xref: /freebsd-src/contrib/llvm-project/llvm/lib/Target/WebAssembly/WebAssemblyFixFunctionBitcasts.cpp (revision 0fca6ea1d4eea4c934cfff25ac9ee8ad6fe95583)
10b57cec5SDimitry Andric //===-- WebAssemblyFixFunctionBitcasts.cpp - Fix function bitcasts --------===//
20b57cec5SDimitry Andric //
30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
60b57cec5SDimitry Andric //
70b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
80b57cec5SDimitry Andric ///
90b57cec5SDimitry Andric /// \file
100b57cec5SDimitry Andric /// Fix bitcasted functions.
110b57cec5SDimitry Andric ///
120b57cec5SDimitry Andric /// WebAssembly requires caller and callee signatures to match, however in LLVM,
130b57cec5SDimitry Andric /// some amount of slop is vaguely permitted. Detect mismatch by looking for
140b57cec5SDimitry Andric /// bitcasts of functions and rewrite them to use wrapper functions instead.
150b57cec5SDimitry Andric ///
160b57cec5SDimitry Andric /// This doesn't catch all cases, such as when a function's address is taken in
170b57cec5SDimitry Andric /// one place and casted in another, but it works for many common cases.
180b57cec5SDimitry Andric ///
190b57cec5SDimitry Andric /// Note that LLVM already optimizes away function bitcasts in common cases by
200b57cec5SDimitry Andric /// dropping arguments as needed, so this pass only ends up getting used in less
210b57cec5SDimitry Andric /// common cases.
220b57cec5SDimitry Andric ///
230b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
240b57cec5SDimitry Andric 
250b57cec5SDimitry Andric #include "WebAssembly.h"
260b57cec5SDimitry Andric #include "llvm/IR/Constants.h"
270b57cec5SDimitry Andric #include "llvm/IR/Instructions.h"
280b57cec5SDimitry Andric #include "llvm/IR/Module.h"
290b57cec5SDimitry Andric #include "llvm/IR/Operator.h"
300b57cec5SDimitry Andric #include "llvm/Pass.h"
310b57cec5SDimitry Andric #include "llvm/Support/Debug.h"
320b57cec5SDimitry Andric #include "llvm/Support/raw_ostream.h"
330b57cec5SDimitry Andric using namespace llvm;
340b57cec5SDimitry Andric 
350b57cec5SDimitry Andric #define DEBUG_TYPE "wasm-fix-function-bitcasts"
360b57cec5SDimitry Andric 
370b57cec5SDimitry Andric namespace {
380b57cec5SDimitry Andric class FixFunctionBitcasts final : public ModulePass {
390b57cec5SDimitry Andric   StringRef getPassName() const override {
400b57cec5SDimitry Andric     return "WebAssembly Fix Function Bitcasts";
410b57cec5SDimitry Andric   }
420b57cec5SDimitry Andric 
430b57cec5SDimitry Andric   void getAnalysisUsage(AnalysisUsage &AU) const override {
440b57cec5SDimitry Andric     AU.setPreservesCFG();
450b57cec5SDimitry Andric     ModulePass::getAnalysisUsage(AU);
460b57cec5SDimitry Andric   }
470b57cec5SDimitry Andric 
480b57cec5SDimitry Andric   bool runOnModule(Module &M) override;
490b57cec5SDimitry Andric 
500b57cec5SDimitry Andric public:
510b57cec5SDimitry Andric   static char ID;
520b57cec5SDimitry Andric   FixFunctionBitcasts() : ModulePass(ID) {}
530b57cec5SDimitry Andric };
540b57cec5SDimitry Andric } // End anonymous namespace
550b57cec5SDimitry Andric 
560b57cec5SDimitry Andric char FixFunctionBitcasts::ID = 0;
570b57cec5SDimitry Andric INITIALIZE_PASS(FixFunctionBitcasts, DEBUG_TYPE,
580b57cec5SDimitry Andric                 "Fix mismatching bitcasts for WebAssembly", false, false)
590b57cec5SDimitry Andric 
600b57cec5SDimitry Andric ModulePass *llvm::createWebAssemblyFixFunctionBitcasts() {
610b57cec5SDimitry Andric   return new FixFunctionBitcasts();
620b57cec5SDimitry Andric }
630b57cec5SDimitry Andric 
640b57cec5SDimitry Andric // Recursively descend the def-use lists from V to find non-bitcast users of
650b57cec5SDimitry Andric // bitcasts of V.
660b57cec5SDimitry Andric static void findUses(Value *V, Function &F,
67349cc55cSDimitry Andric                      SmallVectorImpl<std::pair<CallBase *, Function *>> &Uses) {
68349cc55cSDimitry Andric   for (User *U : V->users()) {
69349cc55cSDimitry Andric     if (auto *BC = dyn_cast<BitCastOperator>(U))
70349cc55cSDimitry Andric       findUses(BC, F, Uses);
71349cc55cSDimitry Andric     else if (auto *A = dyn_cast<GlobalAlias>(U))
72349cc55cSDimitry Andric       findUses(A, F, Uses);
73349cc55cSDimitry Andric     else if (auto *CB = dyn_cast<CallBase>(U)) {
745ffd83dbSDimitry Andric       Value *Callee = CB->getCalledOperand();
750b57cec5SDimitry Andric       if (Callee != V)
760b57cec5SDimitry Andric         // Skip calls where the function isn't the callee
770b57cec5SDimitry Andric         continue;
78349cc55cSDimitry Andric       if (CB->getFunctionType() == F.getValueType())
79349cc55cSDimitry Andric         // Skip uses that are immediately called
800b57cec5SDimitry Andric         continue;
81349cc55cSDimitry Andric       Uses.push_back(std::make_pair(CB, &F));
820b57cec5SDimitry Andric     }
830b57cec5SDimitry Andric   }
840b57cec5SDimitry Andric }
850b57cec5SDimitry Andric 
860b57cec5SDimitry Andric // Create a wrapper function with type Ty that calls F (which may have a
870b57cec5SDimitry Andric // different type). Attempt to support common bitcasted function idioms:
880b57cec5SDimitry Andric //  - Call with more arguments than needed: arguments are dropped
890b57cec5SDimitry Andric //  - Call with fewer arguments than needed: arguments are filled in with undef
900b57cec5SDimitry Andric //  - Return value is not needed: drop it
910b57cec5SDimitry Andric //  - Return value needed but not present: supply an undef
920b57cec5SDimitry Andric //
930b57cec5SDimitry Andric // If the all the argument types of trivially castable to one another (i.e.
940b57cec5SDimitry Andric // I32 vs pointer type) then we don't create a wrapper at all (return nullptr
950b57cec5SDimitry Andric // instead).
960b57cec5SDimitry Andric //
970b57cec5SDimitry Andric // If there is a type mismatch that we know would result in an invalid wasm
980b57cec5SDimitry Andric // module then generate wrapper that contains unreachable (i.e. abort at
990b57cec5SDimitry Andric // runtime).  Such programs are deep into undefined behaviour territory,
1000b57cec5SDimitry Andric // but we choose to fail at runtime rather than generate and invalid module
1010b57cec5SDimitry Andric // or fail at compiler time.  The reason we delay the error is that we want
1020b57cec5SDimitry Andric // to support the CMake which expects to be able to compile and link programs
1030b57cec5SDimitry Andric // that refer to functions with entirely incorrect signatures (this is how
1040b57cec5SDimitry Andric // CMake detects the existence of a function in a toolchain).
1050b57cec5SDimitry Andric //
1060b57cec5SDimitry Andric // For bitcasts that involve struct types we don't know at this stage if they
1070b57cec5SDimitry Andric // would be equivalent at the wasm level and so we can't know if we need to
1080b57cec5SDimitry Andric // generate a wrapper.
1090b57cec5SDimitry Andric static Function *createWrapper(Function *F, FunctionType *Ty) {
1100b57cec5SDimitry Andric   Module *M = F->getParent();
1110b57cec5SDimitry Andric 
1120b57cec5SDimitry Andric   Function *Wrapper = Function::Create(Ty, Function::PrivateLinkage,
1130b57cec5SDimitry Andric                                        F->getName() + "_bitcast", M);
1140b57cec5SDimitry Andric   BasicBlock *BB = BasicBlock::Create(M->getContext(), "body", Wrapper);
115*0fca6ea1SDimitry Andric   const DataLayout &DL = BB->getDataLayout();
1160b57cec5SDimitry Andric 
1170b57cec5SDimitry Andric   // Determine what arguments to pass.
1180b57cec5SDimitry Andric   SmallVector<Value *, 4> Args;
1190b57cec5SDimitry Andric   Function::arg_iterator AI = Wrapper->arg_begin();
1200b57cec5SDimitry Andric   Function::arg_iterator AE = Wrapper->arg_end();
1210b57cec5SDimitry Andric   FunctionType::param_iterator PI = F->getFunctionType()->param_begin();
1220b57cec5SDimitry Andric   FunctionType::param_iterator PE = F->getFunctionType()->param_end();
1230b57cec5SDimitry Andric   bool TypeMismatch = false;
1240b57cec5SDimitry Andric   bool WrapperNeeded = false;
1250b57cec5SDimitry Andric 
1260b57cec5SDimitry Andric   Type *ExpectedRtnType = F->getFunctionType()->getReturnType();
1270b57cec5SDimitry Andric   Type *RtnType = Ty->getReturnType();
1280b57cec5SDimitry Andric 
1290b57cec5SDimitry Andric   if ((F->getFunctionType()->getNumParams() != Ty->getNumParams()) ||
1300b57cec5SDimitry Andric       (F->getFunctionType()->isVarArg() != Ty->isVarArg()) ||
1310b57cec5SDimitry Andric       (ExpectedRtnType != RtnType))
1320b57cec5SDimitry Andric     WrapperNeeded = true;
1330b57cec5SDimitry Andric 
1340b57cec5SDimitry Andric   for (; AI != AE && PI != PE; ++AI, ++PI) {
1350b57cec5SDimitry Andric     Type *ArgType = AI->getType();
1360b57cec5SDimitry Andric     Type *ParamType = *PI;
1370b57cec5SDimitry Andric 
1380b57cec5SDimitry Andric     if (ArgType == ParamType) {
1390b57cec5SDimitry Andric       Args.push_back(&*AI);
1400b57cec5SDimitry Andric     } else {
1410b57cec5SDimitry Andric       if (CastInst::isBitOrNoopPointerCastable(ArgType, ParamType, DL)) {
1420b57cec5SDimitry Andric         Instruction *PtrCast =
1430b57cec5SDimitry Andric             CastInst::CreateBitOrPointerCast(AI, ParamType, "cast");
144bdd1243dSDimitry Andric         PtrCast->insertInto(BB, BB->end());
1450b57cec5SDimitry Andric         Args.push_back(PtrCast);
1460b57cec5SDimitry Andric       } else if (ArgType->isStructTy() || ParamType->isStructTy()) {
1470b57cec5SDimitry Andric         LLVM_DEBUG(dbgs() << "createWrapper: struct param type in bitcast: "
1480b57cec5SDimitry Andric                           << F->getName() << "\n");
1490b57cec5SDimitry Andric         WrapperNeeded = false;
1500b57cec5SDimitry Andric       } else {
1510b57cec5SDimitry Andric         LLVM_DEBUG(dbgs() << "createWrapper: arg type mismatch calling: "
1520b57cec5SDimitry Andric                           << F->getName() << "\n");
1530b57cec5SDimitry Andric         LLVM_DEBUG(dbgs() << "Arg[" << Args.size() << "] Expected: "
1540b57cec5SDimitry Andric                           << *ParamType << " Got: " << *ArgType << "\n");
1550b57cec5SDimitry Andric         TypeMismatch = true;
1560b57cec5SDimitry Andric         break;
1570b57cec5SDimitry Andric       }
1580b57cec5SDimitry Andric     }
1590b57cec5SDimitry Andric   }
1600b57cec5SDimitry Andric 
1610b57cec5SDimitry Andric   if (WrapperNeeded && !TypeMismatch) {
1620b57cec5SDimitry Andric     for (; PI != PE; ++PI)
1630b57cec5SDimitry Andric       Args.push_back(UndefValue::get(*PI));
1640b57cec5SDimitry Andric     if (F->isVarArg())
1650b57cec5SDimitry Andric       for (; AI != AE; ++AI)
1660b57cec5SDimitry Andric         Args.push_back(&*AI);
1670b57cec5SDimitry Andric 
1680b57cec5SDimitry Andric     CallInst *Call = CallInst::Create(F, Args, "", BB);
1690b57cec5SDimitry Andric 
1700b57cec5SDimitry Andric     Type *ExpectedRtnType = F->getFunctionType()->getReturnType();
1710b57cec5SDimitry Andric     Type *RtnType = Ty->getReturnType();
1720b57cec5SDimitry Andric     // Determine what value to return.
1730b57cec5SDimitry Andric     if (RtnType->isVoidTy()) {
1740b57cec5SDimitry Andric       ReturnInst::Create(M->getContext(), BB);
1750b57cec5SDimitry Andric     } else if (ExpectedRtnType->isVoidTy()) {
1760b57cec5SDimitry Andric       LLVM_DEBUG(dbgs() << "Creating dummy return: " << *RtnType << "\n");
1770b57cec5SDimitry Andric       ReturnInst::Create(M->getContext(), UndefValue::get(RtnType), BB);
1780b57cec5SDimitry Andric     } else if (RtnType == ExpectedRtnType) {
1790b57cec5SDimitry Andric       ReturnInst::Create(M->getContext(), Call, BB);
1800b57cec5SDimitry Andric     } else if (CastInst::isBitOrNoopPointerCastable(ExpectedRtnType, RtnType,
1810b57cec5SDimitry Andric                                                     DL)) {
1820b57cec5SDimitry Andric       Instruction *Cast =
1830b57cec5SDimitry Andric           CastInst::CreateBitOrPointerCast(Call, RtnType, "cast");
184bdd1243dSDimitry Andric       Cast->insertInto(BB, BB->end());
1850b57cec5SDimitry Andric       ReturnInst::Create(M->getContext(), Cast, BB);
1860b57cec5SDimitry Andric     } else if (RtnType->isStructTy() || ExpectedRtnType->isStructTy()) {
1870b57cec5SDimitry Andric       LLVM_DEBUG(dbgs() << "createWrapper: struct return type in bitcast: "
1880b57cec5SDimitry Andric                         << F->getName() << "\n");
1890b57cec5SDimitry Andric       WrapperNeeded = false;
1900b57cec5SDimitry Andric     } else {
1910b57cec5SDimitry Andric       LLVM_DEBUG(dbgs() << "createWrapper: return type mismatch calling: "
1920b57cec5SDimitry Andric                         << F->getName() << "\n");
1930b57cec5SDimitry Andric       LLVM_DEBUG(dbgs() << "Expected: " << *ExpectedRtnType
1940b57cec5SDimitry Andric                         << " Got: " << *RtnType << "\n");
1950b57cec5SDimitry Andric       TypeMismatch = true;
1960b57cec5SDimitry Andric     }
1970b57cec5SDimitry Andric   }
1980b57cec5SDimitry Andric 
1990b57cec5SDimitry Andric   if (TypeMismatch) {
2000b57cec5SDimitry Andric     // Create a new wrapper that simply contains `unreachable`.
2010b57cec5SDimitry Andric     Wrapper->eraseFromParent();
2020b57cec5SDimitry Andric     Wrapper = Function::Create(Ty, Function::PrivateLinkage,
2030b57cec5SDimitry Andric                                F->getName() + "_bitcast_invalid", M);
2040b57cec5SDimitry Andric     BasicBlock *BB = BasicBlock::Create(M->getContext(), "body", Wrapper);
2050b57cec5SDimitry Andric     new UnreachableInst(M->getContext(), BB);
2060b57cec5SDimitry Andric     Wrapper->setName(F->getName() + "_bitcast_invalid");
2070b57cec5SDimitry Andric   } else if (!WrapperNeeded) {
2080b57cec5SDimitry Andric     LLVM_DEBUG(dbgs() << "createWrapper: no wrapper needed: " << F->getName()
2090b57cec5SDimitry Andric                       << "\n");
2100b57cec5SDimitry Andric     Wrapper->eraseFromParent();
2110b57cec5SDimitry Andric     return nullptr;
2120b57cec5SDimitry Andric   }
2130b57cec5SDimitry Andric   LLVM_DEBUG(dbgs() << "createWrapper: " << F->getName() << "\n");
2140b57cec5SDimitry Andric   return Wrapper;
2150b57cec5SDimitry Andric }
2160b57cec5SDimitry Andric 
2170b57cec5SDimitry Andric // Test whether a main function with type FuncTy should be rewritten to have
2180b57cec5SDimitry Andric // type MainTy.
2190b57cec5SDimitry Andric static bool shouldFixMainFunction(FunctionType *FuncTy, FunctionType *MainTy) {
2200b57cec5SDimitry Andric   // Only fix the main function if it's the standard zero-arg form. That way,
2210b57cec5SDimitry Andric   // the standard cases will work as expected, and users will see signature
2220b57cec5SDimitry Andric   // mismatches from the linker for non-standard cases.
2230b57cec5SDimitry Andric   return FuncTy->getReturnType() == MainTy->getReturnType() &&
2240b57cec5SDimitry Andric          FuncTy->getNumParams() == 0 &&
2250b57cec5SDimitry Andric          !FuncTy->isVarArg();
2260b57cec5SDimitry Andric }
2270b57cec5SDimitry Andric 
2280b57cec5SDimitry Andric bool FixFunctionBitcasts::runOnModule(Module &M) {
2290b57cec5SDimitry Andric   LLVM_DEBUG(dbgs() << "********** Fix Function Bitcasts **********\n");
2300b57cec5SDimitry Andric 
2310b57cec5SDimitry Andric   Function *Main = nullptr;
2320b57cec5SDimitry Andric   CallInst *CallMain = nullptr;
233349cc55cSDimitry Andric   SmallVector<std::pair<CallBase *, Function *>, 0> Uses;
2340b57cec5SDimitry Andric 
2350b57cec5SDimitry Andric   // Collect all the places that need wrappers.
2360b57cec5SDimitry Andric   for (Function &F : M) {
2375ffd83dbSDimitry Andric     // Skip to fix when the function is swiftcc because swiftcc allows
2385ffd83dbSDimitry Andric     // bitcast type difference for swiftself and swifterror.
2395ffd83dbSDimitry Andric     if (F.getCallingConv() == CallingConv::Swift)
2405ffd83dbSDimitry Andric       continue;
241349cc55cSDimitry Andric     findUses(&F, F, Uses);
2420b57cec5SDimitry Andric 
2430b57cec5SDimitry Andric     // If we have a "main" function, and its type isn't
2440b57cec5SDimitry Andric     // "int main(int argc, char *argv[])", create an artificial call with it
2450b57cec5SDimitry Andric     // bitcasted to that type so that we generate a wrapper for it, so that
2460b57cec5SDimitry Andric     // the C runtime can call it.
2470b57cec5SDimitry Andric     if (F.getName() == "main") {
2480b57cec5SDimitry Andric       Main = &F;
2490b57cec5SDimitry Andric       LLVMContext &C = M.getContext();
2505f757f3fSDimitry Andric       Type *MainArgTys[] = {Type::getInt32Ty(C), PointerType::get(C, 0)};
2510b57cec5SDimitry Andric       FunctionType *MainTy = FunctionType::get(Type::getInt32Ty(C), MainArgTys,
2520b57cec5SDimitry Andric                                                /*isVarArg=*/false);
2530b57cec5SDimitry Andric       if (shouldFixMainFunction(F.getFunctionType(), MainTy)) {
2540b57cec5SDimitry Andric         LLVM_DEBUG(dbgs() << "Found `main` function with incorrect type: "
2550b57cec5SDimitry Andric                           << *F.getFunctionType() << "\n");
2560b57cec5SDimitry Andric         Value *Args[] = {UndefValue::get(MainArgTys[0]),
2570b57cec5SDimitry Andric                          UndefValue::get(MainArgTys[1])};
2585f757f3fSDimitry Andric         CallMain = CallInst::Create(MainTy, Main, Args, "call_main");
259349cc55cSDimitry Andric         Uses.push_back(std::make_pair(CallMain, &F));
2600b57cec5SDimitry Andric       }
2610b57cec5SDimitry Andric     }
2620b57cec5SDimitry Andric   }
2630b57cec5SDimitry Andric 
2640b57cec5SDimitry Andric   DenseMap<std::pair<Function *, FunctionType *>, Function *> Wrappers;
2650b57cec5SDimitry Andric 
2660b57cec5SDimitry Andric   for (auto &UseFunc : Uses) {
267349cc55cSDimitry Andric     CallBase *CB = UseFunc.first;
2680b57cec5SDimitry Andric     Function *F = UseFunc.second;
269349cc55cSDimitry Andric     FunctionType *Ty = CB->getFunctionType();
2700b57cec5SDimitry Andric 
2710b57cec5SDimitry Andric     auto Pair = Wrappers.insert(std::make_pair(std::make_pair(F, Ty), nullptr));
2720b57cec5SDimitry Andric     if (Pair.second)
2730b57cec5SDimitry Andric       Pair.first->second = createWrapper(F, Ty);
2740b57cec5SDimitry Andric 
2750b57cec5SDimitry Andric     Function *Wrapper = Pair.first->second;
2760b57cec5SDimitry Andric     if (!Wrapper)
2770b57cec5SDimitry Andric       continue;
2780b57cec5SDimitry Andric 
279349cc55cSDimitry Andric     CB->setCalledOperand(Wrapper);
2800b57cec5SDimitry Andric   }
2810b57cec5SDimitry Andric 
2820b57cec5SDimitry Andric   // If we created a wrapper for main, rename the wrapper so that it's the
2830b57cec5SDimitry Andric   // one that gets called from startup.
2840b57cec5SDimitry Andric   if (CallMain) {
2850b57cec5SDimitry Andric     Main->setName("__original_main");
2860b57cec5SDimitry Andric     auto *MainWrapper =
2875ffd83dbSDimitry Andric         cast<Function>(CallMain->getCalledOperand()->stripPointerCasts());
2880b57cec5SDimitry Andric     delete CallMain;
2890b57cec5SDimitry Andric     if (Main->isDeclaration()) {
2900b57cec5SDimitry Andric       // The wrapper is not needed in this case as we don't need to export
2910b57cec5SDimitry Andric       // it to anyone else.
2920b57cec5SDimitry Andric       MainWrapper->eraseFromParent();
2930b57cec5SDimitry Andric     } else {
2940b57cec5SDimitry Andric       // Otherwise give the wrapper the same linkage as the original main
2950b57cec5SDimitry Andric       // function, so that it can be called from the same places.
2960b57cec5SDimitry Andric       MainWrapper->setName("main");
2970b57cec5SDimitry Andric       MainWrapper->setLinkage(Main->getLinkage());
2980b57cec5SDimitry Andric       MainWrapper->setVisibility(Main->getVisibility());
2990b57cec5SDimitry Andric     }
3000b57cec5SDimitry Andric   }
3010b57cec5SDimitry Andric 
3020b57cec5SDimitry Andric   return true;
3030b57cec5SDimitry Andric }
304