gen-insn-attr-x86.awk 8.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378
  1. #!/bin/awk -f
  2. # gen-insn-attr-x86.awk: Instruction attribute table generator
  3. # Written by Masami Hiramatsu <mhiramat@redhat.com>
  4. #
  5. # Usage: awk -f gen-insn-attr-x86.awk x86-opcode-map.txt > inat-tables.c
  6. # Awk implementation sanity check
  7. function check_awk_implement() {
  8. if (sprintf("%x", 0) != "0")
  9. return "Your awk has a printf-format problem."
  10. return ""
  11. }
  12. # Clear working vars
  13. function clear_vars() {
  14. delete table
  15. delete lptable2
  16. delete lptable1
  17. delete lptable3
  18. eid = -1 # escape id
  19. gid = -1 # group id
  20. aid = -1 # AVX id
  21. tname = ""
  22. }
  23. BEGIN {
  24. # Implementation error checking
  25. awkchecked = check_awk_implement()
  26. if (awkchecked != "") {
  27. print "Error: " awkchecked > "/dev/stderr"
  28. print "Please try to use gawk." > "/dev/stderr"
  29. exit 1
  30. }
  31. # Setup generating tables
  32. print "/* x86 opcode map generated from x86-opcode-map.txt */"
  33. print "/* Do not change this code. */\n"
  34. ggid = 1
  35. geid = 1
  36. gaid = 0
  37. delete etable
  38. delete gtable
  39. delete atable
  40. opnd_expr = "^[A-Za-z/]"
  41. ext_expr = "^\\("
  42. sep_expr = "^\\|$"
  43. group_expr = "^Grp[0-9A-Za-z]+"
  44. imm_expr = "^[IJAO][a-z]"
  45. imm_flag["Ib"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
  46. imm_flag["Jb"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
  47. imm_flag["Iw"] = "INAT_MAKE_IMM(INAT_IMM_WORD)"
  48. imm_flag["Id"] = "INAT_MAKE_IMM(INAT_IMM_DWORD)"
  49. imm_flag["Iq"] = "INAT_MAKE_IMM(INAT_IMM_QWORD)"
  50. imm_flag["Ap"] = "INAT_MAKE_IMM(INAT_IMM_PTR)"
  51. imm_flag["Iz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
  52. imm_flag["Jz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
  53. imm_flag["Iv"] = "INAT_MAKE_IMM(INAT_IMM_VWORD)"
  54. imm_flag["Ob"] = "INAT_MOFFSET"
  55. imm_flag["Ov"] = "INAT_MOFFSET"
  56. modrm_expr = "^([CDEGMNPQRSUVW/][a-z]+|NTA|T[012])"
  57. force64_expr = "\\([df]64\\)"
  58. rex_expr = "^REX(\\.[XRWB]+)*"
  59. fpu_expr = "^ESC" # TODO
  60. lprefix1_expr = "\\(66\\)"
  61. lprefix2_expr = "\\(F3\\)"
  62. lprefix3_expr = "\\(F2\\)"
  63. max_lprefix = 4
  64. vexok_expr = "\\(VEX\\)"
  65. vexonly_expr = "\\(oVEX\\)"
  66. prefix_expr = "\\(Prefix\\)"
  67. prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ"
  68. prefix_num["REPNE"] = "INAT_PFX_REPNE"
  69. prefix_num["REP/REPE"] = "INAT_PFX_REPE"
  70. prefix_num["LOCK"] = "INAT_PFX_LOCK"
  71. prefix_num["SEG=CS"] = "INAT_PFX_CS"
  72. prefix_num["SEG=DS"] = "INAT_PFX_DS"
  73. prefix_num["SEG=ES"] = "INAT_PFX_ES"
  74. prefix_num["SEG=FS"] = "INAT_PFX_FS"
  75. prefix_num["SEG=GS"] = "INAT_PFX_GS"
  76. prefix_num["SEG=SS"] = "INAT_PFX_SS"
  77. prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ"
  78. prefix_num["2bytes-VEX"] = "INAT_PFX_VEX2"
  79. prefix_num["3bytes-VEX"] = "INAT_PFX_VEX3"
  80. clear_vars()
  81. }
  82. function semantic_error(msg) {
  83. print "Semantic error at " NR ": " msg > "/dev/stderr"
  84. exit 1
  85. }
  86. function debug(msg) {
  87. print "DEBUG: " msg
  88. }
  89. function array_size(arr, i,c) {
  90. c = 0
  91. for (i in arr)
  92. c++
  93. return c
  94. }
  95. /^Table:/ {
  96. print "/* " $0 " */"
  97. if (tname != "")
  98. semantic_error("Hit Table: before EndTable:.");
  99. }
  100. /^Referrer:/ {
  101. if (NF != 1) {
  102. # escape opcode table
  103. ref = ""
  104. for (i = 2; i <= NF; i++)
  105. ref = ref $i
  106. eid = escape[ref]
  107. tname = sprintf("inat_escape_table_%d", eid)
  108. }
  109. }
  110. /^AVXcode:/ {
  111. if (NF != 1) {
  112. # AVX/escape opcode table
  113. aid = $2
  114. if (gaid <= aid)
  115. gaid = aid + 1
  116. if (tname == "") # AVX only opcode table
  117. tname = sprintf("inat_avx_table_%d", $2)
  118. }
  119. if (aid == -1 && eid == -1) # primary opcode table
  120. tname = "inat_primary_table"
  121. }
  122. /^GrpTable:/ {
  123. print "/* " $0 " */"
  124. if (!($2 in group))
  125. semantic_error("No group: " $2 )
  126. gid = group[$2]
  127. tname = "inat_group_table_" gid
  128. }
  129. function print_table(tbl,name,fmt,n)
  130. {
  131. print "const insn_attr_t " name " = {"
  132. for (i = 0; i < n; i++) {
  133. id = sprintf(fmt, i)
  134. if (tbl[id])
  135. print " [" id "] = " tbl[id] ","
  136. }
  137. print "};"
  138. }
  139. /^EndTable/ {
  140. if (gid != -1) {
  141. # print group tables
  142. if (array_size(table) != 0) {
  143. print_table(table, tname "[INAT_GROUP_TABLE_SIZE]",
  144. "0x%x", 8)
  145. gtable[gid,0] = tname
  146. }
  147. if (array_size(lptable1) != 0) {
  148. print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]",
  149. "0x%x", 8)
  150. gtable[gid,1] = tname "_1"
  151. }
  152. if (array_size(lptable2) != 0) {
  153. print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]",
  154. "0x%x", 8)
  155. gtable[gid,2] = tname "_2"
  156. }
  157. if (array_size(lptable3) != 0) {
  158. print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]",
  159. "0x%x", 8)
  160. gtable[gid,3] = tname "_3"
  161. }
  162. } else {
  163. # print primary/escaped tables
  164. if (array_size(table) != 0) {
  165. print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]",
  166. "0x%02x", 256)
  167. etable[eid,0] = tname
  168. if (aid >= 0)
  169. atable[aid,0] = tname
  170. }
  171. if (array_size(lptable1) != 0) {
  172. print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]",
  173. "0x%02x", 256)
  174. etable[eid,1] = tname "_1"
  175. if (aid >= 0)
  176. atable[aid,1] = tname "_1"
  177. }
  178. if (array_size(lptable2) != 0) {
  179. print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]",
  180. "0x%02x", 256)
  181. etable[eid,2] = tname "_2"
  182. if (aid >= 0)
  183. atable[aid,2] = tname "_2"
  184. }
  185. if (array_size(lptable3) != 0) {
  186. print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]",
  187. "0x%02x", 256)
  188. etable[eid,3] = tname "_3"
  189. if (aid >= 0)
  190. atable[aid,3] = tname "_3"
  191. }
  192. }
  193. print ""
  194. clear_vars()
  195. }
  196. function add_flags(old,new) {
  197. if (old && new)
  198. return old " | " new
  199. else if (old)
  200. return old
  201. else
  202. return new
  203. }
  204. # convert operands to flags.
  205. function convert_operands(count,opnd, i,j,imm,mod)
  206. {
  207. imm = null
  208. mod = null
  209. for (j = 1; j <= count; j++) {
  210. i = opnd[j]
  211. if (match(i, imm_expr) == 1) {
  212. if (!imm_flag[i])
  213. semantic_error("Unknown imm opnd: " i)
  214. if (imm) {
  215. if (i != "Ib")
  216. semantic_error("Second IMM error")
  217. imm = add_flags(imm, "INAT_SCNDIMM")
  218. } else
  219. imm = imm_flag[i]
  220. } else if (match(i, modrm_expr))
  221. mod = "INAT_MODRM"
  222. }
  223. return add_flags(imm, mod)
  224. }
  225. /^[0-9a-f]+\:/ {
  226. if (NR == 1)
  227. next
  228. # get index
  229. idx = "0x" substr($1, 1, index($1,":") - 1)
  230. if (idx in table)
  231. semantic_error("Redefine " idx " in " tname)
  232. # check if escaped opcode
  233. if ("escape" == $2) {
  234. if ($3 != "#")
  235. semantic_error("No escaped name")
  236. ref = ""
  237. for (i = 4; i <= NF; i++)
  238. ref = ref $i
  239. if (ref in escape)
  240. semantic_error("Redefine escape (" ref ")")
  241. escape[ref] = geid
  242. geid++
  243. table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")"
  244. next
  245. }
  246. variant = null
  247. # converts
  248. i = 2
  249. while (i <= NF) {
  250. opcode = $(i++)
  251. delete opnds
  252. ext = null
  253. flags = null
  254. opnd = null
  255. # parse one opcode
  256. if (match($i, opnd_expr)) {
  257. opnd = $i
  258. count = split($(i++), opnds, ",")
  259. flags = convert_operands(count, opnds)
  260. }
  261. if (match($i, ext_expr))
  262. ext = $(i++)
  263. if (match($i, sep_expr))
  264. i++
  265. else if (i < NF)
  266. semantic_error($i " is not a separator")
  267. # check if group opcode
  268. if (match(opcode, group_expr)) {
  269. if (!(opcode in group)) {
  270. group[opcode] = ggid
  271. ggid++
  272. }
  273. flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")")
  274. }
  275. # check force(or default) 64bit
  276. if (match(ext, force64_expr))
  277. flags = add_flags(flags, "INAT_FORCE64")
  278. # check REX prefix
  279. if (match(opcode, rex_expr))
  280. flags = add_flags(flags, "INAT_MAKE_PREFIX(INAT_PFX_REX)")
  281. # check coprocessor escape : TODO
  282. if (match(opcode, fpu_expr))
  283. flags = add_flags(flags, "INAT_MODRM")
  284. # check VEX only code
  285. if (match(ext, vexonly_expr))
  286. flags = add_flags(flags, "INAT_VEXOK | INAT_VEXONLY")
  287. # check VEX only code
  288. if (match(ext, vexok_expr))
  289. flags = add_flags(flags, "INAT_VEXOK")
  290. # check prefixes
  291. if (match(ext, prefix_expr)) {
  292. if (!prefix_num[opcode])
  293. semantic_error("Unknown prefix: " opcode)
  294. flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")")
  295. }
  296. if (length(flags) == 0)
  297. continue
  298. # check if last prefix
  299. if (match(ext, lprefix1_expr)) {
  300. lptable1[idx] = add_flags(lptable1[idx],flags)
  301. variant = "INAT_VARIANT"
  302. } else if (match(ext, lprefix2_expr)) {
  303. lptable2[idx] = add_flags(lptable2[idx],flags)
  304. variant = "INAT_VARIANT"
  305. } else if (match(ext, lprefix3_expr)) {
  306. lptable3[idx] = add_flags(lptable3[idx],flags)
  307. variant = "INAT_VARIANT"
  308. } else {
  309. table[idx] = add_flags(table[idx],flags)
  310. }
  311. }
  312. if (variant)
  313. table[idx] = add_flags(table[idx],variant)
  314. }
  315. END {
  316. if (awkchecked != "")
  317. exit 1
  318. # print escape opcode map's array
  319. print "/* Escape opcode map array */"
  320. print "const insn_attr_t const *inat_escape_tables[INAT_ESC_MAX + 1]" \
  321. "[INAT_LSTPFX_MAX + 1] = {"
  322. for (i = 0; i < geid; i++)
  323. for (j = 0; j < max_lprefix; j++)
  324. if (etable[i,j])
  325. print " ["i"]["j"] = "etable[i,j]","
  326. print "};\n"
  327. # print group opcode map's array
  328. print "/* Group opcode map array */"
  329. print "const insn_attr_t const *inat_group_tables[INAT_GRP_MAX + 1]"\
  330. "[INAT_LSTPFX_MAX + 1] = {"
  331. for (i = 0; i < ggid; i++)
  332. for (j = 0; j < max_lprefix; j++)
  333. if (gtable[i,j])
  334. print " ["i"]["j"] = "gtable[i,j]","
  335. print "};\n"
  336. # print AVX opcode map's array
  337. print "/* AVX opcode map array */"
  338. print "const insn_attr_t const *inat_avx_tables[X86_VEX_M_MAX + 1]"\
  339. "[INAT_LSTPFX_MAX + 1] = {"
  340. for (i = 0; i < gaid; i++)
  341. for (j = 0; j < max_lprefix; j++)
  342. if (atable[i,j])
  343. print " ["i"]["j"] = "atable[i,j]","
  344. print "};"
  345. }