[llvm] [NFC][Utils] Extract CloneFunctionBodyInto from CloneFunctionInto (PR #118624)

Artem Pianykh via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 9 08:57:34 PST 2024


https://github.com/artempyanykh updated https://github.com/llvm/llvm-project/pull/118624

>From b2ef3f3d57a5e12e266f897c172966dff6fdbb4c Mon Sep 17 00:00:00 2001
From: Artem Pianykh <arr at fb.com>
Date: Thu, 12 Sep 2024 15:23:43 -0700
Subject: [PATCH 1/3] [NFC][Utils] Extract BuildDebugInfoMDMap from
 CloneFunctionInto

Summary:
Extract the logic to build up a metadata map to use in metadata cloning
into a separate function.

Test Plan:
ninja check-llvm-unit check-llvm

stack-info: PR: https://github.com/llvm/llvm-project/pull/118622, branch: users/artempyanykh/fast-coro-upstream/3
---
 llvm/include/llvm/Transforms/Utils/Cloning.h |  8 ++
 llvm/lib/Transforms/Utils/CloneFunction.cpp  | 89 +++++++++++---------
 2 files changed, 58 insertions(+), 39 deletions(-)

diff --git a/llvm/include/llvm/Transforms/Utils/Cloning.h b/llvm/include/llvm/Transforms/Utils/Cloning.h
index 3c8f2cbfaa9b81..7858c9d9def0da 100644
--- a/llvm/include/llvm/Transforms/Utils/Cloning.h
+++ b/llvm/include/llvm/Transforms/Utils/Cloning.h
@@ -220,6 +220,14 @@ DISubprogram *CollectDebugInfoForCloning(const Function &F,
                                          CloneFunctionChangeType Changes,
                                          DebugInfoFinder &DIFinder);
 
+/// Build a map of debug info to use during Metadata cloning.
+/// Returns true if cloning would need module level changes and false if there
+/// would only be local changes.
+bool BuildDebugInfoMDMap(DenseMap<const Metadata *, TrackingMDRef> &MD,
+                         CloneFunctionChangeType Changes,
+                         DebugInfoFinder &DIFinder,
+                         DISubprogram *SPClonedWithinModule);
+
 /// This class captures the data input to the InlineFunction call, and records
 /// the auxiliary results produced by it.
 class InlineFunctionInfo {
diff --git a/llvm/lib/Transforms/Utils/CloneFunction.cpp b/llvm/lib/Transforms/Utils/CloneFunction.cpp
index d038117090e4cc..6dc5f601b7fcaa 100644
--- a/llvm/lib/Transforms/Utils/CloneFunction.cpp
+++ b/llvm/lib/Transforms/Utils/CloneFunction.cpp
@@ -152,6 +152,54 @@ DISubprogram *llvm::CollectDebugInfoForCloning(const Function &F,
   return SPClonedWithinModule;
 }
 
+bool llvm::BuildDebugInfoMDMap(DenseMap<const Metadata *, TrackingMDRef> &MD,
+                               CloneFunctionChangeType Changes,
+                               DebugInfoFinder &DIFinder,
+                               DISubprogram *SPClonedWithinModule) {
+  bool ModuleLevelChanges = Changes > CloneFunctionChangeType::LocalChangesOnly;
+  if (Changes < CloneFunctionChangeType::DifferentModule &&
+      DIFinder.subprogram_count() > 0) {
+    // Turn on module-level changes, since we need to clone (some of) the
+    // debug info metadata.
+    //
+    // FIXME: Metadata effectively owned by a function should be made
+    // local, and only that local metadata should be cloned.
+    ModuleLevelChanges = true;
+
+    auto mapToSelfIfNew = [&MD](MDNode *N) {
+      // Avoid clobbering an existing mapping.
+      (void)MD.try_emplace(N, N);
+    };
+
+    // Avoid cloning types, compile units, and (other) subprograms.
+    SmallPtrSet<const DISubprogram *, 16> MappedToSelfSPs;
+    for (DISubprogram *ISP : DIFinder.subprograms()) {
+      if (ISP != SPClonedWithinModule) {
+        mapToSelfIfNew(ISP);
+        MappedToSelfSPs.insert(ISP);
+      }
+    }
+
+    // If a subprogram isn't going to be cloned skip its lexical blocks as well.
+    for (DIScope *S : DIFinder.scopes()) {
+      auto *LScope = dyn_cast<DILocalScope>(S);
+      if (LScope && MappedToSelfSPs.count(LScope->getSubprogram()))
+        mapToSelfIfNew(S);
+    }
+
+    for (DICompileUnit *CU : DIFinder.compile_units())
+      mapToSelfIfNew(CU);
+
+    for (DIType *Type : DIFinder.types())
+      mapToSelfIfNew(Type);
+  } else {
+    assert(!SPClonedWithinModule &&
+           "Subprogram should be in DIFinder->subprogram_count()...");
+  }
+
+  return ModuleLevelChanges;
+}
+
 // Clone OldFunc into NewFunc, transforming the old arguments into references to
 // VMap values.
 void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
@@ -210,45 +258,8 @@ void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
   DISubprogram *SPClonedWithinModule =
       CollectDebugInfoForCloning(*OldFunc, Changes, DIFinder);
 
-  if (Changes < CloneFunctionChangeType::DifferentModule &&
-      DIFinder.subprogram_count() > 0) {
-    // Turn on module-level changes, since we need to clone (some of) the
-    // debug info metadata.
-    //
-    // FIXME: Metadata effectively owned by a function should be made
-    // local, and only that local metadata should be cloned.
-    ModuleLevelChanges = true;
-
-    auto mapToSelfIfNew = [&VMap](MDNode *N) {
-      // Avoid clobbering an existing mapping.
-      (void)VMap.MD().try_emplace(N, N);
-    };
-
-    // Avoid cloning types, compile units, and (other) subprograms.
-    SmallPtrSet<const DISubprogram *, 16> MappedToSelfSPs;
-    for (DISubprogram *ISP : DIFinder.subprograms()) {
-      if (ISP != SPClonedWithinModule) {
-        mapToSelfIfNew(ISP);
-        MappedToSelfSPs.insert(ISP);
-      }
-    }
-
-    // If a subprogram isn't going to be cloned skip its lexical blocks as well.
-    for (DIScope *S : DIFinder.scopes()) {
-      auto *LScope = dyn_cast<DILocalScope>(S);
-      if (LScope && MappedToSelfSPs.count(LScope->getSubprogram()))
-        mapToSelfIfNew(S);
-    }
-
-    for (DICompileUnit *CU : DIFinder.compile_units())
-      mapToSelfIfNew(CU);
-
-    for (DIType *Type : DIFinder.types())
-      mapToSelfIfNew(Type);
-  } else {
-    assert(!SPClonedWithinModule &&
-           "Subprogram should be in DIFinder->subprogram_count()...");
-  }
+  ModuleLevelChanges =
+      BuildDebugInfoMDMap(VMap.MD(), Changes, DIFinder, SPClonedWithinModule);
 
   const auto RemapFlag = ModuleLevelChanges ? RF_None : RF_NoModuleLevelChanges;
   // Duplicate the metadata that is attached to the cloned function.

>From 9dcfc56cf669e9b8b571469ba5ba850b7565faa7 Mon Sep 17 00:00:00 2001
From: Artem Pianykh <arr at fb.com>
Date: Thu, 12 Sep 2024 15:35:38 -0700
Subject: [PATCH 2/3] [NFC][Utils] Extract CloneFunctionMetadataInto from
 CloneFunctionInto

Summary:
The new API expects the caller to populate the VMap. We need it this way
for a subsequent change around coroutine cloning.

Test Plan:
ninja check-llvm-unit check-llvm

stack-info: PR: https://github.com/llvm/llvm-project/pull/118623, branch: users/artempyanykh/fast-coro-upstream/4
---
 llvm/include/llvm/Transforms/Utils/Cloning.h | 12 +++++++++
 llvm/lib/Transforms/Utils/CloneFunction.cpp  | 28 +++++++++++++-------
 2 files changed, 31 insertions(+), 9 deletions(-)

diff --git a/llvm/include/llvm/Transforms/Utils/Cloning.h b/llvm/include/llvm/Transforms/Utils/Cloning.h
index 7858c9d9def0da..9a574fc4e4c08e 100644
--- a/llvm/include/llvm/Transforms/Utils/Cloning.h
+++ b/llvm/include/llvm/Transforms/Utils/Cloning.h
@@ -182,6 +182,18 @@ void CloneFunctionAttributesInto(Function *NewFunc, const Function *OldFunc,
                                  ValueMapTypeRemapper *TypeMapper = nullptr,
                                  ValueMaterializer *Materializer = nullptr);
 
+/// Clone OldFunc's metadata into NewFunc.
+///
+/// The caller is expected to populate \p VMap beforehand and set an appropriate
+/// \p RemapFlag.
+///
+/// NOTE: This function doesn't clone !llvm.dbg.cu when cloning into a different
+/// module. Use CloneFunctionInto for that behavior.
+void CloneFunctionMetadataInto(Function *NewFunc, const Function *OldFunc,
+                               ValueToValueMapTy &VMap, RemapFlags RemapFlag,
+                               ValueMapTypeRemapper *TypeMapper = nullptr,
+                               ValueMaterializer *Materializer = nullptr);
+
 void CloneAndPruneIntoFromInst(Function *NewFunc, const Function *OldFunc,
                                const Instruction *StartingInst,
                                ValueToValueMapTy &VMap, bool ModuleLevelChanges,
diff --git a/llvm/lib/Transforms/Utils/CloneFunction.cpp b/llvm/lib/Transforms/Utils/CloneFunction.cpp
index 6dc5f601b7fcaa..c967e78123af1f 100644
--- a/llvm/lib/Transforms/Utils/CloneFunction.cpp
+++ b/llvm/lib/Transforms/Utils/CloneFunction.cpp
@@ -200,6 +200,22 @@ bool llvm::BuildDebugInfoMDMap(DenseMap<const Metadata *, TrackingMDRef> &MD,
   return ModuleLevelChanges;
 }
 
+void llvm::CloneFunctionMetadataInto(Function *NewFunc, const Function *OldFunc,
+                                     ValueToValueMapTy &VMap,
+                                     RemapFlags RemapFlag,
+                                     ValueMapTypeRemapper *TypeMapper,
+                                     ValueMaterializer *Materializer) {
+  // Duplicate the metadata that is attached to the cloned function.
+  // Subprograms/CUs/types that were already mapped to themselves won't be
+  // duplicated.
+  SmallVector<std::pair<unsigned, MDNode *>, 1> MDs;
+  OldFunc->getAllMetadata(MDs);
+  for (auto MD : MDs) {
+    NewFunc->addMetadata(MD.first, *MapMetadata(MD.second, VMap, RemapFlag,
+                                                TypeMapper, Materializer));
+  }
+}
+
 // Clone OldFunc into NewFunc, transforming the old arguments into references to
 // VMap values.
 void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
@@ -262,15 +278,9 @@ void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
       BuildDebugInfoMDMap(VMap.MD(), Changes, DIFinder, SPClonedWithinModule);
 
   const auto RemapFlag = ModuleLevelChanges ? RF_None : RF_NoModuleLevelChanges;
-  // Duplicate the metadata that is attached to the cloned function.
-  // Subprograms/CUs/types that were already mapped to themselves won't be
-  // duplicated.
-  SmallVector<std::pair<unsigned, MDNode *>, 1> MDs;
-  OldFunc->getAllMetadata(MDs);
-  for (auto MD : MDs) {
-    NewFunc->addMetadata(MD.first, *MapMetadata(MD.second, VMap, RemapFlag,
-                                                TypeMapper, Materializer));
-  }
+
+  CloneFunctionMetadataInto(NewFunc, OldFunc, VMap, RemapFlag, TypeMapper,
+                            Materializer);
 
   // Loop over all of the basic blocks in the function, cloning them as
   // appropriate.  Note that we save BE this way in order to handle cloning of

>From 6845960eea8fe58cc5f6e05611ef2920c49a34b9 Mon Sep 17 00:00:00 2001
From: Artem Pianykh <arr at fb.com>
Date: Thu, 12 Sep 2024 15:50:25 -0700
Subject: [PATCH 3/3] [NFC][Utils] Extract CloneFunctionBodyInto from
 CloneFunctionInto

Summary:
This and previously extracted `CloneFunction*Into` functions will be used in later diffs.

Test Plan:
ninja check-llvm-unit check-llvm

stack-info: PR: https://github.com/llvm/llvm-project/pull/118624, branch: users/artempyanykh/fast-coro-upstream/5
---
 llvm/include/llvm/Transforms/Utils/Cloning.h | 34 ++++---
 llvm/lib/Transforms/Utils/CloneFunction.cpp  | 96 +++++++++++---------
 2 files changed, 76 insertions(+), 54 deletions(-)

diff --git a/llvm/include/llvm/Transforms/Utils/Cloning.h b/llvm/include/llvm/Transforms/Utils/Cloning.h
index 9a574fc4e4c08e..50518c746d11ca 100644
--- a/llvm/include/llvm/Transforms/Utils/Cloning.h
+++ b/llvm/include/llvm/Transforms/Utils/Cloning.h
@@ -194,6 +194,15 @@ void CloneFunctionMetadataInto(Function *NewFunc, const Function *OldFunc,
                                ValueMapTypeRemapper *TypeMapper = nullptr,
                                ValueMaterializer *Materializer = nullptr);
 
+/// Clone OldFunc's body into NewFunc.
+void CloneFunctionBodyInto(Function *NewFunc, const Function *OldFunc,
+                           ValueToValueMapTy &VMap, RemapFlags RemapFlag,
+                           SmallVectorImpl<ReturnInst *> &Returns,
+                           const char *NameSuffix = "",
+                           ClonedCodeInfo *CodeInfo = nullptr,
+                           ValueMapTypeRemapper *TypeMapper = nullptr,
+                           ValueMaterializer *Materializer = nullptr);
+
 void CloneAndPruneIntoFromInst(Function *NewFunc, const Function *OldFunc,
                                const Instruction *StartingInst,
                                ValueToValueMapTy &VMap, bool ModuleLevelChanges,
@@ -214,7 +223,7 @@ void CloneAndPruneIntoFromInst(Function *NewFunc, const Function *OldFunc,
 ///
 void CloneAndPruneFunctionInto(Function *NewFunc, const Function *OldFunc,
                                ValueToValueMapTy &VMap, bool ModuleLevelChanges,
-                               SmallVectorImpl<ReturnInst*> &Returns,
+                               SmallVectorImpl<ReturnInst *> &Returns,
                                const char *NameSuffix = "",
                                ClonedCodeInfo *CodeInfo = nullptr);
 
@@ -361,32 +370,31 @@ void updateProfileCallee(
 /// Find the 'llvm.experimental.noalias.scope.decl' intrinsics in the specified
 /// basic blocks and extract their scope. These are candidates for duplication
 /// when cloning.
-void identifyNoAliasScopesToClone(
-    ArrayRef<BasicBlock *> BBs, SmallVectorImpl<MDNode *> &NoAliasDeclScopes);
+void identifyNoAliasScopesToClone(ArrayRef<BasicBlock *> BBs,
+                                  SmallVectorImpl<MDNode *> &NoAliasDeclScopes);
 
 /// Find the 'llvm.experimental.noalias.scope.decl' intrinsics in the specified
 /// instruction range and extract their scope. These are candidates for
 /// duplication when cloning.
-void identifyNoAliasScopesToClone(
-    BasicBlock::iterator Start, BasicBlock::iterator End,
-    SmallVectorImpl<MDNode *> &NoAliasDeclScopes);
+void identifyNoAliasScopesToClone(BasicBlock::iterator Start,
+                                  BasicBlock::iterator End,
+                                  SmallVectorImpl<MDNode *> &NoAliasDeclScopes);
 
 /// Duplicate the specified list of noalias decl scopes.
 /// The 'Ext' string is added as an extension to the name.
 /// Afterwards, the ClonedScopes contains the mapping of the original scope
 /// MDNode onto the cloned scope.
 /// Be aware that the cloned scopes are still part of the original scope domain.
-void cloneNoAliasScopes(
-    ArrayRef<MDNode *> NoAliasDeclScopes,
-    DenseMap<MDNode *, MDNode *> &ClonedScopes,
-    StringRef Ext, LLVMContext &Context);
+void cloneNoAliasScopes(ArrayRef<MDNode *> NoAliasDeclScopes,
+                        DenseMap<MDNode *, MDNode *> &ClonedScopes,
+                        StringRef Ext, LLVMContext &Context);
 
 /// Adapt the metadata for the specified instruction according to the
 /// provided mapping. This is normally used after cloning an instruction, when
 /// some noalias scopes needed to be cloned.
-void adaptNoAliasScopes(
-    llvm::Instruction *I, const DenseMap<MDNode *, MDNode *> &ClonedScopes,
-    LLVMContext &Context);
+void adaptNoAliasScopes(llvm::Instruction *I,
+                        const DenseMap<MDNode *, MDNode *> &ClonedScopes,
+                        LLVMContext &Context);
 
 /// Clone the specified noalias decl scopes. Then adapt all instructions in the
 /// NewBlocks basicblocks to the cloned versions.
diff --git a/llvm/lib/Transforms/Utils/CloneFunction.cpp b/llvm/lib/Transforms/Utils/CloneFunction.cpp
index c967e78123af1f..cf4b1c7a045e05 100644
--- a/llvm/lib/Transforms/Utils/CloneFunction.cpp
+++ b/llvm/lib/Transforms/Utils/CloneFunction.cpp
@@ -216,6 +216,59 @@ void llvm::CloneFunctionMetadataInto(Function *NewFunc, const Function *OldFunc,
   }
 }
 
+void llvm::CloneFunctionBodyInto(Function *NewFunc, const Function *OldFunc,
+                                 ValueToValueMapTy &VMap, RemapFlags RemapFlag,
+                                 SmallVectorImpl<ReturnInst *> &Returns,
+                                 const char *NameSuffix,
+                                 ClonedCodeInfo *CodeInfo,
+                                 ValueMapTypeRemapper *TypeMapper,
+                                 ValueMaterializer *Materializer) {
+  if (OldFunc->isDeclaration())
+    return;
+
+  // Loop over all of the basic blocks in the function, cloning them as
+  // appropriate.  Note that we save BE this way in order to handle cloning of
+  // recursive functions into themselves.
+  for (const BasicBlock &BB : *OldFunc) {
+
+    // Create a new basic block and copy instructions into it!
+    BasicBlock *CBB = CloneBasicBlock(&BB, VMap, NameSuffix, NewFunc, CodeInfo);
+
+    // Add basic block mapping.
+    VMap[&BB] = CBB;
+
+    // It is only legal to clone a function if a block address within that
+    // function is never referenced outside of the function.  Given that, we
+    // want to map block addresses from the old function to block addresses in
+    // the clone. (This is different from the generic ValueMapper
+    // implementation, which generates an invalid blockaddress when
+    // cloning a function.)
+    if (BB.hasAddressTaken()) {
+      Constant *OldBBAddr = BlockAddress::get(const_cast<Function *>(OldFunc),
+                                              const_cast<BasicBlock *>(&BB));
+      VMap[OldBBAddr] = BlockAddress::get(NewFunc, CBB);
+    }
+
+    // Note return instructions for the caller.
+    if (ReturnInst *RI = dyn_cast<ReturnInst>(CBB->getTerminator()))
+      Returns.push_back(RI);
+  }
+
+  // Loop over all of the instructions in the new function, fixing up operand
+  // references as we go. This uses VMap to do all the hard work.
+  for (Function::iterator
+           BB = cast<BasicBlock>(VMap[&OldFunc->front()])->getIterator(),
+           BE = NewFunc->end();
+       BB != BE; ++BB)
+    // Loop over all instructions, fixing each one as we find it, and any
+    // attached debug-info records.
+    for (Instruction &II : *BB) {
+      RemapInstruction(&II, VMap, RemapFlag, TypeMapper, Materializer);
+      RemapDbgRecordRange(II.getModule(), II.getDbgRecordRange(), VMap,
+                          RemapFlag, TypeMapper, Materializer);
+    }
+}
+
 // Clone OldFunc into NewFunc, transforming the old arguments into references to
 // VMap values.
 void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
@@ -282,47 +335,8 @@ void llvm::CloneFunctionInto(Function *NewFunc, const Function *OldFunc,
   CloneFunctionMetadataInto(NewFunc, OldFunc, VMap, RemapFlag, TypeMapper,
                             Materializer);
 
-  // Loop over all of the basic blocks in the function, cloning them as
-  // appropriate.  Note that we save BE this way in order to handle cloning of
-  // recursive functions into themselves.
-  for (const BasicBlock &BB : *OldFunc) {
-
-    // Create a new basic block and copy instructions into it!
-    BasicBlock *CBB = CloneBasicBlock(&BB, VMap, NameSuffix, NewFunc, CodeInfo);
-
-    // Add basic block mapping.
-    VMap[&BB] = CBB;
-
-    // It is only legal to clone a function if a block address within that
-    // function is never referenced outside of the function.  Given that, we
-    // want to map block addresses from the old function to block addresses in
-    // the clone. (This is different from the generic ValueMapper
-    // implementation, which generates an invalid blockaddress when
-    // cloning a function.)
-    if (BB.hasAddressTaken()) {
-      Constant *OldBBAddr = BlockAddress::get(const_cast<Function *>(OldFunc),
-                                              const_cast<BasicBlock *>(&BB));
-      VMap[OldBBAddr] = BlockAddress::get(NewFunc, CBB);
-    }
-
-    // Note return instructions for the caller.
-    if (ReturnInst *RI = dyn_cast<ReturnInst>(CBB->getTerminator()))
-      Returns.push_back(RI);
-  }
-
-  // Loop over all of the instructions in the new function, fixing up operand
-  // references as we go. This uses VMap to do all the hard work.
-  for (Function::iterator
-           BB = cast<BasicBlock>(VMap[&OldFunc->front()])->getIterator(),
-           BE = NewFunc->end();
-       BB != BE; ++BB)
-    // Loop over all instructions, fixing each one as we find it, and any
-    // attached debug-info records.
-    for (Instruction &II : *BB) {
-      RemapInstruction(&II, VMap, RemapFlag, TypeMapper, Materializer);
-      RemapDbgRecordRange(II.getModule(), II.getDbgRecordRange(), VMap,
-                          RemapFlag, TypeMapper, Materializer);
-    }
+  CloneFunctionBodyInto(NewFunc, OldFunc, VMap, RemapFlag, Returns, NameSuffix,
+                        CodeInfo, TypeMapper, Materializer);
 
   // Only update !llvm.dbg.cu for DifferentModule (not CloneModule). In the
   // same module, the compile unit will already be listed (or not). When



More information about the llvm-commits mailing list