ExternalFunctions.cpp revision 207618 
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/DerivedTypes.h"
24#include "llvm/Module.h"
25#include "llvm/Config/config.h"     // Detect libffi
26#include "llvm/Support/ErrorHandling.h"
27#include "llvm/System/DynamicLibrary.h"
28#include "llvm/Target/TargetData.h"
29#include "llvm/Support/ManagedStatic.h"
30#include "llvm/System/Mutex.h"
31#include <csignal>
32#include <cstdio>
33#include <map>
34#include <cmath>
35#include <cstring>
36
37#ifdef HAVE_FFI_CALL
38#ifdef HAVE_FFI_H
39#include <ffi.h>
40#define USE_LIBFFI
41#elif HAVE_FFI_FFI_H
42#include <ffi/ffi.h>
43#define USE_LIBFFI
44#endif
45#endif
46
47using namespace llvm;
48
49static ManagedStatic<sys::Mutex> FunctionsLock;
50
51typedef GenericValue (*ExFunc)(const FunctionType *,
52                               const std::vector<GenericValue> &);
53static ManagedStatic<std::map<const Function *, ExFunc> > ExportedFunctions;
54static 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(const 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  case Type::OpaqueTyID:  return 'O';
82  default: return 'U';
83  }
84}
85
86// Try to find address of external function given a Function object.
87// Please note, that interpreter doesn't know how to assemble a
88// real call in general case (this is JIT job), that's why it assumes,
89// that all external functions has the same (and pretty "general") signature.
90// The typical example of such functions are "lle_X_" ones.
91static ExFunc lookupFunction(const Function *F) {
92  // Function not found, look it up... start by figuring out what the
93  // composite function name should be.
94  std::string ExtName = "lle_";
95  const FunctionType *FT = F->getFunctionType();
96  for (unsigned i = 0, e = FT->getNumContainedTypes(); i != e; ++i)
97    ExtName += getTypeID(FT->getContainedType(i));
98  ExtName + "_" + F->getNameStr();
99
100  sys::ScopedLock Writer(*FunctionsLock);
101  ExFunc FnPtr = FuncNames[ExtName];
102  if (FnPtr == 0)
103    FnPtr = FuncNames["lle_X_" + F->getNameStr()];
104  if (FnPtr == 0)  // Try calling a generic function... if it exists...
105    FnPtr = (ExFunc)(intptr_t)
106      sys::DynamicLibrary::SearchForAddressOfSymbol("lle_X_"+F->getNameStr());
107  if (FnPtr != 0)
108    ExportedFunctions->insert(std::make_pair(F, FnPtr));  // Cache for later
109  return FnPtr;
110}
111
112#ifdef USE_LIBFFI
113static ffi_type *ffiTypeFor(const Type *Ty) {
114  switch (Ty->getTypeID()) {
115    case Type::VoidTyID: return &ffi_type_void;
116    case Type::IntegerTyID:
117      switch (cast<IntegerType>(Ty)->getBitWidth()) {
118        case 8:  return &ffi_type_sint8;
119        case 16: return &ffi_type_sint16;
120        case 32: return &ffi_type_sint32;
121        case 64: return &ffi_type_sint64;
122      }
123    case Type::FloatTyID:   return &ffi_type_float;
124    case Type::DoubleTyID:  return &ffi_type_double;
125    case Type::PointerTyID: return &ffi_type_pointer;
126    default: break;
127  }
128  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
129  report_fatal_error("Type could not be mapped for use with libffi.");
130  return NULL;
131}
132
133static void *ffiValueFor(const Type *Ty, const GenericValue &AV,
134                         void *ArgDataPtr) {
135  switch (Ty->getTypeID()) {
136    case Type::IntegerTyID:
137      switch (cast<IntegerType>(Ty)->getBitWidth()) {
138        case 8: {
139          int8_t *I8Ptr = (int8_t *) ArgDataPtr;
140          *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
141          return ArgDataPtr;
142        }
143        case 16: {
144          int16_t *I16Ptr = (int16_t *) ArgDataPtr;
145          *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
146          return ArgDataPtr;
147        }
148        case 32: {
149          int32_t *I32Ptr = (int32_t *) ArgDataPtr;
150          *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
151          return ArgDataPtr;
152        }
153        case 64: {
154          int64_t *I64Ptr = (int64_t *) ArgDataPtr;
155          *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
156          return ArgDataPtr;
157        }
158      }
159    case Type::FloatTyID: {
160      float *FloatPtr = (float *) ArgDataPtr;
161      *FloatPtr = AV.FloatVal;
162      return ArgDataPtr;
163    }
164    case Type::DoubleTyID: {
165      double *DoublePtr = (double *) ArgDataPtr;
166      *DoublePtr = AV.DoubleVal;
167      return ArgDataPtr;
168    }
169    case Type::PointerTyID: {
170      void **PtrPtr = (void **) ArgDataPtr;
171      *PtrPtr = GVTOP(AV);
172      return ArgDataPtr;
173    }
174    default: break;
175  }
176  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
177  report_fatal_error("Type value could not be mapped for use with libffi.");
178  return NULL;
179}
180
181static bool ffiInvoke(RawFunc Fn, Function *F,
182                      const std::vector<GenericValue> &ArgVals,
183                      const TargetData *TD, GenericValue &Result) {
184  ffi_cif cif;
185  const FunctionType *FTy = F->getFunctionType();
186  const unsigned NumArgs = F->arg_size();
187
188  // TODO: We don't have type information about the remaining arguments, because
189  // this information is never passed into ExecutionEngine::runFunction().
190  if (ArgVals.size() > NumArgs && F->isVarArg()) {
191    report_fatal_error("Calling external var arg function '" + F->getName()
192                      + "' is not supported by the Interpreter.");
193  }
194
195  unsigned ArgBytes = 0;
196
197  std::vector<ffi_type*> args(NumArgs);
198  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
199       A != E; ++A) {
200    const unsigned ArgNo = A->getArgNo();
201    const Type *ArgTy = FTy->getParamType(ArgNo);
202    args[ArgNo] = ffiTypeFor(ArgTy);
203    ArgBytes += TD->getTypeStoreSize(ArgTy);
204  }
205
206  SmallVector<uint8_t, 128> ArgData;
207  ArgData.resize(ArgBytes);
208  uint8_t *ArgDataPtr = ArgData.data();
209  SmallVector<void*, 16> values(NumArgs);
210  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
211       A != E; ++A) {
212    const unsigned ArgNo = A->getArgNo();
213    const Type *ArgTy = FTy->getParamType(ArgNo);
214    values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
215    ArgDataPtr += TD->getTypeStoreSize(ArgTy);
216  }
217
218  const Type *RetTy = FTy->getReturnType();
219  ffi_type *rtype = ffiTypeFor(RetTy);
220
221  if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, &args[0]) == FFI_OK) {
222    SmallVector<uint8_t, 128> ret;
223    if (RetTy->getTypeID() != Type::VoidTyID)
224      ret.resize(TD->getTypeStoreSize(RetTy));
225    ffi_call(&cif, Fn, ret.data(), values.data());
226    switch (RetTy->getTypeID()) {
227      case Type::IntegerTyID:
228        switch (cast<IntegerType>(RetTy)->getBitWidth()) {
229          case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret.data()); break;
230          case 16: Result.IntVal = APInt(16, *(int16_t*) ret.data()); break;
231          case 32: Result.IntVal = APInt(32, *(int32_t*) ret.data()); break;
232          case 64: Result.IntVal = APInt(64, *(int64_t*) ret.data()); break;
233        }
234        break;
235      case Type::FloatTyID:   Result.FloatVal   = *(float *) ret.data(); break;
236      case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret.data(); break;
237      case Type::PointerTyID: Result.PointerVal = *(void **) ret.data(); break;
238      default: break;
239    }
240    return true;
241  }
242
243  return false;
244}
245#endif // USE_LIBFFI
246
247GenericValue Interpreter::callExternalFunction(Function *F,
248                                     const std::vector<GenericValue> &ArgVals) {
249  TheInterpreter = this;
250
251  FunctionsLock->acquire();
252
253  // Do a lookup to see if the function is in our cache... this should just be a
254  // deferred annotation!
255  std::map<const Function *, ExFunc>::iterator FI = ExportedFunctions->find(F);
256  if (ExFunc Fn = (FI == ExportedFunctions->end()) ? lookupFunction(F)
257                                                   : FI->second) {
258    FunctionsLock->release();
259    return Fn(F->getFunctionType(), ArgVals);
260  }
261
262#ifdef USE_LIBFFI
263  std::map<const Function *, RawFunc>::iterator RF = RawFunctions->find(F);
264  RawFunc RawFn;
265  if (RF == RawFunctions->end()) {
266    RawFn = (RawFunc)(intptr_t)
267      sys::DynamicLibrary::SearchForAddressOfSymbol(F->getName());
268    if (!RawFn)
269	RawFn = (RawFunc)(intptr_t)getPointerToGlobalIfAvailable(F);
270    if (RawFn != 0)
271      RawFunctions->insert(std::make_pair(F, RawFn));  // Cache for later
272  } else {
273    RawFn = RF->second;
274  }
275
276  FunctionsLock->release();
277
278  GenericValue Result;
279  if (RawFn != 0 && ffiInvoke(RawFn, F, ArgVals, getTargetData(), Result))
280    return Result;
281#endif // USE_LIBFFI
282
283  if (F->getName() == "__main")
284    errs() << "Tried to execute an unknown external function: "
285      << F->getType()->getDescription() << " __main\n";
286  else
287    report_fatal_error("Tried to execute an unknown external function: " +
288                      F->getType()->getDescription() + " " +F->getName());
289#ifndef USE_LIBFFI
290  errs() << "Recompiling LLVM with --enable-libffi might help.\n";
291#endif
292  return GenericValue();
293}
294
295
296//===----------------------------------------------------------------------===//
297//  Functions "exported" to the running application...
298//
299
300// Visual Studio warns about returning GenericValue in extern "C" linkage
301#ifdef _MSC_VER
302    #pragma warning(disable : 4190)
303#endif
304
305extern "C" {  // Don't add C++ manglings to llvm mangling :)
306
307// void atexit(Function*)
308GenericValue lle_X_atexit(const FunctionType *FT,
309                          const std::vector<GenericValue> &Args) {
310  assert(Args.size() == 1);
311  TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
312  GenericValue GV;
313  GV.IntVal = 0;
314  return GV;
315}
316
317// void exit(int)
318GenericValue lle_X_exit(const FunctionType *FT,
319                        const std::vector<GenericValue> &Args) {
320  TheInterpreter->exitCalled(Args[0]);
321  return GenericValue();
322}
323
324// void abort(void)
325GenericValue lle_X_abort(const FunctionType *FT,
326                         const std::vector<GenericValue> &Args) {
327  //FIXME: should we report or raise here?
328  //report_fatal_error("Interpreted program raised SIGABRT");
329  raise (SIGABRT);
330  return GenericValue();
331}
332
333// int sprintf(char *, const char *, ...) - a very rough implementation to make
334// output useful.
335GenericValue lle_X_sprintf(const FunctionType *FT,
336                           const std::vector<GenericValue> &Args) {
337  char *OutputBuffer = (char *)GVTOP(Args[0]);
338  const char *FmtStr = (const char *)GVTOP(Args[1]);
339  unsigned ArgNo = 2;
340
341  // printf should return # chars printed.  This is completely incorrect, but
342  // close enough for now.
343  GenericValue GV;
344  GV.IntVal = APInt(32, strlen(FmtStr));
345  while (1) {
346    switch (*FmtStr) {
347    case 0: return GV;             // Null terminator...
348    default:                       // Normal nonspecial character
349      sprintf(OutputBuffer++, "%c", *FmtStr++);
350      break;
351    case '\\': {                   // Handle escape codes
352      sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
353      FmtStr += 2; OutputBuffer += 2;
354      break;
355    }
356    case '%': {                    // Handle format specifiers
357      char FmtBuf[100] = "", Buffer[1000] = "";
358      char *FB = FmtBuf;
359      *FB++ = *FmtStr++;
360      char Last = *FB++ = *FmtStr++;
361      unsigned HowLong = 0;
362      while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
363             Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
364             Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
365             Last != 'p' && Last != 's' && Last != '%') {
366        if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
367        Last = *FB++ = *FmtStr++;
368      }
369      *FB = 0;
370
371      switch (Last) {
372      case '%':
373        memcpy(Buffer, "%", 2); break;
374      case 'c':
375        sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
376        break;
377      case 'd': case 'i':
378      case 'u': case 'o':
379      case 'x': case 'X':
380        if (HowLong >= 1) {
381          if (HowLong == 1 &&
382              TheInterpreter->getTargetData()->getPointerSizeInBits() == 64 &&
383              sizeof(long) < sizeof(int64_t)) {
384            // Make sure we use %lld with a 64 bit argument because we might be
385            // compiling LLI on a 32 bit compiler.
386            unsigned Size = strlen(FmtBuf);
387            FmtBuf[Size] = FmtBuf[Size-1];
388            FmtBuf[Size+1] = 0;
389            FmtBuf[Size-1] = 'l';
390          }
391          sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
392        } else
393          sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
394        break;
395      case 'e': case 'E': case 'g': case 'G': case 'f':
396        sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
397      case 'p':
398        sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
399      case 's':
400        sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
401      default:
402        errs() << "<unknown printf code '" << *FmtStr << "'!>";
403        ArgNo++; break;
404      }
405      size_t Len = strlen(Buffer);
406      memcpy(OutputBuffer, Buffer, Len + 1);
407      OutputBuffer += Len;
408      }
409      break;
410    }
411  }
412  return GV;
413}
414
415// int printf(const char *, ...) - a very rough implementation to make output
416// useful.
417GenericValue lle_X_printf(const FunctionType *FT,
418                          const std::vector<GenericValue> &Args) {
419  char Buffer[10000];
420  std::vector<GenericValue> NewArgs;
421  NewArgs.push_back(PTOGV((void*)&Buffer[0]));
422  NewArgs.insert(NewArgs.end(), Args.begin(), Args.end());
423  GenericValue GV = lle_X_sprintf(FT, NewArgs);
424  outs() << Buffer;
425  return GV;
426}
427
428// int sscanf(const char *format, ...);
429GenericValue lle_X_sscanf(const FunctionType *FT,
430                          const std::vector<GenericValue> &args) {
431  assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
432
433  char *Args[10];
434  for (unsigned i = 0; i < args.size(); ++i)
435    Args[i] = (char*)GVTOP(args[i]);
436
437  GenericValue GV;
438  GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
439                        Args[5], Args[6], Args[7], Args[8], Args[9]));
440  return GV;
441}
442
443// int scanf(const char *format, ...);
444GenericValue lle_X_scanf(const FunctionType *FT,
445                         const std::vector<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.
460GenericValue lle_X_fprintf(const FunctionType *FT,
461                           const std::vector<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
473} // End extern "C"
474
475// Done with externals; turn the warning back on
476#ifdef _MSC_VER
477    #pragma warning(default: 4190)
478#endif
479
480
481void Interpreter::initializeExternalFunctions() {
482  sys::ScopedLock Writer(*FunctionsLock);
483  FuncNames["lle_X_atexit"]       = lle_X_atexit;
484  FuncNames["lle_X_exit"]         = lle_X_exit;
485  FuncNames["lle_X_abort"]        = lle_X_abort;
486
487  FuncNames["lle_X_printf"]       = lle_X_printf;
488  FuncNames["lle_X_sprintf"]      = lle_X_sprintf;
489  FuncNames["lle_X_sscanf"]       = lle_X_sscanf;
490  FuncNames["lle_X_scanf"]        = lle_X_scanf;
491  FuncNames["lle_X_fprintf"]      = lle_X_fprintf;
492}
493