diff --git a/llvm/include/llvm/Transforms/Utils/ModuleUtils.h b/llvm/include/llvm/Transforms/Utils/ModuleUtils.h --- a/llvm/include/llvm/Transforms/Utils/ModuleUtils.h +++ b/llvm/include/llvm/Transforms/Utils/ModuleUtils.h @@ -81,6 +81,17 @@ void filterDeadComdatFunctions( Module &M, SmallVectorImpl &DeadComdatFunctions); +/// \brief Produce a unique identifier for this module by taking the MD5 sum of +/// the names of the module's strong external symbols. +/// +/// This identifier is normally guaranteed to be unique, or the program would +/// fail to link due to multiply defined symbols. +/// +/// If the module has no strong external symbols (such a module may still have a +/// semantic effect if it performs global initialization), we cannot produce a +/// unique identifier for this module, so we return the empty string. +std::string getUniqueModuleId(Module *M); + } // End llvm namespace #endif // LLVM_TRANSFORMS_UTILS_MODULEUTILS_H diff --git a/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp b/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp --- a/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp +++ b/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp @@ -28,51 +28,11 @@ #include "llvm/Support/ScopedPrinter.h" #include "llvm/Transforms/IPO/FunctionAttrs.h" #include "llvm/Transforms/Utils/Cloning.h" +#include "llvm/Transforms/Utils/ModuleUtils.h" using namespace llvm; namespace { -// Produce a unique identifier for this module by taking the MD5 sum of the -// names of the module's strong external symbols. This identifier is -// normally guaranteed to be unique, or the program would fail to link due to -// multiply defined symbols. -// -// If the module has no strong external symbols (such a module may still have a -// semantic effect if it performs global initialization), we cannot produce a -// unique identifier for this module, so we return the empty string, which -// causes the entire module to be written as a regular LTO module. -std::string getModuleId(Module *M) { - MD5 Md5; - bool ExportsSymbols = false; - auto AddGlobal = [&](GlobalValue &GV) { - if (GV.isDeclaration() || GV.getName().startswith("llvm.") || - !GV.hasExternalLinkage()) - return; - ExportsSymbols = true; - Md5.update(GV.getName()); - Md5.update(ArrayRef{0}); - }; - - for (auto &F : *M) - AddGlobal(F); - for (auto &GV : M->globals()) - AddGlobal(GV); - for (auto &GA : M->aliases()) - AddGlobal(GA); - for (auto &IF : M->ifuncs()) - AddGlobal(IF); - - if (!ExportsSymbols) - return ""; - - MD5::MD5Result R; - Md5.final(R); - - SmallString<32> Str; - MD5::stringifyResult(R, Str); - return ("$" + Str).str(); -} - // Promote each local-linkage entity defined by ExportM and used by ImportM by // changing visibility and appending the given ModuleId. void promoteInternals(Module &ExportM, Module &ImportM, StringRef ModuleId) { @@ -253,7 +213,7 @@ void splitAndWriteThinLTOBitcode( raw_ostream &OS, function_ref AARGetter, Module &M) { - std::string ModuleId = getModuleId(&M); + std::string ModuleId = getUniqueModuleId(&M); if (ModuleId.empty()) { // We couldn't generate a module ID for this module, just write it out as a // regular LTO module. diff --git a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp --- a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp +++ b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp @@ -101,6 +101,10 @@ "__asan_register_image_globals"; static const char *const kAsanUnregisterImageGlobalsName = "__asan_unregister_image_globals"; +static const char *const kAsanRegisterElfGlobalsName = + "__asan_register_elf_globals"; +static const char *const kAsanUnregisterElfGlobalsName = + "__asan_unregister_elf_globals"; static const char *const kAsanPoisonGlobalsName = "__asan_before_dynamic_init"; static const char *const kAsanUnpoisonGlobalsName = "__asan_after_dynamic_init"; static const char *const kAsanInitName = "__asan_init"; @@ -120,8 +124,11 @@ "__asan_poison_stack_memory"; static const char *const kAsanUnpoisonStackMemoryName = "__asan_unpoison_stack_memory"; + +// ASan version script has __asan_* wildcard. Triple underscore prevents a +// linker (gold) warning about attempting to export a local symbol. static const char *const kAsanGlobalsRegisteredFlagName = - "__asan_globals_registered"; + "___asan_globals_registered"; static const char *const kAsanOptionDetectUseAfterReturn = "__asan_option_detect_stack_use_after_return"; @@ -612,6 +619,10 @@ void InstrumentGlobalsCOFF(IRBuilder<> &IRB, Module &M, ArrayRef ExtendedGlobals, ArrayRef MetadataInitializers); + void InstrumentGlobalsELF(IRBuilder<> &IRB, Module &M, + ArrayRef ExtendedGlobals, + ArrayRef MetadataInitializers, + const std::string &UniqueModuleId); void InstrumentGlobalsMachO(IRBuilder<> &IRB, Module &M, ArrayRef ExtendedGlobals, ArrayRef MetadataInitializers); @@ -622,7 +633,8 @@ GlobalVariable *CreateMetadataGlobal(Module &M, Constant *Initializer, StringRef OriginalName); - void SetComdatForGlobalMetadata(GlobalVariable *G, GlobalVariable *Metadata); + void SetComdatForGlobalMetadata(GlobalVariable *G, GlobalVariable *Metadata, + StringRef InternalSuffix); IRBuilder<> CreateAsanModuleDtor(Module &M); bool ShouldInstrumentGlobal(GlobalVariable *G); @@ -647,6 +659,8 @@ Function *AsanUnregisterGlobals; Function *AsanRegisterImageGlobals; Function *AsanUnregisterImageGlobals; + Function *AsanRegisterElfGlobals; + Function *AsanUnregisterElfGlobals; }; // Stack poisoning does not play well with exception handling. @@ -1596,12 +1610,22 @@ checkSanitizerInterfaceFunction(M.getOrInsertFunction( kAsanUnregisterImageGlobalsName, IRB.getVoidTy(), IntptrTy, nullptr)); AsanUnregisterImageGlobals->setLinkage(Function::ExternalLinkage); + + AsanRegisterElfGlobals = checkSanitizerInterfaceFunction( + M.getOrInsertFunction(kAsanRegisterElfGlobalsName, IRB.getVoidTy(), + IntptrTy, IntptrTy, IntptrTy, nullptr)); + AsanRegisterElfGlobals->setLinkage(Function::ExternalLinkage); + + AsanUnregisterElfGlobals = checkSanitizerInterfaceFunction( + M.getOrInsertFunction(kAsanUnregisterElfGlobalsName, IRB.getVoidTy(), + IntptrTy, IntptrTy, IntptrTy, nullptr)); + AsanUnregisterElfGlobals->setLinkage(Function::ExternalLinkage); } // Put the metadata and the instrumented global in the same group. This ensures // that the metadata is discarded if the instrumented global is discarded. void AddressSanitizerModule::SetComdatForGlobalMetadata( - GlobalVariable *G, GlobalVariable *Metadata) { + GlobalVariable *G, GlobalVariable *Metadata, StringRef InternalSuffix) { Module &M = *G->getParent(); Comdat *C = G->getComdat(); if (!C) { @@ -1611,7 +1635,15 @@ assert(G->hasLocalLinkage()); G->setName(Twine(kAsanGenPrefix) + "_anon_global"); } - C = M.getOrInsertComdat(G->getName()); + + if (!InternalSuffix.empty() && G->hasLocalLinkage()) { + std::string Name = G->getName(); + Name += InternalSuffix; + C = M.getOrInsertComdat(Name); + } else { + C = M.getOrInsertComdat(G->getName()); + } + // Make this IMAGE_COMDAT_SELECT_NODUPLICATES on COFF. if (TargetTriple.isOSBinFormatCOFF()) C->setSelectionKind(Comdat::NoDuplicates); @@ -1666,8 +1698,67 @@ "global metadata will not be padded appropriately"); Metadata->setAlignment(SizeOfGlobalStruct); - SetComdatForGlobalMetadata(G, Metadata); + SetComdatForGlobalMetadata(G, Metadata, ""); + } +} + +void AddressSanitizerModule::InstrumentGlobalsELF( + IRBuilder<> &IRB, Module &M, ArrayRef ExtendedGlobals, + ArrayRef MetadataInitializers, + const std::string &UniqueModuleId) { + assert(ExtendedGlobals.size() == MetadataInitializers.size()); + + SmallVector MetadataGlobals(ExtendedGlobals.size()); + for (size_t i = 0; i < ExtendedGlobals.size(); i++) { + GlobalVariable *G = ExtendedGlobals[i]; + GlobalVariable *Metadata = + CreateMetadataGlobal(M, MetadataInitializers[i], G->getName()); + MDNode *MD = MDNode::get(M.getContext(), ValueAsMetadata::get(G)); + Metadata->setMetadata(LLVMContext::MD_associated, MD); + MetadataGlobals[i] = Metadata; + + SetComdatForGlobalMetadata(G, Metadata, UniqueModuleId); } + + // Update llvm.compiler.used, adding the new metadata globals. This is + // needed so that during LTO these variables stay alive. + if (!MetadataGlobals.empty()) + appendToCompilerUsed(M, MetadataGlobals); + + // RegisteredFlag serves two purposes. First, we can pass it to dladdr() + // to look up the loaded image that contains it. Second, we can store in it + // whether registration has already occurred, to prevent duplicate + // registration. + // + // Common linkage ensures that there is only one global per shared library. + GlobalVariable *RegisteredFlag = new GlobalVariable( + M, IntptrTy, false, GlobalVariable::CommonLinkage, + ConstantInt::get(IntptrTy, 0), kAsanGlobalsRegisteredFlagName); + RegisteredFlag->setVisibility(GlobalVariable::HiddenVisibility); + + // Create start and stop symbols. + GlobalVariable *StartELFMetadata = new GlobalVariable( + M, IntptrTy, false, GlobalVariable::ExternalWeakLinkage, nullptr, + "__start_" + getGlobalMetadataSection()); + StartELFMetadata->setVisibility(GlobalVariable::HiddenVisibility); + GlobalVariable *StopELFMetadata = new GlobalVariable( + M, IntptrTy, false, GlobalVariable::ExternalWeakLinkage, nullptr, + "__stop_" + getGlobalMetadataSection()); + StopELFMetadata->setVisibility(GlobalVariable::HiddenVisibility); + + // Create a call to register the globals with the runtime. + IRB.CreateCall(AsanRegisterElfGlobals, + {IRB.CreatePointerCast(RegisteredFlag, IntptrTy), + IRB.CreatePointerCast(StartELFMetadata, IntptrTy), + IRB.CreatePointerCast(StopELFMetadata, IntptrTy)}); + + // We also need to unregister globals at the end, e.g., when a shared library + // gets closed. + IRBuilder<> IRB_Dtor = CreateAsanModuleDtor(M); + IRB_Dtor.CreateCall(AsanUnregisterElfGlobals, + {IRB.CreatePointerCast(RegisteredFlag, IntptrTy), + IRB.CreatePointerCast(StartELFMetadata, IntptrTy), + IRB.CreatePointerCast(StopELFMetadata, IntptrTy)}); } void AddressSanitizerModule::InstrumentGlobalsMachO( @@ -1912,7 +2003,12 @@ Initializers[i] = Initializer; } - if (TargetTriple.isOSBinFormatCOFF()) { + std::string ELFUniqueModuleId = + TargetTriple.isOSBinFormatELF() ? getUniqueModuleId(&M) : ""; + + if (!ELFUniqueModuleId.empty()) { + InstrumentGlobalsELF(IRB, M, NewGlobals, Initializers, ELFUniqueModuleId); + } else if (TargetTriple.isOSBinFormatCOFF()) { InstrumentGlobalsCOFF(IRB, M, NewGlobals, Initializers); } else if (ShouldUseMachOGlobalsSection()) { InstrumentGlobalsMachO(IRB, M, NewGlobals, Initializers); diff --git a/llvm/lib/Transforms/Utils/ModuleUtils.cpp b/llvm/lib/Transforms/Utils/ModuleUtils.cpp --- a/llvm/lib/Transforms/Utils/ModuleUtils.cpp +++ b/llvm/lib/Transforms/Utils/ModuleUtils.cpp @@ -229,3 +229,35 @@ ComdatEntriesCovered.end(); }); } + +std::string llvm::getUniqueModuleId(Module *M) { + MD5 Md5; + bool ExportsSymbols = false; + auto AddGlobal = [&](GlobalValue &GV) { + if (GV.isDeclaration() || GV.getName().startswith("llvm.") || + !GV.hasExternalLinkage()) + return; + ExportsSymbols = true; + Md5.update(GV.getName()); + Md5.update(ArrayRef{0}); + }; + + for (auto &F : *M) + AddGlobal(F); + for (auto &GV : M->globals()) + AddGlobal(GV); + for (auto &GA : M->aliases()) + AddGlobal(GA); + for (auto &IF : M->ifuncs()) + AddGlobal(IF); + + if (!ExportsSymbols) + return ""; + + MD5::MD5Result R; + Md5.final(R); + + SmallString<32> Str; + MD5::stringifyResult(R, Str); + return ("$" + Str).str(); +} diff --git a/llvm/test/Instrumentation/AddressSanitizer/global_metadata.ll b/llvm/test/Instrumentation/AddressSanitizer/global_metadata.ll --- a/llvm/test/Instrumentation/AddressSanitizer/global_metadata.ll +++ b/llvm/test/Instrumentation/AddressSanitizer/global_metadata.ll @@ -12,17 +12,23 @@ @llvm.global_ctors = appending global [1 x { i32, void ()*, i8* }] [{ i32, void ()*, i8* } { i32 65535, void ()* @_GLOBAL__sub_I_asan_globals.cpp, i8* null }] ; Check that globals were instrumented: -; CHECK: @global = global { i32, [60 x i8] } zeroinitializer, align 32 -; CHECK: @.str = internal unnamed_addr constant { [14 x i8], [50 x i8] } { [14 x i8] c"Hello, world!\00", [50 x i8] zeroinitializer }, align 32 + +; CHECK: @global = global { i32, [60 x i8] } zeroinitializer, comdat, align 32 +; CHECK: @.str = internal unnamed_addr constant { [14 x i8], [50 x i8] } { [14 x i8] c"Hello, world!\00", [50 x i8] zeroinitializer }, comdat($".str${{[01-9a-f]+}}"), align 32 ; Check emitted location descriptions: ; CHECK: [[VARNAME:@__asan_gen_.[0-9]+]] = private unnamed_addr constant [7 x i8] c"global\00", align 1 ; CHECK: [[FILENAME:@__asan_gen_.[0-9]+]] = private unnamed_addr constant [22 x i8] c"/tmp/asan-globals.cpp\00", align 1 ; CHECK: [[LOCDESCR:@__asan_gen_.[0-9]+]] = private unnamed_addr constant { [22 x i8]*, i32, i32 } { [22 x i8]* [[FILENAME]], i32 5, i32 5 } +; CHECK: @__asan_global_global = {{.*}}i64 ptrtoint ({ i32, [60 x i8] }* @global to i64){{.*}} section "asan_globals"{{.*}}, !associated +; CHECK: @__asan_global_.str = {{.*}}i64 ptrtoint ({ [14 x i8], [50 x i8] }* @.str to i64){{.*}} section "asan_globals"{{.*}}, !associated + +; The metadata has to be inserted to llvm.compiler.used to avoid being stripped +; during LTO. +; CHECK: @llvm.compiler.used {{.*}} @__asan_global_global {{.*}} section "llvm.metadata" ; Check that location descriptors and global names were passed into __asan_register_globals: -; CHECK: i64 ptrtoint ([7 x i8]* [[VARNAME]] to i64) -; CHECK: i64 ptrtoint ({ [22 x i8]*, i32, i32 }* [[LOCDESCR]] to i64) +; CHECK: call void @__asan_register_elf_globals(i64 ptrtoint (i64* @___asan_globals_registered to i64), i64 ptrtoint (i64* @__start_asan_globals to i64), i64 ptrtoint (i64* @__stop_asan_globals to i64)) ; Function Attrs: nounwind sanitize_address define internal void @__cxx_global_var_init() #0 section ".text.startup" { diff --git a/llvm/test/Instrumentation/AddressSanitizer/global_metadata_darwin.ll b/llvm/test/Instrumentation/AddressSanitizer/global_metadata_darwin.ll --- a/llvm/test/Instrumentation/AddressSanitizer/global_metadata_darwin.ll +++ b/llvm/test/Instrumentation/AddressSanitizer/global_metadata_darwin.ll @@ -26,16 +26,16 @@ ; CHECK: @llvm.compiler.used {{.*}} @__asan_binder_global {{.*}} section "llvm.metadata" ; Test that there is the flag global variable: -; CHECK: @__asan_globals_registered = common hidden global i64 0 +; CHECK: @___asan_globals_registered = common hidden global i64 0 ; Test that __asan_register_image_globals is invoked from the constructor: ; CHECK-LABEL: define internal void @asan.module_ctor ; CHECK-NOT: ret -; CHECK: call void @__asan_register_image_globals(i64 ptrtoint (i64* @__asan_globals_registered to i64)) +; CHECK: call void @__asan_register_image_globals(i64 ptrtoint (i64* @___asan_globals_registered to i64)) ; CHECK: ret ; Test that __asan_unregister_image_globals is invoked from the destructor: ; CHECK-LABEL: define internal void @asan.module_dtor ; CHECK-NOT: ret -; CHECK: call void @__asan_unregister_image_globals(i64 ptrtoint (i64* @__asan_globals_registered to i64)) +; CHECK: call void @__asan_unregister_image_globals(i64 ptrtoint (i64* @___asan_globals_registered to i64)) ; CHECK: ret diff --git a/llvm/test/Instrumentation/AddressSanitizer/instrument_global.ll b/llvm/test/Instrumentation/AddressSanitizer/instrument_global.ll --- a/llvm/test/Instrumentation/AddressSanitizer/instrument_global.ll +++ b/llvm/test/Instrumentation/AddressSanitizer/instrument_global.ll @@ -73,10 +73,10 @@ ; CHECK-LABEL: define internal void @asan.module_ctor ; CHECK-NOT: ret -; CHECK: call void @__asan_register_globals +; CHECK: call void @__asan_register_elf_globals ; CHECK: ret ; CHECK-LABEL: define internal void @asan.module_dtor ; CHECK-NOT: ret -; CHECK: call void @__asan_unregister_globals +; CHECK: call void @__asan_unregister_elf_globals ; CHECK: ret