llvm-project/llvm/lib/LTO/LTOBackend.cpp

453 lines
16 KiB
C++
Raw Normal View History

//===-LTOBackend.cpp - LLVM Link Time Optimizer Backend -------------------===//
//
// 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 "backend" phase of LTO, i.e. it performs
// optimization and code generation on a loaded module. It is generally used
// internally by the LTO class but can also be used independently, for example
// to implement a standalone ThinLTO backend.
//
//===----------------------------------------------------------------------===//
#include "llvm/LTO/LTOBackend.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/CGSCCPassManager.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Bitcode/BitcodeReader.h"
#include "llvm/Bitcode/BitcodeWriter.h"
#include "llvm/IR/LegacyPassManager.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/Verifier.h"
#include "llvm/LTO/LTO.h"
#include "llvm/MC/SubtargetFeature.h"
#include "llvm/Object/ModuleSymbolTable.h"
#include "llvm/Passes/PassBuilder.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/TargetRegistry.h"
#include "llvm/Support/ThreadPool.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Transforms/IPO.h"
#include "llvm/Transforms/IPO/PassManagerBuilder.h"
#include "llvm/Transforms/Scalar/LoopPassManager.h"
#include "llvm/Transforms/Utils/FunctionImportUtils.h"
#include "llvm/Transforms/Utils/SplitModule.h"
using namespace llvm;
using namespace lto;
LLVM_ATTRIBUTE_NORETURN static void reportOpenError(StringRef Path, Twine Msg) {
errs() << "failed to open " << Path << ": " << Msg << '\n';
errs().flush();
exit(1);
}
Error Config::addSaveTemps(std::string OutputFileName,
bool UseInputModulePath) {
ShouldDiscardValueNames = false;
std::error_code EC;
ResolutionFile = llvm::make_unique<raw_fd_ostream>(
OutputFileName + "resolution.txt", EC, sys::fs::OpenFlags::F_Text);
if (EC)
return errorCodeToError(EC);
auto setHook = [&](std::string PathSuffix, ModuleHookFn &Hook) {
// Keep track of the hook provided by the linker, which also needs to run.
ModuleHookFn LinkerHook = Hook;
Hook = [=](unsigned Task, const Module &M) {
// If the linker's hook returned false, we need to pass that result
// through.
if (LinkerHook && !LinkerHook(Task, M))
return false;
std::string PathPrefix;
// If this is the combined module (not a ThinLTO backend compile) or the
// user hasn't requested using the input module's path, emit to a file
// named from the provided OutputFileName with the Task ID appended.
if (M.getModuleIdentifier() == "ld-temp.o" || !UseInputModulePath) {
PathPrefix = OutputFileName + utostr(Task);
} else
PathPrefix = M.getModuleIdentifier();
std::string Path = PathPrefix + "." + PathSuffix + ".bc";
std::error_code EC;
raw_fd_ostream OS(Path, EC, sys::fs::OpenFlags::F_None);
// Because -save-temps is a debugging feature, we report the error
// directly and exit.
if (EC)
reportOpenError(Path, EC.message());
WriteBitcodeToFile(&M, OS, /*ShouldPreserveUseListOrder=*/false);
return true;
};
};
setHook("0.preopt", PreOptModuleHook);
setHook("1.promote", PostPromoteModuleHook);
setHook("2.internalize", PostInternalizeModuleHook);
setHook("3.import", PostImportModuleHook);
setHook("4.opt", PostOptModuleHook);
setHook("5.precodegen", PreCodeGenModuleHook);
CombinedIndexHook = [=](const ModuleSummaryIndex &Index) {
std::string Path = OutputFileName + "index.bc";
std::error_code EC;
raw_fd_ostream OS(Path, EC, sys::fs::OpenFlags::F_None);
// Because -save-temps is a debugging feature, we report the error
// directly and exit.
if (EC)
reportOpenError(Path, EC.message());
WriteIndexToFile(Index, OS);
return true;
};
return Error::success();
}
namespace {
std::unique_ptr<TargetMachine>
createTargetMachine(Config &Conf, const Target *TheTarget, Module &M) {
StringRef TheTriple = M.getTargetTriple();
SubtargetFeatures Features;
Features.getDefaultSubtargetFeatures(Triple(TheTriple));
for (const std::string &A : Conf.MAttrs)
Features.AddFeature(A);
Reloc::Model RelocModel;
if (Conf.RelocModel)
RelocModel = *Conf.RelocModel;
else
RelocModel =
M.getPICLevel() == PICLevel::NotPIC ? Reloc::Static : Reloc::PIC_;
return std::unique_ptr<TargetMachine>(TheTarget->createTargetMachine(
TheTriple, Conf.CPU, Features.getString(), Conf.Options, RelocModel,
Conf.CodeModel, Conf.CGOptLevel));
}
static void runNewPMPasses(Config &Conf, Module &Mod, TargetMachine *TM,
unsigned OptLevel, bool IsThinLTO) {
Optional<PGOOptions> PGOOpt;
if (!Conf.SampleProfile.empty())
PGOOpt = PGOOptions("", "", Conf.SampleProfile, false, true);
PassBuilder PB(TM, PGOOpt);
AAManager AA;
// Parse a custom AA pipeline if asked to.
if (!PB.parseAAPipeline(AA, "default"))
report_fatal_error("Error parsing default AA pipeline");
LoopAnalysisManager LAM(Conf.DebugPassManager);
FunctionAnalysisManager FAM(Conf.DebugPassManager);
CGSCCAnalysisManager CGAM(Conf.DebugPassManager);
ModuleAnalysisManager MAM(Conf.DebugPassManager);
// Register the AA manager first so that our version is the one used.
FAM.registerPass([&] { return std::move(AA); });
// Register all the basic analyses with the managers.
PB.registerModuleAnalyses(MAM);
PB.registerCGSCCAnalyses(CGAM);
PB.registerFunctionAnalyses(FAM);
PB.registerLoopAnalyses(LAM);
PB.crossRegisterProxies(LAM, FAM, CGAM, MAM);
ModulePassManager MPM(Conf.DebugPassManager);
// FIXME (davide): verify the input.
PassBuilder::OptimizationLevel OL;
switch (OptLevel) {
default:
llvm_unreachable("Invalid optimization level");
case 0:
OL = PassBuilder::O0;
break;
case 1:
OL = PassBuilder::O1;
break;
case 2:
OL = PassBuilder::O2;
break;
case 3:
OL = PassBuilder::O3;
break;
}
[PM/ThinLTO] Port the ThinLTO pipeline (both components) to the new PM. Based on the original patch by Davide, but I've adjusted the API exposed to just be different entry points rather than exposing more state parameters. I've factored all the common logic out so that we don't have any duplicate pipelines, we just stitch them together in different ways. I think this makes the build easier to reason about and understand. This adds a direct method for getting the module simplification pipeline as well as a method to get the optimization pipeline. While not my express goal, this seems nice and gives a good place comment about the restrictions that are imposed on them. I did make some minor changes to the way the pipelines are structured here, but hopefully not ones that are significant or controversial: 1) I sunk the PGO indirect call promotion to only be run when we have PGO enabled (or as part of the special ThinLTO pipeline). 2) I made the extra GlobalOpt run in ThinLTO just happen all the time and at a slightly more powerful place (before we remove available externaly functions). This seems like general goodness and not a big compile time sink, so it didn't make sense to *only* use it in ThinLTO. Fewer differences in the pipeline makes everything simpler IMO. 3) I hoisted the ThinLTO stop point pre-link above the the RPO function attr inference. The RPO inference won't infer anything terribly meaningful pre-link (recursiveness?) so it didn't make a lot of sense. But if the placement of RPO inference starts to matter, we should move it to the canonicalization phase anyways which seems like a better place for it (and there is a FIXME to this effect!). But that seemed a bridge too far for this patch. If we ever need to parameterize these pipelines more heavily, we can always sink the logic to helper functions with parameters to keep those parameters out of the public API. But the changes above seemed minor that we could possible get away without the parameters entirely. I added support for parsing 'thinlto' and 'thinlto-pre-link' names in pass pipelines to make it easy to test these routines and play with them in larger pipelines. I also added a really basic manifest of passes test that will show exactly how the pipelines behave and work as well as making updates to them clear. Lastly, this factoring does introduce a nesting layer of module pass managers in the default pipeline. I don't think this is a big deal and the flexibility of decoupling the pipelines seems easily worth it. Differential Revision: https://reviews.llvm.org/D33540 llvm-svn: 304407
2017-06-01 19:39:39 +08:00
if (IsThinLTO)
MPM = PB.buildThinLTODefaultPipeline(OL, Conf.DebugPassManager);
[PM/ThinLTO] Port the ThinLTO pipeline (both components) to the new PM. Based on the original patch by Davide, but I've adjusted the API exposed to just be different entry points rather than exposing more state parameters. I've factored all the common logic out so that we don't have any duplicate pipelines, we just stitch them together in different ways. I think this makes the build easier to reason about and understand. This adds a direct method for getting the module simplification pipeline as well as a method to get the optimization pipeline. While not my express goal, this seems nice and gives a good place comment about the restrictions that are imposed on them. I did make some minor changes to the way the pipelines are structured here, but hopefully not ones that are significant or controversial: 1) I sunk the PGO indirect call promotion to only be run when we have PGO enabled (or as part of the special ThinLTO pipeline). 2) I made the extra GlobalOpt run in ThinLTO just happen all the time and at a slightly more powerful place (before we remove available externaly functions). This seems like general goodness and not a big compile time sink, so it didn't make sense to *only* use it in ThinLTO. Fewer differences in the pipeline makes everything simpler IMO. 3) I hoisted the ThinLTO stop point pre-link above the the RPO function attr inference. The RPO inference won't infer anything terribly meaningful pre-link (recursiveness?) so it didn't make a lot of sense. But if the placement of RPO inference starts to matter, we should move it to the canonicalization phase anyways which seems like a better place for it (and there is a FIXME to this effect!). But that seemed a bridge too far for this patch. If we ever need to parameterize these pipelines more heavily, we can always sink the logic to helper functions with parameters to keep those parameters out of the public API. But the changes above seemed minor that we could possible get away without the parameters entirely. I added support for parsing 'thinlto' and 'thinlto-pre-link' names in pass pipelines to make it easy to test these routines and play with them in larger pipelines. I also added a really basic manifest of passes test that will show exactly how the pipelines behave and work as well as making updates to them clear. Lastly, this factoring does introduce a nesting layer of module pass managers in the default pipeline. I don't think this is a big deal and the flexibility of decoupling the pipelines seems easily worth it. Differential Revision: https://reviews.llvm.org/D33540 llvm-svn: 304407
2017-06-01 19:39:39 +08:00
else
MPM = PB.buildLTODefaultPipeline(OL, Conf.DebugPassManager);
MPM.run(Mod, MAM);
// FIXME (davide): verify the output.
}
static void runNewPMCustomPasses(Module &Mod, TargetMachine *TM,
std::string PipelineDesc,
std::string AAPipelineDesc,
bool DisableVerify) {
PassBuilder PB(TM);
AAManager AA;
// Parse a custom AA pipeline if asked to.
if (!AAPipelineDesc.empty())
if (!PB.parseAAPipeline(AA, AAPipelineDesc))
report_fatal_error("unable to parse AA pipeline description: " +
AAPipelineDesc);
LoopAnalysisManager LAM;
FunctionAnalysisManager FAM;
CGSCCAnalysisManager CGAM;
ModuleAnalysisManager MAM;
// Register the AA manager first so that our version is the one used.
FAM.registerPass([&] { return std::move(AA); });
// Register all the basic analyses with the managers.
PB.registerModuleAnalyses(MAM);
PB.registerCGSCCAnalyses(CGAM);
PB.registerFunctionAnalyses(FAM);
PB.registerLoopAnalyses(LAM);
PB.crossRegisterProxies(LAM, FAM, CGAM, MAM);
ModulePassManager MPM;
// Always verify the input.
MPM.addPass(VerifierPass());
// Now, add all the passes we've been requested to.
if (!PB.parsePassPipeline(MPM, PipelineDesc))
report_fatal_error("unable to parse pass pipeline description: " +
PipelineDesc);
if (!DisableVerify)
MPM.addPass(VerifierPass());
MPM.run(Mod, MAM);
}
static void runOldPMPasses(Config &Conf, Module &Mod, TargetMachine *TM,
bool IsThinLTO, ModuleSummaryIndex *ExportSummary,
const ModuleSummaryIndex *ImportSummary) {
legacy::PassManager passes;
passes.add(createTargetTransformInfoWrapperPass(TM->getTargetIRAnalysis()));
PassManagerBuilder PMB;
PMB.LibraryInfo = new TargetLibraryInfoImpl(Triple(TM->getTargetTriple()));
PMB.Inliner = createFunctionInliningPass();
PMB.ExportSummary = ExportSummary;
PMB.ImportSummary = ImportSummary;
// Unconditionally verify input since it is not verified before this
// point and has unknown origin.
PMB.VerifyInput = true;
PMB.VerifyOutput = !Conf.DisableVerify;
PMB.LoopVectorize = true;
PMB.SLPVectorize = true;
PMB.OptLevel = Conf.OptLevel;
PMB.PGOSampleUse = Conf.SampleProfile;
if (IsThinLTO)
PMB.populateThinLTOPassManager(passes);
else
PMB.populateLTOPassManager(passes);
passes.run(Mod);
}
bool opt(Config &Conf, TargetMachine *TM, unsigned Task, Module &Mod,
bool IsThinLTO, ModuleSummaryIndex *ExportSummary,
const ModuleSummaryIndex *ImportSummary) {
// FIXME: Plumb the combined index into the new pass manager.
if (!Conf.OptPipeline.empty())
runNewPMCustomPasses(Mod, TM, Conf.OptPipeline, Conf.AAPipeline,
Conf.DisableVerify);
else if (Conf.UseNewPM)
runNewPMPasses(Conf, Mod, TM, Conf.OptLevel, IsThinLTO);
else
runOldPMPasses(Conf, Mod, TM, IsThinLTO, ExportSummary, ImportSummary);
return !Conf.PostOptModuleHook || Conf.PostOptModuleHook(Task, Mod);
}
void codegen(Config &Conf, TargetMachine *TM, AddStreamFn AddStream,
unsigned Task, Module &Mod) {
if (Conf.PreCodeGenModuleHook && !Conf.PreCodeGenModuleHook(Task, Mod))
return;
auto Stream = AddStream(Task);
legacy::PassManager CodeGenPasses;
if (TM->addPassesToEmitFile(CodeGenPasses, *Stream->OS, Conf.CGFileType))
report_fatal_error("Failed to setup codegen");
CodeGenPasses.run(Mod);
}
void splitCodeGen(Config &C, TargetMachine *TM, AddStreamFn AddStream,
unsigned ParallelCodeGenParallelismLevel,
std::unique_ptr<Module> Mod) {
ThreadPool CodegenThreadPool(ParallelCodeGenParallelismLevel);
unsigned ThreadCount = 0;
const Target *T = &TM->getTarget();
SplitModule(
std::move(Mod), ParallelCodeGenParallelismLevel,
[&](std::unique_ptr<Module> MPart) {
// We want to clone the module in a new context to multi-thread the
// codegen. We do it by serializing partition modules to bitcode
// (while still on the main thread, in order to avoid data races) and
// spinning up new threads which deserialize the partitions into
// separate contexts.
// FIXME: Provide a more direct way to do this in LLVM.
SmallString<0> BC;
raw_svector_ostream BCOS(BC);
WriteBitcodeToFile(MPart.get(), BCOS);
// Enqueue the task
CodegenThreadPool.async(
[&](const SmallString<0> &BC, unsigned ThreadId) {
LTOLLVMContext Ctx(C);
Expected<std::unique_ptr<Module>> MOrErr = parseBitcodeFile(
MemoryBufferRef(StringRef(BC.data(), BC.size()), "ld-temp.o"),
Ctx);
if (!MOrErr)
report_fatal_error("Failed to read bitcode");
std::unique_ptr<Module> MPartInCtx = std::move(MOrErr.get());
std::unique_ptr<TargetMachine> TM =
createTargetMachine(C, T, *MPartInCtx);
codegen(C, TM.get(), AddStream, ThreadId, *MPartInCtx);
},
// Pass BC using std::move to ensure that it get moved rather than
// copied into the thread's context.
std::move(BC), ThreadCount++);
},
false);
// Because the inner lambda (which runs in a worker thread) captures our local
// variables, we need to wait for the worker threads to terminate before we
// can leave the function scope.
CodegenThreadPool.wait();
}
Expected<const Target *> initAndLookupTarget(Config &C, Module &Mod) {
if (!C.OverrideTriple.empty())
Mod.setTargetTriple(C.OverrideTriple);
else if (Mod.getTargetTriple().empty())
Mod.setTargetTriple(C.DefaultTriple);
std::string Msg;
const Target *T = TargetRegistry::lookupTarget(Mod.getTargetTriple(), Msg);
if (!T)
return make_error<StringError>(Msg, inconvertibleErrorCode());
return T;
}
}
static void
finalizeOptimizationRemarks(std::unique_ptr<tool_output_file> DiagOutputFile) {
// Make sure we flush the diagnostic remarks file in case the linker doesn't
// call the global destructors before exiting.
if (!DiagOutputFile)
return;
DiagOutputFile->keep();
DiagOutputFile->os().flush();
}
Error lto::backend(Config &C, AddStreamFn AddStream,
unsigned ParallelCodeGenParallelismLevel,
std::unique_ptr<Module> Mod,
ModuleSummaryIndex &CombinedIndex) {
Expected<const Target *> TOrErr = initAndLookupTarget(C, *Mod);
if (!TOrErr)
return TOrErr.takeError();
std::unique_ptr<TargetMachine> TM = createTargetMachine(C, *TOrErr, *Mod);
// Setup optimization remarks.
auto DiagFileOrErr = lto::setupOptimizationRemarks(
Mod->getContext(), C.RemarksFilename, C.RemarksWithHotness);
if (!DiagFileOrErr)
return DiagFileOrErr.takeError();
auto DiagnosticOutputFile = std::move(*DiagFileOrErr);
if (!C.CodeGenOnly) {
if (!opt(C, TM.get(), 0, *Mod, /*IsThinLTO=*/false,
/*ExportSummary=*/&CombinedIndex, /*ImportSummary=*/nullptr)) {
finalizeOptimizationRemarks(std::move(DiagnosticOutputFile));
return Error::success();
}
}
if (ParallelCodeGenParallelismLevel == 1) {
codegen(C, TM.get(), AddStream, 0, *Mod);
} else {
splitCodeGen(C, TM.get(), AddStream, ParallelCodeGenParallelismLevel,
std::move(Mod));
}
finalizeOptimizationRemarks(std::move(DiagnosticOutputFile));
return Error::success();
}
Error lto::thinBackend(Config &Conf, unsigned Task, AddStreamFn AddStream,
Module &Mod, const ModuleSummaryIndex &CombinedIndex,
const FunctionImporter::ImportMapTy &ImportList,
const GVSummaryMapTy &DefinedGlobals,
MapVector<StringRef, BitcodeModule> &ModuleMap) {
Expected<const Target *> TOrErr = initAndLookupTarget(Conf, Mod);
if (!TOrErr)
return TOrErr.takeError();
std::unique_ptr<TargetMachine> TM = createTargetMachine(Conf, *TOrErr, Mod);
if (Conf.CodeGenOnly) {
codegen(Conf, TM.get(), AddStream, Task, Mod);
return Error::success();
}
if (Conf.PreOptModuleHook && !Conf.PreOptModuleHook(Task, Mod))
return Error::success();
renameModuleForThinLTO(Mod, CombinedIndex);
thinLTOResolveWeakForLinkerModule(Mod, DefinedGlobals);
if (Conf.PostPromoteModuleHook && !Conf.PostPromoteModuleHook(Task, Mod))
return Error::success();
if (!DefinedGlobals.empty())
thinLTOInternalizeModule(Mod, DefinedGlobals);
if (Conf.PostInternalizeModuleHook &&
!Conf.PostInternalizeModuleHook(Task, Mod))
return Error::success();
auto ModuleLoader = [&](StringRef Identifier) {
assert(Mod.getContext().isODRUniquingDebugTypes() &&
"ODR Type uniquing should be enabled on the context");
auto I = ModuleMap.find(Identifier);
assert(I != ModuleMap.end());
return I->second.getLazyModule(Mod.getContext(),
/*ShouldLazyLoadMetadata=*/true,
/*IsImporting*/ true);
};
FunctionImporter Importer(CombinedIndex, ModuleLoader);
if (Error Err = Importer.importFunctions(Mod, ImportList).takeError())
return Err;
if (Conf.PostImportModuleHook && !Conf.PostImportModuleHook(Task, Mod))
return Error::success();
if (!opt(Conf, TM.get(), Task, Mod, /*IsThinLTO=*/true,
/*ExportSummary=*/nullptr, /*ImportSummary=*/&CombinedIndex))
return Error::success();
codegen(Conf, TM.get(), AddStream, Task, Mod);
return Error::success();
}