#include "llvm/Analysis/Lint.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AssumptionCache.h"
#include "llvm/Analysis/ConstantFolding.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/Loads.h"
#include "llvm/Analysis/Passes.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/IR/CallSite.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/InstVisitor.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/LegacyPassManager.h"
#include "llvm/Pass.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
namespace {
namespace MemRef {
static const unsigned Read = 1;
static const unsigned Write = 2;
static const unsigned Callee = 4;
static const unsigned Branchee = 8;
}
class Lint : public FunctionPass, public InstVisitor<Lint> {
friend class InstVisitor<Lint>;
void visitFunction(Function &F);
void visitCallSite(CallSite CS);
void visitMemoryReference(Instruction &I, Value *Ptr,
uint64_t Size, unsigned Align,
Type *Ty, unsigned Flags);
void visitEHBeginCatch(IntrinsicInst *II);
void visitEHEndCatch(IntrinsicInst *II);
void visitCallInst(CallInst &I);
void visitInvokeInst(InvokeInst &I);
void visitReturnInst(ReturnInst &I);
void visitLoadInst(LoadInst &I);
void visitStoreInst(StoreInst &I);
void visitXor(BinaryOperator &I);
void visitSub(BinaryOperator &I);
void visitLShr(BinaryOperator &I);
void visitAShr(BinaryOperator &I);
void visitShl(BinaryOperator &I);
void visitSDiv(BinaryOperator &I);
void visitUDiv(BinaryOperator &I);
void visitSRem(BinaryOperator &I);
void visitURem(BinaryOperator &I);
void visitAllocaInst(AllocaInst &I);
void visitVAArgInst(VAArgInst &I);
void visitIndirectBrInst(IndirectBrInst &I);
void visitExtractElementInst(ExtractElementInst &I);
void visitInsertElementInst(InsertElementInst &I);
void visitUnreachableInst(UnreachableInst &I);
Value *findValue(Value *V, bool OffsetOk) const;
Value *findValueImpl(Value *V, bool OffsetOk,
SmallPtrSetImpl<Value *> &Visited) const;
public:
Module *Mod;
const DataLayout *DL;
AliasAnalysis *AA;
AssumptionCache *AC;
DominatorTree *DT;
TargetLibraryInfo *TLI;
std::string Messages;
raw_string_ostream MessagesStr;
static char ID; Lint() : FunctionPass(ID), MessagesStr(Messages) {
initializeLintPass(*PassRegistry::getPassRegistry());
}
bool runOnFunction(Function &F) override;
void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.setPreservesAll();
AU.addRequired<AAResultsWrapperPass>();
AU.addRequired<AssumptionCacheTracker>();
AU.addRequired<TargetLibraryInfoWrapperPass>();
AU.addRequired<DominatorTreeWrapperPass>();
}
void print(raw_ostream &O, const Module *M) const override {}
void WriteValues(ArrayRef<const Value *> Vs) {
for (const Value *V : Vs) {
if (!V)
continue;
if (isa<Instruction>(V)) {
MessagesStr << *V << '\n';
} else {
V->printAsOperand(MessagesStr, true, Mod);
MessagesStr << '\n';
}
}
}
void CheckFailed(const Twine &Message) { MessagesStr << Message << '\n'; }
template <typename T1, typename... Ts>
void CheckFailed(const Twine &Message, const T1 &V1, const Ts &...Vs) {
CheckFailed(Message);
WriteValues({V1, Vs...});
}
};
}
char Lint::ID = 0;
INITIALIZE_PASS_BEGIN(Lint, "lint", "Statically lint-checks LLVM IR",
false, true)
INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
INITIALIZE_PASS_END(Lint, "lint", "Statically lint-checks LLVM IR",
false, true)
#define Assert(C, ...) \
do { if (!(C)) { CheckFailed(__VA_ARGS__); return; } } while (0)
bool Lint::runOnFunction(Function &F) {
Mod = F.getParent();
DL = &F.getParent()->getDataLayout();
AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
visit(F);
dbgs() << MessagesStr.str();
Messages.clear();
return false;
}
void Lint::visitFunction(Function &F) {
Assert(F.hasName() || F.hasLocalLinkage(),
"Unusual: Unnamed function with non-local linkage", &F);
}
void Lint::visitCallSite(CallSite CS) {
Instruction &I = *CS.getInstruction();
Value *Callee = CS.getCalledValue();
visitMemoryReference(I, Callee, MemoryLocation::UnknownSize, 0, nullptr,
MemRef::Callee);
if (Function *F = dyn_cast<Function>(findValue(Callee,
false))) {
Assert(CS.getCallingConv() == F->getCallingConv(),
"Undefined behavior: Caller and callee calling convention differ",
&I);
FunctionType *FT = F->getFunctionType();
unsigned NumActualArgs = CS.arg_size();
Assert(FT->isVarArg() ? FT->getNumParams() <= NumActualArgs
: FT->getNumParams() == NumActualArgs,
"Undefined behavior: Call argument count mismatches callee "
"argument count",
&I);
Assert(FT->getReturnType() == I.getType(),
"Undefined behavior: Call return type mismatches "
"callee return type",
&I);
Function::arg_iterator PI = F->arg_begin(), PE = F->arg_end();
CallSite::arg_iterator AI = CS.arg_begin(), AE = CS.arg_end();
for (; AI != AE; ++AI) {
Value *Actual = *AI;
if (PI != PE) {
Argument *Formal = &*PI++;
Assert(Formal->getType() == Actual->getType(),
"Undefined behavior: Call argument type mismatches "
"callee parameter type",
&I);
if (Formal->hasNoAliasAttr() && Actual->getType()->isPointerTy())
for (CallSite::arg_iterator BI = CS.arg_begin(); BI != AE; ++BI)
if (AI != BI && (*BI)->getType()->isPointerTy()) {
AliasResult Result = AA->alias(*AI, *BI);
Assert(Result != MustAlias && Result != PartialAlias,
"Unusual: noalias argument aliases another argument", &I);
}
if (Formal->hasStructRetAttr() && Actual->getType()->isPointerTy()) {
Type *Ty =
cast<PointerType>(Formal->getType())->getElementType();
visitMemoryReference(I, Actual, DL->getTypeStoreSize(Ty),
DL->getABITypeAlignment(Ty), Ty,
MemRef::Read | MemRef::Write);
}
}
}
}
if (CS.isCall() && cast<CallInst>(CS.getInstruction())->isTailCall())
for (CallSite::arg_iterator AI = CS.arg_begin(), AE = CS.arg_end();
AI != AE; ++AI) {
Value *Obj = findValue(*AI, true);
Assert(!isa<AllocaInst>(Obj),
"Undefined behavior: Call with \"tail\" keyword references "
"alloca",
&I);
}
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(&I))
switch (II->getIntrinsicID()) {
default: break;
case Intrinsic::memcpy: {
MemCpyInst *MCI = cast<MemCpyInst>(&I);
visitMemoryReference(I, MCI->getDest(), MemoryLocation::UnknownSize,
MCI->getAlignment(), nullptr, MemRef::Write);
visitMemoryReference(I, MCI->getSource(), MemoryLocation::UnknownSize,
MCI->getAlignment(), nullptr, MemRef::Read);
uint64_t Size = 0;
if (const ConstantInt *Len =
dyn_cast<ConstantInt>(findValue(MCI->getLength(),
false)))
if (Len->getValue().isIntN(32))
Size = Len->getValue().getZExtValue();
Assert(AA->alias(MCI->getSource(), Size, MCI->getDest(), Size) !=
MustAlias,
"Undefined behavior: memcpy source and destination overlap", &I);
break;
}
case Intrinsic::memmove: {
MemMoveInst *MMI = cast<MemMoveInst>(&I);
visitMemoryReference(I, MMI->getDest(), MemoryLocation::UnknownSize,
MMI->getAlignment(), nullptr, MemRef::Write);
visitMemoryReference(I, MMI->getSource(), MemoryLocation::UnknownSize,
MMI->getAlignment(), nullptr, MemRef::Read);
break;
}
case Intrinsic::memset: {
MemSetInst *MSI = cast<MemSetInst>(&I);
visitMemoryReference(I, MSI->getDest(), MemoryLocation::UnknownSize,
MSI->getAlignment(), nullptr, MemRef::Write);
break;
}
case Intrinsic::vastart:
Assert(I.getParent()->getParent()->isVarArg(),
"Undefined behavior: va_start called in a non-varargs function",
&I);
visitMemoryReference(I, CS.getArgument(0), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Read | MemRef::Write);
break;
case Intrinsic::vacopy:
visitMemoryReference(I, CS.getArgument(0), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Write);
visitMemoryReference(I, CS.getArgument(1), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Read);
break;
case Intrinsic::vaend:
visitMemoryReference(I, CS.getArgument(0), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Read | MemRef::Write);
break;
case Intrinsic::stackrestore:
visitMemoryReference(I, CS.getArgument(0), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Read | MemRef::Write);
break;
}
}
void Lint::visitCallInst(CallInst &I) {
return visitCallSite(&I);
}
void Lint::visitInvokeInst(InvokeInst &I) {
return visitCallSite(&I);
}
void Lint::visitReturnInst(ReturnInst &I) {
Function *F = I.getParent()->getParent();
Assert(!F->doesNotReturn(),
"Unusual: Return statement in function with noreturn attribute", &I);
if (Value *V = I.getReturnValue()) {
Value *Obj = findValue(V, true);
Assert(!isa<AllocaInst>(Obj), "Unusual: Returning alloca value", &I);
}
}
void Lint::visitMemoryReference(Instruction &I,
Value *Ptr, uint64_t Size, unsigned Align,
Type *Ty, unsigned Flags) {
if (Size == 0)
return;
Value *UnderlyingObject = findValue(Ptr, true);
Assert(!isa<ConstantPointerNull>(UnderlyingObject),
"Undefined behavior: Null pointer dereference", &I);
Assert(!isa<UndefValue>(UnderlyingObject),
"Undefined behavior: Undef pointer dereference", &I);
Assert(!isa<ConstantInt>(UnderlyingObject) ||
!cast<ConstantInt>(UnderlyingObject)->isAllOnesValue(),
"Unusual: All-ones pointer dereference", &I);
Assert(!isa<ConstantInt>(UnderlyingObject) ||
!cast<ConstantInt>(UnderlyingObject)->isOne(),
"Unusual: Address one pointer dereference", &I);
if (Flags & MemRef::Write) {
if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(UnderlyingObject))
Assert(!GV->isConstant(), "Undefined behavior: Write to read-only memory",
&I);
Assert(!isa<Function>(UnderlyingObject) &&
!isa<BlockAddress>(UnderlyingObject),
"Undefined behavior: Write to text section", &I);
}
if (Flags & MemRef::Read) {
Assert(!isa<Function>(UnderlyingObject), "Unusual: Load from function body",
&I);
Assert(!isa<BlockAddress>(UnderlyingObject),
"Undefined behavior: Load from block address", &I);
}
if (Flags & MemRef::Callee) {
Assert(!isa<BlockAddress>(UnderlyingObject),
"Undefined behavior: Call to block address", &I);
}
if (Flags & MemRef::Branchee) {
Assert(!isa<Constant>(UnderlyingObject) ||
isa<BlockAddress>(UnderlyingObject),
"Undefined behavior: Branch to non-blockaddress", &I);
}
int64_t Offset = 0;
if (Value *Base = GetPointerBaseWithConstantOffset(Ptr, Offset, *DL)) {
uint64_t BaseSize = MemoryLocation::UnknownSize;
unsigned BaseAlign = 0;
if (AllocaInst *AI = dyn_cast<AllocaInst>(Base)) {
Type *ATy = AI->getAllocatedType();
if (!AI->isArrayAllocation() && ATy->isSized())
BaseSize = DL->getTypeAllocSize(ATy);
BaseAlign = AI->getAlignment();
if (BaseAlign == 0 && ATy->isSized())
BaseAlign = DL->getABITypeAlignment(ATy);
} else if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Base)) {
if (GV->hasDefinitiveInitializer()) {
Type *GTy = GV->getType()->getElementType();
if (GTy->isSized())
BaseSize = DL->getTypeAllocSize(GTy);
BaseAlign = GV->getAlignment();
if (BaseAlign == 0 && GTy->isSized())
BaseAlign = DL->getABITypeAlignment(GTy);
}
}
Assert(Size == MemoryLocation::UnknownSize ||
BaseSize == MemoryLocation::UnknownSize ||
(Offset >= 0 && Offset + Size <= BaseSize),
"Undefined behavior: Buffer overflow", &I);
if (Align == 0 && Ty && Ty->isSized())
Align = DL->getABITypeAlignment(Ty);
Assert(!BaseAlign || Align <= MinAlign(BaseAlign, Offset),
"Undefined behavior: Memory reference address is misaligned", &I);
}
}
void Lint::visitLoadInst(LoadInst &I) {
visitMemoryReference(I, I.getPointerOperand(),
DL->getTypeStoreSize(I.getType()), I.getAlignment(),
I.getType(), MemRef::Read);
}
void Lint::visitStoreInst(StoreInst &I) {
visitMemoryReference(I, I.getPointerOperand(),
DL->getTypeStoreSize(I.getOperand(0)->getType()),
I.getAlignment(),
I.getOperand(0)->getType(), MemRef::Write);
}
void Lint::visitXor(BinaryOperator &I) {
Assert(!isa<UndefValue>(I.getOperand(0)) || !isa<UndefValue>(I.getOperand(1)),
"Undefined result: xor(undef, undef)", &I);
}
void Lint::visitSub(BinaryOperator &I) {
Assert(!isa<UndefValue>(I.getOperand(0)) || !isa<UndefValue>(I.getOperand(1)),
"Undefined result: sub(undef, undef)", &I);
}
void Lint::visitLShr(BinaryOperator &I) {
if (ConstantInt *CI = dyn_cast<ConstantInt>(findValue(I.getOperand(1),
false)))
Assert(CI->getValue().ult(cast<IntegerType>(I.getType())->getBitWidth()),
"Undefined result: Shift count out of range", &I);
}
void Lint::visitAShr(BinaryOperator &I) {
if (ConstantInt *CI =
dyn_cast<ConstantInt>(findValue(I.getOperand(1), false)))
Assert(CI->getValue().ult(cast<IntegerType>(I.getType())->getBitWidth()),
"Undefined result: Shift count out of range", &I);
}
void Lint::visitShl(BinaryOperator &I) {
if (ConstantInt *CI =
dyn_cast<ConstantInt>(findValue(I.getOperand(1), false)))
Assert(CI->getValue().ult(cast<IntegerType>(I.getType())->getBitWidth()),
"Undefined result: Shift count out of range", &I);
}
static bool isZero(Value *V, const DataLayout &DL, DominatorTree *DT,
AssumptionCache *AC) {
if (isa<UndefValue>(V))
return true;
VectorType *VecTy = dyn_cast<VectorType>(V->getType());
if (!VecTy) {
unsigned BitWidth = V->getType()->getIntegerBitWidth();
APInt KnownZero(BitWidth, 0), KnownOne(BitWidth, 0);
computeKnownBits(V, KnownZero, KnownOne, DL, 0, AC,
dyn_cast<Instruction>(V), DT);
return KnownZero.isAllOnesValue();
}
Constant *C = dyn_cast<Constant>(V);
if (!C)
return false;
if (C->isZeroValue())
return true;
unsigned BitWidth = VecTy->getElementType()->getIntegerBitWidth();
for (unsigned I = 0, N = VecTy->getNumElements(); I != N; ++I) {
Constant *Elem = C->getAggregateElement(I);
if (isa<UndefValue>(Elem))
return true;
APInt KnownZero(BitWidth, 0), KnownOne(BitWidth, 0);
computeKnownBits(Elem, KnownZero, KnownOne, DL);
if (KnownZero.isAllOnesValue())
return true;
}
return false;
}
void Lint::visitSDiv(BinaryOperator &I) {
Assert(!isZero(I.getOperand(1), I.getModule()->getDataLayout(), DT, AC),
"Undefined behavior: Division by zero", &I);
}
void Lint::visitUDiv(BinaryOperator &I) {
Assert(!isZero(I.getOperand(1), I.getModule()->getDataLayout(), DT, AC),
"Undefined behavior: Division by zero", &I);
}
void Lint::visitSRem(BinaryOperator &I) {
Assert(!isZero(I.getOperand(1), I.getModule()->getDataLayout(), DT, AC),
"Undefined behavior: Division by zero", &I);
}
void Lint::visitURem(BinaryOperator &I) {
Assert(!isZero(I.getOperand(1), I.getModule()->getDataLayout(), DT, AC),
"Undefined behavior: Division by zero", &I);
}
void Lint::visitAllocaInst(AllocaInst &I) {
if (isa<ConstantInt>(I.getArraySize()))
Assert(&I.getParent()->getParent()->getEntryBlock() == I.getParent(),
"Pessimization: Static alloca outside of entry block", &I);
}
void Lint::visitVAArgInst(VAArgInst &I) {
visitMemoryReference(I, I.getOperand(0), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Read | MemRef::Write);
}
void Lint::visitIndirectBrInst(IndirectBrInst &I) {
visitMemoryReference(I, I.getAddress(), MemoryLocation::UnknownSize, 0,
nullptr, MemRef::Branchee);
Assert(I.getNumDestinations() != 0,
"Undefined behavior: indirectbr with no destinations", &I);
}
void Lint::visitExtractElementInst(ExtractElementInst &I) {
if (ConstantInt *CI = dyn_cast<ConstantInt>(findValue(I.getIndexOperand(),
false)))
Assert(CI->getValue().ult(I.getVectorOperandType()->getNumElements()),
"Undefined result: extractelement index out of range", &I);
}
void Lint::visitInsertElementInst(InsertElementInst &I) {
if (ConstantInt *CI = dyn_cast<ConstantInt>(findValue(I.getOperand(2),
false)))
Assert(CI->getValue().ult(I.getType()->getNumElements()),
"Undefined result: insertelement index out of range", &I);
}
void Lint::visitUnreachableInst(UnreachableInst &I) {
Assert(&I == &I.getParent()->front() ||
std::prev(I.getIterator())->mayHaveSideEffects(),
"Unusual: unreachable immediately preceded by instruction without "
"side effects",
&I);
}
Value *Lint::findValue(Value *V, bool OffsetOk) const {
SmallPtrSet<Value *, 4> Visited;
return findValueImpl(V, OffsetOk, Visited);
}
Value *Lint::findValueImpl(Value *V, bool OffsetOk,
SmallPtrSetImpl<Value *> &Visited) const {
if (!Visited.insert(V).second)
return UndefValue::get(V->getType());
V = OffsetOk ? GetUnderlyingObject(V, *DL) : V->stripPointerCasts();
if (LoadInst *L = dyn_cast<LoadInst>(V)) {
BasicBlock::iterator BBI = L->getIterator();
BasicBlock *BB = L->getParent();
SmallPtrSet<BasicBlock *, 4> VisitedBlocks;
for (;;) {
if (!VisitedBlocks.insert(BB).second)
break;
if (Value *U =
FindAvailableLoadedValue(L->getPointerOperand(),
BB, BBI, DefMaxInstsToScan, AA))
return findValueImpl(U, OffsetOk, Visited);
if (BBI != BB->begin()) break;
BB = BB->getUniquePredecessor();
if (!BB) break;
BBI = BB->end();
}
} else if (PHINode *PN = dyn_cast<PHINode>(V)) {
if (Value *W = PN->hasConstantValue())
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
} else if (CastInst *CI = dyn_cast<CastInst>(V)) {
if (CI->isNoopCast(*DL))
return findValueImpl(CI->getOperand(0), OffsetOk, Visited);
} else if (ExtractValueInst *Ex = dyn_cast<ExtractValueInst>(V)) {
if (Value *W = FindInsertedValue(Ex->getAggregateOperand(),
Ex->getIndices()))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V)) {
if (Instruction::isCast(CE->getOpcode())) {
if (CastInst::isNoopCast(Instruction::CastOps(CE->getOpcode()),
CE->getOperand(0)->getType(), CE->getType(),
DL->getIntPtrType(V->getType())))
return findValueImpl(CE->getOperand(0), OffsetOk, Visited);
} else if (CE->getOpcode() == Instruction::ExtractValue) {
ArrayRef<unsigned> Indices = CE->getIndices();
if (Value *W = FindInsertedValue(CE->getOperand(0), Indices))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
}
}
if (Instruction *Inst = dyn_cast<Instruction>(V)) {
if (Value *W = SimplifyInstruction(Inst, *DL, TLI, DT, AC))
return findValueImpl(W, OffsetOk, Visited);
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V)) {
if (Value *W = ConstantFoldConstantExpression(CE, *DL, TLI))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
}
return V;
}
FunctionPass *llvm::createLintPass() {
return new Lint();
}
void llvm::lintFunction(const Function &f) {
Function &F = const_cast<Function&>(f);
assert(!F.isDeclaration() && "Cannot lint external functions");
legacy::FunctionPassManager FPM(F.getParent());
Lint *V = new Lint();
FPM.add(V);
FPM.run(F);
}
void llvm::lintModule(const Module &M) {
legacy::PassManager PM;
Lint *V = new Lint();
PM.add(V);
PM.run(const_cast<Module&>(M));
}