1 //===- MIParser.cpp - Machine instructions parser implementation ----------===//
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 file implements the parsing of machine instructions.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "MIParser.h"
15 #include "MILexer.h"
16 #include "llvm/ADT/StringMap.h"
17 #include "llvm/AsmParser/SlotMapping.h"
18 #include "llvm/CodeGen/MachineBasicBlock.h"
19 #include "llvm/CodeGen/MachineFunction.h"
20 #include "llvm/CodeGen/MachineFrameInfo.h"
21 #include "llvm/CodeGen/MachineInstr.h"
22 #include "llvm/CodeGen/MachineInstrBuilder.h"
23 #include "llvm/CodeGen/MachineModuleInfo.h"
24 #include "llvm/IR/Instructions.h"
25 #include "llvm/IR/Constants.h"
26 #include "llvm/IR/Module.h"
27 #include "llvm/IR/ModuleSlotTracker.h"
28 #include "llvm/IR/ValueSymbolTable.h"
29 #include "llvm/Support/raw_ostream.h"
30 #include "llvm/Support/SourceMgr.h"
31 #include "llvm/Target/TargetSubtargetInfo.h"
32 #include "llvm/Target/TargetInstrInfo.h"
33 
34 using namespace llvm;
35 
36 namespace {
37 
38 struct StringValueUtility {
39   StringRef String;
40   std::string UnescapedString;
41 
42   StringValueUtility(const MIToken &Token) {
43     if (Token.isStringValueQuoted()) {
44       Token.unescapeQuotedStringValue(UnescapedString);
45       String = UnescapedString;
46       return;
47     }
48     String = Token.stringValue();
49   }
50 
51   operator StringRef() const { return String; }
52 };
53 
54 /// A wrapper struct around the 'MachineOperand' struct that includes a source
55 /// range.
56 struct MachineOperandWithLocation {
57   MachineOperand Operand;
58   StringRef::iterator Begin;
59   StringRef::iterator End;
60 
61   MachineOperandWithLocation(const MachineOperand &Operand,
62                              StringRef::iterator Begin, StringRef::iterator End)
63       : Operand(Operand), Begin(Begin), End(End) {}
64 };
65 
66 class MIParser {
67   SourceMgr &SM;
68   MachineFunction &MF;
69   SMDiagnostic &Error;
70   StringRef Source, CurrentSource;
71   MIToken Token;
72   const PerFunctionMIParsingState &PFS;
73   /// Maps from indices to unnamed global values and metadata nodes.
74   const SlotMapping &IRSlots;
75   /// Maps from instruction names to op codes.
76   StringMap<unsigned> Names2InstrOpCodes;
77   /// Maps from register names to registers.
78   StringMap<unsigned> Names2Regs;
79   /// Maps from register mask names to register masks.
80   StringMap<const uint32_t *> Names2RegMasks;
81   /// Maps from subregister names to subregister indices.
82   StringMap<unsigned> Names2SubRegIndices;
83   /// Maps from slot numbers to function's unnamed basic blocks.
84   DenseMap<unsigned, const BasicBlock *> Slots2BasicBlocks;
85   /// Maps from target index names to target indices.
86   StringMap<int> Names2TargetIndices;
87 
88 public:
89   MIParser(SourceMgr &SM, MachineFunction &MF, SMDiagnostic &Error,
90            StringRef Source, const PerFunctionMIParsingState &PFS,
91            const SlotMapping &IRSlots);
92 
93   void lex();
94 
95   /// Report an error at the current location with the given message.
96   ///
97   /// This function always return true.
98   bool error(const Twine &Msg);
99 
100   /// Report an error at the given location with the given message.
101   ///
102   /// This function always return true.
103   bool error(StringRef::iterator Loc, const Twine &Msg);
104 
105   bool parse(MachineInstr *&MI);
106   bool parseStandaloneMBB(MachineBasicBlock *&MBB);
107   bool parseStandaloneNamedRegister(unsigned &Reg);
108   bool parseStandaloneVirtualRegister(unsigned &Reg);
109   bool parseStandaloneIRBlockReference(const BasicBlock *&BB);
110 
111   bool parseRegister(unsigned &Reg);
112   bool parseRegisterFlag(unsigned &Flags);
113   bool parseSubRegisterIndex(unsigned &SubReg);
114   bool parseRegisterOperand(MachineOperand &Dest, bool IsDef = false);
115   bool parseImmediateOperand(MachineOperand &Dest);
116   bool parseMBBReference(MachineBasicBlock *&MBB);
117   bool parseMBBOperand(MachineOperand &Dest);
118   bool parseStackObjectOperand(MachineOperand &Dest);
119   bool parseFixedStackObjectOperand(MachineOperand &Dest);
120   bool parseGlobalValue(GlobalValue *&GV);
121   bool parseGlobalAddressOperand(MachineOperand &Dest);
122   bool parseConstantPoolIndexOperand(MachineOperand &Dest);
123   bool parseJumpTableIndexOperand(MachineOperand &Dest);
124   bool parseExternalSymbolOperand(MachineOperand &Dest);
125   bool parseMDNode(MDNode *&Node);
126   bool parseMetadataOperand(MachineOperand &Dest);
127   bool parseCFIOffset(int &Offset);
128   bool parseCFIRegister(unsigned &Reg);
129   bool parseCFIOperand(MachineOperand &Dest);
130   bool parseIRBlock(BasicBlock *&BB, const Function &F);
131   bool parseBlockAddressOperand(MachineOperand &Dest);
132   bool parseTargetIndexOperand(MachineOperand &Dest);
133   bool parseMachineOperand(MachineOperand &Dest);
134 
135 private:
136   /// Convert the integer literal in the current token into an unsigned integer.
137   ///
138   /// Return true if an error occurred.
139   bool getUnsigned(unsigned &Result);
140 
141   /// If the current token is of the given kind, consume it and return false.
142   /// Otherwise report an error and return true.
143   bool expectAndConsume(MIToken::TokenKind TokenKind);
144 
145   void initNames2InstrOpCodes();
146 
147   /// Try to convert an instruction name to an opcode. Return true if the
148   /// instruction name is invalid.
149   bool parseInstrName(StringRef InstrName, unsigned &OpCode);
150 
151   bool parseInstruction(unsigned &OpCode, unsigned &Flags);
152 
153   bool verifyImplicitOperands(ArrayRef<MachineOperandWithLocation> Operands,
154                               const MCInstrDesc &MCID);
155 
156   void initNames2Regs();
157 
158   /// Try to convert a register name to a register number. Return true if the
159   /// register name is invalid.
160   bool getRegisterByName(StringRef RegName, unsigned &Reg);
161 
162   void initNames2RegMasks();
163 
164   /// Check if the given identifier is a name of a register mask.
165   ///
166   /// Return null if the identifier isn't a register mask.
167   const uint32_t *getRegMask(StringRef Identifier);
168 
169   void initNames2SubRegIndices();
170 
171   /// Check if the given identifier is a name of a subregister index.
172   ///
173   /// Return 0 if the name isn't a subregister index class.
174   unsigned getSubRegIndex(StringRef Name);
175 
176   void initSlots2BasicBlocks();
177 
178   const BasicBlock *getIRBlock(unsigned Slot);
179 
180   void initNames2TargetIndices();
181 
182   /// Try to convert a name of target index to the corresponding target index.
183   ///
184   /// Return true if the name isn't a name of a target index.
185   bool getTargetIndex(StringRef Name, int &Index);
186 };
187 
188 } // end anonymous namespace
189 
190 MIParser::MIParser(SourceMgr &SM, MachineFunction &MF, SMDiagnostic &Error,
191                    StringRef Source, const PerFunctionMIParsingState &PFS,
192                    const SlotMapping &IRSlots)
193     : SM(SM), MF(MF), Error(Error), Source(Source), CurrentSource(Source),
194       Token(MIToken::Error, StringRef()), PFS(PFS), IRSlots(IRSlots) {}
195 
196 void MIParser::lex() {
197   CurrentSource = lexMIToken(
198       CurrentSource, Token,
199       [this](StringRef::iterator Loc, const Twine &Msg) { error(Loc, Msg); });
200 }
201 
202 bool MIParser::error(const Twine &Msg) { return error(Token.location(), Msg); }
203 
204 bool MIParser::error(StringRef::iterator Loc, const Twine &Msg) {
205   assert(Loc >= Source.data() && Loc <= (Source.data() + Source.size()));
206   Error = SMDiagnostic(
207       SM, SMLoc(),
208       SM.getMemoryBuffer(SM.getMainFileID())->getBufferIdentifier(), 1,
209       Loc - Source.data(), SourceMgr::DK_Error, Msg.str(), Source, None, None);
210   return true;
211 }
212 
213 static const char *toString(MIToken::TokenKind TokenKind) {
214   switch (TokenKind) {
215   case MIToken::comma:
216     return "','";
217   case MIToken::equal:
218     return "'='";
219   case MIToken::lparen:
220     return "'('";
221   case MIToken::rparen:
222     return "')'";
223   default:
224     return "<unknown token>";
225   }
226 }
227 
228 bool MIParser::expectAndConsume(MIToken::TokenKind TokenKind) {
229   if (Token.isNot(TokenKind))
230     return error(Twine("expected ") + toString(TokenKind));
231   lex();
232   return false;
233 }
234 
235 bool MIParser::parse(MachineInstr *&MI) {
236   lex();
237 
238   // Parse any register operands before '='
239   MachineOperand MO = MachineOperand::CreateImm(0);
240   SmallVector<MachineOperandWithLocation, 8> Operands;
241   while (Token.isRegister() || Token.isRegisterFlag()) {
242     auto Loc = Token.location();
243     if (parseRegisterOperand(MO, /*IsDef=*/true))
244       return true;
245     Operands.push_back(MachineOperandWithLocation(MO, Loc, Token.location()));
246     if (Token.isNot(MIToken::comma))
247       break;
248     lex();
249   }
250   if (!Operands.empty() && expectAndConsume(MIToken::equal))
251     return true;
252 
253   unsigned OpCode, Flags = 0;
254   if (Token.isError() || parseInstruction(OpCode, Flags))
255     return true;
256 
257   // TODO: Parse the bundle instruction flags and memory operands.
258 
259   // Parse the remaining machine operands.
260   while (Token.isNot(MIToken::Eof) && Token.isNot(MIToken::kw_debug_location)) {
261     auto Loc = Token.location();
262     if (parseMachineOperand(MO))
263       return true;
264     Operands.push_back(MachineOperandWithLocation(MO, Loc, Token.location()));
265     if (Token.is(MIToken::Eof))
266       break;
267     if (Token.isNot(MIToken::comma))
268       return error("expected ',' before the next machine operand");
269     lex();
270   }
271 
272   DebugLoc DebugLocation;
273   if (Token.is(MIToken::kw_debug_location)) {
274     lex();
275     if (Token.isNot(MIToken::exclaim))
276       return error("expected a metadata node after 'debug-location'");
277     MDNode *Node = nullptr;
278     if (parseMDNode(Node))
279       return true;
280     DebugLocation = DebugLoc(Node);
281   }
282 
283   const auto &MCID = MF.getSubtarget().getInstrInfo()->get(OpCode);
284   if (!MCID.isVariadic()) {
285     // FIXME: Move the implicit operand verification to the machine verifier.
286     if (verifyImplicitOperands(Operands, MCID))
287       return true;
288   }
289 
290   // TODO: Check for extraneous machine operands.
291   MI = MF.CreateMachineInstr(MCID, DebugLocation, /*NoImplicit=*/true);
292   MI->setFlags(Flags);
293   for (const auto &Operand : Operands)
294     MI->addOperand(MF, Operand.Operand);
295   return false;
296 }
297 
298 bool MIParser::parseStandaloneMBB(MachineBasicBlock *&MBB) {
299   lex();
300   if (Token.isNot(MIToken::MachineBasicBlock))
301     return error("expected a machine basic block reference");
302   if (parseMBBReference(MBB))
303     return true;
304   lex();
305   if (Token.isNot(MIToken::Eof))
306     return error(
307         "expected end of string after the machine basic block reference");
308   return false;
309 }
310 
311 bool MIParser::parseStandaloneNamedRegister(unsigned &Reg) {
312   lex();
313   if (Token.isNot(MIToken::NamedRegister))
314     return error("expected a named register");
315   if (parseRegister(Reg))
316     return 0;
317   lex();
318   if (Token.isNot(MIToken::Eof))
319     return error("expected end of string after the register reference");
320   return false;
321 }
322 
323 bool MIParser::parseStandaloneVirtualRegister(unsigned &Reg) {
324   lex();
325   if (Token.isNot(MIToken::VirtualRegister))
326     return error("expected a virtual register");
327   if (parseRegister(Reg))
328     return 0;
329   lex();
330   if (Token.isNot(MIToken::Eof))
331     return error("expected end of string after the register reference");
332   return false;
333 }
334 
335 bool MIParser::parseStandaloneIRBlockReference(const BasicBlock *&BB) {
336   lex();
337   if (Token.isNot(MIToken::IRBlock))
338     return error("expected an IR block reference");
339   unsigned SlotNumber = 0;
340   if (getUnsigned(SlotNumber))
341     return true;
342   BB = getIRBlock(SlotNumber);
343   if (!BB)
344     return error(Twine("use of undefined IR block '%ir-block.") +
345                  Twine(SlotNumber) + "'");
346   lex();
347   if (Token.isNot(MIToken::Eof))
348     return error("expected end of string after the IR block reference");
349   return false;
350 }
351 
352 static const char *printImplicitRegisterFlag(const MachineOperand &MO) {
353   assert(MO.isImplicit());
354   return MO.isDef() ? "implicit-def" : "implicit";
355 }
356 
357 static std::string getRegisterName(const TargetRegisterInfo *TRI,
358                                    unsigned Reg) {
359   assert(TargetRegisterInfo::isPhysicalRegister(Reg) && "expected phys reg");
360   return StringRef(TRI->getName(Reg)).lower();
361 }
362 
363 bool MIParser::verifyImplicitOperands(
364     ArrayRef<MachineOperandWithLocation> Operands, const MCInstrDesc &MCID) {
365   if (MCID.isCall())
366     // We can't verify call instructions as they can contain arbitrary implicit
367     // register and register mask operands.
368     return false;
369 
370   // Gather all the expected implicit operands.
371   SmallVector<MachineOperand, 4> ImplicitOperands;
372   if (MCID.ImplicitDefs)
373     for (const uint16_t *ImpDefs = MCID.getImplicitDefs(); *ImpDefs; ++ImpDefs)
374       ImplicitOperands.push_back(
375           MachineOperand::CreateReg(*ImpDefs, true, true));
376   if (MCID.ImplicitUses)
377     for (const uint16_t *ImpUses = MCID.getImplicitUses(); *ImpUses; ++ImpUses)
378       ImplicitOperands.push_back(
379           MachineOperand::CreateReg(*ImpUses, false, true));
380 
381   const auto *TRI = MF.getSubtarget().getRegisterInfo();
382   assert(TRI && "Expected target register info");
383   size_t I = ImplicitOperands.size(), J = Operands.size();
384   while (I) {
385     --I;
386     if (J) {
387       --J;
388       const auto &ImplicitOperand = ImplicitOperands[I];
389       const auto &Operand = Operands[J].Operand;
390       if (ImplicitOperand.isIdenticalTo(Operand))
391         continue;
392       if (Operand.isReg() && Operand.isImplicit()) {
393         return error(Operands[J].Begin,
394                      Twine("expected an implicit register operand '") +
395                          printImplicitRegisterFlag(ImplicitOperand) + " %" +
396                          getRegisterName(TRI, ImplicitOperand.getReg()) + "'");
397       }
398     }
399     // TODO: Fix source location when Operands[J].end is right before '=', i.e:
400     // insead of reporting an error at this location:
401     //            %eax = MOV32r0
402     //                 ^
403     // report the error at the following location:
404     //            %eax = MOV32r0
405     //                          ^
406     return error(J < Operands.size() ? Operands[J].End : Token.location(),
407                  Twine("missing implicit register operand '") +
408                      printImplicitRegisterFlag(ImplicitOperands[I]) + " %" +
409                      getRegisterName(TRI, ImplicitOperands[I].getReg()) + "'");
410   }
411   return false;
412 }
413 
414 bool MIParser::parseInstruction(unsigned &OpCode, unsigned &Flags) {
415   if (Token.is(MIToken::kw_frame_setup)) {
416     Flags |= MachineInstr::FrameSetup;
417     lex();
418   }
419   if (Token.isNot(MIToken::Identifier))
420     return error("expected a machine instruction");
421   StringRef InstrName = Token.stringValue();
422   if (parseInstrName(InstrName, OpCode))
423     return error(Twine("unknown machine instruction name '") + InstrName + "'");
424   lex();
425   return false;
426 }
427 
428 bool MIParser::parseRegister(unsigned &Reg) {
429   switch (Token.kind()) {
430   case MIToken::underscore:
431     Reg = 0;
432     break;
433   case MIToken::NamedRegister: {
434     StringRef Name = Token.stringValue();
435     if (getRegisterByName(Name, Reg))
436       return error(Twine("unknown register name '") + Name + "'");
437     break;
438   }
439   case MIToken::VirtualRegister: {
440     unsigned ID;
441     if (getUnsigned(ID))
442       return true;
443     const auto RegInfo = PFS.VirtualRegisterSlots.find(ID);
444     if (RegInfo == PFS.VirtualRegisterSlots.end())
445       return error(Twine("use of undefined virtual register '%") + Twine(ID) +
446                    "'");
447     Reg = RegInfo->second;
448     break;
449   }
450   // TODO: Parse other register kinds.
451   default:
452     llvm_unreachable("The current token should be a register");
453   }
454   return false;
455 }
456 
457 bool MIParser::parseRegisterFlag(unsigned &Flags) {
458   switch (Token.kind()) {
459   case MIToken::kw_implicit:
460     Flags |= RegState::Implicit;
461     break;
462   case MIToken::kw_implicit_define:
463     Flags |= RegState::ImplicitDefine;
464     break;
465   case MIToken::kw_dead:
466     Flags |= RegState::Dead;
467     break;
468   case MIToken::kw_killed:
469     Flags |= RegState::Kill;
470     break;
471   case MIToken::kw_undef:
472     Flags |= RegState::Undef;
473     break;
474   // TODO: report an error when we specify the same flag more than once.
475   // TODO: parse the other register flags.
476   default:
477     llvm_unreachable("The current token should be a register flag");
478   }
479   lex();
480   return false;
481 }
482 
483 bool MIParser::parseSubRegisterIndex(unsigned &SubReg) {
484   assert(Token.is(MIToken::colon));
485   lex();
486   if (Token.isNot(MIToken::Identifier))
487     return error("expected a subregister index after ':'");
488   auto Name = Token.stringValue();
489   SubReg = getSubRegIndex(Name);
490   if (!SubReg)
491     return error(Twine("use of unknown subregister index '") + Name + "'");
492   lex();
493   return false;
494 }
495 
496 bool MIParser::parseRegisterOperand(MachineOperand &Dest, bool IsDef) {
497   unsigned Reg;
498   unsigned Flags = IsDef ? RegState::Define : 0;
499   while (Token.isRegisterFlag()) {
500     if (parseRegisterFlag(Flags))
501       return true;
502   }
503   if (!Token.isRegister())
504     return error("expected a register after register flags");
505   if (parseRegister(Reg))
506     return true;
507   lex();
508   unsigned SubReg = 0;
509   if (Token.is(MIToken::colon)) {
510     if (parseSubRegisterIndex(SubReg))
511       return true;
512   }
513   Dest = MachineOperand::CreateReg(
514       Reg, Flags & RegState::Define, Flags & RegState::Implicit,
515       Flags & RegState::Kill, Flags & RegState::Dead, Flags & RegState::Undef,
516       /*isEarlyClobber=*/false, SubReg);
517   return false;
518 }
519 
520 bool MIParser::parseImmediateOperand(MachineOperand &Dest) {
521   assert(Token.is(MIToken::IntegerLiteral));
522   const APSInt &Int = Token.integerValue();
523   if (Int.getMinSignedBits() > 64)
524     // TODO: Replace this with an error when we can parse CIMM Machine Operands.
525     llvm_unreachable("Can't parse large integer literals yet!");
526   Dest = MachineOperand::CreateImm(Int.getExtValue());
527   lex();
528   return false;
529 }
530 
531 bool MIParser::getUnsigned(unsigned &Result) {
532   assert(Token.hasIntegerValue() && "Expected a token with an integer value");
533   const uint64_t Limit = uint64_t(std::numeric_limits<unsigned>::max()) + 1;
534   uint64_t Val64 = Token.integerValue().getLimitedValue(Limit);
535   if (Val64 == Limit)
536     return error("expected 32-bit integer (too large)");
537   Result = Val64;
538   return false;
539 }
540 
541 bool MIParser::parseMBBReference(MachineBasicBlock *&MBB) {
542   assert(Token.is(MIToken::MachineBasicBlock));
543   unsigned Number;
544   if (getUnsigned(Number))
545     return true;
546   auto MBBInfo = PFS.MBBSlots.find(Number);
547   if (MBBInfo == PFS.MBBSlots.end())
548     return error(Twine("use of undefined machine basic block #") +
549                  Twine(Number));
550   MBB = MBBInfo->second;
551   if (!Token.stringValue().empty() && Token.stringValue() != MBB->getName())
552     return error(Twine("the name of machine basic block #") + Twine(Number) +
553                  " isn't '" + Token.stringValue() + "'");
554   return false;
555 }
556 
557 bool MIParser::parseMBBOperand(MachineOperand &Dest) {
558   MachineBasicBlock *MBB;
559   if (parseMBBReference(MBB))
560     return true;
561   Dest = MachineOperand::CreateMBB(MBB);
562   lex();
563   return false;
564 }
565 
566 bool MIParser::parseStackObjectOperand(MachineOperand &Dest) {
567   assert(Token.is(MIToken::StackObject));
568   unsigned ID;
569   if (getUnsigned(ID))
570     return true;
571   auto ObjectInfo = PFS.StackObjectSlots.find(ID);
572   if (ObjectInfo == PFS.StackObjectSlots.end())
573     return error(Twine("use of undefined stack object '%stack.") + Twine(ID) +
574                  "'");
575   StringRef Name;
576   if (const auto *Alloca =
577           MF.getFrameInfo()->getObjectAllocation(ObjectInfo->second))
578     Name = Alloca->getName();
579   if (!Token.stringValue().empty() && Token.stringValue() != Name)
580     return error(Twine("the name of the stack object '%stack.") + Twine(ID) +
581                  "' isn't '" + Token.stringValue() + "'");
582   lex();
583   Dest = MachineOperand::CreateFI(ObjectInfo->second);
584   return false;
585 }
586 
587 bool MIParser::parseFixedStackObjectOperand(MachineOperand &Dest) {
588   assert(Token.is(MIToken::FixedStackObject));
589   unsigned ID;
590   if (getUnsigned(ID))
591     return true;
592   auto ObjectInfo = PFS.FixedStackObjectSlots.find(ID);
593   if (ObjectInfo == PFS.FixedStackObjectSlots.end())
594     return error(Twine("use of undefined fixed stack object '%fixed-stack.") +
595                  Twine(ID) + "'");
596   lex();
597   Dest = MachineOperand::CreateFI(ObjectInfo->second);
598   return false;
599 }
600 
601 bool MIParser::parseGlobalValue(GlobalValue *&GV) {
602   switch (Token.kind()) {
603   case MIToken::NamedGlobalValue:
604   case MIToken::QuotedNamedGlobalValue: {
605     StringValueUtility Name(Token);
606     const Module *M = MF.getFunction()->getParent();
607     GV = M->getNamedValue(Name);
608     if (!GV)
609       return error(Twine("use of undefined global value '@") +
610                    Token.rawStringValue() + "'");
611     break;
612   }
613   case MIToken::GlobalValue: {
614     unsigned GVIdx;
615     if (getUnsigned(GVIdx))
616       return true;
617     if (GVIdx >= IRSlots.GlobalValues.size())
618       return error(Twine("use of undefined global value '@") + Twine(GVIdx) +
619                    "'");
620     GV = IRSlots.GlobalValues[GVIdx];
621     break;
622   }
623   default:
624     llvm_unreachable("The current token should be a global value");
625   }
626   return false;
627 }
628 
629 bool MIParser::parseGlobalAddressOperand(MachineOperand &Dest) {
630   GlobalValue *GV = nullptr;
631   if (parseGlobalValue(GV))
632     return true;
633   Dest = MachineOperand::CreateGA(GV, /*Offset=*/0);
634   // TODO: Parse offset and target flags.
635   lex();
636   return false;
637 }
638 
639 bool MIParser::parseConstantPoolIndexOperand(MachineOperand &Dest) {
640   assert(Token.is(MIToken::ConstantPoolItem));
641   unsigned ID;
642   if (getUnsigned(ID))
643     return true;
644   auto ConstantInfo = PFS.ConstantPoolSlots.find(ID);
645   if (ConstantInfo == PFS.ConstantPoolSlots.end())
646     return error("use of undefined constant '%const." + Twine(ID) + "'");
647   lex();
648   // TODO: Parse offset and target flags.
649   Dest = MachineOperand::CreateCPI(ID, /*Offset=*/0);
650   return false;
651 }
652 
653 bool MIParser::parseJumpTableIndexOperand(MachineOperand &Dest) {
654   assert(Token.is(MIToken::JumpTableIndex));
655   unsigned ID;
656   if (getUnsigned(ID))
657     return true;
658   auto JumpTableEntryInfo = PFS.JumpTableSlots.find(ID);
659   if (JumpTableEntryInfo == PFS.JumpTableSlots.end())
660     return error("use of undefined jump table '%jump-table." + Twine(ID) + "'");
661   lex();
662   // TODO: Parse target flags.
663   Dest = MachineOperand::CreateJTI(JumpTableEntryInfo->second);
664   return false;
665 }
666 
667 bool MIParser::parseExternalSymbolOperand(MachineOperand &Dest) {
668   assert(Token.is(MIToken::ExternalSymbol) ||
669          Token.is(MIToken::QuotedExternalSymbol));
670   StringValueUtility Name(Token);
671   const char *Symbol = MF.createExternalSymbolName(Name);
672   lex();
673   // TODO: Parse the target flags.
674   Dest = MachineOperand::CreateES(Symbol);
675   return false;
676 }
677 
678 bool MIParser::parseMDNode(MDNode *&Node) {
679   assert(Token.is(MIToken::exclaim));
680   auto Loc = Token.location();
681   lex();
682   if (Token.isNot(MIToken::IntegerLiteral) || Token.integerValue().isSigned())
683     return error("expected metadata id after '!'");
684   unsigned ID;
685   if (getUnsigned(ID))
686     return true;
687   auto NodeInfo = IRSlots.MetadataNodes.find(ID);
688   if (NodeInfo == IRSlots.MetadataNodes.end())
689     return error(Loc, "use of undefined metadata '!" + Twine(ID) + "'");
690   lex();
691   Node = NodeInfo->second.get();
692   return false;
693 }
694 
695 bool MIParser::parseMetadataOperand(MachineOperand &Dest) {
696   MDNode *Node = nullptr;
697   if (parseMDNode(Node))
698     return true;
699   Dest = MachineOperand::CreateMetadata(Node);
700   return false;
701 }
702 
703 bool MIParser::parseCFIOffset(int &Offset) {
704   if (Token.isNot(MIToken::IntegerLiteral))
705     return error("expected a cfi offset");
706   if (Token.integerValue().getMinSignedBits() > 32)
707     return error("expected a 32 bit integer (the cfi offset is too large)");
708   Offset = (int)Token.integerValue().getExtValue();
709   lex();
710   return false;
711 }
712 
713 bool MIParser::parseCFIRegister(unsigned &Reg) {
714   if (Token.isNot(MIToken::NamedRegister))
715     return error("expected a cfi register");
716   unsigned LLVMReg;
717   if (parseRegister(LLVMReg))
718     return true;
719   const auto *TRI = MF.getSubtarget().getRegisterInfo();
720   assert(TRI && "Expected target register info");
721   int DwarfReg = TRI->getDwarfRegNum(LLVMReg, true);
722   if (DwarfReg < 0)
723     return error("invalid DWARF register");
724   Reg = (unsigned)DwarfReg;
725   lex();
726   return false;
727 }
728 
729 bool MIParser::parseCFIOperand(MachineOperand &Dest) {
730   auto Kind = Token.kind();
731   lex();
732   auto &MMI = MF.getMMI();
733   int Offset;
734   unsigned Reg;
735   unsigned CFIIndex;
736   switch (Kind) {
737   case MIToken::kw_cfi_offset:
738     if (parseCFIRegister(Reg) || expectAndConsume(MIToken::comma) ||
739         parseCFIOffset(Offset))
740       return true;
741     CFIIndex =
742         MMI.addFrameInst(MCCFIInstruction::createOffset(nullptr, Reg, Offset));
743     break;
744   case MIToken::kw_cfi_def_cfa_register:
745     if (parseCFIRegister(Reg))
746       return true;
747     CFIIndex =
748         MMI.addFrameInst(MCCFIInstruction::createDefCfaRegister(nullptr, Reg));
749     break;
750   case MIToken::kw_cfi_def_cfa_offset:
751     if (parseCFIOffset(Offset))
752       return true;
753     // NB: MCCFIInstruction::createDefCfaOffset negates the offset.
754     CFIIndex = MMI.addFrameInst(
755         MCCFIInstruction::createDefCfaOffset(nullptr, -Offset));
756     break;
757   case MIToken::kw_cfi_def_cfa:
758     if (parseCFIRegister(Reg) || expectAndConsume(MIToken::comma) ||
759         parseCFIOffset(Offset))
760       return true;
761     // NB: MCCFIInstruction::createDefCfa negates the offset.
762     CFIIndex =
763         MMI.addFrameInst(MCCFIInstruction::createDefCfa(nullptr, Reg, -Offset));
764     break;
765   default:
766     // TODO: Parse the other CFI operands.
767     llvm_unreachable("The current token should be a cfi operand");
768   }
769   Dest = MachineOperand::CreateCFIIndex(CFIIndex);
770   return false;
771 }
772 
773 bool MIParser::parseIRBlock(BasicBlock *&BB, const Function &F) {
774   switch (Token.kind()) {
775   case MIToken::NamedIRBlock:
776   case MIToken::QuotedNamedIRBlock: {
777     StringValueUtility Name(Token);
778     BB = dyn_cast_or_null<BasicBlock>(F.getValueSymbolTable().lookup(Name));
779     if (!BB)
780       return error(Twine("use of undefined IR block '%ir-block.") +
781                    Token.rawStringValue() + "'");
782     break;
783   }
784   case MIToken::IRBlock: {
785     unsigned SlotNumber = 0;
786     if (getUnsigned(SlotNumber))
787       return true;
788     BB = const_cast<BasicBlock *>(getIRBlock(SlotNumber));
789     if (!BB)
790       return error(Twine("use of undefined IR block '%ir-block.") +
791                    Twine(SlotNumber) + "'");
792     break;
793   }
794   default:
795     llvm_unreachable("The current token should be an IR block reference");
796   }
797   return false;
798 }
799 
800 bool MIParser::parseBlockAddressOperand(MachineOperand &Dest) {
801   assert(Token.is(MIToken::kw_blockaddress));
802   lex();
803   if (expectAndConsume(MIToken::lparen))
804     return true;
805   if (Token.isNot(MIToken::GlobalValue) &&
806       Token.isNot(MIToken::NamedGlobalValue) &&
807       Token.isNot(MIToken::QuotedNamedGlobalValue))
808     return error("expected a global value");
809   GlobalValue *GV = nullptr;
810   if (parseGlobalValue(GV))
811     return true;
812   auto *F = dyn_cast<Function>(GV);
813   if (!F)
814     return error("expected an IR function reference");
815   lex();
816   if (expectAndConsume(MIToken::comma))
817     return true;
818   BasicBlock *BB = nullptr;
819   if (Token.isNot(MIToken::IRBlock) && Token.isNot(MIToken::NamedIRBlock) &&
820       Token.isNot(MIToken::QuotedNamedIRBlock))
821     return error("expected an IR block reference");
822   if (parseIRBlock(BB, *F))
823     return true;
824   lex();
825   if (expectAndConsume(MIToken::rparen))
826     return true;
827   // TODO: parse offset and target flags.
828   Dest = MachineOperand::CreateBA(BlockAddress::get(F, BB), /*Offset=*/0);
829   return false;
830 }
831 
832 bool MIParser::parseTargetIndexOperand(MachineOperand &Dest) {
833   assert(Token.is(MIToken::kw_target_index));
834   lex();
835   if (expectAndConsume(MIToken::lparen))
836     return true;
837   if (Token.isNot(MIToken::Identifier))
838     return error("expected the name of the target index");
839   int Index = 0;
840   if (getTargetIndex(Token.stringValue(), Index))
841     return error("use of undefined target index '" + Token.stringValue() + "'");
842   lex();
843   if (expectAndConsume(MIToken::rparen))
844     return true;
845   // TODO: Parse the offset and target flags.
846   Dest = MachineOperand::CreateTargetIndex(unsigned(Index), /*Offset=*/0);
847   return false;
848 }
849 
850 bool MIParser::parseMachineOperand(MachineOperand &Dest) {
851   switch (Token.kind()) {
852   case MIToken::kw_implicit:
853   case MIToken::kw_implicit_define:
854   case MIToken::kw_dead:
855   case MIToken::kw_killed:
856   case MIToken::kw_undef:
857   case MIToken::underscore:
858   case MIToken::NamedRegister:
859   case MIToken::VirtualRegister:
860     return parseRegisterOperand(Dest);
861   case MIToken::IntegerLiteral:
862     return parseImmediateOperand(Dest);
863   case MIToken::MachineBasicBlock:
864     return parseMBBOperand(Dest);
865   case MIToken::StackObject:
866     return parseStackObjectOperand(Dest);
867   case MIToken::FixedStackObject:
868     return parseFixedStackObjectOperand(Dest);
869   case MIToken::GlobalValue:
870   case MIToken::NamedGlobalValue:
871   case MIToken::QuotedNamedGlobalValue:
872     return parseGlobalAddressOperand(Dest);
873   case MIToken::ConstantPoolItem:
874     return parseConstantPoolIndexOperand(Dest);
875   case MIToken::JumpTableIndex:
876     return parseJumpTableIndexOperand(Dest);
877   case MIToken::ExternalSymbol:
878   case MIToken::QuotedExternalSymbol:
879     return parseExternalSymbolOperand(Dest);
880   case MIToken::exclaim:
881     return parseMetadataOperand(Dest);
882   case MIToken::kw_cfi_offset:
883   case MIToken::kw_cfi_def_cfa_register:
884   case MIToken::kw_cfi_def_cfa_offset:
885   case MIToken::kw_cfi_def_cfa:
886     return parseCFIOperand(Dest);
887   case MIToken::kw_blockaddress:
888     return parseBlockAddressOperand(Dest);
889   case MIToken::kw_target_index:
890     return parseTargetIndexOperand(Dest);
891   case MIToken::Error:
892     return true;
893   case MIToken::Identifier:
894     if (const auto *RegMask = getRegMask(Token.stringValue())) {
895       Dest = MachineOperand::CreateRegMask(RegMask);
896       lex();
897       break;
898     }
899   // fallthrough
900   default:
901     // TODO: parse the other machine operands.
902     return error("expected a machine operand");
903   }
904   return false;
905 }
906 
907 void MIParser::initNames2InstrOpCodes() {
908   if (!Names2InstrOpCodes.empty())
909     return;
910   const auto *TII = MF.getSubtarget().getInstrInfo();
911   assert(TII && "Expected target instruction info");
912   for (unsigned I = 0, E = TII->getNumOpcodes(); I < E; ++I)
913     Names2InstrOpCodes.insert(std::make_pair(StringRef(TII->getName(I)), I));
914 }
915 
916 bool MIParser::parseInstrName(StringRef InstrName, unsigned &OpCode) {
917   initNames2InstrOpCodes();
918   auto InstrInfo = Names2InstrOpCodes.find(InstrName);
919   if (InstrInfo == Names2InstrOpCodes.end())
920     return true;
921   OpCode = InstrInfo->getValue();
922   return false;
923 }
924 
925 void MIParser::initNames2Regs() {
926   if (!Names2Regs.empty())
927     return;
928   // The '%noreg' register is the register 0.
929   Names2Regs.insert(std::make_pair("noreg", 0));
930   const auto *TRI = MF.getSubtarget().getRegisterInfo();
931   assert(TRI && "Expected target register info");
932   for (unsigned I = 0, E = TRI->getNumRegs(); I < E; ++I) {
933     bool WasInserted =
934         Names2Regs.insert(std::make_pair(StringRef(TRI->getName(I)).lower(), I))
935             .second;
936     (void)WasInserted;
937     assert(WasInserted && "Expected registers to be unique case-insensitively");
938   }
939 }
940 
941 bool MIParser::getRegisterByName(StringRef RegName, unsigned &Reg) {
942   initNames2Regs();
943   auto RegInfo = Names2Regs.find(RegName);
944   if (RegInfo == Names2Regs.end())
945     return true;
946   Reg = RegInfo->getValue();
947   return false;
948 }
949 
950 void MIParser::initNames2RegMasks() {
951   if (!Names2RegMasks.empty())
952     return;
953   const auto *TRI = MF.getSubtarget().getRegisterInfo();
954   assert(TRI && "Expected target register info");
955   ArrayRef<const uint32_t *> RegMasks = TRI->getRegMasks();
956   ArrayRef<const char *> RegMaskNames = TRI->getRegMaskNames();
957   assert(RegMasks.size() == RegMaskNames.size());
958   for (size_t I = 0, E = RegMasks.size(); I < E; ++I)
959     Names2RegMasks.insert(
960         std::make_pair(StringRef(RegMaskNames[I]).lower(), RegMasks[I]));
961 }
962 
963 const uint32_t *MIParser::getRegMask(StringRef Identifier) {
964   initNames2RegMasks();
965   auto RegMaskInfo = Names2RegMasks.find(Identifier);
966   if (RegMaskInfo == Names2RegMasks.end())
967     return nullptr;
968   return RegMaskInfo->getValue();
969 }
970 
971 void MIParser::initNames2SubRegIndices() {
972   if (!Names2SubRegIndices.empty())
973     return;
974   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
975   for (unsigned I = 1, E = TRI->getNumSubRegIndices(); I < E; ++I)
976     Names2SubRegIndices.insert(
977         std::make_pair(StringRef(TRI->getSubRegIndexName(I)).lower(), I));
978 }
979 
980 unsigned MIParser::getSubRegIndex(StringRef Name) {
981   initNames2SubRegIndices();
982   auto SubRegInfo = Names2SubRegIndices.find(Name);
983   if (SubRegInfo == Names2SubRegIndices.end())
984     return 0;
985   return SubRegInfo->getValue();
986 }
987 
988 void MIParser::initSlots2BasicBlocks() {
989   if (!Slots2BasicBlocks.empty())
990     return;
991   const auto &F = *MF.getFunction();
992   ModuleSlotTracker MST(F.getParent());
993   MST.incorporateFunction(F);
994   for (auto &BB : F) {
995     if (BB.hasName())
996       continue;
997     int Slot = MST.getLocalSlot(&BB);
998     if (Slot == -1)
999       continue;
1000     Slots2BasicBlocks.insert(std::make_pair(unsigned(Slot), &BB));
1001   }
1002 }
1003 
1004 const BasicBlock *MIParser::getIRBlock(unsigned Slot) {
1005   initSlots2BasicBlocks();
1006   auto BlockInfo = Slots2BasicBlocks.find(Slot);
1007   if (BlockInfo == Slots2BasicBlocks.end())
1008     return nullptr;
1009   return BlockInfo->second;
1010 }
1011 
1012 void MIParser::initNames2TargetIndices() {
1013   if (!Names2TargetIndices.empty())
1014     return;
1015   const auto *TII = MF.getSubtarget().getInstrInfo();
1016   assert(TII && "Expected target instruction info");
1017   auto Indices = TII->getSerializableTargetIndices();
1018   for (const auto &I : Indices)
1019     Names2TargetIndices.insert(std::make_pair(StringRef(I.second), I.first));
1020 }
1021 
1022 bool MIParser::getTargetIndex(StringRef Name, int &Index) {
1023   initNames2TargetIndices();
1024   auto IndexInfo = Names2TargetIndices.find(Name);
1025   if (IndexInfo == Names2TargetIndices.end())
1026     return true;
1027   Index = IndexInfo->second;
1028   return false;
1029 }
1030 
1031 bool llvm::parseMachineInstr(MachineInstr *&MI, SourceMgr &SM,
1032                              MachineFunction &MF, StringRef Src,
1033                              const PerFunctionMIParsingState &PFS,
1034                              const SlotMapping &IRSlots, SMDiagnostic &Error) {
1035   return MIParser(SM, MF, Error, Src, PFS, IRSlots).parse(MI);
1036 }
1037 
1038 bool llvm::parseMBBReference(MachineBasicBlock *&MBB, SourceMgr &SM,
1039                              MachineFunction &MF, StringRef Src,
1040                              const PerFunctionMIParsingState &PFS,
1041                              const SlotMapping &IRSlots, SMDiagnostic &Error) {
1042   return MIParser(SM, MF, Error, Src, PFS, IRSlots).parseStandaloneMBB(MBB);
1043 }
1044 
1045 bool llvm::parseNamedRegisterReference(unsigned &Reg, SourceMgr &SM,
1046                                        MachineFunction &MF, StringRef Src,
1047                                        const PerFunctionMIParsingState &PFS,
1048                                        const SlotMapping &IRSlots,
1049                                        SMDiagnostic &Error) {
1050   return MIParser(SM, MF, Error, Src, PFS, IRSlots)
1051       .parseStandaloneNamedRegister(Reg);
1052 }
1053 
1054 bool llvm::parseVirtualRegisterReference(unsigned &Reg, SourceMgr &SM,
1055                                          MachineFunction &MF, StringRef Src,
1056                                          const PerFunctionMIParsingState &PFS,
1057                                          const SlotMapping &IRSlots,
1058                                          SMDiagnostic &Error) {
1059   return MIParser(SM, MF, Error, Src, PFS, IRSlots)
1060       .parseStandaloneVirtualRegister(Reg);
1061 }
1062 
1063 bool llvm::parseIRBlockReference(const BasicBlock *&BB, SourceMgr &SM,
1064                                  MachineFunction &MF, StringRef Src,
1065                                  const PerFunctionMIParsingState &PFS,
1066                                  const SlotMapping &IRSlots,
1067                                  SMDiagnostic &Error) {
1068   return MIParser(SM, MF, Error, Src, PFS, IRSlots)
1069       .parseStandaloneIRBlockReference(BB);
1070 }
1071