1 //===- MIRParser.cpp - MIR serialization format 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 class that parses the optional LLVM IR and machine
11 // functions that are stored in MIR files.
12 //
13 //===----------------------------------------------------------------------===//
14 
15 #include "llvm/CodeGen/MIRParser/MIRParser.h"
16 #include "MIParser.h"
17 #include "llvm/ADT/DenseMap.h"
18 #include "llvm/ADT/StringRef.h"
19 #include "llvm/ADT/StringMap.h"
20 #include "llvm/ADT/STLExtras.h"
21 #include "llvm/AsmParser/Parser.h"
22 #include "llvm/AsmParser/SlotMapping.h"
23 #include "llvm/CodeGen/MachineFunction.h"
24 #include "llvm/CodeGen/MachineFrameInfo.h"
25 #include "llvm/CodeGen/MachineRegisterInfo.h"
26 #include "llvm/CodeGen/MIRYamlMapping.h"
27 #include "llvm/IR/BasicBlock.h"
28 #include "llvm/IR/DiagnosticInfo.h"
29 #include "llvm/IR/Instructions.h"
30 #include "llvm/IR/LLVMContext.h"
31 #include "llvm/IR/Module.h"
32 #include "llvm/IR/ValueSymbolTable.h"
33 #include "llvm/Support/LineIterator.h"
34 #include "llvm/Support/SMLoc.h"
35 #include "llvm/Support/SourceMgr.h"
36 #include "llvm/Support/MemoryBuffer.h"
37 #include "llvm/Support/YAMLTraits.h"
38 #include <memory>
39 
40 using namespace llvm;
41 
42 namespace llvm {
43 
44 /// This class implements the parsing of LLVM IR that's embedded inside a MIR
45 /// file.
46 class MIRParserImpl {
47   SourceMgr SM;
48   StringRef Filename;
49   LLVMContext &Context;
50   StringMap<std::unique_ptr<yaml::MachineFunction>> Functions;
51   SlotMapping IRSlots;
52   /// Maps from register class names to register classes.
53   StringMap<const TargetRegisterClass *> Names2RegClasses;
54 
55 public:
56   MIRParserImpl(std::unique_ptr<MemoryBuffer> Contents, StringRef Filename,
57                 LLVMContext &Context);
58 
59   void reportDiagnostic(const SMDiagnostic &Diag);
60 
61   /// Report an error with the given message at unknown location.
62   ///
63   /// Always returns true.
64   bool error(const Twine &Message);
65 
66   /// Report an error with the given message at the given location.
67   ///
68   /// Always returns true.
69   bool error(SMLoc Loc, const Twine &Message);
70 
71   /// Report a given error with the location translated from the location in an
72   /// embedded string literal to a location in the MIR file.
73   ///
74   /// Always returns true.
75   bool error(const SMDiagnostic &Error, SMRange SourceRange);
76 
77   /// Try to parse the optional LLVM module and the machine functions in the MIR
78   /// file.
79   ///
80   /// Return null if an error occurred.
81   std::unique_ptr<Module> parse();
82 
83   /// Parse the machine function in the current YAML document.
84   ///
85   /// \param NoLLVMIR - set to true when the MIR file doesn't have LLVM IR.
86   /// A dummy IR function is created and inserted into the given module when
87   /// this parameter is true.
88   ///
89   /// Return true if an error occurred.
90   bool parseMachineFunction(yaml::Input &In, Module &M, bool NoLLVMIR);
91 
92   /// Initialize the machine function to the state that's described in the MIR
93   /// file.
94   ///
95   /// Return true if error occurred.
96   bool initializeMachineFunction(MachineFunction &MF);
97 
98   /// Initialize the machine basic block using it's YAML representation.
99   ///
100   /// Return true if an error occurred.
101   bool initializeMachineBasicBlock(MachineFunction &MF, MachineBasicBlock &MBB,
102                                    const yaml::MachineBasicBlock &YamlMBB,
103                                    const PerFunctionMIParsingState &PFS);
104 
105   bool initializeRegisterInfo(const MachineFunction &MF,
106                               MachineRegisterInfo &RegInfo,
107                               const yaml::MachineFunction &YamlMF);
108 
109   bool initializeFrameInfo(MachineFrameInfo &MFI,
110                            const yaml::MachineFrameInfo &YamlMFI);
111 
112 private:
113   /// Return a MIR diagnostic converted from an MI string diagnostic.
114   SMDiagnostic diagFromMIStringDiag(const SMDiagnostic &Error,
115                                     SMRange SourceRange);
116 
117   /// Return a MIR diagnostic converted from an LLVM assembly diagnostic.
118   SMDiagnostic diagFromLLVMAssemblyDiag(const SMDiagnostic &Error,
119                                         SMRange SourceRange);
120 
121   /// Create an empty function with the given name.
122   void createDummyFunction(StringRef Name, Module &M);
123 
124   void initNames2RegClasses(const MachineFunction &MF);
125 
126   /// Check if the given identifier is a name of a register class.
127   ///
128   /// Return null if the name isn't a register class.
129   const TargetRegisterClass *getRegClass(const MachineFunction &MF,
130                                          StringRef Name);
131 };
132 
133 } // end namespace llvm
134 
135 MIRParserImpl::MIRParserImpl(std::unique_ptr<MemoryBuffer> Contents,
136                              StringRef Filename, LLVMContext &Context)
137     : SM(), Filename(Filename), Context(Context) {
138   SM.AddNewSourceBuffer(std::move(Contents), SMLoc());
139 }
140 
141 bool MIRParserImpl::error(const Twine &Message) {
142   Context.diagnose(DiagnosticInfoMIRParser(
143       DS_Error, SMDiagnostic(Filename, SourceMgr::DK_Error, Message.str())));
144   return true;
145 }
146 
147 bool MIRParserImpl::error(SMLoc Loc, const Twine &Message) {
148   Context.diagnose(DiagnosticInfoMIRParser(
149       DS_Error, SM.GetMessage(Loc, SourceMgr::DK_Error, Message)));
150   return true;
151 }
152 
153 bool MIRParserImpl::error(const SMDiagnostic &Error, SMRange SourceRange) {
154   assert(Error.getKind() == SourceMgr::DK_Error && "Expected an error");
155   reportDiagnostic(diagFromMIStringDiag(Error, SourceRange));
156   return true;
157 }
158 
159 void MIRParserImpl::reportDiagnostic(const SMDiagnostic &Diag) {
160   DiagnosticSeverity Kind;
161   switch (Diag.getKind()) {
162   case SourceMgr::DK_Error:
163     Kind = DS_Error;
164     break;
165   case SourceMgr::DK_Warning:
166     Kind = DS_Warning;
167     break;
168   case SourceMgr::DK_Note:
169     Kind = DS_Note;
170     break;
171   }
172   Context.diagnose(DiagnosticInfoMIRParser(Kind, Diag));
173 }
174 
175 static void handleYAMLDiag(const SMDiagnostic &Diag, void *Context) {
176   reinterpret_cast<MIRParserImpl *>(Context)->reportDiagnostic(Diag);
177 }
178 
179 std::unique_ptr<Module> MIRParserImpl::parse() {
180   yaml::Input In(SM.getMemoryBuffer(SM.getMainFileID())->getBuffer(),
181                  /*Ctxt=*/nullptr, handleYAMLDiag, this);
182   In.setContext(&In);
183 
184   if (!In.setCurrentDocument()) {
185     if (In.error())
186       return nullptr;
187     // Create an empty module when the MIR file is empty.
188     return llvm::make_unique<Module>(Filename, Context);
189   }
190 
191   std::unique_ptr<Module> M;
192   bool NoLLVMIR = false;
193   // Parse the block scalar manually so that we can return unique pointer
194   // without having to go trough YAML traits.
195   if (const auto *BSN =
196           dyn_cast_or_null<yaml::BlockScalarNode>(In.getCurrentNode())) {
197     SMDiagnostic Error;
198     M = parseAssembly(MemoryBufferRef(BSN->getValue(), Filename), Error,
199                       Context, &IRSlots);
200     if (!M) {
201       reportDiagnostic(diagFromLLVMAssemblyDiag(Error, BSN->getSourceRange()));
202       return M;
203     }
204     In.nextDocument();
205     if (!In.setCurrentDocument())
206       return M;
207   } else {
208     // Create an new, empty module.
209     M = llvm::make_unique<Module>(Filename, Context);
210     NoLLVMIR = true;
211   }
212 
213   // Parse the machine functions.
214   do {
215     if (parseMachineFunction(In, *M, NoLLVMIR))
216       return nullptr;
217     In.nextDocument();
218   } while (In.setCurrentDocument());
219 
220   return M;
221 }
222 
223 bool MIRParserImpl::parseMachineFunction(yaml::Input &In, Module &M,
224                                          bool NoLLVMIR) {
225   auto MF = llvm::make_unique<yaml::MachineFunction>();
226   yaml::yamlize(In, *MF, false);
227   if (In.error())
228     return true;
229   auto FunctionName = MF->Name;
230   if (Functions.find(FunctionName) != Functions.end())
231     return error(Twine("redefinition of machine function '") + FunctionName +
232                  "'");
233   Functions.insert(std::make_pair(FunctionName, std::move(MF)));
234   if (NoLLVMIR)
235     createDummyFunction(FunctionName, M);
236   else if (!M.getFunction(FunctionName))
237     return error(Twine("function '") + FunctionName +
238                  "' isn't defined in the provided LLVM IR");
239   return false;
240 }
241 
242 void MIRParserImpl::createDummyFunction(StringRef Name, Module &M) {
243   auto &Context = M.getContext();
244   Function *F = cast<Function>(M.getOrInsertFunction(
245       Name, FunctionType::get(Type::getVoidTy(Context), false)));
246   BasicBlock *BB = BasicBlock::Create(Context, "entry", F);
247   new UnreachableInst(Context, BB);
248 }
249 
250 bool MIRParserImpl::initializeMachineFunction(MachineFunction &MF) {
251   auto It = Functions.find(MF.getName());
252   if (It == Functions.end())
253     return error(Twine("no machine function information for function '") +
254                  MF.getName() + "' in the MIR file");
255   // TODO: Recreate the machine function.
256   const yaml::MachineFunction &YamlMF = *It->getValue();
257   if (YamlMF.Alignment)
258     MF.setAlignment(YamlMF.Alignment);
259   MF.setExposesReturnsTwice(YamlMF.ExposesReturnsTwice);
260   MF.setHasInlineAsm(YamlMF.HasInlineAsm);
261   if (initializeRegisterInfo(MF, MF.getRegInfo(), YamlMF))
262     return true;
263   if (initializeFrameInfo(*MF.getFrameInfo(), YamlMF.FrameInfo))
264     return true;
265 
266   PerFunctionMIParsingState PFS;
267   const auto &F = *MF.getFunction();
268   for (const auto &YamlMBB : YamlMF.BasicBlocks) {
269     const BasicBlock *BB = nullptr;
270     const yaml::StringValue &Name = YamlMBB.Name;
271     if (!Name.Value.empty()) {
272       BB = dyn_cast_or_null<BasicBlock>(
273           F.getValueSymbolTable().lookup(Name.Value));
274       if (!BB)
275         return error(Name.SourceRange.Start,
276                      Twine("basic block '") + Name.Value +
277                          "' is not defined in the function '" + MF.getName() +
278                          "'");
279     }
280     auto *MBB = MF.CreateMachineBasicBlock(BB);
281     MF.insert(MF.end(), MBB);
282     bool WasInserted =
283         PFS.MBBSlots.insert(std::make_pair(YamlMBB.ID, MBB)).second;
284     if (!WasInserted)
285       return error(Twine("redefinition of machine basic block with id #") +
286                    Twine(YamlMBB.ID));
287   }
288 
289   if (YamlMF.BasicBlocks.empty())
290     return error(Twine("machine function '") + Twine(MF.getName()) +
291                  "' requires at least one machine basic block in its body");
292   // Initialize the machine basic blocks after creating them all so that the
293   // machine instructions parser can resolve the MBB references.
294   unsigned I = 0;
295   for (const auto &YamlMBB : YamlMF.BasicBlocks) {
296     if (initializeMachineBasicBlock(MF, *MF.getBlockNumbered(I++), YamlMBB,
297                                     PFS))
298       return true;
299   }
300   return false;
301 }
302 
303 bool MIRParserImpl::initializeMachineBasicBlock(
304     MachineFunction &MF, MachineBasicBlock &MBB,
305     const yaml::MachineBasicBlock &YamlMBB,
306     const PerFunctionMIParsingState &PFS) {
307   MBB.setAlignment(YamlMBB.Alignment);
308   if (YamlMBB.AddressTaken)
309     MBB.setHasAddressTaken();
310   MBB.setIsLandingPad(YamlMBB.IsLandingPad);
311   SMDiagnostic Error;
312   // Parse the successors.
313   for (const auto &MBBSource : YamlMBB.Successors) {
314     MachineBasicBlock *SuccMBB = nullptr;
315     if (parseMBBReference(SuccMBB, SM, MF, MBBSource.Value, PFS, IRSlots,
316                           Error))
317       return error(Error, MBBSource.SourceRange);
318     // TODO: Report an error when adding the same successor more than once.
319     MBB.addSuccessor(SuccMBB);
320   }
321   // Parse the instructions.
322   for (const auto &MISource : YamlMBB.Instructions) {
323     MachineInstr *MI = nullptr;
324     if (parseMachineInstr(MI, SM, MF, MISource.Value, PFS, IRSlots, Error))
325       return error(Error, MISource.SourceRange);
326     MBB.insert(MBB.end(), MI);
327   }
328   return false;
329 }
330 
331 bool MIRParserImpl::initializeRegisterInfo(
332     const MachineFunction &MF, MachineRegisterInfo &RegInfo,
333     const yaml::MachineFunction &YamlMF) {
334   assert(RegInfo.isSSA());
335   if (!YamlMF.IsSSA)
336     RegInfo.leaveSSA();
337   assert(RegInfo.tracksLiveness());
338   if (!YamlMF.TracksRegLiveness)
339     RegInfo.invalidateLiveness();
340   RegInfo.enableSubRegLiveness(YamlMF.TracksSubRegLiveness);
341 
342   // Parse the virtual register information.
343   for (const auto &VReg : YamlMF.VirtualRegisters) {
344     const auto *RC = getRegClass(MF, VReg.Class.Value);
345     if (!RC)
346       return error(VReg.Class.SourceRange.Start,
347                    Twine("use of undefined register class '") +
348                        VReg.Class.Value + "'");
349     // TODO: create the mapping from IDs to registers so that the virtual
350     // register references can be parsed correctly.
351     RegInfo.createVirtualRegister(RC);
352   }
353   return false;
354 }
355 
356 bool MIRParserImpl::initializeFrameInfo(MachineFrameInfo &MFI,
357                                         const yaml::MachineFrameInfo &YamlMFI) {
358   MFI.setFrameAddressIsTaken(YamlMFI.IsFrameAddressTaken);
359   MFI.setReturnAddressIsTaken(YamlMFI.IsReturnAddressTaken);
360   MFI.setHasStackMap(YamlMFI.HasStackMap);
361   MFI.setHasPatchPoint(YamlMFI.HasPatchPoint);
362   MFI.setStackSize(YamlMFI.StackSize);
363   MFI.setOffsetAdjustment(YamlMFI.OffsetAdjustment);
364   if (YamlMFI.MaxAlignment)
365     MFI.ensureMaxAlignment(YamlMFI.MaxAlignment);
366   MFI.setAdjustsStack(YamlMFI.AdjustsStack);
367   MFI.setHasCalls(YamlMFI.HasCalls);
368   MFI.setMaxCallFrameSize(YamlMFI.MaxCallFrameSize);
369   MFI.setHasOpaqueSPAdjustment(YamlMFI.HasOpaqueSPAdjustment);
370   MFI.setHasVAStart(YamlMFI.HasVAStart);
371   MFI.setHasMustTailInVarArgFunc(YamlMFI.HasMustTailInVarArgFunc);
372   return false;
373 }
374 
375 SMDiagnostic MIRParserImpl::diagFromMIStringDiag(const SMDiagnostic &Error,
376                                                  SMRange SourceRange) {
377   assert(SourceRange.isValid() && "Invalid source range");
378   SMLoc Loc = SourceRange.Start;
379   bool HasQuote = Loc.getPointer() < SourceRange.End.getPointer() &&
380                   *Loc.getPointer() == '\'';
381   // Translate the location of the error from the location in the MI string to
382   // the corresponding location in the MIR file.
383   Loc = Loc.getFromPointer(Loc.getPointer() + Error.getColumnNo() +
384                            (HasQuote ? 1 : 0));
385 
386   // TODO: Translate any source ranges as well.
387   return SM.GetMessage(Loc, Error.getKind(), Error.getMessage(), None,
388                        Error.getFixIts());
389 }
390 
391 SMDiagnostic MIRParserImpl::diagFromLLVMAssemblyDiag(const SMDiagnostic &Error,
392                                                      SMRange SourceRange) {
393   assert(SourceRange.isValid());
394 
395   // Translate the location of the error from the location in the llvm IR string
396   // to the corresponding location in the MIR file.
397   auto LineAndColumn = SM.getLineAndColumn(SourceRange.Start);
398   unsigned Line = LineAndColumn.first + Error.getLineNo() - 1;
399   unsigned Column = Error.getColumnNo();
400   StringRef LineStr = Error.getLineContents();
401   SMLoc Loc = Error.getLoc();
402 
403   // Get the full line and adjust the column number by taking the indentation of
404   // LLVM IR into account.
405   for (line_iterator L(*SM.getMemoryBuffer(SM.getMainFileID()), false), E;
406        L != E; ++L) {
407     if (L.line_number() == Line) {
408       LineStr = *L;
409       Loc = SMLoc::getFromPointer(LineStr.data());
410       auto Indent = LineStr.find(Error.getLineContents());
411       if (Indent != StringRef::npos)
412         Column += Indent;
413       break;
414     }
415   }
416 
417   return SMDiagnostic(SM, Loc, Filename, Line, Column, Error.getKind(),
418                       Error.getMessage(), LineStr, Error.getRanges(),
419                       Error.getFixIts());
420 }
421 
422 void MIRParserImpl::initNames2RegClasses(const MachineFunction &MF) {
423   if (!Names2RegClasses.empty())
424     return;
425   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
426   for (unsigned I = 0, E = TRI->getNumRegClasses(); I < E; ++I) {
427     const auto *RC = TRI->getRegClass(I);
428     Names2RegClasses.insert(
429         std::make_pair(StringRef(TRI->getRegClassName(RC)).lower(), RC));
430   }
431 }
432 
433 const TargetRegisterClass *MIRParserImpl::getRegClass(const MachineFunction &MF,
434                                                       StringRef Name) {
435   initNames2RegClasses(MF);
436   auto RegClassInfo = Names2RegClasses.find(Name);
437   if (RegClassInfo == Names2RegClasses.end())
438     return nullptr;
439   return RegClassInfo->getValue();
440 }
441 
442 MIRParser::MIRParser(std::unique_ptr<MIRParserImpl> Impl)
443     : Impl(std::move(Impl)) {}
444 
445 MIRParser::~MIRParser() {}
446 
447 std::unique_ptr<Module> MIRParser::parseLLVMModule() { return Impl->parse(); }
448 
449 bool MIRParser::initializeMachineFunction(MachineFunction &MF) {
450   return Impl->initializeMachineFunction(MF);
451 }
452 
453 std::unique_ptr<MIRParser> llvm::createMIRParserFromFile(StringRef Filename,
454                                                          SMDiagnostic &Error,
455                                                          LLVMContext &Context) {
456   auto FileOrErr = MemoryBuffer::getFile(Filename);
457   if (std::error_code EC = FileOrErr.getError()) {
458     Error = SMDiagnostic(Filename, SourceMgr::DK_Error,
459                          "Could not open input file: " + EC.message());
460     return nullptr;
461   }
462   return createMIRParser(std::move(FileOrErr.get()), Context);
463 }
464 
465 std::unique_ptr<MIRParser>
466 llvm::createMIRParser(std::unique_ptr<MemoryBuffer> Contents,
467                       LLVMContext &Context) {
468   auto Filename = Contents->getBufferIdentifier();
469   return llvm::make_unique<MIRParser>(
470       llvm::make_unique<MIRParserImpl>(std::move(Contents), Filename, Context));
471 }
472