#include #include "vdp2.h" #include "vdp1.h" #include "scu.h" #include "smpc.h" #include "sh2.h" #include "common/copy.hpp" #include "common/vdp2_func.hpp" #include "common/intback.hpp" #include "input.hpp" #include "gen/maps.hpp" #include "gen/sprites.hpp" #include "map_objects.hpp" #include "coordinates.hpp" #include "render_map.hpp" #include "actor.hpp" constexpr inline uint16_t rgb15(int32_t r, int32_t g, int32_t b) { return ((b & 31) << 10) | ((g & 31) << 5) | ((r & 31) << 0); } void palette_data() { vdp2.cram.u16[3] = rgb15( 0, 0, 0); vdp2.cram.u16[2] = rgb15(10, 10, 10); vdp2.cram.u16[1] = rgb15(21, 21, 21); vdp2.cram.u16[0] = rgb15(31, 31, 31); } static inline void _2bpp_4bpp_vram_copy(uint32_t * vram, const start_size_t& buf) { for (uint32_t ix = 0; ix < buf.size / 4; ix += 1) { const uint32_t pixels = reinterpret_cast(buf.start)[ix]; const uint32_t px0 = pixels >> 16 & 0xffff; const uint32_t px1 = pixels >> 0 & 0xffff; #define lshift(n) ((7 - n) * 2) #define rshift(n) ((7 - n) * 4) #define px(p, n) (((p >> lshift(n)) & 0b11) << rshift(n)) #define p0(n) (px(px0, n)) #define p1(n) (px(px1, n)) vram[ix * 2 + 0] = p0(7) | p0(6) | p0(5) | p0(4) | p0(3) | p0(2) | p0(1) | p0(0); vram[ix * 2 + 1] = p1(7) | p1(6) | p1(5) | p1(4) | p1(3) | p1(2) | p1(1) | p1(0); #undef p1 #undef p0 #undef px #undef lshift #undef rshift } } uint32_t character_pattern_table(const start_size_t& buf, const uint32_t top) { // round to nearest multiple of 32 const uint32_t table_size = ((buf.size * 2) + 0x20 - 1) & (-0x20); const uint32_t base_address = top - table_size; uint32_t * vram = &vdp1.vram.u32[(base_address / 4)]; _2bpp_4bpp_vram_copy(vram, buf); return base_address; } uint32_t cell_data(const start_size_t& buf, const uint32_t top) { // round to nearest multiple of 32 const uint32_t table_size = ((buf.size * 2) + 0x20 - 1) & (-0x20); const uint32_t base_address = top - table_size; // in bytes uint32_t * vram = &vdp2.vram.u32[(base_address / 4)]; _2bpp_4bpp_vram_copy(vram, buf); return base_address; } struct draw_t { struct { uint16_t tilesets[tileset_t::count]; // div 32 uint16_t spritesheets[spritesheet_t::count]; // div 128 } base_pattern; }; struct state_t { enum map_t::map map; draw_t draw; actor_t player; }; static state_t state = { map_t::pallet_town, 0 }; uint32_t load_tileset(uint32_t top, enum tileset_t::tileset tileset) { uint32_t base_address = top = cell_data(tilesets[tileset].tileset, top); state.draw.base_pattern.tilesets[tileset] = base_address / 32; return top; } uint32_t load_sprite(uint32_t top, enum spritesheet_t::spritesheet spritesheet) { const spritesheet_t& s = spritesheets[spritesheet]; uint32_t base_address = top = character_pattern_table(s.spritesheet, top); state.draw.base_pattern.spritesheets[spritesheet] = base_address / 128; return top; } void load_vram() { vdp2.reg.CYCA0 = 0xeeee'eeee; vdp2.reg.CYCA1 = 0xeeee'eeee; vdp2.reg.CYCB0 = 0xeeee'eeee; vdp2.reg.CYCB1 = 0xeeee'eeee; uint32_t vdp2_top = (sizeof (union vdp2_vram)); for (uint32_t i = 0; i < tileset_t::count; i++) vdp2_top = load_tileset(vdp2_top, static_cast(i)); vdp2.reg.CYCA0 = 0x0fff'ffff; vdp2.reg.CYCA1 = 0x0fff'ffff; vdp2.reg.CYCB0 = 0x4fff'ffff; vdp2.reg.CYCB1 = 0x4fff'ffff; uint32_t vdp1_top = (sizeof (union vdp1_vram)); for (uint32_t i = 0; i < spritesheet_t::count; i++) vdp1_top = load_sprite(vdp1_top, static_cast(i)); } void render_sprite(const uint32_t ix, const uint32_t sprite_id, const screen_t& screen, const offset_t& offset) { constexpr uint32_t color_address = 0; const uint32_t character_address = (state.draw.base_pattern.spritesheets[sprite_id] * 128) / 8; vdp1.vram.cmd[ix].CTRL = CTRL__JP__JUMP_NEXT | CTRL__COMM__NORMAL_SPRITE; vdp1.vram.cmd[ix].LINK = 0; // The "end code" is 0xf, which is being used in the mai sprite palette. If // both transparency and end codes are enabled, it seems there are only 14 // usable colors in the 4-bit color mode. vdp1.vram.cmd[ix].PMOD = PMOD__ECD | PMOD__COLOR_MODE__COLOR_BANK_16; // It appears Kronos does not correctly calculate the color address in the // VDP1 debugger. Kronos will report FFFC when the actual color table address // in this example is 7FFE0. vdp1.vram.cmd[ix].COLR = color_address; // non-palettized (rgb15) color data vdp1.vram.cmd[ix].SRCA = character_address; vdp1.vram.cmd[ix].SIZE = SIZE__X(16) | SIZE__Y(16); vdp1.vram.cmd[ix].XA = (cell_offset::x * 8) + screen.x * 16 - offset.x; vdp1.vram.cmd[ix].YA = (cell_offset::y * 8) + screen.y * 16 - 4 - offset.y; } void render_sprites(const offset_t& offset) { uint32_t ix = 2; render_sprite(ix, spritesheet_t::red, {4, 4}, {0, 0}); ix++; const object_t& obj = map_objects[state.map]; for (uint32_t i = 0; i < obj.object_length; i++) { const object_event_t& event = obj.object_events[i]; const world_t world = { event.position.x, event.position.y }; render_sprite(ix, event.sprite_id, world.to_screen(state.player.world), offset); ix++; } constexpr uint16_t top_x = 80 - 1; constexpr uint16_t top_y = 48 - 1; constexpr uint16_t bot_x = 239 + 1; constexpr uint16_t bot_y = 191 + 1; vdp1.vram.cmd[ix].CTRL = CTRL__JP__JUMP_NEXT | CTRL__COMM__POLYLINE; vdp1.vram.cmd[ix].LINK = 0; // "Set [ECD] to '1' for polygons, polylines, and lines" // "Be sure to set [SPD] to '1' for polygons, polylines, and lines" // // The "user clip mode" bit is not set in PMOD here, so setting "user clip // coordinates" has no effect on this draw command. However, "system clip // coordinates" and "local coordinates" are always applied, and must be set to // reasonable values. vdp1.vram.cmd[ix].PMOD = PMOD__ECD | PMOD__SPD; vdp1.vram.cmd[ix].COLR = COLR__RGB | rgb15(255, 0, 255); vdp1.vram.cmd[ix].XA = top_x; vdp1.vram.cmd[ix].YA = top_y; vdp1.vram.cmd[ix].XB = bot_x; vdp1.vram.cmd[ix].YB = top_y; vdp1.vram.cmd[ix].XC = bot_x; vdp1.vram.cmd[ix].YC = bot_y; vdp1.vram.cmd[ix].XD = top_x; vdp1.vram.cmd[ix].YD = bot_y; ix++; vdp1.vram.cmd[ix].CTRL = CTRL__END; } void render_map() { const map_t& map = maps[state.map]; const uint32_t base_pattern = state.draw.base_pattern.tilesets[map.tileset]; vdp2.reg.PNCN0 = PNCN0__N0PNB__1WORD | PNCN0__N0CNSM | PNCN0__N0SCN((base_pattern >> 10) & 0x1f); for (int32_t y = (0 - 1); y < (9 + 1); y++) { for (int32_t x = (0 - 1); x < (10 + 1); x++) { render_screen(base_pattern, map, state.player.world, {x, y} ); } } vdp2.reg.BGON = BGON__N0ON | BGON__N0TPON; } void update() { state.player.tick(); if (event::cursor_left()) { state.player.move(actor_t::left); //state.player.world.x--; } else if (event::cursor_right()) { state.player.move(actor_t::right); //state.player.world.x++; } else if (event::cursor_up()) { state.player.move(actor_t::up); //state.player.world.y--; } else if (event::cursor_down()) { state.player.move(actor_t::down); //state.player.world.y++; } } void render() { const offset_t offset = state.player.offset(); render_sprites(offset); vdp2.reg.SCXIN0 = offset.x; vdp2.reg.SCYIN0 = offset.y; render_map(); } extern "C" void v_blank_in_int(void) __attribute__ ((interrupt_handler)); void v_blank_in_int() { scu.reg.IST &= ~(IST__V_BLANK_IN); scu.reg.IMS = ~(IMS__SMPC | IMS__V_BLANK_IN); sh2.reg.FRC.H = 0; sh2.reg.FRC.L = 0; sh2.reg.FTCSR = 0; // clear flags update(); render(); // wait at least 300us, as specified in the SMPC manual. // It appears reading FRC.H is mandatory and *must* occur before FRC.L on real // hardware. while ((sh2.reg.FTCSR & FTCSR__OVF) == 0 && sh2.reg.FRC.H == 0 && sh2.reg.FRC.L < 63); // on real hardware, SF contains uninitialized garbage bits other than the // lsb. while ((smpc.reg.SF & 1) != 0); smpc.reg.SF = 0; smpc.reg.IREG[0].val = INTBACK__IREG0__STATUS_DISABLE; smpc.reg.IREG[1].val = ( INTBACK__IREG1__PERIPHERAL_DATA_ENABLE | INTBACK__IREG1__PORT2_15BYTE | INTBACK__IREG1__PORT1_15BYTE ); smpc.reg.IREG[2].val = INTBACK__IREG2__MAGIC; smpc.reg.COMREG = COMREG__INTBACK; } extern "C" void smpc_int(void) __attribute__ ((interrupt_handler)); void smpc_int(void) { scu.reg.IST &= ~(IST__SMPC); scu.reg.IMS = ~(IMS__SMPC | IMS__V_BLANK_IN); intback::fsm(digital_callback, nullptr); } void init_vdp1() { /* TVM settings must be performed from the second H-blank IN interrupt after the V-blank IN interrupt to the H-blank IN interrupt immediately after the V-blank OUT interrupt. */ // "normal" display resolution, 16 bits per pixel, 512x256 framebuffer vdp1.reg.TVMR = TVMR__TVM__NORMAL; // swap framebuffers every 1 cycle; non-interlace vdp1.reg.FBCR = 0; // during a framebuffer erase cycle, write the color "black" to each pixel constexpr uint16_t black = 0x0000; vdp1.reg.EWDR = black; // the EWLR/EWRR macros use somewhat nontrivial math for the X coordinates // erase upper-left coordinate vdp1.reg.EWLR = EWLR__16BPP_X1(0) | EWLR__Y1(0); // erase lower-right coordinate vdp1.reg.EWRR = EWRR__16BPP_X3(319) | EWRR__Y3(239); vdp1.vram.cmd[0].CTRL = CTRL__JP__JUMP_NEXT | CTRL__COMM__SYSTEM_CLIP_COORDINATES; vdp1.vram.cmd[0].LINK = 0; vdp1.vram.cmd[0].XC = 319; vdp1.vram.cmd[0].YC = 239; vdp1.vram.cmd[1].CTRL = CTRL__JP__JUMP_NEXT | CTRL__COMM__LOCAL_COORDINATE; vdp1.vram.cmd[1].LINK = 0; vdp1.vram.cmd[1].XA = 0; vdp1.vram.cmd[1].YA = 0; vdp1.vram.cmd[2].CTRL = CTRL__END; // start drawing (execute the command list) on every frame vdp1.reg.PTMR = PTMR__PTM__FRAME_CHANGE; } void init_vdp2() { vdp2.reg.PRISA = PRISA__S0PRIN(7); // Sprite register 0 PRIority Number vdp2.reg.PRINA = PRINA__N0PRIN(5) | PRINA__N1PRIN(6); // DISP: Please make sure to change this bit from 0 to 1 during V blank. vdp2.reg.TVMD = ( TVMD__DISP | TVMD__LSMD__NON_INTERLACE | TVMD__VRESO__240 | TVMD__HRESO__NORMAL_320); vdp2.reg.EXTEN = 0; /* set the color mode to 5bits per channel, 1024 colors */ vdp2.reg.RAMCTL = RAMCTL__CRKTE | RAMCTL__CRMD__RGB_5BIT_1024;// | RAMCTL__VRAMD | RAMCTL__VRBMD; /* disable display of NBG0 */ vdp2.reg.BGON = 0; /* set character format for NBG0 to palettized 16 color set enable "cell format" for NBG0 set character size for NBG0 to 1x1 cell */ vdp2.reg.CHCTLA = CHCTLA__N0CHCN__16_COLOR | CHCTLA__N0BMEN__CELL_FORMAT | CHCTLA__N0CHSZ__1x1_CELL | CHCTLA__N1CHCN__16_COLOR | CHCTLA__N1BMEN__CELL_FORMAT | CHCTLA__N1CHSZ__1x1_CELL; /* plane size */ vdp2.reg.PLSZ = PLSZ__N0PLSZ__1x1 | PLSZ__N1PLSZ__1x1; /* map plane offset 1-word: value of bit 6-0 * 0x2000 2-word: value of bit 5-0 * 0x4000 */ constexpr int nbg0_plane_a = 0; constexpr int nbg1_plane_a = 1; //constexpr int plane_a_offset = plane_a * 0x2000; //constexpr int page_size = 64 * 64 * 2; // N0PNB__1WORD (16-bit) //constexpr int plane_size = page_size * 1; // bits 8~6 vdp2.reg.MPOFN = MPOFN__N0MP(0) | MPOFN__N1MP(0); vdp2.reg.MPABN0 = MPABN0__N0MPB(0) | MPABN0__N0MPA(nbg0_plane_a); // bits 5~0 vdp2.reg.MPCDN0 = MPCDN0__N0MPD(0) | MPCDN0__N0MPC(0); // bits 5~0 vdp2.reg.MPABN1 = MPABN1__N1MPB(0) | MPABN1__N1MPA(nbg1_plane_a); // bits 5~0 vdp2.reg.MPCDN1 = MPCDN1__N1MPD(0) | MPCDN1__N1MPC(0); // bits 5~0 auto& base_pattern = state.draw.base_pattern.tilesets[tileset_t::overworld]; vdp2.reg.PNCN0 = PNCN0__N0PNB__1WORD | PNCN0__N0CNSM | PNCN0__N0SCN((base_pattern >> 10) & 0x1f); vdp2.reg.PNCN1 = PNCN1__N1PNB__1WORD | PNCN1__N1CNSM | PNCN1__N1SCN((base_pattern >> 10) & 0x1f); vdp2.vram.u16[0x2000 / 2 + 0] = (base_pattern & 0xfff) + 1; vdp2.vram.u16[0x2000 / 2 + 1] = (base_pattern & 0xfff) + 2; palette_data(); } void main() { state.map = map_t::pallet_town; load_vram(); v_blank_in(); init_vdp1(); init_vdp2(); // free-running timer sh2.reg.TCR = TCR__CKS__INTERNAL_DIV128; sh2.reg.FTCSR = 0; // initialize smpc smpc.reg.DDR1 = 0; // INPUT smpc.reg.DDR2 = 0; // INPUT smpc.reg.IOSEL = 0; // SMPC control smpc.reg.EXLE = 0; // sh2_vec[SCU_VEC__SMPC] = (u32)(&smpc_int); sh2_vec[SCU_VEC__V_BLANK_IN] = (u32)(&v_blank_in_int); scu.reg.IST = 0; scu.reg.IMS = ~(IMS__SMPC | IMS__V_BLANK_IN); }