ExternalFunctions.cpp revision 288943 
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/Config/config.h"     // Detect libffi
24#include "llvm/IR/DataLayout.h"
25#include "llvm/IR/DerivedTypes.h"
26#include "llvm/IR/Module.h"
27#include "llvm/Support/DynamicLibrary.h"
28#include "llvm/Support/ErrorHandling.h"
29#include "llvm/Support/ManagedStatic.h"
30#include "llvm/Support/Mutex.h"
31#include "llvm/Support/UniqueLock.h"
32#include <cmath>
33#include <csignal>
34#include <cstdio>
35#include <cstring>
36#include <map>
37
38#ifdef HAVE_FFI_CALL
39#ifdef HAVE_FFI_H
40#include <ffi.h>
41#define USE_LIBFFI
42#elif HAVE_FFI_FFI_H
43#include <ffi/ffi.h>
44#define USE_LIBFFI
45#endif
46#endif
47
48using namespace llvm;
49
50static ManagedStatic<sys::Mutex> FunctionsLock;
51
52typedef GenericValue (*ExFunc)(FunctionType *, ArrayRef<GenericValue>);
53static ManagedStatic<std::map<const Function *, ExFunc> > ExportedFunctions;
54static ManagedStatic<std::map<std::string, ExFunc> > FuncNames;
55
56#ifdef USE_LIBFFI
57typedef void (*RawFunc)();
58static ManagedStatic<std::map<const Function *, RawFunc> > RawFunctions;
59#endif
60
61static Interpreter *TheInterpreter;
62
63static char getTypeID(Type *Ty) {
64  switch (Ty->getTypeID()) {
65  case Type::VoidTyID:    return 'V';
66  case Type::IntegerTyID:
67    switch (cast<IntegerType>(Ty)->getBitWidth()) {
68      case 1:  return 'o';
69      case 8:  return 'B';
70      case 16: return 'S';
71      case 32: return 'I';
72      case 64: return 'L';
73      default: return 'N';
74    }
75  case Type::FloatTyID:   return 'F';
76  case Type::DoubleTyID:  return 'D';
77  case Type::PointerTyID: return 'P';
78  case Type::FunctionTyID:return 'M';
79  case Type::StructTyID:  return 'T';
80  case Type::ArrayTyID:   return 'A';
81  default: return 'U';
82  }
83}
84
85// Try to find address of external function given a Function object.
86// Please note, that interpreter doesn't know how to assemble a
87// real call in general case (this is JIT job), that's why it assumes,
88// that all external functions has the same (and pretty "general") signature.
89// The typical example of such functions are "lle_X_" ones.
90static ExFunc lookupFunction(const Function *F) {
91  // Function not found, look it up... start by figuring out what the
92  // composite function name should be.
93  std::string ExtName = "lle_";
94  FunctionType *FT = F->getFunctionType();
95  for (unsigned i = 0, e = FT->getNumContainedTypes(); i != e; ++i)
96    ExtName += getTypeID(FT->getContainedType(i));
97  ExtName += ("_" + F->getName()).str();
98
99  sys::ScopedLock Writer(*FunctionsLock);
100  ExFunc FnPtr = (*FuncNames)[ExtName];
101  if (!FnPtr)
102    FnPtr = (*FuncNames)[("lle_X_" + F->getName()).str()];
103  if (!FnPtr)  // Try calling a generic function... if it exists...
104    FnPtr = (ExFunc)(intptr_t)sys::DynamicLibrary::SearchForAddressOfSymbol(
105        ("lle_X_" + F->getName()).str());
106  if (FnPtr)
107    ExportedFunctions->insert(std::make_pair(F, FnPtr));  // Cache for later
108  return FnPtr;
109}
110
111#ifdef USE_LIBFFI
112static ffi_type *ffiTypeFor(Type *Ty) {
113  switch (Ty->getTypeID()) {
114    case Type::VoidTyID: return &ffi_type_void;
115    case Type::IntegerTyID:
116      switch (cast<IntegerType>(Ty)->getBitWidth()) {
117        case 8:  return &ffi_type_sint8;
118        case 16: return &ffi_type_sint16;
119        case 32: return &ffi_type_sint32;
120        case 64: return &ffi_type_sint64;
121      }
122    case Type::FloatTyID:   return &ffi_type_float;
123    case Type::DoubleTyID:  return &ffi_type_double;
124    case Type::PointerTyID: return &ffi_type_pointer;
125    default: break;
126  }
127  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
128  report_fatal_error("Type could not be mapped for use with libffi.");
129  return NULL;
130}
131
132static void *ffiValueFor(Type *Ty, const GenericValue &AV,
133                         void *ArgDataPtr) {
134  switch (Ty->getTypeID()) {
135    case Type::IntegerTyID:
136      switch (cast<IntegerType>(Ty)->getBitWidth()) {
137        case 8: {
138          int8_t *I8Ptr = (int8_t *) ArgDataPtr;
139          *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
140          return ArgDataPtr;
141        }
142        case 16: {
143          int16_t *I16Ptr = (int16_t *) ArgDataPtr;
144          *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
145          return ArgDataPtr;
146        }
147        case 32: {
148          int32_t *I32Ptr = (int32_t *) ArgDataPtr;
149          *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
150          return ArgDataPtr;
151        }
152        case 64: {
153          int64_t *I64Ptr = (int64_t *) ArgDataPtr;
154          *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
155          return ArgDataPtr;
156        }
157      }
158    case Type::FloatTyID: {
159      float *FloatPtr = (float *) ArgDataPtr;
160      *FloatPtr = AV.FloatVal;
161      return ArgDataPtr;
162    }
163    case Type::DoubleTyID: {
164      double *DoublePtr = (double *) ArgDataPtr;
165      *DoublePtr = AV.DoubleVal;
166      return ArgDataPtr;
167    }
168    case Type::PointerTyID: {
169      void **PtrPtr = (void **) ArgDataPtr;
170      *PtrPtr = GVTOP(AV);
171      return ArgDataPtr;
172    }
173    default: break;
174  }
175  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
176  report_fatal_error("Type value could not be mapped for use with libffi.");
177  return NULL;
178}
179
180static bool ffiInvoke(RawFunc Fn, Function *F, ArrayRef<GenericValue> ArgVals,
181                      const DataLayout *TD, GenericValue &Result) {
182  ffi_cif cif;
183  FunctionType *FTy = F->getFunctionType();
184  const unsigned NumArgs = F->arg_size();
185
186  // TODO: We don't have type information about the remaining arguments, because
187  // this information is never passed into ExecutionEngine::runFunction().
188  if (ArgVals.size() > NumArgs && F->isVarArg()) {
189    report_fatal_error("Calling external var arg function '" + F->getName()
190                      + "' is not supported by the Interpreter.");
191  }
192
193  unsigned ArgBytes = 0;
194
195  std::vector<ffi_type*> args(NumArgs);
196  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
197       A != E; ++A) {
198    const unsigned ArgNo = A->getArgNo();
199    Type *ArgTy = FTy->getParamType(ArgNo);
200    args[ArgNo] = ffiTypeFor(ArgTy);
201    ArgBytes += TD->getTypeStoreSize(ArgTy);
202  }
203
204  SmallVector<uint8_t, 128> ArgData;
205  ArgData.resize(ArgBytes);
206  uint8_t *ArgDataPtr = ArgData.data();
207  SmallVector<void*, 16> values(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    values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
213    ArgDataPtr += TD->getTypeStoreSize(ArgTy);
214  }
215
216  Type *RetTy = FTy->getReturnType();
217  ffi_type *rtype = ffiTypeFor(RetTy);
218
219  if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, &args[0]) == FFI_OK) {
220    SmallVector<uint8_t, 128> ret;
221    if (RetTy->getTypeID() != Type::VoidTyID)
222      ret.resize(TD->getTypeStoreSize(RetTy));
223    ffi_call(&cif, Fn, ret.data(), values.data());
224    switch (RetTy->getTypeID()) {
225      case Type::IntegerTyID:
226        switch (cast<IntegerType>(RetTy)->getBitWidth()) {
227          case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret.data()); break;
228          case 16: Result.IntVal = APInt(16, *(int16_t*) ret.data()); break;
229          case 32: Result.IntVal = APInt(32, *(int32_t*) ret.data()); break;
230          case 64: Result.IntVal = APInt(64, *(int64_t*) ret.data()); break;
231        }
232        break;
233      case Type::FloatTyID:   Result.FloatVal   = *(float *) ret.data(); break;
234      case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret.data(); break;
235      case Type::PointerTyID: Result.PointerVal = *(void **) ret.data(); break;
236      default: break;
237    }
238    return true;
239  }
240
241  return false;
242}
243#endif // USE_LIBFFI
244
245GenericValue Interpreter::callExternalFunction(Function *F,
246                                               ArrayRef<GenericValue> ArgVals) {
247  TheInterpreter = this;
248
249  unique_lock<sys::Mutex> Guard(*FunctionsLock);
250
251  // Do a lookup to see if the function is in our cache... this should just be a
252  // deferred annotation!
253  std::map<const Function *, ExFunc>::iterator FI = ExportedFunctions->find(F);
254  if (ExFunc Fn = (FI == ExportedFunctions->end()) ? lookupFunction(F)
255                                                   : FI->second) {
256    Guard.unlock();
257    return Fn(F->getFunctionType(), ArgVals);
258  }
259
260#ifdef USE_LIBFFI
261  std::map<const Function *, RawFunc>::iterator RF = RawFunctions->find(F);
262  RawFunc RawFn;
263  if (RF == RawFunctions->end()) {
264    RawFn = (RawFunc)(intptr_t)
265      sys::DynamicLibrary::SearchForAddressOfSymbol(F->getName());
266    if (!RawFn)
267      RawFn = (RawFunc)(intptr_t)getPointerToGlobalIfAvailable(F);
268    if (RawFn != 0)
269      RawFunctions->insert(std::make_pair(F, RawFn));  // Cache for later
270  } else {
271    RawFn = RF->second;
272  }
273
274  Guard.unlock();
275
276  GenericValue Result;
277  if (RawFn != 0 && ffiInvoke(RawFn, F, ArgVals, getDataLayout(), Result))
278    return Result;
279#endif // USE_LIBFFI
280
281  if (F->getName() == "__main")
282    errs() << "Tried to execute an unknown external function: "
283      << *F->getType() << " __main\n";
284  else
285    report_fatal_error("Tried to execute an unknown external function: " +
286                       F->getName());
287#ifndef USE_LIBFFI
288  errs() << "Recompiling LLVM with --enable-libffi might help.\n";
289#endif
290  return GenericValue();
291}
292
293
294//===----------------------------------------------------------------------===//
295//  Functions "exported" to the running application...
296//
297
298// void atexit(Function*)
299static GenericValue lle_X_atexit(FunctionType *FT,
300                                 ArrayRef<GenericValue> Args) {
301  assert(Args.size() == 1);
302  TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
303  GenericValue GV;
304  GV.IntVal = 0;
305  return GV;
306}
307
308// void exit(int)
309static GenericValue lle_X_exit(FunctionType *FT, ArrayRef<GenericValue> Args) {
310  TheInterpreter->exitCalled(Args[0]);
311  return GenericValue();
312}
313
314// void abort(void)
315static GenericValue lle_X_abort(FunctionType *FT, ArrayRef<GenericValue> Args) {
316  //FIXME: should we report or raise here?
317  //report_fatal_error("Interpreted program raised SIGABRT");
318  raise (SIGABRT);
319  return GenericValue();
320}
321
322// int sprintf(char *, const char *, ...) - a very rough implementation to make
323// output useful.
324static GenericValue lle_X_sprintf(FunctionType *FT,
325                                  ArrayRef<GenericValue> Args) {
326  char *OutputBuffer = (char *)GVTOP(Args[0]);
327  const char *FmtStr = (const char *)GVTOP(Args[1]);
328  unsigned ArgNo = 2;
329
330  // printf should return # chars printed.  This is completely incorrect, but
331  // close enough for now.
332  GenericValue GV;
333  GV.IntVal = APInt(32, strlen(FmtStr));
334  while (1) {
335    switch (*FmtStr) {
336    case 0: return GV;             // Null terminator...
337    default:                       // Normal nonspecial character
338      sprintf(OutputBuffer++, "%c", *FmtStr++);
339      break;
340    case '\\': {                   // Handle escape codes
341      sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
342      FmtStr += 2; OutputBuffer += 2;
343      break;
344    }
345    case '%': {                    // Handle format specifiers
346      char FmtBuf[100] = "", Buffer[1000] = "";
347      char *FB = FmtBuf;
348      *FB++ = *FmtStr++;
349      char Last = *FB++ = *FmtStr++;
350      unsigned HowLong = 0;
351      while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
352             Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
353             Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
354             Last != 'p' && Last != 's' && Last != '%') {
355        if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
356        Last = *FB++ = *FmtStr++;
357      }
358      *FB = 0;
359
360      switch (Last) {
361      case '%':
362        memcpy(Buffer, "%", 2); break;
363      case 'c':
364        sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
365        break;
366      case 'd': case 'i':
367      case 'u': case 'o':
368      case 'x': case 'X':
369        if (HowLong >= 1) {
370          if (HowLong == 1 &&
371              TheInterpreter->getDataLayout()->getPointerSizeInBits() == 64 &&
372              sizeof(long) < sizeof(int64_t)) {
373            // Make sure we use %lld with a 64 bit argument because we might be
374            // compiling LLI on a 32 bit compiler.
375            unsigned Size = strlen(FmtBuf);
376            FmtBuf[Size] = FmtBuf[Size-1];
377            FmtBuf[Size+1] = 0;
378            FmtBuf[Size-1] = 'l';
379          }
380          sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
381        } else
382          sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
383        break;
384      case 'e': case 'E': case 'g': case 'G': case 'f':
385        sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
386      case 'p':
387        sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
388      case 's':
389        sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
390      default:
391        errs() << "<unknown printf code '" << *FmtStr << "'!>";
392        ArgNo++; break;
393      }
394      size_t Len = strlen(Buffer);
395      memcpy(OutputBuffer, Buffer, Len + 1);
396      OutputBuffer += Len;
397      }
398      break;
399    }
400  }
401  return GV;
402}
403
404// int printf(const char *, ...) - a very rough implementation to make output
405// useful.
406static GenericValue lle_X_printf(FunctionType *FT,
407                                 ArrayRef<GenericValue> Args) {
408  char Buffer[10000];
409  std::vector<GenericValue> NewArgs;
410  NewArgs.push_back(PTOGV((void*)&Buffer[0]));
411  NewArgs.insert(NewArgs.end(), Args.begin(), Args.end());
412  GenericValue GV = lle_X_sprintf(FT, NewArgs);
413  outs() << Buffer;
414  return GV;
415}
416
417// int sscanf(const char *format, ...);
418static GenericValue lle_X_sscanf(FunctionType *FT,
419                                 ArrayRef<GenericValue> args) {
420  assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
421
422  char *Args[10];
423  for (unsigned i = 0; i < args.size(); ++i)
424    Args[i] = (char*)GVTOP(args[i]);
425
426  GenericValue GV;
427  GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
428                    Args[5], Args[6], Args[7], Args[8], Args[9]));
429  return GV;
430}
431
432// int scanf(const char *format, ...);
433static GenericValue lle_X_scanf(FunctionType *FT, ArrayRef<GenericValue> args) {
434  assert(args.size() < 10 && "Only handle up to 10 args to scanf right now!");
435
436  char *Args[10];
437  for (unsigned i = 0; i < args.size(); ++i)
438    Args[i] = (char*)GVTOP(args[i]);
439
440  GenericValue GV;
441  GV.IntVal = APInt(32, scanf( Args[0], Args[1], Args[2], Args[3], Args[4],
442                    Args[5], Args[6], Args[7], Args[8], Args[9]));
443  return GV;
444}
445
446// int fprintf(FILE *, const char *, ...) - a very rough implementation to make
447// output useful.
448static GenericValue lle_X_fprintf(FunctionType *FT,
449                                  ArrayRef<GenericValue> Args) {
450  assert(Args.size() >= 2);
451  char Buffer[10000];
452  std::vector<GenericValue> NewArgs;
453  NewArgs.push_back(PTOGV(Buffer));
454  NewArgs.insert(NewArgs.end(), Args.begin()+1, Args.end());
455  GenericValue GV = lle_X_sprintf(FT, NewArgs);
456
457  fputs(Buffer, (FILE *) GVTOP(Args[0]));
458  return GV;
459}
460
461static GenericValue lle_X_memset(FunctionType *FT,
462                                 ArrayRef<GenericValue> Args) {
463  int val = (int)Args[1].IntVal.getSExtValue();
464  size_t len = (size_t)Args[2].IntVal.getZExtValue();
465  memset((void *)GVTOP(Args[0]), val, len);
466  // llvm.memset.* returns void, lle_X_* returns GenericValue,
467  // so here we return GenericValue with IntVal set to zero
468  GenericValue GV;
469  GV.IntVal = 0;
470  return GV;
471}
472
473static GenericValue lle_X_memcpy(FunctionType *FT,
474                                 ArrayRef<GenericValue> Args) {
475  memcpy(GVTOP(Args[0]), GVTOP(Args[1]),
476         (size_t)(Args[2].IntVal.getLimitedValue()));
477
478  // llvm.memcpy* 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
485void Interpreter::initializeExternalFunctions() {
486  sys::ScopedLock Writer(*FunctionsLock);
487  (*FuncNames)["lle_X_atexit"]       = lle_X_atexit;
488  (*FuncNames)["lle_X_exit"]         = lle_X_exit;
489  (*FuncNames)["lle_X_abort"]        = lle_X_abort;
490
491  (*FuncNames)["lle_X_printf"]       = lle_X_printf;
492  (*FuncNames)["lle_X_sprintf"]      = lle_X_sprintf;
493  (*FuncNames)["lle_X_sscanf"]       = lle_X_sscanf;
494  (*FuncNames)["lle_X_scanf"]        = lle_X_scanf;
495  (*FuncNames)["lle_X_fprintf"]      = lle_X_fprintf;
496  (*FuncNames)["lle_X_memset"]       = lle_X_memset;
497  (*FuncNames)["lle_X_memcpy"]       = lle_X_memcpy;
498}
499