From 19126df7f4b69ca8001b6d2ca6b2c703b7625195 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jakub=20K=C4=85dzio=C5=82ka?= <kuba@kadziolka.net>
Date: Sun, 3 May 2020 22:41:56 +0200
Subject: [PATCH 01/10] Save 8 bytes in the CGB boot ROM

---
 BootROMs/cgb_boot.asm | 20 +++++++-------------
 1 file changed, 7 insertions(+), 13 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 6ae869b2..618e11a4 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -984,14 +984,13 @@ BrightenColor:
     and $1F
     cp $1F
     jr nz, .blueNotMaxed
-    res 0, c
+    dec c
 .blueNotMaxed
 
     ; Is green maxed?
     ld a, e
-    and $E0
     cp $E0
-    jr nz, .greenNotMaxed
+    jr c, .greenNotMaxed
     ld a, d
     and $3
     cp $3
@@ -1007,18 +1006,13 @@ BrightenColor:
     res 2, b
 .redNotMaxed
 
-    ; Add de to bc
-    push hl
-    ld h, d
-    ld l, e
-    add hl, bc
-    ld d, h
-    ld e, l
-    pop hl
-
+    ; add de, bc
+    ; ld [hli], de
     ld a, e
+    add c
     ld [hli], a
     ld a, d
+    adc b
     ld [hli], a
     ret
 
@@ -1155,7 +1149,7 @@ ReplaceColorInAllPalettes:
     dec c
     jr nz, .loop
     ret
-    
+
 LoadDMGTilemap:
     push af
     call WaitFrame

From cb738190be6abca03d1cd3265440908107168a47 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jakub=20K=C4=85dzio=C5=82ka?= <kuba@kadziolka.net>
Date: Sun, 3 May 2020 22:44:13 +0200
Subject: [PATCH 02/10] Add a 2bpp CGB boot ROM logo, pending palettes

---
 BootROMs/SameBoyLogo.png | Bin 14763 -> 479 bytes
 BootROMs/cgb_boot.asm    | 116 ++++++++++++++++++++++++++-------------
 BootROMs/pb12.py         |  68 +++++++++++++++++++++++
 Makefile                 |  13 ++---
 4 files changed, 152 insertions(+), 45 deletions(-)
 create mode 100644 BootROMs/pb12.py

diff --git a/BootROMs/SameBoyLogo.png b/BootROMs/SameBoyLogo.png
index 4bc9706080a1d5e161d506f2d2be6d9c8e1446d4..c7cfc087dbf7a26a2969fc463d116b9e28eb58ac 100644
GIT binary patch
delta 465
zcmV;?0WSWlbKe7y7=Ho-0000_WgT_^000b7OjJex|Nnn~fBygg000170)~4400E9k
zL_t(2&yA8XZo@DPMHzU|sHMA(LKiteI>gABqv&c1I&`dpB6F{x05ZBr`2CFIphJcl
zW-MEe<mXpJ<VBud`*q<_WN0!Vi^B=9$RZIrRB_!_O`HLR?0+i(3%;WHCcHOX+0%f4
zEjKMj&mX2KtK*wH{RSwueVQ0ZiX+}3F<JG*9RiizD8>XKQ3lQsjN*%!4hpQ&S3sG}
z^6VxHz^3k}8BhsS(Jj@7f)Dc4Cz~FVo7>Rf2rx_^Q}w%=4I_(c^4B(0xOv*U`OvMd
zuWE)kbmkB!&3|3H<Is$<03J!q5pHVsrTcUD<`10=1{4pG#Z!Lp1>v*;eF4m9Dn!u~
zO32V{^YirP=K~A~#fBD}H?nKUF%Iq4y9kbaYQ6(34paiPDNr^e8o<%(0$d4jOWOm$
zj=5L>FZ$J|Bc1pQc=KLvfL`!fZ&qn_HC(2BxEwLy5=#35u>6jP-U$>M#%2~9leL~x
z*UVWtYJLF4^3nnX&}Ra14CW{k86!XU`$=!6=YQ&Hl>c47@!eI<`B+HV00000NkvXX
Hu0mjf?upU~

literal 14763
zcmeI3Z)_7~9LEp;L>R#cCPQJ-vp}L?dw1P+TaUJ6w1ZCQoTJDl5V+kv-7ahIj=S61
zO_L2wTtG2om>^;#z5o=>e1nl_l!QP+{1Y%lvWyUvC<!EH5ZH?mqCR)++NWK|1BRDR
zllFdlp6~bfJokBi&$D}TJDZzU&#72YK~dD4#%QDkUI*a!nHlhhTA-ha!pkf@+GbKz
z<r4clqp@Z24T`E;ro>yV*4SD>(o!B#*4lw5lhSP$6<CqcMX3u|bUWx!)SzqZyAv*2
zk%O)cz8D+R!=O`%_8DMZUsGJ_>yrGkYelFckP%>l6tF}(lS-<lkO{h+yaJ5vX2wN3
z600lds<#KETVu_1STg|a^Z48n>*eTLzlZbrYHPhO(_WV28J1(XYB%Q>cwS&xI`?sf
zDqs{a<b==?S(O_O{uOj}T9z&_Ogf$Rq^mue(ZO(jzn@{fjMwXi3b)y-T4Khnnv1eY
zCi6sqDH)1xDVj>#dBt|E+X}i|_CWbh?!HobejwG%v4a$ujHol5hh@r{WGU~_yN#qX
zHCbXn5~P4?nb60T^+BYuSl(K8xKygF)3h3T;2q?Mly#c%UL7zkz|^`82{iP;nJg+d
zdA(D!G_zBi=DzdY)84(MBnpdDG!?sNS{ukH)7b2utj#Hf><R@SEFo+F(b9~#rX@q!
zDr=scJsl1^RYkuT6D38pZ|S7~Q)()#cp)E%h!zOJimY+7)o!*f?qvmz6F7bu>xFfR
zB*`jEa>QWu%c3RTPnIHWX&y~h62148rJzJ|$dZuIjFf1Fl$6*37+viMFoi}$>6D$&
zsG63jN}w?kf(;%;kp(Fs*Cu!%x@&ljck@1$b&C?uyA$m-{yMP+NIt2~!4Zq0_9>|&
zn$&GqelE5AfT<Dtl+>>p3Ou#MB%-lr|DZ#ZNi$`rsA9re>qTl1R-$t*tpnSNTPHh`
zigUi`qG1Ai`vqN-`!9bW<<A@2<7{I=l<bo!B-zCc<N#CDT9{VenX@VXcLB5c5&QIG
zq&tCnudR_y<*24lVrmI1EgGP{1J=a-?b~eLVq@t50meSdlS=VYWLt|@XtHLfu4o{C
zya{0~sTna%1|hiR%QY9q57-Br0BZ=|kLau&7htBemFipVT1a+h^*qnQ55w8NIWO8@
z>TSb#(f(3z!Ma{mtdKW1TuC$Hvd`Zvj&<-+0fcINzG`29DQzxwG%Fd9Y>g;z=bLst
zb6$8@IHuB`vd*@G)61hqq$6j{t}*go-6?@=D|}{vk1I_6d8N2I@_XnZ2ZcL&$cfQ(
zWDt}FHYT{Rh)@B+h0?&r1Q!+&Dj>K}8rYcN!XiQi1Q$vJ8xveuM5ut^LTO-Qf(wfX
z6%bq~4QxzsVG*GMf(xaAjR`I+B2++dp){~D!G%SH3J5Ng1~w+Ru!v9r!G+Sm#sn7@
z5h@_KP#V~n;KCw81q2sL0~-@uSVX9R;6iC&V}c8d2o(@qC=F~(aA6Ul0)h*rfsF|+
zEFx4uaG^A?F~NmJgbD~Qlm<2?xUh&&0l|gRz{Ug@77;2SxKJ9{nBc-9LIngDN&_3;
z6IVsy4M3p6*Z$J*{k}Iw7w&=Y2GUZrB}P$w@Lj@zw<&7kUwFMnQQa&>UEfMk!iN+!
zUmMzTdKJ{(X^hmzGozc%p4k0#ebwokXYXF{E~GCF9sKjqqe};_ZU10k-ppA~y?4qN
zd1Bep<MS#vu={p@IsWAK*vWS<9~@hAseg3K_+Zb`vA3SRb7%Zmm1kf7Ctpl#i~cIa
zcRhdf@@watzCX12`U@Y6zja+|Ixw&Qxo_T?uO!WD(qkVj3IB8M@QClnnO~1I9BDqc
zYvN*+Ieg&o>i)fZq`s#8JI9X?ogR5z*!0Mf_cwl)ur4faY3mu=I6C;s&nthpddy#4
zIeYNp`MYzMpW6M$@R9Ycx0jE6KKtg~TjJwu+?Q8h*{7WJ4d1%W&;9Gn#NR)yT<}X|
q|J65BRq~ZvM=n-=HTda`G&Q(G{(bH7&EG-_RAWO^WO(J)9sdD_A1D9-

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 618e11a4..68376150 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -87,20 +87,24 @@ ELSE
 
 .tilemapRowLoop
 
+    call .write_with_palette
+
+    ; Repeat the 3 tiles common between E and B. This saves 27 bytes after
+    ; compression, with a cost of 17 bytes of code.
     push af
-    ; Switch to second VRAM Bank
-    ld a, 1
-    ldh [$4F], a
-    ld [hl], 8
-    ; Switch to back first VRAM Bank
-    xor a
-    ldh [$4F], a
+    sub $20
+    sub $3
+    jr nc, .notspecial
+    add $20
+    call .write_with_palette
+    dec c
+.notspecial
     pop af
-    ldi [hl], a
-    add d
+
+    add d ; d = 3 for SameBoy logo, d = 1 for Nintendo logo
     dec c
     jr nz, .tilemapRowLoop
-    sub 47
+    sub 44
     push de
     ld de, $10
     add hl, de
@@ -116,6 +120,19 @@ ELSE
     ld l, $a7
     ld bc, $0107
     jr .tilemapRowLoop
+
+.write_with_palette
+    push af
+    ; Switch to second VRAM Bank
+    ld a, 1
+    ldh [$4F], a
+    ld [hl], 8
+    ; Switch to back first VRAM Bank
+    xor a
+    ldh [$4F], a
+    pop af
+    ldi [hl], a
+    ret
 .endTilemap
 ENDC
 
@@ -532,7 +549,7 @@ TrademarkSymbol:
     db $3c,$42,$b9,$a5,$b9,$a5,$42,$3c
 
 SameBoyLogo:
-    incbin "SameBoyLogo.pb8"
+    incbin "SameBoyLogo.pb12"
 
 AnimationColors:
     dw $7FFF ; White
@@ -634,30 +651,28 @@ ReadCGBLogoHalfTile:
     ld a, e
     ret
 
-; LoadTileset using PB8 codec, 2019 Damian Yerrick
-;
-; The logo is compressed using PB8, a form of RLE with unary-coded
-; run lengths.  Each block representing 8 bytes consists of a control
-; byte, where each bit (MSB to LSB) is 0 for literal or 1 for repeat
-; previous, followed by the literals in that block.
+; LoadTileset using PB12 codec, 2020 Jakub Kądziołka
+; (based on PB8 codec, 2019 Damian Yerrick)
 
 SameBoyLogo_dst = $8080
 SameBoyLogo_length = (128 * 24) / 64
 
 LoadTileset:
     ld hl, SameBoyLogo
-    ld de, SameBoyLogo_dst
+    ld de, SameBoyLogo_dst - 1
     ld c, SameBoyLogo_length
-.pb8BlockLoop:
-    ; Register map for PB8 decompression
+.refill
+    ; Register map for PB12 decompression
     ; HL: source address in boot ROM
     ; DE: destination address in VRAM
     ; A: Current literal value
     ; B: Repeat bits, terminated by 1000...
-    ; C: Number of 8-byte blocks left in this block
     ; Source address in HL lets the repeat bits go straight to B,
     ; bypassing A and avoiding spilling registers to the stack.
     ld b, [hl]
+    dec b
+    jr z, .sameboyLogoEnd
+    inc b
     inc hl
 
     ; Shift a 1 into lower bit of shift value.  Once this bit
@@ -665,26 +680,53 @@ LoadTileset:
     scf
     rl b
 
-.pb8BitLoop:
+.loop
     ; If not a repeat, load a literal byte
-    jr c,.pb8Repeat
-    ld a, [hli]
-.pb8Repeat:
-    ; Decompressed data uses colors 0 and 1, so write once, inc twice
-    ld [de], a
-    inc de
-    inc de
+    jr c, .simple_repeat
     sla b
-    jr nz, .pb8BitLoop
-
-    dec c
-    jr nz, .pb8BlockLoop
-
-; End PB8 decoding.  The rest uses HL as the destination
-    ld h, d
-    ld l, e
+    jr c, .shifty_repeat
+    ld a, [hli]
+    jr .got_byte
+.shifty_repeat
+    sla b
+    jr nz, .no_refill_during_shift
+    ld b, [hl] ; see above. Also, no, factoring it out into a callable
+    inc hl ; routine doesn't save bytes, even with conditional calls
+    scf
+    rl b
+.no_refill_during_shift
+    ld c, a
+    jr nc, .shift_left
+    srl a
+    db $fe ; eat the add a with cp d8
+.shift_left
+    add a
+    sla b
+    jr c, .go_and
+    or c
+    db $fe ; eat the and c with cp d8
+.go_and
+    and c
+    jr .got_byte
+.simple_repeat
+    sla b
+    jr c, .got_byte
+    ; far repeat
+    dec de
+    ld a, [de]
+    inc de
+.got_byte
+    inc de
+    ld [de], a
+    sla b
+    jr nz, .loop
+    jr .refill
 
+; End PB12 decoding.  The rest uses HL as the destination
 .sameboyLogoEnd
+    ld h, d
+    ld l, $80
+
 ; Copy (unresized) ROM logo
     ld de, $104
 .CGBROMLogoLoop
diff --git a/BootROMs/pb12.py b/BootROMs/pb12.py
new file mode 100644
index 00000000..0c06538c
--- /dev/null
+++ b/BootROMs/pb12.py
@@ -0,0 +1,68 @@
+import sys
+
+def opts(byte):
+    # top bit: 0 = left, 1 = right
+    # bottom bit: 0 = or, 1 = and
+    if byte is None: return []
+    return [
+        byte | (byte << 1) & 0xff,
+        byte & (byte << 1),
+        byte | (byte >> 1) & 0xff,
+        byte & (byte >> 1),
+    ]
+
+def pb12(data):
+    data = iter(data)
+
+    literals = bytearray()
+    bits = 0
+    control = 0
+    prev = [None, None]
+    gotta_end = False
+
+    chunk = bytearray()
+    while True:
+        try:
+            byte = next(data)
+        except StopIteration:
+            if bits == 0: break
+            byte = 0
+        chunk.append(byte)
+
+        if byte in prev:
+            bits += 2
+            control <<= 1
+            control |= 1
+            control <<= 1
+            if prev[1] == byte:
+                control |= 1 # 10 = out[-2], 11 = out[-1]
+        else:
+            bits += 2
+            control <<= 2
+            options = opts(prev[1])
+            if byte in options:
+                # 01 = modify
+                control |= 1
+
+                bits += 2
+                control <<= 2
+                control |= options.index(byte)
+            else:
+                # 00 = literal
+                literals.append(byte)
+        prev = [prev[1], byte]
+        if bits >= 8:
+            outctl = control >> (bits - 8)
+            assert outctl != 1 # that's the end byte
+            yield bytes([outctl]) + literals
+            bits -= 8
+            control &= (1 << bits) - 1
+            literals = bytearray()
+            chunk = bytearray()
+    yield b'\x01'
+
+_, infile, outfile = sys.argv
+with open(infile, 'rb') as f:
+    data = f.read()
+with open(outfile, 'wb') as f:
+    f.writelines(pb12(data))
diff --git a/Makefile b/Makefile
index afc8d897..1f818628 100644
--- a/Makefile
+++ b/Makefile
@@ -382,21 +382,18 @@ $(BIN)/SDL/Shaders: Shaders
 
 # Boot ROMs
 
-$(OBJ)/%.1bpp: %.png
+$(OBJ)/%.2bpp: %.png
 	-@$(MKDIR) -p $(dir $@)
-	rgbgfx -d 1 -h -o $@ $<
+	rgbgfx -h -u -o $@ $<
 
-$(OBJ)/BootROMs/SameBoyLogo.pb8: $(OBJ)/BootROMs/SameBoyLogo.1bpp $(PB8_COMPRESS)
-	$(realpath $(PB8_COMPRESS)) -l 384 $< $@
-
-$(PB8_COMPRESS): BootROMs/pb8.c
-	$(CC) $< -o $@
+$(OBJ)/BootROMs/SameBoyLogo.pb12: $(OBJ)/BootROMs/SameBoyLogo.2bpp BootROMs/pb12.py
+	python3 BootROMs/pb12.py $< $@
 
 $(BIN)/BootROMs/agb_boot.bin: BootROMs/cgb_boot.asm
 $(BIN)/BootROMs/cgb_boot_fast.bin: BootROMs/cgb_boot.asm
 $(BIN)/BootROMs/sgb2_boot: BootROMs/sgb_boot.asm
 
-$(BIN)/BootROMs/%.bin: BootROMs/%.asm $(OBJ)/BootROMs/SameBoyLogo.pb8
+$(BIN)/BootROMs/%.bin: BootROMs/%.asm $(OBJ)/BootROMs/SameBoyLogo.pb12
 	-@$(MKDIR) -p $(dir $@)
 	rgbasm -i $(OBJ)/BootROMs/ -i BootROMs/ -o $@.tmp $<
 	rgblink -o $@.tmp2 $@.tmp

From b057e0d10af8394671e2af8d463c2f948e860f5f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jakub=20K=C4=85dzio=C5=82ka?= <kuba@kadziolka.net>
Date: Sun, 3 May 2020 23:07:53 +0200
Subject: [PATCH 03/10] Save 4 more bytes in the CGB boot ROM

---
 BootROMs/cgb_boot.asm | 52 ++++++++++++++++++++-----------------------
 1 file changed, 24 insertions(+), 28 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 68376150..1d9cc6a8 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -21,7 +21,7 @@ Start:
     ldh [InputPalette], a
 ; Clear title checksum
     ldh [TitleChecksum], a
-    
+
 ; Clear OAM
     ld h, $fe
     ld c, $a0
@@ -753,29 +753,6 @@ ReadTrademarkSymbol:
     jr nz, .loadTrademarkSymbolLoop
     ret
 
-LoadObjPalettes:
-    ld c, $6A
-    jr LoadPalettes
-
-LoadBGPalettes64:
-    ld d, 64
-
-LoadBGPalettes:
-    ld e, 0
-    ld c, $68
-
-LoadPalettes:
-    ld a, $80
-    or e
-    ld [c], a
-    inc c
-.loop
-    ld a, [hli]
-    ld [c], a
-    dec d
-    jr nz, .loop
-    ret
-
 DoIntroAnimation:
     ; Animate the intro
     ld a, 1
@@ -902,8 +879,7 @@ EmulateDMG:
     call LoadPalettesFromIndex
     ld a, 4
     ; Set the final values for DMG mode
-    ld d, 0
-    ld e, $8
+    ld de, 8
     ld l, $7c
     ret
 
@@ -997,7 +973,8 @@ LoadPalettesFromIndex: ; a = index of combination
     ld c, a
     add hl, bc
     ld d, 8
-    call LoadObjPalettes
+    ld c, $6A
+    call LoadPalettes
     pop hl
     bit 3, e
     jr nz, .loadBGPalette
@@ -1011,7 +988,26 @@ LoadPalettesFromIndex: ; a = index of combination
     ld c, a
     add hl, bc
     ld d, 8
-    jp LoadBGPalettes
+    jr LoadBGPalettes
+
+LoadBGPalettes64:
+    ld d, 64
+
+LoadBGPalettes:
+    ld e, 0
+    ld c, $68
+
+LoadPalettes:
+    ld a, $80
+    or e
+    ld [c], a
+    inc c
+.loop
+    ld a, [hli]
+    ld [c], a
+    dec d
+    jr nz, .loop
+    ret
 
 BrightenColor:
     ld a, [hli]

From 2225fd114c795a4a607fa54267a2a695c80a83d6 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Mon, 4 May 2020 02:07:19 +0300
Subject: [PATCH 04/10] Handle 2bpp palettes

---
 BootROMs/cgb_boot.asm | 74 +++++++++++++++++++------------------------
 BootROMs/pb12.py      |  2 +-
 2 files changed, 33 insertions(+), 43 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 1d9cc6a8..94087884 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -138,41 +138,24 @@ ENDC
 
     ; Expand Palettes
     ld de, AnimationColors
-    ld c, 8
+    ld c, 16
     ld hl, BgPalettes
     xor a
 .expandPalettesLoop:
-IF !DEF(FAST)
     cpl
-ENDC
-    ; One white
+    ; One white or black
     ldi [hl], a
     ldi [hl], a
-
-IF DEF(FAST)
-    ; 3 more whites
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-ELSE    
-    ; The actual color
+    
+    ; One color
+    push af
     ld a, [de]
     inc de
     ldi [hl], a
     ld a, [de]
     inc de
     ldi [hl], a
-
-    xor a
-    ; Two blacks
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-    ldi [hl], a
-ENDC
+    pop af
 
     dec c
     jr nz, .expandPalettesLoop
@@ -551,19 +534,20 @@ TrademarkSymbol:
 SameBoyLogo:
     incbin "SameBoyLogo.pb12"
 
-AnimationColors:
-    dw $7FFF ; White
-    dw $774F ; Cyan
-    dw $22C7 ; Green
-    dw $039F ; Yellow
-    dw $017D ; Orange
-    dw $241D ; Red
-    dw $6D38 ; Purple
-    dw $7102 ; Blue
-AnimationColorsEnd:
+animation_color: MACRO
+    dw ((\1) >> 1) | $4210, (\1)
+ENDM
 
-DMGPalettes:
-    dw $7FFF, $32BF, $00D0, $0000
+AnimationColors:
+    animation_color $7FFF, ($7FFF >> 1) ; White
+    animation_color $774F, ($774F >> 1) ; Cyan
+    animation_color $22C7, ($22C7 >> 1) ; Green
+    animation_color $039F, ($039F >> 1) ; Yellow
+    animation_color $017D, ($017D >> 1) ; Orange
+    animation_color $241D, ($241D >> 1) ; Red
+    animation_color $6D38, ($6D38 >> 1) ; Purple
+    animation_color $7102, ($7102 >> 1) ; Blue
+AnimationColorsEnd:
 
 ; Helper Functions
 DoubleBitsAndWriteRowTwice:
@@ -1140,27 +1124,33 @@ ChangeAnimationPalette:
 .isWhite
     push af
     ld a, [hli]
+    
     push hl
-    ld hl, BgPalettes ; First color, all palette
+    ld hl, BgPalettes ; First color, all palettes
+    call ReplaceColorInAllPalettes
+    ld l, LOW(BgPalettes + 2)  ; Second color, all palettes
     call ReplaceColorInAllPalettes
     pop hl
-    ldh [BgPalettes + 2], a ; Second color, first palette
+    ldh [BgPalettes + 6], a ; Fourth color, first palette
 
     ld a, [hli]
     push hl
-    ld hl, BgPalettes + 1 ; First color, all palette
+    ld hl, BgPalettes + 1 ; First color, all palettes
+    call ReplaceColorInAllPalettes
+    ld l, LOW(BgPalettes + 3) ; Second color, all palettes
     call ReplaceColorInAllPalettes
     pop hl
-    ldh [BgPalettes + 3], a ; Second color, first palette
+    ldh [BgPalettes + 7], a ; Fourth color, first palette
+    
     pop af
     jr z, .isNotWhite
     inc hl
     inc hl
 .isNotWhite
     ld a, [hli]
-    ldh [BgPalettes + 7 * 8 + 2], a ; Second color, 7th palette
+    ldh [BgPalettes + 7 * 8 + 6], a ; Fourth color, 7th palette
     ld a, [hli]
-    ldh [BgPalettes + 7 * 8 + 3], a ; Second color, 7th palette
+    ldh [BgPalettes + 7 * 8 + 7], a ; Fourth color, 7th palette
     ld a, [hli]
     ldh [BgPalettes + 4], a ; Third color, first palette
     ld a, [hl]
@@ -1180,7 +1170,7 @@ ChangeAnimationPalette:
 
 ReplaceColorInAllPalettes:
     ld de, 8
-    ld c, 8
+    ld c, e
 .loop
     ld [hl], a
     add hl, de
diff --git a/BootROMs/pb12.py b/BootROMs/pb12.py
index 0c06538c..dc53d6b9 100644
--- a/BootROMs/pb12.py
+++ b/BootROMs/pb12.py
@@ -63,6 +63,6 @@ def pb12(data):
 
 _, infile, outfile = sys.argv
 with open(infile, 'rb') as f:
-    data = f.read()
+    data = f.read().rstrip(b'\x00')
 with open(outfile, 'wb') as f:
     f.writelines(pb12(data))

From 72a90ba91c8b8ee07859fc25f6b536a5f5afaaad Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Mon, 4 May 2020 02:17:03 +0300
Subject: [PATCH 05/10] Hacky color blending

---
 BootROMs/cgb_boot.asm | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 94087884..cbbcf962 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -1093,12 +1093,11 @@ GetInputPaletteIndex:
     ; Slide into change Animation Palette
 
 ChangeAnimationPalette:
-    push af
     push hl
     push bc
     push de
     ld hl, KeyCombinationPalettes - 1 ; Input palettes are 1-based, 0 means nothing down
-    ld c ,a
+    ld c, a
     ld b, 0
     add hl, bc
     ld a, [hl]
@@ -1149,6 +1148,7 @@ ChangeAnimationPalette:
 .isNotWhite
     ld a, [hli]
     ldh [BgPalettes + 7 * 8 + 6], a ; Fourth color, 7th palette
+    ldh [BgPalettes + 7 * 8 + 2], a ; Second color, half, 7th palette; rough color mixing
     ld a, [hli]
     ldh [BgPalettes + 7 * 8 + 7], a ; Fourth color, 7th palette
     ld a, [hli]
@@ -1165,7 +1165,6 @@ ChangeAnimationPalette:
     pop de
     pop bc
     pop hl
-    pop af
     ret
 
 ReplaceColorInAllPalettes:

From f46f138e9fbda089a1a9f71b88d608b38b932315 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Mon, 4 May 2020 23:54:43 +0300
Subject: [PATCH 06/10] Clear VRAM correctly

---
 BootROMs/cgb_boot.asm | 155 ++++++++++++++++++++----------------------
 1 file changed, 74 insertions(+), 81 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index cbbcf962..33e18e56 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -16,6 +16,15 @@ Start:
     xor a
     call ClearMemoryPage
     ld [c], a
+    
+    ld hl, $FF30
+; Init waveform
+    ld c, $10
+.waveformLoop
+    ldi [hl], a
+    cpl
+    dec c
+    jr nz, .waveformLoop
 
 ; Clear chosen input palette
     ldh [InputPalette], a
@@ -40,8 +49,6 @@ Start:
     ld a, $77
     ldh [$24], a
     
-    call InitWaveform
-
 ; Init BG palette
     ld a, $fc
     ldh [$47], a
@@ -776,25 +783,68 @@ IF !DEF(FAST)
     ld hl, BgPalettes
 .frameLoop
     push bc
-    call BrightenColor
+    
+    ; Brighten Color
+    ld a, [hli]
+    ld e, a
+    ld a, [hld]
+    ld d, a
+    ; RGB(1,1,1)
+    ld bc, $421
+
+   ; Is blue maxed?
+    ld a, e
+    and $1F
+    cp $1F
+    jr nz, .blueNotMaxed
+    dec c
+.blueNotMaxed
+
+    ; Is green maxed?
+    ld a, e
+    cp $E0
+    jr c, .greenNotMaxed
+    ld a, d
+    and $3
+    cp $3
+    jr nz, .greenNotMaxed
+    res 5, c
+.greenNotMaxed
+
+    ; Is red maxed?
+    ld a, d
+    and $7C
+    cp $7C
+    jr nz, .redNotMaxed
+    res 2, b
+.redNotMaxed
+
+    ; add de, bc
+    ; ld [hli], de
+    ld a, e
+    add c
+    ld [hli], a
+    ld a, d
+    adc b
+    ld [hli], a
     pop bc
+    
     dec c
     jr nz, .frameLoop
 
-    call WaitFrame
     call WaitFrame
     ld hl, BgPalettes
     call LoadBGPalettes64
+    call WaitFrame
     dec b
     jr nz, .fadeLoop
 ENDC
+    ld a, 1
     call ClearVRAMViaHDMA
-    ; Select the first bank
-    xor a
-    ldh [$4F], a
-    cpl
+    call _ClearVRAMViaHDMA
+    call ClearVRAMViaHDMA ; A = $40, so it's bank 0
+    ; A should be $FF
     ldh [$00], a
-    call ClearVRAMViaHDMA
     
     ; Final values for CGB mode
     ld de, $ff56
@@ -993,70 +1043,24 @@ LoadPalettes:
     jr nz, .loop
     ret
 
-BrightenColor:
-    ld a, [hli]
-    ld e, a
-    ld a, [hld]
-    ld d, a
-    ; RGB(1,1,1)
-    ld bc, $421
-
-    ; Is blue maxed?
-    ld a, e
-    and $1F
-    cp $1F
-    jr nz, .blueNotMaxed
-    dec c
-.blueNotMaxed
-
-    ; Is green maxed?
-    ld a, e
-    cp $E0
-    jr c, .greenNotMaxed
-    ld a, d
-    and $3
-    cp $3
-    jr nz, .greenNotMaxed
-    res 5, c
-.greenNotMaxed
-
-    ; Is red maxed?
-    ld a, d
-    and $7C
-    cp $7C
-    jr nz, .redNotMaxed
-    res 2, b
-.redNotMaxed
-
-    ; add de, bc
-    ; ld [hli], de
-    ld a, e
-    add c
-    ld [hli], a
-    ld a, d
-    adc b
-    ld [hli], a
-    ret
-
 ClearVRAMViaHDMA:
-    ld hl, $FF51
-
-    ; Src
-    ld a, $88
-    ld [hli], a
-    xor a
-    ld [hli], a
-
-    ; Dest
-    ld a, $98
-    ld [hli], a
-    ld a, $A0
-    ld [hli], a
-
-    ; Do it
-    ld [hl], $12
+    ldh [$4F], a
+    ld hl, HDMAData
+_ClearVRAMViaHDMA:
+    ld c, $51
+    ld b, 5
+.loop
+    ld a, [hli]
+    ldh [c], a
+    inc c
+    dec b
+    jr nz, .loop
     ret
 
+HDMAData:
+    db $88, $00, $98, $A0, $12
+    db $88, $00, $80, $00, $40
+
 GetInputPaletteIndex:
     ld a, $20 ; Select directions
     ldh [$00], a
@@ -1196,17 +1200,6 @@ LoadDMGTilemap:
     pop af
     ret
 
-InitWaveform:
-    ld hl, $FF30
-; Init waveform
-    xor a
-    ld c, $10
-.waveformLoop
-    ldi [hl], a
-    cpl
-    dec c
-    jr nz, .waveformLoop
-    ret
 
 SECTION "ROMMax", ROM0[$900]
     ; Prevent us from overflowing

From a3f261184d2a732b32eb30d6b3538b04fba31988 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Tue, 5 May 2020 01:44:48 +0300
Subject: [PATCH 07/10] Optimize more

---
 BootROMs/cgb_boot.asm | 65 +++++++++++++++++++++----------------------
 1 file changed, 31 insertions(+), 34 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 33e18e56..a83c27a3 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -6,8 +6,7 @@ Start:
     ld sp, $fffe
 
 ; Clear memory VRAM
-    ld hl, $8000
-    call ClearMemoryPage
+    call ClearMemoryPage8000
     ld a, 2
     ld c, $70
     ld [c], a
@@ -74,9 +73,7 @@ Start:
 ; Clear the second VRAM bank
     ld a, 1
     ldh [$4F], a
-    xor a
-    ld hl, $8000
-    call ClearMemoryPage
+    call ClearMemoryPage8000
     call LoadTileset
 
     ld b, 3
@@ -177,8 +174,10 @@ ENDC
 IF !DEF(FAST)
     call DoIntroAnimation
 
+    ld a, 45
+    ldh [WaitLoopCounter], a
 ; Wait ~0.75 seconds
-    ld b, 45
+    ld b, a
     call WaitBFrames
 
     ; Play first sound
@@ -189,10 +188,6 @@ IF !DEF(FAST)
     ; Play second sound
     ld a, $c1
     call PlaySound
-
-; Wait ~0.5 seconds
-    ld a, 30
-    ldh [WaitLoopCounter], a
     
 .waitLoop
     call GetInputPaletteIndex
@@ -602,8 +597,11 @@ PlaySound:
     ldh [$14], a
     ret
 
+ClearMemoryPage8000:
+    ld hl, $8000
 ; Clear from HL to HL | 0x2000
 ClearMemoryPage:
+    xor a
     ldi [hl], a
     bit 5, h
     jr z, ClearMemoryPage
@@ -781,6 +779,7 @@ IF !DEF(FAST)
 .fadeLoop
     ld c, 32 ; 32 colors to fade
     ld hl, BgPalettes
+    push hl
 .frameLoop
     push bc
     
@@ -833,7 +832,7 @@ IF !DEF(FAST)
     jr nz, .frameLoop
 
     call WaitFrame
-    ld hl, BgPalettes
+    pop hl
     call LoadBGPalettes64
     call WaitFrame
     dec b
@@ -888,6 +887,14 @@ ENDC
     ldh [$4C], a
     ld a, $1
     ret
+    
+GetKeyComboPalette:
+    ld hl, KeyCombinationPalettes - 1 ; Return value is 1-based, 0 means nothing down
+    ld c ,a
+    ld b, 0
+    add hl, bc
+    ld a, [hl]
+    ret
 
 EmulateDMG:
     ld a, 1
@@ -900,11 +907,7 @@ EmulateDMG:
     ldh a, [InputPalette]
     and a
     jr z, .nothingDown
-    ld hl, KeyCombinationPalettes - 1 ; Return value is 1-based, 0 means nothing down
-    ld c ,a
-    ld b, 0
-    add hl, bc
-    ld a, [hl]
+    call GetKeyComboPalette
     jr .paletteFromKeys
 .nothingDown
     ld a, b
@@ -985,17 +988,21 @@ GetPaletteIndex:
 .notNintendo
     xor a
     ret
-
-LoadPalettesFromIndex: ; a = index of combination
+    
+GetPaletteCombo:
     ld b, a
     ; Multiply by 3
     add b
     add b
-
+    
     ld hl, PaletteCombinations
     ld b, 0
     ld c, a
     add hl, bc
+    ret
+
+LoadPalettesFromIndex: ; a = index of combination
+    call GetPaletteCombo
 
     ; Obj Palettes
     ld e, 0
@@ -1100,20 +1107,10 @@ ChangeAnimationPalette:
     push hl
     push bc
     push de
-    ld hl, KeyCombinationPalettes - 1 ; Input palettes are 1-based, 0 means nothing down
-    ld c, a
-    ld b, 0
-    add hl, bc
-    ld a, [hl]
-    ld b, a
-    ; Multiply by 3
-    add b
-    add b
-
-    ld hl, PaletteCombinations + 2; Background Palette
-    ld b, 0
-    ld c, a
-    add hl, bc
+    call GetKeyComboPalette
+    call GetPaletteCombo
+    inc l
+    inc l
     ld a, [hl]
     ld hl, Palettes + 1
     ld b, 0
@@ -1164,7 +1161,7 @@ ChangeAnimationPalette:
     ld hl, BgPalettes
     call LoadBGPalettes64
     ; Delay the wait loop while the user is selecting a palette
-    ld a, 30
+    ld a, 45
     ldh [WaitLoopCounter], a
     pop de
     pop bc

From 730567dc609a1213d6395ca92d4be00c072a2817 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Wed, 6 May 2020 01:06:22 +0300
Subject: [PATCH 08/10] Proper color mixing

---
 BootROMs/cgb_boot.asm | 198 ++++++++++++++++++++++++------------------
 1 file changed, 115 insertions(+), 83 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index a83c27a3..332279ca 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -10,35 +10,16 @@ Start:
     ld a, 2
     ld c, $70
     ld [c], a
-; Clear RAM Bank 2 (Like the original boot ROM
+; Clear RAM Bank 2 (Like the original boot ROM)
     ld h, $D0
-    xor a
     call ClearMemoryPage
     ld [c], a
     
-    ld hl, $FF30
-; Init waveform
-    ld c, $10
-.waveformLoop
-    ldi [hl], a
-    cpl
-    dec c
-    jr nz, .waveformLoop
-
 ; Clear chosen input palette
     ldh [InputPalette], a
 ; Clear title checksum
     ldh [TitleChecksum], a
-
-; Clear OAM
-    ld h, $fe
-    ld c, $a0
-.clearOAMLoop
-    ldi [hl], a
-    dec c
-    jr nz, .clearOAMLoop
-
-; Init Audio
+        
     ld a, $80
     ldh [$26], a
     ldh [$11], a
@@ -47,6 +28,23 @@ Start:
     ldh [$25], a
     ld a, $77
     ldh [$24], a
+    ld hl, $FF30
+; Init waveform
+    ld c, $10
+.waveformLoop
+    ldi [hl], a
+    cpl
+    dec c
+    jr nz, .waveformLoop
+        
+
+; Clear OAM
+    ld h, $fe
+    ld c, $a0
+.clearOAMLoop
+    ldi [hl], a
+    dec c
+    jr nz, .clearOAMLoop
     
 ; Init BG palette
     ld a, $fc
@@ -142,25 +140,44 @@ ENDC
 
     ; Expand Palettes
     ld de, AnimationColors
-    ld c, 16
+    ld c, 8
     ld hl, BgPalettes
     xor a
 .expandPalettesLoop:
     cpl
-    ; One white or black
-    ldi [hl], a
-    ldi [hl], a
+    ; One white
+    ld [hli], a
+    ld [hli], a
+    
+    ; Mixed with white
+    ld a, [de]
+    inc e
+    or $20
+    ld b, a
+    
+    ld a, [de]
+    dec e
+    or $84
+    rra
+    rr b
+    ld [hl], b
+    inc l
+    ld [hli], a
+    
+    ; One black
+    xor a
+    ld [hli], a
+    ld [hli], a
     
     ; One color
-    push af
     ld a, [de]
-    inc de
-    ldi [hl], a
+    inc e
+    ld [hli], a
     ld a, [de]
-    inc de
-    ldi [hl], a
-    pop af
-
+    inc e
+    ld [hli], a    
+    
+    xor a
     dec c
     jr nz, .expandPalettesLoop
 
@@ -200,6 +217,9 @@ ELSE
     call PlaySound
 ENDC
     call Preboot
+IF DEF(AGB)
+    ld b, 1
+ENDC
 
 ; Will be filled with NOPs
 
@@ -208,7 +228,6 @@ BootGame:
     ldh [$50], a
 
 SECTION "MoreStuff", ROM0[$200]
-
 ; Game Palettes Data
 TitleChecksums:
     db $00 ; Default
@@ -512,43 +531,40 @@ Palettes:
     dw $4778, $3290, $1D87, $0861 ; DMG LCD
 
 KeyCombinationPalettes
-    db 1 ; Right
-    db 48 ; Left
-    db 5 ; Up
-    db 8 ; Down
-    db 0 ; Right + A
-    db 40 ; Left + A
-    db 43 ; Up + A
-    db 3 ; Down + A
-    db 6 ; Right + B
-    db 7 ; Left + B
-    db 28 ; Up + B
-    db 49 ; Down + B
+    db 1  * 3  ; Right
+    db 48 * 3  ; Left
+    db 5  * 3  ; Up
+    db 8  * 3  ; Down
+    db 0  * 3  ; Right + A
+    db 40 * 3  ; Left + A
+    db 43 * 3  ; Up + A
+    db 3  * 3  ; Down + A
+    db 6  * 3  ; Right + B
+    db 7  * 3  ; Left + B
+    db 28 * 3  ; Up + B
+    db 49 * 3  ; Down + B
     ; SameBoy "Exclusives"
-    db 51 ; Right + A + B
-    db 52 ; Left + A + B
-    db 53 ; Up + A + B
-    db 54 ; Down + A + B
-
+    db 51 * 3 ; Right + A + B
+    db 52 * 3 ; Left + A + B
+    db 53 * 3 ; Up + A + B
+    db 54 * 3 ; Down + A + B
+    
 TrademarkSymbol:
     db $3c,$42,$b9,$a5,$b9,$a5,$42,$3c
 
 SameBoyLogo:
     incbin "SameBoyLogo.pb12"
 
-animation_color: MACRO
-    dw ((\1) >> 1) | $4210, (\1)
-ENDM
 
 AnimationColors:
-    animation_color $7FFF, ($7FFF >> 1) ; White
-    animation_color $774F, ($774F >> 1) ; Cyan
-    animation_color $22C7, ($22C7 >> 1) ; Green
-    animation_color $039F, ($039F >> 1) ; Yellow
-    animation_color $017D, ($017D >> 1) ; Orange
-    animation_color $241D, ($241D >> 1) ; Red
-    animation_color $6D38, ($6D38 >> 1) ; Purple
-    animation_color $7102, ($7102 >> 1) ; Blue
+    dw $7FFF ; White
+    dw $774F ; Cyan
+    dw $22C7 ; Green
+    dw $039F ; Yellow
+    dw $017D ; Orange
+    dw $241D ; Red
+    dw $6D38 ; Purple
+    dw $7102 ; Blue
 AnimationColorsEnd:
 
 ; Helper Functions
@@ -842,11 +858,13 @@ ENDC
     call ClearVRAMViaHDMA
     call _ClearVRAMViaHDMA
     call ClearVRAMViaHDMA ; A = $40, so it's bank 0
+    cpl
     ; A should be $FF
     ldh [$00], a
     
     ; Final values for CGB mode
-    ld de, $ff56
+    ld d, a
+    ld e, c
     ld l, $0d
     
     ld a, [$143]
@@ -870,7 +888,7 @@ IF DEF(AGB)
     ld c, a
     add a, $11
     ld h, c
-    ld b, 1
+    ; B is set to 1 after ret
 ELSE
     ; Set registers to match the original CGB boot
     ; AF = $1180, C = 0
@@ -990,11 +1008,6 @@ GetPaletteIndex:
     ret
     
 GetPaletteCombo:
-    ld b, a
-    ; Multiply by 3
-    add b
-    add b
-    
     ld hl, PaletteCombinations
     ld b, 0
     ld c, a
@@ -1064,10 +1077,6 @@ _ClearVRAMViaHDMA:
     jr nz, .loop
     ret
 
-HDMAData:
-    db $88, $00, $98, $A0, $12
-    db $88, $00, $80, $00, $40
-
 GetInputPaletteIndex:
     ld a, $20 ; Select directions
     ldh [$00], a
@@ -1104,7 +1113,6 @@ GetInputPaletteIndex:
     ; Slide into change Animation Palette
 
 ChangeAnimationPalette:
-    push hl
     push bc
     push de
     call GetKeyComboPalette
@@ -1147,16 +1155,37 @@ ChangeAnimationPalette:
     inc hl
     inc hl
 .isNotWhite
+    ; Mixing code by ISSOtm
+    ldh a, [BgPalettes + 7 * 8 + 2]
+    and ~$21
+    ld b, a
+    ld a, [hli]
+    and ~$21
+    add a, b
+    ld b, a
+    ld a, [BgPalettes + 7 * 8 + 3]
+    res 2, a ; and ~$04, but not touching carry
+    ld c, [hl]
+    res 2, c ; and ~$04, but not touching carry
+    adc a, c
+    rra ; Carry sort of "extends" the accumulator, we're bringing that bit back home
+    ld [BgPalettes + 7 * 8 + 3], a
+    ld a, b
+    rra
+    ld [BgPalettes + 7 * 8 + 2], a
+    dec l
+    
     ld a, [hli]
     ldh [BgPalettes + 7 * 8 + 6], a ; Fourth color, 7th palette
-    ldh [BgPalettes + 7 * 8 + 2], a ; Second color, half, 7th palette; rough color mixing
     ld a, [hli]
     ldh [BgPalettes + 7 * 8 + 7], a ; Fourth color, 7th palette
+    
     ld a, [hli]
     ldh [BgPalettes + 4], a ; Third color, first palette
-    ld a, [hl]
+    ld a, [hli]
     ldh [BgPalettes + 5], a ; Third color, first palette
 
+
     call WaitFrame
     ld hl, BgPalettes
     call LoadBGPalettes64
@@ -1165,7 +1194,6 @@ ChangeAnimationPalette:
     ldh [WaitLoopCounter], a
     pop de
     pop bc
-    pop hl
     ret
 
 ReplaceColorInAllPalettes:
@@ -1181,7 +1209,7 @@ ReplaceColorInAllPalettes:
 LoadDMGTilemap:
     push af
     call WaitFrame
-    ld a,$19      ; Trademark symbol
+    ld a, $19      ; Trademark symbol
     ld [$9910], a ; ... put in the superscript position
     ld hl,$992f   ; Bottom right corner of the logo
     ld c,$c       ; Tiles in a logo row
@@ -1191,16 +1219,20 @@ LoadDMGTilemap:
     ldd [hl], a
     dec c
     jr nz, .tilemapLoop
-    ld l,$0f ; Jump to top row
+    ld l, $0f ; Jump to top row
     jr .tilemapLoop
 .tilemapDone
     pop af
     ret
-
-
-SECTION "ROMMax", ROM0[$900]
-    ; Prevent us from overflowing
-    ds 1
+    
+HDMAData:
+    db $88, $00, $98, $A0, $12
+    db $88, $00, $80, $00, $40
+    
+BootEnd:
+IF BootEnd > $900
+    FAIL "BootROM overflowed: {BootEnd}"
+ENDC
 
 SECTION "HRAM", HRAM[$FF80]
 TitleChecksum:

From 184743637eebf34f5fc673057dbca954037f3599 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Wed, 6 May 2020 01:10:46 +0300
Subject: [PATCH 09/10] Fix silly regression

---
 BootROMs/cgb_boot.asm | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/BootROMs/cgb_boot.asm b/BootROMs/cgb_boot.asm
index 332279ca..f07fbcfd 100644
--- a/BootROMs/cgb_boot.asm
+++ b/BootROMs/cgb_boot.asm
@@ -858,8 +858,7 @@ ENDC
     call ClearVRAMViaHDMA
     call _ClearVRAMViaHDMA
     call ClearVRAMViaHDMA ; A = $40, so it's bank 0
-    cpl
-    ; A should be $FF
+    ld a, $ff
     ldh [$00], a
     
     ; Final values for CGB mode

From 7cff35368d26e87a0638e9be2d56642665af0ac0 Mon Sep 17 00:00:00 2001
From: Lior Halphon <LIJI32@gmail.com>
Date: Wed, 6 May 2020 23:30:01 +0300
Subject: [PATCH 10/10] Port to C to remove the Python dep, remove leftovers

---
 BootROMs/pb12.c  |  95 +++++++++++++
 BootROMs/pb12.py |  68 ----------
 BootROMs/pb8.c   | 341 -----------------------------------------------
 Makefile         |   9 +-
 4 files changed, 101 insertions(+), 412 deletions(-)
 create mode 100644 BootROMs/pb12.c
 delete mode 100644 BootROMs/pb12.py
 delete mode 100644 BootROMs/pb8.c

diff --git a/BootROMs/pb12.c b/BootROMs/pb12.c
new file mode 100644
index 00000000..878dd0d2
--- /dev/null
+++ b/BootROMs/pb12.c
@@ -0,0 +1,95 @@
+#include <stdio.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <assert.h>
+
+void opts(uint8_t byte, uint8_t *options)
+{
+    *(options++) = byte | ((byte << 1) & 0xff);
+    *(options++) = byte & (byte << 1);
+    *(options++) = byte | ((byte >> 1) & 0xff);
+    *(options++) = byte & (byte >> 1);
+}
+
+int main()
+{
+    static uint8_t source[0x4000];
+    size_t size = read(STDIN_FILENO, &source, sizeof(source));
+    unsigned pos = 0;
+    assert(size <= 0x4000);
+    while (size && source[size - 1] == 0) {
+        size--;
+    }
+    
+    uint8_t *literals = NULL;
+    size_t literals_size = 0;
+    unsigned bits = 0;
+    unsigned control = 0;
+    unsigned prev[2] = {-1, -1}; // Unsigned to allow "not set" values
+    
+    while (true) {
+
+        uint8_t byte = 0;
+        if (pos == size){
+            if (bits == 0) break;
+        }
+        else {
+            byte = source[pos++];
+        }
+        
+        if (byte == prev[0] || byte == prev[1]) {
+            bits += 2;
+            control <<= 1;
+            control |= 1;
+            control <<= 1;
+            if (byte == prev[1]) {
+                control |= 1;
+            }
+        }
+        else {
+            bits += 2;
+            control <<= 2;
+            uint8_t options[4];
+            opts(prev[1], options);
+            bool found = false;
+            for (unsigned i = 0; i < 4; i++) {
+                if (options[i] == byte) {
+                    // 01 = modify
+                    control |= 1;
+                    
+                    bits += 2;
+                    control <<= 2;
+                    control |= i;
+                    found = true;
+                    break;
+                }
+            }
+            if (!found) {
+                literals = realloc(literals, literals_size++);
+                literals[literals_size - 1] = byte;
+            }
+        }
+        
+        prev[0] = prev[1];
+        prev[1] = byte;
+        if (bits >= 8) {
+            uint8_t outctl = control >> (bits - 8);
+            assert(outctl != 1);
+            write(STDOUT_FILENO, &outctl, 1);
+            write(STDOUT_FILENO, literals, literals_size);
+            bits -= 8;
+            control &= (1 << bits) - 1;
+            literals_size = 0;
+        }
+    }
+    uint8_t end_byte = 1;
+    write(STDOUT_FILENO, &end_byte, 1);
+
+    if (literals) {
+        free(literals);
+    }
+    
+    return 0;
+}
diff --git a/BootROMs/pb12.py b/BootROMs/pb12.py
deleted file mode 100644
index dc53d6b9..00000000
--- a/BootROMs/pb12.py
+++ /dev/null
@@ -1,68 +0,0 @@
-import sys
-
-def opts(byte):
-    # top bit: 0 = left, 1 = right
-    # bottom bit: 0 = or, 1 = and
-    if byte is None: return []
-    return [
-        byte | (byte << 1) & 0xff,
-        byte & (byte << 1),
-        byte | (byte >> 1) & 0xff,
-        byte & (byte >> 1),
-    ]
-
-def pb12(data):
-    data = iter(data)
-
-    literals = bytearray()
-    bits = 0
-    control = 0
-    prev = [None, None]
-    gotta_end = False
-
-    chunk = bytearray()
-    while True:
-        try:
-            byte = next(data)
-        except StopIteration:
-            if bits == 0: break
-            byte = 0
-        chunk.append(byte)
-
-        if byte in prev:
-            bits += 2
-            control <<= 1
-            control |= 1
-            control <<= 1
-            if prev[1] == byte:
-                control |= 1 # 10 = out[-2], 11 = out[-1]
-        else:
-            bits += 2
-            control <<= 2
-            options = opts(prev[1])
-            if byte in options:
-                # 01 = modify
-                control |= 1
-
-                bits += 2
-                control <<= 2
-                control |= options.index(byte)
-            else:
-                # 00 = literal
-                literals.append(byte)
-        prev = [prev[1], byte]
-        if bits >= 8:
-            outctl = control >> (bits - 8)
-            assert outctl != 1 # that's the end byte
-            yield bytes([outctl]) + literals
-            bits -= 8
-            control &= (1 << bits) - 1
-            literals = bytearray()
-            chunk = bytearray()
-    yield b'\x01'
-
-_, infile, outfile = sys.argv
-with open(infile, 'rb') as f:
-    data = f.read().rstrip(b'\x00')
-with open(outfile, 'wb') as f:
-    f.writelines(pb12(data))
diff --git a/BootROMs/pb8.c b/BootROMs/pb8.c
deleted file mode 100644
index 4ee4524e..00000000
--- a/BootROMs/pb8.c
+++ /dev/null
@@ -1,341 +0,0 @@
-/*
-
-PB8 compressor and decompressor
-
-Copyright 2019 Damian Yerrick
-
-This software is provided 'as-is', without any express or implied
-warranty.  In no event will the authors be held liable for any damages
-arising from the use of this software.
-
-Permission is granted to anyone to use this software for any purpose,
-including commercial applications, and to alter it and redistribute it
-freely, subject to the following restrictions:
-
-1. The origin of this software must not be misrepresented; you must not
-   claim that you wrote the original software. If you use this software
-   in a product, an acknowledgment in the product documentation would be
-   appreciated but is not required.
-2. Altered source versions must be plainly marked as such, and must not be
-   misrepresented as being the original software.
-3. This notice may not be removed or altered from any source distribution.
-
-*/
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdint.h>
-#include <string.h>
-#include <stdbool.h>
-#include <assert.h>
-
-// For setting stdin/stdout to binary mode
-#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
-#include <unistd.h>
-#define fd_isatty isatty
-#elif defined (_WIN32)
-#include <io.h>
-#include <fcntl.h>
-#define fd_isatty _isatty
-#endif
-
-/*
-
-; The logo is compressed using PB8, a form of RLE with unary-coded
-; run lengths.  Each block representing 8 bytes consists of a control
-; byte, where each bit (MSB to LSB) is 0 for literal or 1 for repeat
-; previous, followed by the literals in that block.
-
-SameBoyLogo_dst = $8080
-SameBoyLogo_length = (128 * 24) / 64
-
-LoadTileset:
-    ld hl, SameBoyLogo
-    ld de, SameBoyLogo_dst
-    ld c, SameBoyLogo_length
-.pb8BlockLoop:
-    ; Register map for PB8 decompression
-    ; HL: source address in boot ROM
-    ; DE: destination address in VRAM
-    ; A: Current literal value
-    ; B: Repeat bits, terminated by 1000...
-    ; C: Number of 8-byte blocks left in this block
-    ; Source address in HL lets the repeat bits go straight to B,
-    ; bypassing A and avoiding spilling registers to the stack.
-    ld b, [hl]
-    inc hl
-
-    ; Shift a 1 into lower bit of shift value.  Once this bit
-    ; reaches the carry, B becomes 0 and the byte is over
-    scf
-    rl b
-
-.pb8BitLoop:
-    ; If not a repeat, load a literal byte
-    jr c,.pb8Repeat
-    ld a, [hli]
-.pb8Repeat:
-    ; Decompressed data uses colors 0 and 1, so write once, inc twice
-    ld [de], a
-    inc de
-    inc de
-    sla b
-    jr nz, .pb8BitLoop
-
-    dec c
-    jr nz, .pb8BlockLoop
-    ret
-
-*/
-
-/* Compressor and decompressor *************************************/
-
-/**
- * Compresses an input stream to PB8 data on an output stream.
- * @param infp input stream
- * @param outfp output stream
- * @param blocklength size of an independent input block in bytes
- * @return 0 for reaching infp end of file, or EOF for error
- */
-int pb8(FILE *infp, FILE *outfp, size_t blocklength) 
-{
-  blocklength >>= 3;  // convert bytes to blocks
-  assert(blocklength > 0);
-  while (1) {
-    int last_byte = EOF;  // value that never occurs in a file
-    for (size_t blkleft = blocklength; blkleft > 0; --blkleft) {
-      unsigned int control_byte = 0x0001;
-      unsigned char literals[8];
-      size_t nliterals = 0;
-      while (control_byte < 0x100) {
-        int c = fgetc(infp);
-        if (c == EOF) break;
-
-        control_byte <<= 1;
-        if (c == last_byte) {
-          control_byte |= 0x01;
-        }
-        else {
-          literals[nliterals++] = last_byte = c;
-        }
-      }
-      if (control_byte > 1) {
-        // Fill partial block with repeats
-        while (control_byte < 0x100) {
-          control_byte = (control_byte << 1) | 1;
-        }
-
-        // Write control byte and check for write failure
-        int ok = fputc(control_byte & 0xFF, outfp);
-        if (ok == EOF) return EOF;
-        size_t ok2 = fwrite(literals, 1, nliterals, outfp);
-        if (ok2 < nliterals) return EOF;
-      }
-
-      // If finished, return success or failure
-      if (ferror(infp) || ferror(outfp)) return EOF;
-      if (feof(infp)) return 0;
-    }  // End 8-byte block
-  }  // End packet, resetting last_byte
-}
-
-/**
- * Decompresses PB8 data on an input stream to an output stream.
- * @param infp input stream
- * @param outfp output stream
- * @return 0 for reaching infp end of file, or EOF for error
- */
-int unpb8(FILE *infp, FILE *outfp) 
-{
-  int last_byte = 0;
-  while (1) {
-    int control_byte = fgetc(infp);
-    if (control_byte == EOF) {
-      return feof(infp) ? 0 : EOF;
-    }
-    control_byte &= 0xFF;
-    for (size_t bytesleft = 8; bytesleft > 0; --bytesleft) {
-      if (!(control_byte & 0x80)) {
-        last_byte = fgetc(infp);
-        if (last_byte == EOF) return EOF;  // read error
-      }
-      control_byte <<= 1;
-      int ok = fputc(last_byte, outfp);
-      if (ok == EOF) return EOF;
-    }
-  }
-}
-
-/* CLI frontend ****************************************************/
-
-static inline void set_fd_binary(unsigned int fd) 
-{
-#ifdef _WIN32
-  _setmode(fd, _O_BINARY);
-#else
-  (void) fd;
-#endif
-}
-
-static const char *usage_msg =
-"usage: pb8 [-d] [-l blocklength] [infile [outfile]]\n"
-"Compresses a file using RLE with unary run and literal lengths.\n"
-"\n"
-"options:\n"
-"  -d                decompress\n"
-"  -l blocklength    allow RLE packets to span up to blocklength\n"
-"                      input bytes (multiple of 8; default 8)\n"
-"  -h, -?, --help    show this usage page\n"
-"  --version         show copyright info\n"
-"\n"
-"If infile is - or missing, it is standard input.\n"
-"If outfile is - or missing, it is standard output.\n"
-"You cannot compress to or decompress from a terminal.\n"
-;
-static const char *version_msg =
-"PB8 compressor (C version) v0.01\n"
-"Copyright 2019 Damian Yerrick <https://pineight.com/contact/>\n"
-"This software is provided 'as-is', without any express or implied\n"
-"warranty.\n"
-;
-static const char *toomanyfilenames_msg =
-"pb8: too many filenames; try pb8 --help\n";
-
-int main(int argc, char **argv) 
-{
-  const char *infilename = NULL;
-  const char *outfilename = NULL;
-  bool decompress = false;
-  size_t blocklength = 8;
-
-  for (int i = 1; i < argc; ++i) {
-    if (argv[i][0] == '-' && argv[i][1] != 0) {
-      if (!strcmp(argv[i], "--help")) {
-        fputs(usage_msg, stdout);
-        return 0;
-      }
-      if (!strcmp(argv[i], "--version")) {
-        fputs(version_msg, stdout);
-        return 0;
-      }
-
-      // -t1 or -t 1
-      int argtype = argv[i][1];
-      switch (argtype) {
-        case 'h':
-        case '?':
-          fputs(usage_msg, stdout);
-          return 0;
-
-        case 'd':
-          decompress = true;
-          break;
-
-        case 'l': {
-          const char *argvalue = argv[i][2] ? argv[i] + 2 : argv[++i];
-          const char *endptr = NULL;
-
-          unsigned long tvalue = strtoul(argvalue, (char **)&endptr, 10);
-          if (endptr == argvalue || tvalue == 0 || tvalue > SIZE_MAX) {
-            fprintf(stderr, "pb8: block length %s not a positive integer\n",
-                    argvalue);
-            return EXIT_FAILURE;
-          }
-          if (tvalue % 8 != 0) {
-            fprintf(stderr, "pb8: block length %s not a multiple of 8\n",
-                    argvalue);
-            return EXIT_FAILURE;
-          }
-          blocklength = tvalue;
-        } break;
-
-        default:
-          fprintf(stderr, "pb8: unknown option -%c\n", argtype);
-          return EXIT_FAILURE;
-      }
-    }
-    else if (!infilename) {
-      infilename = argv[i];
-    }
-    else if (!outfilename) {
-      outfilename = argv[i];
-    }
-    else {
-      fputs(toomanyfilenames_msg, stderr);
-      return EXIT_FAILURE;
-    }
-  }
-  if (infilename && !strcmp(infilename, "-")) {
-    infilename = NULL;
-  }
-  if (!infilename && decompress && fd_isatty(0)) {
-    fputs("pb8: cannot decompress from terminal; try redirecting stdin\n",
-          stderr);
-    return EXIT_FAILURE;
-  }
-  if (outfilename && !strcmp(outfilename, "-")) {
-    outfilename = NULL;
-  }
-  if (!outfilename && !decompress && fd_isatty(1)) {
-    fputs("pb8: cannot compress to terminal; try redirecting stdout or pb8 --help\n",
-          stderr);
-    return EXIT_FAILURE;
-  }
-
-  FILE *infp = NULL;
-  if (infilename) {
-    infp = fopen(infilename, "rb");
-    if (!infp) {
-      fprintf(stderr, "pb8: error opening %s ", infilename);
-      perror("for reading");
-      return EXIT_FAILURE;
-    }
-  }
-  else {
-    infp = stdin;
-    set_fd_binary(0);
-  }
-
-  FILE *outfp = NULL;
-  if (outfilename) {
-    outfp = fopen(outfilename, "wb");
-    if (!outfp) {
-      fprintf(stderr, "pb8: error opening %s ", outfilename);
-      perror("for writing");
-      fclose(infp);
-      return EXIT_FAILURE;
-    }
-  }
-  else {
-    outfp = stdout;
-    set_fd_binary(1);
-  }
-
-  int compfailed = 0;
-  int has_ferror = 0;
-  if (decompress) {
-    compfailed = unpb8(infp, outfp);
-  }
-  else {
-    compfailed = pb8(infp, outfp, blocklength);
-  }
-  fflush(outfp);
-  if (ferror(infp)) {
-    fprintf(stderr, "pb8: error reading %s\n",
-            infilename ? infilename : "<stdin>");
-    has_ferror = EOF;
-  }
-  fclose(infp);
-  if (ferror(outfp)) {
-    fprintf(stderr, "pb8: error writing %s\n",
-            outfilename ? outfilename : "<stdout>");
-    has_ferror = EOF;
-  }
-  fclose(outfp);
-
-  if (compfailed && !has_ferror) {
-    fputs("pb8: unknown compression failure\n", stderr);
-  }
-
-  return (compfailed || has_ferror) ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/Makefile b/Makefile
index 1f818628..78978e49 100644
--- a/Makefile
+++ b/Makefile
@@ -20,7 +20,7 @@ else
 EXESUFFIX:=
 endif
 
-PB8_COMPRESS := build/pb8$(EXESUFFIX)
+PB12_COMPRESS := build/pb12$(EXESUFFIX)
 
 ifeq ($(PLATFORM),Darwin)
 DEFAULT := cocoa
@@ -386,8 +386,11 @@ $(OBJ)/%.2bpp: %.png
 	-@$(MKDIR) -p $(dir $@)
 	rgbgfx -h -u -o $@ $<
 
-$(OBJ)/BootROMs/SameBoyLogo.pb12: $(OBJ)/BootROMs/SameBoyLogo.2bpp BootROMs/pb12.py
-	python3 BootROMs/pb12.py $< $@
+$(OBJ)/BootROMs/SameBoyLogo.pb12: $(OBJ)/BootROMs/SameBoyLogo.2bpp $(PB12_COMPRESS)
+	$(PB12_COMPRESS) < $< > $@
+	
+$(PB12_COMPRESS): BootROMs/pb12.c
+	$(CC) -Wall -Werror $< -o $@
 
 $(BIN)/BootROMs/agb_boot.bin: BootROMs/cgb_boot.asm
 $(BIN)/BootROMs/cgb_boot_fast.bin: BootROMs/cgb_boot.asm