This is my Dreamcast stuff.

My major reference for this stuff is Marcus Comstedt's extremely useful
and helpful stuff from http://mc.pp.se/dc/.  (He has been very kind and
helpful to me throughout this exercise; I might have been able to do
this without any assitance from him beyond tatest itself, but at the
very least it would have taken substantially longer.)

My first goal here was to get his tatest program, which is written in
C, translated to assembly and running.  I've now managed that and am
moving on to other things, such as figuring out how to use aspects of
the 3D hardware that tatest _doesn't_ use.

This stuff is available (at this writing) via git clone from
git://sparkle.rodents-montreal.org/Mouse/dreamcast.  At the moment it's
not available any other way without manual assitance; based on the
level of interest I've seen on NetBSD's port-dreamcast list, it seems
unlikely anyone will care, but if anyone does, I can arrange access,
either as a one-shot or more generally.

There are multiple things here.  The list at this writing follows after
my signature.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse@rodents-montreal.org
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B

.gitignore
	git overhead.

16-to-24.c
	Program to convert 16bpp iamge data to 24bpp.  This program is
	very experimental.

3d.s
	Playing around with the 3D hardware.

Makefile
	Makefile for building the assembler.  Doesn't build anything
	else; in particular, doesn't run the assembler on any of the
	various .s files.

all-instructions.s
	List of all 61217 opcodes the SH supports (as far as I know).
	This is less than 64K because some bit patterns don't
	correspond to instructions; it is more than the number of valid
	bit patterns, because bit patterns that are valid but with
	different semantics under multiple conditions are listed
	multiple times (for example, 0xffe6 is "fmov.s @(r0,r14),fr15"
	if fpscr.sz is clear, but "fmov @(r0,r14),xd14" if it's set).

buildbin
	Script to automate the process of assembling a .s file and
	converting the result to a .bin to burn to a CD-R.  This refers
	to executables in a directory outside this tree, so it probably
	will not be much use to you directly.

cdcode-.s
	Former version of cdcode.s.  Human reference only - doesn't
	even assemble under the current assembler.

cdcode.s
	The code I've burnt to a CD-R and generally boot my DC from.
	This is basically a "ROM monitor" except it runs from RAM.
	It's substantially more capable than Marcus's serial upload
	slave.

cdtest1.s through cdtest8.s
	Various programs of possible historical interest which I
	test-booted from CD while developing what is now cdcode.s.

controller.txt
maple.txt
pvr.txt
	Text documentation distilled from Marcus's webpages, mostly by
	just stripping out HTML uglification.

ta-intro.txt
	A copy of http://www.ludd.luth.se/~jlo/dc/ta-intro.txt, saved
	here for my reference.  (Simply using the Web version is not
	suitable, because Web browsers are poor-to-bad as text reading
	tools, because I'm not always on the net when I'm working on
	this stuff, and because versions of things kept off my machines
	have a nasty habit of disappearing.)

dumprom.c
	A program to be run, stdin and stdout connected to the serial
	line to cdcode, to dump out the DC's ROM.  It writes the result
	to "rom.dump".

expand-opcode.c
	Takes instruction summaries with opcode substitutions such as

	cmp/eq  rM,rN           ; 0011nnnnmmmm0000

	and expands them into multiple lines like

	cmp/eq  r0,r0           ; 0011000000000000
	cmp/eq  r0,r1           ; 0011000100000000
	...
	cmp/eq  r15,r15         ; 0011111111110000

find-conflicts.c
	Takes instructions formatted as in all-instructions.s and looks
	for cases where a single bit pattern has multiple
	interpretations.  This was designed to help find typos in my
	transcription of the documentation PDFs into a usable form.

font.s
	A peecee 8x8 console font, converted to .byte directives.

lcd.s
	Speaks to a controller in port A with a VMU in slot 1, moving a
	single dot on the VMU's display to match the controller's
	joystick position.

main.c
parse-action.h
parse-fsm.h
parse.fsm
	My Dreamcast assembler.  This is mostly just a Super-H
	assembler at present, but I have plans to teach it about the
	VMU and possibly ARM instruction sets so I can use it to write
	code for the corresponding hardware.

	You will need at least three other tools to build this.

	The first is gcc.  I don't write in C; I write in gcc
	(actually, gcc with an extension; see below).  Some OSes,
	including most of the open-source ones, come with gcc; it's
	available for almost everything.

	The second is fsm, which converts parse.fsm (a finite state
	machine written in a language loosely patterned after VMS's
	LIB$TPARSE facility) to C code.  This you can get by git clone
	of git://sparkle.rodents-montreal.org/fsm or by anonymous FTP
	from ftp.rodents-montreal.org:/mouseware/local-src/fsm/.

	The third is lcs-cvt.  I actually write in a somewhat extended
	dialect of gcc: gcc plus what I call labeled control structure
	("LCS").  On my systems, LCS is in the compiler.  (I'll be
	happy to send my patches to gcc to add LCS to anyone
	interested.)  But, for the sake of portability to people who
	don't care to hack up their gccs, I have a conversion program
	(lcs-cvt) which runs over preprocessor output, coverting LCS to
	semantically equivalent but substantially less human-friendly
	constructs.  You can get this by anonymous FTP from
	ftp.rodents-montreal.org:/mouseware/local-src/lcs-cvt/; you
	would use it as in

	gcc -E -o foo.i main.c
	lcs-cvt foo.i
	gcc -c -o main.o foo.i

	(with, of course, whatever options and paths are aoppropriate
	for your environment).

	If your system does not have funopen() (which mostly means, if
	you don't have a Torek stdio - Linux is perhaps the commonest
	case), you will need a workalike.  If you can't find anything
	of the sort, you will have a substantial coding effort on your
	hands, because some important parts of the assembler depend on
	it fairly critically.  Most Linux versions have fopencookie;
	while it cannot be used as directly as it might appear, because
	the signature for the seek function doesn't match, it is close
	enough that a naïve wrapper works for the uses I make of it.
	You can find the wrapper I use at
	ftp.rodents-montreal.org:/mouseware/local-src/liblinuxcompat/.

maple-bits.s
	Register addresses and bit definitions and such for the maple
	bus interface hardware.  This is included from the maple-using
	programs.

maple.s
	A first, rudimentary, program to interact with maple-bus
	peripherals.  This just prints the state retrieved from the
	controller, as fast as it can (usually the serial line is the
	limiting factor).  This is useful to see how that state changes
	as you manipulate the controller.

notes
	Some notes on the Super-H instruction set, including page
	numbers in the programmer's reference PDF.

o-to-s
	Takes an S-record file as produced by my assembler and converts
	it to .byte directives in a .text segment for NetBSD
	cross-compilation tools.

regs-BSC.s
regs-CCN.s
regs-CPG.s
regs-CPU.s
regs-DMAC.s
regs-INTC.s
regs-Px.s
regs-RTC.s
regs-SCI.s
regs-SCIF.s
regs-TMU.s
regs-UBC.s
regs-UDI.s
	Assembler definitions related to various pieces of the
	hardware.  These will generally be included by code using the
	relevant hardware.

regs.s
	Simple file which includes all the regs-*.s files.

rom.dis
rom.dump
	Dump and disassembler file for the ROMs in my DC.  (The
	disassembler file is not very useful unless you have my
	disassembler; to discuss that here would be going too far
	afield.  See ftp.rodents-montreal.org:/mouse/disas/ if you're
	interested in it.)

send-s.c
	Program to run on the host, conencted to the serial line to the
	DC, to send an S-record file such as produced by my assembler
	to the DC.

srtest1.s
	An early serial-line test program.  Of historical interest only
	at this point.

ta-cmds.s
	Registers and bit definitions of interest to code using the TA.

tatest.s
	My conversion of Marcus's tatest program to assembly.

test.s
	A file used to test the assembler itself.  In most revisions,
	this contains test code for whatever the latest feature added
	to the assembler is.  (This is not a proper test suite.)

trig-series.c
	A host program to see how the power series for sin(x) and
	cos(x) behave numerically.  This dates from a time when I was
	unaware of the fsca instruction and was doing sines and cosines
	on the DC by table lookup, using the power series to compute
	the tables.

trig.s
	A DC program to compare the power series for sin(x) with the
	DC's fsca instruction.  Prints out 2K lines, each of which
	contains:

	power-series value
	fsca value
	difference between the previous two
	XOR of the bit patterns for the first two values