lib/dis_arm.lua

   1 ----------------------------------------------------------------------------
   2 -- LuaJIT ARM disassembler module.
   3 --
   4 -- Copyright (C) 2005-2012 Mike Pall. All rights reserved.
   5 -- Released under the MIT license. See Copyright Notice in luajit.h
   6 ----------------------------------------------------------------------------
   7 -- This is a helper module used by the LuaJIT machine code dumper module.
   8 --
   9 -- It disassembles most user-mode ARMv7 instructions
  10 -- NYI: Advanced SIMD and VFP instructions.
  11 ------------------------------------------------------------------------------
  12
  13 local type = type
  14 local sub, byte, format = string.sub, string.byte, string.format
  15 local match, gmatch, gsub = string.match, string.gmatch, string.gsub
  16 local concat = table.concat
  17 local bit = require("bit")
  18 local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
  19 local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
  20
  21 ------------------------------------------------------------------------------
  22 -- Opcode maps
  23 ------------------------------------------------------------------------------
  24
  25 local map_loadc = {
  26   shift = 9, mask = 7,
  27   [5] = {
  28     shift = 0, mask = 0 -- NYI VFP load/store.
  29   },
  30   _ = {
  31     shift = 0, mask = 0 -- NYI ldc, mcrr, mrrc.
  32   },
  33 }
  34
  35 local map_datac = {
  36   shift = 24, mask = 1,
  37   [0] = {
  38     shift = 9, mask = 7,
  39     [5] = {
  40       shift = 0, mask = 0 -- NYI VFP data.
  41     },
  42     _ = {
  43       shift = 0, mask = 0 -- NYI cdp, mcr, mrc.
  44     },
  45   },
  46   "svcT",
  47 }
  48
  49 local map_loadcu = {
  50   shift = 0, mask = 0, -- NYI unconditional CP load/store.
  51 }
  52
  53 local map_datacu = {
  54   shift = 0, mask = 0, -- NYI unconditional CP data.
  55 }
  56
  57 local map_simddata = {
  58   shift = 0, mask = 0, -- NYI SIMD data.
  59 }
  60
  61 local map_simdload = {
  62   shift = 0, mask = 0, -- NYI SIMD load/store, preload.
  63 }
  64
  65 local map_preload = {
  66   shift = 0, mask = 0, -- NYI preload.
  67 }
  68
  69 local map_media = {
  70   shift = 20, mask = 31,
  71   [0] = false,
  72   { --01
  73     shift = 5, mask = 7,
  74     [0] = "sadd16DNM", "sasxDNM", "ssaxDNM", "ssub16DNM",
  75     "sadd8DNM", false, false, "ssub8DNM",
  76   },
  77   { --02
  78     shift = 5, mask = 7,
  79     [0] = "qadd16DNM", "qasxDNM", "qsaxDNM", "qsub16DNM",
  80     "qadd8DNM", false, false, "qsub8DNM",
  81   },
  82   { --03
  83     shift = 5, mask = 7,
  84     [0] = "shadd16DNM", "shasxDNM", "shsaxDNM", "shsub16DNM",
  85     "shadd8DNM", false, false, "shsub8DNM",
  86   },
  87   false,
  88   { --05
  89     shift = 5, mask = 7,
  90     [0] = "uadd16DNM", "uasxDNM", "usaxDNM", "usub16DNM",
  91     "uadd8DNM", false, false, "usub8DNM",
  92   },
  93   { --06
  94     shift = 5, mask = 7,
  95     [0] = "uqadd16DNM", "uqasxDNM", "uqsaxDNM", "uqsub16DNM",
  96     "uqadd8DNM", false, false, "uqsub8DNM",
  97   },
  98   { --07
  99     shift = 5, mask = 7,
 100     [0] = "uhadd16DNM", "uhasxDNM", "uhsaxDNM", "uhsub16DNM",
 101     "uhadd8DNM", false, false, "uhsub8DNM",
 102   },
 103   { --08
 104     shift = 5, mask = 7,
 105     [0] = "pkhbtDNMU", false, "pkhtbDNMU",
 106     { shift = 16, mask = 15, [15] = "sxtb16DMU", _ = "sxtab16DNMU", },
 107     "pkhbtDNMU", "selDNM", "pkhtbDNMU",
 108   },
 109   false,
 110   { --0a
 111     shift = 5, mask = 7,
 112     [0] = "ssatDxMu", "ssat16DxM", "ssatDxMu",
 113     { shift = 16, mask = 15, [15] = "sxtbDMU", _ = "sxtabDNMU", },
 114     "ssatDxMu", false, "ssatDxMu",
 115   },
 116   { --0b
 117     shift = 5, mask = 7,
 118     [0] = "ssatDxMu", "revDM", "ssatDxMu",
 119     { shift = 16, mask = 15, [15] = "sxthDMU", _ = "sxtahDNMU", },
 120     "ssatDxMu", "rev16DM", "ssatDxMu",
 121   },
 122   { --0c
 123     shift = 5, mask = 7,
 124     [3] = { shift = 16, mask = 15, [15] = "uxtb16DMU", _ = "uxtab16DNMU", },
 125   },
 126   false,
 127   { --0e
 128     shift = 5, mask = 7,
 129     [0] = "usatDwMu", "usat16DwM", "usatDwMu",
 130     { shift = 16, mask = 15, [15] = "uxtbDMU", _ = "uxtabDNMU", },
 131     "usatDwMu", false, "usatDwMu",
 132   },
 133   { --0f
 134     shift = 5, mask = 7,
 135     [0] = "usatDwMu", "rbitDM", "usatDwMu",
 136     { shift = 16, mask = 15, [15] = "uxthDMU", _ = "uxtahDNMU", },
 137     "usatDwMu", "revshDM", "usatDwMu",
 138   },
 139   { --10
 140     shift = 12, mask = 15,
 141     [15] = {
 142       shift = 5, mask = 7,
 143       "smuadNMS", "smuadxNMS", "smusdNMS", "smusdxNMS",
 144     },
 145     _ = {
 146       shift = 5, mask = 7,
 147       [0] = "smladNMSD", "smladxNMSD", "smlsdNMSD", "smlsdxNMSD",
 148     },
 149   },
 150   false, false, false,
 151   { --14
 152     shift = 5, mask = 7,
 153     [0] = "smlaldDNMS", "smlaldxDNMS", "smlsldDNMS", "smlsldxDNMS",
 154   },
 155   { --15
 156     shift = 5, mask = 7,
 157     [0] = { shift = 12, mask = 15, [15] = "smmulNMS", _ = "smmlaNMSD", },
 158     { shift = 12, mask = 15, [15] = "smmulrNMS", _ = "smmlarNMSD", },
 159     false, false, false, false,
 160     "smmlsNMSD", "smmlsrNMSD",
 161   },
 162   false, false,
 163   { --18
 164     shift = 5, mask = 7,
 165     [0] = { shift = 12, mask = 15, [15] = "usad8NMS", _ = "usada8NMSD", },
 166   },
 167   false,
 168   { --1a
 169     shift = 5, mask = 3, [2] = "sbfxDMvw",
 170   },
 171   { --1b
 172     shift = 5, mask = 3, [2] = "sbfxDMvw",
 173   },
 174   { --1c
 175     shift = 5, mask = 3,
 176     [0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
 177   },
 178   { --1d
 179     shift = 5, mask = 3,
 180     [0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
 181   },
 182   { --1e
 183     shift = 5, mask = 3, [2] = "ubfxDMvw",
 184   },
 185   { --1f
 186     shift = 5, mask = 3, [2] = "ubfxDMvw",
 187   },
 188 }
 189
 190 local map_load = {
 191   shift = 21, mask = 9,
 192   {
 193     shift = 20, mask = 5,
 194     [0] = "strtDL", "ldrtDL", [4] = "strbtDL", [5] = "ldrbtDL",
 195   },
 196   _ = {
 197     shift = 20, mask = 5,
 198     [0] = "strDL", "ldrDL", [4] = "strbDL", [5] = "ldrbDL",
 199   }
 200 }
 201
 202 local map_load1 = {
 203   shift = 4, mask = 1,
 204   [0] = map_load, map_media,
 205 }
 206
 207 local map_loadm = {
 208   shift = 20, mask = 1,
 209   [0] = {
 210     shift = 23, mask = 3,
 211     [0] = "stmdaNR", "stmNR",
 212     { shift = 16, mask = 63, [45] = "pushR", _ = "stmdbNR", }, "stmibNR",
 213   },
 214   {
 215     shift = 23, mask = 3,
 216     [0] = "ldmdaNR", { shift = 16, mask = 63, [61] = "popR", _ = "ldmNR", },
 217     "ldmdbNR", "ldmibNR",
 218   },
 219 }
 220
 221 local map_data = {
 222   shift = 21, mask = 15,
 223   [0] = "andDNPs", "eorDNPs", "subDNPs", "rsbDNPs",
 224   "addDNPs", "adcDNPs", "sbcDNPs", "rscDNPs",
 225   "tstNP", "teqNP", "cmpNP", "cmnNP",
 226   "orrDNPs", "movDPs", "bicDNPs", "mvnDPs",
 227 }
 228
 229 local map_mul = {
 230   shift = 21, mask = 7,
 231   [0] = "mulNMSs", "mlaNMSDs", "umaalDNMS", "mlsDNMS",
 232   "umullDNMSs", "umlalDNMSs", "smullDNMSs", "smlalDNMSs",
 233 }
 234
 235 local map_sync = {
 236   shift = 20, mask = 15, -- NYI: brackets around N. R(D+1) for ldrexd/strexd.
 237   [0] = "swpDMN", false, false, false,
 238   "swpbDMN", false, false, false,
 239   "strexDMN", "ldrexDN", "strexdDN", "ldrexdDN",
 240   "strexbDMN", "ldrexbDN", "strexhDN", "ldrexhDN",
 241 }
 242
 243 local map_mulh = {
 244   shift = 21, mask = 3,
 245   [0] = { shift = 5, mask = 3,
 246     [0] = "smlabbNMSD", "smlatbNMSD", "smlabtNMSD", "smlattNMSD", },
 247   { shift = 5, mask = 3,
 248     [0] = "smlawbNMSD", "smulwbNMS", "smlawtNMSD", "smulwtNMS", },
 249   { shift = 5, mask = 3,
 250     [0] = "smlalbbDNMS", "smlaltbDNMS", "smlalbtDNMS", "smlalttDNMS", },
 251   { shift = 5, mask = 3,
 252     [0] = "smulbbNMS", "smultbNMS", "smulbtNMS", "smulttNMS", },
 253 }
 254
 255 local map_misc = {
 256   shift = 4, mask = 7,
 257   -- NYI: decode PSR bits of msr.
 258   [0] = { shift = 21, mask = 1, [0] = "mrsD", "msrM", },
 259   { shift = 21, mask = 3, "bxM", false, "clzDM", },
 260   { shift = 21, mask = 3, "bxjM", },
 261   { shift = 21, mask = 3, "blxM", },
 262   false,
 263   { shift = 21, mask = 3, [0] = "qaddDMN", "qsubDMN", "qdaddDMN", "qdsubDMN", },
 264   false,
 265   { shift = 21, mask = 3, "bkptK", },
 266 }
 267
 268 local map_datar = {
 269   shift = 4, mask = 9,
 270   [9] = {
 271     shift = 5, mask = 3,
 272     [0] = { shift = 24, mask = 1, [0] = map_mul, map_sync, },
 273     { shift = 20, mask = 1, [0] = "strhDL", "ldrhDL", },
 274     { shift = 20, mask = 1, [0] = "ldrdDL", "ldrsbDL", },
 275     { shift = 20, mask = 1, [0] = "strdDL", "ldrshDL", },
 276   },
 277   _ = {
 278     shift = 20, mask = 25,
 279     [16] = { shift = 7, mask = 1, [0] = map_misc, map_mulh, },
 280     _ = {
 281       shift = 0, mask = 0xffffffff,
 282       [bor(0xe1a00000)] = "nop",
 283       _ = map_data,
 284     }
 285   },
 286 }
 287
 288 local map_datai = {
 289   shift = 20, mask = 31, -- NYI: decode PSR bits of msr. Decode imm12.
 290   [16] = "movwDW", [20] = "movtDW",
 291   [18] = { shift = 0, mask = 0xf00ff, [0] = "nopv6", _ = "msrNW", },
 292   [22] = "msrNW",
 293   _ = map_data,
 294 }
 295
 296 local map_branch = {
 297   shift = 24, mask = 1,
 298   [0] = "bB", "blB"
 299 }
 300
 301 local map_condins = {
 302   [0] = map_datar, map_datai, map_load, map_load1,
 303   map_loadm, map_branch, map_loadc, map_datac
 304 }
 305
 306 -- NYI: setend.
 307 local map_uncondins = {
 308   [0] = false, map_simddata, map_simdload, map_preload,
 309   false, "blxB", map_loadcu, map_datacu,
 310 }
 311
 312 ------------------------------------------------------------------------------
 313
 314 local map_gpr = {
 315   [0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
 316   "r8", "r9", "r10", "r11", "r12", "sp", "lr", "pc",
 317 }
 318
 319 local map_cond = {
 320   [0] = "eq", "ne", "hs", "lo", "mi", "pl", "vs", "vc",
 321   "hi", "ls", "ge", "lt", "gt", "le", "al",
 322 }
 323
 324 local map_shift = { [0] = "lsl", "lsr", "asr", "ror", }
 325
 326 ------------------------------------------------------------------------------
 327
 328 -- Output a nicely formatted line with an opcode and operands.
 329 local function putop(ctx, text, operands)
 330   local pos = ctx.pos
 331   local extra = ""
 332   if ctx.rel then
 333     local sym = ctx.symtab[ctx.rel]
 334     if sym then
 335       extra = "\t->"..sym
 336     elseif band(ctx.op, 0x0e000000) ~= 0x0a000000 then
 337       extra = "\t; 0x"..tohex(ctx.rel)
 338     end
 339   end
 340   if ctx.hexdump > 0 then
 341     ctx.out(format("%08x  %s  %-5s %s%s\n",
 342             ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
 343   else
 344     ctx.out(format("%08x  %-5s %s%s\n",
 345             ctx.addr+pos, text, concat(operands, ", "), extra))
 346   end
 347   ctx.pos = pos + 4
 348 end
 349
 350 -- Fallback for unknown opcodes.
 351 local function unknown(ctx)
 352   return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
 353 end
 354
 355 -- Format operand 2 of load/store opcodes.
 356 local function fmtload(ctx, op, pos)
 357   local base = map_gpr[band(rshift(op, 16), 15)]
 358   local x, ofs
 359   local ext = (band(op, 0x04000000) == 0)
 360   if not ext and band(op, 0x02000000) == 0 then
 361     ofs = band(op, 4095)
 362     if band(op, 0x00800000) == 0 then ofs = -ofs end
 363     if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
 364     ofs = "#"..ofs
 365   elseif ext and band(op, 0x00400000) ~= 0 then
 366     ofs = band(op, 15) + band(rshift(op, 4), 0xf0)
 367     if band(op, 0x00800000) == 0 then ofs = -ofs end
 368     if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
 369     ofs = "#"..ofs
 370   else
 371     ofs = map_gpr[band(op, 15)]
 372     if ext or band(op, 0xfe0) == 0 then
 373     elseif band(op, 0xfe0) == 0x60 then
 374       ofs = format("%s, rrx", ofs)
 375     else
 376       local sh = band(rshift(op, 7), 31)
 377       if sh == 0 then sh = 32 end
 378       ofs = format("%s, %s #%d", ofs, map_shift[band(rshift(op, 5), 3)], sh)
 379     end
 380     if band(op, 0x00800000) == 0 then ofs = "-"..ofs end
 381   end
 382   if ofs == "#0" then
 383     x = format("[%s]", base)
 384   elseif band(op, 0x01000000) == 0 then
 385     x = format("[%s], %s", base, ofs)
 386   else
 387     x = format("[%s, %s]", base, ofs)
 388   end
 389   if band(op, 0x01200000) == 0x01200000 then x = x.."!" end
 390   return x
 391 end
 392
 393 -- Disassemble a single instruction.
 394 local function disass_ins(ctx)
 395   local pos = ctx.pos
 396   local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
 397   local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
 398   local operands = {}
 399   local suffix = ""
 400   local last, name, pat
 401   ctx.op = op
 402   ctx.rel = nil
 403
 404   local cond = rshift(op, 28)
 405   local opat
 406   if cond == 15 then
 407     opat = map_uncondins[band(rshift(op, 25), 7)]
 408   else
 409     if cond ~= 14 then suffix = map_cond[cond] end
 410     opat = map_condins[band(rshift(op, 25), 7)]
 411   end
 412   while type(opat) ~= "string" do
 413     if not opat then return unknown(ctx) end
 414     opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
 415   end
 416   name, pat = match(opat, "^([a-z0-9]*)(.*)")
 417
 418   for p in gmatch(pat, ".") do
 419     local x = nil
 420     if p == "D" then
 421       x = map_gpr[band(rshift(op, 12), 15)]
 422     elseif p == "N" then
 423       x = map_gpr[band(rshift(op, 16), 15)]
 424     elseif p == "S" then
 425       x = map_gpr[band(rshift(op, 8), 15)]
 426     elseif p == "M" then
 427       x = map_gpr[band(op, 15)]
 428     elseif p == "P" then
 429       if band(op, 0x02000000) ~= 0 then
 430         x = ror(band(op, 255), 2*band(rshift(op, 8), 15))
 431       else
 432         x = map_gpr[band(op, 15)]
 433         if band(op, 0xff0) ~= 0 then
 434           operands[#operands+1] = x
 435           local s = map_shift[band(rshift(op, 5), 3)]
 436           local r = nil
 437           if band(op, 0xf90) == 0 then
 438             if s == "ror" then s = "rrx" else r = "#32" end
 439           elseif band(op, 0x10) == 0 then
 440             r = "#"..band(rshift(op, 7), 31)
 441           else
 442             r = map_gpr[band(rshift(op, 8), 15)]
 443           end
 444           if name == "mov" then name = s; x = r
 445           elseif r then x = format("%s %s", s, r)
 446           else x = s end
 447         end
 448       end
 449     elseif p == "L" then
 450       x = fmtload(ctx, op, pos, false)
 451     elseif p == "B" then
 452       local addr = ctx.addr + pos + 8 + arshift(lshift(op, 8), 6)
 453       if cond == 15 then addr = addr + band(rshift(op, 23), 2) end
 454       ctx.rel = addr
 455       x = "0x"..tohex(addr)
 456     elseif p == "R" then
 457       if band(op, 0x00200000) ~= 0 and #operands == 1 then
 458         operands[1] = operands[1].."!"
 459       end
 460       local t = {}
 461       for i=0,15 do
 462         if band(rshift(op, i), 1) == 1 then t[#t+1] = map_gpr[i] end
 463       end
 464       x = "{"..concat(t, ", ").."}"
 465     elseif p == "W" then
 466       x = band(op, 0x0fff) + band(rshift(op, 4), 0xf000)
 467     elseif p == "T" then
 468       x = "#0x"..tohex(band(op, 0x00ffffff), 6)
 469     elseif p == "U" then
 470       x = band(rshift(op, 7), 31)
 471       if x == 0 then x = nil end
 472     elseif p == "u" then
 473       x = band(rshift(op, 7), 31)
 474       if band(op, 0x40) == 0 then
 475         if x == 0 then x = nil else x = "lsl #"..x end
 476       else
 477         if x == 0 then x = "asr #32" else x = "asr #"..x end
 478       end
 479     elseif p == "v" then
 480       x = band(rshift(op, 7), 31)
 481     elseif p == "w" then
 482       x = band(rshift(op, 16), 31)
 483     elseif p == "x" then
 484       x = band(rshift(op, 16), 31) + 1
 485     elseif p == "X" then
 486       x = band(rshift(op, 16), 31) - last + 1
 487     elseif p == "K" then
 488       x = "#0x"..tohex(band(rshift(op, 4), 0x0000fff0) + band(op, 15), 4)
 489     elseif p == "s" then
 490       if band(op, 0x00100000) ~= 0 then suffix = "s"..suffix end
 491     else
 492       assert(false)
 493     end
 494     if x then
 495       last = x
 496       if type(x) == "number" then x = "#"..x end
 497       operands[#operands+1] = x
 498     end
 499   end
 500
 501   return putop(ctx, name..suffix, operands)
 502 end
 503
 504 ------------------------------------------------------------------------------
 505
 506 -- Disassemble a block of code.
 507 local function disass_block(ctx, ofs, len)
 508   if not ofs then ofs = 0 end
 509   local stop = len and ofs+len or #ctx.code
 510   ctx.pos = ofs
 511   ctx.rel = nil
 512   while ctx.pos < stop do disass_ins(ctx) end
 513 end
 514
 515 -- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
 516 local function create_(code, addr, out)
 517   local ctx = {}
 518   ctx.code = code
 519   ctx.addr = addr or 0
 520   ctx.out = out or io.write
 521   ctx.symtab = {}
 522   ctx.disass = disass_block
 523   ctx.hexdump = 8
 524   return ctx
 525 end
 526
 527 -- Simple API: disassemble code (a string) at address and output via out.
 528 local function disass_(code, addr, out)
 529   create_(code, addr, out):disass()
 530 end
 531
 532 -- Return register name for RID.
 533 local function regname_(r)
 534   return map_gpr[r]
 535 end
 536
 537 -- Public module functions.
 538 module(...)
 539
 540 create = create_
 541 disass = disass_
 542 regname = regname_
 543