5 // GCC/SDL port by Niels Wagenaar (Linux/WIN32) and Caz (BeOS)
6 // Cleanups/fixes by James L. Hammons
8 bool specialLog = false;
12 #define REG(A) (((uint32)blitter_ram[(A)] << 24) | ((uint32)blitter_ram[(A)+1] << 16) \
13 | ((uint32)blitter_ram[(A)+2] << 8) | (uint32)blitter_ram[(A)+3])
14 #define WREG(A,D) (blitter_ram[(A)] = ((D)>>24)&0xFF, blitter_ram[(A)+1] = ((D)>>16)&0xFF, \
15 blitter_ram[(A)+2] = ((D)>>8)&0xFF, blitter_ram[(A)+3] = (D)&0xFF)
17 // Blitter registers (offsets from F02200)
19 #define A1_BASE ((UINT32)0x00)
20 #define A1_FLAGS ((UINT32)0x04)
21 #define A1_CLIP ((UINT32)0x08) // Height and width values for clipping
22 #define A1_PIXEL ((UINT32)0x0C) // Integer part of the pixel (Y.i and X.i)
23 #define A1_STEP ((UINT32)0x10) // Integer part of the step
24 #define A1_FSTEP ((UINT32)0x14) // Fractionnal part of the step
25 #define A1_FPIXEL ((UINT32)0x18) // Fractionnal part of the pixel (Y.f and X.f)
26 #define A1_INC ((UINT32)0x1C) // Integer part of the increment
27 #define A1_FINC ((UINT32)0x20) // Fractional part of the increment
28 #define A2_BASE ((UINT32)0x24)
29 #define A2_FLAGS ((UINT32)0x28)
30 #define A2_MASK ((UINT32)0x2C) // Modulo values for x and y (M.y and M.x)
31 #define A2_PIXEL ((UINT32)0x30) // Integer part of the pixel (no fractional part for A2)
32 #define A2_STEP ((UINT32)0x34) // Integer part of the step (no fractional part for A2)
33 #define COMMAND ((UINT32)0x38)
34 #define PIXLINECOUNTER ((UINT32)0x3C)
35 #define SRCDATA ((UINT32)0x40)
36 #define DSTDATA ((UINT32)0x48)
37 #define DSTZ ((UINT32)0x50)
38 #define SRCZINT ((UINT32)0x58)
39 #define SRCZFRAC ((UINT32)0x60)
40 #define PATTERNDATA ((UINT32)0x68)
41 #define INTENSITYINC ((UINT32)0x70)
42 #define ZINC ((UINT32)0x74)
43 #define COLLISIONCTRL ((UINT32)0x78)
44 #define PHRASEINT3 ((UINT32)0x7C)
45 #define PHRASEINT2 ((UINT32)0x80)
46 #define PHRASEINT1 ((UINT32)0x84)
47 #define PHRASEINT0 ((UINT32)0x88)
48 #define PHRASEZ3 ((UINT32)0x8C)
49 #define PHRASEZ2 ((UINT32)0x90)
50 #define PHRASEZ1 ((UINT32)0x94)
51 #define PHRASEZ0 ((UINT32)0x98)
53 // Blitter command bits
55 #define SRCEN (cmd & 0x00000001)
56 #define SRCENZ (cmd & 0x00000002)
57 #define SRCENX (cmd & 0x00000004)
58 #define DSTEN (cmd & 0x00000008)
59 #define DSTENZ (cmd & 0x00000010)
60 #define DSTWRZ (cmd & 0x00000020)
61 #define CLIPA1 (cmd & 0x00000040)
63 #define UPDA1F (cmd & 0x00000100)
64 #define UPDA1 (cmd & 0x00000200)
65 #define UPDA2 (cmd & 0x00000400)
67 #define DSTA2 (cmd & 0x00000800)
69 #define Z_OP_INF (cmd & 0x00040000)
70 #define Z_OP_EQU (cmd & 0x00080000)
71 #define Z_OP_SUP (cmd & 0x00100000)
73 #define LFU_NAN (cmd & 0x00200000)
74 #define LFU_NA (cmd & 0x00400000)
75 #define LFU_AN (cmd & 0x00800000)
76 #define LFU_A (cmd & 0x01000000)
78 #define CMPDST (cmd & 0x02000000)
79 #define BCOMPEN (cmd & 0x04000000)
80 #define DCOMPEN (cmd & 0x08000000)
82 #define PATDSEL (cmd & 0x00010000)
83 #define INTADD (cmd & 0x00020000)
84 #define TOPBEN (cmd & 0x00004000)
85 #define TOPNEN (cmd & 0x00008000)
86 #define BKGWREN (cmd & 0x10000000)
87 #define GOURD (cmd & 0x00001000)
88 #define GOURZ (cmd & 0x00002000)
89 #define SRCSHADE (cmd & 0x40000000)
97 #define XSIGNSUB_A1 (REG(A1_FLAGS)&0x080000)
98 #define XSIGNSUB_A2 (REG(A2_FLAGS)&0x080000)
100 #define YSIGNSUB_A1 (REG(A1_FLAGS)&0x100000)
101 #define YSIGNSUB_A2 (REG(A2_FLAGS)&0x100000)
103 #define YADD1_A1 (REG(A1_FLAGS)&0x040000)
104 #define YADD1_A2 (REG(A2_FLAGS)&0x040000)
106 //Put 'em back, once we fix the problem!!! [KO]
108 #define PIXEL_SHIFT_1(a) (((~a##_x) >> 16) & 7)
109 #define PIXEL_OFFSET_1(a) (((((UINT32)a##_y >> 16) * a##_width / 8) + (((UINT32)a##_x >> 19) & ~7)) * (1 + a##_pitch) + (((UINT32)a##_x >> 19) & 7))
110 #define READ_PIXEL_1(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_1(a), BLITTER) >> PIXEL_SHIFT_1(a)) & 0x01)
111 //#define READ_PIXEL_1(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_1(a)) >> PIXEL_SHIFT_1(a)) & 0x01)
114 #define PIXEL_SHIFT_2(a) (((~a##_x) >> 15) & 6)
115 #define PIXEL_OFFSET_2(a) (((((UINT32)a##_y >> 16) * a##_width / 4) + (((UINT32)a##_x >> 18) & ~7)) * (1 + a##_pitch) + (((UINT32)a##_x >> 18) & 7))
116 #define READ_PIXEL_2(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_2(a), BLITTER) >> PIXEL_SHIFT_2(a)) & 0x03)
117 //#define READ_PIXEL_2(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_2(a)) >> PIXEL_SHIFT_2(a)) & 0x03)
120 #define PIXEL_SHIFT_4(a) (((~a##_x) >> 14) & 4)
121 #define PIXEL_OFFSET_4(a) (((((UINT32)a##_y >> 16) * (a##_width/2)) + (((UINT32)a##_x >> 17) & ~7)) * (1 + a##_pitch) + (((UINT32)a##_x >> 17) & 7))
122 #define READ_PIXEL_4(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_4(a), BLITTER) >> PIXEL_SHIFT_4(a)) & 0x0f)
123 //#define READ_PIXEL_4(a) ((JaguarReadByte(a##_addr+PIXEL_OFFSET_4(a)) >> PIXEL_SHIFT_4(a)) & 0x0f)
126 #define PIXEL_OFFSET_8(a) (((((UINT32)a##_y >> 16) * a##_width) + (((UINT32)a##_x >> 16) & ~7)) * (1 + a##_pitch) + (((UINT32)a##_x >> 16) & 7))
127 #define READ_PIXEL_8(a) (JaguarReadByte(a##_addr+PIXEL_OFFSET_8(a), BLITTER))
128 //#define READ_PIXEL_8(a) (JaguarReadByte(a##_addr+PIXEL_OFFSET_8(a)))
131 #define PIXEL_OFFSET_16(a) (((((UINT32)a##_y >> 16) * a##_width) + (((UINT32)a##_x >> 16) & ~3)) * (1 + a##_pitch) + (((UINT32)a##_x >> 16) & 3))
132 #define READ_PIXEL_16(a) (JaguarReadWord(a##_addr+(PIXEL_OFFSET_16(a)<<1), BLITTER))
133 //#define READ_PIXEL_16(a) (JaguarReadWord(a##_addr+(PIXEL_OFFSET_16(a)<<1)))
136 #define PIXEL_OFFSET_32(a) (((((UINT32)a##_y >> 16) * a##_width) + (((UINT32)a##_x >> 16) & ~1)) * (1 + a##_pitch) + (((UINT32)a##_x >> 16) & 1))
137 #define READ_PIXEL_32(a) (JaguarReadLong(a##_addr+(PIXEL_OFFSET_32(a)<<2), BLITTER))
138 //#define READ_PIXEL_32(a) (JaguarReadLong(a##_addr+(PIXEL_OFFSET_32(a)<<2)))
141 #define READ_PIXEL(a,f) (\
142 (((f>>3)&0x07) == 0) ? (READ_PIXEL_1(a)) : \
143 (((f>>3)&0x07) == 1) ? (READ_PIXEL_2(a)) : \
144 (((f>>3)&0x07) == 2) ? (READ_PIXEL_4(a)) : \
145 (((f>>3)&0x07) == 3) ? (READ_PIXEL_8(a)) : \
146 (((f>>3)&0x07) == 4) ? (READ_PIXEL_16(a)) : \
147 (((f>>3)&0x07) == 5) ? (READ_PIXEL_32(a)) : 0)
149 // 16 bpp z data read
150 #define ZDATA_OFFSET_16(a) (PIXEL_OFFSET_16(a) + a##_zoffs * 4)
151 #define READ_ZDATA_16(a) (JaguarReadWord(a##_addr+(ZDATA_OFFSET_16(a)<<1), BLITTER))
152 //#define READ_ZDATA_16(a) (JaguarReadWord(a##_addr+(ZDATA_OFFSET_16(a)<<1)))
155 #define READ_ZDATA(a,f) (READ_ZDATA_16(a))
157 // 16 bpp z data write
158 #define WRITE_ZDATA_16(a,d) { JaguarWriteWord(a##_addr+(ZDATA_OFFSET_16(a)<<1), d, BLITTER); }
159 //#define WRITE_ZDATA_16(a,d) { JaguarWriteWord(a##_addr+(ZDATA_OFFSET_16(a)<<1), d); }
162 #define WRITE_ZDATA(a,f,d) WRITE_ZDATA_16(a,d);
165 #define READ_RDATA_1(r,a,p) ((p) ? ((REG(r+(((UINT32)a##_x>>19)&4)))>>(((UINT32)a##_x>>16)&0x1f))& 0x1 : (REG(r) & 0x1))
168 #define READ_RDATA_2(r,a,p) ((p) ? ((REG(r+(((UINT32)a##_x>>18)&4)))>>(((UINT32)a##_x>>15)&0x3e))& 0x3 : (REG(r) & 0x3))
171 #define READ_RDATA_4(r,a,p) ((p) ? ((REG(r+(((UINT32)a##_x>>17)&4)))>>(((UINT32)a##_x>>14)&0x28))& 0xf : (REG(r) & 0xf))
174 #define READ_RDATA_8(r,a,p) ((p) ? ((REG(r+(((UINT32)a##_x>>16)&4)))>>(((UINT32)a##_x>>13)&0x18))& 0xff : (REG(r) & 0xff))
176 // 16 bpp r data read
177 #define READ_RDATA_16(r,a,p) ((p) ? ((REG(r+(((UINT32)a##_x>>15)&4)))>>(((UINT32)a##_x>>12)&0x10))&0xffff : (REG(r) & 0xffff))
179 // 32 bpp r data read
180 #define READ_RDATA_32(r,a,p) ((p) ? REG(r+(((UINT32)a##_x>>14)&4)) : REG(r))
182 // register data read
183 #define READ_RDATA(r,a,f,p) (\
184 (((f>>3)&0x07) == 0) ? (READ_RDATA_1(r,a,p)) : \
185 (((f>>3)&0x07) == 1) ? (READ_RDATA_2(r,a,p)) : \
186 (((f>>3)&0x07) == 2) ? (READ_RDATA_4(r,a,p)) : \
187 (((f>>3)&0x07) == 3) ? (READ_RDATA_8(r,a,p)) : \
188 (((f>>3)&0x07) == 4) ? (READ_RDATA_16(r,a,p)) : \
189 (((f>>3)&0x07) == 5) ? (READ_RDATA_32(r,a,p)) : 0)
192 #define WRITE_PIXEL_1(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_1(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_1(a), BLITTER)&(~(0x01 << PIXEL_SHIFT_1(a))))|(d<<PIXEL_SHIFT_1(a)), BLITTER); }
193 //#define WRITE_PIXEL_1(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_1(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_1(a))&(~(0x01 << PIXEL_SHIFT_1(a))))|(d<<PIXEL_SHIFT_1(a))); }
196 #define WRITE_PIXEL_2(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_2(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_2(a), BLITTER)&(~(0x03 << PIXEL_SHIFT_2(a))))|(d<<PIXEL_SHIFT_2(a)), BLITTER); }
197 //#define WRITE_PIXEL_2(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_2(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_2(a))&(~(0x03 << PIXEL_SHIFT_2(a))))|(d<<PIXEL_SHIFT_2(a))); }
200 #define WRITE_PIXEL_4(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_4(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_4(a), BLITTER)&(~(0x0f << PIXEL_SHIFT_4(a))))|(d<<PIXEL_SHIFT_4(a)), BLITTER); }
201 //#define WRITE_PIXEL_4(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_4(a), (JaguarReadByte(a##_addr+PIXEL_OFFSET_4(a))&(~(0x0f << PIXEL_SHIFT_4(a))))|(d<<PIXEL_SHIFT_4(a))); }
204 #define WRITE_PIXEL_8(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_8(a), d, BLITTER); }
205 //#define WRITE_PIXEL_8(a,d) { JaguarWriteByte(a##_addr+PIXEL_OFFSET_8(a), d); }
207 // 16 bpp pixel write
208 //#define WRITE_PIXEL_16(a,d) { JaguarWriteWord(a##_addr+(PIXEL_OFFSET_16(a)<<1),d); }
209 #define WRITE_PIXEL_16(a,d) { JaguarWriteWord(a##_addr+(PIXEL_OFFSET_16(a)<<1), d, BLITTER); if (specialLog) WriteLog("Pixel write address: %08X\n", a##_addr+(PIXEL_OFFSET_16(a)<<1)); }
210 //#define WRITE_PIXEL_16(a,d) { JaguarWriteWord(a##_addr+(PIXEL_OFFSET_16(a)<<1), d); if (specialLog) WriteLog("Pixel write address: %08X\n", a##_addr+(PIXEL_OFFSET_16(a)<<1)); }
212 // 32 bpp pixel write
213 #define WRITE_PIXEL_32(a,d) { JaguarWriteLong(a##_addr+(PIXEL_OFFSET_32(a)<<2), d, BLITTER); }
214 //#define WRITE_PIXEL_32(a,d) { JaguarWriteLong(a##_addr+(PIXEL_OFFSET_32(a)<<2), d); }
217 #define WRITE_PIXEL(a,f,d) {\
218 switch ((f>>3)&0x07) { \
219 case 0: WRITE_PIXEL_1(a,d); break; \
220 case 1: WRITE_PIXEL_2(a,d); break; \
221 case 2: WRITE_PIXEL_4(a,d); break; \
222 case 3: WRITE_PIXEL_8(a,d); break; \
223 case 4: WRITE_PIXEL_16(a,d); break; \
224 case 5: WRITE_PIXEL_32(a,d); break; \
227 // External global variables
229 extern int jaguar_active_memory_dumps;
231 // Local global variables
233 int start_logging = 0;
234 uint8 blitter_working = 0;
236 // Blitter register RAM (most of it is hidden from the user)
238 static uint8 blitter_ram[0x100];
240 // Width in Pixels of a Scanline
241 // This is a pretranslation of the value found in the A1 & A2 flags: It's really a floating point value
242 // of the form EEEEMM where MM is the mantissa with an implied "1." in front of it and the EEEE value is
243 // the exponent. Valid values for the exponent range from 0 to 11 (decimal). It's easiest to think of it
244 // as a floating point bit pattern being followed by a number of zeroes. So, e.g., 001101 translates to
245 // 1.01 (the "1." being implied) x (2 ^ 3) or 1010 -> 10 in base 10 (i.e., 1.01 with the decimal place
246 // being shifted to the right 3 places).
247 static uint32 blitter_scanline_width[48] =
249 0, 0, 0, 0, // Note: This would really translate to 1, 1, 1, 1
259 1024, 1280, 1536, 1792,
260 2048, 2560, 3072, 3584
263 //static uint8 * tom_ram_8;
264 //static uint8 * paletteRam;
274 static uint32 a1_addr;
275 static uint32 a2_addr;
276 static int32 a1_zoffs;
277 static int32 a2_zoffs;
278 static uint32 xadd_a1_control;
279 static uint32 xadd_a2_control;
280 static int32 a1_pitch;
281 static int32 a2_pitch;
282 static uint32 n_pixels;
283 static uint32 n_lines;
286 static int32 a1_width;
289 static int32 a2_width;
290 static int32 a2_mask_x;
291 static int32 a2_mask_y;
292 static int32 a1_xadd;
293 static int32 a1_yadd;
294 static int32 a2_xadd;
295 static int32 a2_yadd;
296 static uint8 a1_phrase_mode;
297 static uint8 a2_phrase_mode;
298 static int32 a1_step_x = 0;
299 static int32 a1_step_y = 0;
300 static int32 a2_step_x = 0;
301 static int32 a2_step_y = 0;
302 static uint32 outer_loop;
303 static uint32 inner_loop;
304 static uint32 a2_psize;
305 static uint32 a1_psize;
306 static uint32 gouraud_add;
307 //static uint32 gouraud_data;
308 //static uint16 gint[4];
309 //static uint16 gfrac[4];
310 //static uint8 gcolour[4];
313 static int gd_ia, gd_ca;
314 static int colour_index = 0;
316 static uint32 z_i[4];
318 static int32 a1_clip_x, a1_clip_y;
320 // In the spirit of "get it right first, *then* optimize" I've taken the liberty
321 // of removing all the unnecessary code caching. If it turns out to be a good way
322 // to optimize the blitter, then we may revisit it in the future...
325 // Generic blit handler
327 void blitter_generic(uint32 cmd)
330 //uint32 logGo = ((cmd == 0x01800E01 && REG(A1_BASE) == 0x898000) ? 1 : 0);
331 uint32 srcdata, srczdata, dstdata, dstzdata, writedata, inhibit;
335 WriteLog("About to do 8x8 blit (BM width is 448 pixels)...\n");
341 WriteLog(" A1_X/Y = %08X/%08X, A2_X/Y = %08X/%08X\n", a1_x, a1_y, a2_x, a2_y);
343 uint32 a1_start = a1_x, a2_start = a2_x;
345 inner_loop = n_pixels;
350 WriteLog(" A1_X/Y = %08X/%08X, A2_X/Y = %08X/%08X\n", a1_x, a1_y, a2_x, a2_y);
352 srcdata = srczdata = dstdata = dstzdata = writedata = inhibit = 0;
356 // load src data and Z
359 srcdata = READ_PIXEL(a2, REG(A2_FLAGS));
361 srczdata = READ_ZDATA(a2, REG(A2_FLAGS));
362 else if (cmd & 0x0001C020)
363 srczdata = READ_RDATA(SRCZINT, a2, REG(A2_FLAGS), a2_phrase_mode);
367 srcdata = READ_RDATA(SRCDATA, a2, REG(A2_FLAGS), a2_phrase_mode);
368 if (cmd & 0x0001C020)
369 srczdata = READ_RDATA(SRCZINT, a2, REG(A2_FLAGS), a2_phrase_mode);
372 // load dst data and Z
375 dstdata = READ_PIXEL(a1, REG(A1_FLAGS));
377 dstzdata = READ_ZDATA(a1, REG(A1_FLAGS));
379 dstzdata = READ_RDATA(DSTZ, a1, REG(A1_FLAGS), a1_phrase_mode);
383 dstdata = READ_RDATA(DSTDATA, a1, REG(A1_FLAGS), a1_phrase_mode);
385 dstzdata = READ_RDATA(DSTZ, a1, REG(A1_FLAGS), a1_phrase_mode);
388 /*This wasn't working... // a1 clipping
389 if (cmd & 0x00000040)
391 if (a1_x < 0 || a1_y < 0 || (a1_x >> 16) >= (REG(A1_CLIP) & 0x7FFF)
392 || (a1_y >> 16) >= ((REG(A1_CLIP) >> 16) & 0x7FFF))
397 srczdata = z_i[colour_index] >> 16;
399 // apply z comparator
400 if (Z_OP_INF) if (srczdata < dstzdata) inhibit = 1;
401 if (Z_OP_EQU) if (srczdata == dstzdata) inhibit = 1;
402 if (Z_OP_SUP) if (srczdata > dstzdata) inhibit = 1;
404 // apply data comparator
405 // Note: DCOMPEN only works in 8/16 bpp modes! !!! FIX !!!
406 // Does BCOMPEN only work in 1 bpp mode???
407 if (DCOMPEN | BCOMPEN)
411 //WriteLog("Blitter: BCOMPEN set on command %08X inhibit prev:%u, now:", cmd, inhibit);
412 // compare source pixel with pattern pixel
414 Blit! (000B8250 <- 0012C3A0) count: 16 x 1, A1/2_FLAGS: 00014420/00012000 [cmd: 05810001]
415 CMD -> src: SRCEN dst: misc: a1ctl: mode: ity: PATDSEL z-op: op: LFU_REPLACE ctrl: BCOMPEN
416 A1 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 384 (22), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
417 A2 -> pitch: 1 phrases, depth: 1bpp, z-off: 0, width: 16 (10), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
421 // AvP is still wrong, could be cuz it's doing A1 -> A2...
423 // Src is the 1bpp bitmap... DST is the PATTERN!!!
424 // This seems to solve at least ONE of the problems with MC3D...
425 // Why should this be inverted???
426 // Bcuz it is. This is supposed to be used only for a bit -> pixel expansion...
427 /* if (srcdata == READ_RDATA(PATTERNDATA, a2, REG(A2_FLAGS), a2_phrase_mode))
428 // if (srcdata != READ_RDATA(PATTERNDATA, a2, REG(A2_FLAGS), a2_phrase_mode))
430 /* uint32 A2bpp = 1 << ((REG(A2_FLAGS) >> 3) & 0x07);
431 if (A2bpp == 1 || A2bpp == 16 || A2bpp == 8)
432 inhibit = (srcdata == 0 ? 1: 0);
433 // inhibit = !srcdata;
435 WriteLog("Blitter: Bad BPP (%u) selected for BCOMPEN mode!\n", A2bpp);//*/
436 // What it boils down to is this:
442 // compare destination pixel with pattern pixel
443 if (dstdata == READ_RDATA(PATTERNDATA, a1, REG(A1_FLAGS), a1_phrase_mode))
444 // if (dstdata != READ_RDATA(PATTERNDATA, a1, REG(A1_FLAGS), a1_phrase_mode))
448 // This is DEFINITELY WRONG
449 // if (a1_phrase_mode || a2_phrase_mode)
450 // inhibit = !inhibit;
455 inhibit |= (((a1_x >> 16) < a1_clip_x && (a1_x >> 16) >= 0
456 && (a1_y >> 16) < a1_clip_y && (a1_y >> 16) >= 0) ? 0 : 1);
459 // compute the write data and store
464 // use pattern data for write data
465 writedata = READ_RDATA(PATTERNDATA, a1, REG(A1_FLAGS), a1_phrase_mode);
469 // intensity addition
470 writedata = (srcdata & 0xFF) + (dstdata & 0xFF);
471 if (!(TOPBEN) && writedata > 0xFF)
473 writedata |= (srcdata & 0xF00) + (dstdata & 0xF00);
474 if (!(TOPNEN) && writedata > 0xFFF)
476 writedata |= (srcdata & 0xF000) + (dstdata & 0xF000);
480 if (LFU_NAN) writedata |= ~srcdata & ~dstdata;
481 if (LFU_NA) writedata |= ~srcdata & dstdata;
482 if (LFU_AN) writedata |= srcdata & ~dstdata;
483 if (LFU_A) writedata |= srcdata & dstdata;
487 writedata = ((gd_c[colour_index]) << 8) | (gd_i[colour_index] >> 16);
491 int intensity = srcdata & 0xFF;
492 int ia = gd_ia >> 16;
494 ia = 0xFFFFFF00 | ia;
498 if (intensity > 0xFF)
500 writedata = (srcdata & 0xFF00) | intensity;
509 if (/*a1_phrase_mode || */BKGWREN || !inhibit)
511 /*if (((REG(A1_FLAGS) >> 3) & 0x07) == 5)
513 uint32 offset = a1_addr+(PIXEL_OFFSET_32(a1)<<2);
514 // (((((UINT32)a##_y >> 16) * a##_width) + (((UINT32)a##_x >> 16) & ~1)) * (1 + a##_pitch) + (((UINT32)a##_x >> 16) & 1))
515 if ((offset >= 0x1FF020 && offset <= 0x1FF03F) || (offset >= 0x1FF820 && offset <= 0x1FF83F))
516 WriteLog("32bpp pixel write: A1 Phrase mode --> ");
518 // write to the destination
519 WRITE_PIXEL(a1, REG(A1_FLAGS), writedata);
521 WRITE_ZDATA(a1, REG(A1_FLAGS), srczdata);
526 // load src data and Z
529 srcdata = READ_PIXEL(a1, REG(A1_FLAGS));
531 srczdata = READ_ZDATA(a1, REG(A1_FLAGS));
532 else if (cmd & 0x0001C020)
533 srczdata = READ_RDATA(SRCZINT, a1, REG(A1_FLAGS), a1_phrase_mode);
537 srcdata = READ_RDATA(SRCDATA, a1, REG(A1_FLAGS), a1_phrase_mode);
539 srczdata = READ_RDATA(SRCZINT, a1, REG(A1_FLAGS), a1_phrase_mode);
542 // load dst data and Z
545 dstdata = READ_PIXEL(a2, REG(A2_FLAGS));
547 dstzdata = READ_ZDATA(a2, REG(A2_FLAGS));
549 dstzdata = READ_RDATA(DSTZ, a2, REG(A2_FLAGS), a2_phrase_mode);
553 dstdata = READ_RDATA(DSTDATA, a2, REG(A2_FLAGS), a2_phrase_mode);
555 dstzdata = READ_RDATA(DSTZ, a2, REG(A2_FLAGS), a2_phrase_mode);
559 srczdata = z_i[colour_index] >> 16;
561 // apply z comparator
562 if (Z_OP_INF) if (srczdata < dstzdata) inhibit = 1;
563 if (Z_OP_EQU) if (srczdata == dstzdata) inhibit = 1;
564 if (Z_OP_SUP) if (srczdata > dstzdata) inhibit = 1;
566 // apply data comparator
567 //NOTE: The bit comparator (BCOMPEN) is NOT the same at the data comparator!
568 if (DCOMPEN | BCOMPEN)
572 // compare source pixel with pattern pixel
573 // AvP: Numbers are correct, but sprites are not!
574 //This doesn't seem to be a problem... But could still be wrong...
575 /* if (srcdata == READ_RDATA(PATTERNDATA, a1, REG(A1_FLAGS), a1_phrase_mode))
576 // if (srcdata != READ_RDATA(PATTERNDATA, a1, REG(A1_FLAGS), a1_phrase_mode))
578 // This is probably not 100% correct... It works in the 1bpp case
579 // (in A1 <- A2 mode, that is...)
580 // AvP: This is causing blocks to be written instead of bit patterns...
582 // NOTE: We really should separate out the BCOMPEN & DCOMPEN stuff!
583 /* uint32 A1bpp = 1 << ((REG(A1_FLAGS) >> 3) & 0x07);
584 if (A1bpp == 1 || A1bpp == 16 || A1bpp == 8)
585 inhibit = (srcdata == 0 ? 1: 0);
587 WriteLog("Blitter: Bad BPP (%u) selected for BCOMPEN mode!\n", A1bpp);//*/
588 // What it boils down to is this:
594 // compare destination pixel with pattern pixel
595 if (dstdata == READ_RDATA(PATTERNDATA, a2, REG(A2_FLAGS), a2_phrase_mode))
596 // if (dstdata != READ_RDATA(PATTERNDATA, a2, REG(A2_FLAGS), a2_phrase_mode))
600 // This is DEFINITELY WRONG
601 // if (a1_phrase_mode || a2_phrase_mode)
602 // inhibit = !inhibit;
607 inhibit |= (((a1_x >> 16) < a1_clip_x && (a1_x >> 16) >= 0
608 && (a1_y >> 16) < a1_clip_y && (a1_y >> 16) >= 0) ? 0 : 1);
611 // compute the write data and store
616 // use pattern data for write data
617 writedata= READ_RDATA(PATTERNDATA, a2, REG(A2_FLAGS), a2_phrase_mode);
621 // intensity addition
622 writedata = (srcdata & 0xFF) + (dstdata & 0xFF);
623 if (!(TOPBEN) && writedata > 0xFF)
625 writedata |= (srcdata & 0xF00) + (dstdata & 0xF00);
626 if (!(TOPNEN) && writedata > 0xFFF)
628 writedata |= (srcdata & 0xF000) + (dstdata & 0xF000);
633 writedata |= ~srcdata & ~dstdata;
635 writedata |= ~srcdata & dstdata;
637 writedata |= srcdata & ~dstdata;
639 writedata |= srcdata & dstdata;
643 writedata = ((gd_c[colour_index]) << 8) | (gd_i[colour_index] >> 16);
647 int intensity = srcdata & 0xFF;
648 int ia = gd_ia >> 16;
650 ia = 0xFFFFFF00 | ia;
654 if (intensity > 0xFF)
656 writedata = (srcdata & 0xFF00) | intensity;
665 if (/*a2_phrase_mode || */BKGWREN || !inhibit)
669 uint32 offset = a2_addr+(PIXEL_OFFSET_16(a2)<<1);
670 // (((((UINT32)a##_y >> 16) * a##_width) + (((UINT32)a##_x >> 16) & ~1)) * (1 + a##_pitch) + (((UINT32)a##_x >> 16) & 1))
671 WriteLog("[%08X:%04X] ", offset, writedata);
673 // write to the destination
674 WRITE_PIXEL(a2, REG(A2_FLAGS), writedata);
676 WRITE_ZDATA(a2, REG(A2_FLAGS), srczdata);
683 a2_x = (a2_x + a2_xadd) & a2_mask_x;
684 a2_y = (a2_y + a2_yadd) & a2_mask_y;
687 z_i[colour_index] += zadd;
689 if (GOURD || SRCSHADE)
691 gd_i[colour_index] += gd_ia;
692 gd_c[colour_index] += gd_ca;
694 if (GOURD || SRCSHADE || GOURZ)
697 colour_index = (colour_index + 1) & 0x03;
701 //New: Phrase mode taken into account! :-p
704 // Bump the pointer to the next phrase boundary
705 // Even though it works, this is crappy... Clean it up!
706 uint32 size = 64 / a1_psize;
708 // Crappy kludge... ('aligning' source to destination)
709 if (a2_phrase_mode && DSTA2)
711 uint32 extra = (a2_start >> 16) % size;
715 uint32 newx = (a1_x >> 16) / size;
716 uint32 newxrem = (a1_x >> 16) % size;
718 a1_x |= (((newx + (newxrem == 0 ? 0 : 1)) * size) & 0xFFFF) << 16;
723 // Bump the pointer to the next phrase boundary
724 // Even though it works, this is crappy... Clean it up!
725 uint32 size = 64 / a2_psize;
727 // Crappy kludge... ('aligning' source to destination)
728 // Prolly should do this for A1 channel as well...
729 if (a1_phrase_mode && !DSTA2)
731 uint32 extra = (a1_start >> 16) % size;
735 uint32 newx = (a2_x >> 16) / size;
736 uint32 newxrem = (a2_x >> 16) % size;
738 a2_x |= (((newx + (newxrem == 0 ? 0 : 1) /*+ extra*/) * size) & 0xFFFF) << 16;
741 //Not entirely: This still mucks things up... !!! FIX !!!
745 a2_y += a2_step_y;//*/
747 /* if (a2_phrase_mode)
749 a1_x+=(64/a1_psize)*a1_xadd;
753 for (int nb=0;nb<(64/a2_psize)+1;nb++)
754 a2_x = (a2_x + a2_xadd) & a2_mask_x;
758 // write values back to registers
759 WREG(A1_PIXEL, (a1_y & 0xFFFF0000) | ((a1_x >> 16) & 0xFFFF));
760 WREG(A1_FPIXEL, (a1_y << 16) | (a1_x & 0xFFFF));
761 WREG(A2_PIXEL, (a2_y & 0xFFFF0000) | ((a2_x >> 16) & 0xFFFF));
765 void blitter_blit(uint32 cmd)
767 uint32 pitchValue[4] = { 0, 1, 3, 2 };
770 dst = (cmd >> 3) & 0x07;
771 misc = (cmd >> 6) & 0x03;
772 a1ctl = (cmd >> 8) & 0x7;
773 mode = (cmd >> 11) & 0x07;
774 ity = (cmd >> 14) & 0x0F;
775 zop = (cmd >> 18) & 0x07;
776 op = (cmd >> 21) & 0x0F;
777 ctrl = (cmd >> 25) & 0x3F;
779 // Addresses in A1/2_BASE are *phrase* aligned, i.e., bottom three bits are ignored!
780 // NOTE: This fixes Rayman's bad collision detection AND keeps T2K working!
781 a1_addr = REG(A1_BASE) & 0xFFFFFFF8;
782 a2_addr = REG(A2_BASE) & 0xFFFFFFF8;
784 a1_zoffs = (REG(A1_FLAGS) >> 6) & 7;
785 a2_zoffs = (REG(A2_FLAGS) >> 6) & 7;
787 xadd_a1_control = (REG(A1_FLAGS) >> 16) & 0x03;
788 xadd_a2_control = (REG(A2_FLAGS) >> 16) & 0x03;
789 // a1_pitch = (REG(A1_FLAGS) & 3) ^ ((REG(A1_FLAGS) & 2) >> 1);
790 // a2_pitch = (REG(A2_FLAGS) & 3) ^ ((REG(A2_FLAGS) & 2) >> 1);
791 a1_pitch = pitchValue[(REG(A1_FLAGS) & 0x03)];
792 a2_pitch = pitchValue[(REG(A2_FLAGS) & 0x03)];
794 n_pixels = REG(PIXLINECOUNTER) & 0xFFFF;
795 n_lines = (REG(PIXLINECOUNTER) >> 16) & 0xFFFF;
797 a1_x = (REG(A1_PIXEL) << 16) | (REG(A1_FPIXEL) & 0xFFFF);
798 a1_y = (REG(A1_PIXEL) & 0xFFFF0000) | (REG(A1_FPIXEL) >> 16);
799 //According to the JTRM, X is restricted to 15 bits and Y is restricted to 12.
800 //But it seems to fuck up T2K! !!! FIX !!!
801 //Could it be sign extended??? Doesn't seem to be so according to JTRM
802 // a1_x &= 0x7FFFFFFF, a1_y &= 0x0FFFFFFF;
804 // a1_width = blitter_scanline_width[((REG(A1_FLAGS) & 0x00007E00) >> 9)];
805 // According to JTRM, this must give a *whole number* of phrases in the current
806 // pixel size (this means the lookup above is WRONG)... !!! FIX !!!
807 UINT32 m = (REG(A1_FLAGS) >> 9) & 0x03, e = (REG(A1_FLAGS) >> 11) & 0x0F;
808 a1_width = ((0x04 | m) << e) >> 2;//*/
810 a2_x = (REG(A2_PIXEL) & 0x0000FFFF) << 16;
811 a2_y = (REG(A2_PIXEL) & 0xFFFF0000);
812 //According to the JTRM, X is restricted to 15 bits and Y is restricted to 12.
813 //But it seems to fuck up T2K! !!! FIX !!!
814 // a2_x &= 0x7FFFFFFF, a2_y &= 0x0FFFFFFF;
816 // a2_width = blitter_scanline_width[((REG(A2_FLAGS) & 0x00007E00) >> 9)];
817 // According to JTRM, this must give a *whole number* of phrases in the current
818 // pixel size (this means the lookup above is WRONG)... !!! FIX !!!
819 m = (REG(A2_FLAGS) >> 9) & 0x03, e = (REG(A2_FLAGS) >> 11) & 0x0F;
820 a2_width = ((0x04 | m) << e) >> 2;//*/
821 a2_mask_x = ((REG(A2_MASK) & 0x0000FFFF) << 16) | 0xFFFF;
822 a2_mask_y = (REG(A2_MASK) & 0xFFFF0000) | 0xFFFF;
824 // Check for "use mask" flag
825 if (!(REG(A2_FLAGS) & 0x8000))
827 a2_mask_x = 0xFFFFFFFF; // must be 16.16
828 a2_mask_y = 0xFFFFFFFF; // must be 16.16
833 // According to the official documentation, a hardware bug ties A2's yadd bit to A1's...
834 a2_yadd = a1_yadd = (YADD1_A1 ? 1 << 16 : 0);
840 switch (xadd_a1_control)
843 // This is a documented Jaguar bug relating to phrase mode and truncation... Look into it!
844 // add phrase offset to X and truncate
849 // add pixelsize (1) to X
853 // add zero (for those nice vertical lines)
857 // add the contents of the increment register
858 a1_xadd = (REG(A1_INC) << 16) | (REG(A1_FINC) & 0xFFFF);
859 a1_yadd = (REG(A1_INC) & 0xFFFF0000) | (REG(A1_FINC) >> 16);
871 switch (xadd_a2_control)
874 // add phrase offset to X and truncate
879 // add pixelsize (1) to X
883 // add zero (for those nice vertical lines)
886 //This really isn't a valid bit combo for A2... Shouldn't this cause the blitter to just say no?
888 WriteLog("BLIT: Asked to use invalid bit combo (XADDINC) for A2...\n");
889 // add the contents of the increment register
890 // since there is no register for a2 we just add 1
891 //Let's do nothing, since it's not listed as a valid bit combo...
892 // a2_xadd = 1 << 16;
898 // Modify outer loop steps based on blitter command
906 a1_step_x = (REG(A1_FSTEP) & 0xFFFF),
907 a1_step_y = (REG(A1_FSTEP) >> 16);
910 a1_step_x |= ((REG(A1_STEP) & 0x0000FFFF) << 16),
911 a1_step_y |= ((REG(A1_STEP) & 0xFFFF0000));
914 a2_step_x = (REG(A2_STEP) & 0x0000FFFF) << 16,
915 a2_step_y = (REG(A2_STEP) & 0xFFFF0000);
917 outer_loop = n_lines;
922 a1_clip_x = REG(A1_CLIP) & 0x7FFF,
923 a1_clip_y = (REG(A1_CLIP) >> 16) & 0x7FFF;
925 // This phrase sizing is incorrect as well... !!! FIX !!!
926 // Err, this is pixel size... (and it's OK)
927 a2_psize = 1 << ((REG(A2_FLAGS) >> 3) & 0x07);
928 a1_psize = 1 << ((REG(A1_FLAGS) >> 3) & 0x07);
933 zadd = JaguarReadLong(0xF02274, BLITTER);
935 for(int v=0; v<4; v++)
936 z_i[v] = (int32)JaguarReadLong(0xF0228C + (v << 2), BLITTER);
938 if (GOURD || GOURZ || SRCSHADE)
941 gouraud_add = JaguarReadLong(0xF02270, BLITTER);
943 gd_c[0] = JaguarReadByte(0xF02268, BLITTER);
944 gd_i[0] = JaguarReadByte(0xF02269, BLITTER);
946 gd_i[0] |= JaguarReadWord(0xF02240, BLITTER);
948 gd_c[1] = JaguarReadByte(0xF0226A, BLITTER);
949 gd_i[1] = JaguarReadByte(0xF0226B, BLITTER);
951 gd_i[1] |= JaguarReadWord(0xF02242, BLITTER);
953 gd_c[2] = JaguarReadByte(0xF0226C, BLITTER);
954 gd_i[2] = JaguarReadByte(0xF0226D, BLITTER);
956 gd_i[2] |= JaguarReadWord(0xF02244, BLITTER);
958 gd_c[3] = JaguarReadByte(0xF0226E, BLITTER);
959 gd_i[3] = JaguarReadByte(0xF0226F, BLITTER);
961 gd_i[3] |= JaguarReadWord(0xF02246, BLITTER);
963 gd_ia = gouraud_add & 0xFFFFFF;
964 if (gd_ia & 0x800000)
965 gd_ia = 0xFF000000 | gd_ia;
967 gd_ca = (gouraud_add>>24) & 0xFF;
969 gd_ca = 0xFFFFFF00 | gd_ca;
972 // fix for zoop! and syndicate
973 /* if ((jaguar_mainRom_crc32==0x501be17c)||
974 (jaguar_mainRom_crc32==0x70895c51)||
975 (jaguar_mainRom_crc32==0x0f1f1497)||
976 (jaguar_mainRom_crc32==0xfc8f0dcd)
980 a1_step_x = (-n_pixels) * 65536;
983 a2_step_x = (-n_pixels) * 65536;;
986 // fix for wolfenstein 3d
987 if (jaguar_mainRom_crc32==0x3966698f)
991 if ((a1_step_x / 65536)==-28)
993 a1_step_x=-24*65536; // au lieu de -28
994 a2_step_x= 0*65536; // au lieu de -8
999 // fix for Tempest 2000
1000 if (jaguar_mainRom_crc32==0x32816d44)
1003 if ((n_lines!=1)&&((n_pixels==288)||(n_pixels==384)))
1005 WriteLog("Blit!\n");
1006 WriteLog(" cmd = 0x%.8x\n",cmd);
1007 WriteLog(" a1_base = %08X\n", a1_addr);
1008 WriteLog(" a1_pitch = %d\n", a1_pitch);
1009 WriteLog(" a1_psize = %d\n", a1_psize);
1010 WriteLog(" a1_width = %d\n", a1_width);
1011 WriteLog(" a1_xadd = %f (phrase=%d)\n", (float)a1_xadd / 65536.0, a1_phrase_mode);
1012 WriteLog(" a1_yadd = %f\n", (float)a1_yadd / 65536.0);
1013 WriteLog(" a1_xstep = %f\n", (float)a1_step_x / 65536.0);
1014 WriteLog(" a1_ystep = %f\n", (float)a1_step_y / 65536.0);
1015 WriteLog(" a1_x = %f\n", (float)a1_x / 65536.0);
1016 WriteLog(" a1_y = %f\n", (float)a1_y / 65536.0);
1017 WriteLog(" a1_zoffs = %i\n",a1_zoffs);
1019 WriteLog(" a2_base = %08X\n", a2_addr);
1020 WriteLog(" a2_pitch = %d\n", a2_pitch);
1021 WriteLog(" a2_psize = %d\n", a2_psize);
1022 WriteLog(" a2_width = %d\n", a2_width);
1023 WriteLog(" a2_xadd = %f (phrase=%d)\n", (float)a2_xadd / 65536.0, a2_phrase_mode);
1024 WriteLog(" a2_yadd = %f\n", (float)a2_yadd / 65536.0);
1025 WriteLog(" a2_xstep = %f\n", (float)a2_step_x / 65536.0);
1026 WriteLog(" a2_ystep = %f\n", (float)a2_step_y / 65536.0);
1027 WriteLog(" a2_x = %f\n", (float)a2_x / 65536.0);
1028 WriteLog(" a2_y = %f\n", (float)a2_y / 65536.0);
1029 WriteLog(" a2_mask_x= 0x%.4x\n",a2_mask_x);
1030 WriteLog(" a2_mask_y= 0x%.4x\n",a2_mask_y);
1031 WriteLog(" a2_zoffs = %i\n",a2_zoffs);
1033 WriteLog(" count = %d x %d\n", n_pixels, n_lines);
1035 WriteLog(" command = %08X\n", cmd);
1036 WriteLog(" dsten = %i\n",DSTEN);
1037 WriteLog(" srcen = %i\n",SRCEN);
1038 WriteLog(" patdsel = %i\n",PATDSEL);
1039 WriteLog(" color = 0x%.8x\n",REG(PATTERNDATA));
1040 WriteLog(" dcompen = %i\n",DCOMPEN);
1041 WriteLog(" bcompen = %i\n",BCOMPEN);
1042 WriteLog(" cmpdst = %i\n",CMPDST);
1043 WriteLog(" GOURZ = %i\n",GOURZ);
1044 WriteLog(" GOURD = %i\n",GOURD);
1045 WriteLog(" SRCSHADE = %i\n",SRCSHADE);
1046 WriteLog(" DSTDATA = 0x%.8x%.8x\n",REG(DSTDATA),REG(DSTDATA+4));
1053 WriteLog("Blit!\n");
1054 WriteLog(" cmd = 0x%.8x\n",cmd);
1055 WriteLog(" a1_base = %08X\n", a1_addr);
1056 WriteLog(" a1_pitch = %d\n", a1_pitch);
1057 WriteLog(" a1_psize = %d\n", a1_psize);
1058 WriteLog(" a1_width = %d\n", a1_width);
1059 WriteLog(" a1_xadd = %f (phrase=%d)\n", (float)a1_xadd / 65536.0, a1_phrase_mode);
1060 WriteLog(" a1_yadd = %f\n", (float)a1_yadd / 65536.0);
1061 WriteLog(" a1_xstep = %f\n", (float)a1_step_x / 65536.0);
1062 WriteLog(" a1_ystep = %f\n", (float)a1_step_y / 65536.0);
1063 WriteLog(" a1_x = %f\n", (float)a1_x / 65536.0);
1064 WriteLog(" a1_y = %f\n", (float)a1_y / 65536.0);
1065 WriteLog(" a1_zoffs = %i\n",a1_zoffs);
1067 WriteLog(" a2_base = %08X\n", a2_addr);
1068 WriteLog(" a2_pitch = %d\n", a2_pitch);
1069 WriteLog(" a2_psize = %d\n", a2_psize);
1070 WriteLog(" a2_width = %d\n", a2_width);
1071 WriteLog(" a2_xadd = %f (phrase=%d)\n", (float)a2_xadd / 65536.0, a2_phrase_mode);
1072 WriteLog(" a2_yadd = %f\n", (float)a2_yadd / 65536.0);
1073 WriteLog(" a2_xstep = %f\n", (float)a2_step_x / 65536.0);
1074 WriteLog(" a2_ystep = %f\n", (float)a2_step_y / 65536.0);
1075 WriteLog(" a2_x = %f\n", (float)a2_x / 65536.0);
1076 WriteLog(" a2_y = %f\n", (float)a2_y / 65536.0);
1077 WriteLog(" a2_mask_x= 0x%.4x\n",a2_mask_x);
1078 WriteLog(" a2_mask_y= 0x%.4x\n",a2_mask_y);
1079 WriteLog(" a2_zoffs = %i\n",a2_zoffs);
1081 WriteLog(" count = %d x %d\n", n_pixels, n_lines);
1083 WriteLog(" command = %08X\n", cmd);
1084 WriteLog(" dsten = %i\n",DSTEN);
1085 WriteLog(" srcen = %i\n",SRCEN);
1086 WriteLog(" patdsel = %i\n",PATDSEL);
1087 WriteLog(" color = 0x%.8x\n",REG(PATTERNDATA));
1088 WriteLog(" dcompen = %i\n",DCOMPEN);
1089 WriteLog(" bcompen = %i\n",BCOMPEN);
1090 WriteLog(" cmpdst = %i\n",CMPDST);
1091 WriteLog(" GOURZ = %i\n",GOURZ);
1092 WriteLog(" GOURD = %i\n",GOURD);
1093 WriteLog(" SRCSHADE= %i\n",SRCSHADE);
1097 //NOTE: Pitch is ignored!
1099 //This *might* be the altimeter blits (they are)...
1100 //On captured screen, x-pos for black (inner) is 259, for pink is 257
1101 //Black is short by 3, pink is short by 1...
1103 Blit! (00110000 <- 000BF010) count: 9 x 31, A1/2_FLAGS: 000042E2/00010020 [cmd: 00010200]
1104 CMD -> src: dst: misc: a1ctl: UPDA1 mode: ity: PATDSEL z-op: op: LFU_CLEAR ctrl:
1105 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1106 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
1107 A1 x/y: 262/124, A2 x/y: 128/0
1108 Blit! (00110000 <- 000BF010) count: 5 x 38, A1/2_FLAGS: 000042E2/00010020 [cmd: 00010200]
1109 CMD -> src: dst: misc: a1ctl: UPDA1 mode: ity: PATDSEL z-op: op: LFU_CLEAR ctrl:
1110 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1111 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
1112 A1 x/y: 264/117, A2 x/y: 407/0
1114 Blit! (00110000 <- 000BF010) count: 9 x 23, A1/2_FLAGS: 000042E2/00010020 [cmd: 00010200]
1115 CMD -> src: dst: misc: a1ctl: UPDA1 mode: ity: PATDSEL z-op: op: LFU_CLEAR ctrl:
1116 A1 step values: -10 (X), 1 (Y)
1117 A1 -> pitch: 4(2) phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1118 A2 -> pitch: 1(0) phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
1119 A1 x/y: 262/132, A2 x/y: 129/0
1120 Blit! (00110000 <- 000BF010) count: 5 x 27, A1/2_FLAGS: 000042E2/00010020 [cmd: 00010200]
1121 CMD -> src: dst: misc: a1ctl: UPDA1 mode: ity: PATDSEL z-op: op: LFU_CLEAR ctrl:
1122 A1 step values: -8 (X), 1 (Y)
1123 A1 -> pitch: 4(2) phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1124 A2 -> pitch: 1(0) phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPIX YADD0 XSIGNADD YSIGNADD
1125 A1 x/y: 264/128, A2 x/y: 336/0
1127 264v vCursor ends up here...
1131 262v vCursor ends up here...
1135 Fixed! Now for more:
1137 ; This looks like the ship icon in the upper left corner...
1139 Blit! (00110000 <- 0010B2A8) count: 11 x 12, A1/2_FLAGS: 000042E2/00000020 [cmd: 09800609]
1140 CMD -> src: SRCEN dst: DSTEN misc: a1ctl: UPDA1 UPDA2 mode: ity: z-op: op: LFU_REPLACE ctrl: DCOMPEN
1141 A1 step values: -12 (X), 1 (Y)
1142 A2 step values: 0 (X), 0 (Y) [mask (unused): 00000000 - FFFFFFFF/FFFFFFFF]
1143 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1144 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1145 A1 x/y: 20/24, A2 x/y: 5780/0
1149 More (not sure this is a blitter problem as much as it's a GPU problem):
1150 All but the "M" are trashed...
1151 This does *NOT* look like a blitter problem, as it's rendering properly...
1152 Actually, if you look at the A1 step values, there IS a discrepancy!
1156 Blit! (00110000 <- 0010B2A8) count: 12 x 12, A1/2_FLAGS: 000042E2/00000020 [cmd: 09800609]
1157 CMD -> src: SRCEN dst: DSTEN misc: a1ctl: UPDA1 UPDA2 mode: ity: z-op: op: LFU_REPLACE ctrl: DCOMPEN
1158 A1 step values: -14 (X), 1 (Y)
1159 A2 step values: -4 (X), 0 (Y) [mask (unused): 00000000 - FFFFFFFF/FFFFFFFF]
1160 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1161 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1162 A1 x/y: 134/144, A2 x/y: 2516/0
1167 Blit! (00110000 <- 0010B2A8) count: 12 x 12, A1/2_FLAGS: 000042E2/00000020 [cmd: 09800609]
1168 CMD -> src: SRCEN dst: DSTEN misc: a1ctl: UPDA1 UPDA2 mode: ity: z-op: op: LFU_REPLACE ctrl: DCOMPEN
1169 A1 step values: -13 (X), 1 (Y)
1170 A2 step values: -4 (X), 0 (Y) [mask (unused): 00000000 - FFFFFFFF/FFFFFFFF]
1171 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1172 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1173 A1 x/y: 147/144, A2 x/y: 2660/0
1177 Blit! (00110000 <- 0010B2A8) count: 12 x 12, A1/2_FLAGS: 000042E2/00000020 [cmd: 09800609]
1178 CMD -> src: SRCEN dst: DSTEN misc: a1ctl: UPDA1 UPDA2 mode: ity: z-op: op: LFU_REPLACE ctrl: DCOMPEN
1179 A1 step values: -12 (X), 1 (Y)
1180 A2 step values: 0 (X), 0 (Y) [mask (unused): 00000000 - FFFFFFFF/FFFFFFFF]
1181 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1182 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1183 A1 x/y: 160/144, A2 x/y: 3764/0
1187 Blit! (00110000 <- 0010B2A8) count: 12 x 12, A1/2_FLAGS: 000042E2/00000020 [cmd: 09800609]
1188 CMD -> src: SRCEN dst: DSTEN misc: a1ctl: UPDA1 UPDA2 mode: ity: z-op: op: LFU_REPLACE ctrl: DCOMPEN
1189 A1 step values: -15 (X), 1 (Y)
1190 A2 step values: -4 (X), 0 (Y) [mask (unused): 00000000 - FFFFFFFF/FFFFFFFF]
1191 A1 -> pitch: 4 phrases, depth: 16bpp, z-off: 3, width: 320 (21), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1192 A2 -> pitch: 1 phrases, depth: 16bpp, z-off: 0, width: 1 (00), addctl: XADDPHR YADD0 XSIGNADD YSIGNADD
1193 A1 x/y: 173/144, A2 x/y: 4052/0
1196 extern int blit_start_log;
1197 extern int op_start_log;
1200 char * ctrlStr[4] = { "XADDPHR\0", "XADDPIX\0", "XADD0\0", "XADDINC\0" };
1201 char * bppStr[8] = { "1bpp\0", "2bpp\0", "4bpp\0", "8bpp\0", "16bpp\0", "32bpp\0", "???\0", "!!!\0" };
1202 char * opStr[16] = { "LFU_CLEAR", "LFU_NSAND", "LFU_NSAD", "LFU_NOTS", "LFU_SAND", "LFU_NOTD", "LFU_N_SXORD", "LFU_NSORND",
1203 "LFU_SAD", "LFU_XOR", "LFU_D", "LFU_NSORD", "LFU_REPLACE", "LFU_SORND", "LFU_SORD", "LFU_ONE" };
1204 uint32 src = cmd & 0x07, dst = (cmd >> 3) & 0x07, misc = (cmd >> 6) & 0x03,
1205 a1ctl = (cmd >> 8) & 0x07, mode = (cmd >> 11) & 0x07, ity = (cmd >> 14) & 0x0F,
1206 zop = (cmd >> 18) & 0x07, op = (cmd >> 21) & 0x0F, ctrl = (cmd >> 25) & 0x3F;
1207 UINT32 a1f = REG(A1_FLAGS), a2f = REG(A2_FLAGS);
1208 uint32 p1 = a1f & 0x07, p2 = a2f & 0x07,
1209 d1 = (a1f >> 3) & 0x07, d2 = (a2f >> 3) & 0x07,
1210 zo1 = (a1f >> 6) & 0x07, zo2 = (a2f >> 6) & 0x07,
1211 w1 = (a1f >> 9) & 0x3F, w2 = (a2f >> 9) & 0x3F,
1212 ac1 = (a1f >> 16) & 0x1F, ac2 = (a2f >> 16) & 0x1F;
1213 UINT32 iw1 = ((0x04 | (w1 & 0x03)) << ((w1 & 0x3C) >> 2)) >> 2;
1214 UINT32 iw2 = ((0x04 | (w2 & 0x03)) << ((w2 & 0x3C) >> 2)) >> 2;
1215 WriteLog("Blit! (%08X %s %08X) count: %d x %d, A1/2_FLAGS: %08X/%08X [cmd: %08X]\n", a1_addr, (mode&0x01 ? "->" : "<-"), a2_addr, n_pixels, n_lines, a1f, a2f, cmd);
1216 // WriteLog(" CMD -> src: %d, dst: %d, misc: %d, a1ctl: %d, mode: %d, ity: %1X, z-op: %d, op: %1X, ctrl: %02X\n", src, dst, misc, a1ctl, mode, ity, zop, op, ctrl);
1218 WriteLog(" CMD -> src: %s%s%s ", (cmd & 0x0001 ? "SRCEN " : ""), (cmd & 0x0002 ? "SRCENZ " : ""), (cmd & 0x0004 ? "SRCENX" : ""));
1219 WriteLog("dst: %s%s%s ", (cmd & 0x0008 ? "DSTEN " : ""), (cmd & 0x0010 ? "DSTENZ " : ""), (cmd & 0x0020 ? "DSTWRZ" : ""));
1220 WriteLog("misc: %s%s ", (cmd & 0x0040 ? "CLIP_A1 " : ""), (cmd & 0x0080 ? "???" : ""));
1221 WriteLog("a1ctl: %s%s%s ", (cmd & 0x0100 ? "UPDA1F " : ""), (cmd & 0x0200 ? "UPDA1 " : ""), (cmd & 0x0400 ? "UPDA2" : ""));
1222 WriteLog("mode: %s%s%s ", (cmd & 0x0800 ? "DSTA2 " : ""), (cmd & 0x1000 ? "GOURD " : ""), (cmd & 0x2000 ? "ZBUFF" : ""));
1223 WriteLog("ity: %s%s%s ", (cmd & 0x4000 ? "TOPBEN " : ""), (cmd & 0x8000 ? "TOPNEN " : ""), (cmd & 0x00010000 ? "PATDSEL" : ""));
1224 WriteLog("z-op: %s%s%s ", (cmd & 0x00040000 ? "ZMODELT " : ""), (cmd & 0x00080000 ? "ZMODEEQ " : ""), (cmd & 0x00100000 ? "ZMODEGT" : ""));
1225 WriteLog("op: %s ", opStr[(cmd >> 21) & 0x0F]);
1226 WriteLog("ctrl: %s%s%s%s%s%s\n", (cmd & 0x02000000 ? "CMPDST " : ""), (cmd & 0x04000000 ? "BCOMPEN " : ""), (cmd & 0x08000000 ? "DCOMPEN " : ""), (cmd & 0x10000000 ? "BKGWREN " : ""), (cmd & 0x20000000 ? "BUSHI " : ""), (cmd & 0x40000000 ? "SRCSHADE" : ""));
1229 WriteLog(" A1 step values: %d (X), %d (Y)\n", a1_step_x >> 16, a1_step_y >> 16);
1232 WriteLog(" A2 step values: %d (X), %d (Y) [mask (%sused): %08X - %08X/%08X]\n", a2_step_x >> 16, a2_step_y >> 16, (a2f & 0x8000 ? "" : "un"), REG(A2_MASK), a2_mask_x, a2_mask_y);
1234 WriteLog(" A1 -> pitch: %d phrases, depth: %s, z-off: %d, width: %d (%02X), addctl: %s %s %s %s\n", 1 << p1, bppStr[d1], zo1, iw1, w1, ctrlStr[ac1&0x03], (ac1&0x04 ? "YADD1" : "YADD0"), (ac1&0x08 ? "XSIGNSUB" : "XSIGNADD"), (ac1&0x10 ? "YSIGNSUB" : "YSIGNADD"));
1235 WriteLog(" A2 -> pitch: %d phrases, depth: %s, z-off: %d, width: %d (%02X), addctl: %s %s %s %s\n", 1 << p2, bppStr[d2], zo2, iw2, w2, ctrlStr[ac2&0x03], (ac2&0x04 ? "YADD1" : "YADD0"), (ac2&0x08 ? "XSIGNSUB" : "XSIGNADD"), (ac2&0x10 ? "YSIGNSUB" : "YSIGNADD"));
1236 WriteLog(" A1 x/y: %d/%d, A2 x/y: %d/%d\n", a1_x >> 16, a1_y >> 16, a2_x >> 16, a2_y >> 16);
1237 // blit_start_log = 0;
1238 // op_start_log = 1;
1241 blitter_working = 1;
1242 //#ifndef USE_GENERIC_BLITTER
1243 // if (!blitter_execute_cached_code(blitter_in_cache(cmd)))
1245 blitter_generic(cmd);
1247 /*if (blit_start_log)
1249 if (a1_addr == 0xF03000 && a2_addr == 0x004D58)
1251 WriteLog("\nBytes at 004D58:\n");
1252 for(int i=0x004D58; i<0x004D58+(10*127*4); i++)
1253 WriteLog("%02X ", JaguarReadByte(i));
1254 WriteLog("\nBytes at F03000:\n");
1255 for(int i=0xF03000; i<0xF03000+(6*127*4); i++)
1256 WriteLog("%02X ", JaguarReadByte(i));
1261 blitter_working = 0;
1264 void blitter_init(void)
1269 void blitter_reset(void)
1271 memset(blitter_ram, 0x00, 0xA0);
1274 void blitter_done(void)
1276 WriteLog("BLIT: Done.\n");
1279 uint8 BlitterReadByte(uint32 offset, uint32 who/*=UNKNOWN*/)
1284 if (offset == (0x38 + 3))
1285 return 0x01; // always idle
1287 return blitter_ram[offset];
1290 uint16 BlitterReadWord(uint32 offset, uint32 who/*=UNKNOWN*/)
1292 return ((uint16)BlitterReadByte(offset, who) << 8) | (uint16)BlitterReadByte(offset+1, who);
1295 uint32 BlitterReadLong(uint32 offset, uint32 who/*=UNKNOWN*/)
1297 return (BlitterReadWord(offset, who) << 16) | BlitterReadWord(offset+2, who);
1300 void BlitterWriteByte(uint32 offset, uint8 data, uint32 who/*=UNKNOWN*/)
1302 /*if (offset & 0xFF == 0x7B)
1303 WriteLog("--> Wrote to B_STOP: value -> %02X\n", data);*/
1306 // if ((offset >= 0x7C) && (offset <= 0x9B))
1307 if ((offset >= 0x7C) && (offset <= 0x8B))
1312 case 0x7D: blitter_ram[0x69] = data; break;
1313 case 0x7E: blitter_ram[0x40] = data; break;
1314 case 0x7F: blitter_ram[0x41] = data; break;
1317 case 0x81: blitter_ram[0x6B] = data; break;
1318 case 0x82: blitter_ram[0x42] = data; break;
1319 case 0x83: blitter_ram[0x43] = data; break;
1322 case 0x85: blitter_ram[0x6D] = data; break;
1323 case 0x86: blitter_ram[0x44] = data; break;
1324 case 0x87: blitter_ram[0x45] = data; break;
1327 case 0x89: blitter_ram[0x6F] = data; break;
1329 // case 0x9A: blitter_ram[0x46] = data; break;
1330 // case 0x9B: blitter_ram[0x47] = data; break;
1331 case 0x8A: blitter_ram[0x46] = data; break;
1332 case 0x8B: blitter_ram[0x47] = data; break;
1336 blitter_ram[offset] = data;
1339 void BlitterWriteWord(uint32 offset, uint16 data, uint32 who/*=UNKNOWN*/)
1341 BlitterWriteByte(offset+0, (data>>8) & 0xFF, who);
1342 BlitterWriteByte(offset+1, data & 0xFF, who);
1344 if ((offset & 0xFF) == 0x3A)
1345 // I.e., the second write of 32-bit value--not convinced this is the best way to do this!
1346 // But then again, according to the Jaguar docs, this is correct...!
1347 blitter_blit(GET32(blitter_ram, 0x38));
1348 // Testing purposes only!
1349 //This does the clipping correctly, but not the Gouraud shading...
1350 // blitter2_exec(GET32(blitter_ram, 0x38));
1354 void BlitterWriteLong(uint32 offset, uint32 data, uint32 who/*=UNKNOWN*/)
1356 BlitterWriteWord(offset, data >> 16, who);
1357 BlitterWriteWord(offset+2, data & 0xFFFF, who);