s/rgb_frame/fb/, add oam_cache, fix several instructions

3 files changed, 111 insertions, 74 deletions

diff --git a/gb.h b/gb.h
index 15bd75e..e027ee7 100644
--- a/gb.h
+++ b/gb.h
@@ -25,8 +25,8 @@ typedef enum {
 #define GB_BTN_SELECT (1 << 6)
 #define GB_BTN_START  (1 << 7)
 
-// frame size, in bytes
-#define GB_FRAME_SIZE (3 * 160 * 144)
+// frame buffer size, in bytes
+#define GB_FB_SIZE (3 * 160 * 144)
 
 typedef struct gb_t_ gb_t;
 
@@ -116,8 +116,22 @@ struct gb_t_ {
             obp1,
             line,
             lyc;
-    uint8_t rgb[GB_FRAME_SIZE];
+
+    // frame buffer
+    uint8_t fb[GB_FB_SIZE];
+
+    // frame counter
     uint32_t frame;
+
+    struct {
+      uint8_t x, y, tile;
+      bool priority,
+           y_flip,
+           x_flip,
+           palette;
+    } oam_cache[40];
+
+    bool oam_dirty;
   } gpu;
 
   struct {
@@ -148,7 +162,7 @@ void gb_set_buttons(gb_t * const, const uint8_t);
 void gb_frame(gb_t * const);
 void gb_step(gb_t * const);
 _Bool gb_disasm(gb_t * const, char * const, size_t * const);
-const uint8_t *gb_get_rgb_frame(const gb_t * const);
+const uint8_t *gb_get_frame(const gb_t * const);
 
 #ifdef __cplusplus
 };
diff --git a/ops.yaml b/ops.yaml
index 61db5ee..6bf9870 100644
--- a/ops.yaml
+++ b/ops.yaml
@@ -1890,7 +1890,7 @@ ops:
         h:
         c:
       code: |
-        mmu_wb(ctx, cpu_rw(ctx, RW_HL), cpu_rb(ctx, RB_H));
+        mmu_wb(ctx, cpu_rw(ctx, RW_HL), cpu_rb(ctx, RB_L));
     - id: HALT
       hex: 0x76
       cat: misc
@@ -2022,7 +2022,7 @@ ops:
       cat: "8-bit load/store/move"
       op: LD
       dst: A
-      src: L
+      src: (HL)
       len: 1
       pc: true
       time: 8
@@ -2144,7 +2144,7 @@ ops:
         h: H
         c: C
       code: |
-        add_rb(ctx, RB_H);
+        add_rb(ctx, RB_L);
     - id: ADD A, (HL)
       hex: 0x86
       cat: "8-bit math"
@@ -3067,6 +3067,7 @@ ops:
         h:
         c:
       code: |
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         if (!FLAG(ctx, Z)) {
           pop_rw(ctx, RW_PC);
           clock = 20;
@@ -3197,7 +3198,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x00);
     - id: RET Z
       hex: 0xC8
@@ -3340,7 +3341,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x08);
     - id: RET NC
       hex: 0xD0
@@ -3429,7 +3430,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 3);
+        cpu_ww(ctx, RW_PC, old_pc + 3);
         if (!FLAG(ctx, C)) {
           call_a16(ctx, old_pc + 1);
           clock = 24;
@@ -3481,7 +3482,7 @@ ops:
         h: H
         c: C
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x10);
     - id: RET C
       hex: 0xD8
@@ -3521,8 +3522,8 @@ ops:
         c:
       code: |
         // pop pc, enable interrupts
-        pop_rw(ctx, RW_PC);
         ctx->cpu.ime = true;
+        pop_rw(ctx, RW_PC);
     - id: JP C, a16
       hex: 0xDA
       cat: "jumps/calls"
@@ -3612,7 +3613,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x18);
     - id: LDH (a8), A
       hex: 0xE0
@@ -3666,6 +3667,8 @@ ops:
       hex: 0xE3
       cat: "invalid"
       op: XX
+      code: |
+        cpu_set_state(ctx, old_pc, GB_CPU_STATE_INVALID);
     - id: XX
       hex: 0xE4
       cat: "invalid"
@@ -3718,7 +3721,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x20);
     - id: ADD SP, r8
       hex: 0xE8
@@ -3815,7 +3818,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x28);
     - id: LDH A, (a8)
       hex: 0xF0
@@ -3931,7 +3934,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x30);
     - id: LD HL, SP+r8
       hex: 0xF8
@@ -4039,7 +4042,7 @@ ops:
         h:
         c:
       code: |
-        cpu_ww(ctx, RW_PC, cpu_rw(ctx, RW_PC) + 1);
+        cpu_ww(ctx, RW_PC, old_pc + 1);
         rst(ctx, 0x38);
 
   cb:
@@ -8561,6 +8564,7 @@ templates:
     static void gpu_set_line(gb_t * const ctx, const uint8_t line);
     static void gpu_set_mode(gb_t * const ctx, const uint8_t mode);
     static bool gpu_is_active(const gb_t * const ctx);
+    static uint8_t gpu_get_mode(const gb_t * const ctx);
 
     static uint8_t
     mmu_rp(
@@ -8820,7 +8824,7 @@ templates:
       const gb_t * const ctx,
       const uint16_t addr
     ) {
-      switch (ctx->gpu.mode & 0x3) {
+      switch (gpu_get_mode(ctx)) {
       case GPU_MODE_VRAM:
         return 0xFF;
       default:
@@ -8834,7 +8838,7 @@ templates:
       const gb_t * const ctx,
       const uint16_t addr
     ) {
-      switch (ctx->gpu.mode & 0x3) {
+      switch (gpu_get_mode(ctx)) {
       case GPU_MODE_HBLANK:
       case GPU_MODE_VBLANK:
         // oam memory (160 bytes):
@@ -8954,7 +8958,7 @@ templates:
       const uint16_t addr,
       const uint8_t val
     ) {
-      switch (ctx->gpu.mode & 0x3) {
+      switch (gpu_get_mode(ctx)) {
       case GPU_MODE_HBLANK:
       case GPU_MODE_VBLANK:
       case GPU_MODE_OAM:
@@ -8973,13 +8977,14 @@ templates:
       const uint16_t addr,
       const uint8_t val
     ) {
-      switch (ctx->gpu.mode & 0x3) {
+      switch (gpu_get_mode(ctx) & 0x3) {
       case GPU_MODE_HBLANK:
       case GPU_MODE_VBLANK:
         // oam memory (160 bytes):
         ctx->mmu.oam[addr & 0xFF] = val;
 
-        // TODO: invalidate OAM cache
+        // invalidate oam cache
+        ctx->gpu.oam_dirty = true;
 
         break;
       default:
@@ -10115,7 +10120,7 @@ templates:
     gpu_get_mode(
       const gb_t * const ctx
     ) {
-      return ctx->gpu.mode & 0x3;
+      return gpu_is_active(ctx) ? (ctx->gpu.mode & 0x3) : GPU_MODE_VBLANK;
     }
 
     static void
@@ -10166,86 +10171,94 @@ templates:
     }
 
     static void
-    gpu_draw_line(
+    gpu_oam_cache(
       gb_t * const ctx
     ) {
-      struct {
-        uint8_t x, y, tile;
-        bool priority,
-             y_flip,
-             x_flip,
-             palette;
-      } objs[40];
+      if (!ctx->gpu.oam_dirty) {
+        return;
+      }
 
       // decode oam
-      for (int i = 0; i < 40; i++) {
-        objs[i].y = (ctx->mmu.oam[i * 4 + 0] + 16) & 0xFF;
-        objs[i].x = (ctx->mmu.oam[i * 4 + 1] + 8) & 0xFF;
-        objs[i].tile = ctx->mmu.oam[i * 4 + 2];
+      for (uint8_t i = 0; i < 40; i++) {
+        ctx->gpu.oam_cache[i].y = (ctx->mmu.oam[i * 4 + 0] + 16) & 0xFF;
+        ctx->gpu.oam_cache[i].x = (ctx->mmu.oam[i * 4 + 1] + 8) & 0xFF;
+        ctx->gpu.oam_cache[i].tile = ctx->mmu.oam[i * 4 + 2];
 
         const uint8_t flags = ctx->mmu.oam[i * 4 + 3];
-        objs[i].priority = flags & (1 << 7);
-        objs[i].y_flip = flags & (1 << 6);
-        objs[i].x_flip = flags & (1 << 5);
-        objs[i].palette = flags & (1 << 4);
+        ctx->gpu.oam_cache[i].priority = flags & (1 << 7);
+        ctx->gpu.oam_cache[i].y_flip = flags & (1 << 6);
+        ctx->gpu.oam_cache[i].x_flip = flags & (1 << 5);
+        ctx->gpu.oam_cache[i].palette = flags & (1 << 4);
       }
 
+      // clear dirty flag
+      ctx->gpu.oam_dirty = false;
+    }
+
+    static void
+    gpu_draw_line(
+      gb_t * const ctx
+    ) {
+      gpu_oam_cache(ctx);
+
       // draw sprites behind bg
-      for (int i = 0; i < 40; i++) {
+      for (uint8_t i = 0; i < 40; i++) {
         if (
           // sprite is below bg
-          objs[i].priority &&
+          ctx->gpu.oam_cache[i].priority &&
 
           // sprite is on screen
-          objs[i].y < 144 && objs[i].x < 160 &&
+          ctx->gpu.oam_cache[i].y < 144 &&
+          ctx->gpu.oam_cache[i].x < 160 &&
 
           // sprite is on current line
-          ctx->gpu.line >= objs[i].y &&
-          ctx->gpu.line <= objs[i].y
+          ctx->gpu.line >= ctx->gpu.oam_cache[i].y &&
+          ctx->gpu.line <= ctx->gpu.oam_cache[i].y
         ) {
-          const uint8_t max_j = MIN(8, 160 - objs[i].x);
+          const uint8_t max_j = MIN(8, 160 - ctx->gpu.oam_cache[i].x);
 
           for (uint8_t j = 0; j < max_j; j++) {
-            const uint32_t ofs = 3 * (objs[i].y * 160 + objs[i].x + j);
+            const uint32_t ofs = 3 * (ctx->gpu.oam_cache[i].y * 160 + ctx->gpu.oam_cache[i].x + j);
 
             // TODO: draw sprite scanline
-            ctx->gpu.rgb[ofs + 0] = 0xff;
-            ctx->gpu.rgb[ofs + 1] = 0xff;
-            ctx->gpu.rgb[ofs + 2] = 0xff;
+            ctx->gpu.fb[ofs + 0] = 0xff;
+            ctx->gpu.fb[ofs + 1] = 0xff;
+            ctx->gpu.fb[ofs + 2] = 0xff;
           }
         }
       }
 
       // TODO: draw background
-      for (int i = 0; i < 160; i++) {
+      for (uint8_t i = 0; i < 160; i++) {
         // interim: set line to red
-        ctx->gpu.rgb[3 * (ctx->gpu.line * 160 + i) + 0] = 0xff;
-        ctx->gpu.rgb[3 * (ctx->gpu.line * 160 + i) + 1] = 0x00;
-        ctx->gpu.rgb[3 * (ctx->gpu.line * 160 + i) + 2] = 0x00;
+        ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 0] = 0xff;
+        ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 1] = 0x00;
+        ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 2] = 0x00;
       }
 
       // draw foreground sprites
-      for (int i = 0; i < 40; i++) {
+      for (uint8_t i = 0; i < 40; i++) {
         if (
           // sprite is above bg
-          !objs[i].priority &&
+          !ctx->gpu.oam_cache[i].priority &&
 
           // sprite is on screen
-          objs[i].y < 144 && objs[i].x < 160 &&
+          ctx->gpu.oam_cache[i].y < 144 &&
+          ctx->gpu.oam_cache[i].x < 160 &&
 
           // sprite is on current line
-          ctx->gpu.line >= objs[i].y &&
-          ctx->gpu.line <= objs[i].y
+          ctx->gpu.line >= ctx->gpu.oam_cache[i].y &&
+          ctx->gpu.line <= ctx->gpu.oam_cache[i].y
         ) {
-          const uint8_t max_j = MIN(8, 160 - objs[i].x);
+          const uint8_t max_j = MIN(8, 160 - ctx->gpu.oam_cache[i].x);
 
           for (int j = 0; j < max_j; j++) {
-            const uint32_t ofs = 3 * (objs[i].y * 160 + objs[i].x + j);
+            const uint32_t ofs = 3 * (ctx->gpu.oam_cache[i].y * 160 + ctx->gpu.oam_cache[i].x + j);
 
             // TODO: draw sprite scanline
-            ctx->gpu.rgb[ofs + 0] = 0xff;
-            ctx->gpu.rgb[ofs + 1] = 0xff;
-            ctx->gpu.rgb[ofs + 2] = 0xff;
+            ctx->gpu.fb[ofs + 0] = 0xff;
+            ctx->gpu.fb[ofs + 1] = 0xff;
+            ctx->gpu.fb[ofs + 2] = 0xff;
           }
         }
       }
@@ -10286,7 +10299,7 @@ templates:
         break;
       case GPU_MODE_HBLANK:
         if (ctx->gpu.clock >= 204) {
-          // clear clock, draw line, increment line
+          // clear clock
           ctx->gpu.clock -= 204;
 
           // draw and increment line
@@ -10565,10 +10578,10 @@ templates:
     }
 
     const uint8_t *
-    gb_get_rgb_frame(
+    gb_get_frame(
       const gb_t * const ctx
     ) {
-      return ctx->gpu.rgb;
+      return ctx->gpu.fb;
     }
 
     static void
@@ -10589,6 +10602,21 @@ templates:
       cpu_set_state(ctx, 0, GB_CPU_STATE_RUN);
     }
 
+    static void
+    gpu_init(
+      gb_t * const ctx
+    ) {
+      gpu_set_mode(ctx, GPU_MODE_OAM);
+      gpu_set_line(ctx, 0);
+      ctx->gpu.lcdc = 0x91;
+      ctx->gpu.bgp = 0xFC;
+      ctx->gpu.obp0 = 0xFF;
+      ctx->gpu.obp1 = 0xFF;
+
+      // flag oam as dirty
+      ctx->gpu.oam_dirty = true;
+    }
+
     void gb_init(
       gb_t * const ctx,
       const gb_config_t * const config,
@@ -10609,10 +10637,5 @@ templates:
       ctx->mmu.rom_size = rom_size;
 
       // init gpu
-      gpu_set_mode(ctx, GPU_MODE_OAM);
-      gpu_set_line(ctx, 0);
-      ctx->gpu.lcdc = 0x91;
-      ctx->gpu.bgp = 0xFC;
-      ctx->gpu.obp0 = 0xFF;
-      ctx->gpu.obp1 = 0xFF;
+      gpu_init(ctx);
     }
diff --git a/test.c b/test.c
index 5900cf7..e2c66f0 100644
--- a/test.c
+++ b/test.c
@@ -61,7 +61,7 @@ load(
 }
 
 // buffer for rendered frames
-static uint8_t frames[NUM_FRAMES * GB_FRAME_SIZE];
+static uint8_t frames[NUM_FRAMES * GB_FB_SIZE];
 
 static void
 on_set_rom_bank(
@@ -193,7 +193,7 @@ test_render_frames(
     gb_frame(&ctx);
 
     // copy frame to buffer
-    memcpy(frames + GB_FRAME_SIZE * j, gb_get_rgb_frame(&ctx), GB_FRAME_SIZE);
+    memcpy(frames + GB_FB_SIZE * j, gb_get_frame(&ctx), GB_FB_SIZE);
   }
 
   // save png