add utf-8 to codepage mapping, change some asm syntax

assembler-8608.lua

@@ -1,13 +1,25 @@
 
 local arch8608 = require("rom-8608-defs")
 
-local aliases = {
+local function loadutf8table(fn)
-	["jpz imm8"] = {"jeq imm8"},
+	local tt = {}
-	["jnz imm8"] = {"jne imm8"},
+	for l in io.lines(fn) do if l~="" then
-	["jmp q"   ] = {"ret"     },
+		local c, d = l:match("^([^ ]+) (.+)$")
-}
+		local t = {}; for v in d:gmatch("[^ ]+") do table.insert(t, tonumber(v, 16)) end;
+		tt[c] = t
+	end end
+	return tt
+end
+local utf8table = loadutf8table("./utf8table.txt")
 
 local function trim(s) return s:gsub("^ +", ""):gsub(" +$", "").."" end
+local function getutf8len(c)
+	local d = c:byte()
+	if     bit.band(d, 0xE0)==0xC0 then return 2
+	elseif bit.band(d, 0xF0)==0xE0 then return 3
+	elseif bit.band(d, 0xF8)==0xF0 then return 4
+	else error("invalid utf8 first byte: "..string.format("%02X", d)) end
+end
 local function validWordsFromInstrs(instrs)
 	local words = {}
 	for mnem, _ in pairs(instrs) do
@@ -303,6 +315,9 @@ local function prefixCode(code, fn) -- fix strings, add line numbers
 	local bracehasmid = {}
 	local lastnl = false
 	
+	local utf8str = ""
+	local utf8len = 0
+	
 	out(".ln 1"); out("\n");
 	for i = 1, #code do
 		local c = code:sub(i, i)
@@ -349,9 +364,22 @@ local function prefixCode(code, fn) -- fix strings, add line numbers
 		elseif state=="string" then
 			if     c=="\\"       then state = "stringesc"
 			elseif c=="\""       then state = "code"
+			elseif c:byte()>=128 then
+				utf8str = c
+				utf8len = getutf8len(c)
+				state = "stringutf8"
 			else outn(c:byte()) end
 		elseif state=="stringesc" then
 			outn(string.byte(stringEscapes[c] or error("invalid escape "..c))); state = "string";
+		elseif state=="stringutf8" then
+			utf8str = utf8str..c
+			if #utf8str == utf8len then
+				local valt = utf8table[utf8str]
+				if not valt then local datastr = ""; for i = 1, #utf8str do datastr = datastr .. string.format("%02X ", utf8str:sub(i, i):byte()) end;
+					error("Unrecognized UTF-8 character: "..datastr); end
+				for i, v in ipairs(valt) do outn(v) end
+				state = "string"
+			end
 		end
 	end
 	assert(#bracestack==0, "unclosed brace")
@@ -397,7 +425,7 @@ local function instrsFromArch(arch)
 			mnem = mnem:gsub("([%*%+%-])", " %1 ")
 			mnem = trim(mnem):gsub(" +", " ")
 			addMnem(mnem, instr.opcode)
-			local alias = aliases[trim(mnem)]
+			local alias = arch.aliases[trim(mnem)]
 			if alias then for _, v in ipairs(alias) do addMnem(v, instr.opcode) end end
 		end
 	end
@@ -566,13 +594,14 @@ local function strtovec(str) local v = {}; for word in str:gmatch("[^ \t\r\n]+")
 function AssembleFile(fn, romsizes, offsets, lens) local offset = tonumber(offsets); local len = tonumber(lens); local romsize = strtovec(romsizes);
 	local arch = arch8608
 	local mem = assembleFile(fn, arch)
+	print(""..fn:match("[^/\\]+$").."\n")
 	printMemory(mem)
 	assert(#romsize==3, "incorrect rom size")
 	buildMemory(mem, romsize, offset, len)
 	disassembleMemory(mem, arch)
 end
 ts.eval [[
-	function AssembleFile(%fn, %romsize, %offset, %len) { luacall("AssembleFile", %fn, %romsize, %offset, %len); }
+	function AssembleFile(%fn, %romsize, %offset, %len) { luacall("AssembleFile", "Add-ons/_misc/rom/8608programs/" @ %fn, %romsize, %offset, %len); }
 ]]
 
 if arg then AssembleFile(arg[1] or "../8608programs/test.asm", "16 16 8", "0", "256") end

instructionList.txt

@@ -5,12 +5,13 @@ hlt          F0  1  Halt non-interrupt execution
 run          F1  1  Resume non-interrupt execution
 brk          F3  1  Trigger interrupt
 irt          F4  1  Return from interrupt
+nop          FF  1  Do nothing
 
 16-bit Inc/Dec (I):
-icp          12  1  P++
+inc p        12  1  P++
-dcp          15  1  P--
+dec p        15  1  P--
-icq          13  1  Q++
+inc q        13  1  Q++
-dcq          16  1  Q--
+dec q        16  1  Q--
 
 8-bit Unary (U):
 inc a        10  1  A++, set flags
@@ -125,16 +126,18 @@ jgt imm8     35  2  I+=imm8 if !Zero & Carry
 jle imm8     36  2  I+=imm8 if Zero | !Carry
 
 Stack (S):
-pha          40  2  *(S++)=A
+psh a        40  2  *(S++)=A
-phb          44  2  *(S++)=B
+psh b        44  2  *(S++)=B
-phc          45  2  *(S++)=C
+psh c        45  2  *(S++)=C
-php          41  3  *(S++++)=P
+psh p        41  3  *(S++++)=P
-phq          46  3  *(S++++)=Q
+psh q        46  3  *(S++++)=Q
-ppa          42  2  A=*(--S)
+pop a        42  2  A=*(--S)
-ppb          47  2  B=*(--S)
+pop b        47  2  B=*(--S)
-ppc          48  2  C=*(--S)
+pop c        48  2  C=*(--S)
-ppp          43  3  P=*(----S)
+pop p        43  3  P=*(----S)
-ppq          49  3  Q=*(----S)
+pop q        49  3  Q=*(----S)
+psh imm8     3B  3  *(S++)=imm8
+psh imm16    3C  5  *(S++++)=imm16
 
 8-bit Load/Store (B):
 lda imm8     20  2  A=imm8, update zero flag
@@ -212,21 +215,26 @@ lda pl       86  1  A=P&FF
 lda ph       87  1  A=P>>8
 lda ql       88  1  A=Q&FF
 lda qh       89  1  A=Q>>8
+ldb pl       37  1  B=P&FF
+ldc ph       38  1  C=P>>8
+ldb ql       39  1  B=Q&FF
+ldc qh       3A  1  C=Q>>8
 ldp q        8A  1  P=Q
 ldp s        8B  1  P=S
 ldp v        8C  1  P=V
 ldp i        8D  1  P=I
-ldp cb       91  1  P=C<<8+B
+ldp cb       91  1  P=(C<<8)+B
+ldq cb       E0  1  Q=(C<<8)+B
 ldq p        8E  1  Q=P
 lds p        8F  1  S=P
 ldv p        90  1  V=P
 
-Opcodes used: 207/255
+Opcodes used: 216/255
      0123456789ABCDEF
 00 | C---------------
 10 | UUIIUIIUUUUUUUUU
 20 | BWWWAWBBBBBUUUUA
-30 | JJJJJJJ-------AA
+30 | JJJJJJJMMMMSSSAA
 40 | SSSSSSSSSSXXXAAA
 50 | BBBBBBBBBBBBBBBB
 60 | JBBJJJJJWWWWWWWW
@@ -237,5 +245,5 @@ A0 | AAAAAAAAAAAAAAAA
 B0 | AAAAAAAAAAAAAAAA
 C0 | BBBBBBBBBBBBWWWW
 D0 | AAAAAAAAAAAAAAAA
-E0 | ----------------
+E0 | M---------------
-F0 | CCCCC-----------
+F0 | CCCCC----------C

rom-8608-defs.lua

@@ -127,12 +127,13 @@ instructions = {
 	{ mnem="int"        , opcode=0xF2, {"instrSwapIV","intFlgVal","intFlgClk","irqFlgClk"}, },
 	{ mnem="brk"        , opcode=0xF3, {"instrSwapIV","adwInc","intFlgVal","intFlgClk"}, desc="Trigger interrupt" },
 	{ mnem="irt"        , opcode=0xF4, {"instrSwapIV","adwInc","intFlgClk"}, desc="Return from interrupt" },
+	{ mnem="nop"        , opcode=0xFF, {"instrNext"}, desc="Do nothing" },
 	
 	{ category = "16-bit Inc/Dec", catlet="I" },
-	{ mnem="icp"        , opcode=0x12, {"adwlP","adwInc","adwSaveP","instrNext"}, desc="P++" },
+	{ mnem="inc p"      , opcode=0x12, {"adwlP","adwInc","adwSaveP","instrNext"}, desc="P++" },
-	{ mnem="dcp"        , opcode=0x15, {"adwlP","adwrm1","adwSaveP","instrNext"}, desc="P--" },
+	{ mnem="dec p"      , opcode=0x15, {"adwlP","adwrm1","adwSaveP","instrNext"}, desc="P--" },
-	{ mnem="icq"        , opcode=0x13, {"adwlQ","adwInc","adwSaveQ","instrNext"}, desc="Q++" },
+	{ mnem="inc q"      , opcode=0x13, {"adwlQ","adwInc","adwSaveQ","instrNext"}, desc="Q++" },
-	{ mnem="dcq"        , opcode=0x16, {"adwlQ","adwrm1","adwSaveQ","instrNext"}, desc="Q--" },
+	{ mnem="dec q"      , opcode=0x16, {"adwlQ","adwrm1","adwSaveQ","instrNext"}, desc="Q--" },
 	
 	{ category = "8-bit Unary", catlet="U" },
 	{ mnem="inc a"      , opcode=0x10,                                                            {"aluA","alur1" ,"aluOpAdd" ,"instrNext"}, desc="A++, set flags" },
@@ -247,16 +248,18 @@ instructions = {
 	{ mnem="jle imm8"   , opcode=0x36, ncycles=2, {"loadImmed","memSaveT","instrSub23Cond","instrNext0NC","instrNext0Z"}, {}, {"instrNext"}, {"jmpRelT"}, desc="I+=imm8 if Zero | !Carry"   },
 	
 	{ category = "Stack", catlet="S" },
-	{ mnem="pha"        , opcode=0x40, {"pushReg","alurA","instrSub1"}, {"instrNext"}, desc="*(S++)=A" },
+	{ mnem="psh a"      , opcode=0x40, {"pushReg","alurA","instrSub1"}, {"instrNext"}, desc="*(S++)=A" },
-	{ mnem="phb"        , opcode=0x44, {"pushReg","alurB","instrSub1"}, {"instrNext"}, desc="*(S++)=B" },
+	{ mnem="psh b"      , opcode=0x44, {"pushReg","alurB","instrSub1"}, {"instrNext"}, desc="*(S++)=B" },
-	{ mnem="phc"        , opcode=0x45, {"pushReg","alurC","instrSub1"}, {"instrNext"}, desc="*(S++)=C" },
+	{ mnem="psh c"      , opcode=0x45, {"pushReg","alurC","instrSub1"}, {"instrNext"}, desc="*(S++)=C" },
-	{ mnem="php"        , opcode=0x41, {"pushReg","alurPH","instrSub1"}, {"pushReg","alurPL","instrSub2"}, {"instrNext"}, desc="*(S++++)=P" },
+	{ mnem="psh p"      , opcode=0x41, {"pushReg","alurPH","instrSub1"}, {"pushReg","alurPL","instrSub2"}, {"instrNext"}, desc="*(S++++)=P" },
-	{ mnem="phq"        , opcode=0x46, {"pushReg","alurQH","instrSub1"}, {"pushReg","alurQL","instrSub2"}, {"instrNext"}, desc="*(S++++)=Q" },
+	{ mnem="psh q"      , opcode=0x46, {"pushReg","alurQH","instrSub1"}, {"pushReg","alurQL","instrSub2"}, {"instrNext"}, desc="*(S++++)=Q" },
-	{ mnem="ppa"        , opcode=0x42, {"popReg","memSaveA","instrSub1"}, {"instrNext"}, desc="A=*(--S)" },
+	{ mnem="pop a"      , opcode=0x42, {"popReg","memSaveA","instrSub1"}, {"instrNext"}, desc="A=*(--S)" },
-	{ mnem="ppb"        , opcode=0x47, {"popReg","memSaveB","instrSub1"}, {"instrNext"}, desc="B=*(--S)" },
+	{ mnem="pop b"      , opcode=0x47, {"popReg","memSaveB","instrSub1"}, {"instrNext"}, desc="B=*(--S)" },
-	{ mnem="ppc"        , opcode=0x48, {"popReg","memSaveC","instrSub1"}, {"instrNext"}, desc="C=*(--S)" },
+	{ mnem="pop c"      , opcode=0x48, {"popReg","memSaveC","instrSub1"}, {"instrNext"}, desc="C=*(--S)" },
-	{ mnem="ppp"        , opcode=0x43, {"pop161","instrSub1"}, {"pop162","instrSub2"}, {"adwrUT","adwSaveP","instrNext"}, desc="P=*(----S)" },
+	{ mnem="pop p"      , opcode=0x43, {"pop161","instrSub1"}, {"pop162","instrSub2"}, {"adwrUT","adwSaveP","instrNext"}, desc="P=*(----S)" },
-	{ mnem="ppq"        , opcode=0x49, {"pop161","instrSub1"}, {"pop162","instrSub2"}, {"adwrUT","adwSaveQ","instrNext"}, desc="Q=*(----S)" },
+	{ mnem="pop q"      , opcode=0x49, {"pop161","instrSub1"}, {"pop162","instrSub2"}, {"adwrUT","adwSaveQ","instrNext"}, desc="Q=*(----S)" },
+	{ mnem="psh imm8"   , opcode=0x3B, {"loadImmedT","instrSub1"}, {"pushReg","alurT","instrSub2"}, {"instrNext"}, desc="*(S++)=imm8" },
+	{ mnem="psh imm16"  , opcode=0x3C, {"loadImm161","instrSub1"}, {"loadImm162","instrSub2"}, {"pushReg","alurU","instrSub3"}, {"pushReg","alurT","instrSub4"}, {"instrNext"}, desc="*(S++++)=imm16" }, -- 0x3D
 	
 	{ category = "8-bit Load/Store", catlet="B" },
 	{ mnem="lda imm8"   , opcode=0x20,                             {"loadImmed",   "memSaveA","memSaveFlags","instrSub1"}, {"instrNext"}, desc="A=imm8, update zero flag" },
@@ -334,14 +337,25 @@ instructions = {
 	{ mnem="lda ph"     , opcode=0x87, {"alurPH","aluOpMov","aluSaveA","instrNext"}, desc="A=P>>8" },
 	{ mnem="lda ql"     , opcode=0x88, {"alurQL","aluOpMov","aluSaveA","instrNext"}, desc="A=Q&FF" },
 	{ mnem="lda qh"     , opcode=0x89, {"alurQH","aluOpMov","aluSaveA","instrNext"}, desc="A=Q>>8" },
+	{ mnem="ldb pl"     , opcode=0x37, {"alurPL","aluOpMov","aluSaveB","instrNext"}, desc="B=P&FF" },
+	{ mnem="ldc ph"     , opcode=0x38, {"alurPH","aluOpMov","aluSaveC","instrNext"}, desc="C=P>>8" },
+	{ mnem="ldb ql"     , opcode=0x39, {"alurQL","aluOpMov","aluSaveB","instrNext"}, desc="B=Q&FF" },
+	{ mnem="ldc qh"     , opcode=0x3A, {"alurQH","aluOpMov","aluSaveC","instrNext"}, desc="C=Q>>8" },
 	{ mnem="ldp q"      , opcode=0x8A, {"adwlQ" ,           "adwSaveP","instrNext"}, desc="P=Q" },
 	{ mnem="ldp s"      , opcode=0x8B, {"adwlS" ,           "adwSaveP","instrNext"}, desc="P=S" },
 	{ mnem="ldp v"      , opcode=0x8C, {"adwlV" ,           "adwSaveP","instrNext"}, desc="P=V" },
 	{ mnem="ldp i"      , opcode=0x8D, {"adwlI" ,           "adwSaveP","instrNext"}, desc="P=I" },
-	{ mnem="ldp cb"     , opcode=0x91, {"adwrCB",           "adwSaveP","instrNext"}, desc="P=C<<8+B" },
+	{ mnem="ldp cb"     , opcode=0x91, {"adwrCB",           "adwSaveP","instrNext"}, desc="P=(C<<8)+B" },
+	{ mnem="ldq cb"     , opcode=0xE0, {"adwrCB",           "adwSaveQ","instrNext"}, desc="Q=(C<<8)+B" },
 	{ mnem="ldq p"      , opcode=0x8E, {"adwlP" ,           "adwSaveQ","instrNext"}, desc="Q=P" },
 	{ mnem="lds p"      , opcode=0x8F, {"adwlP" ,           "adwSaveS","instrNext"}, desc="S=P" },
 	{ mnem="ldv p"      , opcode=0x90, {"adwlP" ,           "adwSaveV","instrNext"}, desc="V=P" },
 },
 
+aliases = {
+	["jpz imm8"] = {"jeq imm8"},
+	["jnz imm8"] = {"jne imm8"},
+	["jmp q"   ] = {"ret"     },
+},
+
 }

utf8table.txt

@@ -0,0 +1,88 @@
+。 A1
+「 A2
+」 A3
+、 A4
+・ A5
+ヲ A6
+ァ A7
+ィ A8
+ゥ A9
+ェ AA
+ォ AB
+ャ AC
+ュ AD
+ョ AE
+ッ AF
+ー B0
+ア B1
+イ B2
+ウ B3
+エ B4
+オ B5
+カ B6
+キ B7
+ク B8
+ケ B9
+コ BA
+サ BB
+シ BC
+ス BD
+セ BE
+ソ BF
+タ C0
+チ C1
+ツ C2
+テ C3
+ト C4
+ナ C5
+ニ C6
+ヌ C7
+ネ C8
+ノ C9
+ハ CA
+ヒ CB
+フ CC
+ヘ CD
+ホ CE
+マ CF
+ミ D0
+ム D1
+メ D2
+モ D3
+ヤ D4
+ユ D5
+ヨ D6
+ラ D7
+リ D8
+ル D9
+レ DA
+ロ DB
+ワ DC
+ン DD
+゛ DE
+゜ DF
+ガ B6 DE
+ギ B7 DE
+グ B8 DE
+ゲ B9 DE
+ゴ BA DE
+ザ BB DE
+ジ BC DE
+ズ BD DE
+ゼ BE DE
+ゾ BF DE
+ダ C0 DE
+ヂ C1 DE
+ヅ C2 DE
+デ C3 DE
+ド C4 DE
+バ CA DE
+ビ CB DE
+ブ CC DE
+ベ CD DE
+ボ CE DE
+パ CA DF
+ピ CB DF
+プ CC DF
+ペ CD DF
+ポ CE DF