src/core/bus/io.zig

@@ -428,6 +428,8 @@ const u8WriteKind = enum { Hi, Lo };
 
 /// Write-only
 pub const WinH = extern union {
+    const Self = @This();
+
     x2: Bitfield(u16, 0, 8),
     x1: Bitfield(u16, 8, 8),
     raw: u16,
@@ -440,6 +442,13 @@ pub const WinV = extern union {
     y2: Bitfield(u16, 0, 8),
     y1: Bitfield(u16, 8, 8),
     raw: u16,
+
+    pub fn set(self: *Self, comptime K: u8WriteKind, value: u8) void {
+        self.raw = switch (K) {
+            .Hi => (@as(u16, value) << 8) | self.raw & 0xFF,
+            .Lo => (self.raw & 0xFF00) | value,
+        };
+    }
 };
 
 pub const WinIn = extern union {

src/core/ppu.zig

@@ -9,6 +9,7 @@ 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 EventKind = @import("scheduler.zig").EventKind;
 const Scheduler = @import("scheduler.zig").Scheduler;
 const Arm7tdmi = @import("cpu.zig").Arm7tdmi;
 const FrameBuffer = @import("../util.zig").FrameBuffer;
@@ -19,6 +20,7 @@ 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;
@@ -263,7 +265,8 @@ pub const Ppu = struct {
     scanline: Scanline,
 
     pub fn init(allocator: Allocator, sched: *Scheduler) !Self {
-        sched.push(.Draw, 240 * 4); // Add first PPU Event to Scheduler
+        // Queue first Hblank
+        sched.push(.Draw, 240 * 4);
 
         const sprites = try allocator.create([128]?Sprite);
         std.mem.set(?Sprite, sprites, null);
@@ -323,16 +326,20 @@ 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
-                var y_pos: i32 = attr0.y.read();
+                const iy = @bitCast(i8, y);
-                if (y_pos >= 160) y_pos -= 256; // fleroviux's solution to negative positions
+
+                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);
 
                 // 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 (y_pos <= y and y < (y_pos + sprite_height)) {
+                if ((start <= y and y < end) or (istart <= iy and iy < iend)) {
                     for (self.scanline_sprites) |*maybe_sprite| {
                         if (maybe_sprite.* == null) {
                             maybe_sprite.* = Sprite.init(attr0, attr1, @bitCast(Attr2, self.oam.read(u16, i + 4)));
@@ -359,6 +366,8 @@ 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();
@@ -367,22 +376,25 @@ 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;
 
-            var x_pos: i32 = sprite.x();
+            const sprite_start = sprite.x();
-            if (x_pos >= 240) x_pos -= 512;
+            const isprite_start = @bitCast(i9, sprite_start);
+            const sprite_end = sprite_start +% sprite.width;
+            const isprite_end = @bitCast(i9, sprite_end);
 
-            if (!(x_pos <= x and x < (x_pos + sprite.width))) continue;
+            const condition = (sprite_start <= x and x < sprite_end) or (isprite_start <= ix and ix < isprite_end);
+            if (!condition) continue;
 
             // Sprite is within bounds and therefore should be rendered
             // std.math.absInt is branchless
-            const tile_x = @bitCast(u32, @as(i32, std.math.absInt(@as(i32, x) - x_pos) catch unreachable));
+            const tile_x = @bitCast(u9, std.math.absInt(ix - @bitCast(i9, sprite.x())) catch unreachable);
-            const tile_y = @bitCast(u32, @as(i32, std.math.absInt(@bitCast(i8, y) -% @bitCast(i8, sprite.y())) catch unreachable));
+            const tile_y = @bitCast(u8, std.math.absInt(iy -% @bitCast(i8, sprite.y())) catch unreachable);
 
             const row = @truncate(u3, tile_y);
             const col = @truncate(u3, tile_x);
@@ -404,6 +416,8 @@ 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();
@@ -412,27 +426,31 @@ 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;
 
-            var x_pos: i32 = sprite.x();
+            const sprite_start = sprite.x();
-            if (x_pos >= 240) x_pos -= 512;
+            const isprite_start = @bitCast(i9, sprite_start);
+            const sprite_end = sprite_start +% sprite.width;
+            const isprite_end = @bitCast(i9, sprite_end);
 
-            if (!(x_pos <= x and x < (x_pos + sprite.width))) continue;
+            const condition = (sprite_start <= x and x < sprite_end) or (isprite_start <= ix and ix < isprite_end);
+            if (!condition) continue;
 
             // Sprite is within bounds and therefore should be rendered
-            const x_diff: i32 = std.math.absInt(@as(i32, x) - x_pos) catch unreachable;
+            // std.math.absInt is branchless
-            const y_diff: i32 = std.math.absInt(@bitCast(i8, y) -% @bitCast(i8, sprite.y())) catch unreachable;
+            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);
 
             // 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_x = @intCast(u9, x_diff) ^ if (sprite.hFlip()) (sprite.width - 1) else 0;
+            const tile_y = y_diff ^ if (sprite.vFlip()) (sprite.height - 1) else 0;
-            const tile_y = @intCast(u8, y_diff) ^ if (sprite.vFlip()) (sprite.height - 1) else 0;
+            const tile_x = x_diff ^ if (sprite.hFlip()) (sprite.width - 1) else 0;
 
             const row = @truncate(u3, tile_y);
             const col = @truncate(u3, tile_x);
@@ -962,35 +980,24 @@ const Window = struct {
     }
 
     fn inRange(self: *const Self, comptime id: u1, x: u9, y: u8) bool {
-        const winh = self.h[id];
+        const h = self.h[id];
-        const winv = self.v[id];
+        const v = self.v[id];
 
-        if (isYInRange(winv, y)) {
+        const y1 = v.y1.read();
-            const x1 = winh.x1.read();
+        const y2 = if (y1 > v.y2.read()) 160 else std.math.min(160, v.y2.read());
-            const x2 = winh.x2.read();
+
+        if (y1 <= y and y < y2) {
+            // Within Y bounds
+            const x1 = h.x1.read();
+            const x2 = if (x1 > h.x2.read()) 240 else std.math.min(240, h.x2.read());
 
             // Within X Bounds
-            return if (x1 < x2) blk: {
+            return x1 <= x and x < x2;
-                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);