llvm/tools/lto/LTOCodeGenerator.cpp

432 lines
14 KiB
C++
Raw Normal View History

//===-LTOCodeGenerator.cpp - LLVM Link Time Optimizer ---------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements the Link Time Optimization library. This library is
// intended to be used by linker to optimize code at link time.
//
//===----------------------------------------------------------------------===//
#include "LTOCodeGenerator.h"
#include "LTOModule.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/Analysis/Passes.h"
#include "llvm/Analysis/Verifier.h"
#include "llvm/Bitcode/ReaderWriter.h"
#include "llvm/Config/config.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/Linker.h"
#include "llvm/MC/MCAsmInfo.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/SubtargetFeature.h"
#include "llvm/PassManager.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/FormattedStream.h"
#include "llvm/Support/Host.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/Signals.h"
#include "llvm/Support/TargetRegistry.h"
#include "llvm/Support/TargetSelect.h"
#include "llvm/Support/ToolOutputFile.h"
#include "llvm/Support/system_error.h"
#include "llvm/Target/Mangler.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Target/TargetOptions.h"
#include "llvm/Target/TargetRegisterInfo.h"
#include "llvm/Transforms/IPO.h"
#include "llvm/Transforms/IPO/PassManagerBuilder.h"
#include "llvm/Transforms/ObjCARC.h"
using namespace llvm;
static cl::opt<bool>
DisableOpt("disable-opt", cl::init(false),
cl::desc("Do not run any optimization passes"));
static cl::opt<bool>
DisableInline("disable-inlining", cl::init(false),
cl::desc("Do not run the inliner pass"));
static cl::opt<bool>
DisableGVNLoadPRE("disable-gvn-loadpre", cl::init(false),
cl::desc("Do not run the GVN load PRE pass"));
const char* LTOCodeGenerator::getVersionString() {
#ifdef LLVM_VERSION_INFO
return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
#else
return PACKAGE_NAME " version " PACKAGE_VERSION;
#endif
}
LTOCodeGenerator::LTOCodeGenerator()
: _context(getGlobalContext()),
_linker("ld-temp.o", _context), _target(NULL),
_emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
_codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
_nativeObjectFile(NULL) {
InitializeAllTargets();
InitializeAllTargetMCs();
InitializeAllAsmPrinters();
}
LTOCodeGenerator::~LTOCodeGenerator() {
delete _target;
delete _nativeObjectFile;
for (std::vector<char*>::iterator I = _codegenOptions.begin(),
E = _codegenOptions.end(); I != E; ++I)
free(*I);
}
bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg) {
bool ret = _linker.LinkInModule(mod->getLLVVMModule(), &errMsg);
const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
for (int i = 0, e = undefs.size(); i != e; ++i)
_asmUndefinedRefs[undefs[i]] = 1;
return ret;
}
bool LTOCodeGenerator::setDebugInfo(lto_debug_model debug,
std::string& errMsg) {
switch (debug) {
case LTO_DEBUG_MODEL_NONE:
_emitDwarfDebugInfo = false;
return false;
case LTO_DEBUG_MODEL_DWARF:
_emitDwarfDebugInfo = true;
return false;
}
llvm_unreachable("Unknown debug format!");
}
bool LTOCodeGenerator::setCodePICModel(lto_codegen_model model,
std::string& errMsg) {
switch (model) {
case LTO_CODEGEN_PIC_MODEL_STATIC:
case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
_codeModel = model;
return false;
}
llvm_unreachable("Unknown PIC model!");
}
bool LTOCodeGenerator::writeMergedModules(const char *path,
std::string &errMsg) {
if (determineTarget(errMsg))
return true;
// mark which symbols can not be internalized
applyScopeRestrictions();
// create output file
std::string ErrInfo;
tool_output_file Out(path, ErrInfo,
raw_fd_ostream::F_Binary);
if (!ErrInfo.empty()) {
errMsg = "could not open bitcode file for writing: ";
errMsg += path;
return true;
}
// write bitcode to it
WriteBitcodeToFile(_linker.getModule(), Out.os());
Out.os().close();
if (Out.os().has_error()) {
errMsg = "could not write bitcode file: ";
errMsg += path;
Out.os().clear_error();
return true;
}
Out.keep();
return false;
}
bool LTOCodeGenerator::compile_to_file(const char** name, std::string& errMsg) {
// make unique temp .o file to put generated object file
sys::PathWithStatus uniqueObjPath("lto-llvm.o");
if (uniqueObjPath.createTemporaryFileOnDisk(false, &errMsg)) {
uniqueObjPath.eraseFromDisk();
return true;
}
sys::RemoveFileOnSignal(uniqueObjPath);
// generate object file
bool genResult = false;
tool_output_file objFile(uniqueObjPath.c_str(), errMsg);
if (!errMsg.empty()) {
uniqueObjPath.eraseFromDisk();
return true;
}
genResult = this->generateObjectFile(objFile.os(), errMsg);
objFile.os().close();
if (objFile.os().has_error()) {
objFile.os().clear_error();
uniqueObjPath.eraseFromDisk();
return true;
}
objFile.keep();
if (genResult) {
uniqueObjPath.eraseFromDisk();
return true;
}
_nativeObjectPath = uniqueObjPath.str();
*name = _nativeObjectPath.c_str();
return false;
}
const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg) {
const char *name;
if (compile_to_file(&name, errMsg))
return NULL;
// remove old buffer if compile() called twice
delete _nativeObjectFile;
// read .o file into memory buffer
OwningPtr<MemoryBuffer> BuffPtr;
if (error_code ec = MemoryBuffer::getFile(name, BuffPtr, -1, false)) {
errMsg = ec.message();
sys::Path(_nativeObjectPath).eraseFromDisk();
return NULL;
}
_nativeObjectFile = BuffPtr.take();
// remove temp files
sys::Path(_nativeObjectPath).eraseFromDisk();
// return buffer, unless error
if (_nativeObjectFile == NULL)
return NULL;
*length = _nativeObjectFile->getBufferSize();
return _nativeObjectFile->getBufferStart();
}
bool LTOCodeGenerator::determineTarget(std::string& errMsg) {
if (_target != NULL)
return false;
std::string TripleStr = _linker.getModule()->getTargetTriple();
if (TripleStr.empty())
TripleStr = sys::getDefaultTargetTriple();
llvm::Triple Triple(TripleStr);
// create target machine from info for merged modules
const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg);
if (march == NULL)
return true;
// The relocation model is actually a static member of TargetMachine and
// needs to be set before the TargetMachine is instantiated.
Reloc::Model RelocModel = Reloc::Default;
switch (_codeModel) {
case LTO_CODEGEN_PIC_MODEL_STATIC:
RelocModel = Reloc::Static;
break;
case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
RelocModel = Reloc::PIC_;
break;
case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
RelocModel = Reloc::DynamicNoPIC;
break;
}
// construct LTOModule, hand over ownership of module and target
SubtargetFeatures Features;
Features.getDefaultSubtargetFeatures(Triple);
std::string FeatureStr = Features.getString();
// Set a default CPU for Darwin triples.
if (_mCpu.empty() && Triple.isOSDarwin()) {
if (Triple.getArch() == llvm::Triple::x86_64)
_mCpu = "core2";
else if (Triple.getArch() == llvm::Triple::x86)
_mCpu = "yonah";
}
TargetOptions Options;
LTOModule::getTargetOptions(Options);
_target = march->createTargetMachine(TripleStr, _mCpu, FeatureStr, Options,
RelocModel, CodeModel::Default,
CodeGenOpt::Aggressive);
return false;
}
void LTOCodeGenerator::
applyRestriction(GlobalValue &GV,
std::vector<const char*> &mustPreserveList,
SmallPtrSet<GlobalValue*, 8> &asmUsed,
Mangler &mangler) {
SmallString<64> Buffer;
mangler.getNameWithPrefix(Buffer, &GV, false);
if (GV.isDeclaration())
return;
if (_mustPreserveSymbols.count(Buffer))
mustPreserveList.push_back(GV.getName().data());
if (_asmUndefinedRefs.count(Buffer))
asmUsed.insert(&GV);
}
static void findUsedValues(GlobalVariable *LLVMUsed,
SmallPtrSet<GlobalValue*, 8> &UsedValues) {
if (LLVMUsed == 0) return;
ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer());
for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
if (GlobalValue *GV =
dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
UsedValues.insert(GV);
}
void LTOCodeGenerator::applyScopeRestrictions() {
if (_scopeRestrictionsDone) return;
Module *mergedModule = _linker.getModule();
// Start off with a verification pass.
PassManager passes;
passes.add(createVerifierPass());
// mark which symbols can not be internalized
MCContext Context(*_target->getMCAsmInfo(), *_target->getRegisterInfo(),NULL);
Mangler mangler(Context, *_target->getDataLayout());
std::vector<const char*> mustPreserveList;
SmallPtrSet<GlobalValue*, 8> asmUsed;
for (Module::iterator f = mergedModule->begin(),
e = mergedModule->end(); f != e; ++f)
applyRestriction(*f, mustPreserveList, asmUsed, mangler);
for (Module::global_iterator v = mergedModule->global_begin(),
e = mergedModule->global_end(); v != e; ++v)
applyRestriction(*v, mustPreserveList, asmUsed, mangler);
for (Module::alias_iterator a = mergedModule->alias_begin(),
e = mergedModule->alias_end(); a != e; ++a)
applyRestriction(*a, mustPreserveList, asmUsed, mangler);
GlobalVariable *LLVMCompilerUsed =
mergedModule->getGlobalVariable("llvm.compiler.used");
findUsedValues(LLVMCompilerUsed, asmUsed);
if (LLVMCompilerUsed)
LLVMCompilerUsed->eraseFromParent();
if (!asmUsed.empty()) {
llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(_context);
std::vector<Constant*> asmUsed2;
for (SmallPtrSet<GlobalValue*, 16>::const_iterator i = asmUsed.begin(),
e = asmUsed.end(); i !=e; ++i) {
GlobalValue *GV = *i;
Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
asmUsed2.push_back(c);
}
llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
LLVMCompilerUsed =
new llvm::GlobalVariable(*mergedModule, ATy, false,
llvm::GlobalValue::AppendingLinkage,
llvm::ConstantArray::get(ATy, asmUsed2),
"llvm.compiler.used");
LLVMCompilerUsed->setSection("llvm.metadata");
}
passes.add(createInternalizePass(mustPreserveList));
// apply scope restrictions
passes.run(*mergedModule);
_scopeRestrictionsDone = true;
}
/// Optimize merged modules using various IPO passes
bool LTOCodeGenerator::generateObjectFile(raw_ostream &out,
std::string &errMsg) {
if (this->determineTarget(errMsg))
return true;
Module* mergedModule = _linker.getModule();
// if options were requested, set them
if (!_codegenOptions.empty())
cl::ParseCommandLineOptions(_codegenOptions.size(),
const_cast<char **>(&_codegenOptions[0]));
// mark which symbols can not be internalized
this->applyScopeRestrictions();
// Instantiate the pass manager to organize the passes.
PassManager passes;
// Start off with a verification pass.
passes.add(createVerifierPass());
// Add an appropriate DataLayout instance for this module...
passes.add(new DataLayout(*_target->getDataLayout()));
Switch TargetTransformInfo from an immutable analysis pass that requires a TargetMachine to construct (and thus isn't always available), to an analysis group that supports layered implementations much like AliasAnalysis does. This is a pretty massive change, with a few parts that I was unable to easily separate (sorry), so I'll walk through it. The first step of this conversion was to make TargetTransformInfo an analysis group, and to sink the nonce implementations in ScalarTargetTransformInfo and VectorTargetTranformInfo into a NoTargetTransformInfo pass. This allows other passes to add a hard requirement on TTI, and assume they will always get at least on implementation. The TargetTransformInfo analysis group leverages the delegation chaining trick that AliasAnalysis uses, where the base class for the analysis group delegates to the previous analysis *pass*, allowing all but tho NoFoo analysis passes to only implement the parts of the interfaces they support. It also introduces a new trick where each pass in the group retains a pointer to the top-most pass that has been initialized. This allows passes to implement one API in terms of another API and benefit when some other pass above them in the stack has more precise results for the second API. The second step of this conversion is to create a pass that implements the TargetTransformInfo analysis using the target-independent abstractions in the code generator. This replaces the ScalarTargetTransformImpl and VectorTargetTransformImpl classes in lib/Target with a single pass in lib/CodeGen called BasicTargetTransformInfo. This class actually provides most of the TTI functionality, basing it upon the TargetLowering abstraction and other information in the target independent code generator. The third step of the conversion adds support to all TargetMachines to register custom analysis passes. This allows building those passes with access to TargetLowering or other target-specific classes, and it also allows each target to customize the set of analysis passes desired in the pass manager. The baseline LLVMTargetMachine implements this interface to add the BasicTTI pass to the pass manager, and all of the tools that want to support target-aware TTI passes call this routine on whatever target machine they end up with to add the appropriate passes. The fourth step of the conversion created target-specific TTI analysis passes for the X86 and ARM backends. These passes contain the custom logic that was previously in their extensions of the ScalarTargetTransformInfo and VectorTargetTransformInfo interfaces. I separated them into their own file, as now all of the interface bits are private and they just expose a function to create the pass itself. Then I extended these target machines to set up a custom set of analysis passes, first adding BasicTTI as a fallback, and then adding their customized TTI implementations. The fourth step required logic that was shared between the target independent layer and the specific targets to move to a different interface, as they no longer derive from each other. As a consequence, a helper functions were added to TargetLowering representing the common logic needed both in the target implementation and the codegen implementation of the TTI pass. While technically this is the only change that could have been committed separately, it would have been a nightmare to extract. The final step of the conversion was just to delete all the old boilerplate. This got rid of the ScalarTargetTransformInfo and VectorTargetTransformInfo classes, all of the support in all of the targets for producing instances of them, and all of the support in the tools for manually constructing a pass based around them. Now that TTI is a relatively normal analysis group, two things become straightforward. First, we can sink it into lib/Analysis which is a more natural layer for it to live. Second, clients of this interface can depend on it *always* being available which will simplify their code and behavior. These (and other) simplifications will follow in subsequent commits, this one is clearly big enough. Finally, I'm very aware that much of the comments and documentation needs to be updated. As soon as I had this working, and plausibly well commented, I wanted to get it committed and in front of the build bots. I'll be doing a few passes over documentation later if it sticks. Commits to update DragonEgg and Clang will be made presently. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171681 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-07 01:37:14 +00:00
_target->addAnalysisPasses(passes);
// Enabling internalize here would use its AllButMain variant. It
// keeps only main if it exists and does nothing for libraries. Instead
// we create the pass ourselves with the symbol list provided by the linker.
if (!DisableOpt) {
PassManagerBuilder().populateLTOPassManager(passes,
/*Internalize=*/false,
!DisableInline,
DisableGVNLoadPRE);
}
// Make sure everything is still good.
passes.add(createVerifierPass());
PassManager codeGenPasses;
codeGenPasses.add(new DataLayout(*_target->getDataLayout()));
_target->addAnalysisPasses(codeGenPasses);
formatted_raw_ostream Out(out);
// If the bitcode files contain ARC code and were compiled with optimization,
// the ObjCARCContractPass must be run, so do it unconditionally here.
codeGenPasses.add(createObjCARCContractPass());
if (_target->addPassesToEmitFile(codeGenPasses, Out,
TargetMachine::CGFT_ObjectFile)) {
errMsg = "target file type not supported";
return true;
}
// Run our queue of passes all at once now, efficiently.
passes.run(*mergedModule);
// Run the code generator, and write assembly file
codeGenPasses.run(*mergedModule);
return false; // success
}
/// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
/// LTO problems.
void LTOCodeGenerator::setCodeGenDebugOptions(const char *options) {
for (std::pair<StringRef, StringRef> o = getToken(options);
!o.first.empty(); o = getToken(o.second)) {
// ParseCommandLineOptions() expects argv[0] to be program name. Lazily add
// that.
if (_codegenOptions.empty())
_codegenOptions.push_back(strdup("libLTO"));
_codegenOptions.push_back(strdup(o.first.str().c_str()));
}
}