more gpu updates, add SKIP_FRAMES

2 files changed, 120 insertions, 96 deletions

diff --git a/ops.yaml b/ops.yaml
index 29fe351..26c1a87 100644
--- a/ops.yaml
+++ b/ops.yaml
@@ -8089,12 +8089,14 @@ templates:
     #define UNUSED(a) ((void) (a))
     #define MIN(a, b) (((a) < (b)) ? (a) : (b))
 
+    // cpu flags
     #define FLAG(ctx, f) (cpu_rb(ctx, RB_F) & (F_##f))
     #define F_Z (1 << 7)
     #define F_N (1 << 6)
     #define F_H (1 << 5)
     #define F_C (1 << 4)
 
+    // interrupts
     #define IRQ_VBLANK  (1)
     #define IRQ_LCDC    (1 << 1)
     #define IRQ_TIMER   (1 << 2)
@@ -8830,10 +8832,9 @@ templates:
       const gb_t * const ctx,
       const uint16_t addr
     ) {
-      switch (gpu_get_mode(ctx)) {
-      case GPU_MODE_VRAM:
+      if (gpu_is_active(ctx) && gpu_get_mode(ctx) == GPU_MODE_VRAM) {
         return 0xFF;
-      default:
+      } else {
         // return vram (8k)
         return ctx->mmu.vram[addr & 0x1FFF];
       }
@@ -8844,13 +8845,18 @@ templates:
       const gb_t * const ctx,
       const uint16_t addr
     ) {
-      switch (gpu_get_mode(ctx)) {
-      case GPU_MODE_HBLANK:
-      case GPU_MODE_VBLANK:
+      if (gpu_is_active(ctx)) {
+        switch (gpu_get_mode(ctx)) {
+        case GPU_MODE_HBLANK:
+        case GPU_MODE_VBLANK:
+          // oam memory (160 bytes):
+          return ctx->mmu.oam[addr & 0xFF];
+        default:
+          return 0;
+        }
+      } else {
         // oam memory (160 bytes):
         return ctx->mmu.oam[addr & 0xFF];
-      default:
-        return 0;
       }
     }
 
@@ -8964,16 +8970,21 @@ templates:
       const uint16_t addr,
       const uint8_t val
     ) {
-      switch (gpu_get_mode(ctx)) {
-      case GPU_MODE_HBLANK:
-      case GPU_MODE_VBLANK:
-      case GPU_MODE_OAM:
+      if (gpu_is_active(ctx)) {
+        switch (gpu_get_mode(ctx)) {
+        case GPU_MODE_HBLANK:
+        case GPU_MODE_VBLANK:
+        case GPU_MODE_OAM:
+          // vram (8k)
+          ctx->mmu.vram[addr & 0x1FFF] = val;
+          break;
+        default:
+          // do nothing
+          break;
+        }
+      } else {
         // vram (8k)
         ctx->mmu.vram[addr & 0x1FFF] = val;
-        break;
-      default:
-        // do nothing
-        break;
       }
     }
 
@@ -8983,19 +8994,21 @@ templates:
       const uint16_t addr,
       const uint8_t val
     ) {
-      switch (gpu_get_mode(ctx) & 0x3) {
-      case GPU_MODE_HBLANK:
-      case GPU_MODE_VBLANK:
-        // oam memory (160 bytes):
-        ctx->mmu.oam[addr & 0xFF] = val;
+      if (gpu_is_active(ctx)) {
+        switch (gpu_get_mode(ctx) & 0x3) {
+        case GPU_MODE_HBLANK:
+        case GPU_MODE_VBLANK:
+          // oam memory (160 bytes):
+          ctx->mmu.oam[addr & 0xFF] = val;
 
-        // invalidate oam cache
-        ctx->gpu.oam_dirty = true;
+          // invalidate oam cache
+          ctx->gpu.oam_dirty = true;
 
-        break;
-      default:
-        // do nothing
-        break;
+          break;
+        default:
+          // do nothing
+          break;
+        }
       }
     }
 
@@ -10180,7 +10193,7 @@ templates:
         // trigger vblank interrupt
         cpu_irq(ctx, IRQ_VBLANK);
 
-        if (ctx->cpu.ime && (mmu_rb(ctx, PORT_IE) & 1)) {
+        if (gpu_is_active(ctx) && ctx->cpu.ime && (mmu_rb(ctx, PORT_IE) & 1)) {
           if (ctx->config && ctx->config->on_vblank) {
             ctx->config->on_vblank(ctx);
           }
@@ -10200,11 +10213,11 @@ templates:
 
       // decode oam
       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];
+        ctx->gpu.oam_cache[i].y = (ctx->mmu.oam[4 * i + 0] + 16) & 0xFF;
+        ctx->gpu.oam_cache[i].x = (ctx->mmu.oam[4 * i + 1] + 8) & 0xFF;
+        ctx->gpu.oam_cache[i].tile = ctx->mmu.oam[4 * i + 2];
 
-        const uint8_t flags = ctx->mmu.oam[i * 4 + 3];
+        const uint8_t flags = ctx->mmu.oam[4 * i + 3];
         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);
@@ -10219,68 +10232,77 @@ templates:
     gpu_draw_line(
       gb_t * const ctx
     ) {
-      gpu_oam_cache(ctx);
+      if (gpu_is_active(ctx)) {
+        gpu_oam_cache(ctx);
 
-      // draw sprites behind bg
-      for (uint8_t i = 0; i < 40; i++) {
-        if (
-          // sprite is below bg
-          ctx->gpu.oam_cache[i].priority &&
-
-          // sprite is on screen
-          ctx->gpu.oam_cache[i].y < 144 &&
-          ctx->gpu.oam_cache[i].x < 160 &&
-
-          // sprite is on current line
-          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 - ctx->gpu.oam_cache[i].x);
-
-          for (uint8_t j = 0; j < max_j; 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.fb[ofs + 0] = 0xff;
-            ctx->gpu.fb[ofs + 1] = 0xff;
-            ctx->gpu.fb[ofs + 2] = 0xff;
+        // draw sprites behind bg
+        for (uint8_t i = 0; i < 40; i++) {
+          if (
+            // sprite is below bg
+            ctx->gpu.oam_cache[i].priority &&
+
+            // sprite is on screen
+            ctx->gpu.oam_cache[i].y < 144 &&
+            ctx->gpu.oam_cache[i].x < 160 &&
+
+            // sprite is on current line
+            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 - ctx->gpu.oam_cache[i].x);
+
+            for (uint8_t j = 0; j < max_j; 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.fb[ofs + 0] = 0xff;
+              ctx->gpu.fb[ofs + 1] = 0xff;
+              ctx->gpu.fb[ofs + 2] = 0xff;
+            }
           }
         }
-      }
 
-      // TODO: draw background
-      for (uint8_t i = 0; i < 160; i++) {
-        // interim: set line to red
-        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;
-      }
+        // TODO: draw background
+        for (uint8_t i = 0; i < 160; i++) {
+          // interim: set line to red
+          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 (uint8_t i = 0; i < 40; i++) {
-        if (
-          // sprite is above bg
-          !ctx->gpu.oam_cache[i].priority &&
-
-          // sprite is on screen
-          ctx->gpu.oam_cache[i].y < 144 &&
-          ctx->gpu.oam_cache[i].x < 160 &&
-
-          // sprite is on current line
-          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 - ctx->gpu.oam_cache[i].x);
-
-          for (int j = 0; j < max_j; 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.fb[ofs + 0] = 0xff;
-            ctx->gpu.fb[ofs + 1] = 0xff;
-            ctx->gpu.fb[ofs + 2] = 0xff;
+        // draw foreground sprites
+        for (uint8_t i = 0; i < 40; i++) {
+          if (
+            // sprite is above bg
+            !ctx->gpu.oam_cache[i].priority &&
+
+            // sprite is on screen
+            ctx->gpu.oam_cache[i].y < 144 &&
+            ctx->gpu.oam_cache[i].x < 160 &&
+
+            // sprite is on current line
+            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 - ctx->gpu.oam_cache[i].x);
+
+            for (int j = 0; j < max_j; 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.fb[ofs + 0] = 0xff;
+              ctx->gpu.fb[ofs + 1] = 0xff;
+              ctx->gpu.fb[ofs + 2] = 0xff;
+            }
           }
         }
+      } else {
+        // write blank line
+        for (uint8_t i = 0; i < 160; i++) {
+          ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 0] = 0xff;
+          ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 1] = 0xff;
+          ctx->gpu.fb[3 * (ctx->gpu.line * 160 + i) + 2] = 0xff;
+        }
       }
     }
 
@@ -10289,10 +10311,6 @@ templates:
       gb_t * const ctx,
       const uint16_t clock
     ) {
-      if (!gpu_is_active(ctx)) {
-        return;
-      }
-
       // increment clock
       ctx->gpu.clock += clock;
 
@@ -10326,7 +10344,7 @@ templates:
           gpu_draw_line(ctx);
           gpu_set_line(ctx, ctx->gpu.line + 1);
 
-          if (ctx->gpu.line < 144) {
+          if (ctx->gpu.line < 143) {
             // set mode
             gpu_set_mode(ctx, GPU_MODE_OAM);
           } else {
@@ -10344,7 +10362,7 @@ templates:
           // clear clock
           ctx->gpu.clock -= 204;
 
-          if (ctx->gpu.line < 154) {
+          if (ctx->gpu.line < 153) {
             // increment line
             gpu_set_line(ctx, ctx->gpu.line + 1);
           } else {
diff --git a/test.c b/test.c
index 929bdd5..c74bf28 100644
--- a/test.c
+++ b/test.c
@@ -9,7 +9,8 @@
 #include "lodepng.h"
 #include "gb.h"
 
-#define NUM_FRAMES 600
+#define SKIP_FRAMES 600
+#define NUM_FRAMES 60
 // #define SKIP_STEPS 1000000
 #define NUM_STEPS 300000
 #define UNUSED(a) ((void) (a))
@@ -187,13 +188,18 @@ test_render_frames(
   gb_init(&ctx, NULL, rom_data, rom_size);
   fprintf(stderr, "gb context initialized\n");
 
+  for (size_t i = 0; i < SKIP_FRAMES; i++) {
+    // render frame
+    gb_frame(&ctx);
+  }
+
   // render frames
-  for (size_t j = 0; j < NUM_FRAMES; j++) {
+  for (size_t i = 0; i < NUM_FRAMES; i++) {
     // render frame
     gb_frame(&ctx);
 
     // copy frame to buffer
-    memcpy(frames + GB_FB_SIZE * j, gb_get_frame(&ctx), GB_FB_SIZE);
+    memcpy(frames + GB_FB_SIZE * i, gb_get_frame(&ctx), GB_FB_SIZE);
   }
 
   // save png