improve "IP" decompilation

.gitignore

@@ -14,6 +14,7 @@ scramble
 cdi4dc
 tools/ttf_outline
 tools/ttf_bitmap
+tools/ftdi_transfer
 k_means_vq
 *.blend1
 *.scramble

burn.sh

@@ -0,0 +1,2 @@
+cdrecord -speed=8 -v dev=/dev/sr0 -dao -multi taudio01.wav
+cdrecord -eject -overburn -speed=8 -v dev=/dev/sr0 -tao -xa tdata02.iso

common.lds

@@ -34,7 +34,7 @@ SECTIONS
     KEEP(*(.ctors.*))
   } > p1ram
 
-  .bss ALIGN(4) (NOLOAD) : SUBALIGN(4)
+  .bss ALIGN(4) (NOLOAD) :
   {
     *(.bss)
     *(.bss.*)

common.mk

@@ -2,7 +2,7 @@ MAKEFILE_PATH := $(abspath $(lastword $(MAKEFILE_LIST)))
 DIR := $(dir $(MAKEFILE_PATH))
 
 LIB ?= .
-OPT ?= -O2
+OPT ?= -Og
 GENERATED ?=
 
 AARCH = --isa=sh4 --little
@@ -21,16 +21,16 @@ IP_OBJ = \
 	systemid.o \
 	toc.o \
 	sg/sg_sec.o \
-	sg/sg_arejp.o \
-	sg/sg_areus.o \
-	sg/sg_areec.o \
-	sg/sg_are00.o \
-	sg/sg_are01.o \
-	sg/sg_are02.o \
-	sg/sg_are03.o \
-	sg/sg_are04.o \
-	sg/sg_ini.o \
-	sg/aip.o
+	sg_arejp.o \
+	sg_areus.o \
+	sg_areec.o \
+	sg_are00.o \
+	sg_are01.o \
+	sg_are02.o \
+	sg_are03.o \
+	sg_are04.o \
+	sg_ini.o \
+	aip.o
 
 %.o: %.obj
 	$(OBJCOPY) -g \
@@ -80,13 +80,13 @@ sine.pcm: common.mk
 		/=./COPYRIGH.TXT \
 		/=./ABSTRACT.TXT \
 		/=./BIBLIOGR.TXT \
-		/=./REIGN.PCM \
-		/=./PILLAR.PCM \
-		/=./RIDDLE.PCM \
-		/=./PRELUDE.PCM \
-		/=./CLOCKTOW.PCM \
-		/=./ELEC.PCM \
-		/=./ECCLESIA.PCM
+		/=./pcm/REIGN.PCM \
+		/=./pcm/PILLAR.PCM \
+		/=./pcm/RIDDLE.PCM \
+		/=./pcm/PRELUDE.PCM \
+		/=./pcm/CLOCKTOW.PCM \
+		/=./pcm/ELEC.PCM \
+		/=./pcm/ECCLESIA.PCM
 
 %.cdi: %.iso
 	./cdi4dc $< $@ >/dev/null

example/aica_gdrom_dft.cpp

@@ -206,14 +206,17 @@ void * memset(void * dest, int c, size_t n)
   return dest;
 }
 
-static fft::complex comp[samples_per_line * 2];
+static fft::complex comp[samples_per_line * 8];
 
-void render_column(int col, int x, const uint32_t * buf)
+int render_column(int col, int x, const uint32_t * buf)
 {
   uint32_t offset = texture_memory_alloc::texture.start + 0;
   auto texture = reinterpret_cast<volatile uint32_t *>(&texture_memory64[offset / 4]);
 
-  const int16_t * src = &((int16_t *)(buf))[col * samples_per_line * 2];
+  const int16_t * src = &((int16_t *)(buf))[col * samples_per_line / 2];
+  if ((src) >= (int16_t*)&buf[(chunk_size - samples_per_line * 2)/4])
+    return 0; // FIXME: this is a hack
+
   fft::int16_to_complex(src, samples_per_line * 2, comp);
   fft::fft(comp, samples_per_line * 2);
   for (int32_t y = 0; y < samples_per_line; y++) {
@@ -229,24 +232,27 @@ void render_column(int col, int x, const uint32_t * buf)
     value |= v << shift;
     texture[ix / 4] = value;
   }
+
+  return 1;
 }
 
 void render();
 void texture_init();
 
-void aica_step(uint32_t& chunk_index, const uint32_t gdrom_buf[2][chunk_size / 4])
+void aica_step(uint32_t& chunk_index, const uint32_t gdrom_buf[2][chunk_size / 4 * 2])
 {
   { // wait for interrupt from arm
     int col = 0;
     while ((system.ISTEXT & (1 << 1)) == 0) {
       if (step_time >= 0) {
 	int32_t dt = step_start - (int32_t)sh7091.TMU.TCNT0;
-	if (step_time * col / lines_per_chunk < dt) {
-	  render_column(col, __x, gdrom_buf[!chunk_index]);
-	  render();
-	  col += 1;
-	  __x += 1;
+	if (step_time * col / (lines_per_chunk * 4) < dt) {
+	  int inc = render_column(col, __x, gdrom_buf[!chunk_index]);
+	  col += inc;
+	  __x += inc;
 	  __x %= samples_per_line;
+	  if (col % 2)
+	    render();
 	}
       }
     };
@@ -362,14 +368,9 @@ uint32_t gdrom_cd_read2(uint16_t * buf,
     length += byte_count;
     gdrom_read_data(buf, byte_count);
 
-    serial::string("status: ");
-    serial::integer<uint8_t>(gdrom_if.status);
-
     while ((gdrom::status::bsy(gdrom_if.status)) != 0); // wait for drive to become not-busy
   }
 
-  serial::string("length: ");
-  serial::integer<uint32_t>(length);
   return length;
 }
 
@@ -466,7 +467,7 @@ struct extent gdrom_find_file()
     serial::string(dr->file_identifier, dr->length_of_file_identifier);
     serial::character('\n');
 
-    const char filename[] = "ELEC.PCM;1";
+    const char filename[] = "RIDDLE.PCM;1";
     bool equal = str_equal(dr->file_identifier, dr->length_of_file_identifier,
                            filename, (sizeof (filename)) - 1);
 
@@ -603,7 +604,9 @@ void render()
 
 void main()
 {
-  serial::init(4);
+  memset(&comp, 0, (sizeof (comp)));
+
+  serial::init(0);
 
   holly.SOFTRESET = softreset::pipeline_soft_reset
 		  | softreset::ta_soft_reset;
@@ -646,7 +649,7 @@ void main()
 
   gdrom_unlock();
   const auto extent = gdrom_find_file();
-  uint32_t gdrom_buf[2][chunk_size / 4];
+  uint32_t gdrom_buf[2][(chunk_size / 4) * 2] = {0};
   gdrom_read_chunk(gdrom_buf[chunk_index], extent.location + segment_index, sectors_per_chunk);
   next_segment(extent, segment_index);
 

example/macaw.cpp

@@ -17,7 +17,7 @@
 #include "holly/video_output.hpp"
 #include "memorymap.hpp"
 
-#include "macaw.hpp"
+#include "texture/macaw/macaw.data.h"
 
 struct vertex {
   float x;
@@ -110,13 +110,13 @@ void init_texture_memory(const struct opb_size& opb_size)
                 480 / 32, // height
                 opb_size
                 );
-  background_parameter(0xff00ff00);
+  background_parameter(0xff004400);
 }
 
 void copy_macaw_texture()
 {
-  auto src = reinterpret_cast<const uint8_t *>(&_binary_macaw_data_start);
-  auto size  = reinterpret_cast<const uint32_t>(&_binary_macaw_data_size);
+  auto src = reinterpret_cast<const uint8_t *>(&_binary_texture_macaw_macaw_data_start);
+  auto size  = reinterpret_cast<const uint32_t>(&_binary_texture_macaw_macaw_data_size);
   auto texture = reinterpret_cast<volatile uint16_t *>(&texture_memory64[texture_memory_alloc::texture.start / 4]);
   for (uint32_t px = 0; px < size / 3; px++) {
     uint8_t r = src[px * 3 + 0];

ip.lds

@@ -42,11 +42,13 @@ SECTIONS
   .text.sg_ini :
   {
     KEEP(*(.text.*sg_ini))
+    . = ORIGIN(sg_ini) + LENGTH(sg_ini);
   } > sg_ini
 
   .text.aip :
   {
     KEEP(*(.text.*aip))
+    . = ORIGIN(aip) + LENGTH(aip);
   } > aip
 
   /DISCARD/ :
@@ -56,4 +58,6 @@ SECTIONS
     *(.bss)
     *(.comment)
   }
+
+  INCLUDE "debug.lds"
 }

math/geometry.hpp

@@ -1,4 +1,4 @@
-#include <functional>
+#include <tuple>
 
 #include "vec.hpp"
 

math/mat2x2.hpp

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <utility>
+#include "vec2.hpp"
 
 template <int R, int C, typename T>
 struct mat;

math/mat4x4.hpp

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <utility>
+#include "vec4.hpp"
 
 template <int R, int C, typename T>
 struct mat;

sg_are00.s

@@ -0,0 +1,5 @@
+	.section	.text.sg_are00
+	bra	_sg_are00_next
+	nop
+	.ascii 	"                            "
+_sg_are00_next:

sg_are01.s

@@ -0,0 +1,5 @@
+	.section	.text.sg_are01
+	bra	_sg_are01_next
+	nop
+	.ascii 	"                            "
+_sg_are01_next:

sg_are02.s

@@ -0,0 +1,5 @@
+	.section	.text.sg_are02
+	bra	_sg_are02_next
+	nop
+	.ascii 	"                            "
+_sg_are02_next:

sg_are03.s

@@ -0,0 +1,5 @@
+	.section	.text.sg_are03
+	bra	_sg_are03_next
+	nop
+	.ascii 	"                            "
+_sg_are03_next:

sg_are04.s

@@ -0,0 +1,5 @@
+	.section	.text.sg_are04
+	bra	_sg_are04_next
+	nop
+	.ascii 	"                            "
+_sg_are04_next:

sg_areec.s

@@ -0,0 +1,5 @@
+	.section 	.text.sg_areec
+	bra	_sg_areec_next
+	nop
+	.ascii 	"For EUROPE.                 "
+_sg_areec_next:

sg_arejp.s

@@ -0,0 +1,5 @@
+	.section 	.text.sg_arejp
+	bra	_sg_arejp_next
+	nop
+	.ascii 	"For JAPAN,TAIWAN,PHILIPINES."
+_sg_arejp_next:

sg_areus.s

@@ -0,0 +1,5 @@
+	.section 	.text.sg_areus
+	bra	_sg_areus_next
+	nop
+	.ascii	"For USA and CANADA.         "
+_sg_areus_next:

sg_ini.s

@@ -2,7 +2,7 @@
 	.global _text_sg_ini_start
 
 _text_sg_ini_start:
-	mov.l	label_18,r0	/* 8d000000 → beginning of system ram, image area */
+	mov.l	label_18,r0	/* 8d000000 → end of system ram */
 	mov	r0,r15
 	nop
 	nop
@@ -23,26 +23,36 @@ label_20:
 	.fill	0x2000,1,0x00
 
 label_2020:
+	/* function that processes the list of addresses/values at label_2284 */
 	mov.l	label_204c,r0	/* 8c00d940 → label_2140 */
 	mov	#0,r1
 	jsr	@r0
 	nop
+
+	/* ???? possibly jump to boot menu? */
 	mov.l	label_2050,r0	/* 8c00d900 → label_2100 */
 	jsr	@r0
 	nop
+
+	/* possibly jump to boot menu? */
 	mov.l	label_2054,r0	/* 8c00d888 → label_2088 */
 	jsr	@r0
 	nop
+
+	/* check for Windows CE? */
 	mov.l	label_2058,r0	/* 8c00dae0 → label_22e0 */
 	jsr	@r0
 	nop
+
 	mov.l	label_205c,r0	/* 8c00db40 → label_2340 */
 	jsr	@r0
 	nop
+
 	nop
 	mov.l	label_2060,r0	/* 8c00d86c → label_206c */
 	jmp	@r0
 	nop
+
 	nop
 	nop
 
@@ -80,7 +90,7 @@ label_2084:
 label_2088:
 	sts.l	pr,@-r15
 	mov.l	label_20a4,r0	/* 8c000048 */
-	mov.l	@r0,r0
+	mov.l	@r0,r0		/* r0 ← 1 */
 	cmp/eq	#6,r0
 	bt	label_209e
 	cmp/eq	#7,r0
@@ -98,6 +108,7 @@ label_20a4:
 label_20a8:
 	.long	0x8c00d8ac
 
+	/* jump back to boot menu? */
 label_20ac:
 	mov.l	label_20b4,r0	/* 8c0000e0 */
 	mov.l	@r0,r0
@@ -108,61 +119,68 @@ label_20b4:
 
 label_20b8:
 	exts.b	r4,r0
-	mov	#57,r1
+	mov	#57,r1		/* ascii '9' */
 	cmp/gt	r1,r0
 	bf	label_20c2
 	add	#-7,r0
 label_20c2:
 	rts
 	add	#-48,r0
+
+	/* called from label_2100 and label_22e0 */
 label_20c6:
 	xor	r2,r2
 	sts.l	pr,@-r15
-	mov.b	@r4,r4
-	mov	r4,r0
-	cmp/eq	#32,r0
-	bt	label_20e0
-	bsr	label_20b8
+	mov.b	@r4,r4		/* r4 ← (0x8c00803e) = 0x30 */
+	mov	r4,r0		/* r0 ← 0x30 */
+	cmp/eq	#0x20,r0	/* check for ascii space in 'Compatible peripherals' of systemid */
+	bt	label_20e0	/* if it is space, return 0 */
+	bsr	label_20b8	/* ascii 0-9 to integer */
 	nop
 	mov	#1,r1
-	tst	r1,r0
-	bt.s	label_20e0
+	tst	r1,r0		/* check for ascii '1' / "uses Windows CE" */
+	bt.s	label_20e0	/* if not "uses Windows CE", take branch to label_20e0 */
 	xor	r2,r2
 	mov	#1,r2
 label_20e0:
 	lds.l	@r15+,pr
 	rts
-	mov	r2,r0
+	mov	r2,r0		/* return 1 if Windows CE, 0 if not */
 	.word 0xffff
 label_20e8:
 	mov.l	label_20fc,r1	! 8c008024
-	mov.l	label_20f8,r3	! 4c494d20
-	mov.l	@r1,r2
+	mov.l	label_20f8,r3	/* ascii ' MIL' ; 0x4c494d20 */
+	mov.l	@r1,r2		/* r2 ← 0x2d444720 ' GD-' */
 	cmp/eq	r3,r2
 	movt	r0
 	rts
 	nop
 	nop
 label_20f8:
-	.long 	0x4c494d20
+	.ascii	" MIL"	/* 4c494d20 */
 label_20fc:
 	.long	0x8c008024
+
+	/* called from label_2020 */
 label_2100:
 	sts.l	pr,@-r15
 	mov.l	label_212c,r3	/* 8c00d8c6 → label_20c6 */
-	mov.l	label_213c,r4	! 8c008000
+	mov.l	label_213c,r4	/* 8c008000 */
 	jsr	@r3
-	add	#62,r4
-	tst	r0,r0
+	add	#0x3e,r4	/* r4 ← 0x8c00803e */
+	tst	r0,r0		/* check return value != 0 */
 	bf	label_2126
+
+	/* check for ' MIL' CD */
 	mov.l	label_2130,r3	/* 8c00d8e8 → label_20e8 */
 	jsr	@r3
 	nop
-	tst	r0,r0
+	tst	r0,r0		/* check return value != 0 */
 	bf	label_2126
+
 	mov.l	label_2138,r3	! 8c010000
 	mov.w	@r3,r0
-	cmp/eq	#34,r0
+	cmp/eq	#34,r0		/* check for 'stc vbr,r0' ?? */
 	bf	label_2126
 	mov.l	label_2134,r1	/* 8c00d8ac → label_20ac */
 	jmp	@r1
@@ -182,22 +200,25 @@ label_2138:
 label_213c:
 	.long	0x8c008000
 
+	/* function that processes the list of addresses/values at label_2284 */
 label_2140:
 	mov.l	r2,@-r15
 	mov.l	r3,@-r15
 	mov.l	r4,@-r15
 	mov.l	r5,@-r15
 	sts.l	pr,@-r15
-	mova	label_2284,r0
-	mov.w	@(r0,r1),r1
-	add	r1,r0
+	mova	label_2284,r0	/* 0x8c00da84 */
+	mov.w	@(r0,r1),r1	/* 0x2 */
+	add	r1,r0		/* 0x8c00da86 */
 label_2150:
 	add	#3,r0
 	shlr2	r0
-	shll2	r0
-	mov.w	@r0+,r4
-	tst	r4,r4
-	bf	label_2172
+	shll2	r0		/* #1 r0 ← 0x8c00da88
+				   #2 r0 ← 0x8c00da90 */
+	mov.w	@r0+,r4		/* #1 r0 ← 0x8c00da8a ; r4 ← 2
+				   #2 r4 ← 1 */
+	tst	r4,r4		/* r4 & r4 != 0 → T unset */
+	bf	label_2172	/* branch taken */
 	mov.w	@r0+,r4
 	tst	r4,r4
 	bf.s	label_2150
@@ -210,20 +231,22 @@ label_2150:
 	rts
 	sett
 label_2172:
-	mov.w	@r0+,r3
-	mov.l	label_2278,r1	! ffff8000
-	and	r3,r1
-	tst	r1,r1
-	bt	label_2180
+	mov.w	@r0+,r3		/* #1 0x8c00da8a ; r0 ← 0x8c00da8c ; r3 ← 0x6
+				   #2 0x8c00da92 ; r0 ← 0x8c00da94 ; r3 ← 0x18
+				*/
+	mov.l	label_2278,r1	/* ffff8000 */
+	and	r3,r1		/* r1 ← 0 */
+	tst	r1,r1		/* r1 & r1 == 0 → T set */
+	bt	label_2180	/* branch taken */
 	xor	r1,r3
 	bf	label_2182
 label_2180:
-	mov.l	@r0+,r2
+	mov.l	@r0+,r2		/* 0x8c00da8c ; r2 ← 0xff000038 */
 label_2182:
-	mov.l	label_227c,r1	! 6000
-	and	r3,r1
-	tst	r1,r1
-	bt	label_21a0
+	mov.l	label_227c,r1	/* r1 ← 6000 */
+	and	r3,r1		/* r1 ← 0 */
+	tst	r1,r1		/* T set */
+	bt	label_21a0	/* branch taken */
 	xor	r1,r3
 	add	#12,r3
 	mov.l	@r0+,r1
@@ -236,27 +259,43 @@ label_2194:
 	add	#4,r2
 	bra	label_2150
 	mov	r5,r0
+	/* branch from label_2182 */
 label_21a0:
-	bsrf	r3
-	dt	r4
-	bf.s	label_21a0
-	add	#4,r2
-	bt	label_2150
+	bsrf	r3		/* #1 0x06 ; PC ← 0x8c00d9aa (label_21aa)
+				   #2 0x18 ; PC ← 0x8c00d9bc (label_2abc)
+				   #3 0x0c ; PC ← 0x8c00d9b0 (label_21b0)
+	                        */
+	dt	r4		/* #1 r4 ← 1 ; T unset
+				   #2 r4 ← 0 ; T set
+				   #3 r4 ← 3 ; T unset
+	                        */
+	bf.s	label_21a0 	/* #1 branch taken ; #2 branch not taken */
+	add	#4,r2		/* 0xff000038 ; r2 ← 0xff00003c */
+	bt	label_2150	/* branch taken */
+label_21aa:
 	mov	#0,r1
 	rts
 	mov.l	r1,@r2
+label_21b0:
 	mov	#0,r1
 	rts
 	mov.w	r1,@r2
 	mov	#0,r1
 	rts
 	mov.b	r1,@r2
-	mov.l	@r0+,r1
+label_21bc:
+	mov.l	@r0+,r1		/* 0x8c00da98 ; r1 ← 0x12c0 */
 	rts
-	mov.l	r1,@r2
+	mov.l	r1,@r2		/* 0xffa0002c ; CHCR2 ← 0x000012c0
+				DM: fixed
+				SM: incremented
+				RS: external request, single address mode
+				*/
+label_21c2:
 	mov.w	@r0+,r1
 	rts
 	mov.l	r1,@r2
+
 	mov.l	@r0+,r1
 	rts
 	mov.b	r1,@r2
@@ -357,50 +396,90 @@ label_2278:
 label_227c:
 	.long	0x00006000
 
-	/* unreachable? */
-	.word 0x8040
-	.word 0xa05f
+	.long	0xa05f8040 /* unused; VO_BORDER_COL */
 label_2284:
 	.word 0x0002
-	.word 0x0009
+label_2286:
+	.word 0x0009	 /* unused garbage data? */
+
+label_2288:
 	.word 0x0002
-	.word 0x0006
-	.word 0x0038
-	.word 0xff00
+label_228a:
+	.word 0x0006	/* r3 */
+label_228c:
+	.long 0xff000038
+
+label_2290:
 	.word 0x0001
-	.word 0x0018
-	.word 0x002c
-	.word 0xffa0
-	.word 0x12c0
-	.word 0x0000
+label_2292:
+	.word 0x0018	/* r3 */
+label_2294:
+	.long 0xffa0002c /* CHCR2 ← 0x000012c0 */
+label_2298:
+	.long 0x000012c0
+
+label_229c:
 	.word 0x0004
+label_229e:
 	.word 0x000c
-	.word 0x0000
-	.word 0xffd0
-	.word 0x0001
-	.word 0x0018
-	.word 0x0024
-	.word 0xff00
-	.word 0x0020
-	.word 0x0000
-	.word 0x0002
-	.word 0x001e
-	.word 0x7490
-	.word 0xa05f
+label_22a0:
+	.long 0xffd00000 /* ICR ← 0
+	                    IPRA ← 0
+	                    IPRB ← 0
+	                    IPRC ← 0
+	                 */
+label_22a4:
+	.word 0x0001     /* r4 */
+label_22a6:
+	.word 0x0018     /* r3 → bsrf label_21bc */
+label_22a8:
+	.long 0xff000024 /* r2
+			    ?? ← 0x00000020 */
+label_22ac:
+	.long 0x00000020 /* r1 */
+
+label_22b0:
+	.word 0x0002	 /* r4 */
+label_22b2:
+	.word 0x001e	 /* r3 → bsrf label_21c2 */
+label_22b4:
+	.long 0xa05f7490 /* r2
+			    SB_G1CRC ← 0x0222
+			    SB_G1CWC ← 0x0222
+	                  */
+label_22b8:
 	.word 0x0222
+label_22ba:
 	.word 0x0222
-	.word 0x0002
-	.word 0x001e
-	.word 0x74a0
-	.word 0xa05f
+
+label_22bc:
+	.word 0x0002	 /* r4 */
+label_22be:
+	.word 0x001e	 /* r3 → bsrf label_21c2 */
+label_22c0:
+	.long 0xa05f74a0 /* r2
+			    SB_G1GDRC ← 0x2001
+			    SB_G1GDWC ← 0x2001
+	*/
+label_22c4:
 	.word 0x2001
+label_22c6:
 	.word 0x2001
-	.word 0x0002
-	.word 0x001e
-	.word 0x7890
-	.word 0xa05f
+
+label_22c8:
+	.word 0x0002	/* r4 */
+label_22ca:
+	.word 0x001e	/* r3 → bsrf label_21c2 */
+label_22cc:
+	.long 0xa05f7890 /* SB_G2DSTO ← 0x001b
+			    SB_G2TRTO ← 0x0271
+			*/
+label_22d0:
 	.word 0x001b
+label_22d2:
 	.word 0x0271
+
+label_22d4:
 	.word 0x0000
 	.word 0x0000
 	nop
@@ -408,14 +487,16 @@ label_2284:
 	nop
 	nop
 
+	/* called from label_2020 */
 label_22e0:
 	sts.l	pr,@-r15
-	mov.l	label_2330,r4	! 8c008000
-	mov.l	label_2328,r3	! 8c00d8c6
+	/* check systemid for "uses Windows CE" */
+	mov.l	label_2330,r4	/* 8c008000 systemid */
+	mov.l	label_2328,r3	/* 8c00d8c6 label_20c6 */
 	jsr	@r3
-	add	#62,r4
-	tst	r0,r0
-	bt	label_2320
+	add	#0x3e,r4
+	tst	r0,r0		/* r0 (1 if Windows CE) */
+	bt	label_2320	/* if not Windows CE, return */
 	mov.l	label_2334,r1	! 8ce01010
 	mov.l	@r1,r0
 	tst	r0,r0
@@ -467,10 +548,11 @@ label_2340:
 	add	#-20,r15
 	mov	r15,r3
 	mov	r15,r5
-	mov.l	r3,@(16,r15)
+	mov.l	r3,@(16,r15)	/* copy stack pointer to stack */
 	add	#16,r5
 	bsr	label_23e0
 	mov	#30,r4
+
 	mov	r0,r4
 	tst	r4,r4
 	bt	label_2384
@@ -554,7 +636,7 @@ label_23e0:
 	mov	#0,r6
 	mov	#0,r7
 	mov.l	label_2404,r0	! 8c0000bc
-	mov.l	@r0,r0
+	mov.l	@r0,r0		/* r0 ← 0x8c001000 */
 label_23e8:
 	jmp	@r0
 	nop
@@ -562,14 +644,14 @@ label_23ec:
 	mov	#0,r6
 	mov	#1,r7
 	mov.l	label_2404,r0	! 8c0000bc
-	mov.l	@r0,r0
+	mov.l	@r0,r0		/* r0 ← 0x8c001000 */
 	jmp	@r0
 	nop
 label_23f8:
 	mov	#0,r6
 	mov	#2,r7
 	mov.l	label_2404,r0	! 8c0000bc
-	mov.l	@r0,r0
+	mov.l	@r0,r0		/* r0 ← 0x8c001000 */
 	jmp	@r0
 	nop
 label_2404:

systemid.s

@@ -34,11 +34,11 @@
 
 	.section .text.systemid
 
-	.ascii "SEGA SEGAKATANA " /* H/W identifier */
-	.ascii "SEGA ENTERPRISES" /* H/W Vendor ID */
-	.ascii "39F1 "            /* Media ID */
+	.ascii "SEGA SEGAKATANA " /* 0x00 H/W identifier */
+	.ascii "SEGA ENTERPRISES" /* 0x10 H/W Vendor ID */
+	.ascii "39F1 "            /* 0x20 Media ID */
 	.ascii      "GD-ROM1/1  " /* Media information */
-	.ascii "JUE     "         /* Compatible Area Symbol */
+	.ascii "JUE     "         /* 0x30 Compatible Area Symbol */
 	.ascii         "C000810 " /* Compatible peripherals */
 	.ascii "HDR-0900  "       /* Product number */
 	.ascii           "V0.000" /* Version number */

texture/macaw/macaw.data.h

@@ -0,0 +1,5 @@
+#pragma once
+#include <cstdint>
+extern uint32_t _binary_texture_macaw_macaw_data_start __asm("_binary_texture_macaw_macaw_data_start");
+extern uint32_t _binary_texture_macaw_macaw_data_end __asm("_binary_texture_macaw_macaw_data_end");
+extern uint32_t _binary_texture_macaw_macaw_data_size __asm("_binary_texture_macaw_macaw_data_size");

texture/macaw/macaw.hpp

@@ -1,5 +0,0 @@
-#include <cstdint>
-
-extern uint32_t _binary_macaw_data_start __asm("_binary_macaw_data_start");
-extern uint32_t _binary_macaw_data_end __asm("_binary_macaw_data_end");
-extern uint32_t _binary_macaw_data_size __asm("_binary_macaw_data_size");

tools/Makefile

@@ -1,16 +1,25 @@
 CFLAGS = -Og -g -gdwarf-4 -Wall -Wextra -Werror -Wfatal-errors -ggdb -Wno-error=unused-parameter -Wno-error=unused-variable -fstack-protector-strong
 CXXFLAGS = -std=c++23
 
-CFLAGS += $(shell pkg-config --cflags freetype2)
-LDFLAGS = $(shell pkg-config --libs freetype2)
+FREETYPE_CFLAGS = $(shell pkg-config --cflags freetype2)
+FREETYPE_LDFLAGS = $(shell pkg-config --libs freetype2)
+
+FTDI_CFLAGS = $(shell pkg-config --cflags libftdi1)
+FTDI_LDFLAGS = $(shell pkg-config --libs libftdi1)
 
 all: ttf_outline
 
-%.o: %.cpp
-	$(CXX) $(CFLAGS) $(CXXFLAGS) -c $< -o $@
+ttf_%.o: ttf_%.cpp
+	$(CXX) $(CFLAGS) $(FREETYPE_CFLAGS) $(CXXFLAGS) -c $< -o $@
 
-%: %.o
-	$(CXX) $(LDFLAGS) $^ -o $@
+ttf_%: ttf_%.o
+	$(CXX) $(LDFLAGS) $(FREETYPE_LDFLAGS) $^ -o $@
+
+ftdi_%.o: ftdi_%.c
+	$(CC) -std=gnu2x $(CFLAGS) $(FTDI_CFLAGS) -c $< -o $@
+
+ftdi_%: ftdi_%.o
+	$(CXX) $(LDFLAGS) $(FTDI_LDFLAGS) $^ -o $@
 
 ttf_outline: ttf_outline.o 2d_pack.o
 
@@ -21,3 +30,9 @@ clean:
 .INTERMEDIATE:
 .SECONDARY:
 .PHONY: all clean
+
+%: RCS/%,v
+%: RCS/%
+%: %,v
+%: s.%
+%: SCCS/s.%

tools/ftdi_transfer.c

@@ -0,0 +1,312 @@
+#include <assert.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <time.h>
+#include <string.h>
+
+#include <ftdi.h>
+#include <libusb.h>
+
+extern int convert_baudrate_UT_export(int baudrate, struct ftdi_context *ftdi,
+				      unsigned short *value, unsigned short *index);
+
+int dreamcast_rates[] = {
+  1562500,
+  781250,
+  520833,
+  390625,
+  312500,
+  260416,
+  223214,
+  195312,
+  173611,
+  156250,
+  142045,
+  130208,
+  120192
+};
+
+int init_ftdi_context(struct ftdi_context * ftdi)
+{
+  ftdi_set_interface(ftdi, INTERFACE_ANY);
+  struct ftdi_device_list * devlist;
+  int res;
+  if ((res = ftdi_usb_find_all(ftdi, &devlist, 0, 0)) < 0) {
+    fprintf(stderr, "ftdi_usb_find_all\n");
+    return -1;
+  }
+
+  if (res == 0) {
+    fprintf(stderr, "no device\n");
+    return -1;
+  }
+
+  struct libusb_device_descriptor desc;
+  struct ftdi_device_list * devlist_item = devlist;
+  for (int i = 0; i < res; i++) {
+    res = libusb_get_device_descriptor(devlist_item->dev, &desc);
+    if (res < 0) {
+      fprintf(stderr, "libusb_get_device_descriptor\n");
+      return -1;
+    }
+    fprintf(stdout, "idVendor: %04x; idProduct: %04x;\n", desc.idVendor, desc.idProduct);
+    fprintf(stdout, "bNumConfigurations: %d;\n", desc.bNumConfigurations);
+    devlist_item = devlist_item->next;
+  }
+
+  res = ftdi_usb_open_dev(ftdi, devlist->dev);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_usb_open_dev\n");
+    return -1;
+  }
+  ftdi_list_free(&devlist);
+
+  unsigned short value;
+  unsigned short index;
+
+  for (unsigned int i = 0; i < (sizeof (dreamcast_rates)) / (sizeof (dreamcast_rates[0])); i++) {
+    int baud = convert_baudrate_UT_export(dreamcast_rates[i], ftdi, &value, &index);
+    float baudf = baud;
+    float ratef = dreamcast_rates[i];
+    float error = (baudf > ratef) ? ratef / baudf : baudf / ratef;
+    fprintf(stdout, "%d: best: %d, error: %f\n", dreamcast_rates[i], baud, (1.f - error) * 100.f);
+  }
+
+  res = ftdi_set_baudrate(ftdi, 1562500);
+  //res = ftdi_set_baudrate(ftdi, 312500);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_set_baudrate\n");
+    return -1;
+  }
+
+  res = ftdi_set_line_property(ftdi, 8, STOP_BIT_1, NONE);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_set_line_property\n");
+    return -1;
+  }
+
+  res = ftdi_set_latency_timer(ftdi, 1);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_set_latency_timer\n");
+    return -1;
+  }
+
+  res = ftdi_tciflush(ftdi);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_tciflush\n");
+    return -1;
+  }
+
+  res = ftdi_tcoflush(ftdi);
+  if (res < 0) {
+    fprintf(stderr, "ftdi_tcoflush\n");
+    return -1;
+  }
+
+
+  return 0;
+}
+
+union data_command {
+  struct {
+    uint8_t command[4];
+    uint32_t size;
+    uint32_t dest;
+  };
+  uint8_t data[4 * 3];
+};
+static_assert((sizeof (union data_command)) == 4 * 3);
+
+uint32_t bswap(const uint32_t n)
+{
+  if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
+    return n;
+  else
+    return __builtin_bswap32(n);
+}
+
+long read_with_timeout(struct ftdi_context * ftdi, uint8_t * read_buf, const long expect_length)
+{
+  int res;
+  struct timespec tp0;
+  res = clock_gettime(CLOCK_MONOTONIC, &tp0);
+  assert(res >= 0);
+
+  long read_length = 0;
+  while (true) {
+    res = ftdi_read_data(ftdi, read_buf, expect_length - read_length);
+    assert(res >= 0);
+
+    read_length += res;
+    if (read_length >= expect_length)
+      break;
+
+    struct timespec tp1;
+    res = clock_gettime(CLOCK_MONOTONIC, &tp1);
+    assert(res >= 0);
+
+    if (tp1.tv_sec - tp0.tv_sec > 1) {
+      fprintf(stderr, "read timeout: %ld expect: %ld\n", read_length, expect_length);
+      break;
+    }
+  }
+  return read_length;
+}
+
+const int chunk_size = 1024;
+
+long min(long a, long b)
+{
+  return a > b ? b : a;
+}
+
+long max(long a, long b)
+{
+  return a > b ? a : b;
+}
+
+void symmetric(struct ftdi_context * ftdi, const uint8_t * tx_buf, const long size)
+{
+  int res;
+  uint8_t rx_buf[size];
+  long tx_offset = 0;
+  long rx_offset = 0;
+
+  while (tx_offset < size) {
+    long txrx_diff = tx_offset - rx_offset;
+    long tx_length = max(min(min(chunk_size, size - tx_offset), chunk_size - txrx_diff), 0);
+
+    if (tx_length > 0) {
+      res = ftdi_write_data(ftdi, &tx_buf[tx_offset], tx_length);
+      assert(res >= 0);
+      tx_offset += res;
+    }
+
+    res = ftdi_read_data(ftdi, &rx_buf[rx_offset], size - rx_offset);
+    assert(res >= 0);
+    rx_offset += res;
+  }
+
+  for (int i = 0; i < size; i++) {
+    if (tx_buf[i] != rx_buf[i]) {
+      fprintf(stderr, "mismatch at %d\n", i);
+      return;
+    }
+  }
+  fprintf(stderr, "equal\n");
+}
+
+double timespec_difference(struct timespec const * const a, struct timespec const * const b)
+{
+  return (double)(a->tv_sec - b->tv_sec) + (double)(a->tv_nsec - b->tv_nsec) / 1'000'000'000.0;
+}
+
+int transfer(struct ftdi_context * ftdi, const uint8_t * buf, const long size)
+{
+  int res;
+
+  union data_command command = {
+    .command = {'D', 'A', 'T', 'A'},
+    .size = bswap(size),
+    .dest = bswap(0xac010000),
+  };
+
+  res = ftdi_write_data(ftdi, command.data, (sizeof (union data_command)));
+  assert(res >= 0);
+
+  const char * expect = "data\n";
+  const long expect_length = 5;
+  uint8_t read_buf[expect_length + 1];
+  read_buf[expect_length] = 0;
+  long read_length = read_with_timeout(ftdi, read_buf, expect_length);
+  if (read_length != expect_length) {
+    fprintf(stderr, "want %ld bytes; received: %ld\n", expect_length, read_length);
+    return -1;
+  }
+  res = memcmp(read_buf, expect, expect_length);
+  if (res != 0) {
+    fprintf(stderr, "expect `%s`; received: `%s`\n", expect, read_buf);
+    return -1;
+  }
+
+  fprintf(stderr, "OK\n");
+
+  struct timespec start;
+  struct timespec end;
+  res = clock_gettime(CLOCK_MONOTONIC, &start);
+  symmetric(ftdi, buf, size);
+  res = clock_gettime(CLOCK_MONOTONIC, &end);
+  fprintf(stderr, "symmetric time: %.03f\n", timespec_difference(&end, &start));
+
+
+  return 0;
+}
+
+int main(int argc, char * argv[])
+{
+  if (argc < 2) {
+    fprintf(stderr, "argc\n");
+    return EXIT_FAILURE;
+  }
+
+  FILE * file = fopen(argv[1], "r");
+  if (file == NULL) {
+    fprintf(stderr, "fopen\n");
+    return EXIT_FAILURE;
+  }
+
+  int ret;
+  ret = fseek(file, 0L, SEEK_END);
+  if (ret < 0) {
+    fprintf(stderr, "seek(SEEK_END)");
+    return EXIT_FAILURE;
+  }
+
+  long off = ftell(file);
+
+  ret = fseek(file, 0L, SEEK_SET);
+  if (ret < 0) {
+    fprintf(stderr, "seek(SEEK_SET)");
+    return EXIT_FAILURE;
+  }
+
+  fprintf(stderr, "%s off %ld\n", argv[1], off);
+  uint8_t buf[off];
+  ssize_t size = fread(buf, 1, off, file);
+  if (size < 0) {
+    fprintf(stderr, "read");
+    return EXIT_FAILURE;
+  }
+
+  ret = fclose(file);
+  if (ret < 0) {
+    fprintf(stderr, "close");
+    return EXIT_FAILURE;
+  }
+
+  struct ftdi_context * ftdi;
+
+  ftdi = ftdi_new();
+  if (ftdi == 0) {
+    fprintf(stderr, "ftdi_new\n");
+    return EXIT_FAILURE;
+  }
+
+  int res;
+  res = init_ftdi_context(ftdi);
+  if (res < 0) {
+    return EXIT_FAILURE;
+  }
+
+  struct timespec start;
+  struct timespec end;
+  res = clock_gettime(CLOCK_MONOTONIC, &start);
+  int transfer_ret = transfer(ftdi, buf, off);
+  res = clock_gettime(CLOCK_MONOTONIC, &end);
+
+  fprintf(stderr, "time: %.03f\n", timespec_difference(&end, &start));
+
+  return transfer_ret;
+}