An emulator, assembler, and disassembler for the Sega Game Gear
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

update_asm_instructions.py 14 KiB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393
  1. #!/usr/bin/env python
  2. # -*- coding: utf-8 -*-
  3. # Copyright (C) 2014-2015 Ben Kurtovic <ben.kurtovic@gmail.com>
  4. # Released under the terms of the MIT License. See LICENSE for details.
  5. """
  6. This script generates 'src/assembler/instructions.inc.c' from
  7. 'src/assembler/instructions.yml'. It should be run automatically by make
  8. when the latter is modified, but can also be run manually.
  9. """
  10. from __future__ import print_function
  11. from itertools import product
  12. import re
  13. import time
  14. SOURCE = "src/assembler/instructions.yml"
  15. DEST = "src/assembler/instructions.inc.c"
  16. ENCODING = "utf8"
  17. TAB = " " * 4
  18. try:
  19. import yaml
  20. except ImportError:
  21. print("Error: PyYAML is required (https://pypi.python.org/pypi/PyYAML)\n"
  22. "If you don't want to rebuild {0}, do:\n`make -t {0}`".format(DEST))
  23. exit(1)
  24. re_date = re.compile(r"^(\s*@AUTOGEN_DATE\s*)(.*?)$", re.M)
  25. re_inst = re.compile(
  26. r"(/\* @AUTOGEN_INST_BLOCK_START \*/\n*)(.*?)"
  27. r"(\n*/\* @AUTOGEN_INST_BLOCK_END \*/)", re.S)
  28. re_lookup = re.compile(
  29. r"(/\* @AUTOGEN_LOOKUP_BLOCK_START \*/\n*)(.*?)"
  30. r"(\n*/\* @AUTOGEN_LOOKUP_BLOCK_END \*/)", re.S)
  31. class Instruction(object):
  32. """
  33. Represent a single ASM instruction mnemonic.
  34. """
  35. ARG_TYPES = {
  36. "register": "AT_REGISTER",
  37. "immediate": "AT_IMMEDIATE",
  38. "indirect": "AT_INDIRECT",
  39. "indexed": "AT_INDEXED",
  40. "condition": "AT_CONDITION",
  41. "port": "AT_PORT"
  42. }
  43. PSEUDO_TYPES = {
  44. "indirect_hl_or_indexed": ["AT_INDIRECT", "AT_INDEXED"]
  45. }
  46. def __init__(self, name, data):
  47. self._name = name
  48. self._data = data
  49. self._has_optional_args = False
  50. def _get_arg_parse_mask(self, num):
  51. """
  52. Return the appropriate mask to parse_args() for the num-th argument.
  53. """
  54. types = set()
  55. optional = False
  56. for case in self._data["cases"]:
  57. if num < len(case["type"]):
  58. atype = case["type"][num]
  59. if atype in self.ARG_TYPES:
  60. types.add(self.ARG_TYPES[atype])
  61. else:
  62. types.update(self.PSEUDO_TYPES[atype])
  63. else:
  64. optional = True
  65. if not types:
  66. return "AT_NONE"
  67. if optional:
  68. types.add("AT_OPTIONAL")
  69. self._has_optional_args = True
  70. return "|".join(sorted(types))
  71. def _handle_return(self, ret, indent=1):
  72. """
  73. Return code to handle an instruction return statement.
  74. """
  75. data = ", ".join("0x%02X" % byte if isinstance(byte, int) else byte
  76. for byte in ret)
  77. return TAB * indent + "INST_RETURN({0}, {1})".format(len(ret), data)
  78. def _build_case_type_check(self, args):
  79. """
  80. Return the test part of an if statement for an instruction case.
  81. """
  82. conds = ["INST_TYPE({0}) == {1}".format(i, self.ARG_TYPES[cond])
  83. for i, cond in enumerate(args)]
  84. check = " && ".join(conds)
  85. if self._has_optional_args:
  86. return "INST_NARGS == {0} && ".format(len(args)) + check
  87. return check
  88. def _build_register_check(self, num, cond):
  89. """
  90. Return an expression to check for a particular register value.
  91. """
  92. return "INST_REG({0}) == REG_{1}".format(num, cond.upper())
  93. def _build_immediate_check(self, num, cond):
  94. """
  95. Return an expression to check for a particular immediate value.
  96. """
  97. if "." in cond:
  98. itype, value = cond.split(".", 1)
  99. try:
  100. value = int(value)
  101. except ValueError:
  102. value = int(value, 16)
  103. vtype = "sval" if itype.upper() in ["S8", "REL"] else "uval"
  104. test1 = "INST_IMM({0}).mask & IMM_{1}".format(num, itype.upper())
  105. if (itype.upper() == "U16"):
  106. test1 += " && !INST_IMM({0}).is_label".format(num)
  107. test2 = "INST_IMM({0}).{1} == {2}".format(num, vtype, value)
  108. return "({0} && {1})".format(test1, test2)
  109. return "INST_IMM({0}).mask & IMM_{1}".format(num, cond.upper())
  110. def _build_indirect_check(self, num, cond):
  111. """
  112. Return an expression to check for a particular indirect value.
  113. """
  114. if cond.startswith("reg."):
  115. test1 = "INST_INDIRECT({0}).type == AT_REGISTER".format(num)
  116. test2 = "INST_INDIRECT({0}).addr.reg == REG_{1}".format(
  117. num, cond[len("reg."):].upper())
  118. return "({0} && {1})".format(test1, test2)
  119. if cond == "imm" or cond == "immediate":
  120. return "INST_INDIRECT({0}).type == AT_IMMEDIATE".format(num)
  121. err = "Unknown condition for indirect argument: {0}"
  122. return RuntimeError(err.format(cond))
  123. def _build_indexed_check(self, num, cond):
  124. """
  125. Return an expression to check for a particular indexed value.
  126. """
  127. raise RuntimeError("The indexed arg type does not support conditions")
  128. def _build_condition_check(self, num, cond):
  129. """
  130. Return an expression to check for a particular condition value.
  131. """
  132. return "INST_COND({0}) == COND_{1}".format(num, cond.upper())
  133. def _build_port_check(self, num, cond):
  134. """
  135. Return an expression to check for a particular port value.
  136. """
  137. if cond == "reg" or cond == "reg.c":
  138. return "INST_PORT({0}).type == AT_REGISTER".format(num)
  139. if cond == "imm" or cond == "immediate":
  140. return "INST_PORT({0}).type == AT_IMMEDIATE".format(num)
  141. err = "Unknown condition for port argument: {0}"
  142. return RuntimeError(err.format(cond))
  143. _SUBCASE_LOOKUP_TABLE = {
  144. "register": _build_register_check,
  145. "immediate": _build_immediate_check,
  146. "indirect": _build_indirect_check,
  147. "indexed": _build_indexed_check,
  148. "condition": _build_condition_check,
  149. "port": _build_port_check
  150. }
  151. def _build_subcase_check(self, types, conds):
  152. """
  153. Return the test part of an if statement for an instruction subcase.
  154. """
  155. conds = [self._SUBCASE_LOOKUP_TABLE[types[i]](self, i, cond)
  156. for i, cond in enumerate(conds) if cond != "_"]
  157. return " && ".join(conds)
  158. def _iter_permutations(self, types, conds):
  159. """
  160. Iterate over all permutations of the given subcase conditions.
  161. """
  162. def split(typ, cond):
  163. if "|" in cond:
  164. splits = [split(typ, c) for c in cond.split("|")]
  165. merged = [choice for s in splits for choice in s]
  166. if len(merged) != len(set(merged)):
  167. msg = "Repeated conditions for {0}: {1}"
  168. raise RuntimeError(msg.format(typ, cond))
  169. return merged
  170. if typ == "register":
  171. if cond == "i":
  172. return ["ix", "iy"]
  173. if cond == "ih":
  174. return ["ixh", "iyh"]
  175. if cond == "il":
  176. return ["ixl", "iyl"]
  177. return [cond]
  178. splits = [split(typ, cond) for typ, cond in zip(types, conds)]
  179. num = max(len(cond) for cond in splits)
  180. if any(1 < len(cond) < num for cond in splits):
  181. msg = "Invalid condition permutations: {0}"
  182. raise RuntimeError(msg.format(conds))
  183. choices = [cond * num if len(cond) == 1 else cond for cond in splits]
  184. return zip(*choices)
  185. def _adapt_return(self, types, conds, ret):
  186. """
  187. Return a modified byte list to accomodate for prefixes and immediates.
  188. """
  189. ret = ret[:]
  190. for i, byte in enumerate(ret):
  191. if not isinstance(byte, int):
  192. if byte == "u8":
  193. try:
  194. index = types.index("immediate")
  195. imm = "INST_IMM({0})".format(index)
  196. except ValueError:
  197. index = types.index("port")
  198. imm = "INST_PORT({0}).port.imm".format(index)
  199. ret[i] = imm + ".uval"
  200. elif byte == "u16":
  201. if i < len(ret) - 1:
  202. raise RuntimeError("U16 return byte must be last")
  203. try:
  204. index = types.index("immediate")
  205. imm = "INST_IMM({0})".format(index)
  206. except ValueError:
  207. indir = types.index("indirect")
  208. if not conds[indir].startswith("imm"):
  209. msg = "Passing non-immediate indirect as immediate"
  210. raise RuntimeError(msg)
  211. imm = "INST_INDIRECT({0}).addr.imm".format(indir)
  212. ret[i] = "INST_IMM_U16_B1({0})".format(imm)
  213. ret.append("INST_IMM_U16_B2({0})".format(imm))
  214. break
  215. elif byte == "rel":
  216. index = types.index("immediate")
  217. ret[i] = "INST_IMM({0}).sval - 2".format(index)
  218. else:
  219. msg = "Unsupported return byte: {0}"
  220. raise RuntimeError(msg.format(byte))
  221. for i, cond in enumerate(conds):
  222. if types[i] == "register" and cond[0] == "i":
  223. prefix = "INST_I{0}_PREFIX".format(cond[1].upper())
  224. if ret[0] != prefix:
  225. ret.insert(0, prefix)
  226. elif types[i] == "indexed":
  227. ret.insert(0, "INST_INDEX_PREFIX({0})".format(i))
  228. ret.insert(2, "INST_INDEX({0}).offset".format(i))
  229. return ret
  230. def _handle_pseudo_case(self, pseudo, case):
  231. """
  232. Return code to handle an instruction pseudo-case.
  233. Pseudo-cases are cases that have pseudo-types as arguments. This means
  234. they are expanded to cover multiple "real" argument types.
  235. """
  236. index = case["type"].index(pseudo)
  237. if pseudo == "indirect_hl_or_indexed":
  238. case["type"][index] = "indexed"
  239. indexed = self._handle_case(case)
  240. case["type"][index] = "indirect"
  241. indirect = self._handle_case(case)
  242. base_cond = self._build_case_type_check(case["type"])
  243. hl_reg = TAB * 3 + self._build_indirect_check(index, "reg.hl")
  244. indirect[0] = TAB + "if ({0} &&\n{1}) {{".format(base_cond, hl_reg)
  245. return indirect + indexed
  246. raise RuntimeError("Unknown pseudo-type: {0}".format(pseudo))
  247. def _handle_case(self, case):
  248. """
  249. Return code to handle an instruction case.
  250. """
  251. ctype = case["type"]
  252. for pseudo in self.PSEUDO_TYPES:
  253. if pseudo in ctype:
  254. return self._handle_pseudo_case(pseudo, case)
  255. lines = []
  256. cond = self._build_case_type_check(ctype)
  257. lines.append(TAB + "if ({0}) {{".format(cond))
  258. subcases = [(perm, sub["return"]) for sub in case["cases"]
  259. for perm in self._iter_permutations(ctype, sub["cond"])]
  260. for cond, ret in subcases:
  261. check = self._build_subcase_check(ctype, cond)
  262. ret = self._adapt_return(ctype, cond, ret)
  263. if check:
  264. lines.append(TAB * 2 + "if ({0})".format(check))
  265. lines.append(self._handle_return(ret, 3))
  266. else:
  267. lines.append(self._handle_return(ret, 2))
  268. break # Unconditional subcase
  269. else:
  270. lines.append(TAB * 2 + "INST_ERROR(ARG_VALUE)")
  271. lines.append(TAB + "}")
  272. return lines
  273. def render(self):
  274. """
  275. Convert data for an individual instruction into a C parse function.
  276. """
  277. lines = []
  278. if self._data["args"]:
  279. lines.append("{tab}INST_TAKES_ARGS(\n{tab2}{0},\n{tab2}{1},"
  280. "\n{tab2}{2}\n{tab})".format(
  281. self._get_arg_parse_mask(0), self._get_arg_parse_mask(1),
  282. self._get_arg_parse_mask(2), tab=TAB, tab2=TAB * 2))
  283. else:
  284. lines.append(TAB + "INST_TAKES_NO_ARGS")
  285. if "return" in self._data:
  286. lines.append(self._handle_return(self._data["return"]))
  287. elif "cases" in self._data:
  288. for case in self._data["cases"]:
  289. lines.extend(self._handle_case(case))
  290. lines.append(TAB + "INST_ERROR(ARG_TYPE)")
  291. else:
  292. msg = "Missing return or case block for {0} instruction"
  293. raise RuntimeError(msg.format(self._name))
  294. contents = "\n".join(lines)
  295. return "INST_FUNC({0})\n{{\n{1}\n}}".format(self._name, contents)
  296. def build_inst_block(data):
  297. """
  298. Return the instruction parser block, given instruction data.
  299. """
  300. return "\n\n".join(
  301. Instruction(k, v).render() for k, v in sorted(data.items()))
  302. def build_lookup_block(data):
  303. """
  304. Return the instruction lookup block, given instruction data.
  305. """
  306. macro = TAB + "HANDLE({0})"
  307. return "\n".join(macro.format(inst) for inst in sorted(data.keys()))
  308. def process(template, data):
  309. """
  310. Return C code generated from a source template and instruction data.
  311. """
  312. inst_block = build_inst_block(data)
  313. lookup_block = build_lookup_block(data)
  314. date = time.asctime(time.gmtime())
  315. result = re_date.sub(r"\1{0} UTC".format(date), template)
  316. result = re_inst.sub(r"\1{0}\3".format(inst_block), result)
  317. result = re_lookup.sub(r"\1{0}\3".format(lookup_block), result)
  318. return result
  319. def main():
  320. """
  321. Main script entry point.
  322. """
  323. with open(SOURCE, "r") as fp:
  324. text = fp.read().decode(ENCODING)
  325. with open(DEST, "r") as fp:
  326. template = fp.read().decode(ENCODING)
  327. data = yaml.load(text)
  328. result = process(template, data)
  329. with open(DEST, "w") as fp:
  330. fp.write(result.encode(ENCODING))
  331. if __name__ == "__main__":
  332. main()