fix(window): proper inRange impl for window

window wrap now works (it's pretty slow though?)

.github/workflows/main.yml

@@ -1,9 +1,14 @@
+name: Nightly
+
 on:
   push:
+    paths:
+      - "**.zig"
     branches:
       - main
   schedule:
     - cron: '0 0 * * *'
+  workflow_dispatch:
 
 jobs:
   build:

lib/zig-clap

@@ -1 +1 @@
-Subproject commit e5d09c4b2d121025ad7195b2de704451e6306807
+Subproject commit 749c43f1f846adc950a5920ed61b40cbc3ec2c54

lib/zig-datetime

@@ -1 +1 @@
-Subproject commit 5ec1c36cf3791b3c6c5b330357bdb6feb93979ba
+Subproject commit 932d2845210644ca736faf35f5bea31eb1a15465

lib/zig-toml

@@ -1 +1 @@
-Subproject commit 5dfa919e03b446c66b295c04bef9bdecabd4276f
+Subproject commit be26d500917b5b3af4708f62ce22b990032a0ad3

src/config.zig

@@ -49,11 +49,11 @@ pub fn config() *const Config {
 }
 
 /// Reads a config file and then loads it into the global state
-pub fn load(allocator: Allocator, config_path: []const u8) !void {
-    var config_file = try std.fs.cwd().openFile(config_path, .{});
+pub fn load(allocator: Allocator, file_path: []const u8) !void {
+    var config_file = try std.fs.cwd().openFile(file_path, .{});
     defer config_file.close();
 
-    log.info("loaded from {s}", .{config_path});
+    log.info("loaded from {s}", .{file_path});
 
     const contents = try config_file.readToEndAlloc(allocator, try config_file.getEndPos());
     defer allocator.free(contents);

src/core/Bus.zig

@@ -102,12 +102,12 @@ pub fn deinit(self: *Self) void {
     // This is so I can deallocate the original `allocator.alloc`. I have to re-make the type
     // since I'm not keeping it around, This is very jank and bad though
     // FIXME: please figure out another way
-    self.allocator.free(@ptrCast([*]const ?*anyopaque, self.write_tables[0][0..])[0 .. 3 * table_len]);
+    self.allocator.free(@ptrCast([*]const ?*anyopaque, self.read_table[0..])[0 .. 3 * table_len]);
     self.* = undefined;
 }
 
 fn fillReadTable(bus: *Self, table: *[table_len]?*const anyopaque) void {
-    const vramMirror = @import("ppu.zig").Vram.mirror;
+    const vramMirror = @import("ppu/Vram.zig").mirror;
 
     for (table) |*ptr, i| {
         const addr = page_size * i;
@@ -134,7 +134,7 @@ fn fillReadTable(bus: *Self, table: *[table_len]?*const anyopaque) void {
 
 fn fillWriteTable(comptime T: type, bus: *Self, table: *[table_len]?*const anyopaque) void {
     comptime std.debug.assert(T == u32 or T == u16 or T == u8);
-    const vramMirror = @import("ppu.zig").Vram.mirror;
+    const vramMirror = @import("ppu/Vram.zig").mirror;
 
     for (table) |*ptr, i| {
         const addr = page_size * i;

src/core/bus/dma.zig

@@ -333,7 +333,7 @@ fn DmaController(comptime id: u2) type {
     };
 }
 
-pub fn pollDmaOnBlank(bus: *Bus, comptime kind: DmaKind) void {
+pub fn onBlanking(bus: *Bus, comptime kind: DmaKind) void {
     comptime var i: usize = 0;
     inline while (i < 4) : (i += 1) {
         bus.dma[i].poll(kind);

src/core/bus/io.zig

@@ -424,6 +424,8 @@ pub const BldY = extern union {
     raw: u16,
 };
 
+const u8WriteKind = enum { Hi, Lo };
+
 /// Write-only
 pub const WinH = extern union {
     x2: Bitfield(u16, 0, 8),
@@ -433,6 +435,8 @@ pub const WinH = extern union {
 
 /// Write-only
 pub const WinV = extern union {
+    const Self = @This();
+
     y2: Bitfield(u16, 0, 8),
     y1: Bitfield(u16, 8, 8),
     raw: u16,
@@ -441,20 +445,20 @@ pub const WinV = extern union {
 pub const WinIn = extern union {
     w0_bg: Bitfield(u16, 0, 4),
     w0_obj: Bit(u16, 4),
-    w0_colour: Bit(u16, 5),
+    w0_bld: Bit(u16, 5),
     w1_bg: Bitfield(u16, 8, 4),
     w1_obj: Bit(u16, 12),
-    w1_colour: Bit(u16, 13),
+    w1_bld: Bit(u16, 13),
     raw: u16,
 };
 
 pub const WinOut = extern union {
     out_bg: Bitfield(u16, 0, 4),
     out_obj: Bit(u16, 4),
-    out_colour: Bit(u16, 5),
+    out_bld: Bit(u16, 5),
     obj_bg: Bitfield(u16, 8, 4),
     obj_obj: Bit(u16, 12),
-    obj_colour: Bit(u16, 13),
+    obj_bld: Bit(u16, 13),
     raw: u16,
 };
 

src/core/ppu.zig

@@ -2,11 +2,16 @@ const std = @import("std");
 const io = @import("bus/io.zig");
 const util = @import("../util.zig");
 
-const Scheduler = @import("scheduler.zig").Scheduler;
-const Arm7tdmi = @import("cpu.zig").Arm7tdmi;
-
 const Bit = @import("bitfield").Bit;
 const Bitfield = @import("bitfield").Bitfield;
+const dma = @import("bus/dma.zig");
+
+const Oam = @import("ppu/Oam.zig");
+const Palette = @import("ppu/Palette.zig");
+const Vram = @import("ppu/Vram.zig");
+const Scheduler = @import("scheduler.zig").Scheduler;
+const Arm7tdmi = @import("cpu.zig").Arm7tdmi;
+const FrameBuffer = @import("../util.zig").FrameBuffer;
 
 const Allocator = std.mem.Allocator;
 const log = std.log.scoped(.PPU);
@@ -14,7 +19,6 @@ const log = std.log.scoped(.PPU);
 const getHalf = util.getHalf;
 const setHalf = util.setHalf;
 const setQuart = util.setQuart;
-const pollDmaOnBlank = @import("bus/dma.zig").pollDmaOnBlank;
 
 pub const width = 240;
 pub const height = 160;
@@ -259,18 +263,17 @@ pub const Ppu = struct {
     scanline: Scanline,
 
     pub fn init(allocator: Allocator, sched: *Scheduler) !Self {
-        // Queue first Hblank
-        sched.push(.Draw, 240 * 4);
+        sched.push(.Draw, 240 * 4); // Add first PPU Event to Scheduler
 
         const sprites = try allocator.create([128]?Sprite);
-        sprites.* = [_]?Sprite{null} ** 128;
+        std.mem.set(?Sprite, sprites, null);
 
         return Self{
             .vram = try Vram.init(allocator),
             .palette = try Palette.init(allocator),
             .oam = try Oam.init(allocator),
             .sched = sched,
-            .framebuf = try FrameBuffer.init(allocator),
+            .framebuf = try FrameBuffer.init(allocator, framebuf_pitch * height),
             .allocator = allocator,
 
             // Registers
@@ -320,20 +323,16 @@ pub const Ppu = struct {
             // Only consider enabled Sprites
             if (attr0.is_affine.read() or !attr0.disabled.read()) {
                 const attr1 = @bitCast(Attr1, self.oam.read(u16, i + 2));
+                const sprite_height = spriteDimensions(attr0.shape.read(), attr1.size.read())[1];
 
                 // When fetching sprites we only care about ones that could be rendered
                 // on this scanline
-                const iy = @bitCast(i8, y);
-
-                const start = attr0.y.read();
-                const istart = @bitCast(i8, start);
-
-                const end = start +% spriteDimensions(attr0.shape.read(), attr1.size.read())[1];
-                const iend = @bitCast(i8, end);
+                var y_pos: i32 = attr0.y.read();
+                if (y_pos >= 160) y_pos -= 256; // fleroviux's solution to negative positions
 
                 // Sprites are expected to be able to wraparound, we perform the same check
                 // for unsigned and signed values so that we handle all valid sprite positions
-                if ((start <= y and y < end) or (istart <= iy and iy < iend)) {
+                if (y_pos <= y and y < (y_pos + sprite_height)) {
                     for (self.scanline_sprites) |*maybe_sprite| {
                         if (maybe_sprite.* == null) {
                             maybe_sprite.* = Sprite.init(attr0, attr1, @bitCast(Attr2, self.oam.read(u16, i + 4)));
@@ -360,8 +359,6 @@ pub const Ppu = struct {
     }
 
     fn drawAffineSprite(self: *Self, sprite: AffineSprite) void {
-        const iy = @bitCast(i8, self.vcount.scanline.read());
-
         const is_8bpp = sprite.is8bpp();
         const tile_id: u32 = sprite.tileId();
         const obj_mapping = self.dispcnt.obj_mapping.read();
@@ -370,25 +367,22 @@ pub const Ppu = struct {
 
         const char_base = 0x4000 * 4;
 
+        const y = self.vcount.scanline.read();
+
         var i: u9 = 0;
         while (i < sprite.width) : (i += 1) {
             const x = (sprite.x() +% i) % width;
-            const ix = @bitCast(i9, x);
-
             if (!shouldDrawSprite(self.bld.cnt, &self.scanline, x)) continue;
 
-            const sprite_start = sprite.x();
-            const isprite_start = @bitCast(i9, sprite_start);
-            const sprite_end = sprite_start +% sprite.width;
-            const isprite_end = @bitCast(i9, sprite_end);
+            var x_pos: i32 = sprite.x();
+            if (x_pos >= 240) x_pos -= 512;
 
-            const condition = (sprite_start <= x and x < sprite_end) or (isprite_start <= ix and ix < isprite_end);
-            if (!condition) continue;
+            if (!(x_pos <= x and x < (x_pos + sprite.width))) continue;
 
             // Sprite is within bounds and therefore should be rendered
             // std.math.absInt is branchless
-            const tile_x = @bitCast(u9, std.math.absInt(ix - @bitCast(i9, sprite.x())) catch unreachable);
-            const tile_y = @bitCast(u8, std.math.absInt(iy -% @bitCast(i8, sprite.y())) catch unreachable);
+            const tile_x = @bitCast(u32, @as(i32, std.math.absInt(@as(i32, x) - x_pos) catch unreachable));
+            const tile_y = @bitCast(u32, @as(i32, std.math.absInt(@bitCast(i8, y) -% @bitCast(i8, sprite.y())) catch unreachable));
 
             const row = @truncate(u3, tile_y);
             const col = @truncate(u3, tile_x);
@@ -410,8 +404,6 @@ pub const Ppu = struct {
     }
 
     fn drawSprite(self: *Self, sprite: Sprite) void {
-        const iy = @bitCast(i8, self.vcount.scanline.read());
-
         const is_8bpp = sprite.is8bpp();
         const tile_id: u32 = sprite.tileId();
         const obj_mapping = self.dispcnt.obj_mapping.read();
@@ -420,31 +412,27 @@ pub const Ppu = struct {
 
         const char_base = 0x4000 * 4;
 
+        const y = self.vcount.scanline.read();
+
         var i: u9 = 0;
         while (i < sprite.width) : (i += 1) {
             const x = (sprite.x() +% i) % width;
-            const ix = @bitCast(i9, x);
-
             if (!shouldDrawSprite(self.bld.cnt, &self.scanline, x)) continue;
 
-            const sprite_start = sprite.x();
-            const isprite_start = @bitCast(i9, sprite_start);
-            const sprite_end = sprite_start +% sprite.width;
-            const isprite_end = @bitCast(i9, sprite_end);
+            var x_pos: i32 = sprite.x();
+            if (x_pos >= 240) x_pos -= 512;
 
-            const condition = (sprite_start <= x and x < sprite_end) or (isprite_start <= ix and ix < isprite_end);
-            if (!condition) continue;
+            if (!(x_pos <= x and x < (x_pos + sprite.width))) continue;
 
             // Sprite is within bounds and therefore should be rendered
-            // std.math.absInt is branchless
-            const x_diff = @bitCast(u9, std.math.absInt(ix - @bitCast(i9, sprite.x())) catch unreachable);
-            const y_diff = @bitCast(u8, std.math.absInt(iy -% @bitCast(i8, sprite.y())) catch unreachable);
+            const x_diff: i32 = std.math.absInt(@as(i32, x) - x_pos) catch unreachable;
+            const y_diff: i32 = std.math.absInt(@bitCast(i8, y) -% @bitCast(i8, sprite.y())) catch unreachable;
 
             // Note that we flip the tile_pos not the (tile_pos % 8) like we do for
             // Background Tiles. By doing this we mirror the entire sprite instead of
             // just a specific tile (see how sprite.width and sprite.height are involved)
-            const tile_y = y_diff ^ if (sprite.vFlip()) (sprite.height - 1) else 0;
-            const tile_x = x_diff ^ if (sprite.hFlip()) (sprite.width - 1) else 0;
+            const tile_x = @intCast(u9, x_diff) ^ if (sprite.hFlip()) (sprite.width - 1) else 0;
+            const tile_y = @intCast(u8, y_diff) ^ if (sprite.vFlip()) (sprite.height - 1) else 0;
 
             const row = @truncate(u3, tile_y);
             const col = @truncate(u3, tile_x);
@@ -487,16 +475,17 @@ pub const Ppu = struct {
             aff_x += self.aff_bg[n - 2].pa;
             aff_y += self.aff_bg[n - 2].pc;
 
-            if (!shouldDrawBackground(n, self.bld.cnt, &self.scanline, i)) continue;
+            const x = @bitCast(u32, ix);
+            const y = @bitCast(u32, iy);
+
+            const win_bounds = self.windowBounds(@truncate(u9, x), @truncate(u8, y));
+            if (!shouldDrawBackground(self, n, win_bounds, i)) continue;
 
             if (self.bg[n].cnt.display_overflow.read()) {
                 ix = if (ix > px_width) @rem(ix, px_width) else if (ix < 0) px_width + @rem(ix, px_width) else ix;
                 iy = if (iy > px_height) @rem(iy, px_height) else if (iy < 0) px_height + @rem(iy, px_height) else iy;
             } else if (ix > px_width or iy > px_height or ix < 0 or iy < 0) continue;
 
-            const x = @bitCast(u32, ix);
-            const y = @bitCast(u32, iy);
-
             const tile_id: u32 = self.vram.read(u8, screen_base + ((y / 8) * @bitCast(u32, tile_width) + (x / 8)));
             const row = y & 7;
             const col = x & 7;
@@ -506,7 +495,7 @@ pub const Ppu = struct {
 
             if (pal_id != 0) {
                 const bgr555 = self.palette.read(u16, pal_id * 2);
-                copyToBackgroundBuffer(n, self.bld.cnt, &self.scanline, i, bgr555);
+                self.copyToBackgroundBuffer(n, win_bounds, i, bgr555);
             }
         }
 
@@ -515,7 +504,7 @@ pub const Ppu = struct {
         self.aff_bg[n - 2].y_latch.? += self.aff_bg[n - 2].pd; // PD is added to BGxY
     }
 
-    fn drawBackround(self: *Self, comptime n: u2) void {
+    fn drawBackground(self: *Self, comptime n: u2) void {
         // A Tile in a charblock is a byte, while a Screen Entry is a halfword
 
         const char_base = 0x4000 * @as(u32, self.bg[n].cnt.char_base.read());
@@ -535,10 +524,11 @@ pub const Ppu = struct {
 
         var i: u32 = 0;
         while (i < width) : (i += 1) {
-            if (!shouldDrawBackground(n, self.bld.cnt, &self.scanline, i)) continue;
-
             const x = hofs + i;
 
+            const win_bounds = self.windowBounds(@truncate(u9, x), @truncate(u8, y));
+            if (!shouldDrawBackground(self, n, win_bounds, i)) continue;
+
             // Grab the Screen Entry from VRAM
             const entry_addr = screen_base + tilemapOffset(size, x, y);
             const entry = @bitCast(ScreenEntry, self.vram.read(u16, entry_addr));
@@ -563,7 +553,7 @@ pub const Ppu = struct {
 
             if (pal_id != 0) {
                 const bgr555 = self.palette.read(u16, pal_id * 2);
-                copyToBackgroundBuffer(n, self.bld.cnt, &self.scanline, i, bgr555);
+                self.copyToBackgroundBuffer(n, win_bounds, i, bgr555);
             }
         }
     }
@@ -589,10 +579,10 @@ pub const Ppu = struct {
                 var layer: usize = 0;
                 while (layer < 4) : (layer += 1) {
                     self.drawSprites(@truncate(u2, layer));
-                    if (layer == self.bg[0].cnt.priority.read() and bg_enable & 1 == 1) self.drawBackround(0);
-                    if (layer == self.bg[1].cnt.priority.read() and bg_enable >> 1 & 1 == 1) self.drawBackround(1);
-                    if (layer == self.bg[2].cnt.priority.read() and bg_enable >> 2 & 1 == 1) self.drawBackround(2);
-                    if (layer == self.bg[3].cnt.priority.read() and bg_enable >> 3 & 1 == 1) self.drawBackround(3);
+                    if (layer == self.bg[0].cnt.priority.read() and bg_enable & 1 == 1) self.drawBackground(0);
+                    if (layer == self.bg[1].cnt.priority.read() and bg_enable >> 1 & 1 == 1) self.drawBackground(1);
+                    if (layer == self.bg[2].cnt.priority.read() and bg_enable >> 2 & 1 == 1) self.drawBackground(2);
+                    if (layer == self.bg[3].cnt.priority.read() and bg_enable >> 3 & 1 == 1) self.drawBackground(3);
                 }
 
                 // Copy Drawn Scanline to Frame Buffer
@@ -617,8 +607,8 @@ pub const Ppu = struct {
                 var layer: usize = 0;
                 while (layer < 4) : (layer += 1) {
                     self.drawSprites(@truncate(u2, layer));
-                    if (layer == self.bg[0].cnt.priority.read() and bg_enable & 1 == 1) self.drawBackround(0);
-                    if (layer == self.bg[1].cnt.priority.read() and bg_enable >> 1 & 1 == 1) self.drawBackround(1);
+                    if (layer == self.bg[0].cnt.priority.read() and bg_enable & 1 == 1) self.drawBackground(0);
+                    if (layer == self.bg[1].cnt.priority.read() and bg_enable >> 1 & 1 == 1) self.drawBackground(1);
                     if (layer == self.bg[2].cnt.priority.read() and bg_enable >> 2 & 1 == 1) self.drawAffineBackground(2);
                 }
 
@@ -696,7 +686,7 @@ pub const Ppu = struct {
                 while (i < width) : (i += 1) {
                     // If we're outside of the bounds of mode 5, draw the background colour
                     const bgr555 =
-                        if (scanline < m5_height and i < m5_width) self.vram.read(u16, vram_base + i * @sizeOf(u16)) else self.palette.getBackdrop();
+                        if (scanline < m5_height and i < m5_width) self.vram.read(u16, vram_base + i * @sizeOf(u16)) else self.palette.backdrop();
 
                     std.mem.writeIntNative(u32, self.framebuf.get(.Emulator)[fb_base + i * @sizeOf(u32) ..][0..@sizeOf(u32)], rgba888(bgr555));
                 }
@@ -740,7 +730,94 @@ pub const Ppu = struct {
         }
 
         if (maybe_top) |top| return top;
-        return self.palette.getBackdrop();
+        return self.palette.backdrop();
+    }
+
+    fn copyToBackgroundBuffer(self: *Self, comptime n: u2, bounds: ?WindowBounds, i: usize, bgr555: u16) void {
+        if (self.bld.cnt.mode.read() != 0b00) {
+            // Standard Alpha Blending
+            const a_layers = self.bld.cnt.layer_a.read();
+            const is_blend_enabled = (a_layers >> n) & 1 == 1;
+
+            // If Alpha Blending is enabled and we've found an eligible layer for
+            // Pixel A, store the pixel in the bottom pixel buffer
+
+            const win_part = if (bounds) |win| blk: {
+                // Window Enabled
+                break :blk switch (win) {
+                    .win0 => self.win.in.w0_bld.read(),
+                    .win1 => self.win.in.w1_bld.read(),
+                    .out => self.win.out.out_bld.read(),
+                };
+            } else true;
+
+            if (win_part and is_blend_enabled) {
+                self.scanline.btm()[i] = bgr555;
+                return;
+            }
+        }
+
+        self.scanline.top()[i] = bgr555;
+    }
+
+    const WindowBounds = enum { win0, win1, out };
+
+    fn windowBounds(self: *Self, x: u9, y: u8) ?WindowBounds {
+        const win0 = self.dispcnt.win_enable.read() & 1 == 1;
+        const win1 = (self.dispcnt.win_enable.read() >> 1) & 1 == 1;
+        const winObj = self.dispcnt.obj_win_enable.read();
+
+        if (!(win0 or win1 or winObj)) return null;
+
+        if (win0 and self.win.inRange(0, x, y)) return .win0;
+        if (win1 and self.win.inRange(1, x, y)) return .win1;
+
+        return .out;
+    }
+
+    fn shouldDrawBackground(self: *Self, comptime n: u2, bounds: ?WindowBounds, i: usize) bool {
+        // If a pixel has been drawn on the top layer, it's because:
+        // 1. The pixel is to be blended with a pixel on the bottom layer
+        // 2. The pixel is not to be blended at all
+        // Also, if we find a pixel on the top layer we don't need to bother with this I think?
+        if (self.scanline.top()[i] != null) return false;
+
+        if (bounds) |win| {
+            switch (win) {
+                .win0 => if ((self.win.in.w0_bg.read() >> n) & 1 == 0) return false,
+                .win1 => if ((self.win.in.w1_bg.read() >> n) & 1 == 0) return false,
+                .out => if ((self.win.out.out_bg.read() >> n) & 1 == 0) return false,
+            }
+        }
+
+        if (self.scanline.btm()[i] != null) {
+            // The pixel found in the bottom layer is:
+            // 1. From a higher priority background
+            // 2. From a background that is marked for blending (Pixel A)
+
+            // If Alpha Blending isn't enabled, then we've already found a higher prio
+            // pixel, we can return early
+            if (self.bld.cnt.mode.read() != 0b01) return false;
+
+            const b_layers = self.bld.cnt.layer_b.read();
+
+            const win_part = if (bounds) |win| blk: {
+                // Window Enabled
+                break :blk switch (win) {
+                    .win0 => self.win.in.w0_bld.read(),
+                    .win1 => self.win.in.w1_bld.read(),
+                    .out => self.win.out.out_bld.read(),
+                };
+            } else true;
+
+            // If the Background is not marked for blending, we've already found
+            // a higher priority pixel, move on.
+
+            const is_blend_enabled = win_part and ((b_layers >> n) & 1 == 1);
+            if (!is_blend_enabled) return false;
+        }
+
+        return true;
     }
 
     // TODO: Comment this + get a better understanding
@@ -782,7 +859,7 @@ pub const Ppu = struct {
         // See if HBlank DMA is present and not enabled
 
         if (!self.dispstat.vblank.read())
-            pollDmaOnBlank(cpu.bus, .HBlank);
+            dma.onBlanking(cpu.bus, .HBlank);
 
         self.dispstat.hblank.set();
         self.sched.push(.HBlank, 68 * 4 -| late);
@@ -824,7 +901,7 @@ pub const Ppu = struct {
                 self.aff_bg[1].latchRefPoints();
 
                 // See if Vblank DMA is present and not enabled
-                pollDmaOnBlank(cpu.bus, .VBlank);
+                dma.onBlanking(cpu.bus, .VBlank);
             }
 
             if (scanline == 227) self.dispstat.vblank.unset();
@@ -833,158 +910,6 @@ pub const Ppu = struct {
     }
 };
 
-const Palette = struct {
-    const palram_size = 0x400;
-    const Self = @This();
-
-    buf: []u8,
-    allocator: Allocator,
-
-    fn init(allocator: Allocator) !Self {
-        const buf = try allocator.alloc(u8, palram_size);
-        std.mem.set(u8, buf, 0);
-
-        return Self{
-            .buf = buf,
-            .allocator = allocator,
-        };
-    }
-
-    fn deinit(self: *Self) void {
-        self.allocator.free(self.buf);
-        self.* = undefined;
-    }
-
-    pub fn read(self: *const Self, comptime T: type, address: usize) T {
-        const addr = address & 0x3FF;
-
-        return switch (T) {
-            u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
-            else => @compileError("PALRAM: Unsupported read width"),
-        };
-    }
-
-    pub fn write(self: *Self, comptime T: type, address: usize, value: T) void {
-        const addr = address & 0x3FF;
-
-        switch (T) {
-            u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)], value),
-            u8 => {
-                const align_addr = addr & ~@as(u32, 1); // Aligned to Halfword boundary
-                std.mem.writeIntSliceLittle(u16, self.buf[align_addr..][0..@sizeOf(u16)], @as(u16, value) * 0x101);
-            },
-            else => @compileError("PALRAM: Unsupported write width"),
-        }
-    }
-
-    fn getBackdrop(self: *const Self) u16 {
-        return self.read(u16, 0);
-    }
-};
-
-pub const Vram = struct {
-    const vram_size = 0x18000;
-    const Self = @This();
-
-    buf: []u8,
-    allocator: Allocator,
-
-    fn init(allocator: Allocator) !Self {
-        const buf = try allocator.alloc(u8, vram_size);
-        std.mem.set(u8, buf, 0);
-
-        return Self{
-            .buf = buf,
-            .allocator = allocator,
-        };
-    }
-
-    fn deinit(self: *Self) void {
-        self.allocator.free(self.buf);
-        self.* = undefined;
-    }
-
-    pub fn read(self: *const Self, comptime T: type, address: usize) T {
-        const addr = Self.mirror(address);
-
-        return switch (T) {
-            u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
-            else => @compileError("VRAM: Unsupported read width"),
-        };
-    }
-
-    pub fn write(self: *Self, comptime T: type, dispcnt: io.DisplayControl, address: usize, value: T) void {
-        const mode: u3 = dispcnt.bg_mode.read();
-        const idx = Self.mirror(address);
-
-        switch (T) {
-            u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[idx..][0..@sizeOf(T)], value),
-            u8 => {
-                // Ignore write if it falls within the boundaries of OBJ VRAM
-                switch (mode) {
-                    0, 1, 2 => if (0x0001_0000 <= idx) return,
-                    else => if (0x0001_4000 <= idx) return,
-                }
-
-                const align_idx = idx & ~@as(u32, 1); // Aligned to a halfword boundary
-                std.mem.writeIntSliceLittle(u16, self.buf[align_idx..][0..@sizeOf(u16)], @as(u16, value) * 0x101);
-            },
-            else => @compileError("VRAM: Unsupported write width"),
-        }
-    }
-
-    pub fn mirror(address: usize) usize {
-        // Mirrored in steps of 128K (64K + 32K + 32K) (abcc)
-        const addr = address & 0x1FFFF;
-
-        // If the address is within 96K we don't do anything,
-        // otherwise we want to mirror the last 32K (addresses between 64K and 96K)
-        return if (addr < vram_size) addr else 0x10000 + (addr & 0x7FFF);
-    }
-};
-
-const Oam = struct {
-    const oam_size = 0x400;
-    const Self = @This();
-
-    buf: []u8,
-    allocator: Allocator,
-
-    fn init(allocator: Allocator) !Self {
-        const buf = try allocator.alloc(u8, oam_size);
-        std.mem.set(u8, buf, 0);
-
-        return Self{
-            .buf = buf,
-            .allocator = allocator,
-        };
-    }
-
-    fn deinit(self: *Self) void {
-        self.allocator.free(self.buf);
-        self.* = undefined;
-    }
-
-    pub fn read(self: *const Self, comptime T: type, address: usize) T {
-        const addr = address & 0x3FF;
-
-        return switch (T) {
-            u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
-            else => @compileError("OAM: Unsupported read width"),
-        };
-    }
-
-    pub fn write(self: *Self, comptime T: type, address: usize, value: T) void {
-        const addr = address & 0x3FF;
-
-        switch (T) {
-            u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)], value),
-            u8 => return, // 8-bit writes are explicitly ignored
-            else => @compileError("OAM: Unsupported write width"),
-        }
-    }
-};
-
 const Blend = struct {
     const Self = @This();
 
@@ -1036,6 +961,36 @@ const Window = struct {
         return self.out.raw & 0x3F3F;
     }
 
+    fn inRange(self: *const Self, comptime id: u1, x: u9, y: u8) bool {
+        const winh = self.h[id];
+        const winv = self.v[id];
+
+        if (isYInRange(winv, y)) {
+            const x1 = winh.x1.read();
+            const x2 = winh.x2.read();
+
+            // Within X Bounds
+            return if (x1 < x2) blk: {
+                break :blk x >= x1 and x < x2;
+            } else blk: {
+                break :blk x >= x1 or x < x2;
+            };
+        }
+
+        return false;
+    }
+
+    inline fn isYInRange(winv: io.WinV, y: u9) bool {
+        const y1 = winv.y1.read();
+        const y2 = winv.y2.read();
+
+        if (y1 < y2) {
+            return y >= y1 and y < y2;
+        } else {
+            return y >= y1 or y < y2;
+        }
+    }
+
     pub fn setH(self: *Self, value: u32) void {
         self.h[0].raw = @truncate(u16, value);
         self.h[1].raw = @truncate(u16, value >> 16);
@@ -1343,37 +1298,6 @@ fn alphaBlend(top: u16, btm: u16, bldalpha: io.BldAlpha) u16 {
     return (bld_b << 10) | (bld_g << 5) | bld_r;
 }
 
-fn shouldDrawBackground(comptime n: u2, bldcnt: io.BldCnt, scanline: *Scanline, i: usize) bool {
-    // If a pixel has been drawn on the top layer, it's because
-    // Either the pixel is to be blended with a pixel on the bottom layer
-    // or the pixel is not to be blended at all
-    // Consequentially, if we find a pixel on the top layer, there's no need
-    // to render anything I think?
-    if (scanline.top()[i] != null) return false;
-
-    if (scanline.btm()[i] != null) {
-        // The Pixel found in the Bottom layer is
-        // 1. From a higher priority
-        // 2. From a Backround that is marked for Blending (Pixel A)
-        //
-        // We now have to confirm whether this current Background can be used
-        // as Pixel B or not.
-
-        // If Alpha Blending isn't enabled, we've aready found a higher
-        // priority pixel to render. Move on
-        if (bldcnt.mode.read() != 0b01) return false;
-
-        const b_layers = bldcnt.layer_b.read();
-        const is_blend_enabled = (b_layers >> n) & 1 == 1;
-
-        // If the Background is not marked for blending, we've already found
-        // a higher priority pixel, move on.
-        if (!is_blend_enabled) return false;
-    }
-
-    return true;
-}
-
 fn shouldDrawSprite(bldcnt: io.BldCnt, scanline: *Scanline, x: u9) bool {
     if (scanline.top()[x] != null) return false;
 
@@ -1388,23 +1312,6 @@ fn shouldDrawSprite(bldcnt: io.BldCnt, scanline: *Scanline, x: u9) bool {
     return true;
 }
 
-fn copyToBackgroundBuffer(comptime n: u2, bldcnt: io.BldCnt, scanline: *Scanline, i: usize, bgr555: u16) void {
-    if (bldcnt.mode.read() != 0b00) {
-        // Standard Alpha Blending
-        const a_layers = bldcnt.layer_a.read();
-        const is_blend_enabled = (a_layers >> n) & 1 == 1;
-
-        // If Alpha Blending is enabled and we've found an eligible layer for
-        // Pixel A, store the pixel in the bottom pixel buffer
-        if (is_blend_enabled) {
-            scanline.btm()[i] = bgr555;
-            return;
-        }
-    }
-
-    scanline.top()[i] = bgr555;
-}
-
 fn copyToSpriteBuffer(bldcnt: io.BldCnt, scanline: *Scanline, x: u9, bgr555: u16) void {
     if (bldcnt.mode.read() != 0b00) {
         // Alpha Blending
@@ -1457,48 +1364,3 @@ const Scanline = struct {
         return self.layers[1];
     }
 };
-
-// Double Buffering Implementation
-const FrameBuffer = struct {
-    const Self = @This();
-
-    layers: [2][]u8,
-    buf: []u8,
-    current: u1,
-
-    allocator: Allocator,
-
-    // TODO: Rename
-    const Device = enum {
-        Emulator,
-        Renderer,
-    };
-
-    pub fn init(allocator: Allocator) !Self {
-        const framebuf_len = framebuf_pitch * height;
-        const buf = try allocator.alloc(u8, framebuf_len * 2);
-        std.mem.set(u8, buf, 0);
-
-        return .{
-            // Front and Back Framebuffers
-            .layers = [_][]u8{ buf[0..][0..framebuf_len], buf[framebuf_len..][0..framebuf_len] },
-            .buf = buf,
-            .current = 0,
-
-            .allocator = allocator,
-        };
-    }
-
-    fn deinit(self: *Self) void {
-        self.allocator.free(self.buf);
-        self.* = undefined;
-    }
-
-    pub fn swap(self: *Self) void {
-        self.current = ~self.current;
-    }
-
-    pub fn get(self: *Self, comptime dev: Device) []u8 {
-        return self.layers[if (dev == .Emulator) self.current else ~self.current];
-    }
-};

src/core/ppu/Oam.zig

@@ -0,0 +1,40 @@
+const std = @import("std");
+
+const Allocator = std.mem.Allocator;
+
+const buf_len = 0x400;
+const Self = @This();
+
+buf: []u8,
+allocator: Allocator,
+
+pub fn read(self: *const Self, comptime T: type, address: usize) T {
+    const addr = address & 0x3FF;
+
+    return switch (T) {
+        u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
+        else => @compileError("OAM: Unsupported read width"),
+    };
+}
+
+pub fn write(self: *Self, comptime T: type, address: usize, value: T) void {
+    const addr = address & 0x3FF;
+
+    switch (T) {
+        u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)], value),
+        u8 => return, // 8-bit writes are explicitly ignored
+        else => @compileError("OAM: Unsupported write width"),
+    }
+}
+
+pub fn init(allocator: Allocator) !Self {
+    const buf = try allocator.alloc(u8, buf_len);
+    std.mem.set(u8, buf, 0);
+
+    return Self{ .buf = buf, .allocator = allocator };
+}
+
+pub fn deinit(self: *Self) void {
+    self.allocator.free(self.buf);
+    self.* = undefined;
+}

src/core/ppu/Palette.zig

@@ -0,0 +1,47 @@
+const std = @import("std");
+
+const Allocator = std.mem.Allocator;
+
+const buf_len = 0x400;
+const Self = @This();
+
+buf: []u8,
+allocator: Allocator,
+
+pub fn read(self: *const Self, comptime T: type, address: usize) T {
+    const addr = address & 0x3FF;
+
+    return switch (T) {
+        u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
+        else => @compileError("PALRAM: Unsupported read width"),
+    };
+}
+
+pub fn write(self: *Self, comptime T: type, address: usize, value: T) void {
+    const addr = address & 0x3FF;
+
+    switch (T) {
+        u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)], value),
+        u8 => {
+            const align_addr = addr & ~@as(u32, 1); // Aligned to Halfword boundary
+            std.mem.writeIntSliceLittle(u16, self.buf[align_addr..][0..@sizeOf(u16)], @as(u16, value) * 0x101);
+        },
+        else => @compileError("PALRAM: Unsupported write width"),
+    }
+}
+
+pub fn init(allocator: Allocator) !Self {
+    const buf = try allocator.alloc(u8, buf_len);
+    std.mem.set(u8, buf, 0);
+
+    return Self{ .buf = buf, .allocator = allocator };
+}
+
+pub fn deinit(self: *Self) void {
+    self.allocator.free(self.buf);
+    self.* = undefined;
+}
+
+pub fn backdrop(self: *const Self) u16 {
+    return self.read(u16, 0);
+}

src/core/ppu/Vram.zig

@@ -0,0 +1,60 @@
+const std = @import("std");
+const io = @import("../bus/io.zig");
+
+const Allocator = std.mem.Allocator;
+
+const buf_len = 0x18000;
+const Self = @This();
+
+buf: []u8,
+allocator: Allocator,
+
+pub fn read(self: *const Self, comptime T: type, address: usize) T {
+    const addr = Self.mirror(address);
+
+    return switch (T) {
+        u32, u16, u8 => std.mem.readIntSliceLittle(T, self.buf[addr..][0..@sizeOf(T)]),
+        else => @compileError("VRAM: Unsupported read width"),
+    };
+}
+
+pub fn write(self: *Self, comptime T: type, dispcnt: io.DisplayControl, address: usize, value: T) void {
+    const mode: u3 = dispcnt.bg_mode.read();
+    const idx = Self.mirror(address);
+
+    switch (T) {
+        u32, u16 => std.mem.writeIntSliceLittle(T, self.buf[idx..][0..@sizeOf(T)], value),
+        u8 => {
+            // Ignore write if it falls within the boundaries of OBJ VRAM
+            switch (mode) {
+                0, 1, 2 => if (0x0001_0000 <= idx) return,
+                else => if (0x0001_4000 <= idx) return,
+            }
+
+            const align_idx = idx & ~@as(u32, 1); // Aligned to a halfword boundary
+            std.mem.writeIntSliceLittle(u16, self.buf[align_idx..][0..@sizeOf(u16)], @as(u16, value) * 0x101);
+        },
+        else => @compileError("VRAM: Unsupported write width"),
+    }
+}
+
+pub fn init(allocator: Allocator) !Self {
+    const buf = try allocator.alloc(u8, buf_len);
+    std.mem.set(u8, buf, 0);
+
+    return Self{ .buf = buf, .allocator = allocator };
+}
+
+pub fn deinit(self: *Self) void {
+    self.allocator.free(self.buf);
+    self.* = undefined;
+}
+
+pub fn mirror(address: usize) usize {
+    // Mirrored in steps of 128K (64K + 32K + 32K) (abcc)
+    const addr = address & 0x1FFFF;
+
+    // If the address is within 96K we don't do anything,
+    // otherwise we want to mirror the last 32K (addresses between 64K and 96K)
+    return if (addr < buf_len) addr else 0x10000 + (addr & 0x7FFF);
+}

src/main.zig

@@ -26,33 +26,44 @@ const params = clap.parseParamsComptime(
     \\
 );
 
-pub fn main() anyerror!void {
+pub fn main() void {
     // Main Allocator for ZBA
     var gpa = std.heap.GeneralPurposeAllocator(.{}){};
     defer std.debug.assert(!gpa.deinit());
 
     const allocator = gpa.allocator();
 
-    // Determine the Data Directory (stores saves, config file, etc.)
+    // Determine the Data Directory (stores saves)
     const data_path = blk: {
         const result = known_folders.getPath(allocator, .data);
-        const option = result catch |e| exitln("interrupted while attempting to find a data directory: {}", .{e});
-        const path = option orelse exitln("no valid data directory could be found", .{});
-        ensureDirectoriesExist(path) catch |e| exitln("failed to create directories under \"{s}\": {}", .{ path, e });
+        const option = result catch |e| exitln("interrupted while determining the data folder: {}", .{e});
+        const path = option orelse exitln("no valid data folder found", .{});
+        ensureDataDirsExist(path) catch |e| exitln("failed to create folders under \"{s}\": {}", .{ path, e });
 
         break :blk path;
     };
     defer allocator.free(data_path);
 
+    // Determine the Config Directory
+    const config_path = blk: {
+        const result = known_folders.getPath(allocator, .roaming_configuration);
+        const option = result catch |e| exitln("interreupted while determining the config folder: {}", .{e});
+        const path = option orelse exitln("no valid config folder found", .{});
+        ensureConfigDirExists(path) catch |e| exitln("failed to create required folder \"{s}\": {}", .{ path, e });
+
+        break :blk path;
+    };
+    defer allocator.free(config_path);
+
     // Parse CLI
     const result = clap.parse(clap.Help, &params, clap.parsers.default, .{}) catch |e| exitln("failed to parse cli: {}", .{e});
     defer result.deinit();
 
     // TODO: Move config file to XDG Config directory?
-    const config_path = configFilePath(allocator, data_path) catch |e| exitln("failed to determine the config file path for ZBA: {}", .{e});
-    defer allocator.free(config_path);
+    const cfg_file_path = configFilePath(allocator, config_path) catch |e| exitln("failed to ready config file for access: {}", .{e});
+    defer allocator.free(cfg_file_path);
 
-    config.load(allocator, config_path) catch |e| exitln("failed to read config file: {}", .{e});
+    config.load(allocator, cfg_file_path) catch |e| exitln("failed to load config file: {}", .{e});
 
     const paths = handleArguments(allocator, data_path, &result) catch |e| exitln("failed to handle cli arguments: {}", .{e});
     defer if (paths.save) |path| allocator.free(path);
@@ -99,8 +110,8 @@ pub fn handleArguments(allocator: Allocator, data_path: []const u8, result: *con
     };
 }
 
-fn configFilePath(allocator: Allocator, data_path: []const u8) ![]const u8 {
-    const path = try std.fs.path.join(allocator, &[_][]const u8{ data_path, "zba", "config.toml" });
+fn configFilePath(allocator: Allocator, config_path: []const u8) ![]const u8 {
+    const path = try std.fs.path.join(allocator, &[_][]const u8{ config_path, "zba", "config.toml" });
     errdefer allocator.free(path);
 
     // We try to create the file exclusively, meaning that we err out if the file already exists.
@@ -109,7 +120,7 @@ fn configFilePath(allocator: Allocator, data_path: []const u8) ![]const u8 {
     std.fs.accessAbsolute(path, .{}) catch |e| {
         if (e != error.FileNotFound) return e;
 
-        const config_file = try std.fs.createFileAbsolute(path, .{});
+        const config_file = std.fs.createFileAbsolute(path, .{}) catch |err| exitln("failed to create \"{s}\": {}", .{ path, err });
         defer config_file.close();
 
         try config_file.writeAll(@embedFile("../example.toml"));
@@ -118,17 +129,21 @@ fn configFilePath(allocator: Allocator, data_path: []const u8) ![]const u8 {
     return path;
 }
 
-fn ensureDirectoriesExist(data_path: []const u8) !void {
+fn ensureDataDirsExist(data_path: []const u8) !void {
     var dir = try std.fs.openDirAbsolute(data_path, .{});
     defer dir.close();
 
-    // We want to make sure: %APPDATA%/zba and %APPDATA%/zba/save exist
-    // (~/.local/share/zba/save for linux, ??? for macOS)
-
     // Will recursively create directories
     try dir.makePath("zba" ++ std.fs.path.sep_str ++ "save");
 }
 
+fn ensureConfigDirExists(config_path: []const u8) !void {
+    var dir = try std.fs.openDirAbsolute(config_path, .{});
+    defer dir.close();
+
+    try dir.makePath("zba");
+}
+
 fn romPath(result: *const clap.Result(clap.Help, &params, clap.parsers.default)) []const u8 {
     return switch (result.positionals.len) {
         1 => result.positionals[0],

src/util.zig

@@ -5,6 +5,8 @@ const config = @import("config.zig");
 const Log2Int = std.math.Log2Int;
 const Arm7tdmi = @import("core/cpu.zig").Arm7tdmi;
 
+const Allocator = std.mem.Allocator;
+
 // Sign-Extend value of type `T` to type `U`
 pub fn sext(comptime T: type, comptime U: type, value: T) T {
     // U must have less bits than T
@@ -174,6 +176,7 @@ pub const io = struct {
 
 pub const Logger = struct {
     const Self = @This();
+    const FmtArgTuple = std.meta.Tuple(&.{ u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32 });
 
     buf: std.io.BufferedWriter(4096 << 2, std.fs.File.Writer),
 
@@ -232,8 +235,6 @@ pub const Logger = struct {
     }
 };
 
-const FmtArgTuple = std.meta.Tuple(&.{ u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32, u32 });
-
 pub const audio = struct {
     const _io = @import("core/bus/io.zig");
 
@@ -326,3 +327,44 @@ fn HalfInt(comptime T: type) type {
 
     return std.meta.Int(type_info.Int.signedness, type_info.Int.bits >> 1);
 }
+
+/// Double Buffering Implementation
+pub const FrameBuffer = struct {
+    const Self = @This();
+
+    layers: [2][]u8,
+    buf: []u8,
+    current: u1,
+
+    allocator: Allocator,
+
+    // TODO: Rename
+    const Device = enum { Emulator, Renderer };
+
+    pub fn init(allocator: Allocator, comptime len: comptime_int) !Self {
+        const buf = try allocator.alloc(u8, len * 2);
+        std.mem.set(u8, buf, 0);
+
+        return .{
+            // Front and Back Framebuffers
+            .layers = [_][]u8{ buf[0..][0..len], buf[len..][0..len] },
+            .buf = buf,
+            .current = 0,
+
+            .allocator = allocator,
+        };
+    }
+
+    pub fn deinit(self: *Self) void {
+        self.allocator.free(self.buf);
+        self.* = undefined;
+    }
+
+    pub fn swap(self: *Self) void {
+        self.current = ~self.current;
+    }
+
+    pub fn get(self: *Self, comptime dev: Device) []u8 {
+        return self.layers[if (dev == .Emulator) self.current else ~self.current];
+    }
+};