[llvm][mlir][OMPIRBuilder] Translate omp.single's copyprivate (#80488)

Use the new copyprivate list from omp.single to emit calls to
__kmpc_copyprivate, during the creation of the single operation
in OMPIRBuilder.

This is patch 4 of 4, to add support for COPYPRIVATE in Flang.
Original PR: #73128

flang/test/Integration/OpenMP/copyprivate.f90

@@ -0,0 +1,97 @@
+!===----------------------------------------------------------------------===!
+! This directory can be used to add Integration tests involving multiple
+! stages of the compiler (for eg. from Fortran to LLVM IR). It should not
+! contain executable tests. We should only add tests here sparingly and only
+! if there is no other way to test. Repeat this message in each test that is
+! added to this directory and sub-directories.
+!===----------------------------------------------------------------------===!
+
+!RUN: %flang_fc1 -emit-llvm -fopenmp %s -o - | FileCheck %s
+
+!CHECK-DAG: define void @_copy_box_Uxi32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_10xi32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_i64(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_box_Uxi64(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_f32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_2x3xf32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_z32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_10xz32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_l32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_5xl32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_c8x8(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_10xc8x8(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_c16x5(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_rec__QFtest_typesTdt(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_box_heap_Uxi32(ptr %{{.*}}, ptr %{{.*}})
+!CHECK-DAG: define void @_copy_box_ptr_Uxc8x9(ptr %{{.*}}, ptr %{{.*}})
+
+!CHECK-LABEL: define void @_copy_i32(
+!CHECK-SAME:                         ptr %[[DST:.*]], ptr %[[SRC:.*]]) {
+!CHECK-NEXT:    %[[SRC_VAL:.*]] = load i32, ptr %[[SRC]]
+!CHECK-NEXT:    store i32 %[[SRC_VAL]], ptr %[[DST]]
+!CHECK-NEXT:    ret void
+!CHECK-NEXT:  }
+
+!CHECK-LABEL: define internal void @test_scalar_..omp_par({{.*}})
+!CHECK:         %[[I:.*]] = alloca i32, i64 1
+!CHECK:         %[[J:.*]] = alloca i32, i64 1
+!CHECK:         %[[DID_IT:.*]] = alloca i32
+!CHECK:         store i32 0, ptr %[[DID_IT]]
+!CHECK:         %[[THREAD_NUM1:.*]] = call i32 @__kmpc_global_thread_num(ptr @[[LOC:.*]])
+!CHECK:         %[[RET:.*]] = call i32 @__kmpc_single({{.*}})
+!CHECK:         %[[NOT_ZERO:.*]] = icmp ne i32 %[[RET]], 0
+!CHECK:         br i1 %[[NOT_ZERO]], label %[[OMP_REGION_BODY:.*]], label %[[OMP_REGION_END:.*]]
+
+!CHECK:       [[OMP_REGION_END]]:
+!CHECK:         %[[THREAD_NUM2:.*]] = call i32 @__kmpc_global_thread_num(ptr @[[LOC:.*]])
+!CHECK:         %[[DID_IT_VAL:.*]] = load i32, ptr %[[DID_IT]]
+!CHECK:         call void @__kmpc_copyprivate(ptr @[[LOC]], i32 %[[THREAD_NUM2]], i64 0, ptr %[[I]], ptr @_copy_i32, i32 %[[DID_IT_VAL]])
+!CHECK:         %[[THREAD_NUM3:.*]] = call i32 @__kmpc_global_thread_num(ptr @[[LOC]])
+!CHECK:         %[[DID_IT_VAL2:.*]] = load i32, ptr %[[DID_IT]]
+!CHECK:         call void @__kmpc_copyprivate(ptr @[[LOC]], i32 %[[THREAD_NUM3]], i64 0, ptr %[[J]], ptr @_copy_i32, i32 %[[DID_IT_VAL2]])
+
+!CHECK:       [[OMP_REGION_BODY]]:
+!CHECK:         br label %[[OMP_SINGLE_REGION:.*]]
+!CHECK:       [[OMP_SINGLE_REGION]]:
+!CHECK:         store i32 11, ptr %[[I]]
+!CHECK:         store i32 22, ptr %[[J]]
+!CHECK:         br label %[[OMP_REGION_CONT3:.*]]
+!CHECK:       [[OMP_REGION_CONT3:.*]]:
+!CHECK:         store i32 1, ptr %[[DID_IT]]
+!CHECK:         call void @__kmpc_end_single(ptr @[[LOC]], i32 %[[THREAD_NUM1]])
+!CHECK:         br label %[[OMP_REGION_END]]
+subroutine test_scalar()
+  integer :: i, j
+
+  !$omp parallel private(i, j)
+  !$omp single
+  i = 11
+  j = 22
+  !$omp end single copyprivate(i, j)
+  !$omp end parallel
+end subroutine
+
+subroutine test_types(a, n)
+  integer :: a(:), n
+  integer(4) :: i4, i4a(10)
+  integer(8) :: i8, i8a(n)
+  real :: r, ra(2, 3)
+  complex :: z, za(10)
+  logical :: l, la(5)
+  character(kind=1, len=8) :: c1, c1a(10)
+  character(kind=2, len=5) :: c2
+
+  type dt
+    integer :: i
+    real :: r
+  end type
+  type(dt) :: t
+
+  integer, allocatable :: aloc(:)
+  character(kind=1, len=9), pointer :: ptr(:)
+
+  !$omp parallel private(a, i4, i4a, i8, i8a, r, ra, z, za, l, la, c1, c1a, c2, t, aloc, ptr)
+  !$omp single
+  !$omp end single copyprivate(a, i4, i4a, i8, i8a, r, ra, z, za, l, la, c1, c1a, c2, t, aloc, ptr)
+  !$omp end parallel
+end subroutine

llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h

@@ -1834,13 +1834,15 @@ class OpenMPIRBuilder {
   /// \param BodyGenCB Callback that will generate the region code.
   /// \param FiniCB Callback to finalize variable copies.
   /// \param IsNowait If false, a barrier is emitted.
-  /// \param DidIt Local variable used as a flag to indicate 'single' thread
+  /// \param CPVars copyprivate variables.
+  /// \param CPFuncs copy functions to use for each copyprivate variable.
   ///
   /// \returns The insertion position *after* the single call.
   InsertPointTy createSingle(const LocationDescription &Loc,
                              BodyGenCallbackTy BodyGenCB,
                              FinalizeCallbackTy FiniCB, bool IsNowait,
-                             llvm::Value *DidIt);
+                             ArrayRef<llvm::Value *> CPVars = {},
+                             ArrayRef<llvm::Function *> CPFuncs = {});
 
   /// Generator for '#omp master'
   ///

llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

@@ -4047,13 +4047,17 @@ OpenMPIRBuilder::createCopyPrivate(const LocationDescription &Loc,
 
 OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createSingle(
     const LocationDescription &Loc, BodyGenCallbackTy BodyGenCB,
-    FinalizeCallbackTy FiniCB, bool IsNowait, llvm::Value *DidIt) {
+    FinalizeCallbackTy FiniCB, bool IsNowait, ArrayRef<llvm::Value *> CPVars,
+    ArrayRef<llvm::Function *> CPFuncs) {
 
   if (!updateToLocation(Loc))
     return Loc.IP;
 
-  // If needed (i.e. not null), initialize `DidIt` with 0
-  if (DidIt) {
+  // If needed allocate and initialize `DidIt` with 0.
+  // DidIt: flag variable: 1=single thread; 0=not single thread.
+  llvm::Value *DidIt = nullptr;
+  if (!CPVars.empty()) {
+    DidIt = Builder.CreateAlloca(llvm::Type::getInt32Ty(Builder.getContext()));
     Builder.CreateStore(Builder.getInt32(0), DidIt);
   }
 
@@ -4070,17 +4074,36 @@ OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createSingle(
   Function *ExitRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_single);
   Instruction *ExitCall = Builder.CreateCall(ExitRTLFn, Args);
 
+  auto FiniCBWrapper = [&](InsertPointTy IP) {
+    FiniCB(IP);
+
+    // The thread that executes the single region must set `DidIt` to 1.
+    // This is used by __kmpc_copyprivate, to know if the caller is the
+    // single thread or not.
+    if (DidIt)
+      Builder.CreateStore(Builder.getInt32(1), DidIt);
+  };
+
   // generates the following:
   // if (__kmpc_single()) {
   //		.... single region ...
   // 		__kmpc_end_single
   // }
+  // __kmpc_copyprivate
   // __kmpc_barrier
 
-  EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
+  EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCBWrapper,
                        /*Conditional*/ true,
                        /*hasFinalize*/ true);
-  if (!IsNowait)
+
+  if (DidIt) {
+    for (size_t I = 0, E = CPVars.size(); I < E; ++I)
+      // NOTE BufSize is currently unused, so just pass 0.
+      createCopyPrivate(LocationDescription(Builder.saveIP(), Loc.DL),
+                        /*BufSize=*/ConstantInt::get(Int64, 0), CPVars[I],
+                        CPFuncs[I], DidIt);
+    // NOTE __kmpc_copyprivate already inserts a barrier
+  } else if (!IsNowait)
     createBarrier(LocationDescription(Builder.saveIP(), Loc.DL),
                   omp::Directive::OMPD_unknown, /* ForceSimpleCall */ false,
                   /* CheckCancelFlag */ false);

llvm/unittests/Frontend/OpenMPIRBuilderTest.cpp

@@ -3327,8 +3327,8 @@ TEST_F(OpenMPIRBuilderTest, SingleDirective) {
     EXPECT_NE(IPBB->end(), IP.getPoint());
   };
 
-  Builder.restoreIP(OMPBuilder.createSingle(
-      Builder, BodyGenCB, FiniCB, /*IsNowait*/ false, /*DidIt*/ nullptr));
+  Builder.restoreIP(
+      OMPBuilder.createSingle(Builder, BodyGenCB, FiniCB, /*IsNowait*/ false));
   Value *EntryBBTI = EntryBB->getTerminator();
   EXPECT_NE(EntryBBTI, nullptr);
   EXPECT_TRUE(isa<BranchInst>(EntryBBTI));
@@ -3417,8 +3417,8 @@ TEST_F(OpenMPIRBuilderTest, SingleDirectiveNowait) {
     EXPECT_NE(IPBB->end(), IP.getPoint());
   };
 
-  Builder.restoreIP(OMPBuilder.createSingle(
-      Builder, BodyGenCB, FiniCB, /*IsNowait*/ true, /*DidIt*/ nullptr));
+  Builder.restoreIP(
+      OMPBuilder.createSingle(Builder, BodyGenCB, FiniCB, /*IsNowait*/ true));
   Value *EntryBBTI = EntryBB->getTerminator();
   EXPECT_NE(EntryBBTI, nullptr);
   EXPECT_TRUE(isa<BranchInst>(EntryBBTI));
@@ -3464,6 +3464,151 @@ TEST_F(OpenMPIRBuilderTest, SingleDirectiveNowait) {
   EXPECT_EQ(ExitBarrier, nullptr);
 }
 
+// Helper class to check each instruction of a BB.
+class BBInstIter {
+  BasicBlock *BB;
+  BasicBlock::iterator BBI;
+
+public:
+  BBInstIter(BasicBlock *BB) : BB(BB), BBI(BB->begin()) {}
+
+  bool hasNext() const { return BBI != BB->end(); }
+
+  template <typename InstTy> InstTy *next() {
+    if (!hasNext())
+      return nullptr;
+    Instruction *Cur = &*BBI++;
+    if (!isa<InstTy>(Cur))
+      return nullptr;
+    return cast<InstTy>(Cur);
+  }
+};
+
+TEST_F(OpenMPIRBuilderTest, SingleDirectiveCopyPrivate) {
+  using InsertPointTy = OpenMPIRBuilder::InsertPointTy;
+  OpenMPIRBuilder OMPBuilder(*M);
+  OMPBuilder.initialize();
+  F->setName("func");
+  IRBuilder<> Builder(BB);
+
+  OpenMPIRBuilder::LocationDescription Loc({Builder.saveIP(), DL});
+
+  AllocaInst *PrivAI = nullptr;
+
+  BasicBlock *EntryBB = nullptr;
+  BasicBlock *ThenBB = nullptr;
+
+  Value *CPVar = Builder.CreateAlloca(F->arg_begin()->getType());
+  Builder.CreateStore(F->arg_begin(), CPVar);
+
+  FunctionType *CopyFuncTy = FunctionType::get(
+      Builder.getVoidTy(), {Builder.getPtrTy(), Builder.getPtrTy()}, false);
+  Function *CopyFunc =
+      Function::Create(CopyFuncTy, Function::PrivateLinkage, "copy_var", *M);
+
+  auto BodyGenCB = [&](InsertPointTy AllocaIP, InsertPointTy CodeGenIP) {
+    if (AllocaIP.isSet())
+      Builder.restoreIP(AllocaIP);
+    else
+      Builder.SetInsertPoint(&*(F->getEntryBlock().getFirstInsertionPt()));
+    PrivAI = Builder.CreateAlloca(F->arg_begin()->getType());
+    Builder.CreateStore(F->arg_begin(), PrivAI);
+
+    llvm::BasicBlock *CodeGenIPBB = CodeGenIP.getBlock();
+    llvm::Instruction *CodeGenIPInst = &*CodeGenIP.getPoint();
+    EXPECT_EQ(CodeGenIPBB->getTerminator(), CodeGenIPInst);
+
+    Builder.restoreIP(CodeGenIP);
+
+    // collect some info for checks later
+    ThenBB = Builder.GetInsertBlock();
+    EntryBB = ThenBB->getUniquePredecessor();
+
+    // simple instructions for body
+    Value *PrivLoad =
+        Builder.CreateLoad(PrivAI->getAllocatedType(), PrivAI, "local.use");
+    Builder.CreateICmpNE(F->arg_begin(), PrivLoad);
+  };
+
+  auto FiniCB = [&](InsertPointTy IP) {
+    BasicBlock *IPBB = IP.getBlock();
+    // IP must be before the unconditional branch to ExitBB
+    EXPECT_NE(IPBB->end(), IP.getPoint());
+  };
+
+  Builder.restoreIP(OMPBuilder.createSingle(Builder, BodyGenCB, FiniCB,
+                                            /*IsNowait*/ false, {CPVar},
+                                            {CopyFunc}));
+  Value *EntryBBTI = EntryBB->getTerminator();
+  EXPECT_NE(EntryBBTI, nullptr);
+  EXPECT_TRUE(isa<BranchInst>(EntryBBTI));
+  BranchInst *EntryBr = cast<BranchInst>(EntryBB->getTerminator());
+  EXPECT_TRUE(EntryBr->isConditional());
+  EXPECT_EQ(EntryBr->getSuccessor(0), ThenBB);
+  BasicBlock *ExitBB = ThenBB->getUniqueSuccessor();
+  EXPECT_EQ(EntryBr->getSuccessor(1), ExitBB);
+
+  CmpInst *CondInst = cast<CmpInst>(EntryBr->getCondition());
+  EXPECT_TRUE(isa<CallInst>(CondInst->getOperand(0)));
+
+  CallInst *SingleEntryCI = cast<CallInst>(CondInst->getOperand(0));
+  EXPECT_EQ(SingleEntryCI->arg_size(), 2U);
+  EXPECT_EQ(SingleEntryCI->getCalledFunction()->getName(), "__kmpc_single");
+  EXPECT_TRUE(isa<GlobalVariable>(SingleEntryCI->getArgOperand(0)));
+
+  // check ThenBB
+  BBInstIter ThenBBI(ThenBB);
+  // load PrivAI
+  auto *PrivLI = ThenBBI.next<LoadInst>();
+  EXPECT_NE(PrivLI, nullptr);
+  EXPECT_EQ(PrivLI->getPointerOperand(), PrivAI);
+  // icmp
+  EXPECT_TRUE(ThenBBI.next<ICmpInst>());
+  // store 1, DidIt
+  auto *DidItSI = ThenBBI.next<StoreInst>();
+  EXPECT_NE(DidItSI, nullptr);
+  EXPECT_EQ(DidItSI->getValueOperand(),
+            ConstantInt::get(Type::getInt32Ty(Ctx), 1));
+  Value *DidIt = DidItSI->getPointerOperand();
+  // call __kmpc_end_single
+  auto *SingleEndCI = ThenBBI.next<CallInst>();
+  EXPECT_NE(SingleEndCI, nullptr);
+  EXPECT_EQ(SingleEndCI->getCalledFunction()->getName(), "__kmpc_end_single");
+  EXPECT_EQ(SingleEndCI->arg_size(), 2U);
+  EXPECT_TRUE(isa<GlobalVariable>(SingleEndCI->getArgOperand(0)));
+  EXPECT_EQ(SingleEndCI->getArgOperand(1), SingleEntryCI->getArgOperand(1));
+  // br ExitBB
+  auto *ExitBBBI = ThenBBI.next<BranchInst>();
+  EXPECT_NE(ExitBBBI, nullptr);
+  EXPECT_TRUE(ExitBBBI->isUnconditional());
+  EXPECT_EQ(ExitBBBI->getOperand(0), ExitBB);
+  EXPECT_FALSE(ThenBBI.hasNext());
+
+  // check ExitBB
+  BBInstIter ExitBBI(ExitBB);
+  // call __kmpc_global_thread_num
+  auto *ThreadNumCI = ExitBBI.next<CallInst>();
+  EXPECT_NE(ThreadNumCI, nullptr);
+  EXPECT_EQ(ThreadNumCI->getCalledFunction()->getName(),
+            "__kmpc_global_thread_num");
+  // load DidIt
+  auto *DidItLI = ExitBBI.next<LoadInst>();
+  EXPECT_NE(DidItLI, nullptr);
+  EXPECT_EQ(DidItLI->getPointerOperand(), DidIt);
+  // call __kmpc_copyprivate
+  auto *CopyPrivateCI = ExitBBI.next<CallInst>();
+  EXPECT_NE(CopyPrivateCI, nullptr);
+  EXPECT_EQ(CopyPrivateCI->arg_size(), 6U);
+  EXPECT_TRUE(isa<AllocaInst>(CopyPrivateCI->getArgOperand(3)));
+  EXPECT_EQ(CopyPrivateCI->getArgOperand(3), CPVar);
+  EXPECT_TRUE(isa<Function>(CopyPrivateCI->getArgOperand(4)));
+  EXPECT_EQ(CopyPrivateCI->getArgOperand(4), CopyFunc);
+  EXPECT_TRUE(isa<LoadInst>(CopyPrivateCI->getArgOperand(5)));
+  DidItLI = cast<LoadInst>(CopyPrivateCI->getArgOperand(5));
+  EXPECT_EQ(DidItLI->getOperand(0), DidIt);
+  EXPECT_FALSE(ExitBBI.hasNext());
+}
+
 TEST_F(OpenMPIRBuilderTest, OMPAtomicReadFlt) {
   OpenMPIRBuilder OMPBuilder(*M);
   OMPBuilder.initialize();

mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp

@@ -667,8 +667,22 @@ convertOmpSingle(omp::SingleOp &singleOp, llvm::IRBuilderBase &builder,
                         moduleTranslation, bodyGenStatus);
   };
   auto finiCB = [&](InsertPointTy codeGenIP) {};
+
+  // Handle copyprivate
+  Operation::operand_range cpVars = singleOp.getCopyprivateVars();
+  std::optional<ArrayAttr> cpFuncs = singleOp.getCopyprivateFuncs();
+  llvm::SmallVector<llvm::Value *> llvmCPVars;
+  llvm::SmallVector<llvm::Function *> llvmCPFuncs;
+  for (size_t i = 0, e = cpVars.size(); i < e; ++i) {
+    llvmCPVars.push_back(moduleTranslation.lookupValue(cpVars[i]));
+    auto llvmFuncOp = SymbolTable::lookupNearestSymbolFrom<LLVM::LLVMFuncOp>(
+        singleOp, cast<SymbolRefAttr>((*cpFuncs)[i]));
+    llvmCPFuncs.push_back(
+        moduleTranslation.lookupFunction(llvmFuncOp.getName()));
+  }
+
   builder.restoreIP(moduleTranslation.getOpenMPBuilder()->createSingle(
-      ompLoc, bodyCB, finiCB, singleOp.getNowait(), /*DidIt=*/nullptr));
+      ompLoc, bodyCB, finiCB, singleOp.getNowait(), llvmCPVars, llvmCPFuncs));
   return bodyGenStatus;
 }