[flang-commits] [flang] [OpenMP][Flang] Add "IsolatedFromAbove" trait to omp.target (PR #67164)

Akash Banerjee via flang-commits flang-commits at lists.llvm.org
Thu Oct 26 04:40:31 PDT 2023


https://github.com/TIFitis updated https://github.com/llvm/llvm-project/pull/67164

>From 8277dd9d0f8e01b756b611bfdef92b96ce698f22 Mon Sep 17 00:00:00 2001
From: Akash Banerjee <Akash.Banerjee at amd.com>
Date: Fri, 22 Sep 2023 18:12:06 +0100
Subject: [PATCH] [OpenMP][Flang] Add "IsolatedFromAbove" trait to omp.target

This patch adds the PFT lowering changes required for adding the IsolatedFromAbove and OutlineableOpenMPOpInterface traits to omp.target.

Key Changes:
	- Add IsolatedFromAbove and OutlineableOpenMPOpInterface traits to target op in MLIR.
	- Main reason for this change is to prevent CSE and other similar optimisations from crossing region boundaries for target operations. The link below has the discourse discussion surrounding this issue.
	- Move implicit operand capturing to the PFT lowering stage.
	- Update related tests.

Related discussion: https://discourse.llvm.org/t/rfc-prevent-cse-from-removing-expressions-inside-some-non-isolatedfromabove-operation-regions/73150
---
 flang/lib/Lower/OpenMP.cpp                    | 270 +++++++++++++++---
 .../Driver/OpenMP/map-types-and-sizes.f90     |  52 ++++
 .../Fir/convert-to-llvm-openmp-and-fir.fir    |  10 +-
 flang/test/Lower/OpenMP/FIR/array-bounds.f90  | 119 +++-----
 flang/test/Lower/OpenMP/FIR/location.f90      |   2 +-
 flang/test/Lower/OpenMP/FIR/target.f90        | 106 +++++--
 flang/test/Lower/OpenMP/array-bounds.f90      |   6 +-
 flang/test/Lower/OpenMP/location.f90          |   2 +-
 flang/test/Lower/OpenMP/target.f90            | 118 +++++++-
 mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td |  35 +--
 mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp  |  70 +++++
 11 files changed, 621 insertions(+), 169 deletions(-)
 create mode 100644 flang/test/Driver/OpenMP/map-types-and-sizes.f90

diff --git a/flang/lib/Lower/OpenMP.cpp b/flang/lib/Lower/OpenMP.cpp
index 0faaae6c08e0476..b2a32498ad390c6 100644
--- a/flang/lib/Lower/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP.cpp
@@ -560,11 +560,21 @@ class ClauseProcessor {
             mlir::Value &result) const;
   bool
   processLink(llvm::SmallVectorImpl<DeclareTargetCapturePair> &result) const;
+
+  // This method is used to process a map clause.
+  // The optional parameters - mapSymTypes, mapSymLocs & mapSymbols are used to
+  // store the original type, location and Fortran symbol for the map operands.
+  // They may be used later on to create the block_arguments for some of the
+  // target directives that require it.
   bool processMap(mlir::Location currentLocation,
                   const llvm::omp::Directive &directive,
                   Fortran::semantics::SemanticsContext &semanticsContext,
                   Fortran::lower::StatementContext &stmtCtx,
-                  llvm::SmallVectorImpl<mlir::Value> &mapOperands) const;
+                  llvm::SmallVectorImpl<mlir::Value> &mapOperands,
+                  llvm::SmallVectorImpl<mlir::Type> *mapSymTypes = nullptr,
+                  llvm::SmallVectorImpl<mlir::Location> *mapSymLocs = nullptr,
+                  llvm::SmallVectorImpl<const Fortran::semantics::Symbol *>
+                      *mapSymbols = nullptr) const;
   bool processReduction(
       mlir::Location currentLocation,
       llvm::SmallVectorImpl<mlir::Value> &reductionVars,
@@ -1691,31 +1701,32 @@ static mlir::omp::MapInfoOp
 createMapInfoOp(fir::FirOpBuilder &builder, mlir::Location loc,
                 mlir::Value baseAddr, std::stringstream &name,
                 mlir::SmallVector<mlir::Value> bounds, uint64_t mapType,
-                mlir::omp::VariableCaptureKind mapCaptureType, bool implicit,
-                mlir::Type retTy) {
-  mlir::Value varPtrPtr;
+                mlir::omp::VariableCaptureKind mapCaptureType, mlir::Type retTy,
+                bool isVal = false) {
+  mlir::Value val, varPtr, varPtrPtr;
+  mlir::TypeAttr varType;
+
   if (auto boxTy = baseAddr.getType().dyn_cast<fir::BaseBoxType>()) {
     baseAddr = builder.create<fir::BoxAddrOp>(loc, baseAddr);
     retTy = baseAddr.getType();
   }
 
-  mlir::omp::MapInfoOp op =
-      builder.create<mlir::omp::MapInfoOp>(loc, retTy, baseAddr);
-  op.setVarTypeAttr(mlir::TypeAttr::get(
-      llvm::dyn_cast<mlir::omp::PointerLikeType>(retTy).getElementType()));
-  op.setNameAttr(builder.getStringAttr(name.str()));
-  op.setImplicit(implicit);
-  op.setMapType(mapType);
-  op.setMapCaptureType(mapCaptureType);
-
-  unsigned insPos = 1;
-  if (varPtrPtr)
-    op->insertOperands(insPos++, varPtrPtr);
-  if (bounds.size() > 0)
-    op->insertOperands(insPos, bounds);
-  op->setAttr(mlir::omp::MapInfoOp::getOperandSegmentSizeAttr(),
-              builder.getDenseI32ArrayAttr(
-                  {1, varPtrPtr ? 1 : 0, static_cast<int32_t>(bounds.size())}));
+  if (isVal)
+    val = baseAddr;
+  else
+    varPtr = baseAddr;
+
+  if (auto ptrType = llvm::dyn_cast<mlir::omp::PointerLikeType>(retTy)) {
+    varType = mlir::TypeAttr::get(ptrType.getElementType());
+  } else {
+    varType = mlir::TypeAttr::get(retTy);
+  }
+
+  mlir::omp::MapInfoOp op = builder.create<mlir::omp::MapInfoOp>(
+      loc, retTy, val, varPtr, varType, varPtrPtr, bounds,
+      builder.getIntegerAttr(builder.getIntegerType(64, false), mapType),
+      builder.getAttr<mlir::omp::VariableCaptureKindAttr>(mapCaptureType),
+      builder.getStringAttr(name.str()));
   return op;
 }
 
@@ -1723,7 +1734,11 @@ bool ClauseProcessor::processMap(
     mlir::Location currentLocation, const llvm::omp::Directive &directive,
     Fortran::semantics::SemanticsContext &semanticsContext,
     Fortran::lower::StatementContext &stmtCtx,
-    llvm::SmallVectorImpl<mlir::Value> &mapOperands) const {
+    llvm::SmallVectorImpl<mlir::Value> &mapOperands,
+    llvm::SmallVectorImpl<mlir::Type> *mapSymTypes,
+    llvm::SmallVectorImpl<mlir::Location> *mapSymLocs,
+    llvm::SmallVectorImpl<const Fortran::semantics::Symbol *> *mapSymbols)
+    const {
   fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
   return findRepeatableClause<ClauseTy::Map>(
       [&](const ClauseTy::Map *mapClause,
@@ -1783,13 +1798,20 @@ bool ClauseProcessor::processMap(
           // Explicit map captures are captured ByRef by default,
           // optimisation passes may alter this to ByCopy or other capture
           // types to optimise
-          mapOperands.push_back(createMapInfoOp(
+          mlir::Value mapOp = createMapInfoOp(
               firOpBuilder, clauseLocation, baseAddr, asFortran, bounds,
               static_cast<
                   std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
                   mapTypeBits),
-              mlir::omp::VariableCaptureKind::ByRef, false,
-              baseAddr.getType()));
+              mlir::omp::VariableCaptureKind::ByRef, baseAddr.getType());
+
+          mapOperands.push_back(mapOp);
+          if (mapSymTypes)
+            mapSymTypes->push_back(baseAddr.getType());
+          if (mapSymLocs)
+            mapSymLocs->push_back(baseAddr.getLoc());
+          if (mapSymbols)
+            mapSymbols->push_back(getOmpObjectSymbol(ompObject));
         }
       });
 }
@@ -2172,7 +2194,7 @@ static void createBodyOfOp(
   }
 }
 
-static void createBodyOfTargetDataOp(
+static void genBodyOfTargetDataOp(
     Fortran::lower::AbstractConverter &converter, mlir::omp::DataOp &dataOp,
     const llvm::SmallVector<mlir::Type> &useDeviceTypes,
     const llvm::SmallVector<mlir::Location> &useDeviceLocs,
@@ -2189,18 +2211,17 @@ static void createBodyOfTargetDataOp(
   unsigned argIndex = 0;
   for (const Fortran::semantics::Symbol *sym : useDeviceSymbols) {
     const mlir::BlockArgument &arg = region.front().getArgument(argIndex);
-    mlir::Value val = fir::getBase(arg);
     fir::ExtendedValue extVal = converter.getSymbolExtendedValue(*sym);
-    if (auto refType = val.getType().dyn_cast<fir::ReferenceType>()) {
+    if (auto refType = arg.getType().dyn_cast<fir::ReferenceType>()) {
       if (fir::isa_builtin_cptr_type(refType.getElementType())) {
-        converter.bindSymbol(*sym, val);
+        converter.bindSymbol(*sym, arg);
       } else {
         extVal.match(
             [&](const fir::MutableBoxValue &mbv) {
               converter.bindSymbol(
                   *sym,
                   fir::MutableBoxValue(
-                      val, fir::factory::getNonDeferredLenParams(extVal), {}));
+                      arg, fir::factory::getNonDeferredLenParams(extVal), {}));
             },
             [&](const auto &) {
               TODO(converter.getCurrentLocation(),
@@ -2386,8 +2407,8 @@ genDataOp(Fortran::lower::AbstractConverter &converter,
   auto dataOp = converter.getFirOpBuilder().create<mlir::omp::DataOp>(
       currentLocation, ifClauseOperand, deviceOperand, devicePtrOperands,
       deviceAddrOperands, mapOperands);
-  createBodyOfTargetDataOp(converter, dataOp, useDeviceTypes, useDeviceLocs,
-                           useDeviceSymbols, currentLocation);
+  genBodyOfTargetDataOp(converter, dataOp, useDeviceTypes, useDeviceLocs,
+                        useDeviceSymbols, currentLocation);
   return dataOp;
 }
 
@@ -2430,6 +2451,75 @@ genEnterExitDataOp(Fortran::lower::AbstractConverter &converter,
                                    deviceOperand, nowaitAttr, mapOperands);
 }
 
+static void genBodyOfTargetOp(
+    Fortran::lower::AbstractConverter &converter, mlir::omp::TargetOp &targetOp,
+    const llvm::SmallVector<mlir::Type> &mapSymTypes,
+    const llvm::SmallVector<mlir::Location> &mapSymLocs,
+    const llvm::SmallVector<const Fortran::semantics::Symbol *> &mapSymbols,
+    const mlir::Location &currentLocation) {
+  assert(mapSymTypes.size() == mapSymLocs.size());
+
+  fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
+  mlir::Region &region = targetOp.getRegion();
+
+  firOpBuilder.createBlock(&region, {}, mapSymTypes, mapSymLocs);
+  firOpBuilder.create<mlir::omp::TerminatorOp>(currentLocation);
+  firOpBuilder.setInsertionPointToStart(&region.front());
+
+  unsigned argIndex = 0;
+  unsigned blockArgsIndex = mapSymbols.size();
+
+  auto extractBoundArgs = [&](auto n) {
+    llvm::SmallVector<mlir::Value> argExtents;
+    while (n--) {
+      argExtents.push_back(fir::getBase(region.getArgument(blockArgsIndex)));
+      blockArgsIndex++;
+    }
+    return argExtents;
+  };
+
+  for (const Fortran::semantics::Symbol *sym : mapSymbols) {
+    const mlir::BlockArgument &arg = region.getArgument(argIndex);
+    fir::ExtendedValue extVal = converter.getSymbolExtendedValue(*sym);
+    extVal.match(
+        [&](const fir::BoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::BoxValue(arg, extractBoundArgs(v.getLBounds().size()),
+                            v.getExplicitParameters(), v.getExplicitExtents()));
+        },
+        [&](const fir::MutableBoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::MutableBoxValue(arg, extractBoundArgs(v.getLBounds().size()),
+                                   v.getMutableProperties()));
+        },
+        [&](const fir::ArrayBoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::ArrayBoxValue(arg, extractBoundArgs(v.getExtents().size()),
+                                 extractBoundArgs(v.getLBounds().size()),
+                                 v.getSourceBox()));
+        },
+        [&](const fir::CharArrayBoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::CharArrayBoxValue(arg, v.getLen(),
+                                     extractBoundArgs(v.getExtents().size()),
+                                     extractBoundArgs(v.getLBounds().size())));
+        },
+        [&](const fir::CharBoxValue &v) {
+          converter.bindSymbol(*sym, fir::CharBoxValue(arg, v.getLen()));
+        },
+        [&](const fir::UnboxedValue &v) { converter.bindSymbol(*sym, arg); },
+        [&](const auto &) {
+          TODO(converter.getCurrentLocation(),
+               "target map clause operand unsupported type");
+        });
+    argIndex++;
+  }
+}
+
 static mlir::omp::TargetOp
 genTargetOp(Fortran::lower::AbstractConverter &converter,
             Fortran::lower::pft::Evaluation &eval,
@@ -2441,6 +2531,9 @@ genTargetOp(Fortran::lower::AbstractConverter &converter,
   mlir::Value ifClauseOperand, deviceOperand, threadLimitOperand;
   mlir::UnitAttr nowaitAttr;
   llvm::SmallVector<mlir::Value> mapOperands;
+  llvm::SmallVector<mlir::Type> mapSymTypes;
+  llvm::SmallVector<mlir::Location> mapSymLocs;
+  llvm::SmallVector<const Fortran::semantics::Symbol *> mapSymbols;
 
   ClauseProcessor cp(converter, clauseList);
   cp.processIf(stmtCtx,
@@ -2450,7 +2543,7 @@ genTargetOp(Fortran::lower::AbstractConverter &converter,
   cp.processThreadLimit(stmtCtx, threadLimitOperand);
   cp.processNowait(nowaitAttr);
   cp.processMap(currentLocation, directive, semanticsContext, stmtCtx,
-                mapOperands);
+                mapOperands, &mapSymTypes, &mapSymLocs, &mapSymbols);
   cp.processTODO<Fortran::parser::OmpClause::Private,
                  Fortran::parser::OmpClause::Depend,
                  Fortran::parser::OmpClause::Firstprivate,
@@ -2463,10 +2556,113 @@ genTargetOp(Fortran::lower::AbstractConverter &converter,
                  Fortran::parser::OmpClause::Defaultmap>(
       currentLocation, llvm::omp::Directive::OMPD_target);
 
-  return genOpWithBody<mlir::omp::TargetOp>(
-      converter, eval, currentLocation, outerCombined, &clauseList,
-      ifClauseOperand, deviceOperand, threadLimitOperand, nowaitAttr,
-      mapOperands);
+  auto captureImplicitMap = [&](const Fortran::semantics::Symbol &sym) {
+    if (llvm::find(mapSymbols, &sym) == mapSymbols.end()) {
+      mlir::Value baseOp = converter.getSymbolAddress(sym);
+      if (!baseOp)
+        if (const auto *details = sym.template detailsIf<
+                                  Fortran::semantics::HostAssocDetails>()) {
+          baseOp = converter.getSymbolAddress(details->symbol());
+          converter.copySymbolBinding(details->symbol(), sym);
+        }
+
+      if (baseOp) {
+        llvm::SmallVector<mlir::Value> bounds;
+        std::stringstream name;
+        fir::ExtendedValue dataExv = converter.getSymbolExtendedValue(sym);
+        name << sym.name().ToString();
+
+        mlir::Value baseAddr =
+            getDataOperandBaseAddr(converter, converter.getFirOpBuilder(), sym,
+                                   converter.getCurrentLocation());
+        if (fir::unwrapRefType(baseAddr.getType()).isa<fir::BaseBoxType>())
+          bounds =
+              Fortran::lower::genBoundsOpsFromBox<mlir::omp::DataBoundsOp,
+                                                  mlir::omp::DataBoundsType>(
+                  converter.getFirOpBuilder(), converter.getCurrentLocation(),
+                  converter, dataExv, baseAddr);
+        if (fir::unwrapRefType(baseAddr.getType()).isa<fir::SequenceType>())
+          bounds = Fortran::lower::genBaseBoundsOps<mlir::omp::DataBoundsOp,
+                                                    mlir::omp::DataBoundsType>(
+              converter.getFirOpBuilder(), converter.getCurrentLocation(),
+              converter, dataExv, baseAddr);
+
+        llvm::omp::OpenMPOffloadMappingFlags mapFlag =
+            llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT;
+        mlir::omp::VariableCaptureKind captureKind =
+            mlir::omp::VariableCaptureKind::ByRef;
+        if (auto refType = baseOp.getType().dyn_cast<fir::ReferenceType>()) {
+          auto eleType = refType.getElementType();
+          if (fir::isa_trivial(eleType) || fir::isa_char(eleType)) {
+            // mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_LITERAL;
+            captureKind = mlir::omp::VariableCaptureKind::ByCopy;
+          } else if (!fir::isa_builtin_cptr_type(eleType)) {
+            mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TO;
+            mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_FROM;
+          }
+        }
+
+        mlir::Value mapOp = createMapInfoOp(
+            converter.getFirOpBuilder(), baseOp.getLoc(), baseOp, name, bounds,
+            static_cast<
+                std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
+                mapFlag),
+            captureKind, baseOp.getType());
+
+        mapOperands.push_back(mapOp);
+        mapSymTypes.push_back(baseOp.getType());
+        mapSymLocs.push_back(baseOp.getLoc());
+        mapSymbols.push_back(&sym);
+      }
+    }
+  };
+  Fortran::lower::pft::visitAllSymbols(eval, captureImplicitMap);
+
+  // Add the bounds and extents for box values to mapOperands
+  auto addBoundsInfo = [&](const auto &bounds) {
+    for (auto &val : bounds) {
+      mapSymLocs.push_back(val.getLoc());
+      mapSymTypes.push_back(val.getType());
+
+      llvm::SmallVector<mlir::Value> bounds;
+      std::stringstream name;
+
+      mlir::Value mapOp = createMapInfoOp(
+          converter.getFirOpBuilder(), val.getLoc(), val, name, bounds,
+          static_cast<
+              std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
+              llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT),
+          mlir::omp::VariableCaptureKind::ByCopy, val.getType(), true);
+      mapOperands.push_back(mapOp);
+    }
+  };
+
+  for (const Fortran::semantics::Symbol *sym : mapSymbols) {
+    fir::ExtendedValue extVal = converter.getSymbolExtendedValue(*sym);
+    extVal.match(
+        [&](const fir::BoxValue &v) { addBoundsInfo(v.getLBounds()); },
+        [&](const fir::MutableBoxValue &v) { addBoundsInfo(v.getLBounds()); },
+        [&](const fir::ArrayBoxValue &v) {
+          addBoundsInfo(v.getExtents());
+          addBoundsInfo(v.getLBounds());
+        },
+        [&](const fir::CharArrayBoxValue &v) {
+          addBoundsInfo(v.getExtents());
+          addBoundsInfo(v.getLBounds());
+        },
+        [&](const auto &) {
+          // Nothing to do for non-box values.
+        });
+  }
+
+  auto targetOp = converter.getFirOpBuilder().create<mlir::omp::TargetOp>(
+      currentLocation, ifClauseOperand, deviceOperand, threadLimitOperand,
+      nowaitAttr, mapOperands);
+
+  genBodyOfTargetOp(converter, targetOp, mapSymTypes, mapSymLocs, mapSymbols,
+                    currentLocation);
+
+  return targetOp;
 }
 
 static mlir::omp::TeamsOp
diff --git a/flang/test/Driver/OpenMP/map-types-and-sizes.f90 b/flang/test/Driver/OpenMP/map-types-and-sizes.f90
new file mode 100644
index 000000000000000..cbbe4eb9a507628
--- /dev/null
+++ b/flang/test/Driver/OpenMP/map-types-and-sizes.f90
@@ -0,0 +1,52 @@
+!RUN: %flang_fc1 -emit-llvm -fopenmp %s -o - | FileCheck %s
+
+!===============================================================================
+! Check MapTypes for target implicit captures
+!===============================================================================
+
+!CHECK: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 4]
+!CHECK: @.offload_maptypes = private unnamed_addr constant [1 x i64] [i64 800]
+subroutine mapType_scalar
+  integer :: a
+  !$omp target
+     a = 10
+  !$omp end target
+end subroutine mapType_scalar
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 4096]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 547]
+subroutine mapType_array
+  integer :: a(1024)
+  !$omp target
+     a(10) = 20
+  !$omp end target
+end subroutine mapType_array
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 8]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 547]
+subroutine mapType_ptr
+  integer, pointer :: a
+  !$omp target
+     a = 10
+  !$omp end target
+end subroutine mapType_ptr
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [2 x i64] [i64 8, i64 4]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [2 x i64] [i64 544, i64 800]
+subroutine mapType_c_ptr
+  use iso_c_binding, only : c_ptr, c_loc
+  type(c_ptr) :: a
+  integer, target :: b
+  !$omp target
+     a = c_loc(b)
+  !$omp end target
+end subroutine mapType_c_ptr
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 1]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 800]
+subroutine mapType_char
+  character :: a
+  !$omp target
+     a = 'b'
+  !$omp end target
+end subroutine mapType_char
\ No newline at end of file
diff --git a/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
index ecfb8e52cae783f..6efa4d0a095869a 100644
--- a/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
+++ b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
@@ -433,12 +433,13 @@ func.func @_QPomp_target() {
   %1 = arith.subi %c512, %c1 : index
   %2 = omp.bounds   lower_bound(%c0 : index) upper_bound(%1 : index) extent(%c512 : index) stride(%c1 : index) start_idx(%c1 : index)
   %3 = omp.map_info var_ptr(%0 : !fir.ref<!fir.array<512xi32>>, !fir.array<512xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%2) -> !fir.ref<!fir.array<512xi32>> {name = "a"}
-  omp.target   thread_limit(%c64_i32 : i32) map_entries(%3 : !fir.ref<!fir.array<512xi32>>) {
+  omp.target   thread_limit(%c64_i32 : i32) map_entries(%3 -> %arg0 : !fir.ref<!fir.array<512xi32>>) {
+    ^bb0(%arg0: !fir.ref<!fir.array<512xi32>>):
     %c10_i32 = arith.constant 10 : i32
     %c1_i64 = arith.constant 1 : i64
     %c1_i64_0 = arith.constant 1 : i64
     %4 = arith.subi %c1_i64, %c1_i64_0 : i64
-    %5 = fir.coordinate_of %0, %4 : (!fir.ref<!fir.array<512xi32>>, i64) -> !fir.ref<i32>
+    %5 = fir.coordinate_of %arg0, %4 : (!fir.ref<!fir.array<512xi32>>, i64) -> !fir.ref<i32>
     fir.store %c10_i32 to %5 : !fir.ref<i32>
     omp.terminator
   }
@@ -455,12 +456,13 @@ func.func @_QPomp_target() {
 // CHECK:           %[[UPPER:.*]] = llvm.mlir.constant(511 : index) : i64
 // CHECK:           %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[LOWER]] : i64) upper_bound(%[[UPPER]] : i64) extent(%[[EXTENT]] : i64) stride(%[[STRIDE]] : i64) start_idx(%[[STRIDE]] : i64)
 // CHECK:           %[[MAP:.*]] = omp.map_info var_ptr(%2 : !llvm.ptr, !llvm.array<512 x i32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !llvm.ptr {name = "a"}
-// CHECK:           omp.target   thread_limit(%[[VAL_2]] : i32) map_entries(%[[MAP]] : !llvm.ptr) {
+// CHECK:           omp.target   thread_limit(%[[VAL_2]] : i32) map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !llvm.ptr) {
+// CHECK:           ^bb0(%[[ARG_0]]: !llvm.ptr):
 // CHECK:             %[[VAL_3:.*]] = llvm.mlir.constant(10 : i32) : i32
 // CHECK:             %[[VAL_4:.*]] = llvm.mlir.constant(1 : i64) : i64
 // CHECK:             %[[VAL_5:.*]] = llvm.mlir.constant(1 : i64) : i64
 // CHECK:             %[[VAL_6:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:             %[[VAL_7:.*]] = llvm.getelementptr %[[VAL_1]][0, %[[VAL_6]]] : (!llvm.ptr, i64) -> !llvm.ptr
+// CHECK:             %[[VAL_7:.*]] = llvm.getelementptr %[[ARG_0]][0, %[[VAL_6]]] : (!llvm.ptr, i64) -> !llvm.ptr
 // CHECK:             llvm.store %[[VAL_3]], %[[VAL_7]] : i32, !llvm.ptr
 // CHECK:             omp.terminator
 // CHECK:           }
diff --git a/flang/test/Lower/OpenMP/FIR/array-bounds.f90 b/flang/test/Lower/OpenMP/FIR/array-bounds.f90
index a17e4b9dfad36d0..abef31af22ba663 100644
--- a/flang/test/Lower/OpenMP/FIR/array-bounds.f90
+++ b/flang/test/Lower/OpenMP/FIR/array-bounds.f90
@@ -1,37 +1,22 @@
-!RUN: %flang_fc1 -emit-fir -fopenmp %s -o - | FileCheck %s --check-prefixes HOST
-!RUN: %flang_fc1 -emit-fir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s --check-prefixes DEVICE
+!RUN: %flang_fc1 -emit-fir -fopenmp %s -o - | FileCheck %s --check-prefixes=HOST,ALL
+!RUN: %flang_fc1 -emit-fir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s --check-prefixes=DEVICE,ALL
 
-!DEVICE-LABEL: func.func @_QPread_write_section_omp_outline_0(
-!DEVICE-SAME: %[[ARG0:.*]]: !fir.ref<i32>, %[[ARG1:.*]]: !fir.ref<!fir.array<10xi32>>, %[[ARG2:.*]]: !fir.ref<!fir.array<10xi32>>) attributes {omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>, omp.outline_parent_name = "_QPread_write_section"} {
-!DEVICE:  %[[C1:.*]] = arith.constant 1 : index
-!DEVICE:  %[[C2:.*]] = arith.constant 4 : index
-!DEVICE:  %[[C3:.*]] = arith.constant 1 : index
-!DEVICE:  %[[C4:.*]] = arith.constant 1 : index
-!DEVICE:  %[[BOUNDS0:.*]] = omp.bounds   lower_bound(%[[C1]] : index) upper_bound(%[[C2]] : index) stride(%[[C4]] : index) start_idx(%[[C4]] : index)
-!DEVICE:  %[[MAP0:.*]] = omp.map_info var_ptr(%[[ARG1]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS0]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_read(2:5)"}
-!DEVICE:  %[[C5:.*]] = arith.constant 1 : index
-!DEVICE:  %[[C6:.*]] = arith.constant 4 : index
-!DEVICE:  %[[C7:.*]] = arith.constant 1 : index
-!DEVICE:  %[[C8:.*]] = arith.constant 1 : index
-!DEVICE:  %[[BOUNDS1:.*]] = omp.bounds   lower_bound(%[[C5]] : index) upper_bound(%[[C6]] : index) stride(%[[C8]] : index) start_idx(%[[C8]] : index)
-!DEVICE:  %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG2]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS1]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_write(2:5)"}
-!DEVICE:  omp.target   map_entries(%[[MAP0]], %[[MAP1]] : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>) {
-
-!HOST-LABEL:  func.func @_QPread_write_section() {
-!HOST:  %0 = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFread_write_sectionEi"}
-!HOST:  %[[READ:.*]] = fir.address_of(@_QFread_write_sectionEsp_read) : !fir.ref<!fir.array<10xi32>>
-!HOST:  %[[WRITE:.*]] = fir.address_of(@_QFread_write_sectionEsp_write) : !fir.ref<!fir.array<10xi32>>
-!HOST:  %[[C1:.*]] = arith.constant 1 : index
-!HOST:  %[[C2:.*]] = arith.constant 1 : index
-!HOST:  %[[C3:.*]] = arith.constant 4 : index
-!HOST:  %[[BOUNDS0:.*]] = omp.bounds   lower_bound(%[[C2]] : index) upper_bound(%[[C3]] : index) stride(%[[C1]] : index) start_idx(%[[C1]] : index)
-!HOST:  %[[MAP0:.*]] = omp.map_info var_ptr(%[[READ]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS0]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_read(2:5)"}
-!HOST:  %[[C4:.*]] = arith.constant 1 : index
-!HOST:  %[[C5:.*]] = arith.constant 1 : index
-!HOST:  %[[C6:.*]] = arith.constant 4 : index
-!HOST:  %[[BOUNDS1:.*]] = omp.bounds   lower_bound(%[[C5]] : index) upper_bound(%[[C6]] : index) stride(%[[C4]] : index) start_idx(%[[C4]] : index)
-!HOST:  %[[MAP1:.*]] = omp.map_info var_ptr(%[[WRITE]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS1]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_write(2:5)"}
-!HOST:  omp.target   map_entries(%[[MAP0]], %[[MAP1]] : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>) {
+!ALL-LABEL: func.func @_QPread_write_section(
+!ALL:  %[[ITER:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFread_write_sectionEi"}
+!ALL:  %[[READ:.*]] = fir.address_of(@_QFread_write_sectionEsp_read) : !fir.ref<!fir.array<10xi32>>
+!ALL:  %[[WRITE:.*]] = fir.address_of(@_QFread_write_sectionEsp_write) : !fir.ref<!fir.array<10xi32>>
+!ALL:  %[[C1:.*]] = arith.constant 1 : index
+!ALL:  %[[C2:.*]] = arith.constant 1 : index
+!ALL:  %[[C3:.*]] = arith.constant 4 : index
+!ALL:  %[[BOUNDS0:.*]] = omp.bounds   lower_bound(%[[C2]] : index) upper_bound(%[[C3]] : index) stride(%[[C1]] : index) start_idx(%[[C1]] : index)
+!ALL:  %[[MAP0:.*]] = omp.map_info var_ptr(%[[READ]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS0]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_read(2:5)"}
+!ALL:  %[[C4:.*]] = arith.constant 1 : index
+!ALL:  %[[C5:.*]] = arith.constant 1 : index
+!ALL:  %[[C6:.*]] = arith.constant 4 : index
+!ALL:  %[[BOUNDS1:.*]] = omp.bounds   lower_bound(%[[C5]] : index) upper_bound(%[[C6]] : index) stride(%[[C4]] : index) start_idx(%[[C4]] : index)
+!ALL:  %[[MAP1:.*]] = omp.map_info var_ptr(%[[WRITE]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS1]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_write(2:5)"}
+!ALL:  %[[MAP2:.*]] = omp.map_info var_ptr(%[[ITER]] : !fir.ref<i32>, i32)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !fir.ref<i32> {name = "i"}
+!ALL: omp.target map_entries(%[[MAP0]] -> %{{.*}}, %[[MAP1]] -> %{{.*}}, %[[MAP2]] -> %{{.*}}, %{{.*}} -> %{{.*}}, %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>, !fir.ref<i32>, index, index) {
 
 subroutine read_write_section()
     integer :: sp_read(10) = (/1,2,3,4,5,6,7,8,9,10/)
@@ -44,33 +29,22 @@ subroutine read_write_section()
 !$omp end target
 end subroutine read_write_section
 
-
 module assumed_array_routines
-    contains
-!DEVICE-LABEL: func.func @_QMassumed_array_routinesPassumed_shape_array_omp_outline_0(
-!DEVICE-SAME: %[[ARG0:.*]]: !fir.ref<i32>, %[[ARG1:.*]]: !fir.box<!fir.array<?xi32>>, %[[ARG2:.*]]: !fir.ref<!fir.array<?xi32>>) attributes {omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>, omp.outline_parent_name = "_QMassumed_array_routinesPassumed_shape_array"} {
-!DEVICE: %[[C0:.*]] = arith.constant 1 : index
-!DEVICE: %[[C1:.*]] = arith.constant 4 : index
-!DEVICE: %[[C2:.*]] = arith.constant 0 : index
-!DEVICE: %[[C3:.*]]:3 = fir.box_dims %[[ARG1]], %[[C2]] : (!fir.box<!fir.array<?xi32>>, index) -> (index, index, index)
-!DEVICE: %[[C4:.*]] = arith.constant 1 : index
-!DEVICE: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C0]] : index) upper_bound(%[[C1]] : index) stride(%[[C3]]#2 : index) start_idx(%[[C4]] : index) {stride_in_bytes = true}
-!DEVICE: %[[ARGADDR:.*]] = fir.box_addr %[[ARG1]] : (!fir.box<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
-!DEVICE: %[[MAP:.*]] = omp.map_info var_ptr(%[[ARGADDR]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!DEVICE: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
+contains
+!ALL-LABEL: func.func @_QMassumed_array_routinesPassumed_shape_array(
+!ALL-SAME: %[[ARG0:.*]]: !fir.box<!fir.array<?xi32>> {fir.bindc_name = "arr_read_write"}) {
+!ALL: %[[ALLOCA:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QMassumed_array_routinesFassumed_shape_arrayEi"}
+!ALL: %[[C0:.*]] = arith.constant 1 : index
+!ALL: %[[C1:.*]] = arith.constant 0 : index
+!ALL: %[[C2:.*]]:3 = fir.box_dims %arg0, %[[C1]] : (!fir.box<!fir.array<?xi32>>, index) -> (index, index, index)
+!ALL: %[[C3:.*]] = arith.constant 1 : index
+!ALL: %[[C4:.*]] = arith.constant 4 : index
+!ALL: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C3]] : index) upper_bound(%[[C4]] : index) stride(%[[C2]]#2 : index) start_idx(%[[C0]] : index) {stride_in_bytes = true}
+!ALL: %[[ADDROF:.*]] = fir.box_addr %arg0 : (!fir.box<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
+!ALL: %[[MAP:.*]] = omp.map_info var_ptr(%[[ADDROF]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
+!ALL: %[[MAP2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<i32>, i32)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !fir.ref<i32> {name = "i"}
+!ALL: omp.target map_entries(%[[MAP]] -> %{{.*}}, %[[MAP2]] -> %{{.*}} : !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>) {
 
-!HOST-LABEL: func.func @_QMassumed_array_routinesPassumed_shape_array(
-!HOST-SAME: %[[ARG0:.*]]: !fir.box<!fir.array<?xi32>> {fir.bindc_name = "arr_read_write"}) {
-!HOST: %[[ALLOCA:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QMassumed_array_routinesFassumed_shape_arrayEi"}
-!HOST: %[[C0:.*]] = arith.constant 1 : index
-!HOST: %[[C1:.*]] = arith.constant 0 : index
-!HOST: %[[C2:.*]]:3 = fir.box_dims %arg0, %[[C1]] : (!fir.box<!fir.array<?xi32>>, index) -> (index, index, index)
-!HOST: %[[C3:.*]] = arith.constant 1 : index
-!HOST: %[[C4:.*]] = arith.constant 4 : index
-!HOST: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C3]] : index) upper_bound(%[[C4]] : index) stride(%[[C2]]#2 : index) start_idx(%[[C0]] : index) {stride_in_bytes = true}
-!HOST: %[[ADDROF:.*]] = fir.box_addr %arg0 : (!fir.box<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
-!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[ADDROF]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!HOST: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
     subroutine assumed_shape_array(arr_read_write)
             integer, intent(inout) :: arr_read_write(:)
 
@@ -81,25 +55,17 @@ subroutine assumed_shape_array(arr_read_write)
         !$omp end target
         end subroutine assumed_shape_array
 
-!DEVICE-LABEL:   func.func @_QMassumed_array_routinesPassumed_size_array_omp_outline_0(
-!DEVICE-SAME:    %[[ARG0:.*]]: !fir.ref<i32>, %[[ARG1:.*]]: !fir.ref<!fir.array<?xi32>>) attributes {omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>, omp.outline_parent_name = "_QMassumed_array_routinesPassumed_size_array"} {
-!DEVICE: %[[C0:.*]] = arith.constant 1 : index
-!DEVICE: %[[C1:.*]] = arith.constant 4 : index
-!DEVICE: %[[C2:.*]] = arith.constant 1 : index
-!DEVICE: %[[C3:.*]] = arith.constant 1 : index
-!DEVICE: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C0]] : index) upper_bound(%[[C1]] : index) stride(%[[C3]] : index) start_idx(%[[C3]] : index)
-!DEVICE: %[[MAP:.*]] = omp.map_info var_ptr(%[[ARG1]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!DEVICE: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
+!ALL-LABEL:   func.func @_QMassumed_array_routinesPassumed_size_array(
+!ALL-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xi32>> {fir.bindc_name = "arr_read_write"}) {
+!ALL: %[[ALLOCA:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QMassumed_array_routinesFassumed_size_arrayEi"}
+!ALL: %[[C0:.*]] = arith.constant 1 : index
+!ALL: %[[C1:.*]] = arith.constant 1 : index
+!ALL: %[[C2:.*]] = arith.constant 4 : index
+!ALL: %[[BOUNDS:.*]]  = omp.bounds   lower_bound(%[[C1]] : index) upper_bound(%[[C2]] : index) stride(%[[C0]] : index) start_idx(%[[C0]] : index)
+!ALL: %[[MAP:.*]] = omp.map_info var_ptr(%[[ARG0]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
+!ALL: %[[MAP2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<i32>, i32)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !fir.ref<i32> {name = "i"}
+!ALL: omp.target map_entries(%[[MAP]] -> %{{.*}}, %[[MAP2]] -> %{{.*}}, %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>, index) {
 
-!HOST-LABEL: func.func @_QMassumed_array_routinesPassumed_size_array(
-!HOST-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xi32>> {fir.bindc_name = "arr_read_write"}) {
-!HOST: %[[ALLOCA:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QMassumed_array_routinesFassumed_size_arrayEi"}
-!HOST: %[[C0:.*]] = arith.constant 1 : index
-!HOST: %[[C1:.*]] = arith.constant 1 : index
-!HOST: %[[C2:.*]] = arith.constant 4 : index
-!HOST: %[[BOUNDS:.*]]  = omp.bounds   lower_bound(%[[C1]] : index) upper_bound(%[[C2]] : index) stride(%[[C0]] : index) start_idx(%[[C0]] : index)
-!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[ARG0]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!HOST: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
         subroutine assumed_size_array(arr_read_write)
             integer, intent(inout) :: arr_read_write(*)
 
@@ -111,6 +77,7 @@ subroutine assumed_size_array(arr_read_write)
         end subroutine assumed_size_array
     end module assumed_array_routines
 
+!DEVICE-NOT:func.func @_QPcall_assumed_shape_and_size_array() {
 
 !HOST-LABEL:func.func @_QPcall_assumed_shape_and_size_array() {
 !HOST:%{{.*}} = arith.constant 20 : index
diff --git a/flang/test/Lower/OpenMP/FIR/location.f90 b/flang/test/Lower/OpenMP/FIR/location.f90
index 0e36e09b19e1942..648377837670329 100644
--- a/flang/test/Lower/OpenMP/FIR/location.f90
+++ b/flang/test/Lower/OpenMP/FIR/location.f90
@@ -17,7 +17,7 @@ subroutine sub_parallel()
 !CHECK-LABEL: sub_target
 subroutine sub_target()
   print *, x
-!CHECK: omp.target  {
+!CHECK: omp.target {{.*}} {
   !$omp target
     print *, x
 !CHECK:   omp.terminator loc(#[[TAR_LOC:.*]])
diff --git a/flang/test/Lower/OpenMP/FIR/target.f90 b/flang/test/Lower/OpenMP/FIR/target.f90
index 95c57c9a2240142..d03fc7756ef290d 100644
--- a/flang/test/Lower/OpenMP/FIR/target.f90
+++ b/flang/test/Lower/OpenMP/FIR/target.f90
@@ -189,13 +189,14 @@ subroutine omp_target
    integer :: a(1024)
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound({{.*}}) upper_bound({{.*}}) extent({{.*}}) stride({{.*}}) start_idx({{.*}})
    !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]], %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<1024xi32>>, index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %{{.*}}: index):
    !$omp target map(tofrom: a)
       !CHECK: %[[VAL_1:.*]] = arith.constant 10 : i32
       !CHECK: %[[VAL_2:.*]] = arith.constant 1 : i64
       !CHECK: %[[VAL_3:.*]] = arith.constant 1 : i64
       !CHECK: %[[VAL_4:.*]] = arith.subi %[[VAL_2]], %[[VAL_3]] : i64
-      !CHECK: %[[VAL_5:.*]] = fir.coordinate_of %[[VAL_0]], %[[VAL_4]] : (!fir.ref<!fir.array<1024xi32>>, i64) -> !fir.ref<i32>
+      !CHECK: %[[VAL_5:.*]] = fir.coordinate_of %[[ARG_0]], %[[VAL_4]] : (!fir.ref<!fir.array<1024xi32>>, i64) -> !fir.ref<i32>
       !CHECK: fir.store %[[VAL_1]] to %[[VAL_5]] : !fir.ref<i32>
       a(1) = 10
    !CHECK: omp.terminator
@@ -203,6 +204,72 @@ subroutine omp_target
    !CHECK: }
 end subroutine omp_target
 
+!===============================================================================
+! Target implicit capture
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit() {
+subroutine omp_target_implicit
+   !CHECK: %[[VAL_0:.*]] = fir.alloca !fir.array<1024xi32> {bindc_name = "a", uniq_name = "_QFomp_target_implicitEa"}
+   integer :: a(1024)
+   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(implicit, tofrom) capture(ByRef) bounds(%{{.*}}) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]], %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<1024xi32>>, index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %{{.*}}: index):
+   !$omp target
+      !CHECK: %[[VAL_5:.*]] = fir.coordinate_of %[[ARG_0]], %{{.*}} : (!fir.ref<!fir.array<1024xi32>>, i64) -> !fir.ref<i32>
+      a(1) = 10
+   !CHECK: omp.terminator
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target_implicit
+
+!===============================================================================
+! Target implicit capture nested
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit_nested() {
+subroutine omp_target_implicit_nested
+   integer::a, b
+   !CHECK: omp.target   map_entries(%{{.*}} -> %[[ARG0:.*]], %{{.*}} -> %[[ARG1:.*]] : !fir.ref<i32>, !fir.ref<i32>) {
+   !CHECK: ^bb0(%[[ARG0]]: !fir.ref<i32>, %[[ARG1]]: !fir.ref<i32>):
+   !$omp target
+      !CHECK: fir.store %{{.*}} to %[[ARG0]] : !fir.ref<i32>
+      a = 10
+      !$omp parallel
+         !CHECK: fir.store %{{.*}} to %[[ARG1]] : !fir.ref<i32>
+         b = 20
+         !CHECK: omp.terminator
+      !$omp end parallel
+   !CHECK: omp.terminator
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target_implicit_nested
+
+!===============================================================================
+! Target implicit capture with bounds
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit_bounds(%{{.*}}: !fir.ref<i32> {fir.bindc_name = "n"}) {
+subroutine omp_target_implicit_bounds(n)
+   !CHECK: %[[VAL_1:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_2:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_3:.*]] = fir.alloca !fir.array<?x1024xi32>, %[[VAL_1]] {bindc_name = "a", uniq_name = "_QFomp_target_implicit_boundsEa"}
+   integer :: n
+   integer :: a(n, 1024)
+   !CHECK: %[[VAL_4:.*]] = omp.map_info var_ptr(%[[VAL_3]] : !fir.ref<!fir.array<?x1024xi32>>, !fir.array<?x1024xi32>)   map_clauses(implicit, tofrom) capture(ByRef) bounds(%{{.*}}) -> !fir.ref<!fir.array<?x1024xi32>> {name = "a"}
+   !CHECK: %[[VAL_5:.*]] = omp.map_info val(%[[VAL_1]] : index) var_ptr( : , index)  map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> index {name = ""}
+   !CHECK: %[[VAL_6:.*]] = omp.map_info val(%[[VAL_2]] : index) var_ptr( : , index)  map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> index {name = ""}
+   !CHECK: omp.target   map_entries(%[[VAL_4]] -> %[[ARG_1:.*]], %[[VAL_5]] -> %[[ARG_2:.*]], %[[VAL_6]] -> %[[ARG_3:.*]] : !fir.ref<!fir.array<?x1024xi32>>, index, index) {
+   !CHECK: ^bb0(%[[ARG_1]]: !fir.ref<!fir.array<?x1024xi32>>, %[[ARG_2]]: index, %[[ARG_3]]: index):
+   !$omp target
+      !CHECK: %{{.*}} = fir.convert %[[ARG_1]] : (!fir.ref<!fir.array<?x1024xi32>>) -> !fir.ref<!fir.array<?xi32>>
+      !CHECK: %{{.*}} = arith.muli %{{.*}}, %[[ARG_2]] : index
+      a(11,22) = 33
+      !CHECK: omp.terminator
+   !$omp end target
+!CHECK: }
+end subroutine omp_target_implicit_bounds
+
 !===============================================================================
 ! Target `thread_limit` clause
 !===============================================================================
@@ -212,7 +279,8 @@ subroutine omp_target_thread_limit
    integer :: a
    !CHECK: %[[VAL_1:.*]] = arith.constant 64 : i32
    !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}})   map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "a"}
-   !CHECK: omp.target   thread_limit(%[[VAL_1]] : i32) map_entries(%[[MAP]] : !fir.ref<i32>) {
+   !CHECK: omp.target   thread_limit(%[[VAL_1]] : i32) map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !fir.ref<i32>) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<i32>):
    !$omp target map(tofrom: a) thread_limit(64)
       a = 10
    !CHECK: omp.terminator
@@ -274,23 +342,25 @@ subroutine omp_target_parallel_do
    !CHECK: %[[C0:.*]] = arith.constant 0 : index
    !CHECK: %[[SUB:.*]] = arith.subi %[[C1024]], %[[C1]] : index
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C0]] : index) upper_bound(%[[SUB]] : index) extent(%[[C1024]] : index) stride(%[[C1]] : index) start_idx(%[[C1]] : index)
-   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
+   !CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: %[[MAP2:.*]] = omp.map_info var_ptr(%[[VAL_1]] : !fir.ref<i32>, i32)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !fir.ref<i32> {name = "i"}
+   !CHECK: omp.target map_entries(%[[MAP1]] -> %[[VAL_2:.*]], %[[MAP2]] -> %[[VAL_3:.*]], %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<1024xi32>>, !fir.ref<i32>, index) {
+   !CHECK: ^bb0(%[[VAL_2]]: !fir.ref<!fir.array<1024xi32>>, %[[VAL_3]]: !fir.ref<i32>, %{{.*}}: index):
       !CHECK-NEXT: omp.parallel
       !$omp target parallel do map(tofrom: a)
-         !CHECK: %[[VAL_2:.*]] = fir.alloca i32 {adapt.valuebyref, pinned}
-         !CHECK: %[[VAL_3:.*]] = arith.constant 1 : i32
-         !CHECK: %[[VAL_4:.*]] = arith.constant 1024 : i32
+         !CHECK: %[[VAL_4:.*]] = fir.alloca i32 {adapt.valuebyref, pinned}
          !CHECK: %[[VAL_5:.*]] = arith.constant 1 : i32
-         !CHECK: omp.wsloop   for  (%[[VAL_6:.*]]) : i32 = (%[[VAL_3]]) to (%[[VAL_4]]) inclusive step (%[[VAL_5]]) {
-         !CHECK: fir.store %[[VAL_6]] to %[[VAL_2]] : !fir.ref<i32>
-         !CHECK: %[[VAL_7:.*]] = arith.constant 10 : i32
-         !CHECK: %[[VAL_8:.*]] = fir.load %[[VAL_2]] : !fir.ref<i32>
-         !CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_8]] : (i32) -> i64
-         !CHECK: %[[VAL_10:.*]] = arith.constant 1 : i64
-         !CHECK: %[[VAL_11:.*]] = arith.subi %[[VAL_9]], %[[VAL_10]] : i64
-         !CHECK: %[[VAL_12:.*]] = fir.coordinate_of %[[VAL_0]], %[[VAL_11]] : (!fir.ref<!fir.array<1024xi32>>, i64) -> !fir.ref<i32>
-         !CHECK: fir.store %[[VAL_7]] to %[[VAL_12]] : !fir.ref<i32>
+         !CHECK: %[[VAL_6:.*]] = arith.constant 1024 : i32
+         !CHECK: %[[VAL_7:.*]] = arith.constant 1 : i32
+         !CHECK: omp.wsloop   for  (%[[VAL_8:.*]]) : i32 = (%[[VAL_5]]) to (%[[VAL_6]]) inclusive step (%[[VAL_7]]) {
+         !CHECK: fir.store %[[VAL_8]] to %[[VAL_4]] : !fir.ref<i32>
+         !CHECK: %[[VAL_9:.*]] = arith.constant 10 : i32
+         !CHECK: %[[VAL_10:.*]] = fir.load %[[VAL_4]] : !fir.ref<i32>
+         !CHECK: %[[VAL_11:.*]] = fir.convert %[[VAL_10]] : (i32) -> i64
+         !CHECK: %[[VAL_12:.*]] = arith.constant 1 : i64
+         !CHECK: %[[VAL_13:.*]] = arith.subi %[[VAL_11]], %[[VAL_12]] : i64
+         !CHECK: %[[VAL_14:.*]] = fir.coordinate_of %[[VAL_2]], %[[VAL_13]] : (!fir.ref<!fir.array<1024xi32>>, i64) -> !fir.ref<i32>
+         !CHECK: fir.store %[[VAL_9]] to %[[VAL_14]] : !fir.ref<i32>
          do i = 1, 1024
             a(i) = 10
          end do
@@ -301,4 +371,4 @@ subroutine omp_target_parallel_do
    !CHECK: omp.terminator
    !CHECK: }
    !$omp end target parallel do
-end subroutine omp_target_parallel_do
+ end subroutine omp_target_parallel_do
diff --git a/flang/test/Lower/OpenMP/array-bounds.f90 b/flang/test/Lower/OpenMP/array-bounds.f90
index e65f17f74f59097..d0c584bec6044ad 100644
--- a/flang/test/Lower/OpenMP/array-bounds.f90
+++ b/flang/test/Lower/OpenMP/array-bounds.f90
@@ -22,7 +22,7 @@
 !HOST:  %[[C6:.*]] = arith.constant 4 : index
 !HOST:  %[[BOUNDS1:.*]] = omp.bounds   lower_bound(%[[C5]] : index) upper_bound(%[[C6]] : index) stride(%[[C4]] : index) start_idx(%[[C4]] : index)
 !HOST:  %[[MAP1:.*]] = omp.map_info var_ptr(%[[WRITE_DECL]]#1 : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS1]]) -> !fir.ref<!fir.array<10xi32>> {name = "sp_write(2:5)"}
-!HOST:  omp.target   map_entries(%[[MAP0]], %[[MAP1]] : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>) {
+!HOST:  omp.target map_entries(%[[MAP0]] -> %{{.*}}, %[[MAP1]] -> %{{.*}}, {{.*}} -> {{.*}}, {{.*}} -> {{.*}}, {{.*}} -> {{.*}} : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.array<10xi32>>, !fir.ref<i32>, index, index) {
 
 subroutine read_write_section()
     integer :: sp_read(10) = (/1,2,3,4,5,6,7,8,9,10/)
@@ -50,7 +50,7 @@ module assumed_array_routines
 !HOST: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C3]] : index) upper_bound(%[[C4]] : index) stride(%[[C2]]#2 : index) start_idx(%[[C0]] : index) {stride_in_bytes = true}
 !HOST: %[[ADDROF:.*]] = fir.box_addr %[[ARG0_DECL]]#1 : (!fir.box<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
 !HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[ADDROF]] : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!HOST: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
+!HOST: omp.target   map_entries(%[[MAP]] -> %{{.*}}, {{.*}} -> {{.*}} : !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>) {
     subroutine assumed_shape_array(arr_read_write)
             integer, intent(inout) :: arr_read_write(:)
 
@@ -73,7 +73,7 @@ end subroutine assumed_shape_array
 !HOST: %[[C2:.*]] = arith.constant 4 : index
 !HOST: %[[BOUNDS:.*]]  = omp.bounds   lower_bound(%[[C1]] : index) upper_bound(%[[C2]] : index) stride(%[[C0]] : index) start_idx(%[[C0]] : index)
 !HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[ARG0_DECL]]#1 : !fir.ref<!fir.array<?xi32>>, !fir.array<?xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
-!HOST: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<?xi32>>) {
+!HOST: omp.target map_entries(%[[MAP]] -> %{{.*}}, {{.*}} -> {{.*}}, {{.*}} -> {{.*}} : !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>, index) {
         subroutine assumed_size_array(arr_read_write)
             integer, intent(inout) :: arr_read_write(*)
 
diff --git a/flang/test/Lower/OpenMP/location.f90 b/flang/test/Lower/OpenMP/location.f90
index c87bf038e967215..1e01a4828dd9e1b 100644
--- a/flang/test/Lower/OpenMP/location.f90
+++ b/flang/test/Lower/OpenMP/location.f90
@@ -17,7 +17,7 @@ subroutine sub_parallel()
 !CHECK-LABEL: sub_target
 subroutine sub_target()
   print *, x
-!CHECK: omp.target  {
+!CHECK: omp.target {{.*}} {
   !$omp target
     print *, x
 !CHECK:   omp.terminator loc(#[[TAR_LOC:.*]])
diff --git a/flang/test/Lower/OpenMP/target.f90 b/flang/test/Lower/OpenMP/target.f90
index a94859a3a1aa7d8..498e8a6b6b654bf 100644
--- a/flang/test/Lower/OpenMP/target.f90
+++ b/flang/test/Lower/OpenMP/target.f90
@@ -187,23 +187,112 @@ end subroutine omp_target_data_mt
 
 !CHECK-LABEL: func.func @_QPomp_target() {
 subroutine omp_target
-   !CHECK: %[[VAL_0:.*]] = fir.alloca !fir.array<1024xi32> {bindc_name = "a", uniq_name = "_QFomp_targetEa"}
-   !CHECK: %[[VAL_0_DECL:.*]]:2 = hlfir.declare %[[VAL_0]](%{{.*}}) {uniq_name = "_QFomp_targetEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+   !CHECK: %[[VAL_1:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFomp_targetEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
    integer :: a(1024)
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound({{.*}}) upper_bound({{.*}}) extent({{.*}}) stride({{.*}}) start_idx({{.*}})
-   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0_DECL]]#1 : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
+   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_1]]#1 : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]], %{{.*}} -> %[[ARG_1:.*]] : !fir.ref<!fir.array<1024xi32>>, index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %[[ARG_1]]: index):
    !$omp target map(tofrom: a)
-      !CHECK: %[[C10:.*]] = arith.constant 10 : i32
-      !CHECK: %[[C1:.*]] = arith.constant 1 : index
-      !CHECK: %[[A_1:.*]] = hlfir.designate %[[VAL_0_DECL]]#0 (%[[C1]])  : (!fir.ref<!fir.array<1024xi32>>, index) -> !fir.ref<i32>
-      !CHECK: hlfir.assign %[[C10]] to %[[A_1]] : i32, !fir.ref<i32>
+      !CHECK: %[[VAL_2:.*]] = fir.shape %[[ARG_1]] : (index) -> !fir.shape<1>
+      !CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[ARG_0]](%[[VAL_2]]) {uniq_name = "_QFomp_targetEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+      !CHECK: %[[VAL_4:.*]] = arith.constant 10 : i32
+      !CHECK: %[[VAL_5:.*]] = arith.constant 1 : index
+      !CHECK: %[[VAL_6:.*]] = hlfir.designate %[[VAL_3]]#0 (%[[VAL_5]])  : (!fir.ref<!fir.array<1024xi32>>, index) -> !fir.ref<i32>
+      !CHECK: hlfir.assign %[[VAL_4]] to %[[VAL_6]] : i32, !fir.ref<i32>
       a(1) = 10
-   !CHECK: omp.terminator
+      !CHECK: omp.terminator
    !$omp end target
    !CHECK: }
 end subroutine omp_target
 
+!===============================================================================
+! Target implicit capture
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit() {
+subroutine omp_target_implicit
+   !CHECK: %[[VAL_0:.*]] = arith.constant 1024 : index
+   !CHECK: %[[VAL_1:.*]] = fir.alloca !fir.array<1024xi32> {bindc_name = "a", uniq_name = "_QFomp_target_implicitEa"}
+   !CHECK: %[[VAL_2:.*]] = fir.shape %[[VAL_0]] : (index) -> !fir.shape<1>
+   !CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_1]](%[[VAL_2]]) {uniq_name = "_QFomp_target_implicitEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+   integer :: a(1024)
+   !CHECK: %[[VAL_4:.*]] = omp.map_info var_ptr(%[[VAL_3]]#1 : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(implicit, tofrom) capture(ByRef) bounds(%{{.*}}) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: %[[VAL_5:.*]] = omp.map_info val(%[[VAL_0]] : index) var_ptr( : , index)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> index {name = ""}
+   !CHECK: omp.target   map_entries(%[[VAL_4]] -> %[[VAL_6:.*]], %[[VAL_5]] -> %[[VAL_7:.*]] : !fir.ref<!fir.array<1024xi32>>, index) {
+   !CHECK: ^bb0(%[[VAL_6]]: !fir.ref<!fir.array<1024xi32>>, %[[VAL_7]]: index):
+   !$omp target
+      !CHECK: %[[VAL_8:.*]] = fir.shape %[[VAL_7]] : (index) -> !fir.shape<1>
+      !CHECK: %[[VAL_9:.*]]:2 = hlfir.declare %[[VAL_6]](%[[VAL_8]]) {uniq_name = "_QFomp_target_implicitEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+      !CHECK: %[[VAL_10:.*]] = arith.constant 10 : i32
+      !CHECK: %[[VAL_11:.*]] = arith.constant 1 : index
+      !CHECK: %[[VAL_12:.*]] = hlfir.designate %[[VAL_9]]#0 (%[[VAL_11]])  : (!fir.ref<!fir.array<1024xi32>>, index) -> !fir.ref<i32>
+      !CHECK: hlfir.assign %[[VAL_10]] to %[[VAL_12]] : i32, !fir.ref<i32>
+      a(1) = 10
+      !CHECK: omp.terminator
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target_implicit
+
+!===============================================================================
+! Target implicit capture nested
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit_nested() {
+subroutine omp_target_implicit_nested
+   integer::a, b
+   !CHECK: omp.target   map_entries(%{{.*}} -> %[[ARG0:.*]], %{{.*}} -> %[[ARG1:.*]] : !fir.ref<i32>, !fir.ref<i32>) {
+   !CHECK: ^bb0(%[[ARG0]]: !fir.ref<i32>, %[[ARG1]]: !fir.ref<i32>):
+   !$omp target
+      !CHECK: %[[VAL_8:.*]]:2 = hlfir.declare %[[ARG0]] {uniq_name = "_QFomp_target_implicit_nestedEa"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+      !CHECK: %[[VAL_9:.*]]:2 = hlfir.declare %[[ARG1]] {uniq_name = "_QFomp_target_implicit_nestedEb"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+      !CHECK: %[[VAL_10:.*]] = arith.constant 10 : i32
+      !CHECK: hlfir.assign %[[VAL_10]] to %[[VAL_8]]#0 : i32, !fir.ref<i32>
+      a = 10
+      !CHECK: omp.parallel
+      !$omp parallel
+         !CHECK: %[[VAL_11:.*]] = arith.constant 20 : i32
+         !CHECK: hlfir.assign %[[VAL_11]] to %[[VAL_9]]#0 : i32, !fir.ref<i32>
+         b = 20
+         !CHECK: omp.terminator
+      !$omp end parallel
+      !CHECK: omp.terminator
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target_implicit_nested
+
+!===============================================================================
+! Target implicit capture with bounds
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit_bounds(%{{.*}}: !fir.ref<i32> {fir.bindc_name = "n"}) {
+subroutine omp_target_implicit_bounds(n)
+   !CHECK: %[[VAL_1:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_2:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_3:.*]] = fir.alloca !fir.array<?x1024xi32>, %[[VAL_1]] {bindc_name = "a", uniq_name = "_QFomp_target_implicit_boundsEa"}
+   !CHECK: %[[VAL_4:.*]] = fir.shape %[[VAL_1]], %[[VAL_2]] : (index, index) -> !fir.shape<2>
+   !CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %[[VAL_3]](%[[VAL_4]]) {uniq_name = "_QFomp_target_implicit_boundsEa"} : (!fir.ref<!fir.array<?x1024xi32>>, !fir.shape<2>) -> (!fir.box<!fir.array<?x1024xi32>>, !fir.ref<!fir.array<?x1024xi32>>)
+   integer :: n
+   integer :: a(n, 1024)
+   !CHECK: %[[VAL_6:.*]] = omp.map_info var_ptr(%[[VAL_5]]#1 : !fir.ref<!fir.array<?x1024xi32>>, !fir.array<?x1024xi32>)   map_clauses(implicit, tofrom) capture(ByRef) bounds(%{{.*}}) -> !fir.ref<!fir.array<?x1024xi32>> {name = "a"}
+   !CHECK: %[[VAL_7:.*]] = omp.map_info val(%[[VAL_1]] : index) var_ptr( : , index)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> index {name = ""}
+   !CHECK: %[[VAL_8:.*]] = omp.map_info val(%[[VAL_2]] : index) var_ptr( : , index)   map_clauses(implicit, exit_release_or_enter_alloc) capture(ByCopy) -> index {name = ""}
+   !CHECK: omp.target   map_entries(%[[VAL_6]] -> %[[ARG_1:.*]], %[[VAL_7]] -> %[[ARG_2:.*]], %[[VAL_8]] -> %[[ARG_3:.*]] : !fir.ref<!fir.array<?x1024xi32>>, index, index) {
+   !CHECK: ^bb0(%[[ARG_1]]: !fir.ref<!fir.array<?x1024xi32>>, %[[ARG_2]]: index, %[[ARG_3]]: index):
+   !$omp target
+      !CHECK: %[[VAL_9:.*]] = fir.shape %[[ARG_2]], %[[ARG_3]] : (index, index) -> !fir.shape<2>
+      !CHECK: %[[VAL_10:.*]]:2 = hlfir.declare %[[ARG_1]](%[[VAL_9]]) {uniq_name = "_QFomp_target_implicit_boundsEa"} : (!fir.ref<!fir.array<?x1024xi32>>, !fir.shape<2>) -> (!fir.box<!fir.array<?x1024xi32>>, !fir.ref<!fir.array<?x1024xi32>>)
+      !CHECK: %[[VAL_11:.*]] = arith.constant 33 : i32
+      !CHECK: %[[VAL_12:.*]] = arith.constant 11 : index
+      !CHECK: %[[VAL_13:.*]] = arith.constant 22 : index
+      !CHECK: %[[VAL_14:.*]] = hlfir.designate %[[VAL_10]]#0 (%[[VAL_12]], %[[VAL_13]])  : (!fir.box<!fir.array<?x1024xi32>>, index, index) -> !fir.ref<i32>
+      !CHECK: hlfir.assign %[[VAL_11]] to %[[VAL_14]] : i32, !fir.ref<i32>
+      a(11, 22) = 33
+      !CHECK: omp.terminator
+   !$omp end target
+!CHECK: }
+end subroutine omp_target_implicit_bounds
+
 !===============================================================================
 ! Target `thread_limit` clause
 !===============================================================================
@@ -213,7 +302,8 @@ subroutine omp_target_thread_limit
    integer :: a
    !CHECK: %[[VAL_1:.*]] = arith.constant 64 : i32
    !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}})   map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "a"}
-   !CHECK: omp.target   thread_limit(%[[VAL_1]] : i32) map_entries(%[[MAP]] : !fir.ref<i32>) {
+   !CHECK: omp.target   thread_limit(%[[VAL_1]] : i32) map_entries(%[[MAP]] -> %{{.*}} : !fir.ref<i32>) {
+   !CHECK: ^bb0(%{{.*}}: !fir.ref<i32>):
    !$omp target map(tofrom: a) thread_limit(64)
       a = 10
    !CHECK: omp.terminator
@@ -283,8 +373,10 @@ subroutine omp_target_parallel_do
    !CHECK: %[[SUB:.*]] = arith.subi %[[C1024]], %[[C1]] : index
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C0]] : index) upper_bound(%[[SUB]] : index) extent(%[[C1024]] : index) stride(%[[C1]] : index) start_idx(%[[C1]] : index)
    !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0_DECL]]#1 : !fir.ref<!fir.array<1024xi32>>, !fir.array<1024xi32>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
-      !CHECK-NEXT: omp.parallel
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]], %{{.*}} -> %{{.*}}, %{{.*}} -> %{{.*}} : !fir.ref<!fir.array<1024xi32>>, !fir.ref<i32>, index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %{{.*}}: !fir.ref<i32>, %{{.*}}: index):
+      !CHECK: %[[VAL_0_DECL:.*]]:2 = hlfir.declare %[[ARG_0]](%{{.*}}) {uniq_name = "_QFomp_target_parallel_doEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+      !CHECK: omp.parallel
       !$omp target parallel do map(tofrom: a)
          !CHECK: %[[I_PVT_ALLOCA:.*]] = fir.alloca i32 {adapt.valuebyref, pinned}
          !CHECK: %[[I_PVT_DECL:.*]]:2 = hlfir.declare %[[I_PVT_ALLOCA]] {uniq_name = "_QFomp_target_parallel_doEi"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
@@ -305,4 +397,4 @@ subroutine omp_target_parallel_do
    !CHECK: omp.terminator
    !CHECK: }
    !$omp end target parallel do
-end subroutine omp_target_parallel_do
+end subroutine omp_target_parallel_do
\ No newline at end of file
diff --git a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
index 72121ad4f9e47a5..ee9386eefc8177c 100644
--- a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
+++ b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
@@ -1144,15 +1144,15 @@ def DataBoundsOp : OpenMP_Op<"bounds",
 }
 
 def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
-  let arguments = (ins OpenMP_PointerLikeType:$var_ptr,
-                       TypeAttr:$var_type,
+  let arguments = (ins Optional<AnyType>:$val,
+                       Optional<OpenMP_PointerLikeType>:$var_ptr,
+                       OptionalAttr<TypeAttr>:$var_type,
                        Optional<OpenMP_PointerLikeType>:$var_ptr_ptr,
                        Variadic<DataBoundsType>:$bounds, /* rank-0 to rank-{n-1} */
                        OptionalAttr<UI64Attr>:$map_type,
                        OptionalAttr<VariableCaptureKindAttr>:$map_capture_type,
-                       DefaultValuedAttr<BoolAttr, "false">:$implicit,
                        OptionalAttr<StrAttr>:$name);
-  let results = (outs OpenMP_PointerLikeType:$omp_ptr);
+  let results = (outs AnyType:$omp_ptr);
 
   let description = [{
     The MapInfoOp captures information relating to individual OpenMP map clauses
@@ -1178,11 +1178,12 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
     ```
     =>
     ```mlir
-    omp.map_info var_ptr(%index_ssa) map_type(to) map_capture_type(ByRef) implicit(false)
+    omp.map_info var_ptr(%index_ssa) map_type(to) map_capture_type(ByRef)
       name(index)
     ```
 
     Description of arguments:
+    - `val`: The value to copy.
     - `var_ptr`: The address of variable to copy.
     - `var_type`: The type of the variable to copy.
     - `var_ptr_ptr`: Used when the variable copied is a member of a class, structure
@@ -1191,9 +1192,6 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
       objects (e.g. derived types or classes), indicates the bounds to be copied
       of the variable. When it's an array slice it is in rank order where rank 0
       is the inner-most dimension.
-    - `implicit`: indicates where the map item has been specified explicitly in a
-      map clause or captured implicitly by being used in a target region with no
-      map or other data mapping construct.
     - 'map_clauses': OpenMP map type for this map capture, for example: from, to and
       always. It's a bitfield composed of the OpenMP runtime flags stored in
       OpenMPOffloadMappingFlags.
@@ -1203,9 +1201,10 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
   }];
 
   let assemblyFormat = [{
-    `var_ptr` `(` $var_ptr `:` type($var_ptr) `,` $var_type`)`
     oilist(
-        `var_ptr_ptr` `(` $var_ptr_ptr `:` type($var_ptr_ptr) `)`
+        `val` `(` $val `:` type($val) `)`
+      | `var_ptr` `(` $var_ptr `:` type($var_ptr) `,` $var_type`)`
+      | `var_ptr_ptr` `(` $var_ptr_ptr `:` type($var_ptr_ptr) `)`
       | `map_clauses` `(` custom<MapClause>($map_type) `)`
       | `capture` `(` custom<CaptureType>($map_capture_type) `)`
       | `bounds` `(` $bounds `)`
@@ -1265,7 +1264,7 @@ def Target_DataOp: OpenMP_Op<"target_data", [AttrSizedOperandSegments]>{
                        Optional<AnyInteger>:$device,
                        Variadic<OpenMP_PointerLikeType>:$use_device_ptr,
                        Variadic<OpenMP_PointerLikeType>:$use_device_addr,
-                       Variadic<OpenMP_PointerLikeType>:$map_operands);
+                       Variadic<AnyType>:$map_operands);
 
   let regions = (region AnyRegion:$region);
 
@@ -1314,7 +1313,7 @@ def Target_EnterDataOp: OpenMP_Op<"target_enter_data",
   let arguments = (ins Optional<I1>:$if_expr,
                        Optional<AnyInteger>:$device,
                        UnitAttr:$nowait,
-                       Variadic<OpenMP_PointerLikeType>:$map_operands);
+                       Variadic<AnyType>:$map_operands);
 
   let assemblyFormat = [{
     oilist(`if` `(` $if_expr `:` type($if_expr) `)`
@@ -1360,7 +1359,7 @@ def Target_ExitDataOp: OpenMP_Op<"target_exit_data",
   let arguments = (ins Optional<I1>:$if_expr,
                        Optional<AnyInteger>:$device,
                        UnitAttr:$nowait,
-                       Variadic<OpenMP_PointerLikeType>:$map_operands);
+                       Variadic<AnyType>:$map_operands);
 
   let assemblyFormat = [{
     oilist(`if` `(` $if_expr `:` type($if_expr) `)`
@@ -1377,7 +1376,7 @@ def Target_ExitDataOp: OpenMP_Op<"target_exit_data",
 // 2.14.5 target construct
 //===----------------------------------------------------------------------===//
 
-def TargetOp : OpenMP_Op<"target",[OutlineableOpenMPOpInterface, AttrSizedOperandSegments]> {
+def TargetOp : OpenMP_Op<"target",[IsolatedFromAbove, OutlineableOpenMPOpInterface, AttrSizedOperandSegments]> {
   let summary = "target construct";
   let description = [{
     The target construct includes a region of code which is to be executed
@@ -1395,6 +1394,10 @@ def TargetOp : OpenMP_Op<"target",[OutlineableOpenMPOpInterface, AttrSizedOperan
     The optional $nowait elliminates the implicit barrier so the parent task can make progress
     even if the target task is not yet completed.
 
+    The optional $block_args stores the mapping between the block_arguments and their
+    corresponding value outside the region. It only stores those values which haven't already
+    been mapped in $map_operands.
+
     TODO:  is_device_ptr, depend, defaultmap, in_reduction
 
   }];
@@ -1403,7 +1406,7 @@ def TargetOp : OpenMP_Op<"target",[OutlineableOpenMPOpInterface, AttrSizedOperan
                        Optional<AnyInteger>:$device,
                        Optional<AnyInteger>:$thread_limit,
                        UnitAttr:$nowait,
-                       Variadic<OpenMP_PointerLikeType>:$map_operands);
+                       Variadic<AnyType>:$map_operands);
 
   let regions = (region AnyRegion:$region);
 
@@ -1412,7 +1415,7 @@ def TargetOp : OpenMP_Op<"target",[OutlineableOpenMPOpInterface, AttrSizedOperan
     | `device` `(` $device `:` type($device) `)`
     | `thread_limit` `(` $thread_limit `:` type($thread_limit) `)`
     | `nowait` $nowait
-    | `map_entries` `(` $map_operands `:` type($map_operands) `)`
+    | `map_entries` `(` custom<MapEntries>($map_operands, type($map_operands)) `)`
     ) $region attr-dict
   }];
 
diff --git a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
index 2bf9355ed62676b..a389a6630d47b94 100644
--- a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
+++ b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
@@ -693,6 +693,9 @@ static ParseResult parseMapClause(OpAsmParser &parser, IntegerAttr &mapType) {
     if (mapTypeMod == "always")
       mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_ALWAYS;
 
+    if (mapTypeMod == "implicit")
+      mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT;
+
     if (mapTypeMod == "close")
       mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_CLOSE;
 
@@ -740,6 +743,9 @@ static void printMapClause(OpAsmPrinter &p, Operation *op,
   if (mapTypeToBitFlag(mapTypeBits,
                        llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_ALWAYS))
     mapTypeStrs.push_back("always");
+  if (mapTypeToBitFlag(mapTypeBits,
+                       llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT))
+    mapTypeStrs.push_back("implicit");
   if (mapTypeToBitFlag(mapTypeBits,
                        llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_CLOSE))
     mapTypeStrs.push_back("close");
@@ -780,6 +786,64 @@ static void printMapClause(OpAsmPrinter &p, Operation *op,
   }
 }
 
+static ParseResult
+parseMapEntries(OpAsmParser &parser,
+                SmallVectorImpl<OpAsmParser::UnresolvedOperand> &mapOperands,
+                SmallVectorImpl<Type> &mapOperandTypes) {
+  OpAsmParser::UnresolvedOperand arg;
+  OpAsmParser::UnresolvedOperand blockArg;
+  Type argType;
+  auto parseEntries = [&]() -> ParseResult {
+    if (parser.parseOperand(arg) || parser.parseArrow() ||
+        parser.parseOperand(blockArg))
+      return failure();
+    mapOperands.push_back(arg);
+    return success();
+  };
+
+  auto parseTypes = [&]() -> ParseResult {
+    if (parser.parseType(argType))
+      return failure();
+    mapOperandTypes.push_back(argType);
+    return success();
+  };
+
+  if (parser.parseCommaSeparatedList(parseEntries))
+    return failure();
+
+  if (parser.parseColon())
+    return failure();
+
+  if (parser.parseCommaSeparatedList(parseTypes))
+    return failure();
+
+  return success();
+}
+
+static void printMapEntries(OpAsmPrinter &p, Operation *op,
+                            OperandRange mapOperands,
+                            TypeRange mapOperandTypes) {
+  auto &region = op->getRegion(0);
+  unsigned argIndex = 0;
+
+  for (const auto &mapOp : mapOperands) {
+    const auto &blockArg = region.front().getArgument(argIndex);
+    p << mapOp << " -> " << blockArg;
+    argIndex++;
+    if (argIndex < mapOperands.size())
+      p << ", ";
+  }
+  p << " : ";
+
+  argIndex = 0;
+  for (const auto &mapType : mapOperandTypes) {
+    p << mapType;
+    argIndex++;
+    if (argIndex < mapOperands.size())
+      p << ", ";
+  }
+}
+
 static void printCaptureType(OpAsmPrinter &p, Operation *op,
                              VariableCaptureKindAttr mapCaptureType) {
   std::string typeCapStr;
@@ -826,6 +890,12 @@ static LogicalResult verifyMapClause(Operation *op, OperandRange mapOperands) {
     if (auto MapInfoOp =
             mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp())) {
 
+      if (MapInfoOp.getVal() && MapInfoOp.getVarPtr())
+        emitError(op->getLoc(), "only one of val or var_ptr must be used");
+
+      if (!MapInfoOp.getVal() && !MapInfoOp.getVarPtr())
+        emitError(op->getLoc(), "missing val or var_ptr");
+
       if (!MapInfoOp.getMapType().has_value())
         emitError(op->getLoc(), "missing map type for map operand");
 



More information about the flang-commits mailing list