From 43e75276dfb805a9a405c1ff2b14ca69022ee568 Mon Sep 17 00:00:00 2001
From: Karol Stasiak <karol.m.stasiak@gmail.com>
Date: Mon, 18 Mar 2019 23:15:05 +0100
Subject: [PATCH] Preliminary and experimental Game Boy support

---
 .gitignore                  |   1 +
 examples/README.md          |   4 +
 examples/gb/gbtest.mfk      | 250 ++++++++++++++++++++++++++++++++++++
 include/gb_hardware.mfk     | 171 ++++++++++++++++++++++++
 include/gb_header_small.mfk |  22 ++++
 include/gb_joy.mfk          |  47 +++++++
 include/gb_small.ini        |  34 +++++
 7 files changed, 529 insertions(+)
 create mode 100644 examples/gb/gbtest.mfk
 create mode 100644 include/gb_hardware.mfk
 create mode 100644 include/gb_header_small.mfk
 create mode 100644 include/gb_joy.mfk
 create mode 100644 include/gb_small.ini

diff --git a/.gitignore b/.gitignore
index 551c892e..5f059d8c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -36,6 +36,7 @@ examples/lunix/
 *.tap
 *.d88
 *.com
+*.gb
 HELLO
 HELLOCPC
 
diff --git a/examples/README.md b/examples/README.md
index fe63dc0d..c37f10ec 100644
--- a/examples/README.md
+++ b/examples/README.md
@@ -39,6 +39,10 @@ how to create a program made of multiple files loaded on demand
 
 * [MMC4 example](nes/nestest_mmc4.mfk) – the same thing as above, but uses a MMC4 mapper just to test bankswitching
 
+## Game Boy examples
+
+* [GB test example](gb/gbtest.mfk) – a partial port of the NES example, with a rudimentary experimental text output implementation
+
 ## Atari 2600 examples
 
 * [Colors](vcs/colors.mfk) – simple static rasterbars
diff --git a/examples/gb/gbtest.mfk b/examples/gb/gbtest.mfk
new file mode 100644
index 00000000..95dd71b0
--- /dev/null
+++ b/examples/gb/gbtest.mfk
@@ -0,0 +1,250 @@
+// compile with -t gb_small
+
+import gb_hardware
+import gb_joy
+import stdio
+
+array oam_buffer[$a0] align(256)
+
+sbyte sprite_direction
+byte old_a
+
+void on_vblank(){
+    flush_putchar_buffer()
+    sprite_dma(oam_buffer.addr.hi)
+    oam_buffer[1] += sprite_direction
+    react_to_input()
+    if oam_buffer[1] == 160 { sprite_direction = -1 }
+    if oam_buffer[1] == 8 { sprite_direction = 1 }
+}
+void on_lcdc(){
+
+}
+void on_timer(){
+
+}
+void on_serial(){
+
+}
+void on_joypad(){
+
+}
+
+void react_to_input() {
+    read_joy()
+    if old_a == 0 && input_a != 0 { // A button
+        sprite_direction = 0 - sprite_direction
+        low_c()
+    }
+    old_a = input_a
+    if input_dy < 0 { // Up button
+        if oam_buffer[0] > 16 { oam_buffer[0] -= 1}
+    }
+    if input_dy > 0 { // Down button
+        if oam_buffer[0] < 152 { oam_buffer[0] += 1}
+    }
+}
+
+void low_c() {
+}
+
+void high_c() {
+}
+
+
+void main() {
+    byte i
+    for i,0,paralleluntil,$a0 {
+        oam_buffer[i] = 0
+    }
+    reg_obj0_palette = 0q2222
+    load_charset()
+    init_putchar()
+    reg_lcd_ctrl = lcd_tile_8000 | lcd_bg_on | lcd_obj_on | lcd_win_off | lcd_map_9800
+    enable_lcd()
+    enable_irq()
+    putstrz("HELLO WORLD"z)
+    new_line()
+    sprite_direction = 1
+    oam_buffer[0] = 50
+    oam_buffer[1] = 8
+    oam_buffer[2] = '@'
+    oam_buffer[3] = 0
+    while(true){}
+}
+
+void load_charset() {
+    pointer s
+    pointer d
+    s = charset.addr
+    d = $8200
+    while s < charset.addr + $300 {
+        d[0] = s[0]
+        d += 1
+        d[0] = s[0]
+        d += 1
+        s += 1
+    }
+}
+
+array charset = [
+		$00,$00,$00,$00,$00,$00,$00,$00,
+		$18,$18,$18,$18,$00,$00,$18,$00,
+		$66,$66,$66,$00,$00,$00,$00,$00,
+		$66,$66,$FF,$66,$FF,$66,$66,$00,
+		$18,$3E,$60,$3C,$06,$7C,$18,$00,
+		$62,$66,$0C,$18,$30,$66,$46,$00,
+		$3C,$66,$3C,$38,$67,$66,$3F,$00,
+		$06,$0C,$18,$00,$00,$00,$00,$00,
+		$0C,$18,$30,$30,$30,$18,$0C,$00,
+		$30,$18,$0C,$0C,$0C,$18,$30,$00,
+		$00,$66,$3C,$FF,$3C,$66,$00,$00,
+		$00,$18,$18,$7E,$18,$18,$00,$00,
+		$00,$00,$00,$00,$00,$18,$18,$30,
+		$00,$00,$00,$7E,$00,$00,$00,$00,
+		$00,$00,$00,$00,$00,$18,$18,$00,
+		$00,$03,$06,$0C,$18,$30,$60,$00,
+		$3C,$66,$6E,$76,$66,$66,$3C,$00,
+		$18,$18,$38,$18,$18,$18,$7E,$00,
+		$3C,$66,$06,$0C,$30,$60,$7E,$00,
+		$3C,$66,$06,$1C,$06,$66,$3C,$00,
+		$06,$0E,$1E,$66,$7F,$06,$06,$00,
+		$7E,$60,$7C,$06,$06,$66,$3C,$00,
+		$3C,$66,$60,$7C,$66,$66,$3C,$00,
+		$7E,$66,$0C,$18,$18,$18,$18,$00,
+		$3C,$66,$66,$3C,$66,$66,$3C,$00,
+		$3C,$66,$66,$3E,$06,$66,$3C,$00,
+		$00,$00,$18,$00,$00,$18,$00,$00,
+		$00,$00,$18,$00,$00,$18,$18,$30,
+		$0E,$18,$30,$60,$30,$18,$0E,$00,
+		$00,$00,$7E,$00,$7E,$00,$00,$00,
+		$70,$18,$0C,$06,$0C,$18,$70,$00,
+		$3C,$66,$06,$0C,$18,$00,$18,$00,
+		$3C,$66,$6E,$6E,$60,$62,$3C,$00,
+		$18,$3C,$66,$7E,$66,$66,$66,$00,
+		$7C,$66,$66,$7C,$66,$66,$7C,$00,
+		$3C,$66,$60,$60,$60,$66,$3C,$00,
+		$78,$6C,$66,$66,$66,$6C,$78,$00,
+		$7E,$60,$60,$78,$60,$60,$7E,$00,
+		$7E,$60,$60,$78,$60,$60,$60,$00,
+		$3C,$66,$60,$6E,$66,$66,$3C,$00,
+		$66,$66,$66,$7E,$66,$66,$66,$00,
+		$3C,$18,$18,$18,$18,$18,$3C,$00,
+		$1E,$0C,$0C,$0C,$0C,$6C,$38,$00,
+		$66,$6C,$78,$70,$78,$6C,$66,$00,
+		$60,$60,$60,$60,$60,$60,$7E,$00,
+		$63,$77,$7F,$6B,$63,$63,$63,$00,
+		$66,$76,$7E,$7E,$6E,$66,$66,$00,
+		$3C,$66,$66,$66,$66,$66,$3C,$00,
+		$7C,$66,$66,$7C,$60,$60,$60,$00,
+		$3C,$66,$66,$66,$66,$3C,$0E,$00,
+		$7C,$66,$66,$7C,$78,$6C,$66,$00,
+		$3C,$66,$60,$3C,$06,$66,$3C,$00,
+		$7E,$18,$18,$18,$18,$18,$18,$00,
+		$66,$66,$66,$66,$66,$66,$3C,$00,
+		$66,$66,$66,$66,$66,$3C,$18,$00,
+		$63,$63,$63,$6B,$7F,$77,$63,$00,
+		$66,$66,$3C,$18,$3C,$66,$66,$00,
+		$66,$66,$66,$3C,$18,$18,$18,$00,
+		$7E,$06,$0C,$18,$30,$60,$7E,$00,
+		$3C,$30,$30,$30,$30,$30,$3C,$00,
+		$0C,$12,$30,$7C,$30,$62,$FC,$00,
+		$3C,$0C,$0C,$0C,$0C,$0C,$3C,$00,
+		$00,$18,$3C,$7E,$18,$18,$18,$18,
+		$00,$10,$30,$7F,$7F,$30,$10,$00,$FF,$FF,$FF,$FF,$FF,$FF,$FF,$FF
+]
+
+
+
+
+// the following is an experimental implementation of text output for Game Boy
+// if it matures, it might get mainlined into the standard distribution
+
+array putchar_buffer[256]
+byte putchar_buffer_start
+byte putchar_buffer_size
+byte putchar_column
+byte putchar_row
+volatile byte putchar_lock
+
+void init_putchar() {
+    putchar_buffer_start = 0
+    putchar_buffer_size = 0
+    putchar_column = 0
+    putchar_row = 0
+    putchar_lock = 0
+}
+
+asm void putchar(byte e) {
+    ld a,(putchar_buffer_size)
+    cp $ff
+    jr nz, __putchar_do
+    halt
+    jr putchar
+    __putchar_do:
+    ld b,a
+    ld a,1
+    ld (putchar_lock),a
+    ld a,(putchar_buffer_start)
+    add a,b
+    ld c,a
+    ld b,0
+    ld hl,putchar_buffer.addr
+    add hl,bc
+    ld (hl),e
+    ld hl,putchar_buffer_size.addr
+    inc (hl)
+    xor a
+    ld (putchar_lock),a
+    ret
+}
+
+
+// TODO: ???
+noinline pointer calculate_cursor() {
+    pointer p
+    p = putchar_column
+    p <<= 5
+    p += putchar_row
+    p += $9800
+    return p
+}
+
+void new_line() {
+    putchar(13)
+    putchar(10)
+}
+
+void flush_putchar_buffer() {
+    pointer dest
+    pointer src
+    if putchar_lock != 0 { return }
+    if putchar_buffer_size == 0 { return }
+    src = putchar_buffer.addr + putchar_buffer_start
+    dest = calculate_cursor()
+    byte i
+    byte c
+    for i,0,paralleluntil,putchar_buffer_size {
+        c = src[0]
+        src += 1
+        if c == 13 {
+            putchar_column = 0
+            dest.lo &= $E0
+        } else if c == 10 {
+            putchar_row += 1
+            dest += 32
+        } else {
+            dest[0] = c
+            dest += 1
+            putchar_column += 1
+            if putchar_column == 20 {
+                putchar_column = 0
+                putchar_row += 1
+                dest.lo &= $E0
+                dest += 32
+            }
+        }
+    }
+    putchar_buffer_start += putchar_buffer_size
+    putchar_buffer_size = 0
+}
diff --git a/include/gb_hardware.mfk b/include/gb_hardware.mfk
new file mode 100644
index 00000000..d7aa4319
--- /dev/null
+++ b/include/gb_hardware.mfk
@@ -0,0 +1,171 @@
+#if not(GAMEBOY)
+#warn gb_hardware module should be only used on NES/Famicom targets
+#endif
+
+#pragma zilog_syntax
+
+// TODO: optimize?
+array __vectors @ $40 = [
+    $c3, __on_vblank.addr.lo, __on_vblank.addr.hi, 0,0,0,0,0,
+    $c3, __on_lcdc.addr.lo, __on_lcdc.addr.hi, 0,0,0,0,0,
+    $c3, __on_timer.addr.lo, __on_timer.addr.hi, 0,0,0,0,0,
+    $c3, __on_serial.addr.lo, __on_serial.addr.hi, 0,0,0,0,0,
+    $c3, __on_joypad.addr.lo, __on_joypad.addr.hi, 0,0,0,0,0
+]
+
+
+asm void __start() @ $150 {
+    di
+
+    ld sp, $DFFF
+    xor a
+    ldh (reg_irq_flag), a
+    ? ld a,ief_vblank
+    ldh (reg_irq_enable), a
+
+
+    ? call disable_lcd
+    ? call clear_ram
+
+    // copy the DMA routine to hi RAM
+    ld hl, __sprite_dma_template
+    ld de, __sprite_dma_actual
+    ld c, __sprite_dma_actual.length
+    __copy_sprite_dma_template:
+    ld a,(hli)
+    ld (de),a
+    inc de
+    dec c
+    jr nz, __copy_sprite_dma_template
+
+
+    ? jp main
+    ? jp __start
+}
+
+asm void clear_ram() {
+    ? xor a
+    ? ld hl,$c000
+    ? ld bc,$1ff0
+    ? call memset
+    ? ld hl,$8000
+    ? ld bc,$2000
+    ? call memset
+    ? ld hl,$fe00
+    ? ld bc,$a0
+    ? call memset
+    ? ld hl,$ff80
+    ? ld bc,$7f
+    ? call memset
+    ? ret
+}
+
+inline asm void memset(word hl, word bc) {
+    ? inc c
+    ? inc b
+    ? jr __memzero_start
+    __memzero_loop:
+    ld (hli), a
+    __memzero_start:
+    dec c
+    jr nz, __memzero_loop
+    dec b
+    jr nz, __memzero_loop
+    ? ret
+}
+
+interrupt void __on_vblank(){
+    on_vblank()
+}
+interrupt void __on_lcdc(){
+    on_lcdc()
+}
+interrupt void __on_timer(){
+    on_timer()
+}
+interrupt void __on_serial(){
+    on_serial()
+}
+interrupt void __on_joypad(){
+    on_joypad()
+}
+
+array oam[$a0] @$fe00
+
+volatile byte reg_joypad @$ff00
+volatile byte reg_irq_flag @$ff0f
+volatile byte reg_lcd_ctrl @$ff40
+volatile byte reg_lcd_status @$ff41
+volatile byte reg_scroll_y @$ff42
+volatile byte reg_scroll_x @$ff43
+volatile byte reg_dma @$ff46
+volatile byte reg_bg_palette @$ff47
+volatile byte reg_obj0_palette @$ff48
+volatile byte reg_obj1_palette @$ff49
+volatile byte reg_window_y @$ff4A
+volatile byte reg_window_x @$ff4B
+volatile byte reg_vram_bank @$ff4f
+volatile byte reg_irq_enable @$ffff
+
+const byte lcd_on = $80
+const byte lcd_off = 0
+const byte lcd_win_9c00 = $40
+const byte lcd_win_9800 = 0
+const byte lcd_win_on = $20
+const byte lcd_win_off = 0
+const byte lcd_tile_8000 = $10
+const byte lcd_tile_8800 = 0
+const byte lcd_map_9c00 = 8
+const byte lcd_map_9800 = 0
+const byte lcd_obj_16 = 4
+const byte lcd_obj_8 = 0
+const byte lcd_obj_on = 2
+const byte lcd_obj_off = 0
+const byte lcd_bg_on = 1
+const byte lcd_bg_off = 1
+
+const byte ief_vblank = 1
+
+segment(hiram)
+array __sprite_dma_actual[8]
+
+asm void __sprite_dma_template() {
+    ldh (reg_dma),a
+    ld a, $28
+__sprite_dma_wait:
+    dec a
+    jr nz, __sprite_dma_wait
+    ret
+    __sprite_dma_end:
+}
+
+asm void sprite_dma(byte a) {
+    jp __sprite_dma_actual
+}
+
+asm byte wait_for_vblank() {
+    ld hl, reg_irq_flag
+    __wait_for_vblank_loop:
+    ld a, ief_vblank
+    halt
+    and (hl)
+    jr z, __wait_for_vblank_loop
+    ? ld a, $ff ^ ief_vblank
+    ? and (hl)
+    ? ret
+}
+
+asm void disable_lcd() {
+    ? call wait_for_vblank
+    ldh a,(reg_lcd_ctrl)
+    res 7,a
+    ldh (reg_lcd_ctrl),a
+    ? ret
+}
+asm void enable_lcd() {
+    ? call wait_for_vblank
+    ldh a,(reg_lcd_ctrl)
+    set 7,a
+    ldh (reg_lcd_ctrl),a
+    ? ret
+}
\ No newline at end of file
diff --git a/include/gb_header_small.mfk b/include/gb_header_small.mfk
new file mode 100644
index 00000000..62fb0752
--- /dev/null
+++ b/include/gb_header_small.mfk
@@ -0,0 +1,22 @@
+#if not(GAMEBOY)
+#warn gb_header_small module should be only used on NES/Famicom targets
+#endif
+
+#pragma zilog_syntax
+
+array __header @ $100 = [
+    $00, $C3, $50, $01, $CE, $ED, $66, $66,   $CC, $0D, $00, $0B, $03, $73, $00, $83,
+    $00, $0C, $00, $0D, $00, $08, $11, $1F,   $88, $89, $00, $0E, $DC, $CC, $6E, $E6,
+    $DD, $DD, $D9, $99, $BB, $BB, $67, $63,   $6E, $0E, $EC, $CC, $DD, $DC, $99, $9F,
+    $BB, $B9, $33, $3E,
+    "PROGRAM" ascii, 0,0,0,0,0,0,0,0,
+    0,0, // no new license code
+    0, // no SGB support
+    0, // ROM only
+    0, // 32 kB
+    0, // no RAM
+    0, // Japan
+    0, // unlicensed
+    0,0, // checksum, will patch later
+    0,0 // global checksum, will patch later
+]
\ No newline at end of file
diff --git a/include/gb_joy.mfk b/include/gb_joy.mfk
new file mode 100644
index 00000000..163cd1bd
--- /dev/null
+++ b/include/gb_joy.mfk
@@ -0,0 +1,47 @@
+#if not(GAMEBOY)
+#warn gb_joy module should be only used on NES/Famicom targets
+#endif
+
+import gb_hardware
+
+#pragma zilog_syntax
+
+// standard joystick driver for Game Boy
+
+import joy
+
+alias input_a = input_btn
+byte input_b
+byte input_select
+byte input_start
+
+void read_joy() {
+    byte tmp
+    reg_joypad = $20
+    asm {
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+    }
+    tmp = reg_joypad
+    input_dx = 0
+    input_dy = 0
+    if tmp & 1 == 0 { input_dx += 1 }
+    if tmp & 2 == 0 { input_dx -= 1 }
+    if tmp & 4 == 0 { input_dy -= 1 }
+    if tmp & 8 == 0 { input_dy += 1 }
+    reg_joypad = $10
+    asm {
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+        ld a,(reg_joypad)
+    }
+    byte tmp
+    tmp = reg_joypad ^ $ff
+    input_a = (tmp & 1) >> 0
+    input_b = (tmp & 2) >> 1
+    input_select = (tmp & 4) >> 2
+    input_start = (tmp & 8) >> 3
+}
\ No newline at end of file
diff --git a/include/gb_small.ini b/include/gb_small.ini
new file mode 100644
index 00000000..8d9bb012
--- /dev/null
+++ b/include/gb_small.ini
@@ -0,0 +1,34 @@
+[compilation]
+arch=lr35902
+encoding=ascii
+screen_encoding=ascii
+modules=default_panic,stdlib,gb_hardware,gb_header_small
+
+
+[allocation]
+
+segments=default,rom,hiram
+default_code_segment=rom
+
+segment_default_start=$c000
+segment_default_end=$dfff
+
+segment_rom_start=$0150
+segment_rom_end=$7fff
+segment_hiram_start=$ff80
+segment_hiram_end=$fffe
+
+[define]
+GAMEBOY=1
+WIDESCREEN=0
+KEYBOARD=0
+JOYSTICKS=1
+HAS_BITMAP_MODE=0
+
+[output]
+style=single
+format=rom:0:$7fff
+gb_checksum=true
+extension=gb
+
+