llvm/lib/ExecutionEngine/Interpreter/ExternalFunctions.cpp

   1 //===-- ExternalFunctions.cpp - Implement External Functions --------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 //  This file contains both code to deal with invoking "external" functions, but
  10 //  also contains code that implements "exported" external functions.
  11 //
  12 //  There are currently two mechanisms for handling external functions in the
  13 //  Interpreter.  The first is to implement lle_* wrapper functions that are
  14 //  specific to well-known library functions which manually translate the
  15 //  arguments from GenericValues and make the call.  If such a wrapper does
  16 //  not exist, and libffi is available, then the Interpreter will attempt to
  17 //  invoke the function using libffi, after finding its address.
  18 //
  19 //===----------------------------------------------------------------------===//
  20
  21 #include "Interpreter.h"
  22 #include "llvm/ADT/APInt.h"
  23 #include "llvm/ADT/ArrayRef.h"
  24 #include "llvm/Config/config.h" // Detect libffi
  25 #include "llvm/ExecutionEngine/GenericValue.h"
  26 #include "llvm/IR/DataLayout.h"
  27 #include "llvm/IR/DerivedTypes.h"
  28 #include "llvm/IR/Function.h"
  29 #include "llvm/IR/Type.h"
  30 #include "llvm/Support/Casting.h"
  31 #include "llvm/Support/DynamicLibrary.h"
  32 #include "llvm/Support/ErrorHandling.h"
  33 #include "llvm/Support/Mutex.h"
  34 #include "llvm/Support/raw_ostream.h"
  35 #include <cassert>
  36 #include <cmath>
  37 #include <csignal>
  38 #include <cstdint>
  39 #include <cstdio>
  40 #include <cstring>
  41 #include <map>
  42 #include <mutex>
  43 #include <string>
  44 #include <utility>
  45 #include <vector>
  46
  47 #ifdef HAVE_FFI_CALL
  48 #ifdef HAVE_FFI_H
  49 #include <ffi.h>
  50 #define USE_LIBFFI
  51 #elif HAVE_FFI_FFI_H
  52 #include <ffi/ffi.h>
  53 #define USE_LIBFFI
  54 #endif
  55 #endif
  56
  57 using namespace llvm;
  58
  59 namespace {
  60
  61 typedef GenericValue (*ExFunc)(FunctionType *, ArrayRef<GenericValue>);
  62 typedef void (*RawFunc)();
  63
  64 struct Functions {
  65   sys::Mutex Lock;
  66   std::map<const Function *, ExFunc> ExportedFunctions;
  67   std::map<std::string, ExFunc> FuncNames;
  68 #ifdef USE_LIBFFI
  69   std::map<const Function *, RawFunc> RawFunctions;
  70 #endif
  71 };
  72
  73 Functions &getFunctions() {
  74   static Functions F;
  75   return F;
  76 }
  77
  78 } // anonymous namespace
  79
  80 static Interpreter *TheInterpreter;
  81
  82 static char getTypeID(Type *Ty) {
  83   switch (Ty->getTypeID()) {
  84   case Type::VoidTyID:    return 'V';
  85   case Type::IntegerTyID:
  86     switch (cast<IntegerType>(Ty)->getBitWidth()) {
  87       case 1:  return 'o';
  88       case 8:  return 'B';
  89       case 16: return 'S';
  90       case 32: return 'I';
  91       case 64: return 'L';
  92       default: return 'N';
  93     }
  94   case Type::FloatTyID:   return 'F';
  95   case Type::DoubleTyID:  return 'D';
  96   case Type::PointerTyID: return 'P';
  97   case Type::FunctionTyID:return 'M';
  98   case Type::StructTyID:  return 'T';
  99   case Type::ArrayTyID:   return 'A';
 100   default: return 'U';
 101   }
 102 }
 103
 104 // Try to find address of external function given a Function object.
 105 // Please note, that interpreter doesn't know how to assemble a
 106 // real call in general case (this is JIT job), that's why it assumes,
 107 // that all external functions has the same (and pretty "general") signature.
 108 // The typical example of such functions are "lle_X_" ones.
 109 static ExFunc lookupFunction(const Function *F) {
 110   // Function not found, look it up... start by figuring out what the
 111   // composite function name should be.
 112   std::string ExtName = "lle_";
 113   FunctionType *FT = F->getFunctionType();
 114   ExtName += getTypeID(FT->getReturnType());
 115   for (Type *T : FT->params())
 116     ExtName += getTypeID(T);
 117   ExtName += ("_" + F->getName()).str();
 118
 119   auto &Fns = getFunctions();
 120   sys::ScopedLock Writer(Fns.Lock);
 121   ExFunc FnPtr = Fns.FuncNames[ExtName];
 122   if (!FnPtr)
 123     FnPtr = Fns.FuncNames[("lle_X_" + F->getName()).str()];
 124   if (!FnPtr)  // Try calling a generic function... if it exists...
 125     FnPtr = (ExFunc)(intptr_t)sys::DynamicLibrary::SearchForAddressOfSymbol(
 126         ("lle_X_" + F->getName()).str());
 127   if (FnPtr)
 128     Fns.ExportedFunctions.insert(std::make_pair(F, FnPtr)); // Cache for later
 129   return FnPtr;
 130 }
 131
 132 #ifdef USE_LIBFFI
 133 static ffi_type *ffiTypeFor(Type *Ty) {
 134   switch (Ty->getTypeID()) {
 135     case Type::VoidTyID: return &ffi_type_void;
 136     case Type::IntegerTyID:
 137       switch (cast<IntegerType>(Ty)->getBitWidth()) {
 138         case 8:  return &ffi_type_sint8;
 139         case 16: return &ffi_type_sint16;
 140         case 32: return &ffi_type_sint32;
 141         case 64: return &ffi_type_sint64;
 142       }
 143       llvm_unreachable("Unhandled integer type bitwidth");
 144     case Type::FloatTyID:   return &ffi_type_float;
 145     case Type::DoubleTyID:  return &ffi_type_double;
 146     case Type::PointerTyID: return &ffi_type_pointer;
 147     default: break;
 148   }
 149   // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
 150   report_fatal_error("Type could not be mapped for use with libffi.");
 151   return NULL;
 152 }
 153
 154 static void *ffiValueFor(Type *Ty, const GenericValue &AV,
 155                          void *ArgDataPtr) {
 156   switch (Ty->getTypeID()) {
 157     case Type::IntegerTyID:
 158       switch (cast<IntegerType>(Ty)->getBitWidth()) {
 159         case 8: {
 160           int8_t *I8Ptr = (int8_t *) ArgDataPtr;
 161           *I8Ptr = (int8_t) AV.IntVal.getZExtValue();
 162           return ArgDataPtr;
 163         }
 164         case 16: {
 165           int16_t *I16Ptr = (int16_t *) ArgDataPtr;
 166           *I16Ptr = (int16_t) AV.IntVal.getZExtValue();
 167           return ArgDataPtr;
 168         }
 169         case 32: {
 170           int32_t *I32Ptr = (int32_t *) ArgDataPtr;
 171           *I32Ptr = (int32_t) AV.IntVal.getZExtValue();
 172           return ArgDataPtr;
 173         }
 174         case 64: {
 175           int64_t *I64Ptr = (int64_t *) ArgDataPtr;
 176           *I64Ptr = (int64_t) AV.IntVal.getZExtValue();
 177           return ArgDataPtr;
 178         }
 179       }
 180       llvm_unreachable("Unhandled integer type bitwidth");
 181     case Type::FloatTyID: {
 182       float *FloatPtr = (float *) ArgDataPtr;
 183       *FloatPtr = AV.FloatVal;
 184       return ArgDataPtr;
 185     }
 186     case Type::DoubleTyID: {
 187       double *DoublePtr = (double *) ArgDataPtr;
 188       *DoublePtr = AV.DoubleVal;
 189       return ArgDataPtr;
 190     }
 191     case Type::PointerTyID: {
 192       void **PtrPtr = (void **) ArgDataPtr;
 193       *PtrPtr = GVTOP(AV);
 194       return ArgDataPtr;
 195     }
 196     default: break;
 197   }
 198   // TODO: Support other types such as StructTyID, ArrayTyID, OpaqueTyID, etc.
 199   report_fatal_error("Type value could not be mapped for use with libffi.");
 200   return NULL;
 201 }
 202
 203 static bool ffiInvoke(RawFunc Fn, Function *F, ArrayRef<GenericValue> ArgVals,
 204                       const DataLayout &TD, GenericValue &Result) {
 205   ffi_cif cif;
 206   FunctionType *FTy = F->getFunctionType();
 207   const unsigned NumArgs = F->arg_size();
 208
 209   // TODO: We don't have type information about the remaining arguments, because
 210   // this information is never passed into ExecutionEngine::runFunction().
 211   if (ArgVals.size() > NumArgs && F->isVarArg()) {
 212     report_fatal_error("Calling external var arg function '" + F->getName()
 213                       + "' is not supported by the Interpreter.");
 214   }
 215
 216   unsigned ArgBytes = 0;
 217
 218   std::vector<ffi_type*> args(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     args[ArgNo] = ffiTypeFor(ArgTy);
 224     ArgBytes += TD.getTypeStoreSize(ArgTy);
 225   }
 226
 227   SmallVector<uint8_t, 128> ArgData;
 228   ArgData.resize(ArgBytes);
 229   uint8_t *ArgDataPtr = ArgData.data();
 230   SmallVector<void*, 16> values(NumArgs);
 231   for (Function::const_arg_iterator A = F->arg_begin(), E = F->arg_end();
 232        A != E; ++A) {
 233     const unsigned ArgNo = A->getArgNo();
 234     Type *ArgTy = FTy->getParamType(ArgNo);
 235     values[ArgNo] = ffiValueFor(ArgTy, ArgVals[ArgNo], ArgDataPtr);
 236     ArgDataPtr += TD.getTypeStoreSize(ArgTy);
 237   }
 238
 239   Type *RetTy = FTy->getReturnType();
 240   ffi_type *rtype = ffiTypeFor(RetTy);
 241
 242   if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, NumArgs, rtype, args.data()) ==
 243       FFI_OK) {
 244     SmallVector<uint8_t, 128> ret;
 245     if (RetTy->getTypeID() != Type::VoidTyID)
 246       ret.resize(TD.getTypeStoreSize(RetTy));
 247     ffi_call(&cif, Fn, ret.data(), values.data());
 248     switch (RetTy->getTypeID()) {
 249       case Type::IntegerTyID:
 250         switch (cast<IntegerType>(RetTy)->getBitWidth()) {
 251           case 8:  Result.IntVal = APInt(8 , *(int8_t *) ret.data()); break;
 252           case 16: Result.IntVal = APInt(16, *(int16_t*) ret.data()); break;
 253           case 32: Result.IntVal = APInt(32, *(int32_t*) ret.data()); break;
 254           case 64: Result.IntVal = APInt(64, *(int64_t*) ret.data()); break;
 255         }
 256         break;
 257       case Type::FloatTyID:   Result.FloatVal   = *(float *) ret.data(); break;
 258       case Type::DoubleTyID:  Result.DoubleVal  = *(double*) ret.data(); break;
 259       case Type::PointerTyID: Result.PointerVal = *(void **) ret.data(); break;
 260       default: break;
 261     }
 262     return true;
 263   }
 264
 265   return false;
 266 }
 267 #endif // USE_LIBFFI
 268
 269 GenericValue Interpreter::callExternalFunction(Function *F,
 270                                                ArrayRef<GenericValue> ArgVals) {
 271   TheInterpreter = this;
 272
 273   auto &Fns = getFunctions();
 274   std::unique_lock<sys::Mutex> Guard(Fns.Lock);
 275
 276   // Do a lookup to see if the function is in our cache... this should just be a
 277   // deferred annotation!
 278   std::map<const Function *, ExFunc>::iterator FI =
 279       Fns.ExportedFunctions.find(F);
 280   if (ExFunc Fn = (FI == Fns.ExportedFunctions.end()) ? lookupFunction(F)
 281                                                       : FI->second) {
 282     Guard.unlock();
 283     return Fn(F->getFunctionType(), ArgVals);
 284   }
 285
 286 #ifdef USE_LIBFFI
 287   std::map<const Function *, RawFunc>::iterator RF = Fns.RawFunctions.find(F);
 288   RawFunc RawFn;
 289   if (RF == Fns.RawFunctions.end()) {
 290     RawFn = (RawFunc)(intptr_t)
 291       sys::DynamicLibrary::SearchForAddressOfSymbol(std::string(F->getName()));
 292     if (!RawFn)
 293       RawFn = (RawFunc)(intptr_t)getPointerToGlobalIfAvailable(F);
 294     if (RawFn != 0)
 295       Fns.RawFunctions.insert(std::make_pair(F, RawFn)); // Cache for later
 296   } else {
 297     RawFn = RF->second;
 298   }
 299
 300   Guard.unlock();
 301
 302   GenericValue Result;
 303   if (RawFn != 0 && ffiInvoke(RawFn, F, ArgVals, getDataLayout(), Result))
 304     return Result;
 305 #endif // USE_LIBFFI
 306
 307   if (F->getName() == "__main")
 308     errs() << "Tried to execute an unknown external function: "
 309       << *F->getType() << " __main\n";
 310   else
 311     report_fatal_error("Tried to execute an unknown external function: " +
 312                        F->getName());
 313 #ifndef USE_LIBFFI
 314   errs() << "Recompiling LLVM with --enable-libffi might help.\n";
 315 #endif
 316   return GenericValue();
 317 }
 318
 319 //===----------------------------------------------------------------------===//
 320 //  Functions "exported" to the running application...
 321 //
 322
 323 // void atexit(Function*)
 324 static GenericValue lle_X_atexit(FunctionType *FT,
 325                                  ArrayRef<GenericValue> Args) {
 326   assert(Args.size() == 1);
 327   TheInterpreter->addAtExitHandler((Function*)GVTOP(Args[0]));
 328   GenericValue GV;
 329   GV.IntVal = 0;
 330   return GV;
 331 }
 332
 333 // void exit(int)
 334 static GenericValue lle_X_exit(FunctionType *FT, ArrayRef<GenericValue> Args) {
 335   TheInterpreter->exitCalled(Args[0]);
 336   return GenericValue();
 337 }
 338
 339 // void abort(void)
 340 static GenericValue lle_X_abort(FunctionType *FT, ArrayRef<GenericValue> Args) {
 341   //FIXME: should we report or raise here?
 342   //report_fatal_error("Interpreted program raised SIGABRT");
 343   raise (SIGABRT);
 344   return GenericValue();
 345 }
 346
 347 // Silence warnings about sprintf. (See also
 348 // https://github.com/llvm/llvm-project/issues/58086)
 349 #if defined(__clang__)
 350 #pragma clang diagnostic push
 351 #pragma clang diagnostic ignored "-Wdeprecated-declarations"
 352 #endif
 353 // int sprintf(char *, const char *, ...) - a very rough implementation to make
 354 // output useful.
 355 static GenericValue lle_X_sprintf(FunctionType *FT,
 356                                   ArrayRef<GenericValue> Args) {
 357   char *OutputBuffer = (char *)GVTOP(Args[0]);
 358   const char *FmtStr = (const char *)GVTOP(Args[1]);
 359   unsigned ArgNo = 2;
 360
 361   // printf should return # chars printed.  This is completely incorrect, but
 362   // close enough for now.
 363   GenericValue GV;
 364   GV.IntVal = APInt(32, strlen(FmtStr));
 365   while (true) {
 366     switch (*FmtStr) {
 367     case 0: return GV;             // Null terminator...
 368     default:                       // Normal nonspecial character
 369       sprintf(OutputBuffer++, "%c", *FmtStr++);
 370       break;
 371     case '\\': {                   // Handle escape codes
 372       sprintf(OutputBuffer, "%c%c", *FmtStr, *(FmtStr+1));
 373       FmtStr += 2; OutputBuffer += 2;
 374       break;
 375     }
 376     case '%': {                    // Handle format specifiers
 377       char FmtBuf[100] = "", Buffer[1000] = "";
 378       char *FB = FmtBuf;
 379       *FB++ = *FmtStr++;
 380       char Last = *FB++ = *FmtStr++;
 381       unsigned HowLong = 0;
 382       while (Last != 'c' && Last != 'd' && Last != 'i' && Last != 'u' &&
 383              Last != 'o' && Last != 'x' && Last != 'X' && Last != 'e' &&
 384              Last != 'E' && Last != 'g' && Last != 'G' && Last != 'f' &&
 385              Last != 'p' && Last != 's' && Last != '%') {
 386         if (Last == 'l' || Last == 'L') HowLong++;  // Keep track of l's
 387         Last = *FB++ = *FmtStr++;
 388       }
 389       *FB = 0;
 390
 391       switch (Last) {
 392       case '%':
 393         memcpy(Buffer, "%", 2); break;
 394       case 'c':
 395         sprintf(Buffer, FmtBuf, uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
 396         break;
 397       case 'd': case 'i':
 398       case 'u': case 'o':
 399       case 'x': case 'X':
 400         if (HowLong >= 1) {
 401           if (HowLong == 1 &&
 402               TheInterpreter->getDataLayout().getPointerSizeInBits() == 64 &&
 403               sizeof(long) < sizeof(int64_t)) {
 404             // Make sure we use %lld with a 64 bit argument because we might be
 405             // compiling LLI on a 32 bit compiler.
 406             unsigned Size = strlen(FmtBuf);
 407             FmtBuf[Size] = FmtBuf[Size-1];
 408             FmtBuf[Size+1] = 0;
 409             FmtBuf[Size-1] = 'l';
 410           }
 411           sprintf(Buffer, FmtBuf, Args[ArgNo++].IntVal.getZExtValue());
 412         } else
 413           sprintf(Buffer, FmtBuf,uint32_t(Args[ArgNo++].IntVal.getZExtValue()));
 414         break;
 415       case 'e': case 'E': case 'g': case 'G': case 'f':
 416         sprintf(Buffer, FmtBuf, Args[ArgNo++].DoubleVal); break;
 417       case 'p':
 418         sprintf(Buffer, FmtBuf, (void*)GVTOP(Args[ArgNo++])); break;
 419       case 's':
 420         sprintf(Buffer, FmtBuf, (char*)GVTOP(Args[ArgNo++])); break;
 421       default:
 422         errs() << "<unknown printf code '" << *FmtStr << "'!>";
 423         ArgNo++; break;
 424       }
 425       size_t Len = strlen(Buffer);
 426       memcpy(OutputBuffer, Buffer, Len + 1);
 427       OutputBuffer += Len;
 428       }
 429       break;
 430     }
 431   }
 432   return GV;
 433 }
 434 #if defined(__clang__)
 435 #pragma clang diagnostic pop
 436 #endif
 437
 438 // int printf(const char *, ...) - a very rough implementation to make output
 439 // useful.
 440 static GenericValue lle_X_printf(FunctionType *FT,
 441                                  ArrayRef<GenericValue> Args) {
 442   char Buffer[10000];
 443   std::vector<GenericValue> NewArgs;
 444   NewArgs.push_back(PTOGV((void*)&Buffer[0]));
 445   llvm::append_range(NewArgs, Args);
 446   GenericValue GV = lle_X_sprintf(FT, NewArgs);
 447   outs() << Buffer;
 448   return GV;
 449 }
 450
 451 // int sscanf(const char *format, ...);
 452 static GenericValue lle_X_sscanf(FunctionType *FT,
 453                                  ArrayRef<GenericValue> args) {
 454   assert(args.size() < 10 && "Only handle up to 10 args to sscanf right now!");
 455
 456   char *Args[10];
 457   for (unsigned i = 0; i < args.size(); ++i)
 458     Args[i] = (char*)GVTOP(args[i]);
 459
 460   GenericValue GV;
 461   GV.IntVal = APInt(32, sscanf(Args[0], Args[1], Args[2], Args[3], Args[4],
 462                     Args[5], Args[6], Args[7], Args[8], Args[9]));
 463   return GV;
 464 }
 465
 466 // int scanf(const char *format, ...);
 467 static GenericValue lle_X_scanf(FunctionType *FT, ArrayRef<GenericValue> args) {
 468   assert(args.size() < 10 && "Only handle up to 10 args to scanf right now!");
 469
 470   char *Args[10];
 471   for (unsigned i = 0; i < args.size(); ++i)
 472     Args[i] = (char*)GVTOP(args[i]);
 473
 474   GenericValue GV;
 475   GV.IntVal = APInt(32, scanf( Args[0], Args[1], Args[2], Args[3], Args[4],
 476                     Args[5], Args[6], Args[7], Args[8], Args[9]));
 477   return GV;
 478 }
 479
 480 // int fprintf(FILE *, const char *, ...) - a very rough implementation to make
 481 // output useful.
 482 static GenericValue lle_X_fprintf(FunctionType *FT,
 483                                   ArrayRef<GenericValue> Args) {
 484   assert(Args.size() >= 2);
 485   char Buffer[10000];
 486   std::vector<GenericValue> NewArgs;
 487   NewArgs.push_back(PTOGV(Buffer));
 488   NewArgs.insert(NewArgs.end(), Args.begin()+1, Args.end());
 489   GenericValue GV = lle_X_sprintf(FT, NewArgs);
 490
 491   fputs(Buffer, (FILE *) GVTOP(Args[0]));
 492   return GV;
 493 }
 494
 495 static GenericValue lle_X_memset(FunctionType *FT,
 496                                  ArrayRef<GenericValue> Args) {
 497   int val = (int)Args[1].IntVal.getSExtValue();
 498   size_t len = (size_t)Args[2].IntVal.getZExtValue();
 499   memset((void *)GVTOP(Args[0]), val, len);
 500   // llvm.memset.* returns void, lle_X_* returns GenericValue,
 501   // so here we return GenericValue with IntVal set to zero
 502   GenericValue GV;
 503   GV.IntVal = 0;
 504   return GV;
 505 }
 506
 507 static GenericValue lle_X_memcpy(FunctionType *FT,
 508                                  ArrayRef<GenericValue> Args) {
 509   memcpy(GVTOP(Args[0]), GVTOP(Args[1]),
 510          (size_t)(Args[2].IntVal.getLimitedValue()));
 511
 512   // llvm.memcpy* returns void, lle_X_* returns GenericValue,
 513   // so here we return GenericValue with IntVal set to zero
 514   GenericValue GV;
 515   GV.IntVal = 0;
 516   return GV;
 517 }
 518
 519 void Interpreter::initializeExternalFunctions() {
 520   auto &Fns = getFunctions();
 521   sys::ScopedLock Writer(Fns.Lock);
 522   Fns.FuncNames["lle_X_atexit"]       = lle_X_atexit;
 523   Fns.FuncNames["lle_X_exit"]         = lle_X_exit;
 524   Fns.FuncNames["lle_X_abort"]        = lle_X_abort;
 525
 526   Fns.FuncNames["lle_X_printf"]       = lle_X_printf;
 527   Fns.FuncNames["lle_X_sprintf"]      = lle_X_sprintf;
 528   Fns.FuncNames["lle_X_sscanf"]       = lle_X_sscanf;
 529   Fns.FuncNames["lle_X_scanf"]        = lle_X_scanf;
 530   Fns.FuncNames["lle_X_fprintf"]      = lle_X_fprintf;
 531   Fns.FuncNames["lle_X_memset"]       = lle_X_memset;
 532   Fns.FuncNames["lle_X_memcpy"]       = lle_X_memcpy;
 533 }