[clang][llvm][fatlto] Avoid cloning modules in FatLTO (#72180)

#70703 pointed out that
cloning LLVM modules could lead to miscompiles when using FatLTO.

This is due to an existing issue when cloning modules with labels (see
#55991 and #47769). Since this can lead to miscompilation, we can avoid
cloning the LLVM modules, which was desirable anyway.

This patch modifies the EmbedBitcodePass to no longer clone the module
or run an input pipeline over it. Further, it make FatLTO always perform
UnifiedLTO, so we can still defer the Thin/Full LTO decision to
link-time. Lastly, it removes dead/obsolete code related to now defunct
options that do not work with the EmbedBitcodePass implementation any
longer.

clang/lib/CodeGen/BackendUtil.cpp

@@ -996,9 +996,8 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
     }
 
     if (CodeGenOpts.FatLTO) {
-      MPM.addPass(PB.buildFatLTODefaultPipeline(
-          Level, PrepareForThinLTO,
-          PrepareForThinLTO || shouldEmitRegularLTOSummary()));
+      assert(CodeGenOpts.UnifiedLTO && "FatLTO requires UnifiedLTO");
+      MPM.addPass(PB.buildFatLTODefaultPipeline(Level));
     } else if (PrepareForThinLTO) {
       MPM.addPass(PB.buildThinLTOPreLinkDefaultPipeline(Level));
     } else if (PrepareForLTO) {
@@ -1042,7 +1041,6 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
         MPM.addPass(PrintModulePass(*OS, "", CodeGenOpts.EmitLLVMUseLists,
                                     /*EmitLTOSummary=*/true));
       }
-
     } else {
       // Emit a module summary by default for Regular LTO except for ld64
       // targets
@@ -1073,7 +1071,8 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
     if (!TheModule->getModuleFlag("EnableSplitLTOUnit"))
       TheModule->addModuleFlag(llvm::Module::Error, "EnableSplitLTOUnit",
                                uint32_t(CodeGenOpts.EnableSplitLTOUnit));
-    if (CodeGenOpts.UnifiedLTO && !TheModule->getModuleFlag("UnifiedLTO"))
+    // FatLTO always means UnifiedLTO
+    if (!TheModule->getModuleFlag("UnifiedLTO"))
       TheModule->addModuleFlag(llvm::Module::Error, "UnifiedLTO", uint32_t(1));
   }
 

clang/lib/Driver/ToolChains/Clang.cpp

@@ -4862,7 +4862,9 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
   bool UnifiedLTO = false;
   if (IsUsingLTO) {
     UnifiedLTO = Args.hasFlag(options::OPT_funified_lto,
-                              options::OPT_fno_unified_lto, Triple.isPS());
+                              options::OPT_fno_unified_lto, Triple.isPS()) ||
+                 Args.hasFlag(options::OPT_ffat_lto_objects,
+                              options::OPT_fno_fat_lto_objects, false);
     if (UnifiedLTO)
       CmdArgs.push_back("-funified-lto");
   }

clang/lib/Frontend/CompilerInvocation.cpp

@@ -1861,6 +1861,20 @@ bool CompilerInvocation::ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args,
     if (Args.hasArg(OPT_funified_lto))
       Opts.PrepareForThinLTO = true;
   }
+  if (Arg *A = Args.getLastArg(options::OPT_ffat_lto_objects,
+                               options::OPT_fno_fat_lto_objects)) {
+    if (A->getOption().matches(options::OPT_ffat_lto_objects)) {
+      if (Arg *Uni = Args.getLastArg(options::OPT_funified_lto,
+                                     options::OPT_fno_unified_lto)) {
+        if (Uni->getOption().matches(options::OPT_fno_unified_lto))
+          Diags.Report(diag::err_drv_incompatible_options)
+              << A->getAsString(Args) << "-fno-unified-lto";
+      } else
+        Diags.Report(diag::err_drv_argument_only_allowed_with)
+            << A->getAsString(Args) << "-funified-lto";
+    }
+  }
+
   if (Arg *A = Args.getLastArg(OPT_fthinlto_index_EQ)) {
     if (IK.getLanguage() != Language::LLVM_IR)
       Diags.Report(diag::err_drv_argument_only_allowed_with)

clang/test/CodeGen/fat-lto-objects.c

@@ -1,49 +1,47 @@
 // REQUIRES: x86-registered-target
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -fsplit-lto-unit -emit-llvm < %s  | FileCheck %s --check-prefixes=FULL,SPLIT
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=FULL,SPLIT
+// RUN: not %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -fsplit-lto-unit -emit-llvm < %s 2>&1 | FileCheck %s --check-prefixes=NO-UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -fsplit-lto-unit -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=THIN,SPLIT
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=THIN,NOSPLIT
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -funified-lto -ffat-lto-objects -fsplit-lto-unit -emit-llvm < %s  | FileCheck %s --check-prefixes=FULL,SPLIT,UNIFIED
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -funified-lto -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=FULL,NOSPLIT,UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -fsplit-lto-unit -emit-obj < %s -o %t.full.split.o
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -funified-lto -fsplit-lto-unit -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=THIN,SPLIT,UNIFIED
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -funified-lto -ffat-lto-objects -emit-llvm < %s  | FileCheck %s --check-prefixes=THIN,NOSPLIT,UNIFIED
+
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -funified-lto -ffat-lto-objects -fsplit-lto-unit -emit-obj < %s -o %t.full.split.o
 // RUN: llvm-readelf -S %t.full.split.o | FileCheck %s --check-prefixes=ELF
 // RUN: llvm-objcopy --dump-section=.llvm.lto=%t.full.split.bc %t.full.split.o
-// RUN: llvm-dis %t.full.split.bc -o - | FileCheck %s --check-prefixes=FULL,SPLIT,NOUNIFIED
+// RUN: llvm-dis %t.full.split.bc -o - | FileCheck %s --check-prefixes=THIN,SPLIT,UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -emit-obj < %s -o %t.full.nosplit.o
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -funified-lto -ffat-lto-objects -emit-obj < %s -o %t.full.nosplit.o
 // RUN: llvm-readelf -S %t.full.nosplit.o | FileCheck %s --check-prefixes=ELF
 // RUN: llvm-objcopy --dump-section=.llvm.lto=%t.full.nosplit.bc %t.full.nosplit.o
-// RUN: llvm-dis %t.full.nosplit.bc -o - | FileCheck %s --check-prefixes=FULL,NOSPLIT,NOUNIFIED
+// RUN: llvm-dis %t.full.nosplit.bc -o - | FileCheck %s --check-prefixes=THIN,NOSPLIT,UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -fsplit-lto-unit -ffat-lto-objects -emit-obj < %s -o %t.thin.split.o
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -funified-lto -fsplit-lto-unit -ffat-lto-objects -emit-obj < %s -o %t.thin.split.o
 // RUN: llvm-readelf -S %t.thin.split.o | FileCheck %s --check-prefixes=ELF
 // RUN: llvm-objcopy --dump-section=.llvm.lto=%t.thin.split.bc %t.thin.split.o
-// RUN: llvm-dis %t.thin.split.bc -o - | FileCheck %s --check-prefixes=THIN,SPLIT,NOUNIFIED
+// RUN: llvm-dis %t.thin.split.bc -o - | FileCheck %s --check-prefixes=THIN,SPLIT,UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -ffat-lto-objects -emit-obj < %s -o %t.thin.nosplit.o
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -funified-lto -ffat-lto-objects -emit-obj < %s -o %t.thin.nosplit.o
 // RUN: llvm-readelf -S %t.thin.nosplit.o | FileCheck %s --check-prefixes=ELF
 // RUN: llvm-objcopy --dump-section=.llvm.lto=%t.thin.nosplit.bc %t.thin.nosplit.o
-// RUN: llvm-dis %t.thin.nosplit.bc -o -  | FileCheck %s --check-prefixes=THIN,NOSPLIT,NOUNIFIED
-
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=thin -funified-lto -ffat-lto-objects -emit-obj < %s -o %t.unified.o
-// RUN: llvm-readelf -S %t.unified.o | FileCheck %s --check-prefixes=ELF
-// RUN: llvm-objcopy --dump-section=.llvm.lto=%t.unified.bc %t.unified.o
-// RUN: llvm-dis %t.unified.bc -o - | FileCheck %s --check-prefixes=THIN,NOSPLIT,UNIFIED
+// RUN: llvm-dis %t.thin.nosplit.bc -o -  | FileCheck %s --check-prefixes=THIN,NOSPLIT,UNIFIED
 
-// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -ffat-lto-objects -fsplit-lto-unit -S < %s -o - \
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -flto=full -funified-lto -ffat-lto-objects -fsplit-lto-unit -S < %s -o - \
 // RUN: | FileCheck %s --check-prefixes=ASM
 
-/// Check that the ThinLTO metadata is only set false for full LTO.
-// FULL: ![[#]] = !{i32 1, !"ThinLTO", i32 0}
-// THIN-NOT: ![[#]] = !{i32 1, !"ThinLTO", i32 0}
-
 /// Be sure we enable split LTO units correctly under -ffat-lto-objects.
-// SPLIT: ![[#]] = !{i32 1, !"EnableSplitLTOUnit", i32 1}
+//   SPLIT: ![[#]] = !{i32 1, !"EnableSplitLTOUnit", i32 1}
 // NOSPLIT: ![[#]] = !{i32 1, !"EnableSplitLTOUnit", i32 0}
 
-// UNIFIED: ![[#]] = !{i32 1, !"UnifiedLTO", i32 1}
-// NOUNIFIED-NOT: ![[#]] = !{i32 1, !"UnifiedLTO", i32 1}
+/// Check that the ThinLTO metadata is set true for both full and thin LTO, since FatLTO is based on UnifiedLTO.
+//     FULL: ![[#]] = !{i32 1, !"ThinLTO", i32 1}
+// THIN-NOT: ![[#]] = !{i32 1, !"ThinLTO", i32 0}
+
+/// FatLTO always uses UnifiedLTO. It's an error if they aren't set together
+//    UNIFIED: ![[#]] = !{i32 1, !"UnifiedLTO", i32 1}
+// NO-UNIFIED: error: invalid argument '-ffat-lto-objects' only allowed with '-funified-lto'
 
 // ELF: .llvm.lto
 

clang/test/Driver/fat-lto-objects.c

@@ -1,5 +1,6 @@
 // RUN: %clang --target=x86_64-unknown-linux-gnu -flto -ffat-lto-objects -### %s -c 2>&1 | FileCheck %s -check-prefix=CHECK-CC
 // CHECK-CC: -cc1
+// CHECK-CC-SAME: -funified-lto
 // CHECK-CC-SAME: -emit-obj
 // CHECK-CC-SAME: -ffat-lto-objects
 
@@ -15,6 +16,7 @@
 // RUN: %clang --target=x86_64-unknown-linux-gnu -flto -ffat-lto-objects -### %s -S 2>&1 | FileCheck %s -check-prefix=CHECK-CC-S-LTO
 // RUN: %clang --target=x86_64-unknown-linux-gnu -flto -ffat-lto-objects -### %s -S -emit-llvm 2>&1 | FileCheck %s -check-prefix=CHECK-CC-S-LTO
 // CHECK-CC-S-LTO: -cc1
+// CHECK-CC-S-LTO-SAME: -funified-lto
 // CHECK-CC-S-LTO-SAME: -emit-llvm
 // CHECK-CC-S-LTO-SAME: -ffat-lto-objects
 
@@ -32,3 +34,13 @@
 // RUN:   -fuse-ld=lld -fno-lto -ffat-lto-objects -### 2>&1 | FileCheck --check-prefix=NOLTO %s
 // LTO: "--fat-lto-objects"
 // NOLTO-NOT: "--fat-lto-objects"
+
+/// Make sure that incompatible options emit the correct diagnostics, since -ffat-lto-objects requires -funified-lto
+// RUN: %clang -cc1 -triple=x86_64-unknown-linux-gnu -flto -ffat-lto-objects -funified-lto -emit-llvm-only %s  2>&1 | FileCheck %s -check-prefix=UNIFIED --allow-empty
+// UNIFIED-NOT: error:
+
+// RUN: not %clang -cc1 -triple=x86_64-unknown-linux-gnu -flto -ffat-lto-objects -emit-llvm-only %s  2>&1 | FileCheck %s -check-prefix=MISSING_UNIFIED
+// MISSING_UNIFIED: error: invalid argument '-ffat-lto-objects' only allowed with '-funified-lto'
+
+// RUN: not %clang -cc1 -triple=x86_64-unknown-linux-gnu -flto -fno-unified-lto -ffat-lto-objects -emit-llvm-only %s  2>&1 | FileCheck %s -check-prefix=NO-UNIFIED
+// NO-UNIFIED: error: the combination of '-ffat-lto-objects' and '-fno-unified-lto' is incompatible

llvm/docs/FatLTO.rst

@@ -29,30 +29,31 @@ Overview
 Within LLVM, FatLTO is supported by choosing the ``FatLTODefaultPipeline``.
 This pipeline will:
 
-#) Clone the IR module.
-#) Run the pre-link (Thin)LTO pipeline using the cloned module.
+#) Run the pre-link UnifiedLTO pipeline on the current module.
 #) Embed the pre-link bitcode in a special ``.llvm.lto`` section.
-#) Optimize the unmodified copy of the module using the normal compilation pipeline.
+#) Finish optimizing the module using the post-link ThinLTO pipeline.
 #) Emit the object file, including the new ``.llvm.lto`` section.
 
 .. NOTE
 
-   At the time of writing, we conservatively run independent pipelines to
-   generate the bitcode section and the object code, which happen to be
-   identical to those used outside of FatLTO. This results in  compiled
-   artifacts that are identical to those produced by the default and (Thin)LTO
-   pipelines. However, this is not a guarantee, and we reserve the right to
-   change this at any time. Explicitly, users should not rely on the produced
-   bitcode or object code to mach their non-LTO counterparts precisely. They
-   will exhibit similar performance characteristics, but may not be bit-for-bit
-   the same.
+   Previously, we conservatively ran independent pipelines on separate copies
+   of the LLVM module to generate the bitcode section and the object code,
+   which happen to be identical to those used outside of FatLTO. While that
+   resulted in  compiled artifacts that were identical to those produced by the
+   default and (Thin)LTO pipelines, module cloning led to some cases of
+   miscompilation, and we have moved away from trying to keep bitcode
+   generation and optimization completely disjoint.
+
+   Bit-for-bit compatibility is not (and never was) a guarantee, and we reserve
+   the right to change this at any time. Explicitly, users should not rely on
+   the produced bitcode or object code to match their non-LTO counterparts
+   precisely. They will exhibit similar performance characteristics, but may
+   not be bit-for-bit the same.
 
 Internally, the ``.llvm.lto`` section is created by running the
-``EmbedBitcodePass`` at the start of the ``PerModuleDefaultPipeline``. This
-pass is responsible for cloning and optimizing the module with the appropriate
-LTO pipeline and emitting the ``.llvm.lto`` section. Afterwards, the
-``PerModuleDefaultPipeline`` runs normally and the compiler can emit the fat
-object file.
+``EmbedBitcodePass`` after the ``ThinLTOPreLinkDefaultPipeline``. This pass is
+responsible for emitting the ``.llvm.lto`` section. Afterwards, the
+``ThinLTODefaultPipeline`` runs and the compiler can emit the fat object file.
 
 Limitations
 ===========

llvm/include/llvm/Passes/PassBuilder.h

@@ -246,8 +246,7 @@ class PassBuilder {
   /// separately to avoid any inconsistencies with an ad-hoc pipeline that tries
   /// to approximate the PerModuleDefaultPipeline from the pre-link LTO
   /// pipelines.
-  ModulePassManager buildFatLTODefaultPipeline(OptimizationLevel Level,
-                                               bool ThinLTO, bool EmitSummary);
+  ModulePassManager buildFatLTODefaultPipeline(OptimizationLevel Level);
 
   /// Build a pre-link, ThinLTO-targeting default optimization pipeline to
   /// a pass manager.

llvm/include/llvm/Transforms/IPO/EmbedBitcodePass.h

@@ -25,28 +25,13 @@ class Module;
 class ModulePass;
 class Pass;
 
-struct EmbedBitcodeOptions {
-  EmbedBitcodeOptions() : EmbedBitcodeOptions(false, false) {}
-  EmbedBitcodeOptions(bool IsThinLTO, bool EmitLTOSummary)
-      : IsThinLTO(IsThinLTO), EmitLTOSummary(EmitLTOSummary) {}
-  bool IsThinLTO;
-  bool EmitLTOSummary;
-};
-
 /// Pass embeds a copy of the module optimized with the provided pass pipeline
 /// into a global variable.
 class EmbedBitcodePass : public PassInfoMixin<EmbedBitcodePass> {
-  bool IsThinLTO;
-  bool EmitLTOSummary;
   ModulePassManager MPM;
 
 public:
-  EmbedBitcodePass(EmbedBitcodeOptions Opts)
-      : EmbedBitcodePass(Opts.IsThinLTO, Opts.EmitLTOSummary,
-                         ModulePassManager()) {}
-  EmbedBitcodePass(bool IsThinLTO, bool EmitLTOSummary, ModulePassManager &&MPM)
-      : IsThinLTO(IsThinLTO), EmitLTOSummary(EmitLTOSummary),
-        MPM(std::move(MPM)) {}
+  EmbedBitcodePass() {}
 
   PreservedAnalyses run(Module &M, ModuleAnalysisManager &);
 

llvm/lib/Passes/PassBuilder.cpp

@@ -759,26 +759,6 @@ Expected<HWAddressSanitizerOptions> parseHWASanPassOptions(StringRef Params) {
   return Result;
 }
 
-Expected<EmbedBitcodeOptions> parseEmbedBitcodePassOptions(StringRef Params) {
-  EmbedBitcodeOptions Result;
-  while (!Params.empty()) {
-    StringRef ParamName;
-    std::tie(ParamName, Params) = Params.split(';');
-
-    if (ParamName == "thinlto") {
-      Result.IsThinLTO = true;
-    } else if (ParamName == "emit-summary") {
-      Result.EmitLTOSummary = true;
-    } else {
-      return make_error<StringError>(
-          formatv("invalid EmbedBitcode pass parameter '{0}' ", ParamName)
-              .str(),
-          inconvertibleErrorCode());
-    }
-  }
-  return Result;
-}
-
 Expected<MemorySanitizerOptions> parseMSanPassOptions(StringRef Params) {
   MemorySanitizerOptions Result;
   while (!Params.empty()) {

llvm/lib/Passes/PassBuilderPipelines.cpp

@@ -1530,14 +1530,22 @@ PassBuilder::buildPerModuleDefaultPipeline(OptimizationLevel Level,
 }
 
 ModulePassManager
-PassBuilder::buildFatLTODefaultPipeline(OptimizationLevel Level, bool ThinLTO,
-                                        bool EmitSummary) {
+PassBuilder::buildFatLTODefaultPipeline(OptimizationLevel Level) {
   ModulePassManager MPM;
-  MPM.addPass(EmbedBitcodePass(ThinLTO, EmitSummary,
-                               ThinLTO
-                                   ? buildThinLTOPreLinkDefaultPipeline(Level)
-                                   : buildLTOPreLinkDefaultPipeline(Level)));
-  MPM.addPass(buildPerModuleDefaultPipeline(Level));
+  // FatLTO always uses UnifiedLTO, so use the ThinLTOPreLink pipeline
+  MPM.addPass(buildThinLTOPreLinkDefaultPipeline(Level));
+  MPM.addPass(EmbedBitcodePass());
+
+  // Use the ThinLTO post-link pipeline with sample profiling, other
+  if (PGOOpt && PGOOpt->Action == PGOOptions::SampleUse)
+    MPM.addPass(buildThinLTODefaultPipeline(Level, /*ImportSummary=*/nullptr));
+  else {
+    // otherwise, just use module optimization
+    MPM.addPass(
+        buildModuleOptimizationPipeline(Level, ThinOrFullLTOPhase::None));
+    // Emit annotation remarks.
+    addAnnotationRemarksPass(MPM);
+  }
   return MPM;
 }
 

llvm/lib/Passes/PassRegistry.def

@@ -58,6 +58,7 @@ MODULE_PASS("dfsan", DataFlowSanitizerPass())
 MODULE_PASS("dot-callgraph", CallGraphDOTPrinterPass())
 MODULE_PASS("dxil-upgrade", DXILUpgradePass())
 MODULE_PASS("elim-avail-extern", EliminateAvailableExternallyPass())
+MODULE_PASS("embed-bitcode", EmbedBitcodePass())
 MODULE_PASS("extract-blocks", BlockExtractorPass({}, false))
 MODULE_PASS("forceattrs", ForceFunctionAttrsPass())
 MODULE_PASS("function-import", FunctionImportPass())
@@ -145,10 +146,6 @@ MODULE_PASS_WITH_PARAMS(
     "asan", "AddressSanitizerPass",
     [](AddressSanitizerOptions Opts) { return AddressSanitizerPass(Opts); },
     parseASanPassOptions, "kernel")
-MODULE_PASS_WITH_PARAMS(
-    "embed-bitcode", "EmbedBitcodePass",
-    [](EmbedBitcodeOptions Opts) { return EmbedBitcodePass(Opts); },
-    parseEmbedBitcodePassOptions, "thinlto;emit-summary")
 MODULE_PASS_WITH_PARAMS(
     "globaldce", "GlobalDCEPass",
     [](bool InLTOPostLink) { return GlobalDCEPass(InLTOPostLink); },

llvm/lib/Transforms/IPO/EmbedBitcodePass.cpp

@@ -7,19 +7,15 @@
 //===----------------------------------------------------------------------===//
 
 #include "llvm/Transforms/IPO/EmbedBitcodePass.h"
-#include "llvm/Bitcode/BitcodeWriter.h"
-#include "llvm/Bitcode/BitcodeWriterPass.h"
 #include "llvm/IR/PassManager.h"
 #include "llvm/Pass.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/MemoryBufferRef.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/TargetParser/Triple.h"
 #include "llvm/Transforms/IPO/ThinLTOBitcodeWriter.h"
-#include "llvm/Transforms/Utils/Cloning.h"
 #include "llvm/Transforms/Utils/ModuleUtils.h"
 
-#include <memory>
 #include <string>
 
 using namespace llvm;
@@ -34,19 +30,9 @@ PreservedAnalyses EmbedBitcodePass::run(Module &M, ModuleAnalysisManager &AM) {
     report_fatal_error(
         "EmbedBitcode pass currently only supports ELF object format",
         /*gen_crash_diag=*/false);
-
-  std::unique_ptr<Module> NewModule = CloneModule(M);
-  MPM.run(*NewModule, AM);
-
   std::string Data;
   raw_string_ostream OS(Data);
-  if (IsThinLTO)
-    ThinLTOBitcodeWriterPass(OS, /*ThinLinkOS=*/nullptr).run(*NewModule, AM);
-  else
-    BitcodeWriterPass(OS, /*ShouldPreserveUseListOrder=*/false, EmitLTOSummary)
-        .run(*NewModule, AM);
-
+  ThinLTOBitcodeWriterPass(OS, /*ThinLinkOS=*/nullptr).run(M, AM);
   embedBufferInModule(M, MemoryBufferRef(Data, "ModuleData"), ".llvm.lto");
-
   return PreservedAnalyses::all();
 }

llvm/test/CodeGen/X86/fat-lto-section.ll

@@ -1,5 +1,5 @@
 ;; Ensure that the .llvm.lto section has SHT_EXCLUDE set.
-; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode<thinlto;emit-summary>" -S \
+; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode" -S \
 ; RUN:   | llc --mtriple x86_64-unknown-linux-gnu -filetype=obj \
 ; RUN:   | llvm-readelf - --sections \
 ; RUN:   | FileCheck %s --check-prefix=EXCLUDE

llvm/test/Transforms/EmbedBitcode/embed.ll

@@ -1,7 +1,4 @@
 ; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode" -S | FileCheck %s
-; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode<thinlto>" -S | FileCheck %s
-; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode<emit-summary>" -S | FileCheck %s
-; RUN: opt --mtriple x86_64-unknown-linux-gnu < %s -passes="embed-bitcode<thinlto;emit-summary>" -S | FileCheck %s
 
 @a = global i32 1