ExternalFunctions.cpp revision 194710 
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/Streams.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)(void);
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->getName();
99
100  sys::ScopedLock Writer(&*FunctionsLock);
101  ExFunc FnPtr = FuncNames[ExtName];
102  if (FnPtr == 0)
103    FnPtr = FuncNames["lle_X_"+F->getName()];
104  if (FnPtr == 0)  // Try calling a generic function... if it exists...
105    FnPtr = (ExFunc)(intptr_t)sys::DynamicLibrary::SearchForAddressOfSymbol(
106            ("lle_X_"+F->getName()).c_str());
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  cerr << "Type could not be mapped for use with libffi.\n";
130  abort();
131  return NULL;
132}
133
134static void *ffiValueFor(const Type *Ty, const GenericValue &AV,
135                         void *ArgDataPtr) {
136  switch (Ty->getTypeID()) {
137    case Type::IntegerTyID:
138      switch (cast<IntegerType>(Ty)->getBitWidth()) {
139        case 8: {
140          int8_t *I8Ptr = (int8_t *) ArgDataPtr;
141          *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
142          return ArgDataPtr;
143        }
144        case 16: {
145          int16_t *I16Ptr = (int16_t *) ArgDataPtr;
146          *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
147          return ArgDataPtr;
148        }
149        case 32: {
150          int32_t *I32Ptr = (int32_t *) ArgDataPtr;
151          *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
152          return ArgDataPtr;
153        }
154        case 64: {
155          int64_t *I64Ptr = (int64_t *) ArgDataPtr;
156          *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
157          return ArgDataPtr;
158        }
159      }
160    case Type::FloatTyID: {
161      float *FloatPtr = (float *) ArgDataPtr;
162      *FloatPtr = AV.DoubleVal;
163      return ArgDataPtr;
164    }
165    case Type::DoubleTyID: {
166      double *DoublePtr = (double *) ArgDataPtr;
167      *DoublePtr = AV.DoubleVal;
168      return ArgDataPtr;
169    }
170    case Type::PointerTyID: {
171      void **PtrPtr = (void **) ArgDataPtr;
172      *PtrPtr = GVTOP(AV);
173      return ArgDataPtr;
174    }
175    default: break;
176  }
177  // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
178  cerr << "Type value could not be mapped for use with libffi.\n";
179  abort();
180  return NULL;
181}
182
183static bool ffiInvoke(RawFunc Fn, Function *F,
184                      const std::vector<GenericValue> &ArgVals,
185                      const TargetData *TD, GenericValue &Result) {
186  ffi_cif cif;
187  const FunctionType *FTy = F->getFunctionType();
188  const unsigned NumArgs = F->arg_size();
189
190  // TODO: We don't have type information about the remaining arguments, because
191  // this information is never passed into ExecutionEngine::runFunction().
192  if (ArgVals.size() > NumArgs && F->isVarArg()) {
193    cerr << "Calling external var arg function '" << F->getName()
194         << "' is not supported by the Interpreter.\n";
195    abort();
196  }
197
198  unsigned ArgBytes = 0;
199
200  std::vector<ffi_type*> args(NumArgs);
201  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
202       A != E; ++A) {
203    const unsigned ArgNo = A->getArgNo();
204    const Type *ArgTy = FTy->getParamType(ArgNo);
205    args[ArgNo] = ffiTypeFor(ArgTy);
206    ArgBytes += TD->getTypeStoreSize(ArgTy);
207  }
208
209  uint8_t *ArgData = (uint8_t*) alloca(ArgBytes);
210  uint8_t *ArgDataPtr = ArgData;
211  std::vector<void*> values(NumArgs);
212  for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
213       A != E; ++A) {
214    const unsigned ArgNo = A->getArgNo();
215    const Type *ArgTy = FTy->getParamType(ArgNo);
216    values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
217    ArgDataPtr += TD->getTypeStoreSize(ArgTy);
218  }
219
220  const Type *RetTy = FTy->getReturnType();
221  ffi_type *rtype = ffiTypeFor(RetTy);
222
223  if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, &args[0]) == FFI_OK) {
224    void *ret = NULL;
225    if (RetTy->getTypeID() != Type::VoidTyID)
226      ret = alloca(TD->getTypeStoreSize(RetTy));
227    ffi_call(&cif, Fn, ret, &values[0]);
228    switch (RetTy->getTypeID()) {
229      case Type::IntegerTyID:
230        switch (cast<IntegerType>(RetTy)->getBitWidth()) {
231          case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret); break;
232          case 16: Result.IntVal = APInt(16, *(int16_t*) ret); break;
233          case 32: Result.IntVal = APInt(32, *(int32_t*) ret); break;
234          case 64: Result.IntVal = APInt(64, *(int64_t*) ret); break;
235        }
236        break;
237      case Type::FloatTyID:   Result.FloatVal   = *(float *) ret; break;
238      case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret; break;
239      case Type::PointerTyID: Result.PointerVal = *(void **) ret; break;
240      default: break;
241    }
242    return true;
243  }
244
245  return false;
246}
247#endif // USE_LIBFFI
248
249GenericValue Interpreter::callExternalFunction(Function *F,
250                                     const std::vector<GenericValue> &ArgVals) {
251  TheInterpreter = this;
252
253  FunctionsLock->acquire();
254
255  // Do a lookup to see if the function is in our cache... this should just be a
256  // deferred annotation!
257  std::map<const Function *, ExFunc>::iterator FI = ExportedFunctions->find(F);
258  if (ExFunc Fn = (FI == ExportedFunctions->end()) ? lookupFunction(F)
259                                                   : FI->second) {
260    FunctionsLock->release();
261    return Fn(F->getFunctionType(), ArgVals);
262  }
263
264#ifdef USE_LIBFFI
265  std::map<const Function *, RawFunc>::iterator RF = RawFunctions->find(F);
266  RawFunc RawFn;
267  if (RF == RawFunctions->end()) {
268    RawFn = (RawFunc)(intptr_t)
269      sys::DynamicLibrary::SearchForAddressOfSymbol(F->getName());
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  cerr << "Tried to execute an unknown external function: "
284       << F->getType()->getDescription() << " " << F->getName() << "\n";
285  if (F->getName() != "__main")
286    abort();
287  return GenericValue();
288}
289
290
291//===----------------------------------------------------------------------===//
292//  Functions "exported" to the running application...
293//
294extern "C" {  // Don't add C++ manglings to llvm mangling :)
295
296// void atexit(Function*)
297GenericValue lle_X_atexit(const FunctionType *FT,
298                          const std::vector<GenericValue> &Args) {
299  assert(Args.size() == 1);
300  TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
301  GenericValue GV;
302  GV.IntVal = 0;
303  return GV;
304}
305
306// void exit(int)
307GenericValue lle_X_exit(const FunctionType *FT,
308                        const std::vector<GenericValue> &Args) {
309  TheInterpreter->exitCalled(Args[0]);
310  return GenericValue();
311}
312
313// void abort(void)
314GenericValue lle_X_abort(const FunctionType *FT,
315                         const std::vector<GenericValue> &Args) {
316  raise (SIGABRT);
317  return GenericValue();
318}
319
320// int sprintf(char *, const char *, ...) - a very rough implementation to make
321// output useful.
322GenericValue lle_X_sprintf(const FunctionType *FT,
323                           const std::vector<GenericValue> &Args) {
324  char *OutputBuffer = (char *)GVTOP(Args[0]);
325  const char *FmtStr = (const char *)GVTOP(Args[1]);
326  unsigned ArgNo = 2;
327
328  // printf should return # chars printed.  This is completely incorrect, but
329  // close enough for now.
330  GenericValue GV;
331  GV.IntVal = APInt(32, strlen(FmtStr));
332  while (1) {
333    switch (*FmtStr) {
334    case 0: return GV;             // Null terminator...
335    default:                       // Normal nonspecial character
336      sprintf(OutputBuffer++, "%c", *FmtStr++);
337      break;
338    case '\\': {                   // Handle escape codes
339      sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
340      FmtStr += 2; OutputBuffer += 2;
341      break;
342    }
343    case '%': {                    // Handle format specifiers
344      char FmtBuf[100] = "", Buffer[1000] = "";
345      char *FB = FmtBuf;
346      *FB++ = *FmtStr++;
347      char Last = *FB++ = *FmtStr++;
348      unsigned HowLong = 0;
349      while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
350             Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
351             Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
352             Last != 'p' && Last != 's' && Last != '%') {
353        if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
354        Last = *FB++ = *FmtStr++;
355      }
356      *FB = 0;
357
358      switch (Last) {
359      case '%':
360        strcpy(Buffer, "%"); break;
361      case 'c':
362        sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
363        break;
364      case 'd': case 'i':
365      case 'u': case 'o':
366      case 'x': case 'X':
367        if (HowLong >= 1) {
368          if (HowLong == 1 &&
369              TheInterpreter->getTargetData()->getPointerSizeInBits() == 64 &&
370              sizeof(long) < sizeof(int64_t)) {
371            // Make sure we use %lld with a 64 bit argument because we might be
372            // compiling LLI on a 32 bit compiler.
373            unsigned Size = strlen(FmtBuf);
374            FmtBuf[Size] = FmtBuf[Size-1];
375            FmtBuf[Size+1] = 0;
376            FmtBuf[Size-1] = 'l';
377          }
378          sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
379        } else
380          sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
381        break;
382      case 'e': case 'E': case 'g': case 'G': case 'f':
383        sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
384      case 'p':
385        sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
386      case 's':
387        sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
388      default:  cerr << "<unknown printf code '" << *FmtStr << "'!>";
389        ArgNo++; break;
390      }
391      strcpy(OutputBuffer, Buffer);
392      OutputBuffer += strlen(Buffer);
393      }
394      break;
395    }
396  }
397  return GV;
398}
399
400// int printf(const char *, ...) - a very rough implementation to make output
401// useful.
402GenericValue lle_X_printf(const FunctionType *FT,
403                          const std::vector<GenericValue> &Args) {
404  char Buffer[10000];
405  std::vector<GenericValue> NewArgs;
406  NewArgs.push_back(PTOGV((void*)&Buffer[0]));
407  NewArgs.insert(NewArgs.end(), Args.begin(), Args.end());
408  GenericValue GV = lle_X_sprintf(FT, NewArgs);
409  cout << Buffer;
410  return GV;
411}
412
413static void ByteswapSCANFResults(const char *Fmt, void *Arg0, void *Arg1,
414                                 void *Arg2, void *Arg3, void *Arg4, void *Arg5,
415                                 void *Arg6, void *Arg7, void *Arg8) {
416  void *Args[] = { Arg0, Arg1, Arg2, Arg3, Arg4, Arg5, Arg6, Arg7, Arg8, 0 };
417
418  // Loop over the format string, munging read values as appropriate (performs
419  // byteswaps as necessary).
420  unsigned ArgNo = 0;
421  while (*Fmt) {
422    if (*Fmt++ == '%') {
423      // Read any flag characters that may be present...
424      bool Suppress = false;
425      bool Half = false;
426      bool Long = false;
427      bool LongLong = false;  // long long or long double
428
429      while (1) {
430        switch (*Fmt++) {
431        case '*': Suppress = true; break;
432        case 'a': /*Allocate = true;*/ break;  // We don't need to track this
433        case 'h': Half = true; break;
434        case 'l': Long = true; break;
435        case 'q':
436        case 'L': LongLong = true; break;
437        default:
438          if (Fmt[-1] > '9' || Fmt[-1] < '0')   // Ignore field width specs
439            goto Out;
440        }
441      }
442    Out:
443
444      // Read the conversion character
445      if (!Suppress && Fmt[-1] != '%') { // Nothing to do?
446        unsigned Size = 0;
447        const Type *Ty = 0;
448
449        switch (Fmt[-1]) {
450        case 'i': case 'o': case 'u': case 'x': case 'X': case 'n': case 'p':
451        case 'd':
452          if (Long || LongLong) {
453            Size = 8; Ty = Type::Int64Ty;
454          } else if (Half) {
455            Size = 4; Ty = Type::Int16Ty;
456          } else {
457            Size = 4; Ty = Type::Int32Ty;
458          }
459          break;
460
461        case 'e': case 'g': case 'E':
462        case 'f':
463          if (Long || LongLong) {
464            Size = 8; Ty = Type::DoubleTy;
465          } else {
466            Size = 4; Ty = Type::FloatTy;
467          }
468          break;
469
470        case 's': case 'c': case '[':  // No byteswap needed
471          Size = 1;
472          Ty = Type::Int8Ty;
473          break;
474
475        default: break;
476        }
477
478        if (Size) {
479          GenericValue GV;
480          void *Arg = Args[ArgNo++];
481          memcpy(&GV, Arg, Size);
482          TheInterpreter->StoreValueToMemory(GV, (GenericValue*)Arg, Ty);
483        }
484      }
485    }
486  }
487}
488
489// int sscanf(const char *format, ...);
490GenericValue lle_X_sscanf(const FunctionType *FT,
491                          const std::vector<GenericValue> &args) {
492  assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
493
494  char *Args[10];
495  for (unsigned i = 0; i < args.size(); ++i)
496    Args[i] = (char*)GVTOP(args[i]);
497
498  GenericValue GV;
499  GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
500                        Args[5], Args[6], Args[7], Args[8], Args[9]));
501  ByteswapSCANFResults(Args[1], Args[2], Args[3], Args[4],
502                       Args[5], Args[6], Args[7], Args[8], Args[9], 0);
503  return GV;
504}
505
506// int scanf(const char *format, ...);
507GenericValue lle_X_scanf(const FunctionType *FT,
508                         const std::vector<GenericValue> &args) {
509  assert(args.size() < 10 && "Only handle up to 10 args to scanf right now!");
510
511  char *Args[10];
512  for (unsigned i = 0; i < args.size(); ++i)
513    Args[i] = (char*)GVTOP(args[i]);
514
515  GenericValue GV;
516  GV.IntVal = APInt(32, scanf( Args[0], Args[1], Args[2], Args[3], Args[4],
517                        Args[5], Args[6], Args[7], Args[8], Args[9]));
518  ByteswapSCANFResults(Args[0], Args[1], Args[2], Args[3], Args[4],
519                       Args[5], Args[6], Args[7], Args[8], Args[9]);
520  return GV;
521}
522
523// int fprintf(FILE *, const char *, ...) - a very rough implementation to make
524// output useful.
525GenericValue lle_X_fprintf(const FunctionType *FT,
526                           const std::vector<GenericValue> &Args) {
527  assert(Args.size() >= 2);
528  char Buffer[10000];
529  std::vector<GenericValue> NewArgs;
530  NewArgs.push_back(PTOGV(Buffer));
531  NewArgs.insert(NewArgs.end(), Args.begin()+1, Args.end());
532  GenericValue GV = lle_X_sprintf(FT, NewArgs);
533
534  fputs(Buffer, (FILE *) GVTOP(Args[0]));
535  return GV;
536}
537
538} // End extern "C"
539
540
541void Interpreter::initializeExternalFunctions() {
542  sys::ScopedLock Writer(&*FunctionsLock);
543  FuncNames["lle_X_atexit"]       = lle_X_atexit;
544  FuncNames["lle_X_exit"]         = lle_X_exit;
545  FuncNames["lle_X_abort"]        = lle_X_abort;
546
547  FuncNames["lle_X_printf"]       = lle_X_printf;
548  FuncNames["lle_X_sprintf"]      = lle_X_sprintf;
549  FuncNames["lle_X_sscanf"]       = lle_X_sscanf;
550  FuncNames["lle_X_scanf"]        = lle_X_scanf;
551  FuncNames["lle_X_fprintf"]      = lle_X_fprintf;
552}
553
554