// // 9918Base.hpp // Clock Signal // // Created by Thomas Harte on 14/12/2017. // Copyright 2017 Thomas Harte. All rights reserved. // #ifndef TMS9918Base_hpp #define TMS9918Base_hpp #include "ClockConverter.hpp" #include "../../../ClockReceiver/ClockReceiver.hpp" #include "../../../Numeric/BitReverse.hpp" #include "../../../Outputs/CRT/CRT.hpp" #include "AccessEnums.hpp" #include "LineBuffer.hpp" #include "PersonalityTraits.hpp" #include "Storage.hpp" #include "YamahaCommands.hpp" #include #include #include #include #include #include namespace TI::TMS { constexpr uint8_t StatusInterrupt = 0x80; constexpr uint8_t StatusSpriteOverflow = 0x40; constexpr int StatusSpriteCollisionShift = 5; constexpr uint8_t StatusSpriteCollision = 0x20; template struct Base: public Storage { Base(); static constexpr int output_lag = 11; // i.e. pixel output will occur 11 cycles // after corresponding data read. static constexpr uint32_t palette_pack(uint8_t r, uint8_t g, uint8_t b) { #if TARGET_RT_BIG_ENDIAN return uint32_t((r << 24) | (g << 16) | (b << 8)); #else return uint32_t((b << 16) | (g << 8) | r); #endif } // The default TMS palette. static constexpr std::array default_palette { palette_pack(0, 0, 0), palette_pack(0, 0, 0), palette_pack(33, 200, 66), palette_pack(94, 220, 120), palette_pack(84, 85, 237), palette_pack(125, 118, 252), palette_pack(212, 82, 77), palette_pack(66, 235, 245), palette_pack(252, 85, 84), palette_pack(255, 121, 120), palette_pack(212, 193, 84), palette_pack(230, 206, 128), palette_pack(33, 176, 59), palette_pack(201, 91, 186), palette_pack(204, 204, 204), palette_pack(255, 255, 255) }; const std::array &palette() { if constexpr (is_yamaha_vdp(personality)) { return Storage::solid_background_ ? Storage::palette_ : Storage::background_palette_; } return default_palette; } Outputs::CRT::CRT crt_; TVStandard tv_standard_ = TVStandard::NTSC; using AddressT = typename Storage::AddressT; /// Mutates @c target such that @c source replaces the @c length bits that currently start /// at bit @c shift . Subsequently ensures @c target is constrained by the /// applicable @c memory_mask. template void install_field(AddressT &target, uint8_t source) { static_assert(length > 0 && length <= 8); constexpr auto source_mask = (1 << length) - 1; constexpr auto mask = AddressT(~(source_mask << shift)); target = ( (target & mask) | AddressT((source & source_mask) << shift) ) & memory_mask(personality); } // Personality-specific metrics and converters. ClockConverter clock_converter_; // This VDP's DRAM. std::array ram_; // State of the DRAM/CRAM-access mechanism. AddressT ram_pointer_ = 0; uint8_t read_ahead_buffer_ = 0; MemoryAccess queued_access_ = MemoryAccess::None; int minimum_access_column_ = 0; // The main status register. uint8_t status_ = 0; // Current state of programmer input. bool write_phase_ = false; // Determines whether the VDP is expecting the low or high byte of a write. uint8_t low_write_ = 0; // Buffers the low byte of a write. // Various programmable flags. bool mode1_enable_ = false; bool mode2_enable_ = false; bool mode3_enable_ = false; bool blank_display_ = false; bool sprites_16x16_ = false; bool sprites_magnified_ = false; bool generate_interrupts_ = false; uint8_t sprite_height_ = 8; // Programmer-specified addresses. // // The TMS and descendants combine various parts of the address with AND operations, // e.g. the fourth byte in the pattern name table will be at `pattern_name_address_ & 4`; // ordinarily the difference between that and plain substitution is invisible because // the programmer mostly can't set low-enough-order bits. That's not universally true // though, so this implementation uses AND throughout. // // ... therefore, all programmer-specified addresses are seeded as all '1's. As and when // actual addresses are specified, the relevant bits will be substituted in. // // Cf. install_field. AddressT pattern_name_address_ = memory_mask(personality); // Address of the tile map. AddressT colour_table_address_ = memory_mask(personality); // Address of the colour map (if applicable). AddressT pattern_generator_table_address_ = memory_mask(personality); // Address of the tile contents. AddressT sprite_attribute_table_address_ = memory_mask(personality); // Address of the sprite list. AddressT sprite_generator_table_address_ = memory_mask(personality); // Address of the sprite contents. // Default colours. uint8_t text_colour_ = 0; uint8_t background_colour_ = 0; // Internal mechanisms for position tracking. int latched_column_ = 0; // A struct to contain timing information that is a function of the current mode. struct { /* Vertical layout: Lines 0 to [pixel_lines]: standard data fetch and drawing will occur. ... to [first_vsync_line]: refresh fetches will occur and border will be output. .. to [2.5 or 3 lines later]: vertical sync is output. ... to [total lines - 1]: refresh fetches will occur and border will be output. ... for one line: standard data fetch will occur, without drawing. */ int total_lines = 262; int pixel_lines = 192; int first_vsync_line = 227; // Maximum number of sprite slots to populate; // if sprites beyond this number should be visible // then the appropriate status information will be set. int maximum_visible_sprites = 4; // Set the position, in cycles, of the two interrupts, // within a line. // // TODO: redetermine where this number came from. struct { int column = 313; int row = 192; } end_of_frame_interrupt_position; int line_interrupt_position = -1; // Enables or disabled the recognition of the sprite // list terminator, and sets the terminator value. bool allow_sprite_terminator = true; uint8_t sprite_terminator(ScreenMode mode) { switch(mode) { default: return 0xd0; case ScreenMode::YamahaGraphics3: case ScreenMode::YamahaGraphics4: case ScreenMode::YamahaGraphics5: case ScreenMode::YamahaGraphics6: case ScreenMode::YamahaGraphics7: return 0xd8; } } } mode_timing_; uint8_t line_interrupt_target_ = 0xff; uint8_t line_interrupt_counter_ = 0; bool enable_line_interrupts_ = false; bool line_interrupt_pending_ = false; bool vertical_active_ = false; ScreenMode screen_mode_, underlying_mode_; using LineBufferArray = std::array; LineBufferArray line_buffers_; LineBufferArray::iterator fetch_line_buffer_; LineBufferArray::iterator draw_line_buffer_; void advance(LineBufferArray::iterator &iterator) { ++iterator; if(iterator == line_buffers_.end()) { iterator = line_buffers_.begin(); } } using SpriteBufferArray = std::array; SpriteBufferArray sprite_buffers_; SpriteBufferArray::iterator fetch_sprite_buffer_; SpriteBuffer *fetched_sprites_ = nullptr; void advance(SpriteBufferArray::iterator &iterator) { ++iterator; if(iterator == sprite_buffers_.end()) { iterator = sprite_buffers_.begin(); } } void regress(SpriteBufferArray::iterator &iterator) { if(iterator == sprite_buffers_.begin()) { iterator = sprite_buffers_.end(); } --iterator; } AddressT tile_offset_ = 0; uint8_t name_[4]{}; void posit_sprite(int sprite_number, int sprite_y, uint8_t screen_row); // There is a delay between reading into the line buffer and outputting from there to the screen. That delay // is observeable because reading time affects availability of memory accesses and therefore time in which // to update sprites and tiles, but writing time affects when the palette is used and when the collision flag // may end up being set. So the two processes are slightly decoupled. The end of reading one line may overlap // with the beginning of writing the next, hence the two separate line buffers. LineBufferPointer output_pointer_, fetch_pointer_; int fetch_line() const; bool is_horizontal_blank() const; VerticalState vertical_state() const; int masked_address(int address) const; void write_vram(uint8_t); void write_register(uint8_t); void write_palette(uint8_t); void write_register_indirect(uint8_t); uint8_t read_vram(); uint8_t read_register(); void commit_register(int reg, uint8_t value); template ScreenMode current_screen_mode() const { if(check_blank && blank_display_) { return ScreenMode::Blank; } if constexpr (is_sega_vdp(personality)) { if(Storage::mode4_enable_) { return ScreenMode::SMSMode4; } } if constexpr (is_yamaha_vdp(personality)) { switch(Storage::mode_) { case 0b00001: return ScreenMode::Text; case 0b01001: return ScreenMode::YamahaText80; case 0b00010: return ScreenMode::MultiColour; case 0b00000: return ScreenMode::YamahaGraphics1; case 0b00100: return ScreenMode::YamahaGraphics2; case 0b01000: return ScreenMode::YamahaGraphics3; case 0b01100: return ScreenMode::YamahaGraphics4; case 0b10000: return ScreenMode::YamahaGraphics5; case 0b10100: return ScreenMode::YamahaGraphics6; case 0b11100: return ScreenMode::YamahaGraphics7; } } if(!mode1_enable_ && !mode2_enable_ && !mode3_enable_) { return ScreenMode::ColouredText; } if(mode1_enable_ && !mode2_enable_ && !mode3_enable_) { return ScreenMode::Text; } if(!mode1_enable_ && mode2_enable_ && !mode3_enable_) { return ScreenMode::Graphics; } if(!mode1_enable_ && !mode2_enable_ && mode3_enable_) { return ScreenMode::MultiColour; } // TODO: undocumented TMS modes. return ScreenMode::Blank; } static AddressT rotate(AddressT address) { return AddressT((address >> 1) | (address << 16)) & memory_mask(personality); } AddressT command_address(Vector location, bool expansion) const { if constexpr (is_yamaha_vdp(personality)) { switch(this->underlying_mode_) { default: case ScreenMode::YamahaGraphics4: // 256 pixels @ 4bpp return AddressT( ((location.v[0] >> 1) & 127) + (location.v[1] << 7) ); case ScreenMode::YamahaGraphics5: // 512 pixels @ 2bpp return AddressT( ((location.v[0] >> 2) & 127) + (location.v[1] << 7) ); case ScreenMode::YamahaGraphics6: { // 512 pixels @ 4bpp const auto linear_address = AddressT( ((location.v[0] >> 1) & 255) + (location.v[1] << 8) ); return expansion ? linear_address : rotate(linear_address); } case ScreenMode::YamahaGraphics7: { // 256 pixels @ 8bpp const auto linear_address = AddressT( ((location.v[0] >> 0) & 255) + (location.v[1] << 8) ); return expansion ? linear_address : rotate(linear_address); } } } else { return 0; } } uint8_t extract_colour(uint8_t byte, Vector location) const { switch(this->screen_mode_) { default: case ScreenMode::YamahaGraphics4: // 256 pixels @ 4bpp case ScreenMode::YamahaGraphics6: // 512 pixels @ 4bpp return (byte >> (((location.v[0] & 1) ^ 1) << 2)) & 0xf; case ScreenMode::YamahaGraphics5: // 512 pixels @ 2bpp return (byte >> (((location.v[0] & 3) ^ 3) << 1)) & 0x3; case ScreenMode::YamahaGraphics7: // 256 pixels @ 8bpp return byte; } } std::pair command_colour_mask(Vector location) const { if constexpr (is_yamaha_vdp(personality)) { auto &context = Storage::command_context_; auto colour = context.latched_colour.has_value() ? context.latched_colour : context.colour; switch(this->screen_mode_) { default: case ScreenMode::YamahaGraphics4: // 256 pixels @ 4bpp case ScreenMode::YamahaGraphics6: // 512 pixels @ 4bpp return std::make_pair( 0xf0 >> ((location.v[0] & 1) << 2), colour.colour4bpp ); case ScreenMode::YamahaGraphics5: // 512 pixels @ 2bpp return std::make_pair( 0xc0 >> ((location.v[0] & 3) << 1), colour.colour2bpp ); case ScreenMode::YamahaGraphics7: // 256 pixels @ 8bpp return std::make_pair( 0xff, colour.colour ); } } else { return std::make_pair(0, 0); } } void do_external_slot(int access_column) { // Don't do anything if the required time for the access to become executable // has yet to pass. if(queued_access_ == MemoryAccess::None || access_column < minimum_access_column_) { if constexpr (is_yamaha_vdp(personality)) { using CommandStep = typename Storage::CommandStep; if( Storage::next_command_step_ == CommandStep::None || access_column < Storage::minimum_command_column_ ) { return; } auto &context = Storage::command_context_; const uint8_t *const source = (context.arguments & 0x10) ? Storage::expansion_ram_.data() : ram_.data(); const AddressT source_mask = (context.arguments & 0x10) ? 0xfff : 0x1ffff; uint8_t *const destination = (context.arguments & 0x20) ? Storage::expansion_ram_.data() : ram_.data(); const AddressT destination_mask = (context.arguments & 0x20) ? 0xfff : 0x1ffff; switch(Storage::next_command_step_) { // Duplicative, but keeps the compiler happy. case CommandStep::None: break; case CommandStep::CopySourcePixelToStatus: Storage::colour_status_ = extract_colour( source[command_address(context.source, context.arguments & 0x10) & source_mask], context.source ); Storage::command_->advance(); Storage::update_command_step(access_column); break; case CommandStep::ReadSourcePixel: context.latched_colour.set( extract_colour( source[command_address(context.source, context.arguments & 0x10)] & source_mask, context.source) ); Storage::minimum_command_column_ = access_column + 32; Storage::next_command_step_ = CommandStep::ReadDestinationPixel; break; case CommandStep::ReadDestinationPixel: Storage::command_latch_ = source[command_address(context.destination, context.arguments & 0x20) & source_mask]; Storage::minimum_command_column_ = access_column + 24; Storage::next_command_step_ = CommandStep::WritePixel; break; case CommandStep::WritePixel: { const auto [mask, unmasked_colour] = command_colour_mask(context.destination); const auto address = command_address(context.destination, context.arguments & 0x20) & destination_mask; const uint8_t colour = unmasked_colour & mask; context.latched_colour.reset(); using LogicalOperation = CommandContext::LogicalOperation; if(!context.test_source || colour) { switch(context.pixel_operation) { default: case LogicalOperation::Copy: Storage::command_latch_ &= ~mask; Storage::command_latch_ |= colour; break; case LogicalOperation::And: Storage::command_latch_ &= ~mask | colour; break; case LogicalOperation::Or: Storage::command_latch_ |= colour; break; case LogicalOperation::Xor: Storage::command_latch_ ^= colour; break; case LogicalOperation::Not: Storage::command_latch_ &= ~mask; Storage::command_latch_ |= colour ^ mask; break; } } destination[address] = Storage::command_latch_; Storage::command_->advance(); Storage::update_command_step(access_column); } break; case CommandStep::ReadSourceByte: { Vector source_vector = context.source; if(Storage::command_->y_only) { source_vector.v[0] = context.destination.v[0]; } context.latched_colour.set(source[command_address(source_vector, context.arguments & 0x10) & source_mask]); Storage::minimum_command_column_ = access_column + 24; Storage::next_command_step_ = CommandStep::WriteByte; } break; case CommandStep::WriteByte: destination[command_address(context.destination, context.arguments & 0x20) & destination_mask] = context.latched_colour.has_value() ? context.latched_colour.colour : context.colour.colour; context.latched_colour.reset(); Storage::command_->advance(); Storage::update_command_step(access_column); break; } } return; } // Copy and mutate the RAM pointer. AddressT address = ram_pointer_; ++ram_pointer_; // Determine the relevant RAM and its mask. uint8_t *ram = ram_.data(); AddressT mask = memory_mask(personality); if constexpr (is_yamaha_vdp(personality)) { // The Yamaha increments only 14 bits of the address in TMS-compatible modes. if(this->underlying_mode_ < ScreenMode::YamahaText80) { ram_pointer_ = (ram_pointer_ & 0x3fff) | (address & AddressT(~0x3fff)); } if(this->underlying_mode_ == ScreenMode::YamahaGraphics6 || this->underlying_mode_ == ScreenMode::YamahaGraphics7) { // Rotate address one to the right as the hardware accesses // the underlying banks of memory alternately but presents // them as if linear. address = rotate(address); } // Also check whether expansion RAM is the true target here. if(Storage::command_context_.arguments & 0x40) { ram = Storage::expansion_ram_.data(); mask = AddressT(Storage::expansion_ram_.size() - 1); } } switch(queued_access_) { default: break; case MemoryAccess::Write: if constexpr (is_sega_vdp(personality)) { if(Storage::cram_is_selected_) { // Adjust the palette. In a Master System blue has a slightly different // scale; cf. https://www.retrorgb.com/sega-master-system-non-linear-blue-channel-findings.html constexpr uint8_t rg_scale[] = {0, 85, 170, 255}; constexpr uint8_t b_scale[] = {0, 104, 170, 255}; Storage::colour_ram_[address & 0x1f] = palette_pack( rg_scale[(read_ahead_buffer_ >> 0) & 3], rg_scale[(read_ahead_buffer_ >> 2) & 3], b_scale[(read_ahead_buffer_ >> 4) & 3] ); // Schedule a CRAM dot; this is scheduled for wherever it should appear // on screen. So it's wherever the output stream would be now. Which // is output_lag cycles ago from the point of view of the input stream. auto &dot = Storage::upcoming_cram_dots_.emplace_back(); dot.location.column = fetch_pointer_.column - output_lag; dot.location.row = fetch_pointer_.row; // Handle before this row conditionally; then handle after (or, more realistically, // exactly at the end of) naturally. if(dot.location.column < 0) { --dot.location.row; dot.location.column += 342; } dot.location.row += dot.location.column / 342; dot.location.column %= 342; dot.value = Storage::colour_ram_[address & 0x1f]; break; } } ram[address & mask] = read_ahead_buffer_; break; case MemoryAccess::Read: read_ahead_buffer_ = ram[address & mask]; break; } queued_access_ = MemoryAccess::None; } /// Helper for TMS dispatches; contains a switch statement with cases 0 to 170, each of the form: /// /// if constexpr (use_end && end == n) return; [[fallthrough]]; case n: fetcher.fetch(); /// /// i.e. it provides standard glue to enter a fetch sequence at any point, while the fetches themselves are templated on the cycle /// at which they appear for neater expression. template void dispatch(Fetcher &fetcher, int start, int end); // Various fetchers. template void fetch_tms_refresh(uint8_t y, int start, int end); template void fetch_tms_text(uint8_t y, int start, int end); template void fetch_tms_character(uint8_t y, int start, int end); template void fetch_yamaha(uint8_t y, int start, int end); template void fetch_yamaha(uint8_t y, int end); template void fetch_sms(uint8_t y, int start, int end); // A helper function to output the current border colour for // the number of cycles supplied. void output_border(int cycles, uint32_t cram_dot); // Output serialisation state. uint32_t *pixel_target_ = nullptr, *pixel_origin_ = nullptr; bool asked_for_write_area_ = false; // Output serialisers. template void draw_tms_character(int start, int end); template void draw_tms_text(int start, int end); void draw_sms(int start, int end, uint32_t cram_dot); template void draw_yamaha(uint8_t y, int start, int end); void draw_yamaha(uint8_t y, int start, int end); template void draw_sprites(uint8_t y, int start, int end, const std::array &palette, int *colour_buffer = nullptr); }; #include "Fetch.hpp" #include "Draw.hpp" } #endif /* TMS9918Base_hpp */