#define DEBUG_TYPE "asan"
+#include "FunctionBlackList.h"
+#include "llvm/Function.h"
+#include "llvm/IRBuilder.h"
+#include "llvm/InlineAsm.h"
+#include "llvm/IntrinsicInst.h"
+#include "llvm/LLVMContext.h"
+#include "llvm/Module.h"
+#include "llvm/Type.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/OwningPtr.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/StringExtras.h"
-#include "llvm/Function.h"
-#include "llvm/InlineAsm.h"
-#include "llvm/IntrinsicInst.h"
-#include "llvm/LLVMContext.h"
-#include "llvm/Module.h"
+#include "llvm/ADT/Triple.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/DataTypes.h"
#include "llvm/Support/Debug.h"
-#include "llvm/Support/IRBuilder.h"
-#include "llvm/Support/MemoryBuffer.h"
-#include "llvm/Support/Regex.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Support/system_error.h"
#include "llvm/Target/TargetData.h"
#include "llvm/Transforms/Instrumentation.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include "llvm/Transforms/Utils/ModuleUtils.h"
-#include "llvm/Type.h"
#include <string>
#include <algorithm>
static const uint64_t kDefaultShadowScale = 3;
static const uint64_t kDefaultShadowOffset32 = 1ULL << 29;
static const uint64_t kDefaultShadowOffset64 = 1ULL << 44;
+static const uint64_t kDefaultShadowOffsetAndroid = 0;
static const size_t kMaxStackMallocSize = 1 << 16; // 64K
static const uintptr_t kCurrentStackFrameMagic = 0x41B58AB3;
static const uintptr_t kRetiredStackFrameMagic = 0x45E0360E;
static const char *kAsanModuleCtorName = "asan.module_ctor";
+static const char *kAsanModuleDtorName = "asan.module_dtor";
+static const int kAsanCtorAndCtorPriority = 1;
static const char *kAsanReportErrorTemplate = "__asan_report_";
static const char *kAsanRegisterGlobalsName = "__asan_register_globals";
+static const char *kAsanUnregisterGlobalsName = "__asan_unregister_globals";
static const char *kAsanInitName = "__asan_init";
+static const char *kAsanHandleNoReturnName = "__asan_handle_no_return";
static const char *kAsanMappingOffsetName = "__asan_mapping_offset";
static const char *kAsanMappingScaleName = "__asan_mapping_scale";
static const char *kAsanStackMallocName = "__asan_stack_malloc";
static const int kAsanStackRightRedzoneMagic = 0xf3;
static const int kAsanStackPartialRedzoneMagic = 0xf4;
+// Accesses sizes are powers of two: 1, 2, 4, 8, 16.
+static const size_t kNumberOfAccessSizes = 5;
+
// Command-line flags.
// This flag may need to be replaced with -f[no-]asan-reads.
cl::desc("instrument read instructions"), cl::Hidden, cl::init(true));
static cl::opt<bool> ClInstrumentWrites("asan-instrument-writes",
cl::desc("instrument write instructions"), cl::Hidden, cl::init(true));
+static cl::opt<bool> ClInstrumentAtomics("asan-instrument-atomics",
+ cl::desc("instrument atomic instructions (rmw, cmpxchg)"),
+ cl::Hidden, cl::init(true));
+static cl::opt<bool> ClMergeCallbacks("asan-merge-callbacks",
+ cl::desc("merge __asan_report_ callbacks to create fewer BBs"),
+ cl::Hidden, cl::init(false));
+// This flag limits the number of instructions to be instrumented
+// in any given BB. Normally, this should be set to unlimited (INT_MAX),
+// but due to http://llvm.org/bugs/show_bug.cgi?id=12652 we temporary
+// set it to 10000.
+static cl::opt<int> ClMaxInsnsToInstrumentPerBB("asan-max-ins-per-bb",
+ cl::init(10000),
+ cl::desc("maximal number of instructions to instrument in any given BB"),
+ cl::Hidden);
// This flag may need to be replaced with -f[no]asan-stack.
static cl::opt<bool> ClStack("asan-stack",
cl::desc("Handle stack memory"), cl::Hidden, cl::init(true));
static cl::opt<std::string> ClBlackListFile("asan-blacklist",
cl::desc("File containing the list of functions to ignore "
"during instrumentation"), cl::Hidden);
-static cl::opt<bool> ClUseCall("asan-use-call",
- cl::desc("Use function call to generate a crash"), cl::Hidden,
- cl::init(true));
// These flags allow to change the shadow mapping.
// The shadow mapping looks like
namespace {
-// Blacklisted functions are not instrumented.
-// The blacklist file contains one or more lines like this:
-// ---
-// fun:FunctionWildCard
-// ---
-// This is similar to the "ignore" feature of ThreadSanitizer.
-// http://code.google.com/p/data-race-test/wiki/ThreadSanitizerIgnores
-class BlackList {
- public:
- BlackList(const std::string &Path);
- bool isIn(const Function &F);
- private:
- Regex *Functions;
+/// When the crash callbacks are merged, they receive some amount of arguments
+/// that are merged in a PHI node. This struct represents arguments from one
+/// call site.
+struct CrashArg {
+ Value *Arg1;
+ Value *Arg2;
+};
+
+/// An object of this type is created while instrumenting every function.
+struct AsanFunctionContext {
+ AsanFunctionContext(Function &Function) : F(Function), CrashBlock() { }
+
+ Function &F;
+ // These are initially zero. If we require at least one call to
+ // __asan_report_{read,write}{1,2,4,8,16}, an appropriate BB is created.
+ BasicBlock *CrashBlock[2][kNumberOfAccessSizes];
+ typedef SmallVector<CrashArg, 8> CrashArgsVec;
+ CrashArgsVec CrashArgs[2][kNumberOfAccessSizes];
};
/// AddressSanitizer: instrument the code in module to find memory bugs.
struct AddressSanitizer : public ModulePass {
AddressSanitizer();
- void instrumentMop(Instruction *I);
- void instrumentAddress(Instruction *OrigIns, IRBuilder<> &IRB,
+ virtual const char *getPassName() const;
+ void instrumentMop(AsanFunctionContext &AFC, Instruction *I);
+ void instrumentAddress(AsanFunctionContext &AFC,
+ Instruction *OrigIns, IRBuilder<> &IRB,
Value *Addr, uint32_t TypeSize, bool IsWrite);
- Instruction *generateCrashCode(IRBuilder<> &IRB, Value *Addr,
- bool IsWrite, uint32_t TypeSize);
- bool instrumentMemIntrinsic(MemIntrinsic *MI);
- void instrumentMemIntrinsicParam(Instruction *OrigIns, Value *Addr,
- Value *Size,
+ Value *createSlowPathCmp(IRBuilder<> &IRB, Value *AddrLong,
+ Value *ShadowValue, uint32_t TypeSize);
+ Instruction *generateCrashCode(BasicBlock *BB, Value *Addr, Value *PC,
+ bool IsWrite, size_t AccessSizeIndex);
+ bool instrumentMemIntrinsic(AsanFunctionContext &AFC, MemIntrinsic *MI);
+ void instrumentMemIntrinsicParam(AsanFunctionContext &AFC,
+ Instruction *OrigIns, Value *Addr,
+ Value *Size,
Instruction *InsertBefore, bool IsWrite);
Value *memToShadow(Value *Shadow, IRBuilder<> &IRB);
bool handleFunction(Module &M, Function &F);
+ bool maybeInsertAsanInitAtFunctionEntry(Function &F);
bool poisonStackInFunction(Module &M, Function &F);
virtual bool runOnModule(Module &M);
bool insertGlobalRedzones(Module &M);
- BranchInst *splitBlockAndInsertIfThen(Instruction *SplitBefore, Value *Cmp);
static char ID; // Pass identification, replacement for typeid
private:
uint64_t getAllocaSizeInBytes(AllocaInst *AI) {
Type *Ty = AI->getAllocatedType();
- uint64_t SizeInBytes = TD->getTypeStoreSizeInBits(Ty) / 8;
+ uint64_t SizeInBytes = TD->getTypeAllocSize(Ty);
return SizeInBytes;
}
uint64_t getAlignedSize(uint64_t SizeInBytes) {
return getAlignedSize(SizeInBytes);
}
+ Function *checkInterfaceFunction(Constant *FuncOrBitcast);
void PoisonStack(const ArrayRef<AllocaInst*> &AllocaVec, IRBuilder<> IRB,
Value *ShadowBase, bool DoPoison);
bool LooksLikeCodeInBug11395(Instruction *I);
- Module *CurrentModule;
LLVMContext *C;
TargetData *TD;
uint64_t MappingOffset;
Function *AsanCtorFunction;
Function *AsanInitFunction;
Instruction *CtorInsertBefore;
- OwningPtr<BlackList> BL;
+ OwningPtr<FunctionBlackList> BL;
+ // This array is indexed by AccessIsWrite and log2(AccessSize).
+ Function *AsanErrorCallback[2][kNumberOfAccessSizes];
+ InlineAsm *EmptyAsm;
};
+
} // namespace
char AddressSanitizer::ID = 0;
return new AddressSanitizer();
}
+const char *AddressSanitizer::getPassName() const {
+ return "AddressSanitizer";
+}
+
+static size_t TypeSizeToSizeIndex(uint32_t TypeSize) {
+ size_t Res = CountTrailingZeros_32(TypeSize / 8);
+ assert(Res < kNumberOfAccessSizes);
+ return Res;
+}
+
// Create a constant for Str so that we can pass it to the run-time lib.
static GlobalVariable *createPrivateGlobalForString(Module &M, StringRef Str) {
- Constant *StrConst = ConstantArray::get(M.getContext(), Str);
+ Constant *StrConst = ConstantDataArray::getString(M.getContext(), Str);
return new GlobalVariable(M, StrConst->getType(), true,
GlobalValue::PrivateLinkage, StrConst, "");
}
// Split the basic block and insert an if-then code.
// Before:
// Head
-// SplitBefore
+// Cmp
// Tail
// After:
// Head
// if (Cmp)
-// NewBasicBlock
-// SplitBefore
+// ThenBlock
// Tail
//
-// Returns the NewBasicBlock's terminator.
-BranchInst *AddressSanitizer::splitBlockAndInsertIfThen(
- Instruction *SplitBefore, Value *Cmp) {
+// If ThenBlock is zero, a new block is created and its terminator is returned.
+// Otherwize 0 is returned.
+static BranchInst *splitBlockAndInsertIfThen(Value *Cmp,
+ BasicBlock *ThenBlock = 0) {
+ Instruction *SplitBefore = cast<Instruction>(Cmp)->getNextNode();
BasicBlock *Head = SplitBefore->getParent();
BasicBlock *Tail = Head->splitBasicBlock(SplitBefore);
TerminatorInst *HeadOldTerm = Head->getTerminator();
- BasicBlock *NewBasicBlock =
- BasicBlock::Create(*C, "", Head->getParent());
- BranchInst *HeadNewTerm = BranchInst::Create(/*ifTrue*/NewBasicBlock,
- /*ifFalse*/Tail,
- Cmp);
+ BranchInst *CheckTerm = 0;
+ if (!ThenBlock) {
+ LLVMContext &C = Head->getParent()->getParent()->getContext();
+ ThenBlock = BasicBlock::Create(C, "", Head->getParent(), Tail);
+ CheckTerm = BranchInst::Create(Tail, ThenBlock);
+ }
+ BranchInst *HeadNewTerm =
+ BranchInst::Create(/*ifTrue*/ThenBlock, /*ifFalse*/Tail, Cmp);
ReplaceInstWithInst(HeadOldTerm, HeadNewTerm);
- BranchInst *CheckTerm = BranchInst::Create(Tail, NewBasicBlock);
return CheckTerm;
}
MappingOffset));
}
-void AddressSanitizer::instrumentMemIntrinsicParam(Instruction *OrigIns,
+void AddressSanitizer::instrumentMemIntrinsicParam(
+ AsanFunctionContext &AFC, Instruction *OrigIns,
Value *Addr, Value *Size, Instruction *InsertBefore, bool IsWrite) {
// Check the first byte.
{
IRBuilder<> IRB(InsertBefore);
- instrumentAddress(OrigIns, IRB, Addr, 8, IsWrite);
+ instrumentAddress(AFC, OrigIns, IRB, Addr, 8, IsWrite);
}
// Check the last byte.
{
SizeMinusOne = IRB.CreateIntCast(SizeMinusOne, IntptrTy, false);
Value *AddrLong = IRB.CreatePointerCast(Addr, IntptrTy);
Value *AddrPlusSizeMinisOne = IRB.CreateAdd(AddrLong, SizeMinusOne);
- instrumentAddress(OrigIns, IRB, AddrPlusSizeMinisOne, 8, IsWrite);
+ instrumentAddress(AFC, OrigIns, IRB, AddrPlusSizeMinisOne, 8, IsWrite);
}
}
// Instrument memset/memmove/memcpy
-bool AddressSanitizer::instrumentMemIntrinsic(MemIntrinsic *MI) {
+bool AddressSanitizer::instrumentMemIntrinsic(AsanFunctionContext &AFC,
+ MemIntrinsic *MI) {
Value *Dst = MI->getDest();
MemTransferInst *MemTran = dyn_cast<MemTransferInst>(MI);
- Value *Src = MemTran ? MemTran->getSource() : NULL;
+ Value *Src = MemTran ? MemTran->getSource() : 0;
Value *Length = MI->getLength();
Constant *ConstLength = dyn_cast<Constant>(Length);
IRBuilder<> IRB(InsertBefore);
Value *Cmp = IRB.CreateICmpNE(Length,
- Constant::getNullValue(Length->getType()));
- InsertBefore = splitBlockAndInsertIfThen(InsertBefore, Cmp);
+ Constant::getNullValue(Length->getType()));
+ InsertBefore = splitBlockAndInsertIfThen(Cmp);
}
- instrumentMemIntrinsicParam(MI, Dst, Length, InsertBefore, true);
+ instrumentMemIntrinsicParam(AFC, MI, Dst, Length, InsertBefore, true);
if (Src)
- instrumentMemIntrinsicParam(MI, Src, Length, InsertBefore, false);
+ instrumentMemIntrinsicParam(AFC, MI, Src, Length, InsertBefore, false);
return true;
}
-static Value *getLDSTOperand(Instruction *I) {
+// If I is an interesting memory access, return the PointerOperand
+// and set IsWrite. Otherwise return NULL.
+static Value *isInterestingMemoryAccess(Instruction *I, bool *IsWrite) {
if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
+ if (!ClInstrumentReads) return NULL;
+ *IsWrite = false;
return LI->getPointerOperand();
}
- return cast<StoreInst>(*I).getPointerOperand();
+ if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
+ if (!ClInstrumentWrites) return NULL;
+ *IsWrite = true;
+ return SI->getPointerOperand();
+ }
+ if (AtomicRMWInst *RMW = dyn_cast<AtomicRMWInst>(I)) {
+ if (!ClInstrumentAtomics) return NULL;
+ *IsWrite = true;
+ return RMW->getPointerOperand();
+ }
+ if (AtomicCmpXchgInst *XCHG = dyn_cast<AtomicCmpXchgInst>(I)) {
+ if (!ClInstrumentAtomics) return NULL;
+ *IsWrite = true;
+ return XCHG->getPointerOperand();
+ }
+ return NULL;
}
-void AddressSanitizer::instrumentMop(Instruction *I) {
- int IsWrite = isa<StoreInst>(*I);
- Value *Addr = getLDSTOperand(I);
+void AddressSanitizer::instrumentMop(AsanFunctionContext &AFC, Instruction *I) {
+ bool IsWrite;
+ Value *Addr = isInterestingMemoryAccess(I, &IsWrite);
+ assert(Addr);
if (ClOpt && ClOptGlobals && isa<GlobalVariable>(Addr)) {
// We are accessing a global scalar variable. Nothing to catch here.
return;
}
IRBuilder<> IRB(I);
- instrumentAddress(I, IRB, Addr, TypeSize, IsWrite);
+ instrumentAddress(AFC, I, IRB, Addr, TypeSize, IsWrite);
+}
+
+// Validate the result of Module::getOrInsertFunction called for an interface
+// function of AddressSanitizer. If the instrumented module defines a function
+// with the same name, their prototypes must match, otherwise
+// getOrInsertFunction returns a bitcast.
+Function *AddressSanitizer::checkInterfaceFunction(Constant *FuncOrBitcast) {
+ if (isa<Function>(FuncOrBitcast)) return cast<Function>(FuncOrBitcast);
+ FuncOrBitcast->dump();
+ report_fatal_error("trying to redefine an AddressSanitizer "
+ "interface function");
}
Instruction *AddressSanitizer::generateCrashCode(
- IRBuilder<> &IRB, Value *Addr, bool IsWrite, uint32_t TypeSize) {
-
- if (ClUseCall) {
- // Here we use a call instead of arch-specific asm to report an error.
- // This is almost always slower (because the codegen needs to generate
- // prologue/epilogue for otherwise leaf functions) and generates more code.
- // This mode could be useful if we can not use SIGILL for some reason.
- //
- // IsWrite and TypeSize are encoded in the function name.
- std::string FunctionName = std::string(kAsanReportErrorTemplate) +
- (IsWrite ? "store" : "load") + itostr(TypeSize / 8);
- Value *ReportWarningFunc = CurrentModule->getOrInsertFunction(
- FunctionName, IRB.getVoidTy(), IntptrTy, NULL);
- CallInst *Call = IRB.CreateCall(ReportWarningFunc, Addr);
- Call->setDoesNotReturn();
- return Call;
- }
-
- uint32_t LogOfSizeInBytes = CountTrailingZeros_32(TypeSize / 8);
- assert(8U * (1 << LogOfSizeInBytes) == TypeSize);
- uint8_t TelltaleValue = IsWrite * 8 + LogOfSizeInBytes;
- assert(TelltaleValue < 16);
-
- // Move the failing address to %rax/%eax
- FunctionType *Fn1Ty = FunctionType::get(
- IRB.getVoidTy(), ArrayRef<Type*>(IntptrTy), false);
- const char *MovStr = LongSize == 32
- ? "mov $0, %eax" : "mov $0, %rax";
- Value *AsmMov = InlineAsm::get(
- Fn1Ty, StringRef(MovStr), StringRef("r"), true);
- IRB.CreateCall(AsmMov, Addr);
-
- // crash with ud2; could use int3, but it is less friendly to gdb.
- // after ud2 put a 1-byte instruction that encodes the access type and size.
-
- const char *TelltaleInsns[16] = {
- "push %eax", // 0x50
- "push %ecx", // 0x51
- "push %edx", // 0x52
- "push %ebx", // 0x53
- "push %esp", // 0x54
- "push %ebp", // 0x55
- "push %esi", // 0x56
- "push %edi", // 0x57
- "pop %eax", // 0x58
- "pop %ecx", // 0x59
- "pop %edx", // 0x5a
- "pop %ebx", // 0x5b
- "pop %esp", // 0x5c
- "pop %ebp", // 0x5d
- "pop %esi", // 0x5e
- "pop %edi" // 0x5f
- };
-
- std::string AsmStr = "ud2;";
- AsmStr += TelltaleInsns[TelltaleValue];
- Value *MyAsm = InlineAsm::get(FunctionType::get(Type::getVoidTy(*C), false),
- StringRef(AsmStr), StringRef(""), true);
- CallInst *AsmCall = IRB.CreateCall(MyAsm);
-
- // This saves us one jump, but triggers a bug in RA (or somewhere else):
- // while building 483.xalancbmk the compiler goes into infinite loop in
- // llvm::SpillPlacement::iterate() / RAGreedy::growRegion
- // AsmCall->setDoesNotReturn();
- return AsmCall;
+ BasicBlock *BB, Value *Addr, Value *PC,
+ bool IsWrite, size_t AccessSizeIndex) {
+ IRBuilder<> IRB(BB->getFirstNonPHI());
+ CallInst *Call;
+ if (PC)
+ Call = IRB.CreateCall2(AsanErrorCallback[IsWrite][AccessSizeIndex],
+ Addr, PC);
+ else
+ Call = IRB.CreateCall(AsanErrorCallback[IsWrite][AccessSizeIndex], Addr);
+ // We don't do Call->setDoesNotReturn() because the BB already has
+ // UnreachableInst at the end.
+ // This EmptyAsm is required to avoid callback merge.
+ IRB.CreateCall(EmptyAsm);
+ return Call;
}
-void AddressSanitizer::instrumentAddress(Instruction *OrigIns,
+Value *AddressSanitizer::createSlowPathCmp(IRBuilder<> &IRB, Value *AddrLong,
+ Value *ShadowValue,
+ uint32_t TypeSize) {
+ size_t Granularity = 1 << MappingScale;
+ // Addr & (Granularity - 1)
+ Value *LastAccessedByte = IRB.CreateAnd(
+ AddrLong, ConstantInt::get(IntptrTy, Granularity - 1));
+ // (Addr & (Granularity - 1)) + size - 1
+ if (TypeSize / 8 > 1)
+ LastAccessedByte = IRB.CreateAdd(
+ LastAccessedByte, ConstantInt::get(IntptrTy, TypeSize / 8 - 1));
+ // (uint8_t) ((Addr & (Granularity-1)) + size - 1)
+ LastAccessedByte = IRB.CreateIntCast(
+ LastAccessedByte, IRB.getInt8Ty(), false);
+ // ((uint8_t) ((Addr & (Granularity-1)) + size - 1)) >= ShadowValue
+ return IRB.CreateICmpSGE(LastAccessedByte, ShadowValue);
+}
+
+void AddressSanitizer::instrumentAddress(AsanFunctionContext &AFC,
+ Instruction *OrigIns,
IRBuilder<> &IRB, Value *Addr,
uint32_t TypeSize, bool IsWrite) {
Value *AddrLong = IRB.CreatePointerCast(Addr, IntptrTy);
Value *Cmp = IRB.CreateICmpNE(ShadowValue, CmpVal);
- Instruction *CheckTerm = splitBlockAndInsertIfThen(
- cast<Instruction>(Cmp)->getNextNode(), Cmp);
- IRBuilder<> IRB2(CheckTerm);
+ BasicBlock *CrashBlock = 0;
+ if (ClMergeCallbacks) {
+ size_t AccessSizeIndex = TypeSizeToSizeIndex(TypeSize);
+ BasicBlock **Cached = &AFC.CrashBlock[IsWrite][AccessSizeIndex];
+ if (!*Cached) {
+ std::string BBName("crash_bb-");
+ BBName += (IsWrite ? "w-" : "r-") + itostr(1 << AccessSizeIndex);
+ BasicBlock *BB = BasicBlock::Create(*C, BBName, &AFC.F);
+ new UnreachableInst(*C, BB);
+ *Cached = BB;
+ }
+ CrashBlock = *Cached;
+ // We need to pass the PC as the second parameter to __asan_report_*.
+ // There are few problems:
+ // - Some architectures (e.g. x86_32) don't have a cheap way to get the PC.
+ // - LLVM doesn't have the appropriate intrinsic.
+ // For now, put a random number into the PC, just to allow experiments.
+ Value *PC = ConstantInt::get(IntptrTy, rand());
+ CrashArg Arg = {AddrLong, PC};
+ AFC.CrashArgs[IsWrite][AccessSizeIndex].push_back(Arg);
+ } else {
+ CrashBlock = BasicBlock::Create(*C, "crash_bb", &AFC.F);
+ new UnreachableInst(*C, CrashBlock);
+ size_t AccessSizeIndex = TypeSizeToSizeIndex(TypeSize);
+ Instruction *Crash =
+ generateCrashCode(CrashBlock, AddrLong, 0, IsWrite, AccessSizeIndex);
+ Crash->setDebugLoc(OrigIns->getDebugLoc());
+ }
size_t Granularity = 1 << MappingScale;
if (TypeSize < 8 * Granularity) {
- // Addr & (Granularity - 1)
- Value *Lower3Bits = IRB2.CreateAnd(
- AddrLong, ConstantInt::get(IntptrTy, Granularity - 1));
- // (Addr & (Granularity - 1)) + size - 1
- Value *LastAccessedByte = IRB2.CreateAdd(
- Lower3Bits, ConstantInt::get(IntptrTy, TypeSize / 8 - 1));
- // (uint8_t) ((Addr & (Granularity-1)) + size - 1)
- LastAccessedByte = IRB2.CreateIntCast(
- LastAccessedByte, IRB.getInt8Ty(), false);
- // ((uint8_t) ((Addr & (Granularity-1)) + size - 1)) >= ShadowValue
- Value *Cmp2 = IRB2.CreateICmpSGE(LastAccessedByte, ShadowValue);
-
- CheckTerm = splitBlockAndInsertIfThen(CheckTerm, Cmp2);
- }
-
- IRBuilder<> IRB1(CheckTerm);
- Instruction *Crash = generateCrashCode(IRB1, AddrLong, IsWrite, TypeSize);
- Crash->setDebugLoc(OrigIns->getDebugLoc());
- ReplaceInstWithInst(CheckTerm, new UnreachableInst(*C));
+ BranchInst *CheckTerm = splitBlockAndInsertIfThen(Cmp);
+ assert(CheckTerm->isUnconditional());
+ BasicBlock *NextBB = CheckTerm->getSuccessor(0);
+ IRB.SetInsertPoint(CheckTerm);
+ Value *Cmp2 = createSlowPathCmp(IRB, AddrLong, ShadowValue, TypeSize);
+ BranchInst *NewTerm = BranchInst::Create(CrashBlock, NextBB, Cmp2);
+ ReplaceInstWithInst(CheckTerm, NewTerm);
+ } else {
+ splitBlockAndInsertIfThen(Cmp, CrashBlock);
+ }
}
// This function replaces all global variables with new variables that have
continue;
}
- // Ignore the globals from the __OBJC section. The ObjC runtime assumes
- // those conform to /usr/lib/objc/runtime.h, so we can't add redzones to
- // them.
if (G->hasSection()) {
StringRef Section(G->getSection());
+ // Ignore the globals from the __OBJC section. The ObjC runtime assumes
+ // those conform to /usr/lib/objc/runtime.h, so we can't add redzones to
+ // them.
if ((Section.find("__OBJC,") == 0) ||
(Section.find("__DATA, __objc_") == 0)) {
DEBUG(dbgs() << "Ignoring ObjC runtime global: " << *G);
continue;
}
+ // See http://code.google.com/p/address-sanitizer/issues/detail?id=32
+ // Constant CFString instances are compiled in the following way:
+ // -- the string buffer is emitted into
+ // __TEXT,__cstring,cstring_literals
+ // -- the constant NSConstantString structure referencing that buffer
+ // is placed into __DATA,__cfstring
+ // Therefore there's no point in placing redzones into __DATA,__cfstring.
+ // Moreover, it causes the linker to crash on OS X 10.7
+ if (Section.find("__DATA,__cfstring") == 0) {
+ DEBUG(dbgs() << "Ignoring CFString: " << *G);
+ continue;
+ }
}
GlobalsToChange.push_back(G);
GlobalVariable *G = GlobalsToChange[i];
PointerType *PtrTy = cast<PointerType>(G->getType());
Type *Ty = PtrTy->getElementType();
- uint64_t SizeInBytes = TD->getTypeStoreSizeInBits(Ty) / 8;
+ uint64_t SizeInBytes = TD->getTypeAllocSize(Ty);
uint64_t RightRedzoneSize = RedzoneSize +
(RedzoneSize - (SizeInBytes % RedzoneSize));
Type *RightRedZoneTy = ArrayType::get(IRB.getInt8Ty(), RightRedzoneSize);
NewTy, G->getInitializer(),
Constant::getNullValue(RightRedZoneTy), NULL);
- GlobalVariable *Name = createPrivateGlobalForString(M, G->getName());
+ SmallString<2048> DescriptionOfGlobal = G->getName();
+ DescriptionOfGlobal += " (";
+ DescriptionOfGlobal += M.getModuleIdentifier();
+ DescriptionOfGlobal += ")";
+ GlobalVariable *Name = createPrivateGlobalForString(M, DescriptionOfGlobal);
// Create a new global variable with enough space for a redzone.
GlobalVariable *NewGlobal = new GlobalVariable(
M, NewTy, G->isConstant(), G->getLinkage(),
- NewInitializer, "", G, G->isThreadLocal());
+ NewInitializer, "", G, G->getThreadLocalMode());
NewGlobal->copyAttributesFrom(G);
NewGlobal->setAlignment(RedzoneSize);
Indices2[1] = IRB.getInt32(0);
G->replaceAllUsesWith(
- ConstantExpr::getGetElementPtr(NewGlobal, Indices2, 2));
+ ConstantExpr::getGetElementPtr(NewGlobal, Indices2, true));
NewGlobal->takeName(G);
G->eraseFromParent();
M, ArrayOfGlobalStructTy, false, GlobalVariable::PrivateLinkage,
ConstantArray::get(ArrayOfGlobalStructTy, Initializers), "");
- Function *AsanRegisterGlobals = cast<Function>(M.getOrInsertFunction(
+ Function *AsanRegisterGlobals = checkInterfaceFunction(M.getOrInsertFunction(
kAsanRegisterGlobalsName, IRB.getVoidTy(), IntptrTy, IntptrTy, NULL));
AsanRegisterGlobals->setLinkage(Function::ExternalLinkage);
IRB.CreatePointerCast(AllGlobals, IntptrTy),
ConstantInt::get(IntptrTy, n));
+ // We also need to unregister globals at the end, e.g. when a shared library
+ // gets closed.
+ Function *AsanDtorFunction = Function::Create(
+ FunctionType::get(Type::getVoidTy(*C), false),
+ GlobalValue::InternalLinkage, kAsanModuleDtorName, &M);
+ BasicBlock *AsanDtorBB = BasicBlock::Create(*C, "", AsanDtorFunction);
+ IRBuilder<> IRB_Dtor(ReturnInst::Create(*C, AsanDtorBB));
+ Function *AsanUnregisterGlobals =
+ checkInterfaceFunction(M.getOrInsertFunction(
+ kAsanUnregisterGlobalsName,
+ IRB.getVoidTy(), IntptrTy, IntptrTy, NULL));
+ AsanUnregisterGlobals->setLinkage(Function::ExternalLinkage);
+
+ IRB_Dtor.CreateCall2(AsanUnregisterGlobals,
+ IRB.CreatePointerCast(AllGlobals, IntptrTy),
+ ConstantInt::get(IntptrTy, n));
+ appendToGlobalDtors(M, AsanDtorFunction, kAsanCtorAndCtorPriority);
+
DEBUG(dbgs() << M);
return true;
}
TD = getAnalysisIfAvailable<TargetData>();
if (!TD)
return false;
- BL.reset(new BlackList(ClBlackListFile));
+ BL.reset(new FunctionBlackList(ClBlackListFile));
- CurrentModule = &M;
C = &(M.getContext());
LongSize = TD->getPointerSizeInBits();
IntptrTy = Type::getIntNTy(*C, LongSize);
// call __asan_init in the module ctor.
IRBuilder<> IRB(CtorInsertBefore);
- AsanInitFunction = cast<Function>(
+ AsanInitFunction = checkInterfaceFunction(
M.getOrInsertFunction(kAsanInitName, IRB.getVoidTy(), NULL));
AsanInitFunction->setLinkage(Function::ExternalLinkage);
IRB.CreateCall(AsanInitFunction);
- MappingOffset = LongSize == 32
- ? kDefaultShadowOffset32 : kDefaultShadowOffset64;
+ // Create __asan_report* callbacks.
+ for (size_t AccessIsWrite = 0; AccessIsWrite <= 1; AccessIsWrite++) {
+ for (size_t AccessSizeIndex = 0; AccessSizeIndex < kNumberOfAccessSizes;
+ AccessSizeIndex++) {
+ // IsWrite and TypeSize are encoded in the function name.
+ std::string FunctionName = std::string(kAsanReportErrorTemplate) +
+ (AccessIsWrite ? "store" : "load") + itostr(1 << AccessSizeIndex);
+ // If we are merging crash callbacks, they have two parameters.
+ if (ClMergeCallbacks)
+ AsanErrorCallback[AccessIsWrite][AccessSizeIndex] = cast<Function>(
+ M.getOrInsertFunction(FunctionName, IRB.getVoidTy(), IntptrTy,
+ IntptrTy, NULL));
+ else
+ AsanErrorCallback[AccessIsWrite][AccessSizeIndex] = cast<Function>(
+ M.getOrInsertFunction(FunctionName, IRB.getVoidTy(), IntptrTy, NULL));
+ }
+ }
+ // We insert an empty inline asm after __asan_report* to avoid callback merge.
+ EmptyAsm = InlineAsm::get(FunctionType::get(IRB.getVoidTy(), false),
+ StringRef(""), StringRef(""),
+ /*hasSideEffects=*/true);
+
+ llvm::Triple targetTriple(M.getTargetTriple());
+ bool isAndroid = targetTriple.getEnvironment() == llvm::Triple::ANDROIDEABI;
+
+ MappingOffset = isAndroid ? kDefaultShadowOffsetAndroid :
+ (LongSize == 32 ? kDefaultShadowOffset32 : kDefaultShadowOffset64);
if (ClMappingOffsetLog >= 0) {
if (ClMappingOffsetLog == 0) {
// special case
if (ClGlobals)
Res |= insertGlobalRedzones(M);
- // Tell the run-time the current values of mapping offset and scale.
- GlobalValue *asan_mapping_offset =
- new GlobalVariable(M, IntptrTy, true, GlobalValue::LinkOnceODRLinkage,
- ConstantInt::get(IntptrTy, MappingOffset),
- kAsanMappingOffsetName);
- GlobalValue *asan_mapping_scale =
- new GlobalVariable(M, IntptrTy, true, GlobalValue::LinkOnceODRLinkage,
- ConstantInt::get(IntptrTy, MappingScale),
- kAsanMappingScaleName);
- // Read these globals, otherwise they may be optimized away.
- IRB.CreateLoad(asan_mapping_scale, true);
- IRB.CreateLoad(asan_mapping_offset, true);
+ if (ClMappingOffsetLog >= 0) {
+ // Tell the run-time the current values of mapping offset and scale.
+ GlobalValue *asan_mapping_offset =
+ new GlobalVariable(M, IntptrTy, true, GlobalValue::LinkOnceODRLinkage,
+ ConstantInt::get(IntptrTy, MappingOffset),
+ kAsanMappingOffsetName);
+ // Read the global, otherwise it may be optimized away.
+ IRB.CreateLoad(asan_mapping_offset, true);
+ }
+ if (ClMappingScale) {
+ GlobalValue *asan_mapping_scale =
+ new GlobalVariable(M, IntptrTy, true, GlobalValue::LinkOnceODRLinkage,
+ ConstantInt::get(IntptrTy, MappingScale),
+ kAsanMappingScaleName);
+ // Read the global, otherwise it may be optimized away.
+ IRB.CreateLoad(asan_mapping_scale, true);
+ }
for (Module::iterator F = M.begin(), E = M.end(); F != E; ++F) {
Res |= handleFunction(M, *F);
}
- appendToGlobalCtors(M, AsanCtorFunction, 1 /*high priority*/);
+ appendToGlobalCtors(M, AsanCtorFunction, kAsanCtorAndCtorPriority);
return Res;
}
+bool AddressSanitizer::maybeInsertAsanInitAtFunctionEntry(Function &F) {
+ // For each NSObject descendant having a +load method, this method is invoked
+ // by the ObjC runtime before any of the static constructors is called.
+ // Therefore we need to instrument such methods with a call to __asan_init
+ // at the beginning in order to initialize our runtime before any access to
+ // the shadow memory.
+ // We cannot just ignore these methods, because they may call other
+ // instrumented functions.
+ if (F.getName().find(" load]") != std::string::npos) {
+ IRBuilder<> IRB(F.begin()->begin());
+ IRB.CreateCall(AsanInitFunction);
+ return true;
+ }
+ return false;
+}
+
bool AddressSanitizer::handleFunction(Module &M, Function &F) {
if (BL->isIn(F)) return false;
if (&F == AsanCtorFunction) return false;
+ // If needed, insert __asan_init before checking for AddressSafety attr.
+ maybeInsertAsanInitAtFunctionEntry(F);
+
+ if (!F.hasFnAttr(Attribute::AddressSafety)) return false;
+
if (!ClDebugFunc.empty() && ClDebugFunc != F.getName())
return false;
// We want to instrument every address only once per basic block
// (unless there are calls between uses).
SmallSet<Value*, 16> TempsToInstrument;
SmallVector<Instruction*, 16> ToInstrument;
+ SmallVector<Instruction*, 8> NoReturnCalls;
+ bool IsWrite;
// Fill the set of memory operations to instrument.
for (Function::iterator FI = F.begin(), FE = F.end();
FI != FE; ++FI) {
TempsToInstrument.clear();
+ int NumInsnsPerBB = 0;
for (BasicBlock::iterator BI = FI->begin(), BE = FI->end();
BI != BE; ++BI) {
- if ((isa<LoadInst>(BI) && ClInstrumentReads) ||
- (isa<StoreInst>(BI) && ClInstrumentWrites)) {
- Value *Addr = getLDSTOperand(BI);
+ if (LooksLikeCodeInBug11395(BI)) return false;
+ if (Value *Addr = isInterestingMemoryAccess(BI, &IsWrite)) {
if (ClOpt && ClOptSameTemp) {
if (!TempsToInstrument.insert(Addr))
continue; // We've seen this temp in the current BB.
} else if (isa<MemIntrinsic>(BI) && ClMemIntrin) {
// ok, take it.
} else {
- if (isa<CallInst>(BI)) {
+ if (CallInst *CI = dyn_cast<CallInst>(BI)) {
// A call inside BB.
TempsToInstrument.clear();
+ if (CI->doesNotReturn()) {
+ NoReturnCalls.push_back(CI);
+ }
}
continue;
}
ToInstrument.push_back(BI);
+ NumInsnsPerBB++;
+ if (NumInsnsPerBB >= ClMaxInsnsToInstrumentPerBB)
+ break;
}
}
+ AsanFunctionContext AFC(F);
+
// Instrument.
int NumInstrumented = 0;
for (size_t i = 0, n = ToInstrument.size(); i != n; i++) {
Instruction *Inst = ToInstrument[i];
if (ClDebugMin < 0 || ClDebugMax < 0 ||
(NumInstrumented >= ClDebugMin && NumInstrumented <= ClDebugMax)) {
- if (isa<StoreInst>(Inst) || isa<LoadInst>(Inst))
- instrumentMop(Inst);
+ if (isInterestingMemoryAccess(Inst, &IsWrite))
+ instrumentMop(AFC, Inst);
else
- instrumentMemIntrinsic(cast<MemIntrinsic>(Inst));
+ instrumentMemIntrinsic(AFC, cast<MemIntrinsic>(Inst));
}
NumInstrumented++;
}
+ // Create PHI nodes and crash callbacks if we are merging crash callbacks.
+ if (NumInstrumented) {
+ for (size_t IsWrite = 0; IsWrite <= 1; IsWrite++) {
+ for (size_t AccessSizeIndex = 0; AccessSizeIndex < kNumberOfAccessSizes;
+ AccessSizeIndex++) {
+ BasicBlock *BB = AFC.CrashBlock[IsWrite][AccessSizeIndex];
+ if (!BB) continue;
+ assert(ClMergeCallbacks);
+ AsanFunctionContext::CrashArgsVec &Args =
+ AFC.CrashArgs[IsWrite][AccessSizeIndex];
+ IRBuilder<> IRB(BB->getFirstNonPHI());
+ size_t n = Args.size();
+ PHINode *PN1 = IRB.CreatePHI(IntptrTy, n);
+ PHINode *PN2 = IRB.CreatePHI(IntptrTy, n);
+ // We need to match crash parameters and the predecessors.
+ for (pred_iterator PI = pred_begin(BB), PE = pred_end(BB);
+ PI != PE; ++PI) {
+ n--;
+ PN1->addIncoming(Args[n].Arg1, *PI);
+ PN2->addIncoming(Args[n].Arg2, *PI);
+ }
+ assert(n == 0);
+ generateCrashCode(BB, PN1, PN2, IsWrite, AccessSizeIndex);
+ }
+ }
+ }
+
DEBUG(dbgs() << F);
bool ChangedStack = poisonStackInFunction(M, F);
- // For each NSObject descendant having a +load method, this method is invoked
- // by the ObjC runtime before any of the static constructors is called.
- // Therefore we need to instrument such methods with a call to __asan_init
- // at the beginning in order to initialize our runtime before any access to
- // the shadow memory.
- // We cannot just ignore these methods, because they may call other
- // instrumented functions.
- if (F.getName().find(" load]") != std::string::npos) {
- IRBuilder<> IRB(F.begin()->begin());
- IRB.CreateCall(AsanInitFunction);
+ // We must unpoison the stack before every NoReturn call (throw, _exit, etc).
+ // See e.g. http://code.google.com/p/address-sanitizer/issues/detail?id=37
+ for (size_t i = 0, n = NoReturnCalls.size(); i != n; i++) {
+ Instruction *CI = NoReturnCalls[i];
+ IRBuilder<> IRB(CI);
+ IRB.CreateCall(M.getOrInsertFunction(kAsanHandleNoReturnName,
+ IRB.getVoidTy(), NULL));
}
- return NumInstrumented > 0 || ChangedStack;
+ return NumInstrumented > 0 || ChangedStack || !NoReturnCalls.empty();
}
static uint64_t ValueForPoison(uint64_t PoisonByte, size_t ShadowRedzoneSize) {
if (ShadowRedzoneSize == 4)
return (PoisonByte << 24) + (PoisonByte << 16) +
(PoisonByte << 8) + (PoisonByte);
- assert(0 && "ShadowRedzoneSize is either 1, 2 or 4");
- return 0;
+ llvm_unreachable("ShadowRedzoneSize is either 1, 2 or 4");
}
static void PoisonShadowPartialRightRedzone(uint8_t *Shadow,
BasicBlock &BB = *FI;
for (BasicBlock::iterator BI = BB.begin(), BE = BB.end();
BI != BE; ++BI) {
- if (LooksLikeCodeInBug11395(BI)) return false;
if (isa<ReturnInst>(BI)) {
RetVec.push_back(BI);
continue;
return true;
}
-
-BlackList::BlackList(const std::string &Path) {
- Functions = NULL;
- const char *kFunPrefix = "fun:";
- if (!ClBlackListFile.size()) return;
- std::string Fun;
-
- OwningPtr<MemoryBuffer> File;
- if (error_code EC = MemoryBuffer::getFile(ClBlackListFile.c_str(), File)) {
- report_fatal_error("Can't open blacklist file " + ClBlackListFile + ": " +
- EC.message());
- }
- MemoryBuffer *Buff = File.take();
- const char *Data = Buff->getBufferStart();
- size_t DataLen = Buff->getBufferSize();
- SmallVector<StringRef, 16> Lines;
- SplitString(StringRef(Data, DataLen), Lines, "\n\r");
- for (size_t i = 0, numLines = Lines.size(); i < numLines; i++) {
- if (Lines[i].startswith(kFunPrefix)) {
- std::string ThisFunc = Lines[i].substr(strlen(kFunPrefix));
- std::string ThisFuncRE;
- // add ThisFunc replacing * with .*
- for (size_t j = 0, n = ThisFunc.size(); j < n; j++) {
- if (ThisFunc[j] == '*')
- ThisFuncRE += '.';
- ThisFuncRE += ThisFunc[j];
- }
- // Check that the regexp is valid.
- Regex CheckRE(ThisFuncRE);
- std::string Error;
- if (!CheckRE.isValid(Error))
- report_fatal_error("malformed blacklist regex: " + ThisFunc +
- ": " + Error);
- // Append to the final regexp.
- if (Fun.size())
- Fun += "|";
- Fun += ThisFuncRE;
- }
- }
- if (Fun.size()) {
- Functions = new Regex(Fun);
- }
-}
-
-bool BlackList::isIn(const Function &F) {
- if (Functions) {
- bool Res = Functions->match(F.getName());
- return Res;
- }
- return false;
-}
projects / oota-llvm.git / blobdiff