load32(&memory->d, source, 0) & 0xFFFFFF00) >> 8;
333 int padding = (4 - remaining) & 0x3;
334 // We assume the signature byte (0x30) is correct
335 int blockheader;
336 int block;
337 uint32_t sPointer = source + 4;
338 uint8_t* dPointer = dest;
339 while (remaining > 0) {
340 blockheader = memory->d.loadU8(&memory->d, sPointer++, 0);
341 if (blockheader & 0x80) {
342 // Compressed
343 blockheader &= 0x7F;
344 blockheader += 3;
345 block = memory->d.loadU8(&memory->d, sPointer++, 0);
346 while (blockheader-- && remaining) {
347 --remaining;
348 *dPointer = block;
349 ++dPointer;
350 }
351 } else {
352 // Uncompressed
353 blockheader++;
354 while (blockheader-- && remaining) {
355 --remaining;
356 *dPointer = memory->d.loadU8(&memory->d, sPointer++, 0);
357 ++dPointer;
358 }
359 }
360 }
361 while (padding--) {
362 *dPointer = 0;
363 ++dPointer;
364 }
365}