[flang-commits] [flang] [OpenMP][MLIR] Add "IsolatedFromAbove" and "OutlineableOpenMPOpInterface" trait to omp.target (PR #67318)

Akash Banerjee via flang-commits flang-commits at lists.llvm.org
Fri Oct 20 09:04:19 PDT 2023


https://github.com/TIFitis updated https://github.com/llvm/llvm-project/pull/67318

>From 52d311370122f878939d800e8bb8617abd304319 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 1/2] [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                    | 227 +++++++++++++++---
 .../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        | 104 ++++++--
 flang/test/Lower/OpenMP/array-bounds.f90      |   6 +-
 flang/test/Lower/OpenMP/location.f90          |   2 +-
 flang/test/Lower/OpenMP/target.f90            | 110 +++++++++
 mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td |  33 +--
 mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp  |  76 ++++++
 10 files changed, 540 insertions(+), 149 deletions(-)
 create mode 100644 flang/test/Lower/OpenMP/target.f90

diff --git a/flang/lib/Lower/OpenMP.cpp b/flang/lib/Lower/OpenMP.cpp
index 5f5e968eaaa6414..07236cbd0e31efa 100644
--- a/flang/lib/Lower/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP.cpp
@@ -534,11 +534,21 @@ 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,
@@ -1665,29 +1675,25 @@ 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;
+
   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.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;
+
+  mlir::omp::MapInfoOp op = builder.create<mlir::omp::MapInfoOp>(
+      loc, retTy, val, varPtr, varPtrPtr, bounds,
+      builder.getIntegerAttr(builder.getIntegerType(64, false), mapType),
+      builder.getAttr<mlir::omp::VariableCaptureKindAttr>(mapCaptureType),
+      builder.getStringAttr(name.str()));
   return op;
 }
 
@@ -1695,7 +1701,11 @@ 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,
@@ -1755,13 +1765,20 @@ 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));
         }
       });
 }
@@ -2142,7 +2159,7 @@ 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,
@@ -2356,8 +2373,8 @@ 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;
 }
 
@@ -2400,6 +2417,66 @@ 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);
+    mlir::Value val = fir::getBase(arg);
+    extVal.match(
+        [&](const fir::BoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::BoxValue(val, extractBoundArgs(v.getLBounds().size()),
+                            v.getExplicitParameters(), v.getExplicitExtents()));
+        },
+        [&](const fir::MutableBoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::MutableBoxValue(val, extractBoundArgs(v.getLBounds().size()),
+                                   v.getMutableProperties()));
+        },
+        [&](const fir::ArrayBoxValue &v) {
+          converter.bindSymbol(
+              *sym,
+              fir::ArrayBoxValue(val, extractBoundArgs(v.getExtents().size()),
+                                 extractBoundArgs(v.getLBounds().size()),
+                                 v.getSourceBox()));
+        },
+        [&](const fir::UnboxedValue &v) { converter.bindSymbol(*sym, val); },
+        [&](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,
@@ -2411,6 +2488,9 @@ 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,
@@ -2420,7 +2500,7 @@ 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,
@@ -2433,10 +2513,95 @@ 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 = genBoundsOpsFromBox<mlir::omp::DataBoundsOp,
+                                       mlir::omp::DataBoundsType>(
+              converter.getFirOpBuilder(), converter.getCurrentLocation(),
+              converter, dataExv, baseAddr);
+        if (fir::unwrapRefType(baseAddr.getType()).isa<fir::SequenceType>())
+          bounds = genBaseBoundsOps<mlir::omp::DataBoundsOp,
+                                    mlir::omp::DataBoundsType>(
+              converter.getFirOpBuilder(), converter.getCurrentLocation(),
+              converter, dataExv, baseAddr);
+
+        mlir::Value mapOp = createMapInfoOp(
+            converter.getFirOpBuilder(), baseOp.getLoc(), baseOp, name, bounds,
+            static_cast<
+                std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
+                llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_LITERAL |
+                llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT),
+            mlir::omp::VariableCaptureKind::ByCopy, 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_LITERAL |
+              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 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/Fir/convert-to-llvm-openmp-and-fir.fir b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
index ba1a15bf773d2ef..1123e07e6c690aa 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>>)   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<array<512 x i32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !llvm.ptr<array<512 x i32>> {name = "a"}
-// CHECK:           omp.target   thread_limit(%[[VAL_2]] : i32) map_entries(%[[MAP]] : !llvm.ptr<array<512 x i32>>) {
+// CHECK:           omp.target   thread_limit(%[[VAL_2]] : i32) map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !llvm.ptr<array<512 x i32>>) {
+// CHECK:           ^bb0(%[[ARG_0]]: !llvm.ptr<array<512 x i32>>):
 // 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<array<512 x i32>>, i64) -> !llvm.ptr<i32>
+// CHECK:             %[[VAL_7:.*]] = llvm.getelementptr %[[ARG_0]][0, %[[VAL_6]]] : (!llvm.ptr<array<512 x i32>>, i64) -> !llvm.ptr<i32>
 // CHECK:             llvm.store %[[VAL_3]], %[[VAL_7]] : !llvm.ptr<i32>
 // 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 697ff44176b3f10..f8346b49ae8a120 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>>)   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>>)   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>>)   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>>)   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>>)   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>>)   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>)   map_clauses(literal, 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>) block_args(%{{.*}}, %{{.*}} : 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>>) 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>>)   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>>)   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>)   map_clauses(literal, 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>>)   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>>)   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>>)   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>)   map_clauses(literal, 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>) block_args(%{{.*}} : 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>>)   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 {{.*}} {
   !$omp target
     print *, x
 !CHECK:   omp.terminator loc(#[[TAR_LOC:.*]])
diff --git a/flang/test/Lower/OpenMP/FIR/target.f90 b/flang/test/Lower/OpenMP/FIR/target.f90
index 9b1fb5c15ac1d2d..85634e185e37de8 100644
--- a/flang/test/Lower/OpenMP/FIR/target.f90
+++ b/flang/test/Lower/OpenMP/FIR/target.f90
@@ -189,13 +189,14 @@ subroutine omp_target
    integer :: a(1024)
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound({{.*}}) upper_bound({{.*}}) extent({{.*}}) stride({{.*}}) start_idx({{.*}})
    !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !fir.ref<!fir.array<1024xi32>>) block_args(%{{.*}} : index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %{{.*}}: index):
    !$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 +204,70 @@ 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>>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) bounds(%{{.*}}) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !fir.ref<!fir.array<1024xi32>>) block_args(%{{.*}} : 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>>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) bounds(%{{.*}}) -> !fir.ref<!fir.array<?x1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[VAL_4]] -> %[[ARG_1:.*]] : !fir.ref<!fir.array<?x1024xi32>>) block_args(%[[VAL_5:.*]], %[[VAL_6:.*]] : 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 +277,8 @@ 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>):
    !$omp target map(tofrom: a) thread_limit(64)
       a = 10
    !CHECK: omp.terminator
@@ -274,23 +340,25 @@ subroutine omp_target_parallel_do
    !CHECK: %[[C0:.*]] = arith.constant 0 : index
    !CHECK: %[[SUB:.*]] = arith.subi %[[C1024]], %[[C1]] : index
    !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound(%[[C0]] : index) upper_bound(%[[SUB]] : index) extent(%[[C1024]] : index) stride(%[[C1]] : index) start_idx(%[[C1]] : index)
-   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
-   !CHECK: omp.target   map_entries(%[[MAP]] : !fir.ref<!fir.array<1024xi32>>) {
+   !CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[VAL_0]] : !fir.ref<!fir.array<1024xi32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: %[[MAP2:.*]] = omp.map_info var_ptr(%[[VAL_1]] : !fir.ref<i32>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !fir.ref<i32> {name = "i"}
+   !CHECK: omp.target   map_entries(%[[MAP1]] -> %[[VAL_2:.*]], %[[MAP2]] -> %[[VAL_3:.*]] : !fir.ref<!fir.array<1024xi32>>, !fir.ref<i32>) block_args(%{{.*}} : index) {
+   !CHECK: ^bb0(%[[VAL_2]]: !fir.ref<!fir.array<1024xi32>>, %[[VAL_3]]: !fir.ref<i32>, %{{.*}}: index):
       !CHECK-NEXT: omp.parallel
       !$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 +369,4 @@ 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 831c7021d6b2ed9..45db381b5672745 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>>)   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>) block_args(%{{.*}}, %{{.*}} : 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>>)   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>>)   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>) block_args(%{{.*}} : 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 {{.*}} {
   !$omp target
     print *, x
 !CHECK:   omp.terminator loc(#[[TAR_LOC:.*]])
diff --git a/flang/test/Lower/OpenMP/target.f90 b/flang/test/Lower/OpenMP/target.f90
new file mode 100644
index 000000000000000..0e68bf12f05688d
--- /dev/null
+++ b/flang/test/Lower/OpenMP/target.f90
@@ -0,0 +1,110 @@
+!RUN: %flang_fc1 -emit-hlfir -fopenmp %s -o - | FileCheck %s
+
+!===============================================================================
+! Target with region
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target() {
+subroutine omp_target
+   !CHECK: %[[VAL_1:.*]]:2 = hlfir.declare %{{.*}}(%{{.*}}) {uniq_name = "_QFomp_targetEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+   integer :: a(1024)
+   !CHECK: %[[BOUNDS:.*]] = omp.bounds   lower_bound({{.*}}) upper_bound({{.*}}) extent({{.*}}) stride({{.*}}) start_idx({{.*}})
+   !CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[VAL_1]]#1 : !fir.ref<!fir.array<1024xi32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[MAP]] -> %[[ARG_0:.*]] : !fir.ref<!fir.array<1024xi32>>) block_args(%{{.*}} : index) {
+   !CHECK: ^bb0(%[[ARG_0]]: !fir.ref<!fir.array<1024xi32>>, %[[ARG_1:.*]]: index):
+   !$omp target map(tofrom: a)
+   !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
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target
+
+!===============================================================================
+! Target implicit capture
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit() {
+subroutine omp_target_implicit
+   !CHECK: %[[VAL_0:.*]] = arith.constant 1024 : index
+   !CHECK: %[[VAL_1:.*]] = fir.alloca !fir.array<1024xi32> {bindc_name = "a", uniq_name = "_QFomp_target_implicitEa"}
+   !CHECK: %[[VAL_2:.*]] = fir.shape %[[VAL_0]] : (index) -> !fir.shape<1>
+   !CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_1]](%[[VAL_2]]) {uniq_name = "_QFomp_target_implicitEa"} : (!fir.ref<!fir.array<1024xi32>>, !fir.shape<1>) -> (!fir.ref<!fir.array<1024xi32>>, !fir.ref<!fir.array<1024xi32>>)
+   integer :: a(1024)
+   !CHECK: %[[VAL_4:.*]] = omp.map_info var_ptr(%[[VAL_3]]#1 : !fir.ref<!fir.array<1024xi32>>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) bounds(%{{.*}}) -> !fir.ref<!fir.array<1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[VAL_4]] -> %[[VAL_6:.*]] : !fir.ref<!fir.array<1024xi32>>) block_args(%[[VAL_5:.*]] : index) {
+   !CHECK: ^bb0(%[[VAL_6]]: !fir.ref<!fir.array<1024xi32>>, %[[VAL_7:.*]]: index):
+   !$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
+   !$omp end target
+   !CHECK: }
+end subroutine omp_target_implicit_nested
+
+!===============================================================================
+! Target implicit capture with bounds
+!===============================================================================
+
+!CHECK-LABEL: func.func @_QPomp_target_implicit_bounds(%{{.*}}: !fir.ref<i32> {fir.bindc_name = "n"}) {
+subroutine omp_target_implicit_bounds(n)
+   !CHECK: %[[VAL_1:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_2:.*]] = arith.select %{{.*}}, %{{.*}}, %{{.*}} : index
+   !CHECK: %[[VAL_3:.*]] = fir.alloca !fir.array<?x1024xi32>, %[[VAL_1]] {bindc_name = "a", uniq_name = "_QFomp_target_implicit_boundsEa"}
+   !CHECK: %[[VAL_4:.*]] = fir.shape %[[VAL_1]], %[[VAL_2]] : (index, index) -> !fir.shape<2>
+   !CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %[[VAL_3]](%[[VAL_4]]) {uniq_name = "_QFomp_target_implicit_boundsEa"} : (!fir.ref<!fir.array<?x1024xi32>>, !fir.shape<2>) -> (!fir.box<!fir.array<?x1024xi32>>, !fir.ref<!fir.array<?x1024xi32>>)
+   integer :: n
+   integer :: a(n, 1024)
+   !CHECK: %[[VAL_6:.*]] = omp.map_info var_ptr(%[[VAL_5]]#1 : !fir.ref<!fir.array<?x1024xi32>>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) bounds(%{{.*}}) -> !fir.ref<!fir.array<?x1024xi32>> {name = "a"}
+   !CHECK: omp.target   map_entries(%[[VAL_6]] -> %[[ARG_1:.*]] : !fir.ref<!fir.array<?x1024xi32>>) block_args(%[[VAL_7:.*]], %[[VAL_8:.*]] : index, index) {
+   !CHECK: ^bb0(%[[ARG_1]]: !fir.ref<!fir.array<?x1024xi32>>, %[[ARG_2:.*]]: index, %[[ARG_3:.*]]: index):
+   !$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
diff --git a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
index 5da05476a683725..89c363b192f40c8 100644
--- a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
+++ b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
@@ -1144,14 +1144,14 @@ def DataBoundsOp : OpenMP_Op<"bounds",
 }
 
 def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
-  let arguments = (ins OpenMP_PointerLikeType:$var_ptr,
+  let arguments = (ins Optional<AnyType>:$val,
+                       Optional<OpenMP_PointerLikeType>:$var_ptr,
                        Optional<OpenMP_PointerLikeType>:$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
@@ -1177,11 +1177,12 @@ 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_ptr_ptr`: Used when the variable copied is a member of a class, structure
       or derived type and refers to the originating struct.
@@ -1189,9 +1190,6 @@ 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.
@@ -1201,9 +1199,10 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
   }];
 
   let assemblyFormat = [{
-    `var_ptr` `(` $var_ptr `:` type($var_ptr) `)`
     oilist(
-        `var_ptr_ptr` `(` $var_ptr_ptr `:` type($var_ptr_ptr) `)`
+        `val` `(` $val `:` type($val) `)`
+      | `var_ptr` `(` $var_ptr `:` type($var_ptr) `)`
+      | `var_ptr_ptr` `(` $var_ptr_ptr `:` type($var_ptr_ptr) `)`
       | `map_clauses` `(` custom<MapClause>($map_type) `)`
       | `capture` `(` custom<CaptureType>($map_capture_type) `)`
       | `bounds` `(` $bounds `)`
@@ -1263,7 +1262,7 @@ def Target_DataOp: OpenMP_Op<"target_data", [AttrSizedOperandSegments]>{
                        Optional<AnyInteger>:$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);
 
@@ -1312,7 +1311,7 @@ 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) `)`
@@ -1358,7 +1357,7 @@ 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) `)`
@@ -1375,7 +1374,7 @@ 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
@@ -1393,6 +1392,10 @@ 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
 
   }];
@@ -1401,7 +1404,7 @@ 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);
 
@@ -1410,7 +1413,7 @@ 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..ff4bd9d632ed04f 100644
--- a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
+++ b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
@@ -693,6 +693,12 @@ static ParseResult parseMapClause(OpAsmParser &parser, IntegerAttr &mapType) {
     if (mapTypeMod == "always")
       mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_ALWAYS;
 
+    if (mapTypeMod == "literal")
+      mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_LITERAL;
+
+    if (mapTypeMod == "implicit")
+      mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT;
+
     if (mapTypeMod == "close")
       mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_CLOSE;
 
@@ -740,6 +746,12 @@ 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_LITERAL))
+    mapTypeStrs.push_back("literal");
+  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 +792,64 @@ 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 +896,12 @@ 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");
 

>From e7afe1294edcd0c70ebc62e961ef5c73ec0ad95a Mon Sep 17 00:00:00 2001
From: Akash Banerjee <Akash.Banerjee at amd.com>
Date: Fri, 22 Sep 2023 18:13:46 +0100
Subject: [PATCH 2/2] [OpenMP][MLIR] Add "IsolatedFromAbove" trait to
 omp.target

This patch adds the MLIR translation changes required for add the IsolatedFromAbove and OutlineableOpenMPOpInterface traits to omp.target. It links the newly added block arguments to their corresponding llvm values.
---
 .../OpenMP/OpenMPToLLVMIRTranslation.cpp      | 59 +++++++++++++------
 .../OpenMPToLLVM/convert-to-llvmir.mlir       | 20 ++++---
 mlir/test/Dialect/OpenMP/canonicalize.mlir    |  5 +-
 mlir/test/Dialect/OpenMP/invalid.mlir         |  4 +-
 mlir/test/Dialect/OpenMP/ops.mlir             | 23 +++++---
 .../omptarget-declare-target-llvm-device.mlir | 13 ++--
 .../LLVMIR/omptarget-region-device-llvm.mlir  |  9 +--
 .../omptarget-region-llvm-target-device.mlir  |  9 +--
 .../Target/LLVMIR/omptarget-region-llvm.mlir  |  9 +--
 .../omptarget-region-parallel-llvm.mlir       |  9 +--
 10 files changed, 102 insertions(+), 58 deletions(-)

diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index e3dc68a1b8b7d72..8c687012031ba6c 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -1681,18 +1681,21 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
     if (!mapInfoOp.getType().isa<LLVM::LLVMPointerType>())
       return fail();
 
-    llvm::Value *mapOpValue =
-        moduleTranslation.lookupValue(mapInfoOp.getVarPtr());
+    llvm::Value *mapOpValue = moduleTranslation.lookupValue(
+        mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr() : mapInfoOp.getVal());
 
-    llvm::Value *refPtr =
-        getRefPtrIfDeclareTarget(mapInfoOp.getVarPtr(), moduleTranslation);
+    llvm::Value *refPtr = getRefPtrIfDeclareTarget(
+        mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr() : mapInfoOp.getVal(),
+        moduleTranslation);
 
     combinedInfo.BasePointers.emplace_back(refPtr ? refPtr : mapOpValue);
     combinedInfo.Pointers.emplace_back(mapOpValue);
     combinedInfo.DevicePointers.emplace_back(
         llvm::OpenMPIRBuilder::DeviceInfoTy::None);
     combinedInfo.Names.emplace_back(LLVM::createMappingInformation(
-        mapInfoOp.getVarPtr().getLoc(), *ompBuilder));
+        mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr().getLoc()
+                              : mapInfoOp.getVal().getLoc(),
+        *ompBuilder));
 
     auto mapFlag = llvm::omp::OpenMPOffloadMappingFlags(
         mapTypes[index].cast<IntegerAttr>().getUInt());
@@ -1706,8 +1709,9 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
       mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TARGET_PARAM;
 
     combinedInfo.Types.emplace_back(mapFlag);
-    combinedInfo.Sizes.emplace_back(
-        builder.getInt64(getSizeInBytes(DL, mapInfoOp.getVarPtr().getType())));
+    combinedInfo.Sizes.emplace_back(builder.getInt64(getSizeInBytes(
+        DL, mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr().getType()
+                                  : mapInfoOp.getVal().getType())));
     index++;
   }
 
@@ -2003,10 +2007,11 @@ handleDeclareTargetMapVar(llvm::ArrayRef<Value> mapOperands,
   for (const mlir::Value &mapOp : mapOperands) {
     auto mapInfoOp =
         mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp());
-    llvm::Value *mapOpValue =
-        moduleTranslation.lookupValue(mapInfoOp.getVarPtr());
-    if (auto *declareTarget = getRefPtrIfDeclareTarget(mapInfoOp.getVarPtr(),
-                                                       moduleTranslation)) {
+    llvm::Value *mapOpValue = moduleTranslation.lookupValue(
+        mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr() : mapInfoOp.getVal());
+    if (auto *declareTarget = getRefPtrIfDeclareTarget(
+            mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr() : mapInfoOp.getVal(),
+            moduleTranslation)) {
       // The user's iterator will get invalidated if we modify an element,
       // so we populate this vector of uses to alter each user on an individual
       // basis to emit its own load (rather than one load for all).
@@ -2017,7 +2022,9 @@ handleDeclareTargetMapVar(llvm::ArrayRef<Value> mapOperands,
       for (llvm::User *user : userVec) {
         if (auto *insn = dyn_cast<llvm::Instruction>(user)) {
           auto *load = builder.CreateLoad(
-              moduleTranslation.convertType(mapInfoOp.getVarPtr().getType()),
+              moduleTranslation.convertType(
+                  mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr().getType()
+                                        : mapInfoOp.getVal().getType()),
               declareTarget);
           load->moveBefore(insn);
           user->replaceUsesOfWith(mapOpValue, load);
@@ -2061,6 +2068,8 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
 
   auto targetOp = cast<omp::TargetOp>(opInst);
   auto &targetRegion = targetOp.getRegion();
+  DataLayout DL = DataLayout(opInst.getParentOfType<ModuleOp>());
+  SmallVector<Value> mapOperands = targetOp.getMapOperands();
 
   // This function filters out kernel data that will not show up as kernel
   // input arguments to the generated kernel function but will still need
@@ -2068,7 +2077,7 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
   // (declare target). It also prepares some data used for generating the
   // kernel and populating the associated OpenMP runtime data structures.
   auto getKernelArguments =
-      [&](const llvm::SetVector<Value> &operandSet,
+      [&](const llvm::SmallVectorImpl<Value> &operandSet,
           llvm::SmallVectorImpl<llvm::Value *> &llvmInputs) {
         for (Value operand : operandSet) {
           if (!getRefPtrIfDeclareTarget(operand, moduleTranslation))
@@ -2076,11 +2085,16 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
         }
       };
 
-  llvm::SetVector<Value> operandSet;
-  getUsedValuesDefinedAbove(targetRegion, operandSet);
+  llvm::SmallVector<Value> mapVals;
+  for (Value mapOp : mapOperands) {
+    auto mapInfoOp =
+        mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp());
+    mapVals.push_back(mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr()
+                                            : mapInfoOp.getVal());
+  }
 
   llvm::SmallVector<llvm::Value *> inputs;
-  getKernelArguments(operandSet, inputs);
+  getKernelArguments(mapVals, inputs);
 
   LogicalResult bodyGenStatus = success();
 
@@ -2088,6 +2102,16 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
   auto bodyCB = [&](InsertPointTy allocaIP,
                     InsertPointTy codeGenIP) -> InsertPointTy {
     builder.restoreIP(codeGenIP);
+    unsigned argIndex = 0;
+    for (auto &mapOp : mapOperands) {
+      auto mapInfoOp =
+          mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp());
+      llvm::Value *mapOpValue = moduleTranslation.lookupValue(
+          mapInfoOp.getVarPtr() ? mapInfoOp.getVarPtr() : mapInfoOp.getVal());
+      const auto &arg = targetRegion.front().getArgument(argIndex);
+      moduleTranslation.mapValue(arg, mapOpValue);
+      argIndex++;
+    }
     llvm::BasicBlock *exitBlock = convertOmpOpRegions(
         targetRegion, "omp.target", builder, moduleTranslation, bodyGenStatus);
     builder.SetInsertPoint(exitBlock);
@@ -2115,9 +2139,6 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
   llvm::OpenMPIRBuilder::InsertPointTy allocaIP =
       findAllocaInsertPoint(builder, moduleTranslation);
 
-  DataLayout DL = DataLayout(opInst.getParentOfType<ModuleOp>());
-  SmallVector<Value> mapOperands = targetOp.getMapOperands();
-
   auto getMapTypes = [](mlir::OperandRange mapOperands,
                         mlir::MLIRContext *ctx) {
     SmallVector<mlir::Attribute> mapTypes;
diff --git a/mlir/test/Conversion/OpenMPToLLVM/convert-to-llvmir.mlir b/mlir/test/Conversion/OpenMPToLLVM/convert-to-llvmir.mlir
index 881d738b413ef15..37a984f77a527a8 100644
--- a/mlir/test/Conversion/OpenMPToLLVM/convert-to-llvmir.mlir
+++ b/mlir/test/Conversion/OpenMPToLLVM/convert-to-llvmir.mlir
@@ -244,10 +244,12 @@ llvm.func @_QPomp_target_data_region(%a : !llvm.ptr<array<1024 x i32>>, %i : !ll
 // CHECK:                             %[[ARG_0:.*]]: !llvm.ptr<array<1024 x i32>>,
 // CHECK:                             %[[ARG_1:.*]]: !llvm.ptr<i32>) {
 // CHECK:           %[[VAL_0:.*]] = llvm.mlir.constant(64 : i32) : i32
-// CHECK:           %[[MAP:.*]] = omp.map_info var_ptr(%[[ARG_0]] : !llvm.ptr<array<1024 x i32>>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
-// CHECK:           omp.target   thread_limit(%[[VAL_0]] : i32) map_entries(%[[MAP]] : !llvm.ptr<array<1024 x i32>>) {
+// CHECK:           %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG_0]] : !llvm.ptr<array<1024 x i32>>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
+// CHECK:           %[[MAP2:.*]] = omp.map_info var_ptr(%[[ARG_1]] : !llvm.ptr<i32>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !llvm.ptr<i32> {name = ""}
+// CHECK:           omp.target   thread_limit(%[[VAL_0]] : i32) map_entries(%[[MAP1]] -> %[[BB_ARG0:.*]], %[[MAP2]] -> %[[BB_ARG1:.*]] : !llvm.ptr<array<1024 x i32>>, !llvm.ptr<i32>) {
+// CHECK:           ^bb0(%[[BB_ARG0]]: !llvm.ptr<array<1024 x i32>>, %[[BB_ARG1]]: !llvm.ptr<i32>):
 // CHECK:             %[[VAL_1:.*]] = llvm.mlir.constant(10 : i32) : i32
-// CHECK:             llvm.store %[[VAL_1]], %[[ARG_1]] : !llvm.ptr<i32>
+// CHECK:             llvm.store %[[VAL_1]], %[[BB_ARG1]] : !llvm.ptr<i32>
 // CHECK:             omp.terminator
 // CHECK:           }
 // CHECK:           llvm.return
@@ -256,9 +258,11 @@ llvm.func @_QPomp_target_data_region(%a : !llvm.ptr<array<1024 x i32>>, %i : !ll
 llvm.func @_QPomp_target(%a : !llvm.ptr<array<1024 x i32>>, %i : !llvm.ptr<i32>) {
   %0 = llvm.mlir.constant(64 : i32) : i32
   %1 = omp.map_info var_ptr(%a : !llvm.ptr<array<1024 x i32>>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
-  omp.target   thread_limit(%0 : i32) map_entries(%1 : !llvm.ptr<array<1024 x i32>>) {
+  %3 = omp.map_info var_ptr(%i : !llvm.ptr<i32>)   map_clauses(literal, implicit, exit_release_or_enter_alloc) capture(ByCopy) -> !llvm.ptr<i32> {name = ""}
+  omp.target   thread_limit(%0 : i32) map_entries(%1 -> %arg0, %3 -> %arg1 : !llvm.ptr<array<1024 x i32>>, !llvm.ptr<i32>) {
+    ^bb0(%arg0: !llvm.ptr<array<1024 x i32>>, %arg1: !llvm.ptr<i32>):
     %2 = llvm.mlir.constant(10 : i32) : i32
-    llvm.store %2, %i : !llvm.ptr<i32>
+    llvm.store %2, %arg1 : !llvm.ptr<i32>
     omp.terminator
   }
   llvm.return
@@ -451,7 +455,8 @@ llvm.func @sub_() {
 // CHECK: %[[C_14:.*]] = llvm.mlir.constant(1 : index) : i64
 // CHECK: %[[BOUNDS1:.*]] = omp.bounds   lower_bound(%[[C_12]] : i64) upper_bound(%[[C_11]] : i64) stride(%[[C_14]] : i64) start_idx(%[[C_14]] : i64)
 // CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG_2]] : !llvm.ptr<array<10 x i32>>)   map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS1]]) -> !llvm.ptr<array<10 x i32>> {name = ""}
-// CHECK: omp.target   map_entries(%[[MAP0]], %[[MAP1]] : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>) {
+// CHECK: omp.target   map_entries(%[[MAP0]] -> %[[BB_ARG0:.*]], %[[MAP1]]  -> %[[BB_ARG1:.*]] : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>) {
+// CHECK: ^bb0(%[[BB_ARG0]]: !llvm.ptr<array<10 x i32>>, %[[BB_ARG1]]: !llvm.ptr<array<10 x i32>>):
 // CHECK:   omp.terminator
 // CHECK: }
 // CHECK: llvm.return
@@ -470,7 +475,8 @@ llvm.func @_QPtarget_map_with_bounds(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<arr
   %9 = llvm.mlir.constant(1 : index) : i64
   %10 = omp.bounds   lower_bound(%7 : i64) upper_bound(%6 : i64) stride(%9 : i64) start_idx(%9 : i64)
   %11 = omp.map_info var_ptr(%arg2 : !llvm.ptr<array<10 x i32>>)   map_clauses(tofrom) capture(ByRef) bounds(%10) -> !llvm.ptr<array<10 x i32>> {name = ""}
-  omp.target   map_entries(%5, %11 : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>) {
+  omp.target   map_entries(%5 -> %arg3, %11 -> %arg4: !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>) {
+    ^bb0(%arg3: !llvm.ptr<array<10 x i32>>, %arg4: !llvm.ptr<array<10 x i32>>):
     omp.terminator
   }
   llvm.return
diff --git a/mlir/test/Dialect/OpenMP/canonicalize.mlir b/mlir/test/Dialect/OpenMP/canonicalize.mlir
index 68f5bacb1def178..4ecbb027ba47f24 100644
--- a/mlir/test/Dialect/OpenMP/canonicalize.mlir
+++ b/mlir/test/Dialect/OpenMP/canonicalize.mlir
@@ -131,8 +131,9 @@ func.func private @foo() -> ()
 
 func.func @constant_hoisting_target(%x : !llvm.ptr<i32>) {
   omp.target {
+    ^bb0(%arg0: !llvm.ptr<i32>):
     %c1 = arith.constant 10 : i32
-    llvm.store %c1, %x : i32, !llvm.ptr<i32>
+    llvm.store %c1, %arg0 : i32, !llvm.ptr<i32>
     omp.terminator
   }
   return
@@ -141,4 +142,4 @@ func.func @constant_hoisting_target(%x : !llvm.ptr<i32>) {
 // CHECK-LABEL: func.func @constant_hoisting_target
 // CHECK-NOT: arith.constant
 // CHECK: omp.target
-// CHECK-NEXT: arith.constant
+// CHECK: arith.constant
diff --git a/mlir/test/Dialect/OpenMP/invalid.mlir b/mlir/test/Dialect/OpenMP/invalid.mlir
index c8025249e27004e..36528e18b236f37 100644
--- a/mlir/test/Dialect/OpenMP/invalid.mlir
+++ b/mlir/test/Dialect/OpenMP/invalid.mlir
@@ -1617,7 +1617,9 @@ func.func @omp_threadprivate() {
 func.func @omp_target(%map1: memref<?xi32>) {
   %mapv = omp.map_info var_ptr(%map1 : memref<?xi32>)   map_clauses(delete) capture(ByRef) -> memref<?xi32> {name = ""}
   // expected-error @below {{to, from, tofrom and alloc map types are permitted}}
-  omp.target map_entries(%mapv : memref<?xi32>){}
+  omp.target map_entries(%mapv -> %arg0: memref<?xi32>) {
+    ^bb0(%arg0: memref<?xi32>):
+  }
   return
 }
 
diff --git a/mlir/test/Dialect/OpenMP/ops.mlir b/mlir/test/Dialect/OpenMP/ops.mlir
index 27c31824c0506df..63d3701cb13cd76 100644
--- a/mlir/test/Dialect/OpenMP/ops.mlir
+++ b/mlir/test/Dialect/OpenMP/ops.mlir
@@ -487,21 +487,27 @@ func.func @omp_target(%if_cond : i1, %device : si32,  %num_threads : i32, %map1:
     "omp.target"(%if_cond, %device, %num_threads) ({
        // CHECK: omp.terminator
        omp.terminator
-    }) {nowait, operandSegmentSizes = array<i32: 1,1,1,0>} : ( i1, si32, i32 ) -> ()
+    }) {nowait, operandSegmentSizes = array<i32: 1,1,1,0,0>} : ( i1, si32, i32 ) -> ()
 
     // Test with optional map clause.
     // CHECK: %[[MAP_A:.*]] = omp.map_info var_ptr(%[[VAL_1:.*]] : memref<?xi32>)   map_clauses(tofrom) capture(ByRef) -> memref<?xi32> {name = ""}
     // CHECK: %[[MAP_B:.*]] = omp.map_info var_ptr(%[[VAL_2:.*]] : memref<?xi32>)   map_clauses(exit_release_or_enter_alloc) capture(ByRef) -> memref<?xi32> {name = ""}
-    // CHECK: omp.target map_entries(%[[MAP_A]], %[[MAP_B]] : memref<?xi32>, memref<?xi32>) {
+    // CHECK: omp.target map_entries(%[[MAP_A]] -> {{.*}}, %[[MAP_B]] -> {{.*}} : memref<?xi32>, memref<?xi32>) {
     %mapv1 = omp.map_info var_ptr(%map1 : memref<?xi32>)   map_clauses(tofrom) capture(ByRef) -> memref<?xi32> {name = ""}
     %mapv2 = omp.map_info var_ptr(%map2 : memref<?xi32>)   map_clauses(exit_release_or_enter_alloc) capture(ByRef) -> memref<?xi32> {name = ""}
-    omp.target map_entries(%mapv1, %mapv2 : memref<?xi32>, memref<?xi32>){}
+    omp.target map_entries(%mapv1 -> %arg0, %mapv2 -> %arg1 : memref<?xi32>, memref<?xi32>) {
+    ^bb0(%arg0: memref<?xi32>, %arg1: memref<?xi32>):
+      omp.terminator
+    }
     // CHECK: %[[MAP_C:.*]] = omp.map_info var_ptr(%[[VAL_1:.*]] : memref<?xi32>)   map_clauses(to) capture(ByRef) -> memref<?xi32> {name = ""}
     // CHECK: %[[MAP_D:.*]] = omp.map_info var_ptr(%[[VAL_2:.*]] : memref<?xi32>)   map_clauses(always, from) capture(ByRef) -> memref<?xi32> {name = ""}
-    // CHECK: omp.target map_entries(%[[MAP_C]], %[[MAP_D]] : memref<?xi32>, memref<?xi32>) {
+    // CHECK: omp.target map_entries(%[[MAP_C]] -> {{.*}}, %[[MAP_D]] -> {{.*}} : memref<?xi32>, memref<?xi32>) {
     %mapv3 = omp.map_info var_ptr(%map1 : memref<?xi32>)   map_clauses(to) capture(ByRef) -> memref<?xi32> {name = ""}
     %mapv4 = omp.map_info var_ptr(%map2 : memref<?xi32>)   map_clauses(always, from) capture(ByRef) -> memref<?xi32> {name = ""}
-    omp.target map_entries(%mapv3, %mapv4 : memref<?xi32>, memref<?xi32>) {}
+    omp.target map_entries(%mapv3 -> %arg0, %mapv4 -> %arg1 : memref<?xi32>, memref<?xi32>) {
+    ^bb0(%arg0: memref<?xi32>, %arg1: memref<?xi32>):
+      omp.terminator
+    }
     // CHECK: omp.barrier
     omp.barrier
 
@@ -2055,8 +2061,11 @@ func.func @omp_targets_with_map_bounds(%arg0: !llvm.ptr<array<10 x i32>>, %arg1:
     %10 = omp.bounds   lower_bound(%7 : i64) upper_bound(%6 : i64) stride(%8 : i64) start_idx(%9 : i64)
     %mapv2 = omp.map_info var_ptr(%arg1 : !llvm.ptr<array<10 x i32>>)   map_clauses(exit_release_or_enter_alloc) capture(ByCopy) bounds(%10) -> !llvm.ptr<array<10 x i32>> {name = ""}
 
-    // CHECK: omp.target map_entries(%[[MAP0]], %[[MAP1]] : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>)
-    omp.target map_entries(%mapv1, %mapv2 : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>){}
+    // CHECK: omp.target map_entries(%[[MAP0]] -> {{.*}}, %[[MAP1]] -> {{.*}} : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>)
+    omp.target map_entries(%mapv1 -> %arg2, %mapv2 -> %arg3 : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>) {
+    ^bb0(%arg2: !llvm.ptr<array<10 x i32>>, %arg3: !llvm.ptr<array<10 x i32>>):
+      omp.terminator
+    }
 
     // CHECK: omp.target_data map_entries(%[[MAP0]], %[[MAP1]] : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>)
     omp.target_data map_entries(%mapv1, %mapv2 : !llvm.ptr<array<10 x i32>>, !llvm.ptr<array<10 x i32>>){}
diff --git a/mlir/test/Target/LLVMIR/omptarget-declare-target-llvm-device.mlir b/mlir/test/Target/LLVMIR/omptarget-declare-target-llvm-device.mlir
index f279772bd6c0969..f4c7a626398f5a9 100644
--- a/mlir/test/Target/LLVMIR/omptarget-declare-target-llvm-device.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-declare-target-llvm-device.mlir
@@ -1,10 +1,10 @@
 // RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
 
 // This test the generation of additional load operations for declare target link variables
-// inside of target op regions when lowering to IR for device. Unfortunately as the host file is not 
+// inside of target op regions when lowering to IR for device. Unfortunately as the host file is not
 // passed as a module attribute, we miss out on the metadata and entryinfo.
 //
-// Unfortunately, only so much can be tested as the device side is dependent on a *.bc 
+// Unfortunately, only so much can be tested as the device side is dependent on a *.bc
 // file created by the host and appended as an attribute to the module.
 
 module attributes {omp.is_target_device = true} {
@@ -13,18 +13,19 @@ module attributes {omp.is_target_device = true} {
     %0 = llvm.mlir.constant(0 : i32) : i32
     llvm.return %0 : i32
   }
-                                                            
+
   llvm.func @_QQmain() attributes {} {
     %0 = llvm.mlir.addressof @_QMtest_0Esp : !llvm.ptr<i32>
-  
+
   // CHECK-DAG:   omp.target:                                       ; preds = %user_code.entry
   // CHECK-DAG: %1 = load ptr, ptr @_QMtest_0Esp_decl_tgt_ref_ptr, align 8
   // CHECK-DAG: store i32 1, ptr %1, align 4
   // CHECK-DAG: br label %omp.region.cont
     %map = omp.map_info var_ptr(%0 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    omp.target   map_entries(%map : !llvm.ptr<i32>) {
+    omp.target   map_entries(%map -> %arg0 : !llvm.ptr<i32>) {
+      ^bb0(%arg0: !llvm.ptr<i32>):
       %1 = llvm.mlir.constant(1 : i32) : i32
-      llvm.store %1, %0 : !llvm.ptr<i32>
+      llvm.store %1, %arg0 : !llvm.ptr<i32>
       omp.terminator
     }
 
diff --git a/mlir/test/Target/LLVMIR/omptarget-region-device-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-region-device-llvm.mlir
index 99f1a3b072ad8fe..d02e554ccd4d525 100644
--- a/mlir/test/Target/LLVMIR/omptarget-region-device-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-region-device-llvm.mlir
@@ -15,11 +15,12 @@ module attributes {omp.is_target_device = true} {
     %map1 = omp.map_info var_ptr(%3 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map2 = omp.map_info var_ptr(%5 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map3 = omp.map_info var_ptr(%7 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    omp.target map_entries(%map1, %map2, %map3 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
-      %8 = llvm.load %3 : !llvm.ptr<i32>
-      %9 = llvm.load %5 : !llvm.ptr<i32>
+    omp.target map_entries(%map1 -> %arg0, %map2 -> %arg1, %map3 -> %arg2 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
+    ^bb0(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<i32>, %arg2: !llvm.ptr<i32>):
+      %8 = llvm.load %arg0 : !llvm.ptr<i32>
+      %9 = llvm.load %arg1 : !llvm.ptr<i32>
       %10 = llvm.add %8, %9  : i32
-      llvm.store %10, %7 : !llvm.ptr<i32>
+      llvm.store %10, %arg2 : !llvm.ptr<i32>
       omp.terminator
     }
     llvm.return
diff --git a/mlir/test/Target/LLVMIR/omptarget-region-llvm-target-device.mlir b/mlir/test/Target/LLVMIR/omptarget-region-llvm-target-device.mlir
index 01a3bd556294f3e..1032a8d8bc999bb 100644
--- a/mlir/test/Target/LLVMIR/omptarget-region-llvm-target-device.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-region-llvm-target-device.mlir
@@ -3,10 +3,11 @@
 // RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
 
 module attributes {omp.is_target_device = true} {
-  llvm.func @writeindex_omp_outline_0_(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<i32>) attributes {omp.outline_parent_name = "writeindex_"} {
-    %0 = omp.map_info var_ptr(%arg0 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    %1 = omp.map_info var_ptr(%arg1 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    omp.target   map_entries(%0, %1 : !llvm.ptr<i32>, !llvm.ptr<i32>) {
+  llvm.func @writeindex_omp_outline_0_(%val0: !llvm.ptr<i32>, %val1: !llvm.ptr<i32>) attributes {omp.outline_parent_name = "writeindex_"} {
+    %0 = omp.map_info var_ptr(%val0 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
+    %1 = omp.map_info var_ptr(%val1 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
+    omp.target   map_entries(%0 -> %arg0, %1 -> %arg1 : !llvm.ptr<i32>, !llvm.ptr<i32>) {
+    ^bb0(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<i32>):
       %2 = llvm.mlir.constant(20 : i32) : i32
       %3 = llvm.mlir.constant(10 : i32) : i32
       llvm.store %3, %arg0 : !llvm.ptr<i32>
diff --git a/mlir/test/Target/LLVMIR/omptarget-region-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-region-llvm.mlir
index 51506386d83782d..1c4a756eabef71e 100644
--- a/mlir/test/Target/LLVMIR/omptarget-region-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-region-llvm.mlir
@@ -15,11 +15,12 @@ module attributes {omp.is_target_device = false} {
     %map1 = omp.map_info var_ptr(%3 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map2 = omp.map_info var_ptr(%5 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map3 = omp.map_info var_ptr(%7 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    omp.target map_entries(%map1, %map2, %map3 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
-      %8 = llvm.load %3 : !llvm.ptr<i32>
-      %9 = llvm.load %5 : !llvm.ptr<i32>
+    omp.target map_entries(%map1 -> %arg0, %map2 -> %arg1, %map3 -> %arg2 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
+    ^bb0(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<i32>, %arg2: !llvm.ptr<i32>):
+      %8 = llvm.load %arg0 : !llvm.ptr<i32>
+      %9 = llvm.load %arg1 : !llvm.ptr<i32>
       %10 = llvm.add %8, %9  : i32
-      llvm.store %10, %7 : !llvm.ptr<i32>
+      llvm.store %10, %arg2 : !llvm.ptr<i32>
       omp.terminator
     }
     llvm.return
diff --git a/mlir/test/Target/LLVMIR/omptarget-region-parallel-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-region-parallel-llvm.mlir
index 5741e44862d8b3f..80ebee56171735f 100644
--- a/mlir/test/Target/LLVMIR/omptarget-region-parallel-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-region-parallel-llvm.mlir
@@ -15,12 +15,13 @@ module attributes {omp.is_target_device = false} {
     %map1 = omp.map_info var_ptr(%3 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map2 = omp.map_info var_ptr(%5 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
     %map3 = omp.map_info var_ptr(%7 : !llvm.ptr<i32>)   map_clauses(tofrom) capture(ByRef) -> !llvm.ptr<i32> {name = ""}
-    omp.target map_entries( %map1, %map2, %map3 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
+    omp.target map_entries( %map1 -> %arg0, %map2 -> %arg1, %map3 -> %arg2 : !llvm.ptr<i32>, !llvm.ptr<i32>, !llvm.ptr<i32>) {
+    ^bb0(%arg0: !llvm.ptr<i32>, %arg1: !llvm.ptr<i32>, %arg2: !llvm.ptr<i32>):
       omp.parallel {
-        %8 = llvm.load %3 : !llvm.ptr<i32>
-        %9 = llvm.load %5 : !llvm.ptr<i32>
+        %8 = llvm.load %arg0 : !llvm.ptr<i32>
+        %9 = llvm.load %arg1 : !llvm.ptr<i32>
         %10 = llvm.add %8, %9  : i32
-        llvm.store %10, %7 : !llvm.ptr<i32>
+        llvm.store %10, %arg2 : !llvm.ptr<i32>
         omp.terminator
         }
       omp.terminator



More information about the flang-commits mailing list