Sean Callanan | 95fcebd | 2010-01-29 00:21:04 +0000 | [diff] [blame] | 1 | //===- EDEmitter.cpp - Generate instruction descriptions for ED -*- C++ -*-===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This tablegen backend is responsible for emitting a description of each |
| 11 | // instruction in a format that the enhanced disassembler can use to tokenize |
| 12 | // and parse instructions. |
| 13 | // |
| 14 | //===----------------------------------------------------------------------===// |
| 15 | |
| 16 | #include "EDEmitter.h" |
| 17 | |
| 18 | #include "CodeGenTarget.h" |
| 19 | #include "Record.h" |
| 20 | |
| 21 | #include "llvm/Support/ErrorHandling.h" |
| 22 | #include "llvm/Support/Format.h" |
| 23 | #include "llvm/Support/raw_ostream.h" |
| 24 | |
| 25 | #include <vector> |
| 26 | #include <string> |
| 27 | |
| 28 | #define MAX_OPERANDS 5 |
| 29 | #define MAX_SYNTAXES 2 |
| 30 | |
| 31 | using namespace llvm; |
| 32 | |
| 33 | /////////////////////////////////////////////////////////// |
| 34 | // Support classes for emitting nested C data structures // |
| 35 | /////////////////////////////////////////////////////////// |
| 36 | |
| 37 | namespace { |
| 38 | |
| 39 | class EnumEmitter { |
| 40 | private: |
| 41 | std::string Name; |
| 42 | std::vector<std::string> Entries; |
| 43 | public: |
| 44 | EnumEmitter(const char *N) : Name(N) { |
| 45 | } |
| 46 | int addEntry(const char *e) { |
| 47 | Entries.push_back(std::string(e)); |
| 48 | return Entries.size() - 1; |
| 49 | } |
| 50 | void emit(raw_ostream &o, unsigned int &i) { |
| 51 | o.indent(i) << "enum " << Name.c_str() << " {" << "\n"; |
| 52 | i += 2; |
| 53 | |
| 54 | unsigned int index = 0; |
| 55 | unsigned int numEntries = Entries.size(); |
| 56 | for(index = 0; index < numEntries; ++index) { |
| 57 | o.indent(i) << Entries[index]; |
| 58 | if(index < (numEntries - 1)) |
| 59 | o << ","; |
| 60 | o << "\n"; |
| 61 | } |
| 62 | |
| 63 | i -= 2; |
| 64 | o.indent(i) << "};" << "\n"; |
| 65 | } |
| 66 | |
| 67 | void emitAsFlags(raw_ostream &o, unsigned int &i) { |
| 68 | o.indent(i) << "enum " << Name.c_str() << " {" << "\n"; |
| 69 | i += 2; |
| 70 | |
| 71 | unsigned int index = 0; |
| 72 | unsigned int numEntries = Entries.size(); |
| 73 | unsigned int flag = 1; |
| 74 | for (index = 0; index < numEntries; ++index) { |
| 75 | o.indent(i) << Entries[index] << " = " << format("0x%x", flag); |
| 76 | if (index < (numEntries - 1)) |
| 77 | o << ","; |
| 78 | o << "\n"; |
| 79 | flag <<= 1; |
| 80 | } |
| 81 | |
| 82 | i -= 2; |
| 83 | o.indent(i) << "};" << "\n"; |
| 84 | } |
| 85 | }; |
| 86 | |
| 87 | class StructEmitter { |
| 88 | private: |
| 89 | std::string Name; |
| 90 | std::vector<std::string> MemberTypes; |
| 91 | std::vector<std::string> MemberNames; |
| 92 | public: |
| 93 | StructEmitter(const char *N) : Name(N) { |
| 94 | } |
| 95 | void addMember(const char *t, const char *n) { |
| 96 | MemberTypes.push_back(std::string(t)); |
| 97 | MemberNames.push_back(std::string(n)); |
| 98 | } |
| 99 | void emit(raw_ostream &o, unsigned int &i) { |
| 100 | o.indent(i) << "struct " << Name.c_str() << " {" << "\n"; |
| 101 | i += 2; |
| 102 | |
| 103 | unsigned int index = 0; |
| 104 | unsigned int numMembers = MemberTypes.size(); |
| 105 | for (index = 0; index < numMembers; ++index) { |
| 106 | o.indent(i) << MemberTypes[index] << " " << MemberNames[index] << ";"; |
| 107 | o << "\n"; |
| 108 | } |
| 109 | |
| 110 | i -= 2; |
| 111 | o.indent(i) << "};" << "\n"; |
| 112 | } |
| 113 | }; |
| 114 | |
| 115 | class ConstantEmitter { |
| 116 | public: |
| 117 | virtual ~ConstantEmitter() { } |
| 118 | virtual void emit(raw_ostream &o, unsigned int &i) = 0; |
| 119 | }; |
| 120 | |
| 121 | class LiteralConstantEmitter : public ConstantEmitter { |
| 122 | private: |
| 123 | std::string Literal; |
| 124 | public: |
| 125 | LiteralConstantEmitter(const char *literal) : Literal(literal) { |
| 126 | } |
| 127 | LiteralConstantEmitter(int literal) { |
| 128 | char buf[256]; |
| 129 | snprintf(buf, 256, "%d", literal); |
| 130 | Literal = buf; |
| 131 | } |
| 132 | void emit(raw_ostream &o, unsigned int &i) { |
| 133 | o << Literal; |
| 134 | } |
| 135 | }; |
| 136 | |
| 137 | class CompoundConstantEmitter : public ConstantEmitter { |
| 138 | private: |
| 139 | std::vector<ConstantEmitter*> Entries; |
| 140 | public: |
| 141 | CompoundConstantEmitter() { |
| 142 | } |
| 143 | ~CompoundConstantEmitter() { |
| 144 | unsigned int index; |
| 145 | unsigned int numEntries = Entries.size(); |
| 146 | for (index = 0; index < numEntries; ++index) { |
| 147 | delete Entries[index]; |
| 148 | } |
| 149 | } |
| 150 | CompoundConstantEmitter &addEntry(ConstantEmitter *e) { |
| 151 | Entries.push_back(e); |
| 152 | return *this; |
| 153 | } |
| 154 | void emit(raw_ostream &o, unsigned int &i) { |
| 155 | o << "{" << "\n"; |
| 156 | i += 2; |
| 157 | |
| 158 | unsigned int index; |
| 159 | unsigned int numEntries = Entries.size(); |
| 160 | for (index = 0; index < numEntries; ++index) { |
| 161 | o.indent(i); |
| 162 | Entries[index]->emit(o, i); |
| 163 | if (index < (numEntries - 1)) |
| 164 | o << ","; |
| 165 | o << "\n"; |
| 166 | } |
| 167 | |
| 168 | i -= 2; |
| 169 | o.indent(i) << "}"; |
| 170 | } |
| 171 | }; |
| 172 | |
| 173 | class FlagsConstantEmitter : public ConstantEmitter { |
| 174 | private: |
| 175 | std::vector<std::string> Flags; |
| 176 | public: |
| 177 | FlagsConstantEmitter() { |
| 178 | } |
| 179 | FlagsConstantEmitter &addEntry(const char *f) { |
| 180 | Flags.push_back(std::string(f)); |
| 181 | return *this; |
| 182 | } |
| 183 | void emit(raw_ostream &o, unsigned int &i) { |
| 184 | unsigned int index; |
| 185 | unsigned int numFlags = Flags.size(); |
| 186 | if (numFlags == 0) |
| 187 | o << "0"; |
| 188 | |
| 189 | for (index = 0; index < numFlags; ++index) { |
| 190 | o << Flags[index].c_str(); |
| 191 | if (index < (numFlags - 1)) |
| 192 | o << " | "; |
| 193 | } |
| 194 | } |
| 195 | }; |
| 196 | } |
| 197 | |
| 198 | EDEmitter::EDEmitter(RecordKeeper &R) : Records(R) { |
| 199 | } |
| 200 | |
| 201 | ////////////////////////////////////////////// |
| 202 | // Support functions for parsing AsmStrings // |
| 203 | ////////////////////////////////////////////// |
| 204 | |
| 205 | /// parseError - A better error reporter for use in AsmString parsers |
| 206 | /// |
| 207 | /// @arg asmString - The original assembly string, for use in the error report |
| 208 | /// @arg index - The character where the error occurred |
| 209 | /// @arg err - The text of the error itself |
| 210 | static void parseError(const std::string& asmString, |
| 211 | unsigned int index, |
| 212 | const char* err) { |
| 213 | errs() << "In: " << asmString.c_str() << "\n"; |
| 214 | errs() << "Error at " << format("%d", index) << ": " << err << "\n"; |
| 215 | llvm_unreachable("Parse error"); |
| 216 | } |
| 217 | |
| 218 | /// resolveBraces - Interprets the brace syntax in an AsmString in favor of just |
| 219 | /// one syntax, and returns the result. "{A}" is resolved to "A" for syntax 0 |
| 220 | /// and "" for all others; "{A|B}" is resolved to "A" for syntax 0, "B" for |
| 221 | /// syntax 1, and "" for all others; and so on. |
| 222 | /// |
| 223 | /// @arg asmString - The original string, as loaded from the .td file |
| 224 | /// @arg syntaxIndex - The index to use |
| 225 | static std::string resolveBraces(const std::string &asmString, |
| 226 | unsigned int syntaxIndex) { |
| 227 | std::string ret; |
| 228 | |
| 229 | unsigned int index; |
| 230 | unsigned int numChars = asmString.length(); |
| 231 | |
| 232 | // Brace parsing countable-state transducer |
| 233 | // |
| 234 | // STATES - -1, 0, 1, ..., error |
| 235 | // SYMBOLS - '{', '|', '}', ?, EOF |
| 236 | // START STATE - -1 |
| 237 | // |
| 238 | // state input -> state output |
| 239 | // -1 '{' -> 0 |
| 240 | // -1 '|' -> error |
| 241 | // -1 '}' -> error |
| 242 | // -1 ? -> -1 ? |
| 243 | // -1 EOF -> -1 |
| 244 | // n '{' -> error |
| 245 | // n '|' -> n+1 |
| 246 | // n '}' -> -1 |
| 247 | // n ? -> n ? if n == syntaxIndex |
| 248 | // if not |
| 249 | // n EOF -> error |
| 250 | |
| 251 | int state = -1; |
| 252 | |
| 253 | for (index = 0; index < numChars; ++index) { |
| 254 | char input = asmString[index]; |
| 255 | |
| 256 | switch (state) { |
| 257 | default: |
| 258 | switch (input) { |
| 259 | default: |
| 260 | if (state == (int)syntaxIndex) |
| 261 | ret.push_back(input); |
| 262 | break; |
| 263 | case '{': |
| 264 | parseError(asmString, index, "Nested { in AsmString"); |
| 265 | break; |
| 266 | case '|': |
| 267 | state++; |
| 268 | break; |
| 269 | case '}': |
| 270 | state = -1; |
| 271 | break; |
| 272 | } |
| 273 | break; |
| 274 | case -1: |
| 275 | switch (input) { |
| 276 | default: |
| 277 | ret.push_back(input); |
| 278 | break; |
| 279 | case '{': |
| 280 | state = 0; |
| 281 | break; |
| 282 | case '|': |
| 283 | parseError(asmString, index, "| outside braces in AsmString"); |
| 284 | break; |
| 285 | case '}': |
| 286 | parseError(asmString, index, "Unmatched } in AsmString"); |
| 287 | break; |
| 288 | } |
| 289 | break; |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | if (state != -1) |
| 294 | parseError(asmString, index, "Unmatched { in AsmString"); |
| 295 | |
| 296 | return ret; |
| 297 | } |
| 298 | |
| 299 | /// getOperandIndex - looks up a named operand in an instruction and determines |
| 300 | /// its index in the operand descriptor array, returning the index or -1 if it |
| 301 | /// is not present. |
| 302 | /// |
| 303 | /// @arg asmString - The assembly string for the instruction, for errors only |
| 304 | /// @arg operand - The operand's name |
| 305 | /// @arg inst - The instruction to use when looking up the operand |
| 306 | static int8_t getOperandIndex(const std::string &asmString, |
| 307 | const std::string &operand, |
| 308 | const CodeGenInstruction &inst) { |
| 309 | int8_t operandIndex; |
| 310 | |
| 311 | if(operand.length() == 0) { |
| 312 | errs() << "In: " << asmString << "\n"; |
| 313 | errs() << "Operand: " << operand << "\n"; |
| 314 | llvm_unreachable("Empty operand"); |
| 315 | } |
| 316 | |
| 317 | try { |
| 318 | operandIndex = inst.getOperandNamed(operand); |
| 319 | } |
| 320 | catch (...) { |
| 321 | return -1; |
| 322 | } |
| 323 | |
| 324 | return operandIndex; |
| 325 | } |
| 326 | |
| 327 | /// isAlphanumeric - returns true if a character is a valid alphanumeric |
| 328 | /// character, and false otherwise |
| 329 | /// |
| 330 | /// input - The character to query |
| 331 | static inline bool isAlphanumeric(char input) { |
| 332 | if((input >= 'a' && input <= 'z') || |
| 333 | (input >= 'A' && input <= 'Z') || |
| 334 | (input >= '0' && input <= '9') || |
| 335 | (input == '_')) |
| 336 | return true; |
| 337 | else |
| 338 | return false; |
| 339 | } |
| 340 | |
| 341 | /// populateOperandOrder - reads a resolved AsmString (see resolveBraces) and |
| 342 | /// records the index into the operand descriptor array for each operand in |
| 343 | /// that string, in the order of appearance. |
| 344 | /// |
| 345 | /// @arg operandOrder - The array that will be populated with the operand |
| 346 | /// mapping. Each entry will contain -1 (invalid index |
| 347 | /// into the operands present in the AsmString) or a number |
| 348 | /// representing an index in the operand descriptor array. |
| 349 | /// @arg asmString - The operand's name |
| 350 | /// @arg inst - The instruction to use when looking up the operand |
| 351 | void populateOperandOrder(CompoundConstantEmitter *operandOrder, |
| 352 | const std::string &asmString, |
| 353 | const CodeGenInstruction &inst) { |
| 354 | std::string aux; |
| 355 | |
| 356 | unsigned int index; |
| 357 | unsigned int numChars = asmString.length(); |
| 358 | unsigned int numArgs = 0; |
| 359 | |
| 360 | // Argument processing finite-state transducer |
| 361 | // |
| 362 | // STATES - 0, 1, error |
| 363 | // SYMBOLS - A(lphanumeric), '$', ?, EOF |
| 364 | // START STATE - 0 |
| 365 | // |
| 366 | // state input -> state aux |
| 367 | // 0 A -> 0 |
| 368 | // 0 '$' -> 1 |
| 369 | // 0 ? -> 0 |
| 370 | // 0 EOF -> 0 |
| 371 | // 1 A -> 1 A |
| 372 | // 1 '$' -> error |
| 373 | // 1 ? -> 0 clear |
| 374 | // 1 EOF -> 0 clear |
| 375 | |
| 376 | unsigned int state = 0; |
| 377 | |
| 378 | for (index = 0; index < numChars; ++index) { |
| 379 | char input = asmString[index]; |
| 380 | |
| 381 | switch (state) { |
| 382 | default: |
| 383 | parseError(asmString, index, "Parser in unreachable state"); |
| 384 | case 0: |
| 385 | if (input == '$') { |
| 386 | state = 1; |
| 387 | } |
| 388 | break; |
| 389 | case 1: |
| 390 | if (isAlphanumeric(input)) { |
| 391 | aux.push_back(input); |
| 392 | } |
| 393 | else if (input == '$') { |
| 394 | parseError(asmString, index, "$ found in argument name"); |
| 395 | } |
| 396 | else { |
| 397 | int8_t operandIndex = getOperandIndex(asmString, aux, inst); |
| 398 | char buf[3]; |
| 399 | snprintf(buf, sizeof(buf), "%d", operandIndex); |
| 400 | operandOrder->addEntry(new LiteralConstantEmitter(buf)); |
| 401 | aux.clear(); |
| 402 | state = 0; |
| 403 | numArgs++; |
| 404 | } |
| 405 | break; |
| 406 | } |
| 407 | } |
| 408 | |
| 409 | if (state == 1) { |
| 410 | int8_t operandIndex = getOperandIndex(asmString, aux, inst); |
| 411 | char buf[2]; |
| 412 | snprintf(buf, 2, "%d", operandIndex); |
| 413 | operandOrder->addEntry(new LiteralConstantEmitter(buf)); |
| 414 | aux.clear(); |
| 415 | numArgs++; |
| 416 | } |
| 417 | |
| 418 | for(; numArgs < MAX_OPERANDS; numArgs++) { |
| 419 | operandOrder->addEntry(new LiteralConstantEmitter("-1")); |
| 420 | } |
| 421 | } |
| 422 | |
| 423 | ///////////////////////////////////////////////////// |
| 424 | // Support functions for handling X86 instructions // |
| 425 | ///////////////////////////////////////////////////// |
| 426 | |
| 427 | #define ADDFLAG(flag) flags->addEntry(flag) |
| 428 | |
| 429 | #define REG(str) if (name == str) { ADDFLAG("kOperandFlagRegister"); return 0; } |
| 430 | #define MEM(str) if (name == str) { ADDFLAG("kOperandFlagMemory"); return 0; } |
| 431 | #define LEA(str) if (name == str) { ADDFLAG("kOperandFlagEffectiveAddress"); \ |
| 432 | return 0; } |
| 433 | #define IMM(str) if (name == str) { ADDFLAG("kOperandFlagImmediate"); \ |
| 434 | return 0; } |
| 435 | #define PCR(str) if (name == str) { ADDFLAG("kOperandFlagMemory"); \ |
| 436 | ADDFLAG("kOperandFlagPCRelative"); \ |
| 437 | return 0; } |
| 438 | |
| 439 | /// X86FlagFromOpName - Processes the name of a single X86 operand (which is |
| 440 | /// actually its type) and translates it into an operand flag |
| 441 | /// |
| 442 | /// @arg flags - The flags object to add the flag to |
| 443 | /// @arg name - The name of the operand |
| 444 | static int X86FlagFromOpName(FlagsConstantEmitter *flags, |
| 445 | const std::string &name) { |
| 446 | REG("GR8"); |
| 447 | REG("GR8_NOREX"); |
| 448 | REG("GR16"); |
| 449 | REG("GR32"); |
| 450 | REG("GR32_NOREX"); |
| 451 | REG("FR32"); |
| 452 | REG("RFP32"); |
| 453 | REG("GR64"); |
| 454 | REG("FR64"); |
| 455 | REG("VR64"); |
| 456 | REG("RFP64"); |
| 457 | REG("RFP80"); |
| 458 | REG("VR128"); |
| 459 | REG("RST"); |
| 460 | REG("SEGMENT_REG"); |
| 461 | REG("DEBUG_REG"); |
| 462 | REG("CONTROL_REG_32"); |
| 463 | REG("CONTROL_REG_64"); |
| 464 | |
| 465 | MEM("i8mem"); |
| 466 | MEM("i8mem_NOREX"); |
| 467 | MEM("i16mem"); |
| 468 | MEM("i32mem"); |
| 469 | MEM("f32mem"); |
| 470 | MEM("ssmem"); |
| 471 | MEM("opaque32mem"); |
| 472 | MEM("opaque48mem"); |
| 473 | MEM("i64mem"); |
| 474 | MEM("f64mem"); |
| 475 | MEM("sdmem"); |
| 476 | MEM("f80mem"); |
| 477 | MEM("opaque80mem"); |
| 478 | MEM("i128mem"); |
| 479 | MEM("f128mem"); |
| 480 | MEM("opaque512mem"); |
| 481 | |
| 482 | LEA("lea32mem"); |
| 483 | LEA("lea64_32mem"); |
| 484 | LEA("lea64mem"); |
| 485 | |
| 486 | IMM("i8imm"); |
| 487 | IMM("i16imm"); |
| 488 | IMM("i16i8imm"); |
| 489 | IMM("i32imm"); |
| 490 | IMM("i32imm_pcrel"); |
| 491 | IMM("i32i8imm"); |
| 492 | IMM("i64imm"); |
| 493 | IMM("i64i8imm"); |
| 494 | IMM("i64i32imm"); |
| 495 | IMM("i64i32imm_pcrel"); |
| 496 | IMM("SSECC"); |
| 497 | |
| 498 | PCR("brtarget8"); |
| 499 | PCR("offset8"); |
| 500 | PCR("offset16"); |
| 501 | PCR("offset32"); |
| 502 | PCR("offset64"); |
| 503 | PCR("brtarget"); |
| 504 | |
| 505 | return 1; |
| 506 | } |
| 507 | |
| 508 | #undef REG |
| 509 | #undef MEM |
| 510 | #undef LEA |
| 511 | #undef IMM |
| 512 | #undef PCR |
| 513 | #undef ADDFLAG |
| 514 | |
| 515 | /// X86PopulateOperands - Handles all the operands in an X86 instruction, adding |
| 516 | /// the appropriate flags to their descriptors |
| 517 | /// |
| 518 | /// @operandFlags - A reference the array of operand flag objects |
| 519 | /// @inst - The instruction to use as a source of information |
| 520 | static void X86PopulateOperands( |
| 521 | FlagsConstantEmitter *(&operandFlags)[MAX_OPERANDS], |
| 522 | const CodeGenInstruction &inst) { |
| 523 | if (!inst.TheDef->isSubClassOf("X86Inst")) |
| 524 | return; |
| 525 | |
| 526 | unsigned int index; |
| 527 | unsigned int numOperands = inst.OperandList.size(); |
| 528 | |
| 529 | for (index = 0; index < numOperands; ++index) { |
| 530 | const CodeGenInstruction::OperandInfo &operandInfo = |
| 531 | inst.OperandList[index]; |
| 532 | Record &rec = *operandInfo.Rec; |
| 533 | |
| 534 | if (X86FlagFromOpName(operandFlags[index], rec.getName())) { |
| 535 | errs() << "Operand type: " << rec.getName().c_str() << "\n"; |
| 536 | errs() << "Operand name: " << operandInfo.Name.c_str() << "\n"; |
| 537 | errs() << "Instruction mame: " << inst.TheDef->getName().c_str() << "\n"; |
| 538 | llvm_unreachable("Unhandled type"); |
| 539 | } |
| 540 | } |
| 541 | } |
| 542 | |
| 543 | /// decorate1 - Decorates a named operand with a new flag |
| 544 | /// |
| 545 | /// @operandFlags - The array of operand flag objects, which don't have names |
| 546 | /// @inst - The CodeGenInstruction, which provides a way to translate |
| 547 | /// between names and operand indices |
| 548 | /// @opName - The name of the operand |
| 549 | /// @flag - The name of the flag to add |
| 550 | static inline void decorate1(FlagsConstantEmitter *(&operandFlags)[MAX_OPERANDS], |
| 551 | const CodeGenInstruction &inst, |
| 552 | const char *opName, |
| 553 | const char *opFlag) { |
| 554 | unsigned opIndex; |
| 555 | |
| 556 | try { |
| 557 | opIndex = inst.getOperandNamed(std::string(opName)); |
| 558 | } |
| 559 | catch (...) { |
| 560 | errs() << "Instruction: " << inst.TheDef->getName().c_str() << "\n"; |
| 561 | errs() << "Operand name: " << opName << "\n"; |
| 562 | llvm_unreachable("Couldn't find operand"); |
| 563 | } |
| 564 | |
| 565 | operandFlags[opIndex]->addEntry(opFlag); |
| 566 | } |
| 567 | |
| 568 | #define DECORATE1(opName, opFlag) decorate1(operandFlags, inst, opName, opFlag) |
| 569 | |
| 570 | #define MOV(source, target) { \ |
| 571 | instFlags.addEntry("kInstructionFlagMove"); \ |
| 572 | DECORATE1(source, "kOperandFlagSource"); \ |
| 573 | DECORATE1(target, "kOperandFlagTarget"); \ |
| 574 | } |
| 575 | |
| 576 | #define BRANCH(target) { \ |
| 577 | instFlags.addEntry("kInstructionFlagBranch"); \ |
| 578 | DECORATE1(target, "kOperandFlagTarget"); \ |
| 579 | } |
| 580 | |
| 581 | #define PUSH(source) { \ |
| 582 | instFlags.addEntry("kInstructionFlagPush"); \ |
| 583 | DECORATE1(source, "kOperandFlagSource"); \ |
| 584 | } |
| 585 | |
| 586 | #define POP(target) { \ |
| 587 | instFlags.addEntry("kInstructionFlagPop"); \ |
| 588 | DECORATE1(target, "kOperandFlagTarget"); \ |
| 589 | } |
| 590 | |
| 591 | #define CALL(target) { \ |
| 592 | instFlags.addEntry("kInstructionFlagCall"); \ |
| 593 | DECORATE1(target, "kOperandFlagTarget"); \ |
| 594 | } |
| 595 | |
| 596 | #define RETURN() { \ |
| 597 | instFlags.addEntry("kInstructionFlagReturn"); \ |
| 598 | } |
| 599 | |
| 600 | /// X86ExtractSemantics - Performs various checks on the name of an X86 |
| 601 | /// instruction to determine what sort of an instruction it is and then adds |
| 602 | /// the appropriate flags to the instruction and its operands |
| 603 | /// |
| 604 | /// @arg instFlags - A reference to the flags for the instruction as a whole |
| 605 | /// @arg operandFlags - A reference to the array of operand flag object pointers |
| 606 | /// @arg inst - A reference to the original instruction |
| 607 | static void X86ExtractSemantics(FlagsConstantEmitter &instFlags, |
| 608 | FlagsConstantEmitter *(&operandFlags)[MAX_OPERANDS], |
| 609 | const CodeGenInstruction &inst) { |
| 610 | const std::string &name = inst.TheDef->getName(); |
| 611 | |
| 612 | if (name.find("MOV") != name.npos) { |
| 613 | if (name.find("MOV_V") != name.npos) { |
| 614 | // ignore (this is a pseudoinstruction) |
| 615 | } |
| 616 | else if (name.find("MASK") != name.npos) { |
| 617 | // ignore (this is a masking move) |
| 618 | } |
| 619 | else if (name.find("r0") != name.npos) { |
| 620 | // ignore (this is a pseudoinstruction) |
| 621 | } |
| 622 | else if (name.find("PS") != name.npos || |
| 623 | name.find("PD") != name.npos) { |
| 624 | // ignore (this is a shuffling move) |
| 625 | } |
| 626 | else if (name.find("MOVS") != name.npos) { |
| 627 | // ignore (this is a string move) |
| 628 | } |
| 629 | else if (name.find("_F") != name.npos) { |
| 630 | // TODO handle _F moves to ST(0) |
| 631 | } |
| 632 | else if (name.find("a") != name.npos) { |
| 633 | // TODO handle moves to/from %ax |
| 634 | } |
| 635 | else if (name.find("CMOV") != name.npos) { |
| 636 | MOV("src2", "dst"); |
| 637 | } |
| 638 | else if (name.find("PC") != name.npos) { |
| 639 | MOV("label", "reg") |
| 640 | } |
| 641 | else { |
| 642 | MOV("src", "dst"); |
| 643 | } |
| 644 | } |
| 645 | |
| 646 | if (name.find("JMP") != name.npos || |
| 647 | name.find("J") == 0) { |
| 648 | if (name.find("FAR") != name.npos && name.find("i") != name.npos) { |
| 649 | BRANCH("off"); |
| 650 | } |
| 651 | else { |
| 652 | BRANCH("dst"); |
| 653 | } |
| 654 | } |
| 655 | |
| 656 | if (name.find("PUSH") != name.npos) { |
| 657 | if (name.find("FS") != name.npos || |
| 658 | name.find("GS") != name.npos) { |
| 659 | instFlags.addEntry("kInstructionFlagPush"); |
| 660 | // TODO add support for fixed operands |
| 661 | } |
| 662 | else if (name.find("F") != name.npos) { |
| 663 | // ignore (this pushes onto the FP stack) |
| 664 | } |
| 665 | else if (name[name.length() - 1] == 'm') { |
| 666 | PUSH("src"); |
| 667 | } |
| 668 | else if (name.find("i") != name.npos) { |
| 669 | PUSH("imm"); |
| 670 | } |
| 671 | else { |
| 672 | PUSH("reg"); |
| 673 | } |
| 674 | } |
| 675 | |
| 676 | if (name.find("POP") != name.npos) { |
| 677 | if (name.find("POPCNT") != name.npos) { |
| 678 | // ignore (not a real pop) |
| 679 | } |
| 680 | else if (name.find("FS") != name.npos || |
| 681 | name.find("GS") != name.npos) { |
| 682 | instFlags.addEntry("kInstructionFlagPop"); |
| 683 | // TODO add support for fixed operands |
| 684 | } |
| 685 | else if (name.find("F") != name.npos) { |
| 686 | // ignore (this pops from the FP stack) |
| 687 | } |
| 688 | else if (name[name.length() - 1] == 'm') { |
| 689 | POP("dst"); |
| 690 | } |
| 691 | else { |
| 692 | POP("reg"); |
| 693 | } |
| 694 | } |
| 695 | |
| 696 | if (name.find("CALL") != name.npos) { |
| 697 | if (name.find("ADJ") != name.npos) { |
| 698 | // ignore (not a call) |
| 699 | } |
| 700 | else if (name.find("SYSCALL") != name.npos) { |
| 701 | // ignore (doesn't go anywhere we know about) |
| 702 | } |
| 703 | else if (name.find("VMCALL") != name.npos) { |
| 704 | // ignore (rather different semantics than a regular call) |
| 705 | } |
| 706 | else if (name.find("FAR") != name.npos && name.find("i") != name.npos) { |
| 707 | CALL("off"); |
| 708 | } |
| 709 | else { |
| 710 | CALL("dst"); |
| 711 | } |
| 712 | } |
| 713 | |
| 714 | if (name.find("RET") != name.npos) { |
| 715 | RETURN(); |
| 716 | } |
| 717 | } |
| 718 | |
| 719 | #undef MOV |
| 720 | #undef BRANCH |
| 721 | #undef PUSH |
| 722 | #undef POP |
| 723 | #undef CALL |
| 724 | #undef RETURN |
| 725 | |
| 726 | #undef COND_DECORATE_2 |
| 727 | #undef COND_DECORATE_1 |
| 728 | #undef DECORATE1 |
| 729 | |
| 730 | /// populateInstInfo - Fills an array of InstInfos with information about each |
| 731 | /// instruction in a target |
| 732 | /// |
| 733 | /// @arg infoArray - The array of InstInfo objects to populate |
| 734 | /// @arg target - The CodeGenTarget to use as a source of instructions |
| 735 | static void populateInstInfo(CompoundConstantEmitter &infoArray, |
| 736 | CodeGenTarget &target) { |
| 737 | std::vector<const CodeGenInstruction*> numberedInstructions; |
| 738 | target.getInstructionsByEnumValue(numberedInstructions); |
| 739 | |
| 740 | unsigned int index; |
| 741 | unsigned int numInstructions = numberedInstructions.size(); |
| 742 | |
| 743 | for (index = 0; index < numInstructions; ++index) { |
| 744 | const CodeGenInstruction& inst = *numberedInstructions[index]; |
| 745 | |
| 746 | CompoundConstantEmitter *infoStruct = new CompoundConstantEmitter; |
| 747 | infoArray.addEntry(infoStruct); |
| 748 | |
| 749 | FlagsConstantEmitter *instFlags = new FlagsConstantEmitter; |
| 750 | infoStruct->addEntry(instFlags); |
| 751 | |
| 752 | LiteralConstantEmitter *numOperandsEmitter = |
| 753 | new LiteralConstantEmitter(inst.OperandList.size()); |
| 754 | infoStruct->addEntry(numOperandsEmitter); |
| 755 | |
| 756 | CompoundConstantEmitter *operandFlagArray = new CompoundConstantEmitter; |
| 757 | infoStruct->addEntry(operandFlagArray); |
| 758 | |
| 759 | FlagsConstantEmitter *operandFlags[MAX_OPERANDS]; |
| 760 | |
| 761 | for (unsigned operandIndex = 0; operandIndex < MAX_OPERANDS; ++operandIndex) { |
| 762 | operandFlags[operandIndex] = new FlagsConstantEmitter; |
| 763 | operandFlagArray->addEntry(operandFlags[operandIndex]); |
| 764 | } |
| 765 | |
| 766 | unsigned numSyntaxes = 0; |
| 767 | |
| 768 | if (target.getName() == "X86") { |
| 769 | X86PopulateOperands(operandFlags, inst); |
| 770 | X86ExtractSemantics(*instFlags, operandFlags, inst); |
| 771 | numSyntaxes = 2; |
| 772 | } |
| 773 | |
| 774 | CompoundConstantEmitter *operandOrderArray = new CompoundConstantEmitter; |
| 775 | infoStruct->addEntry(operandOrderArray); |
| 776 | |
| 777 | for (unsigned syntaxIndex = 0; syntaxIndex < MAX_SYNTAXES; ++syntaxIndex) { |
| 778 | CompoundConstantEmitter *operandOrder = new CompoundConstantEmitter; |
| 779 | operandOrderArray->addEntry(operandOrder); |
| 780 | |
| 781 | if (syntaxIndex < numSyntaxes) { |
| 782 | std::string asmString = inst.AsmString; |
| 783 | asmString = resolveBraces(asmString, syntaxIndex); |
| 784 | populateOperandOrder(operandOrder, asmString, inst); |
| 785 | } |
| 786 | else { |
| 787 | for (unsigned operandIndex = 0; |
| 788 | operandIndex < MAX_OPERANDS; |
| 789 | ++operandIndex) { |
| 790 | operandOrder->addEntry(new LiteralConstantEmitter("-1")); |
| 791 | } |
| 792 | } |
| 793 | } |
| 794 | } |
| 795 | } |
| 796 | |
| 797 | void EDEmitter::run(raw_ostream &o) { |
| 798 | unsigned int i = 0; |
| 799 | |
| 800 | CompoundConstantEmitter infoArray; |
| 801 | CodeGenTarget target; |
| 802 | |
| 803 | populateInstInfo(infoArray, target); |
| 804 | |
| 805 | o << "InstInfo instInfo" << target.getName().c_str() << "[] = "; |
| 806 | infoArray.emit(o, i); |
| 807 | o << ";" << "\n"; |
| 808 | } |
| 809 | |
| 810 | void EDEmitter::runHeader(raw_ostream &o) { |
Sean Callanan | 9988ab0 | 2010-01-29 01:34:29 +0000 | [diff] [blame^] | 811 | EmitSourceFileHeader("Enhanced Disassembly Info Header", o); |
Sean Callanan | 95fcebd | 2010-01-29 00:21:04 +0000 | [diff] [blame] | 812 | |
Sean Callanan | 9988ab0 | 2010-01-29 01:34:29 +0000 | [diff] [blame^] | 813 | o << "#ifndef EDInfo_" << "\n"; |
| 814 | o << "#define EDInfo_" << "\n"; |
Sean Callanan | 95fcebd | 2010-01-29 00:21:04 +0000 | [diff] [blame] | 815 | o << "\n"; |
| 816 | o << "#include <inttypes.h>" << "\n"; |
| 817 | o << "\n"; |
| 818 | o << "#define MAX_OPERANDS " << format("%d", MAX_OPERANDS) << "\n"; |
| 819 | o << "#define MAX_SYNTAXES " << format("%d", MAX_SYNTAXES) << "\n"; |
| 820 | o << "\n"; |
| 821 | |
| 822 | unsigned int i = 0; |
| 823 | |
| 824 | EnumEmitter operandFlags("OperandFlags"); |
| 825 | operandFlags.addEntry("kOperandFlagImmediate"); |
| 826 | operandFlags.addEntry("kOperandFlagRegister"); |
| 827 | operandFlags.addEntry("kOperandFlagMemory"); |
| 828 | operandFlags.addEntry("kOperandFlagEffectiveAddress"); |
| 829 | operandFlags.addEntry("kOperandFlagPCRelative"); |
| 830 | operandFlags.addEntry("kOperandFlagSource"); |
| 831 | operandFlags.addEntry("kOperandFlagTarget"); |
| 832 | operandFlags.emitAsFlags(o, i); |
| 833 | |
| 834 | o << "\n"; |
| 835 | |
| 836 | EnumEmitter instructionFlags("InstructionFlags"); |
| 837 | instructionFlags.addEntry("kInstructionFlagMove"); |
| 838 | instructionFlags.addEntry("kInstructionFlagBranch"); |
| 839 | instructionFlags.addEntry("kInstructionFlagPush"); |
| 840 | instructionFlags.addEntry("kInstructionFlagPop"); |
| 841 | instructionFlags.addEntry("kInstructionFlagCall"); |
| 842 | instructionFlags.addEntry("kInstructionFlagReturn"); |
| 843 | instructionFlags.emitAsFlags(o, i); |
| 844 | |
| 845 | o << "\n"; |
| 846 | |
| 847 | StructEmitter instInfo("InstInfo"); |
| 848 | instInfo.addMember("uint32_t", "instructionFlags"); |
| 849 | instInfo.addMember("uint8_t", "numOperands"); |
| 850 | instInfo.addMember("uint8_t", "operandFlags[MAX_OPERANDS]"); |
| 851 | instInfo.addMember("const char", "operandOrders[MAX_SYNTAXES][MAX_OPERANDS]"); |
| 852 | instInfo.emit(o, i); |
| 853 | |
| 854 | o << "\n"; |
| 855 | o << "#endif" << "\n"; |
| 856 | } |