ExternalFunctions.cpp revision 321369 
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
59using namespace llvm;
60
61static ManagedStatic<sys::Mutex> FunctionsLock;
62
63typedef GenericValue (*ExFunc)(FunctionType *, ArrayRef<GenericValue>);
64static ManagedStatic<std::map<const Function *, ExFunc> > ExportedFunctions;
65static ManagedStatic<std::map<std::string, ExFunc> > FuncNames;
66
67#ifdef USE_LIBFFI
68typedef void (*RawFunc)();
69static ManagedStatic<std::map<const Function *, RawFunc> > RawFunctions;
70#endif
71
72static Interpreter *TheInterpreter;
73
74static 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.
101static 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  for (unsigned i = 0, e = FT->getNumContainedTypes(); i != e; ++i)
107    ExtName += getTypeID(FT->getContainedType(i));
108  ExtName += ("_" + F->getName()).str();
109
110  sys::ScopedLock Writer(*FunctionsLock);
111  ExFunc FnPtr = (*FuncNames)[ExtName];
112  if (!FnPtr)
113    FnPtr = (*FuncNames)[("lle_X_" + F->getName()).str()];
114  if (!FnPtr)  // Try calling a generic function... if it exists...
115    FnPtr = (ExFunc)(intptr_t)sys::DynamicLibrary::SearchForAddressOfSymbol(
116        ("lle_X_" + F->getName()).str());
117  if (FnPtr)
118    ExportedFunctions->insert(std::make_pair(F, FnPtr));  // Cache for later
119  return FnPtr;
120}
121
122#ifdef USE_LIBFFI
123static ffi_type *ffiTypeFor(Type *Ty) {
124  switch (Ty->getTypeID()) {
125    case Type::VoidTyID: return &ffi_type_void;
126    case Type::IntegerTyID:
127      switch (cast<IntegerType>(Ty)->getBitWidth()) {
128        case 8:  return &ffi_type_sint8;
129        case 16: return &ffi_type_sint16;
130        case 32: return &ffi_type_sint32;
131        case 64: return &ffi_type_sint64;
132      }
133    case Type::FloatTyID:   return &ffi_type_float;
134    case Type::DoubleTyID:  return &ffi_type_double;
135    case Type::PointerTyID: return &ffi_type_pointer;
136    default: break;
137  }
138  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
139  report_fatal_error("Type could not be mapped for use with libffi.");
140  return NULL;
141}
142
143static void *ffiValueFor(Type *Ty, const GenericValue &AV,
144                         void *ArgDataPtr) {
145  switch (Ty->getTypeID()) {
146    case Type::IntegerTyID:
147      switch (cast<IntegerType>(Ty)->getBitWidth()) {
148        case 8: {
149          int8_t *I8Ptr = (int8_t *) ArgDataPtr;
150          *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
151          return ArgDataPtr;
152        }
153        case 16: {
154          int16_t *I16Ptr = (int16_t *) ArgDataPtr;
155          *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
156          return ArgDataPtr;
157        }
158        case 32: {
159          int32_t *I32Ptr = (int32_t *) ArgDataPtr;
160          *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
161          return ArgDataPtr;
162        }
163        case 64: {
164          int64_t *I64Ptr = (int64_t *) ArgDataPtr;
165          *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
166          return ArgDataPtr;
167        }
168      }
169    case Type::FloatTyID: {
170      float *FloatPtr = (float *) ArgDataPtr;
171      *FloatPtr = AV.FloatVal;
172      return ArgDataPtr;
173    }
174    case Type::DoubleTyID: {
175      double *DoublePtr = (double *) ArgDataPtr;
176      *DoublePtr = AV.DoubleVal;
177      return ArgDataPtr;
178    }
179    case Type::PointerTyID: {
180      void **PtrPtr = (void **) ArgDataPtr;
181      *PtrPtr = GVTOP(AV);
182      return ArgDataPtr;
183    }
184    default: break;
185  }
186  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
187  report_fatal_error("Type value could not be mapped for use with libffi.");
188  return NULL;
189}
190
191static bool ffiInvoke(RawFunc Fn, Function *F, ArrayRef<GenericValue> ArgVals,
192                      const DataLayout &TD, GenericValue &Result) {
193  ffi_cif cif;
194  FunctionType *FTy = F->getFunctionType();
195  const unsigned NumArgs = F->arg_size();
196
197  // TODO: We don't have type information about the remaining arguments, because
198  // this information is never passed into ExecutionEngine::runFunction().
199  if (ArgVals.size() > NumArgs && F->isVarArg()) {
200    report_fatal_error("Calling external var arg function '" + F->getName()
201                      + "' is not supported by the Interpreter.");
202  }
203
204  unsigned ArgBytes = 0;
205
206  std::vector<ffi_type*> args(NumArgs);
207  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
208       A != E; ++A) {
209    const unsigned ArgNo = A->getArgNo();
210    Type *ArgTy = FTy->getParamType(ArgNo);
211    args[ArgNo] = ffiTypeFor(ArgTy);
212    ArgBytes += TD.getTypeStoreSize(ArgTy);
213  }
214
215  SmallVector<uint8_t, 128> ArgData;
216  ArgData.resize(ArgBytes);
217  uint8_t *ArgDataPtr = ArgData.data();
218  SmallVector<void*, 16> values(NumArgs);
219  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
220       A != E; ++A) {
221    const unsigned ArgNo = A->getArgNo();
222    Type *ArgTy = FTy->getParamType(ArgNo);
223    values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
224    ArgDataPtr += TD.getTypeStoreSize(ArgTy);
225  }
226
227  Type *RetTy = FTy->getReturnType();
228  ffi_type *rtype = ffiTypeFor(RetTy);
229
230  if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, &args[0]) == FFI_OK) {
231    SmallVector<uint8_t, 128> ret;
232    if (RetTy->getTypeID() != Type::VoidTyID)
233      ret.resize(TD.getTypeStoreSize(RetTy));
234    ffi_call(&cif, Fn, ret.data(), values.data());
235    switch (RetTy->getTypeID()) {
236      case Type::IntegerTyID:
237        switch (cast<IntegerType>(RetTy)->getBitWidth()) {
238          case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret.data()); break;
239          case 16: Result.IntVal = APInt(16, *(int16_t*) ret.data()); break;
240          case 32: Result.IntVal = APInt(32, *(int32_t*) ret.data()); break;
241          case 64: Result.IntVal = APInt(64, *(int64_t*) ret.data()); break;
242        }
243        break;
244      case Type::FloatTyID:   Result.FloatVal   = *(float *) ret.data(); break;
245      case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret.data(); break;
246      case Type::PointerTyID: Result.PointerVal = *(void **) ret.data(); break;
247      default: break;
248    }
249    return true;
250  }
251
252  return false;
253}
254#endif // USE_LIBFFI
255
256GenericValue Interpreter::callExternalFunction(Function *F,
257                                               ArrayRef<GenericValue> ArgVals) {
258  TheInterpreter = this;
259
260  unique_lock<sys::Mutex> Guard(*FunctionsLock);
261
262  // Do a lookup to see if the function is in our cache... this should just be a
263  // deferred annotation!
264  std::map<const Function *, ExFunc>::iterator FI = ExportedFunctions->find(F);
265  if (ExFunc Fn = (FI == ExportedFunctions->end()) ? lookupFunction(F)
266                                                   : FI->second) {
267    Guard.unlock();
268    return Fn(F->getFunctionType(), ArgVals);
269  }
270
271#ifdef USE_LIBFFI
272  std::map<const Function *, RawFunc>::iterator RF = RawFunctions->find(F);
273  RawFunc RawFn;
274  if (RF == RawFunctions->end()) {
275    RawFn = (RawFunc)(intptr_t)
276      sys::DynamicLibrary::SearchForAddressOfSymbol(F->getName());
277    if (!RawFn)
278      RawFn = (RawFunc)(intptr_t)getPointerToGlobalIfAvailable(F);
279    if (RawFn != 0)
280      RawFunctions->insert(std::make_pair(F, RawFn));  // Cache for later
281  } else {
282    RawFn = RF->second;
283  }
284
285  Guard.unlock();
286
287  GenericValue Result;
288  if (RawFn != 0 && ffiInvoke(RawFn, F, ArgVals, getDataLayout(), Result))
289    return Result;
290#endif // USE_LIBFFI
291
292  if (F->getName() == "__main")
293    errs() << "Tried to execute an unknown external function: "
294      << *F->getType() << " __main\n";
295  else
296    report_fatal_error("Tried to execute an unknown external function: " +
297                       F->getName());
298#ifndef USE_LIBFFI
299  errs() << "Recompiling LLVM with --enable-libffi might help.\n";
300#endif
301  return GenericValue();
302}
303
304//===----------------------------------------------------------------------===//
305//  Functions "exported" to the running application...
306//
307
308// void atexit(Function*)
309static GenericValue lle_X_atexit(FunctionType *FT,
310                                 ArrayRef<GenericValue> Args) {
311  assert(Args.size() == 1);
312  TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
313  GenericValue GV;
314  GV.IntVal = 0;
315  return GV;
316}
317
318// void exit(int)
319static GenericValue lle_X_exit(FunctionType *FT, ArrayRef<GenericValue> Args) {
320  TheInterpreter->exitCalled(Args[0]);
321  return GenericValue();
322}
323
324// void abort(void)
325static GenericValue lle_X_abort(FunctionType *FT, ArrayRef<GenericValue> Args) {
326  //FIXME: should we report or raise here?
327  //report_fatal_error("Interpreted program raised SIGABRT");
328  raise (SIGABRT);
329  return GenericValue();
330}
331
332// int sprintf(char *, const char *, ...) - a very rough implementation to make
333// output useful.
334static GenericValue lle_X_sprintf(FunctionType *FT,
335                                  ArrayRef<GenericValue> Args) {
336  char *OutputBuffer = (char *)GVTOP(Args[0]);
337  const char *FmtStr = (const char *)GVTOP(Args[1]);
338  unsigned ArgNo = 2;
339
340  // printf should return # chars printed.  This is completely incorrect, but
341  // close enough for now.
342  GenericValue GV;
343  GV.IntVal = APInt(32, strlen(FmtStr));
344  while (true) {
345    switch (*FmtStr) {
346    case 0: return GV;             // Null terminator...
347    default:                       // Normal nonspecial character
348      sprintf(OutputBuffer++, "%c", *FmtStr++);
349      break;
350    case '\\': {                   // Handle escape codes
351      sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
352      FmtStr += 2; OutputBuffer += 2;
353      break;
354    }
355    case '%': {                    // Handle format specifiers
356      char FmtBuf[100] = "", Buffer[1000] = "";
357      char *FB = FmtBuf;
358      *FB++ = *FmtStr++;
359      char Last = *FB++ = *FmtStr++;
360      unsigned HowLong = 0;
361      while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
362             Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
363             Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
364             Last != 'p' && Last != 's' && Last != '%') {
365        if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
366        Last = *FB++ = *FmtStr++;
367      }
368      *FB = 0;
369
370      switch (Last) {
371      case '%':
372        memcpy(Buffer, "%", 2); break;
373      case 'c':
374        sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
375        break;
376      case 'd': case 'i':
377      case 'u': case 'o':
378      case 'x': case 'X':
379        if (HowLong >= 1) {
380          if (HowLong == 1 &&
381              TheInterpreter->getDataLayout().getPointerSizeInBits() == 64 &&
382              sizeof(long) < sizeof(int64_t)) {
383            // Make sure we use %lld with a 64 bit argument because we might be
384            // compiling LLI on a 32 bit compiler.
385            unsigned Size = strlen(FmtBuf);
386            FmtBuf[Size] = FmtBuf[Size-1];
387            FmtBuf[Size+1] = 0;
388            FmtBuf[Size-1] = 'l';
389          }
390          sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
391        } else
392          sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
393        break;
394      case 'e': case 'E': case 'g': case 'G': case 'f':
395        sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
396      case 'p':
397        sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
398      case 's':
399        sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
400      default:
401        errs() << "<unknown printf code '" << *FmtStr << "'!>";
402        ArgNo++; break;
403      }
404      size_t Len = strlen(Buffer);
405      memcpy(OutputBuffer, Buffer, Len + 1);
406      OutputBuffer += Len;
407      }
408      break;
409    }
410  }
411  return GV;
412}
413
414// int printf(const char *, ...) - a very rough implementation to make output
415// useful.
416static GenericValue lle_X_printf(FunctionType *FT,
417                                 ArrayRef<GenericValue> Args) {
418  char Buffer[10000];
419  std::vector<GenericValue> NewArgs;
420  NewArgs.push_back(PTOGV((void*)&Buffer[0]));
421  NewArgs.insert(NewArgs.end(), Args.begin(), Args.end());
422  GenericValue GV = lle_X_sprintf(FT, NewArgs);
423  outs() << Buffer;
424  return GV;
425}
426
427// int sscanf(const char *format, ...);
428static GenericValue lle_X_sscanf(FunctionType *FT,
429                                 ArrayRef<GenericValue> args) {
430  assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
431
432  char *Args[10];
433  for (unsigned i = 0; i < args.size(); ++i)
434    Args[i] = (char*)GVTOP(args[i]);
435
436  GenericValue GV;
437  GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
438                    Args[5], Args[6], Args[7], Args[8], Args[9]));
439  return GV;
440}
441
442// int scanf(const char *format, ...);
443static GenericValue lle_X_scanf(FunctionType *FT, ArrayRef<GenericValue> args) {
444  assert(args.size() < 10 && "Only handle up to 10 args to scanf right now!");
445
446  char *Args[10];
447  for (unsigned i = 0; i < args.size(); ++i)
448    Args[i] = (char*)GVTOP(args[i]);
449
450  GenericValue GV;
451  GV.IntVal = APInt(32, scanf( Args[0], Args[1], Args[2], Args[3], Args[4],
452                    Args[5], Args[6], Args[7], Args[8], Args[9]));
453  return GV;
454}
455
456// int fprintf(FILE *, const char *, ...) - a very rough implementation to make
457// output useful.
458static GenericValue lle_X_fprintf(FunctionType *FT,
459                                  ArrayRef<GenericValue> Args) {
460  assert(Args.size() >= 2);
461  char Buffer[10000];
462  std::vector<GenericValue> NewArgs;
463  NewArgs.push_back(PTOGV(Buffer));
464  NewArgs.insert(NewArgs.end(), Args.begin()+1, Args.end());
465  GenericValue GV = lle_X_sprintf(FT, NewArgs);
466
467  fputs(Buffer, (FILE *) GVTOP(Args[0]));
468  return GV;
469}
470
471static GenericValue lle_X_memset(FunctionType *FT,
472                                 ArrayRef<GenericValue> Args) {
473  int val = (int)Args[1].IntVal.getSExtValue();
474  size_t len = (size_t)Args[2].IntVal.getZExtValue();
475  memset((void *)GVTOP(Args[0]), val, len);
476  // llvm.memset.* returns void, lle_X_* returns GenericValue,
477  // so here we return GenericValue with IntVal set to zero
478  GenericValue GV;
479  GV.IntVal = 0;
480  return GV;
481}
482
483static GenericValue lle_X_memcpy(FunctionType *FT,
484                                 ArrayRef<GenericValue> Args) {
485  memcpy(GVTOP(Args[0]), GVTOP(Args[1]),
486         (size_t)(Args[2].IntVal.getLimitedValue()));
487
488  // llvm.memcpy* returns void, lle_X_* returns GenericValue,
489  // so here we return GenericValue with IntVal set to zero
490  GenericValue GV;
491  GV.IntVal = 0;
492  return GV;
493}
494
495void Interpreter::initializeExternalFunctions() {
496  sys::ScopedLock Writer(*FunctionsLock);
497  (*FuncNames)["lle_X_atexit"]       = lle_X_atexit;
498  (*FuncNames)["lle_X_exit"]         = lle_X_exit;
499  (*FuncNames)["lle_X_abort"]        = lle_X_abort;
500
501  (*FuncNames)["lle_X_printf"]       = lle_X_printf;
502  (*FuncNames)["lle_X_sprintf"]      = lle_X_sprintf;
503  (*FuncNames)["lle_X_sscanf"]       = lle_X_sscanf;
504  (*FuncNames)["lle_X_scanf"]        = lle_X_scanf;
505  (*FuncNames)["lle_X_fprintf"]      = lle_X_fprintf;
506  (*FuncNames)["lle_X_memset"]       = lle_X_memset;
507  (*FuncNames)["lle_X_memcpy"]       = lle_X_memcpy;
508}
509