Minor progress on gb_ppu_render_continue

include/gb.h

@@ -25,7 +25,7 @@ typedef struct gb_timers_t {
 } GBTimers;
 
 GBTimers *gb_timers_open(RIO *io);
-void gb_timers_update(GBTimers *timers, ut32 cycles);
+void gb_timers_continue(GBTimers *timers, ut32 cycles);
 void gb_timers_close(GBTimers *timers, RIO *io);
 
 typedef struct gb_joypad_t {
@@ -44,7 +44,7 @@ typedef struct gb_joypad_t {
 } GBJoypad;
 
 GBJoypad *gb_joypad_open(RIO *io);
-void gb_joypad_update(GBJoypad *joypad);
+void gb_joypad_continue(GBJoypad *joypad);
 void gb_joypad_close(GBJoypad *joypad, RIO *io);
 
 typedef struct gb_dma_t {
@@ -69,7 +69,7 @@ typedef struct gb_dma_t {
 
 GBDMA *gb_dma_open(RIO *io);
 //void gb_dma_enable_cgb(GBDMA *dma);
-void gb_dma_update(GBDMA *dma, RIO *io, ut32 cycles, bool pre_exec);
+void gb_dma_continue(GBDMA *dma, RIO *io, ut32 cycles, bool pre_exec);
 void gb_dma_close(GBDMA *dma, RIO *io);
 
 enum {
@@ -121,6 +121,8 @@ typedef struct pixel_fifo_fetcher_t {
 #define	GB_PIXEL_FIFO_FETCH_OBJECT	0x10
 #define	GB_PIXEL_FIFO_FETCH_WINDOW	0x20
 #define	GB_PIXEL_FIFO_FETCH_READY	0x40
+//specifies if scx & 0x7 and wy are copied into fifo
+#define GB_PIXEL_FIFO_REG_DATA_LOADED	0x80
 
 #define	GB_OAM_PALLET	0x10
 #define	GB_OAM_FLIP_X	0x20
@@ -132,6 +134,7 @@ typedef struct pixel_fifo_fetcher_t {
 #define	GB_FIFO_OAM_FLIP_Y	(GB_OAM_FLIP_Y << 24)
 #define	GB_FIFO_OAM_PRIORITY	(GB_OAM_PRIORITY << 24)
 
+
 typedef struct pixel_fifo_t {
 	ut64 data;
 	PixelFifoFetcher fetcher[2];
@@ -149,8 +152,9 @@ obj p1	-	11
 		ut8 flags;
 	};
 	ut8 n_fpixel;	//number of pixel that are currently in the fifo
-	ut8 wy;
 	ut8 x;
+	ut8 wy;
+	ut8 dx;		//discard x
 } PixelFifo;
 
 typedef struct gb_dmg_ppu_t {
@@ -178,7 +182,7 @@ typedef struct gameboy_t {
 } GB;
 
 GBPPU *gb_ppu_open (RIO *io, SDL_Renderer *renderer);
-void gb_ppu_update (GB *gb, ut32 cycles);
+void gb_ppu_continue (GB *gb, ut32 cycles);
 void gb_ppu_close (GBPPU *ppu, RIO *io);
 
 extern RIOPlugin r_io_plugin_gb_timers;

io/dma.c

@@ -221,7 +221,7 @@ GBDMA *gb_dma_open (RIO *io) {
 	return dma;
 }
 
-void gb_dma_update (GBDMA *dma, RIO *io, ut32 cycles, bool pre_exec) {
+void gb_dma_continue (GBDMA *dma, RIO *io, ut32 cycles, bool pre_exec) {
 	if (!(dma->seek & GB_DMA_ACTIVE)) {
 		return;
 	}

io/joypad.c

@@ -108,7 +108,7 @@ GBJoypad *gb_joypad_open (RIO *io) {
 	return joypad;
 }
 
-void gb_joypad_update(GBJoypad *joypad) {
+void gb_joypad_continue(GBJoypad *joypad) {
 	joypad->odata &= 0xc0;
 	joypad->odata |= joypad->data & 0x3f;
 	SDL_PumpEvents ();

io/ppu.c

@@ -124,7 +124,7 @@ GBPPU *gb_ppu_open (RIO *io, SDL_Renderer *renderer) {
 	return ppu;
 }
 
-static ut32 gb_ppu_oam_scan_update (GB *gb, ut32 cycles) {
+static ut32 gb_ppu_oam_scan_continue (GB *gb, ut32 cycles) {
 	if (gb->ppu->ost.addr >= 0xa0) {
 		gb->ppu->ost.addr = 0;
 		gb->ppu->ost.n_entries = 0;
@@ -279,6 +279,9 @@ static void gb_pixel_fifo_fetch_continue (GB *gb) {
 			}
 		}
 	case 5:
+		if (!fetch_obj) {
+			fifo->flags |= GB_PIXEL_FIFO_FETCH_READY;
+		}
 		break;
 	case 6:
 		if (fetch_obj) {
@@ -311,37 +314,74 @@ static void gb_pixel_fifo_fetch_continue (GB *gb) {
 		}
 		break;
 	case 7:
-		fifo->flags |= GB_PIXEL_FIFO_FETCH_READY;
+		if (fetch_obj) {
+			fifo->flags |= GB_PIXEL_FIFO_FETCH_READY;
+		}
 	}
 	fifo->fetcher[fetch_obj].state_ctr = (fifo->fetcher[fetch_obj].state_ctr + 1) & 0x7;
 }
 
-static ut32 gb_ppu_render_update (GB *gb, ut32 cycles) {
+static void gb_ppu_render_launch (GB *gb) {
+	RIOMap *vram = r_io_map_get (gb->io, gb->ppu->vram_mapid);
+	vram->perm = 0;	//disable vram access for rendering
+	gb->ppu->fifo = (const PixelFifo){0};
+	gb->ppu->fifo.remaining_cycles = 376;
+}
+
+static ut32 gb_ppu_render_continue (GB *gb, ut32 cycles) {
+	PixelFifo *fifo = &gb->ppu->fifo;
+	if (R_UNLIKELY (!(fifo->flags & GB_PIXEL_FIFO_REG_DATA_LOADED))) {
+		fifo->dx = gb->ppu->buf[GB_PPU_SCX] & 0x7;
+		fifo->wy = gb->ppu->buf[GB_PPU_WY];
+		fifo->flags |= GB_PIXEL_FIFO_REG_DATA_LOADED;
+	}
+	while (cycles) {
+		if (fifo->n_fpixel < 9) {
+			gb_pixel_fifo_fetch_continue (gb);
+			if (fifo->flags & GB_PIXEL_FIFO_FETCH_READY) {
+				if (!fifo->n_fpixel) {
+					fifo->data = ((ut64)fifo->fetcher[0].fetched) << 32;
+				} else {
+					fifo->data |= fifo->fetcher[0].fetched;
+				}
+				fifo->n_fpixel += 8;
+				//I suspect the initial 16 pixels fetch on each line only takes 12 cycles instead of 16,
+				//because the pixel fifo is empty
+				fifo->flags ^= GB_PIXEL_FIFO_FETCH_READY;
+				fifo->fetcher[0].state_ctr = 0;
+			}
+		}
+		fifo->remaining_cycles--;
+		cycles--;
+	}
 	return 0;
 }
 
-static ut32 gb_ppu_hblank_update (GB *gb, ut32 cycles) {
+static ut32 gb_ppu_hblank_continue (GB *gb, ut32 cycles) {
 	return 0;
 }
 
-static ut32 gb_ppu_vblank_update (GB *gb, ut32 cycles) {
+static ut32 gb_ppu_vblank_continue (GB *gb, ut32 cycles) {
 	return 0;
 }
 
-void gb_ppu_update (GB *gb, ut32 cycles) {
+void gb_ppu_continue (GB *gb, ut32 cycles) {
 	while (cycles) {
 		switch (gb->ppu->buf[GB_PPU_STAT] & GB_PPU_STAT_MODE_MASK) {
 		case GB_PPU_STAT_MODE_OAM_SCAN:
-			cycles = gb_ppu_oam_scan_update (gb, cycles);
+			cycles = gb_ppu_oam_scan_continue (gb, cycles);
+			if ((gb->ppu->buf[GB_PPU_STAT] & GB_PPU_STAT_MODE_MASK) == GB_PPU_STAT_MODE_RENDER) {
+				gb_ppu_render_launch (gb);
+			}
 			break;
 		case GB_PPU_STAT_MODE_RENDER:
-			cycles = gb_ppu_render_update (gb, cycles);
+			cycles = gb_ppu_render_continue (gb, cycles);
 			break;
 		case GB_PPU_STAT_MODE_HBLANK:
-			cycles = gb_ppu_hblank_update (gb, cycles);
+			cycles = gb_ppu_hblank_continue (gb, cycles);
 			break;
 		case GB_PPU_STAT_MODE_VBLANK:
-			cycles = gb_ppu_vblank_update (gb, cycles);
+			cycles = gb_ppu_vblank_continue (gb, cycles);
 			break;
 		}
 	}

io/timers.c

@@ -130,7 +130,7 @@ void gb_timers_close (GBTimers *timers, RIO *io) {
 	free (timers);
 }
 
-void gb_timers_update (GBTimers *timers, ut32 cycles) {
+void gb_timers_continue (GBTimers *timers, ut32 cycles) {
 	bool request_interrupt = false;
 	do {
 		//ensure all falling edges are detected