[Support] Change SetVector's default template parameter to SmallVector<*, 0>

Similar to D156016 for MapVector.

This brings back commit fae7b98 with a
fix to llvm/unittests/Support/ThreadPool.cpp's `_WIN32` code path.

Author: MaskRay

clang/lib/CodeGen/CodeGenModule.cpp

@@ -2236,7 +2236,7 @@ static bool requiresMemberFunctionPointerTypeMetadata(CodeGenModule &CGM,
          !isa<CXXDestructorDecl>(MD);
 }
 
-std::vector<const CXXRecordDecl *>
+SmallVector<const CXXRecordDecl *, 0>
 CodeGenModule::getMostBaseClasses(const CXXRecordDecl *RD) {
   llvm::SetVector<const CXXRecordDecl *> MostBases;
 

clang/lib/CodeGen/CodeGenModule.h

@@ -1505,7 +1505,7 @@ class CodeGenModule : public CodeGenTypeCache {
   ///
   /// A most-base class of a class C is defined as a recursive base class of C,
   /// including C itself, that does not have any bases.
-  std::vector<const CXXRecordDecl *>
+  SmallVector<const CXXRecordDecl *, 0>
   getMostBaseClasses(const CXXRecordDecl *RD);
 
   /// Get the declaration of std::terminate for the platform.

llvm/include/llvm/ADT/SetVector.h

@@ -23,10 +23,10 @@
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/DenseSet.h"
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/Compiler.h"
 #include <cassert>
 #include <iterator>
-#include <vector>
 
 namespace llvm {
 
@@ -52,7 +52,7 @@ namespace llvm {
 /// when searching for elements instead of checking Set, due to it being better
 /// for performance. A value of 0 means that this mode of operation is not used,
 /// and is the default value.
-template <typename T, typename Vector = std::vector<T>,
+template <typename T, typename Vector = SmallVector<T, 0>,
           typename Set = DenseSet<T>, unsigned N = 0>
 class SetVector {
   // Much like in SmallPtrSet, this value should not be too high to prevent

llvm/lib/Analysis/ModuleSummaryAnalysis.cpp

@@ -93,7 +93,7 @@ extern cl::opt<bool> ScalePartialSampleProfileWorkingSetSize;
 // instruction in it takes an address of any basic block, because instruction
 // can only take an address of basic block located in the same function.
 static bool findRefEdges(ModuleSummaryIndex &Index, const User *CurUser,
-                         SetVector<ValueInfo> &RefEdges,
+                         SetVector<ValueInfo, std::vector<ValueInfo>> &RefEdges,
                          SmallPtrSet<const User *, 8> &Visited) {
   bool HasBlockAddress = false;
   SmallVector<const User *, 32> Worklist;
@@ -144,9 +144,12 @@ static bool isNonRenamableLocal(const GlobalValue &GV) {
 
 /// Determine whether this call has all constant integer arguments (excluding
 /// "this") and summarize it to VCalls or ConstVCalls as appropriate.
-static void addVCallToSet(DevirtCallSite Call, GlobalValue::GUID Guid,
-                          SetVector<FunctionSummary::VFuncId> &VCalls,
-                          SetVector<FunctionSummary::ConstVCall> &ConstVCalls) {
+static void addVCallToSet(
+    DevirtCallSite Call, GlobalValue::GUID Guid,
+    SetVector<FunctionSummary::VFuncId, std::vector<FunctionSummary::VFuncId>>
+        &VCalls,
+    SetVector<FunctionSummary::ConstVCall,
+              std::vector<FunctionSummary::ConstVCall>> &ConstVCalls) {
   std::vector<uint64_t> Args;
   // Start from the second argument to skip the "this" pointer.
   for (auto &Arg : drop_begin(Call.CB.args())) {
@@ -163,11 +166,18 @@ static void addVCallToSet(DevirtCallSite Call, GlobalValue::GUID Guid,
 /// If this intrinsic call requires that we add information to the function
 /// summary, do so via the non-constant reference arguments.
 static void addIntrinsicToSummary(
-    const CallInst *CI, SetVector<GlobalValue::GUID> &TypeTests,
-    SetVector<FunctionSummary::VFuncId> &TypeTestAssumeVCalls,
-    SetVector<FunctionSummary::VFuncId> &TypeCheckedLoadVCalls,
-    SetVector<FunctionSummary::ConstVCall> &TypeTestAssumeConstVCalls,
-    SetVector<FunctionSummary::ConstVCall> &TypeCheckedLoadConstVCalls,
+    const CallInst *CI,
+    SetVector<GlobalValue::GUID, std::vector<GlobalValue::GUID>> &TypeTests,
+    SetVector<FunctionSummary::VFuncId, std::vector<FunctionSummary::VFuncId>>
+        &TypeTestAssumeVCalls,
+    SetVector<FunctionSummary::VFuncId, std::vector<FunctionSummary::VFuncId>>
+        &TypeCheckedLoadVCalls,
+    SetVector<FunctionSummary::ConstVCall,
+              std::vector<FunctionSummary::ConstVCall>>
+        &TypeTestAssumeConstVCalls,
+    SetVector<FunctionSummary::ConstVCall,
+              std::vector<FunctionSummary::ConstVCall>>
+        &TypeCheckedLoadConstVCalls,
     DominatorTree &DT) {
   switch (CI->getCalledFunction()->getIntrinsicID()) {
   case Intrinsic::type_test:
@@ -269,12 +279,14 @@ static void computeFunctionSummary(
   MapVector<ValueInfo, CalleeInfo, DenseMap<ValueInfo, unsigned>,
             std::vector<std::pair<ValueInfo, CalleeInfo>>>
       CallGraphEdges;
-  SetVector<ValueInfo> RefEdges, LoadRefEdges, StoreRefEdges;
-  SetVector<GlobalValue::GUID> TypeTests;
-  SetVector<FunctionSummary::VFuncId> TypeTestAssumeVCalls,
-      TypeCheckedLoadVCalls;
-  SetVector<FunctionSummary::ConstVCall> TypeTestAssumeConstVCalls,
-      TypeCheckedLoadConstVCalls;
+  SetVector<ValueInfo, std::vector<ValueInfo>> RefEdges, LoadRefEdges,
+      StoreRefEdges;
+  SetVector<GlobalValue::GUID, std::vector<GlobalValue::GUID>> TypeTests;
+  SetVector<FunctionSummary::VFuncId, std::vector<FunctionSummary::VFuncId>>
+      TypeTestAssumeVCalls, TypeCheckedLoadVCalls;
+  SetVector<FunctionSummary::ConstVCall,
+            std::vector<FunctionSummary::ConstVCall>>
+      TypeTestAssumeConstVCalls, TypeCheckedLoadConstVCalls;
   ICallPromotionAnalysis ICallAnalysis;
   SmallPtrSet<const User *, 8> Visited;
 
@@ -505,7 +517,7 @@ static void computeFunctionSummary(
   std::vector<ValueInfo> Refs;
   if (IsThinLTO) {
     auto AddRefEdges = [&](const std::vector<const Instruction *> &Instrs,
-                           SetVector<ValueInfo> &Edges,
+                           SetVector<ValueInfo, std::vector<ValueInfo>> &Edges,
                            SmallPtrSet<const User *, 8> &Cache) {
       for (const auto *I : Instrs) {
         Cache.erase(I);
@@ -710,7 +722,7 @@ static void computeVariableSummary(ModuleSummaryIndex &Index,
                                    DenseSet<GlobalValue::GUID> &CantBePromoted,
                                    const Module &M,
                                    SmallVectorImpl<MDNode *> &Types) {
-  SetVector<ValueInfo> RefEdges;
+  SetVector<ValueInfo, std::vector<ValueInfo>> RefEdges;
   SmallPtrSet<const User *, 8> Visited;
   bool HasBlockAddress = findRefEdges(Index, &V, RefEdges, Visited);
   bool NonRenamableLocal = isNonRenamableLocal(V);

llvm/tools/llvm-reduce/deltas/ReduceOperandsSkip.cpp

@@ -155,7 +155,7 @@ opportunities(Function &F,
 
       // After all candidates have been added, it doesn't need to be a set
       // anymore.
-      std::vector<Value *> Candidates = ReferencedVals.takeVector();
+      auto Candidates = ReferencedVals.takeVector();
 
       // Remove ineligible candidates.
       llvm::erase_if(Candidates, [&, OpVal](Value *V) {

llvm/unittests/Support/ThreadPool.cpp

@@ -99,7 +99,7 @@ class ThreadPoolTest : public testing::Test {
 
   void SetUp() override { CurrentPhase = 0; }
 
-  std::vector<llvm::BitVector> RunOnAllSockets(ThreadPoolStrategy S);
+  SmallVector<llvm::BitVector, 0> RunOnAllSockets(ThreadPoolStrategy S);
 
   std::condition_variable CurrentPhaseCondition;
   std::mutex CurrentPhaseMutex;
@@ -346,7 +346,7 @@ TEST_F(ThreadPoolTest, RecursiveWaitDeadlock) {
 // isn't implemented for Unix (need AffinityMask in Support/Unix/Program.inc).
 #ifdef _WIN32
 
-std::vector<llvm::BitVector>
+SmallVector<llvm::BitVector, 0>
 ThreadPoolTest::RunOnAllSockets(ThreadPoolStrategy S) {
   llvm::SetVector<llvm::BitVector> ThreadsUsed;
   std::mutex Lock;
@@ -387,7 +387,7 @@ TEST_F(ThreadPoolTest, AllThreads_UseAllRessources) {
   // therefore this test should not run.
   if (llvm::RunningWindows11OrGreater())
     GTEST_SKIP();
-  std::vector<llvm::BitVector> ThreadsUsed = RunOnAllSockets({});
+  auto ThreadsUsed = RunOnAllSockets({});
   ASSERT_EQ(llvm::get_cpus(), ThreadsUsed.size());
 }
 
@@ -398,8 +398,7 @@ TEST_F(ThreadPoolTest, AllThreads_OneThreadPerCore) {
   // therefore this test should not run.
   if (llvm::RunningWindows11OrGreater())
     GTEST_SKIP();
-  std::vector<llvm::BitVector> ThreadsUsed =
-      RunOnAllSockets(llvm::heavyweight_hardware_concurrency());
+  auto ThreadsUsed = RunOnAllSockets(llvm::heavyweight_hardware_concurrency());
   ASSERT_EQ(llvm::get_cpus(), ThreadsUsed.size());
 }
 
@@ -422,7 +421,7 @@ TEST_F(ThreadPoolTest, AffinityMask) {
 
   using namespace llvm::sys;
   if (getenv("LLVM_THREADPOOL_AFFINITYMASK")) {
-    std::vector<llvm::BitVector> ThreadsUsed = RunOnAllSockets({});
+    auto ThreadsUsed = RunOnAllSockets({});
     // Ensure the threads only ran on CPUs 0-3.
     // NOTE: Don't use ASSERT* here because this runs in a subprocess,
     // and will show up as un-executed in the parent.

mlir/include/mlir/Support/LLVM.h

@@ -122,7 +122,7 @@ template <typename KeyT, typename ValueT,
 using DenseMap = llvm::DenseMap<KeyT, ValueT, KeyInfoT, BucketT>;
 template <typename ValueT, typename ValueInfoT = DenseMapInfo<ValueT>>
 using DenseSet = llvm::DenseSet<ValueT, ValueInfoT>;
-template <typename T, typename Vector = std::vector<T>,
+template <typename T, typename Vector = llvm::SmallVector<T, 0>,
           typename Set = DenseSet<T>, unsigned N = 0>
 using SetVector = llvm::SetVector<T, Vector, Set, N>;
 template <typename AllocatorTy = llvm::MallocAllocator>

mlir/lib/Analysis/SliceAnalysis.cpp

@@ -62,7 +62,7 @@ void mlir::getForwardSlice(Operation *op, SetVector<Operation *> *forwardSlice,
   // Reverse to get back the actual topological order.
   // std::reverse does not work out of the box on SetVector and I want an
   // in-place swap based thing (the real std::reverse, not the LLVM adapter).
-  std::vector<Operation *> v(forwardSlice->takeVector());
+  SmallVector<Operation *, 0> v(forwardSlice->takeVector());
   forwardSlice->insert(v.rbegin(), v.rend());
 }
 
@@ -74,7 +74,7 @@ void mlir::getForwardSlice(Value root, SetVector<Operation *> *forwardSlice,
   // Reverse to get back the actual topological order.
   // std::reverse does not work out of the box on SetVector and I want an
   // in-place swap based thing (the real std::reverse, not the LLVM adapter).
-  std::vector<Operation *> v(forwardSlice->takeVector());
+  SmallVector<Operation *, 0> v(forwardSlice->takeVector());
   forwardSlice->insert(v.rbegin(), v.rend());
 }