1 //===-- ExternalFunctions.cpp - Implement External Functions --------------===//
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 contains both code to deal with invoking "external" functions, but
11 //  also contains code that implements "exported" external functions.
12 //
13 //  There are currently two mechanisms for handling external functions in the
14 //  Interpreter.  The first is to implement lle_* wrapper functions that are
15 //  specific to well-known library functions which manually translate the
16 //  arguments from GenericValues and make the call.  If such a wrapper does
17 //  not exist, and libffi is available, then the Interpreter will attempt to
18 //  invoke the function using libffi, after finding its address.
19 //
20 //===----------------------------------------------------------------------===//
21 
22 #include "Interpreter.h"
23 #include "llvm/ADT/APInt.h"
24 #include "llvm/ADT/ArrayRef.h"
25 #include "llvm/Config/config.h" // Detect libffi
26 #include "llvm/ExecutionEngine/GenericValue.h"
27 #include "llvm/IR/DataLayout.h"
28 #include "llvm/IR/DerivedTypes.h"
29 #include "llvm/IR/Function.h"
30 #include "llvm/IR/Type.h"
31 #include "llvm/Support/Casting.h"
32 #include "llvm/Support/DynamicLibrary.h"
33 #include "llvm/Support/ErrorHandling.h"
34 #include "llvm/Support/ManagedStatic.h"
35 #include "llvm/Support/Mutex.h"
36 #include "llvm/Support/UniqueLock.h"
37 #include "llvm/Support/raw_ostream.h"
38 #include <cassert>
39 #include <cmath>
40 #include <csignal>
41 #include <cstdint>
42 #include <cstdio>
43 #include <cstring>
44 #include <map>
45 #include <string>
46 #include <utility>
47 #include <vector>
48 
49 #ifdef HAVE_FFI_CALL
50 #ifdef HAVE_FFI_H
51 #include <ffi.h>
52 #define USE_LIBFFI
53 #elif HAVE_FFI_FFI_H
54 #include <ffi/ffi.h>
55 #define USE_LIBFFI
56 #endif
57 #endif
58 
59 using namespace llvm;
60 
61 static ManagedStatic<sys::Mutex> FunctionsLock;
62 
63 typedef GenericValue (*ExFunc)(FunctionType *, ArrayRef<GenericValue>);
64 static ManagedStatic<std::map<const Function *, ExFunc> > ExportedFunctions;
65 static ManagedStatic<std::map<std::string, ExFunc> > FuncNames;
66 
67 #ifdef USE_LIBFFI
68 typedef void (*RawFunc)();
69 static ManagedStatic<std::map<const Function *, RawFunc> > RawFunctions;
70 #endif
71 
72 static Interpreter *TheInterpreter;
73 
getTypeID(Type * Ty)74 static char getTypeID(Type *Ty) {
75   switch (Ty->getTypeID()) {
76   case Type::VoidTyID:    return 'V';
77   case Type::IntegerTyID:
78     switch (cast<IntegerType>(Ty)->getBitWidth()) {
79       case 1:  return 'o';
80       case 8:  return 'B';
81       case 16: return 'S';
82       case 32: return 'I';
83       case 64: return 'L';
84       default: return 'N';
85     }
86   case Type::FloatTyID:   return 'F';
87   case Type::DoubleTyID:  return 'D';
88   case Type::PointerTyID: return 'P';
89   case Type::FunctionTyID:return 'M';
90   case Type::StructTyID:  return 'T';
91   case Type::ArrayTyID:   return 'A';
92   default: return 'U';
93   }
94 }
95 
96 // Try to find address of external function given a Function object.
97 // Please note, that interpreter doesn't know how to assemble a
98 // real call in general case (this is JIT job), that's why it assumes,
99 // that all external functions has the same (and pretty "general") signature.
100 // The typical example of such functions are "lle_X_" ones.
lookupFunction(const Function * F)101 static ExFunc lookupFunction(const Function *F) {
102   // Function not found, look it up... start by figuring out what the
103   // composite function name should be.
104   std::string ExtName = "lle_";
105   FunctionType *FT = F->getFunctionType();
106   ExtName += getTypeID(FT->getReturnType());
107   for (Type *T : FT->params())
108     ExtName += getTypeID(T);
109   ExtName += ("_" + F->getName()).str();
110 
111   sys::ScopedLock Writer(*FunctionsLock);
112   ExFunc FnPtr = (*FuncNames)[ExtName];
113   if (!FnPtr)
114     FnPtr = (*FuncNames)[("lle_X_" + F->getName()).str()];
115   if (!FnPtr)  // Try calling a generic function... if it exists...
116     FnPtr = (ExFunc)(intptr_t)sys::DynamicLibrary::SearchForAddressOfSymbol(
117         ("lle_X_" + F->getName()).str());
118   if (FnPtr)
119     ExportedFunctions->insert(std::make_pair(F, FnPtr));  // Cache for later
120   return FnPtr;
121 }
122 
123 #ifdef USE_LIBFFI
ffiTypeFor(Type * Ty)124 static ffi_type *ffiTypeFor(Type *Ty) {
125   switch (Ty->getTypeID()) {
126     case Type::VoidTyID: return &ffi_type_void;
127     case Type::IntegerTyID:
128       switch (cast<IntegerType>(Ty)->getBitWidth()) {
129         case 8:  return &ffi_type_sint8;
130         case 16: return &ffi_type_sint16;
131         case 32: return &ffi_type_sint32;
132         case 64: return &ffi_type_sint64;
133       }
134     case Type::FloatTyID:   return &ffi_type_float;
135     case Type::DoubleTyID:  return &ffi_type_double;
136     case Type::PointerTyID: return &ffi_type_pointer;
137     default: break;
138   }
139   // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
140   report_fatal_error("Type could not be mapped for use with libffi.");
141   return NULL;
142 }
143 
ffiValueFor(Type * Ty,const GenericValue & AV,void * ArgDataPtr)144 static void *ffiValueFor(Type *Ty, const GenericValue &AV,
145                          void *ArgDataPtr) {
146   switch (Ty->getTypeID()) {
147     case Type::IntegerTyID:
148       switch (cast<IntegerType>(Ty)->getBitWidth()) {
149         case 8: {
150           int8_t *I8Ptr = (int8_t *) ArgDataPtr;
151           *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
152           return ArgDataPtr;
153         }
154         case 16: {
155           int16_t *I16Ptr = (int16_t *) ArgDataPtr;
156           *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
157           return ArgDataPtr;
158         }
159         case 32: {
160           int32_t *I32Ptr = (int32_t *) ArgDataPtr;
161           *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
162           return ArgDataPtr;
163         }
164         case 64: {
165           int64_t *I64Ptr = (int64_t *) ArgDataPtr;
166           *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
167           return ArgDataPtr;
168         }
169       }
170     case Type::FloatTyID: {
171       float *FloatPtr = (float *) ArgDataPtr;
172       *FloatPtr = AV.FloatVal;
173       return ArgDataPtr;
174     }
175     case Type::DoubleTyID: {
176       double *DoublePtr = (double *) ArgDataPtr;
177       *DoublePtr = AV.DoubleVal;
178       return ArgDataPtr;
179     }
180     case Type::PointerTyID: {
181       void **PtrPtr = (void **) ArgDataPtr;
182       *PtrPtr = GVTOP(AV);
183       return ArgDataPtr;
184     }
185     default: break;
186   }
187   // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
188   report_fatal_error("Type value could not be mapped for use with libffi.");
189   return NULL;
190 }
191 
ffiInvoke(RawFunc Fn,Function * F,ArrayRef<GenericValue> ArgVals,const DataLayout & TD,GenericValue & Result)192 static bool ffiInvoke(RawFunc Fn, Function *F, ArrayRef<GenericValue> ArgVals,
193                       const DataLayout &TD, GenericValue &Result) {
194   ffi_cif cif;
195   FunctionType *FTy = F->getFunctionType();
196   const unsigned NumArgs = F->arg_size();
197 
198   // TODO: We don't have type information about the remaining arguments, because
199   // this information is never passed into ExecutionEngine::runFunction().
200   if (ArgVals.size() > NumArgs && F->isVarArg()) {
201     report_fatal_error("Calling external var arg function '" + F->getName()
202                       + "' is not supported by the Interpreter.");
203   }
204 
205   unsigned ArgBytes = 0;
206 
207   std::vector<ffi_type*> args(NumArgs);
208   for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
209        A != E; ++A) {
210     const unsigned ArgNo = A->getArgNo();
211     Type *ArgTy = FTy->getParamType(ArgNo);
212     args[ArgNo] = ffiTypeFor(ArgTy);
213     ArgBytes += TD.getTypeStoreSize(ArgTy);
214   }
215 
216   SmallVector<uint8_t, 128> ArgData;
217   ArgData.resize(ArgBytes);
218   uint8_t *ArgDataPtr = ArgData.data();
219   SmallVector<void*, 16> values(NumArgs);
220   for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
221        A != E; ++A) {
222     const unsigned ArgNo = A->getArgNo();
223     Type *ArgTy = FTy->getParamType(ArgNo);
224     values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
225     ArgDataPtr += TD.getTypeStoreSize(ArgTy);
226   }
227 
228   Type *RetTy = FTy->getReturnType();
229   ffi_type *rtype = ffiTypeFor(RetTy);
230 
231   if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, args.data()) ==
232       FFI_OK) {
233     SmallVector<uint8_t, 128> ret;
234     if (RetTy->getTypeID() != Type::VoidTyID)
235       ret.resize(TD.getTypeStoreSize(RetTy));
236     ffi_call(&cif, Fn, ret.data(), values.data());
237     switch (RetTy->getTypeID()) {
238       case Type::IntegerTyID:
239         switch (cast<IntegerType>(RetTy)->getBitWidth()) {
240           case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret.data()); break;
241           case 16: Result.IntVal = APInt(16, *(int16_t*) ret.data()); break;
242           case 32: Result.IntVal = APInt(32, *(int32_t*) ret.data()); break;
243           case 64: Result.IntVal = APInt(64, *(int64_t*) ret.data()); break;
244         }
245         break;
246       case Type::FloatTyID:   Result.FloatVal   = *(float *) ret.data(); break;
247       case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret.data(); break;
248       case Type::PointerTyID: Result.PointerVal = *(void **) ret.data(); break;
249       default: break;
250     }
251     return true;
252   }
253 
254   return false;
255 }
256 #endif // USE_LIBFFI
257 
callExternalFunction(Function * F,ArrayRef<GenericValue> ArgVals)258 GenericValue Interpreter::callExternalFunction(Function *F,
259                                                ArrayRef<GenericValue> ArgVals) {
260   TheInterpreter = this;
261 
262   unique_lock<sys::Mutex> Guard(*FunctionsLock);
263 
264   // Do a lookup to see if the function is in our cache... this should just be a
265   // deferred annotation!
266   std::map<const Function *, ExFunc>::iterator FI = ExportedFunctions->find(F);
267   if (ExFunc Fn = (FI == ExportedFunctions->end()) ? lookupFunction(F)
268                                                    : FI->second) {
269     Guard.unlock();
270     return Fn(F->getFunctionType(), ArgVals);
271   }
272 
273 #ifdef USE_LIBFFI
274   std::map<const Function *, RawFunc>::iterator RF = RawFunctions->find(F);
275   RawFunc RawFn;
276   if (RF == RawFunctions->end()) {
277     RawFn = (RawFunc)(intptr_t)
278       sys::DynamicLibrary::SearchForAddressOfSymbol(F->getName());
279     if (!RawFn)
280       RawFn = (RawFunc)(intptr_t)getPointerToGlobalIfAvailable(F);
281     if (RawFn != 0)
282       RawFunctions->insert(std::make_pair(F, RawFn));  // Cache for later
283   } else {
284     RawFn = RF->second;
285   }
286 
287   Guard.unlock();
288 
289   GenericValue Result;
290   if (RawFn != 0 && ffiInvoke(RawFn, F, ArgVals, getDataLayout(), Result))
291     return Result;
292 #endif // USE_LIBFFI
293 
294   if (F->getName() == "__main")
295     errs() << "Tried to execute an unknown external function: "
296       << *F->getType() << " __main\n";
297   else
298     report_fatal_error("Tried to execute an unknown external function: " +
299                        F->getName());
300 #ifndef USE_LIBFFI
301   errs() << "Recompiling LLVM with --enable-libffi might help.\n";
302 #endif
303   return GenericValue();
304 }
305 
306 //===----------------------------------------------------------------------===//
307 //  Functions "exported" to the running application...
308 //
309 
310 // void atexit(Function*)
lle_X_atexit(FunctionType * FT,ArrayRef<GenericValue> Args)311 static GenericValue lle_X_atexit(FunctionType *FT,
312                                  ArrayRef<GenericValue> Args) {
313   assert(Args.size() == 1);
314   TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
315   GenericValue GV;
316   GV.IntVal = 0;
317   return GV;
318 }
319 
320 // void exit(int)
lle_X_exit(FunctionType * FT,ArrayRef<GenericValue> Args)321 static GenericValue lle_X_exit(FunctionType *FT, ArrayRef<GenericValue> Args) {
322   TheInterpreter->exitCalled(Args[0]);
323   return GenericValue();
324 }
325 
326 // void abort(void)
lle_X_abort(FunctionType * FT,ArrayRef<GenericValue> Args)327 static GenericValue lle_X_abort(FunctionType *FT, ArrayRef<GenericValue> Args) {
328   //FIXME: should we report or raise here?
329   //report_fatal_error("Interpreted program raised SIGABRT");
330   raise (SIGABRT);
331   return GenericValue();
332 }
333 
334 // int sprintf(char *, const char *, ...) - a very rough implementation to make
335 // output useful.
lle_X_sprintf(FunctionType * FT,ArrayRef<GenericValue> Args)336 static GenericValue lle_X_sprintf(FunctionType *FT,
337                                   ArrayRef<GenericValue> Args) {
338   char *OutputBuffer = (char *)GVTOP(Args[0]);
339   const char *FmtStr = (const char *)GVTOP(Args[1]);
340   unsigned ArgNo = 2;
341 
342   // printf should return # chars printed.  This is completely incorrect, but
343   // close enough for now.
344   GenericValue GV;
345   GV.IntVal = APInt(32, strlen(FmtStr));
346   while (true) {
347     switch (*FmtStr) {
348     case 0: return GV;             // Null terminator...
349     default:                       // Normal nonspecial character
350       sprintf(OutputBuffer++, "%c", *FmtStr++);
351       break;
352     case '\\': {                   // Handle escape codes
353       sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
354       FmtStr += 2; OutputBuffer += 2;
355       break;
356     }
357     case '%': {                    // Handle format specifiers
358       char FmtBuf[100] = "", Buffer[1000] = "";
359       char *FB = FmtBuf;
360       *FB++ = *FmtStr++;
361       char Last = *FB++ = *FmtStr++;
362       unsigned HowLong = 0;
363       while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
364              Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
365              Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
366              Last != 'p' && Last != 's' && Last != '%') {
367         if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
368         Last = *FB++ = *FmtStr++;
369       }
370       *FB = 0;
371 
372       switch (Last) {
373       case '%':
374         memcpy(Buffer, "%", 2); break;
375       case 'c':
376         sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
377         break;
378       case 'd': case 'i':
379       case 'u': case 'o':
380       case 'x': case 'X':
381         if (HowLong >= 1) {
382           if (HowLong == 1 &&
383               TheInterpreter->getDataLayout().getPointerSizeInBits() == 64 &&
384               sizeof(long) < sizeof(int64_t)) {
385             // Make sure we use %lld with a 64 bit argument because we might be
386             // compiling LLI on a 32 bit compiler.
387             unsigned Size = strlen(FmtBuf);
388             FmtBuf[Size] = FmtBuf[Size-1];
389             FmtBuf[Size+1] = 0;
390             FmtBuf[Size-1] = 'l';
391           }
392           sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
393         } else
394           sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
395         break;
396       case 'e': case 'E': case 'g': case 'G': case 'f':
397         sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
398       case 'p':
399         sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
400       case 's':
401         sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
402       default:
403         errs() << "<unknown printf code '" << *FmtStr << "'!>";
404         ArgNo++; break;
405       }
406       size_t Len = strlen(Buffer);
407       memcpy(OutputBuffer, Buffer, Len + 1);
408       OutputBuffer += Len;
409       }
410       break;
411     }
412   }
413   return GV;
414 }
415 
416 // int printf(const char *, ...) - a very rough implementation to make output
417 // useful.
lle_X_printf(FunctionType * FT,ArrayRef<GenericValue> Args)418 static GenericValue lle_X_printf(FunctionType *FT,
419                                  ArrayRef<GenericValue> Args) {
420   char Buffer[10000];
421   std::vector<GenericValue> NewArgs;
422   NewArgs.push_back(PTOGV((void*)&Buffer[0]));
423   NewArgs.insert(NewArgs.end(), Args.begin(), Args.end());
424   GenericValue GV = lle_X_sprintf(FT, NewArgs);
425   outs() << Buffer;
426   return GV;
427 }
428 
429 // int sscanf(const char *format, ...);
lle_X_sscanf(FunctionType * FT,ArrayRef<GenericValue> args)430 static GenericValue lle_X_sscanf(FunctionType *FT,
431                                  ArrayRef<GenericValue> args) {
432   assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
433 
434   char *Args[10];
435   for (unsigned i = 0; i < args.size(); ++i)
436     Args[i] = (char*)GVTOP(args[i]);
437 
438   GenericValue GV;
439   GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
440                     Args[5], Args[6], Args[7], Args[8], Args[9]));
441   return GV;
442 }
443 
444 // int scanf(const char *format, ...);
lle_X_scanf(FunctionType * FT,ArrayRef<GenericValue> args)445 static GenericValue lle_X_scanf(FunctionType *FT, ArrayRef<GenericValue> args) {
446   assert(args.size() < 10 && "Only handle up to 10 args to scanf right now!");
447 
448   char *Args[10];
449   for (unsigned i = 0; i < args.size(); ++i)
450     Args[i] = (char*)GVTOP(args[i]);
451 
452   GenericValue GV;
453   GV.IntVal = APInt(32, scanf( Args[0], Args[1], Args[2], Args[3], Args[4],
454                     Args[5], Args[6], Args[7], Args[8], Args[9]));
455   return GV;
456 }
457 
458 // int fprintf(FILE *, const char *, ...) - a very rough implementation to make
459 // output useful.
lle_X_fprintf(FunctionType * FT,ArrayRef<GenericValue> Args)460 static GenericValue lle_X_fprintf(FunctionType *FT,
461                                   ArrayRef<GenericValue> Args) {
462   assert(Args.size() >= 2);
463   char Buffer[10000];
464   std::vector<GenericValue> NewArgs;
465   NewArgs.push_back(PTOGV(Buffer));
466   NewArgs.insert(NewArgs.end(), Args.begin()+1, Args.end());
467   GenericValue GV = lle_X_sprintf(FT, NewArgs);
468 
469   fputs(Buffer, (FILE *) GVTOP(Args[0]));
470   return GV;
471 }
472 
lle_X_memset(FunctionType * FT,ArrayRef<GenericValue> Args)473 static GenericValue lle_X_memset(FunctionType *FT,
474                                  ArrayRef<GenericValue> Args) {
475   int val = (int)Args[1].IntVal.getSExtValue();
476   size_t len = (size_t)Args[2].IntVal.getZExtValue();
477   memset((void *)GVTOP(Args[0]), val, len);
478   // llvm.memset.* returns void, lle_X_* returns GenericValue,
479   // so here we return GenericValue with IntVal set to zero
480   GenericValue GV;
481   GV.IntVal = 0;
482   return GV;
483 }
484 
lle_X_memcpy(FunctionType * FT,ArrayRef<GenericValue> Args)485 static GenericValue lle_X_memcpy(FunctionType *FT,
486                                  ArrayRef<GenericValue> Args) {
487   memcpy(GVTOP(Args[0]), GVTOP(Args[1]),
488          (size_t)(Args[2].IntVal.getLimitedValue()));
489 
490   // llvm.memcpy* returns void, lle_X_* returns GenericValue,
491   // so here we return GenericValue with IntVal set to zero
492   GenericValue GV;
493   GV.IntVal = 0;
494   return GV;
495 }
496 
initializeExternalFunctions()497 void Interpreter::initializeExternalFunctions() {
498   sys::ScopedLock Writer(*FunctionsLock);
499   (*FuncNames)["lle_X_atexit"]       = lle_X_atexit;
500   (*FuncNames)["lle_X_exit"]         = lle_X_exit;
501   (*FuncNames)["lle_X_abort"]        = lle_X_abort;
502 
503   (*FuncNames)["lle_X_printf"]       = lle_X_printf;
504   (*FuncNames)["lle_X_sprintf"]      = lle_X_sprintf;
505   (*FuncNames)["lle_X_sscanf"]       = lle_X_sscanf;
506   (*FuncNames)["lle_X_scanf"]        = lle_X_scanf;
507   (*FuncNames)["lle_X_fprintf"]      = lle_X_fprintf;
508   (*FuncNames)["lle_X_memset"]       = lle_X_memset;
509   (*FuncNames)["lle_X_memcpy"]       = lle_X_memcpy;
510 }
511