Skip to content

Commit

Permalink
[asan] Fix dead stripping of globals on Linux.
Browse files Browse the repository at this point in the history 
Use a combination of !associated, comdat, @llvm.compiler.used and
custom sections to allow dead stripping of globals and their asan
metadata. Sometimes.

Currently this works on LLD, which supports SHF_LINK_ORDER with
sh_link pointing to the associated section.

This also works on BFD, which seems to treat comdats as
all-or-nothing with respect to linker GC. There is a weird quirk
where the "first" global in each link is never GC-ed because of the
section symbols.

At this moment it does not work on Gold (as in the globals are never
stripped).

This is a re-land of r298158 rebased on D31358. This time,
asan.module_ctor is put in a comdat as well to avoid quadratic
behavior in Gold.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299697 91177308-0d34-0410-b5e6-96231b3b80d8
  • Loading branch information
@eugenis
eugenis committed Apr 6, 2017
1 parent 9d13481 commit 7bfedca
Show file tree
Hide file tree
Showing 7 changed files with 163 additions and 48 deletions.
11 changes: 11 additions & 0 deletions include/llvm/Transforms/Utils/ModuleUtils.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -84,6 +84,17 @@ void appendToCompilerUsed(Module &M, ArrayRef<GlobalValue *> Values);
void filterDeadComdatFunctions(
Module &M, SmallVectorImpl<Function *> &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
35 changes: 2 additions & 33 deletions lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,42 +30,11 @@
#include "llvm/Transforms/IPO.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;
for (auto &GV : M->global_values()) {
if (GV.isDeclaration() || GV.getName().startswith("llvm.") ||
!GV.hasExternalLinkage())
continue;
ExportsSymbols = true;
Md5.update(GV.getName());
Md5.update(ArrayRef<uint8_t>{0});
}

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) {
Expand Down Expand Up @@ -237,7 +206,7 @@ void forEachVirtualFunction(Constant *C, function_ref<void(Function *)> Fn) {
void splitAndWriteThinLTOBitcode(
raw_ostream &OS, raw_ostream *ThinLinkOS,
function_ref<AAResults &(Function &)> 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.
Expand Down
109 changes: 103 additions & 6 deletions lib/Transforms/Instrumentation/AddressSanitizer.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,10 @@ static const char *const kAsanRegisterImageGlobalsName =
"__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";
Expand All @@ -120,8 +124,11 @@ static const char *const kAsanPoisonStackMemoryName =
"__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";
Expand Down Expand Up @@ -610,6 +617,10 @@ class AddressSanitizerModule : public ModulePass {
void InstrumentGlobalsCOFF(IRBuilder<> &IRB, Module &M,
ArrayRef<GlobalVariable *> ExtendedGlobals,
ArrayRef<Constant *> MetadataInitializers);
void InstrumentGlobalsELF(IRBuilder<> &IRB, Module &M,
ArrayRef<GlobalVariable *> ExtendedGlobals,
ArrayRef<Constant *> MetadataInitializers,
const std::string &UniqueModuleId);
void InstrumentGlobalsMachO(IRBuilder<> &IRB, Module &M,
ArrayRef<GlobalVariable *> ExtendedGlobals,
ArrayRef<Constant *> MetadataInitializers);
Expand All @@ -620,7 +631,8 @@ class AddressSanitizerModule : public ModulePass {

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);
Expand All @@ -645,6 +657,8 @@ class AddressSanitizerModule : public ModulePass {
Function *AsanUnregisterGlobals;
Function *AsanRegisterImageGlobals;
Function *AsanUnregisterImageGlobals;
Function *AsanRegisterElfGlobals;
Function *AsanUnregisterElfGlobals;

Function *AsanCtorFunction = nullptr;
Function *AsanDtorFunction = nullptr;
Expand Down Expand Up @@ -1602,12 +1616,22 @@ void AddressSanitizerModule::initializeCallbacks(Module &M) {
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) {
Expand All @@ -1617,7 +1641,15 @@ void AddressSanitizerModule::SetComdatForGlobalMetadata(
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);
Expand Down Expand Up @@ -1671,10 +1703,69 @@ void AddressSanitizerModule::InstrumentGlobalsCOFF(
"global metadata will not be padded appropriately");
Metadata->setAlignment(SizeOfGlobalStruct);

SetComdatForGlobalMetadata(G, Metadata);
SetComdatForGlobalMetadata(G, Metadata, "");
}
}

void AddressSanitizerModule::InstrumentGlobalsELF(
IRBuilder<> &IRB, Module &M, ArrayRef<GlobalVariable *> ExtendedGlobals,
ArrayRef<Constant *> MetadataInitializers,
const std::string &UniqueModuleId) {
assert(ExtendedGlobals.size() == MetadataInitializers.size());

SmallVector<GlobalValue *, 16> 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(
IRBuilder<> &IRB, Module &M, ArrayRef<GlobalVariable *> ExtendedGlobals,
ArrayRef<Constant *> MetadataInitializers) {
Expand Down Expand Up @@ -1923,7 +2014,13 @@ bool AddressSanitizerModule::InstrumentGlobals(IRBuilder<> &IRB, Module &M, bool
Initializers[i] = Initializer;
}

if (TargetTriple.isOSBinFormatCOFF()) {
std::string ELFUniqueModuleId =
TargetTriple.isOSBinFormatELF() ? getUniqueModuleId(&M) : "";

if (!ELFUniqueModuleId.empty()) {
InstrumentGlobalsELF(IRB, M, NewGlobals, Initializers, ELFUniqueModuleId);
*CtorComdat = true;
} else if (TargetTriple.isOSBinFormatCOFF()) {
InstrumentGlobalsCOFF(IRB, M, NewGlobals, Initializers);
} else if (ShouldUseMachOGlobalsSection()) {
InstrumentGlobalsMachO(IRB, M, NewGlobals, Initializers);
Expand Down
32 changes: 32 additions & 0 deletions lib/Transforms/Utils/ModuleUtils.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -237,3 +237,35 @@ void llvm::filterDeadComdatFunctions(
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<uint8_t>{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();
}
14 changes: 10 additions & 4 deletions test/Instrumentation/AddressSanitizer/global_metadata.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -12,17 +12,23 @@ target triple = "x86_64-unknown-linux-gnu"
@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" {
Expand Down
6 changes: 3 additions & 3 deletions test/Instrumentation/AddressSanitizer/global_metadata_darwin.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -26,16 +26,16 @@ target triple = "x86_64-apple-macosx10.11.0"
; 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
4 changes: 2 additions & 2 deletions test/Instrumentation/AddressSanitizer/instrument_global.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -73,10 +73,10 @@ entry:

; 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

0 comments on commit 7bfedca

Please sign in to comment.