[flang-commits] [flang] [mlir] [openmp] [Flang][OpenMP][MLIR] Extend derived (record) type map support in Flang and OpenMP dialect (PR #81328)
via flang-commits
flang-commits at lists.llvm.org
Fri Feb 9 14:47:53 PST 2024
https://github.com/agozillon created https://github.com/llvm/llvm-project/pull/81328
This patch seeks to refactor slightly and extend the current record type map support that was put in place for Fortran's descriptor types to handle explicit member mapping for record types at a single level of depth.
For example, the below case where two members of a Fortran derived type are mapped explicitly:
''''
type :: scalar_and_array
real(4) :: real
integer(4) :: array(10)
integer(4) :: int
end type scalar_and_array
type(scalar_and_array) :: scalar_arr
!$omp target map(tofrom: scalar_arr%int, scalar_arr%real)
''''
Current cases of derived type mapping left for future work are:
> explicit member mapping of nested members (e.g. two layers of
record types where we explicitly map a member from the internal
record type)
> Fortran's automagical mapping of all elements and nested elements
of a derived type
> explicit member mapping of a derived type and then constituent members
(redundant in Fortran due to former case but still legal as far as I am aware)
> explicit member mapping of a record type (may be handled reasonably, just
not fully tested in this iteration)
> explicit member mapping for Fortran allocatable types (a variation of nested
record types)
This patch seeks to support this by extending the OpenMPToLLVMIRTranslation phase to more generally support record types, building on the prior groundwork in the Fortran allocatables/pointers patch. It now supports different kinds of record type mapping, in this case full record type mapping and then explicit member mapping in which there is a special case for certain types when mapped individually to not require any parent map link in the kernel argument structure. To facilitate this required:
> The movement of the setting of the map flag type "ptr_and_obj" to respective
frontends, now supporting it as a possible flag that can be read and printed
in mlir form. Some minor changes to declare target map type setting was
necessary for this.
> The addition of a member index array operand, which tracks the position
of the member in the parent, required for calculating the appropriate size
to offload to the target, alongside the parents offload pointer (always the
first member currently being mapped).
> A partial mapping attribute operand, to indicate if the entire record type is
being mapped or just member components, aiding the ability to lower
record types in the different manners that are possible.
> Refactoring bounds calculation for record types and general arrays to one
location (as well as load/store generation prior to assigning to the kernel
argument structure), as a side affect enter/exit/update/data mapping
should now be more correct and fully support bounds mapping, previously
this would have only worked for target.
It also extends the Flang-new OpenMP lowering to support generation of this newly required information, creating the necessary parent <-to-> member map_info links, calculating the member indices and setting if it's a partial map.
The OMPDescriptorMapInfoGen pass has been generalized into a map finalization phase, now named OMPMapInfoFinalization. This pass was extended to support the insertion of member maps into the BlockArg and MapOperands of relevant map carrying operations. Similar to the method in which descriptor types are expanded and constituent members inserted.
Otherwise, this patch contains a large amount of tests, both runtime and compilation related, which will hopefully help showcase the cases this patch aims to cover and the IR changes necessary.
>From e22a1bf10e9c6eccf47cad2fc853c3439a70b8da Mon Sep 17 00:00:00 2001
From: Andrew Gozillon <Andrew.Gozillon at amd.com>
Date: Fri, 9 Feb 2024 15:51:27 -0600
Subject: [PATCH] [Flang][OpenMP][MLIR] Extend derived (record) type map
support in Flang and OpenMP dialect
This patch seeks to refactor slightly and extend the current
record type map support that was put in place for Fortran's
descriptor types to handle explicit member mapping for
record types at a single level of depth.
For example, the below case where two members of a Fortran
derived type are mapped explicitly:
''''
type :: scalar_and_array
real(4) :: real
integer(4) :: array(10)
integer(4) :: int
end type scalar_and_array
type(scalar_and_array) :: scalar_arr
!$omp target map(tofrom: scalar_arr%int, scalar_arr%real)
''''
Current cases of derived type mapping left for future work are:
> explicit member mapping of nested members (e.g. two layers of
record types where we explicitly map a member from the internal
record type)
> Fortran's automagical mapping of all elements and nested elements
of a derived type
> explicit member mapping of a derived type and then constituient members
(redundant in Fortran due to former case but still legal as far as I am aware)
> explicit member mapping of a record type (may be handled reasonably, just
not fully tested in this iteration)
> explicit member mapping for Fortran allocatable types (a variation of nested
record types)
This patch seeks to support this by extending the OpenMPToLLVMIRTranslation phase
to more generally support record types, building on the prior groundwork in the
Fortran allocatables/pointers patch. It now supports different kinds of record type
mapping, in this case full record type mapping and then explicit member mapping
in which there is a special case for certain types when mapped individually to not
require any parent map link in the kernel argument structure. To facilitate this
required:
> The movement of the setting of the map flag type "ptr_and_obj" to respective
frontends, now supporting it as a possible flag that can be read and printed
in mlir form. Some minor changes to declare target map type setting was
neccesary for this.
> The addition of a member index array operand, which tracks the position
of the member in the parent, required for caclulating the appropriate size
to offload to the target, alongside the parents offload pointer (always the
first member currently being mapped).
> A partial mapping attribute operand, to indicate if the entire record type is
being mapped or just member components, aiding the ability to lower
record types in the different manners that are possible.
> Refactoring bounds calculation for record types and general arrays to one
location (as well as load/store generation prior to assigning to the kernel
argument structure), as a side affect enter/exit/update/data mapping
should now be more correct and fully support bounds mapping, previously
this would have only worked for target.
It also extends the Flang-new OpenMP lowering to support generation of this newly
required information, creating the neccessary parent <-to-> member map_info links,
calculating the member indices and setting if it's a partial map.
The OMPDescriptorMapInfoGen pass has been generalized into a map finalization
phase, now named OMPMapInfoFinalization. This pass was extended to support
the insertion of member maps into the BlockArg and MapOperands of relevant
map carrying operations. Similar to the method in which descriptor types are
expanded and constituient members inserted.
Otherwise, this patch contains a large amount of tests, both runtime and
compilation related, which will hopefully help showcase the cases this
patch aims to cover and the IR changes neccesary.
---
flang/docs/OpenMP-descriptor-management.md | 4 +-
.../flang/Optimizer/Transforms/Passes.h | 2 +-
.../flang/Optimizer/Transforms/Passes.td | 6 +-
flang/include/flang/Tools/CLOptions.inc | 2 +-
flang/lib/Lower/OpenMP.cpp | 235 +++++++-
flang/lib/Optimizer/Transforms/CMakeLists.txt | 2 +-
.../Transforms/OMPDescriptorMapInfoGen.cpp | 168 ------
.../Transforms/OMPMapInfoFinalization.cpp | 262 +++++++++
.../Fir/convert-to-llvm-openmp-and-fir.fir | 35 +-
.../OpenMP/map-types-and-sizes.f90 | 154 +++++-
flang/test/Lower/OpenMP/FIR/array-bounds.f90 | 4 +-
.../Lower/OpenMP/FIR/map-component-ref.f90 | 4 +-
flang/test/Lower/OpenMP/FIR/target.f90 | 4 +-
.../Lower/OpenMP/allocatable-array-bounds.f90 | 12 +-
flang/test/Lower/OpenMP/allocatable-map.f90 | 8 +-
flang/test/Lower/OpenMP/array-bounds.f90 | 8 +-
flang/test/Lower/OpenMP/derived-type-map.f90 | 105 ++++
flang/test/Lower/OpenMP/map-component-ref.f90 | 2 +-
flang/test/Lower/OpenMP/target.f90 | 4 +-
...-gen.fir => omp-map-info-finalization.fir} | 34 +-
mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td | 16 +-
mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp | 7 +
.../OpenMP/OpenMPToLLVMIRTranslation.cpp | 504 +++++++++++-------
mlir/test/Dialect/OpenMP/ops.mlir | 28 +-
...target-fortran-allocatable-types-host.mlir | 30 +-
mlir/test/Target/LLVMIR/omptarget-llvm.mlir | 49 +-
.../omptarget-record-type-mapping-host.mlir | 63 +++
.../target-map-derived-type-full-1.f90 | 45 ++
.../target-map-derived-type-full-2.f90 | 60 +++
...arget-map-derived-type-full-implicit-1.f90 | 46 ++
...arget-map-derived-type-full-implicit-2.f90 | 61 +++
...pe-arr-bounds-member-enter-exit-update.f90 | 49 ++
...map-dtype-arr-bounds-member-enter-exit.f90 | 49 ++
...dtype-explicit-individual-array-member.f90 | 33 ++
...-multi-explicit-array-3D-member-bounds.f90 | 45 ++
...ype-multi-explicit-array-member-bounds.f90 | 38 ++
...-map-dtype-multi-explicit-array-member.f90 | 39 ++
...target-map-dtype-multi-explicit-member.f90 | 33 ++
.../fortran/target-map-enter-exit-array-2.f90 | 39 ++
.../target-map-enter-exit-array-bounds.f90 | 44 ++
.../fortran/target-map-enter-exit-scalar.f90 | 33 ++
...target-map-individual-dtype-member-map.f90 | 33 ++
.../target-map-two-dtype-explicit-member.f90 | 35 ++
...type-individual-member-array-1D-bounds.f90 | 39 ++
...type-mixed-implicit-explicit-capture-1.f90 | 35 ++
...type-mixed-implicit-explicit-capture-2.f90 | 41 ++
...two-dtype-multi-member-array-1D-bounds.f90 | 51 ++
47 files changed, 2114 insertions(+), 486 deletions(-)
delete mode 100644 flang/lib/Optimizer/Transforms/OMPDescriptorMapInfoGen.cpp
create mode 100644 flang/lib/Optimizer/Transforms/OMPMapInfoFinalization.cpp
create mode 100644 flang/test/Lower/OpenMP/derived-type-map.f90
rename flang/test/Transforms/{omp-descriptor-map-info-gen.fir => omp-map-info-finalization.fir} (53%)
create mode 100644 mlir/test/Target/LLVMIR/omptarget-record-type-mapping-host.mlir
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-1.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-2.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-1.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-2.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit-update.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-explicit-individual-array-member.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-3D-member-bounds.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member-bounds.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-member.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-2.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-bounds.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-scalar.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-individual-dtype-member-map.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-explicit-member.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-individual-member-array-1D-bounds.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-1.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-2.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-multi-member-array-1D-bounds.f90
diff --git a/flang/docs/OpenMP-descriptor-management.md b/flang/docs/OpenMP-descriptor-management.md
index 90a20282e05126..af02b3a99cb07d 100644
--- a/flang/docs/OpenMP-descriptor-management.md
+++ b/flang/docs/OpenMP-descriptor-management.md
@@ -44,7 +44,7 @@ Currently, Flang will lower these descriptor types in the OpenMP lowering (lower
to all other map types, generating an omp.MapInfoOp containing relevant information required for lowering
the OpenMP dialect to LLVM-IR during the final stages of the MLIR lowering. However, after
the lowering to FIR/HLFIR has been performed an OpenMP dialect specific pass for Fortran,
-`OMPDescriptorMapInfoGenPass` (Optimizer/OMPDescriptorMapInfoGen.cpp) will expand the
+`OMPMapInfoFinalizationPass` (Optimizer/OMPMapInfoFinalization.cpp) will expand the
`omp.MapInfoOp`'s containing descriptors (which currently will be a `BoxType` or `BoxAddrOp`) into multiple
mappings, with one extra per pointer member in the descriptor that is supported on top of the original
descriptor map operation. These pointers members are linked to the parent descriptor by adding them to
@@ -52,7 +52,7 @@ the member field of the original descriptor map operation, they are then inserte
owning operation's (`omp.TargetOp`, `omp.DataOp` etc.) map operand list and in cases where the owning operation
is `IsolatedFromAbove`, it also inserts them as `BlockArgs` to canonicalize the mappings and simplify lowering.
-An example transformation by the `OMPDescriptorMapInfoGenPass`:
+An example transformation by the `OMPMapInfoFinalizationPass`:
```
diff --git a/flang/include/flang/Optimizer/Transforms/Passes.h b/flang/include/flang/Optimizer/Transforms/Passes.h
index e1d22c8c986da7..fc9a098c3931d3 100644
--- a/flang/include/flang/Optimizer/Transforms/Passes.h
+++ b/flang/include/flang/Optimizer/Transforms/Passes.h
@@ -76,7 +76,7 @@ std::unique_ptr<mlir::Pass>
createAlgebraicSimplificationPass(const mlir::GreedyRewriteConfig &config);
std::unique_ptr<mlir::Pass> createPolymorphicOpConversionPass();
-std::unique_ptr<mlir::Pass> createOMPDescriptorMapInfoGenPass();
+std::unique_ptr<mlir::Pass> createOMPMapInfoFinalizationPass();
std::unique_ptr<mlir::Pass> createOMPFunctionFilteringPass();
std::unique_ptr<mlir::OperationPass<mlir::ModuleOp>>
createOMPMarkDeclareTargetPass();
diff --git a/flang/include/flang/Optimizer/Transforms/Passes.td b/flang/include/flang/Optimizer/Transforms/Passes.td
index 5fb576fd876254..0638ae49f5f4ea 100644
--- a/flang/include/flang/Optimizer/Transforms/Passes.td
+++ b/flang/include/flang/Optimizer/Transforms/Passes.td
@@ -318,15 +318,15 @@ def LoopVersioning : Pass<"loop-versioning", "mlir::func::FuncOp"> {
let dependentDialects = [ "fir::FIROpsDialect" ];
}
-def OMPDescriptorMapInfoGenPass
- : Pass<"omp-descriptor-map-info-gen", "mlir::func::FuncOp"> {
+def OMPMapInfoFinalizationPass
+ : Pass<"omp-map-info-finalization", "mlir::func::FuncOp"> {
let summary = "expands OpenMP MapInfo operations containing descriptors";
let description = [{
Expands MapInfo operations containing descriptor types into multiple
MapInfo's for each pointer element in the descriptor that requires
explicit individual mapping by the OpenMP runtime.
}];
- let constructor = "::fir::createOMPDescriptorMapInfoGenPass()";
+ let constructor = "::fir::createOMPMapInfoFinalizationPass()";
let dependentDialects = ["mlir::omp::OpenMPDialect"];
}
diff --git a/flang/include/flang/Tools/CLOptions.inc b/flang/include/flang/Tools/CLOptions.inc
index 68e504d0ccb512..ec3d634ac0264b 100644
--- a/flang/include/flang/Tools/CLOptions.inc
+++ b/flang/include/flang/Tools/CLOptions.inc
@@ -274,7 +274,7 @@ inline void createHLFIRToFIRPassPipeline(
/// rather than the host device.
inline void createOpenMPFIRPassPipeline(
mlir::PassManager &pm, bool isTargetDevice) {
- pm.addPass(fir::createOMPDescriptorMapInfoGenPass());
+ pm.addPass(fir::createOMPMapInfoFinalizationPass());
pm.addPass(fir::createOMPMarkDeclareTargetPass());
if (isTargetDevice)
pm.addPass(fir::createOMPFunctionFilteringPass());
diff --git a/flang/lib/Lower/OpenMP.cpp b/flang/lib/Lower/OpenMP.cpp
index fd18b212bad515..86fdc51602cf12 100644
--- a/flang/lib/Lower/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP.cpp
@@ -49,14 +49,16 @@ using DeclareTargetCapturePair =
//===----------------------------------------------------------------------===//
static Fortran::semantics::Symbol *
-getOmpObjectSymbol(const Fortran::parser::OmpObject &ompObject) {
+getOmpObjParentSymbol(const Fortran::parser::OmpObject &ompObject) {
Fortran::semantics::Symbol *sym = nullptr;
std::visit(
Fortran::common::visitors{
[&](const Fortran::parser::Designator &designator) {
- if (auto *arrayEle =
- Fortran::parser::Unwrap<Fortran::parser::ArrayElement>(
- designator)) {
+ if (auto *structComp = Fortran::parser::Unwrap<
+ Fortran::parser::StructureComponent>(designator)) {
+ sym = GetFirstName(structComp->base).symbol;
+ } else if (auto *arrayEle = Fortran::parser::Unwrap<
+ Fortran::parser::ArrayElement>(designator)) {
sym = GetFirstName(arrayEle->base).symbol;
} else if (auto *structComp = Fortran::parser::Unwrap<
Fortran::parser::StructureComponent>(designator)) {
@@ -72,6 +74,29 @@ getOmpObjectSymbol(const Fortran::parser::OmpObject &ompObject) {
return sym;
}
+static Fortran::semantics::Symbol *
+getOmpObjectSymbol(const Fortran::parser::OmpObject &ompObject) {
+ Fortran::semantics::Symbol *sym = nullptr;
+ std::visit(
+ Fortran::common::visitors{
+ [&](const Fortran::parser::Designator &designator) {
+ if (auto *structComp = Fortran::parser::Unwrap<
+ Fortran::parser::StructureComponent>(designator)) {
+ sym = structComp->component.symbol;
+ } else if (auto *arrayEle = Fortran::parser::Unwrap<
+ Fortran::parser::ArrayElement>(designator)) {
+ sym = GetLastName(arrayEle->base).symbol;
+ } else if (const Fortran::parser::Name *name =
+ Fortran::semantics::getDesignatorNameIfDataRef(
+ designator)) {
+ sym = name->symbol;
+ }
+ },
+ [&](const Fortran::parser::Name &name) { sym = name.symbol; }},
+ ompObject.u);
+ return sym;
+}
+
static void genObjectList(const Fortran::parser::OmpObjectList &objectList,
Fortran::lower::AbstractConverter &converter,
llvm::SmallVectorImpl<mlir::Value> &operands) {
@@ -1829,9 +1854,10 @@ static mlir::omp::MapInfoOp
createMapInfoOp(fir::FirOpBuilder &builder, mlir::Location loc,
mlir::Value baseAddr, mlir::Value varPtrPtr, std::string name,
mlir::SmallVector<mlir::Value> bounds,
- mlir::SmallVector<mlir::Value> members, uint64_t mapType,
+ mlir::SmallVector<mlir::Value> members,
+ mlir::ArrayAttr membersIndex, uint64_t mapType,
mlir::omp::VariableCaptureKind mapCaptureType, mlir::Type retTy,
- bool isVal = false) {
+ bool partialMap = false) {
if (auto boxTy = baseAddr.getType().dyn_cast<fir::BaseBoxType>()) {
baseAddr = builder.create<fir::BoxAddrOp>(loc, baseAddr);
retTy = baseAddr.getType();
@@ -1841,14 +1867,112 @@ createMapInfoOp(fir::FirOpBuilder &builder, mlir::Location loc,
llvm::cast<mlir::omp::PointerLikeType>(retTy).getElementType());
mlir::omp::MapInfoOp op = builder.create<mlir::omp::MapInfoOp>(
- loc, retTy, baseAddr, varType, varPtrPtr, members, bounds,
+ loc, retTy, baseAddr, varType, varPtrPtr, members, membersIndex, bounds,
builder.getIntegerAttr(builder.getIntegerType(64, false), mapType),
builder.getAttr<mlir::omp::VariableCaptureKindAttr>(mapCaptureType),
- builder.getStringAttr(name));
+ builder.getStringAttr(name), builder.getBoolAttr(partialMap));
return op;
}
+int findComponenetMemberPlacement(
+ const Fortran::semantics::Symbol *dTypeSym,
+ const Fortran::semantics::Symbol *componentSym) {
+ int placement = -1;
+ if (const auto *derived{
+ dTypeSym->detailsIf<Fortran::semantics::DerivedTypeDetails>()}) {
+ for (auto t : derived->componentNames()) {
+ placement++;
+ if (t == componentSym->name())
+ return placement;
+ }
+ }
+ return placement;
+}
+
+static void
+checkAndApplyDeclTargetMapFlags(Fortran::lower::AbstractConverter &converter,
+ llvm::omp::OpenMPOffloadMappingFlags &mapFlags,
+ Fortran::semantics::Symbol *symbol) {
+ mlir::Operation *op =
+ converter.getModuleOp().lookupSymbol(converter.mangleName(*symbol));
+ if (op)
+ if (auto declareTargetOp =
+ llvm::dyn_cast<mlir::omp::DeclareTargetInterface>(op)) {
+ // only Link clauses have OMP_MAP_PTR_AND_OBJ applied, To clause
+ // functions fairly different.
+ if (declareTargetOp.getDeclareTargetCaptureClause() ==
+ mlir::omp::DeclareTargetCaptureClause::link)
+ mapFlags |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
+ }
+}
+
+static void insertChildMapInfoIntoParent(
+ Fortran::lower::AbstractConverter &converter,
+ llvm::SmallVector<const Fortran::semantics::Symbol *> &memberParentSyms,
+ llvm::SmallVector<mlir::Value> &memberMaps,
+ llvm::SmallVector<mlir::Attribute> &memberPlacementIndices,
+ llvm::SmallVectorImpl<mlir::Value> &mapOperands,
+ llvm::SmallVectorImpl<mlir::Type> *mapSymTypes,
+ llvm::SmallVectorImpl<mlir::Location> *mapSymLocs,
+ llvm::SmallVectorImpl<const Fortran::semantics::Symbol *> *mapSymbols) {
+ // TODO: For multi-nested record types the top level parent is currently
+ // the containing parent for all member operations.
+ for (auto [idx, sym] : llvm::enumerate(memberParentSyms)) {
+ bool parentExists = false;
+ size_t parentIdx = 0;
+ for (size_t i = 0; i < mapSymbols->size(); ++i) {
+ if ((*mapSymbols)[i] == sym) {
+ parentExists = true;
+ parentIdx = i;
+ }
+ }
+
+ if (parentExists) {
+ // found a parent, append.
+ if (auto mapOp = mlir::dyn_cast<mlir::omp::MapInfoOp>(
+ mapOperands[parentIdx].getDefiningOp())) {
+ mapOp.getMembersMutable().append(memberMaps[idx]);
+ llvm::SmallVector<mlir::Attribute> memberIndexTmp{
+ mapOp.getMembersIndexAttr().begin(),
+ mapOp.getMembersIndexAttr().end()};
+ memberIndexTmp.push_back(memberPlacementIndices[idx]);
+ mapOp.setMembersIndexAttr(mlir::ArrayAttr::get(
+ converter.getFirOpBuilder().getContext(), memberIndexTmp));
+ }
+ } else {
+ // NOTE: We take the map type of the first child, this may not
+ // be the correct thing to do, however, we shall see. For the moment
+ // it allows this to work with enter and exit without causing MLIR
+ // verification issues. The more appropriate thing may be to take
+ // the "main" map type clause from the directive being used.
+ uint64_t mapType = 0;
+ if (auto mapOp = mlir::dyn_cast<mlir::omp::MapInfoOp>(
+ memberMaps[idx].getDefiningOp()))
+ mapType = mapOp.getMapType().value_or(0);
+
+ // create parent to emplace and bind members
+ auto origSymbol = converter.getSymbolAddress(*sym);
+ mlir::Value mapOp = createMapInfoOp(
+ converter.getFirOpBuilder(),
+ converter.getFirOpBuilder().getUnknownLoc(), origSymbol,
+ mlir::Value(), sym->name().ToString(), {}, {memberMaps[idx]},
+ mlir::ArrayAttr::get(converter.getFirOpBuilder().getContext(),
+ memberPlacementIndices[idx]),
+ mapType, mlir::omp::VariableCaptureKind::ByRef, origSymbol.getType(),
+ true);
+
+ mapOperands.push_back(mapOp);
+ if (mapSymTypes)
+ mapSymTypes->push_back(mapOp.getType());
+ if (mapSymLocs)
+ mapSymLocs->push_back(mapOp.getLoc());
+ if (mapSymbols)
+ mapSymbols->push_back(sym);
+ }
+ }
+}
+
bool ClauseProcessor::processMap(
mlir::Location currentLocation, const llvm::omp::Directive &directive,
Fortran::semantics::SemanticsContext &semanticsContext,
@@ -1859,7 +1983,13 @@ bool ClauseProcessor::processMap(
llvm::SmallVectorImpl<const Fortran::semantics::Symbol *> *mapSymbols)
const {
fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
- return findRepeatableClause<ClauseTy::Map>(
+
+ llvm::SmallVector<mlir::Value> memberMaps;
+ llvm::SmallVector<mlir::Attribute> memberPlacementIndices;
+ llvm::SmallVector<const Fortran::semantics::Symbol *> memberParentSyms,
+ mapSyms;
+
+ bool clauseFound = findRepeatableClause<ClauseTy::Map>(
[&](const ClauseTy::Map *mapClause,
const Fortran::parser::CharBlock &source) {
mlir::Location clauseLocation = converter.genLocation(source);
@@ -1906,8 +2036,22 @@ bool ClauseProcessor::processMap(
for (const Fortran::parser::OmpObject &ompObject :
std::get<Fortran::parser::OmpObjectList>(mapClause->v.t).v) {
+ llvm::omp::OpenMPOffloadMappingFlags objectsMapTypeBits = mapTypeBits;
+ checkAndApplyDeclTargetMapFlags(converter, objectsMapTypeBits,
+ getOmpObjectSymbol(ompObject));
+
llvm::SmallVector<mlir::Value> bounds;
std::stringstream asFortran;
+ const Fortran::semantics::Symbol *parentSym = nullptr;
+
+ if (getOmpObjectSymbol(ompObject)->owner().IsDerivedType()) {
+ memberPlacementIndices.push_back(
+ firOpBuilder.getI64IntegerAttr(findComponenetMemberPlacement(
+ getOmpObjectSymbol(ompObject)->owner().symbol(),
+ getOmpObjectSymbol(ompObject))));
+ parentSym = getOmpObjParentSymbol(ompObject);
+ memberParentSyms.push_back(parentSym);
+ }
Fortran::lower::AddrAndBoundsInfo info =
Fortran::lower::gatherDataOperandAddrAndBounds<
@@ -1927,22 +2071,33 @@ bool ClauseProcessor::processMap(
// types to optimise
mlir::Value mapOp = createMapInfoOp(
firOpBuilder, clauseLocation, symAddr, mlir::Value{},
- asFortran.str(), bounds, {},
+ asFortran.str(), bounds, {}, mlir::ArrayAttr{},
static_cast<
std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
- mapTypeBits),
+ objectsMapTypeBits),
mlir::omp::VariableCaptureKind::ByRef, symAddr.getType());
- mapOperands.push_back(mapOp);
- if (mapSymTypes)
- mapSymTypes->push_back(symAddr.getType());
- if (mapSymLocs)
- mapSymLocs->push_back(symAddr.getLoc());
-
- if (mapSymbols)
- mapSymbols->push_back(getOmpObjectSymbol(ompObject));
+ if (parentSym) {
+ memberMaps.push_back(mapOp);
+ } else {
+ mapOperands.push_back(mapOp);
+ mapSyms.push_back(getOmpObjectSymbol(ompObject));
+ if (mapSymTypes)
+ mapSymTypes->push_back(symAddr.getType());
+ if (mapSymLocs)
+ mapSymLocs->push_back(symAddr.getLoc());
+ }
}
});
+
+ insertChildMapInfoIntoParent(converter, memberParentSyms, memberMaps,
+ memberPlacementIndices, mapOperands, mapSymTypes,
+ mapSymLocs, &mapSyms);
+
+ if (mapSymbols)
+ *mapSymbols = mapSyms;
+
+ return clauseFound;
}
bool ClauseProcessor::processReduction(
@@ -2021,7 +2176,12 @@ bool ClauseProcessor::processMotionClauses(
Fortran::semantics::SemanticsContext &semanticsContext,
Fortran::lower::StatementContext &stmtCtx,
llvm::SmallVectorImpl<mlir::Value> &mapOperands) {
- return findRepeatableClause<T>(
+ llvm::SmallVector<mlir::Value> memberMaps;
+ llvm::SmallVector<mlir::Attribute> memberPlacementIndices;
+ llvm::SmallVector<const Fortran::semantics::Symbol *> memberParentSyms,
+ mapSymbols;
+
+ bool clauseFound = findRepeatableClause<T>(
[&](const T *motionClause, const Fortran::parser::CharBlock &source) {
mlir::Location clauseLocation = converter.genLocation(source);
fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
@@ -2036,8 +2196,23 @@ bool ClauseProcessor::processMotionClauses(
: llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_FROM;
for (const Fortran::parser::OmpObject &ompObject : motionClause->v.v) {
+ llvm::omp::OpenMPOffloadMappingFlags objectsMapTypeBits = mapTypeBits;
+ checkAndApplyDeclTargetMapFlags(converter, objectsMapTypeBits,
+ getOmpObjectSymbol(ompObject));
+
llvm::SmallVector<mlir::Value> bounds;
std::stringstream asFortran;
+ const Fortran::semantics::Symbol *parentSym = nullptr;
+
+ if (getOmpObjectSymbol(ompObject)->owner().IsDerivedType()) {
+ memberPlacementIndices.push_back(
+ firOpBuilder.getI64IntegerAttr(findComponenetMemberPlacement(
+ getOmpObjectSymbol(ompObject)->owner().symbol(),
+ getOmpObjectSymbol(ompObject))));
+ parentSym = getOmpObjParentSymbol(ompObject);
+ memberParentSyms.push_back(parentSym);
+ }
+
Fortran::lower::AddrAndBoundsInfo info =
Fortran::lower::gatherDataOperandAddrAndBounds<
Fortran::parser::OmpObject, mlir::omp::DataBoundsOp,
@@ -2056,15 +2231,25 @@ bool ClauseProcessor::processMotionClauses(
// types to optimise
mlir::Value mapOp = createMapInfoOp(
firOpBuilder, clauseLocation, symAddr, mlir::Value{},
- asFortran.str(), bounds, {},
+ asFortran.str(), bounds, {}, mlir::ArrayAttr{},
static_cast<
std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
- mapTypeBits),
+ objectsMapTypeBits),
mlir::omp::VariableCaptureKind::ByRef, symAddr.getType());
- mapOperands.push_back(mapOp);
+ if (parentSym) {
+ memberMaps.push_back(mapOp);
+ } else {
+ mapOperands.push_back(mapOp);
+ mapSymbols.push_back(getOmpObjectSymbol(ompObject));
+ }
}
});
+
+ insertChildMapInfoIntoParent(converter, memberParentSyms, memberMaps,
+ memberPlacementIndices, mapOperands, nullptr,
+ nullptr, &mapSymbols);
+ return clauseFound;
}
template <typename... Ts>
@@ -2882,7 +3067,7 @@ static void genBodyOfTargetOp(
firOpBuilder.setInsertionPoint(targetOp);
mlir::Value mapOp = createMapInfoOp(
firOpBuilder, copyVal.getLoc(), copyVal, mlir::Value{}, name.str(),
- bounds, llvm::SmallVector<mlir::Value>{},
+ bounds, llvm::SmallVector<mlir::Value>{}, mlir::ArrayAttr{},
static_cast<
std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_IMPLICIT),
@@ -3018,7 +3203,7 @@ genTargetOp(Fortran::lower::AbstractConverter &converter,
mlir::Value mapOp = createMapInfoOp(
converter.getFirOpBuilder(), baseOp.getLoc(), baseOp, mlir::Value{},
- name.str(), bounds, {},
+ name.str(), bounds, {}, mlir::ArrayAttr{},
static_cast<
std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
mapFlag),
diff --git a/flang/lib/Optimizer/Transforms/CMakeLists.txt b/flang/lib/Optimizer/Transforms/CMakeLists.txt
index ba2e267996150e..ce5ce3ed1bc48d 100644
--- a/flang/lib/Optimizer/Transforms/CMakeLists.txt
+++ b/flang/lib/Optimizer/Transforms/CMakeLists.txt
@@ -17,7 +17,7 @@ add_flang_library(FIRTransforms
AddDebugFoundation.cpp
PolymorphicOpConversion.cpp
LoopVersioning.cpp
- OMPDescriptorMapInfoGen.cpp
+ OMPMapInfoFinalization.cpp
OMPFunctionFiltering.cpp
OMPMarkDeclareTarget.cpp
VScaleAttr.cpp
diff --git a/flang/lib/Optimizer/Transforms/OMPDescriptorMapInfoGen.cpp b/flang/lib/Optimizer/Transforms/OMPDescriptorMapInfoGen.cpp
deleted file mode 100644
index 6ffcf0746c76fc..00000000000000
--- a/flang/lib/Optimizer/Transforms/OMPDescriptorMapInfoGen.cpp
+++ /dev/null
@@ -1,168 +0,0 @@
-//===- OMPDescriptorMapInfoGen.cpp
-//---------------------------------------------------===//
-//
-// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-//
-//===----------------------------------------------------------------------===//
-
-//===----------------------------------------------------------------------===//
-/// \file
-/// An OpenMP dialect related pass for FIR/HLFIR which expands MapInfoOp's
-/// containing descriptor related types (fir::BoxType's) into multiple
-/// MapInfoOp's containing the parent descriptor and pointer member components
-/// for individual mapping, treating the descriptor type as a record type for
-/// later lowering in the OpenMP dialect.
-//===----------------------------------------------------------------------===//
-
-#include "flang/Optimizer/Builder/FIRBuilder.h"
-#include "flang/Optimizer/Dialect/FIRType.h"
-#include "flang/Optimizer/Dialect/Support/KindMapping.h"
-#include "flang/Optimizer/Transforms/Passes.h"
-#include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/Dialect/OpenMP/OpenMPDialect.h"
-#include "mlir/IR/BuiltinDialect.h"
-#include "mlir/IR/BuiltinOps.h"
-#include "mlir/IR/Operation.h"
-#include "mlir/IR/SymbolTable.h"
-#include "mlir/Pass/Pass.h"
-#include "mlir/Support/LLVM.h"
-#include "llvm/ADT/SmallPtrSet.h"
-#include <iterator>
-
-namespace fir {
-#define GEN_PASS_DEF_OMPDESCRIPTORMAPINFOGENPASS
-#include "flang/Optimizer/Transforms/Passes.h.inc"
-} // namespace fir
-
-namespace {
-class OMPDescriptorMapInfoGenPass
- : public fir::impl::OMPDescriptorMapInfoGenPassBase<
- OMPDescriptorMapInfoGenPass> {
-
- void genDescriptorMemberMaps(mlir::omp::MapInfoOp op,
- fir::FirOpBuilder &builder,
- mlir::Operation *target) {
- mlir::Location loc = builder.getUnknownLoc();
- mlir::Value descriptor = op.getVarPtr();
-
- // If we enter this function, but the mapped type itself is not the
- // descriptor, then it's likely the address of the descriptor so we
- // must retrieve the descriptor SSA.
- if (!fir::isTypeWithDescriptor(op.getVarType())) {
- if (auto addrOp = mlir::dyn_cast_if_present<fir::BoxAddrOp>(
- op.getVarPtr().getDefiningOp())) {
- descriptor = addrOp.getVal();
- }
- }
-
- // The fir::BoxOffsetOp only works with !fir.ref<!fir.box<...>> types, as
- // allowing it to access non-reference box operations can cause some
- // problematic SSA IR. However, in the case of assumed shape's the type
- // is not a !fir.ref, in these cases to retrieve the appropriate
- // !fir.ref<!fir.box<...>> to access the data we need to map we must
- // perform an alloca and then store to it and retrieve the data from the new
- // alloca.
- if (mlir::isa<fir::BaseBoxType>(descriptor.getType())) {
- mlir::OpBuilder::InsertPoint insPt = builder.saveInsertionPoint();
- builder.setInsertionPointToStart(builder.getAllocaBlock());
- auto alloca = builder.create<fir::AllocaOp>(loc, descriptor.getType());
- builder.restoreInsertionPoint(insPt);
- builder.create<fir::StoreOp>(loc, descriptor, alloca);
- descriptor = alloca;
- }
-
- mlir::Value baseAddrAddr = builder.create<fir::BoxOffsetOp>(
- loc, descriptor, fir::BoxFieldAttr::base_addr);
-
- // Member of the descriptor pointing at the allocated data
- mlir::Value baseAddr = builder.create<mlir::omp::MapInfoOp>(
- loc, baseAddrAddr.getType(), descriptor,
- llvm::cast<mlir::omp::PointerLikeType>(
- fir::unwrapRefType(baseAddrAddr.getType()))
- .getElementType(),
- baseAddrAddr, mlir::SmallVector<mlir::Value>{}, op.getBounds(),
- builder.getIntegerAttr(builder.getIntegerType(64, false),
- op.getMapType().value()),
- builder.getAttr<mlir::omp::VariableCaptureKindAttr>(
- mlir::omp::VariableCaptureKind::ByRef),
- builder.getStringAttr("") /*name*/);
-
- // TODO: map the addendum segment of the descriptor, similarly to the
- // above base address/data pointer member.
-
- if (auto mapClauseOwner =
- llvm::dyn_cast<mlir::omp::MapClauseOwningOpInterface>(target)) {
- llvm::SmallVector<mlir::Value> newMapOps;
- mlir::OperandRange mapOperandsArr = mapClauseOwner.getMapOperands();
-
- for (size_t i = 0; i < mapOperandsArr.size(); ++i) {
- if (mapOperandsArr[i] == op) {
- // Push new implicit maps generated for the descriptor.
- newMapOps.push_back(baseAddr);
-
- // for TargetOp's which have IsolatedFromAbove we must align the
- // new additional map operand with an appropriate BlockArgument,
- // as the printing and later processing currently requires a 1:1
- // mapping of BlockArgs to MapInfoOp's at the same placement in
- // each array (BlockArgs and MapOperands).
- if (auto targetOp = llvm::dyn_cast<mlir::omp::TargetOp>(target))
- targetOp.getRegion().insertArgument(i, baseAddr.getType(), loc);
- }
- newMapOps.push_back(mapOperandsArr[i]);
- }
- mapClauseOwner.getMapOperandsMutable().assign(newMapOps);
- }
-
- mlir::Value newDescParentMapOp = builder.create<mlir::omp::MapInfoOp>(
- op->getLoc(), op.getResult().getType(), descriptor,
- fir::unwrapRefType(descriptor.getType()), mlir::Value{},
- mlir::SmallVector<mlir::Value>{baseAddr},
- mlir::SmallVector<mlir::Value>{},
- builder.getIntegerAttr(builder.getIntegerType(64, false),
- op.getMapType().value()),
- op.getMapCaptureTypeAttr(), op.getNameAttr());
- op.replaceAllUsesWith(newDescParentMapOp);
- op->erase();
- }
-
- // This pass executes on mlir::ModuleOp's finding omp::MapInfoOp's containing
- // descriptor based types (allocatables, pointers, assumed shape etc.) and
- // expanding them into multiple omp::MapInfoOp's for each pointer member
- // contained within the descriptor.
- void runOnOperation() override {
- mlir::func::FuncOp func = getOperation();
- mlir::ModuleOp module = func->getParentOfType<mlir::ModuleOp>();
- fir::KindMapping kindMap = fir::getKindMapping(module);
- fir::FirOpBuilder builder{module, std::move(kindMap)};
-
- func->walk([&](mlir::omp::MapInfoOp op) {
- if (fir::isTypeWithDescriptor(op.getVarType()) ||
- mlir::isa_and_present<fir::BoxAddrOp>(
- op.getVarPtr().getDefiningOp())) {
- builder.setInsertionPoint(op);
- // TODO: Currently only supports a single user for the MapInfoOp, this
- // is fine for the moment as the Fortran Frontend will generate a
- // new MapInfoOp per Target operation for the moment. However, when/if
- // we optimise/cleanup the IR, it likely isn't too difficult to
- // extend this function, it would require some modification to create a
- // single new MapInfoOp per new MapInfoOp generated and share it across
- // all users appropriately, making sure to only add a single member link
- // per new generation for the original originating descriptor MapInfoOp.
- assert(llvm::hasSingleElement(op->getUsers()) &&
- "OMPDescriptorMapInfoGen currently only supports single users "
- "of a MapInfoOp");
- genDescriptorMemberMaps(op, builder, *op->getUsers().begin());
- }
- });
- }
-};
-
-} // namespace
-
-namespace fir {
-std::unique_ptr<mlir::Pass> createOMPDescriptorMapInfoGenPass() {
- return std::make_unique<OMPDescriptorMapInfoGenPass>();
-}
-} // namespace fir
diff --git a/flang/lib/Optimizer/Transforms/OMPMapInfoFinalization.cpp b/flang/lib/Optimizer/Transforms/OMPMapInfoFinalization.cpp
new file mode 100644
index 00000000000000..c3c32078e918c1
--- /dev/null
+++ b/flang/lib/Optimizer/Transforms/OMPMapInfoFinalization.cpp
@@ -0,0 +1,262 @@
+//===- OMPMapInfoFinalization.cpp
+//---------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+//===----------------------------------------------------------------------===//
+/// \file
+/// An OpenMP dialect related pass for FIR/HLFIR which performs some
+/// pre-processing of MapInfoOp's after the module has been lowered to
+/// finalize them.
+///
+/// For example, it expands MapInfoOp's containing descriptor related
+/// types (fir::BoxType's) into multiple MapInfoOp's containing the parent
+/// descriptor and pointer member components for individual mapping,
+/// treating the descriptor type as a record type for later lowering in the
+/// OpenMP dialect.
+///
+/// The pass also adds MapInfoOp's that are members of a parent object but are
+/// not directly used in the body of a target region to it's BlockArgument list
+/// to maintain consistency across all MapInfoOp's tied to a region directly or
+/// indirectly via an parent object.
+//===----------------------------------------------------------------------===//
+
+#include "flang/Optimizer/Builder/FIRBuilder.h"
+#include "flang/Optimizer/Dialect/FIRType.h"
+#include "flang/Optimizer/Dialect/Support/KindMapping.h"
+#include "flang/Optimizer/Transforms/Passes.h"
+#include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Dialect/OpenMP/OpenMPDialect.h"
+#include "mlir/IR/BuiltinDialect.h"
+#include "mlir/IR/BuiltinOps.h"
+#include "mlir/IR/Operation.h"
+#include "mlir/IR/SymbolTable.h"
+#include "mlir/Pass/Pass.h"
+#include "mlir/Support/LLVM.h"
+#include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/Frontend/OpenMP/OMPConstants.h"
+#include <iterator>
+
+namespace fir {
+#define GEN_PASS_DEF_OMPMAPINFOFINALIZATIONPASS
+#include "flang/Optimizer/Transforms/Passes.h.inc"
+} // namespace fir
+
+namespace {
+class OMPMapInfoFinalizationPass
+ : public fir::impl::OMPMapInfoFinalizationPassBase<
+ OMPMapInfoFinalizationPass> {
+
+ void genDescriptorMemberMaps(mlir::omp::MapInfoOp op,
+ fir::FirOpBuilder &builder,
+ mlir::Operation *target) {
+ mlir::Location loc = builder.getUnknownLoc();
+ mlir::Value descriptor = op.getVarPtr();
+
+ // If we enter this function, but the mapped type itself is not the
+ // descriptor, then it's likely the address of the descriptor so we
+ // must retrieve the descriptor SSA.
+ if (!fir::isTypeWithDescriptor(op.getVarType())) {
+ if (auto addrOp = mlir::dyn_cast_if_present<fir::BoxAddrOp>(
+ op.getVarPtr().getDefiningOp())) {
+ descriptor = addrOp.getVal();
+ }
+ }
+
+ // The fir::BoxOffsetOp only works with !fir.ref<!fir.box<...>> types, as
+ // allowing it to access non-reference box operations can cause some
+ // problematic SSA IR. However, in the case of assumed shape's the type
+ // is not a !fir.ref, in these cases to retrieve the appropriate
+ // !fir.ref<!fir.box<...>> to access the data we need to map we must
+ // perform an alloca and then store to it and retrieve the data from the new
+ // alloca.
+ if (mlir::isa<fir::BaseBoxType>(descriptor.getType())) {
+ mlir::OpBuilder::InsertPoint insPt = builder.saveInsertionPoint();
+ builder.setInsertionPointToStart(builder.getAllocaBlock());
+ auto alloca = builder.create<fir::AllocaOp>(loc, descriptor.getType());
+ builder.restoreInsertionPoint(insPt);
+ builder.create<fir::StoreOp>(loc, descriptor, alloca);
+ descriptor = alloca;
+ }
+
+ mlir::Value baseAddrAddr = builder.create<fir::BoxOffsetOp>(
+ loc, descriptor, fir::BoxFieldAttr::base_addr);
+
+ llvm::omp::OpenMPOffloadMappingFlags baseAddrMapFlag =
+ llvm::omp::OpenMPOffloadMappingFlags(op.getMapType().value());
+ baseAddrMapFlag |=
+ llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
+
+ // Member of the descriptor pointing at the allocated data
+ mlir::Value baseAddr = builder.create<mlir::omp::MapInfoOp>(
+ loc, baseAddrAddr.getType(), descriptor,
+ mlir::TypeAttr::get(llvm::cast<mlir::omp::PointerLikeType>(
+ fir::unwrapRefType(baseAddrAddr.getType()))
+ .getElementType()),
+ baseAddrAddr, mlir::SmallVector<mlir::Value>{}, mlir::ArrayAttr{},
+ op.getBounds(),
+ builder.getIntegerAttr(
+ builder.getIntegerType(64, false),
+ static_cast<
+ std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
+ baseAddrMapFlag)),
+ builder.getAttr<mlir::omp::VariableCaptureKindAttr>(
+ mlir::omp::VariableCaptureKind::ByRef),
+ builder.getStringAttr("") /*name*/,
+ builder.getBoolAttr(false) /*partial_map*/);
+
+ // TODO: map the addendum segment of the descriptor, similarly to the
+ // above base address/data pointer member.
+
+ if (auto mapClauseOwner =
+ llvm::dyn_cast<mlir::omp::MapClauseOwningOpInterface>(target)) {
+ llvm::SmallVector<mlir::Value> newMapOps;
+ mlir::OperandRange mapOperandsArr = mapClauseOwner.getMapOperands();
+
+ for (size_t i = 0; i < mapOperandsArr.size(); ++i) {
+ if (mapOperandsArr[i] == op) {
+ // Push new implicit maps generated for the descriptor.
+ newMapOps.push_back(baseAddr);
+
+ // for TargetOp's which have IsolatedFromAbove we must align the
+ // new additional map operand with an appropriate BlockArgument,
+ // as the printing and later processing currently requires a 1:1
+ // mapping of BlockArgs to MapInfoOp's at the same placement in
+ // each array (BlockArgs and MapOperands).
+ if (auto targetOp = llvm::dyn_cast<mlir::omp::TargetOp>(target))
+ targetOp.getRegion().insertArgument(i, baseAddr.getType(), loc);
+ }
+ newMapOps.push_back(mapOperandsArr[i]);
+ }
+ mapClauseOwner.getMapOperandsMutable().assign(newMapOps);
+ }
+
+ mlir::Value newDescParentMapOp = builder.create<mlir::omp::MapInfoOp>(
+ op->getLoc(), op.getResult().getType(), descriptor,
+ mlir::TypeAttr::get(fir::unwrapRefType(descriptor.getType())),
+ mlir::Value{}, mlir::SmallVector<mlir::Value>{baseAddr},
+ mlir::ArrayAttr::get(builder.getContext(),
+ builder.getI64IntegerAttr(0)) /*members_index*/,
+ mlir::SmallVector<mlir::Value>{},
+ builder.getIntegerAttr(builder.getIntegerType(64, false),
+ op.getMapType().value()),
+ op.getMapCaptureTypeAttr(), op.getNameAttr(), op.getPartialMapAttr());
+ op.replaceAllUsesWith(newDescParentMapOp);
+ op->erase();
+ }
+
+ // For all mapped record members not directly used in the target region
+ // we add them to the block arguments in front of their parent and place
+ // them into the map operands list for consistency.
+ //
+ // These indirect uses (via accesses to their parent) will still be
+ // mapped individually in most cases, and a parent mapping doesn't
+ // guarantee the parent will be mapped in its totality, partial
+ // mapping is common.
+ //
+ // For example:
+ // map(tofrom: x%y)
+ //
+ // Will generate a mapping for "x" (the parent) and "y" (the member),
+ // the parent "x" will not be mapped, only the member "y" will,
+ // however, we must have the parent as a BlockArg and MapOperand in
+ // these cases to maintain the correct uses within the region and
+ // it helps to track that the member is part of a larger object.
+ //
+ // In the case of:
+ // map(tofrom: x%y, x%z)
+ //
+ // The parent member becomes more critical, as we perform a partial
+ // structure mapping, where we link the mapping of x and y together
+ // via the parent (at a kernel argument level in LLVM IR not just
+ // MLIR, important to maintain similarity to Clang and for the runtime
+ // to do the correct thing), however, we still do not map the structure
+ // in its totality, we do however, generate an un-sized "binding"
+ // map entry for it.
+ //
+ // In the case of:
+ // map(tofrom: x, x%y, x%z)
+ //
+ // We do actually map the entirety of "x", so the explicit
+ // mapping of x%y, x%z becomes unneccesary. It also doesn't
+ // quite make sense to write this from a Fortran OpenMP
+ // perspective (although it is legal), as even if the members were
+ // allocatables or pointers, we are mandated by the specification
+ // to map these (and any recursive components) in their entirety,
+ // which is different to the C++ equivelant, which requires
+ // explicit mapping of these segments.
+ void addImplicitMembersToTarget(mlir::omp::MapInfoOp op,
+ fir::FirOpBuilder &builder,
+ mlir::Operation *target) {
+ if (auto mapClauseOwner =
+ llvm::dyn_cast<mlir::omp::MapClauseOwningOpInterface>(target)) {
+ llvm::SmallVector<mlir::Value> newMapOps;
+ mlir::OperandRange mapOperandsArr = mapClauseOwner.getMapOperands();
+
+ for (size_t i = 0; i < mapOperandsArr.size(); ++i) {
+ if (mapOperandsArr[i] == op) {
+ // Push member maps
+ for (auto member : op.getMembers()) {
+ newMapOps.push_back(member);
+ // for TargetOp's which have IsolatedFromAbove we must align the
+ // new additional map operand with an appropriate BlockArgument,
+ // as the printing and later processing currently requires a 1:1
+ // mapping of BlockArgs to MapInfoOp's at the same placement in
+ // each array (BlockArgs and MapOperands).
+ if (auto targetOp = llvm::dyn_cast<mlir::omp::TargetOp>(target))
+ targetOp.getRegion().insertArgument(i, member.getType(),
+ builder.getUnknownLoc());
+ }
+ }
+ newMapOps.push_back(mapOperandsArr[i]);
+ }
+ mapClauseOwner.getMapOperandsMutable().assign(newMapOps);
+ }
+ }
+
+ // This pass executes on mlir::ModuleOp's finding omp::MapInfoOp's containing
+ // descriptor based types (allocatables, pointers, assumed shape etc.) and
+ // expanding them into multiple omp::MapInfoOp's for each pointer member
+ // contained within the descriptor.
+ void runOnOperation() override {
+ mlir::func::FuncOp func = getOperation();
+ mlir::ModuleOp module = func->getParentOfType<mlir::ModuleOp>();
+ fir::KindMapping kindMap = fir::getKindMapping(module);
+ fir::FirOpBuilder builder{module, std::move(kindMap)};
+
+ func->walk([&](mlir::omp::MapInfoOp op) {
+ // TODO: Currently only supports a single user for the MapInfoOp, this
+ // is fine for the moment as the Fortran Frontend will generate a
+ // new MapInfoOp per Target operation for the moment. However, when/if
+ // we optimise/cleanup the IR, it likely isn't too difficult to
+ // extend this function, it would require some modification to create a
+ // single new MapInfoOp per new MapInfoOp generated and share it across
+ // all users appropriately, making sure to only add a single member link
+ // per new generation for the original originating descriptor MapInfoOp.
+ assert(llvm::hasSingleElement(op->getUsers()) &&
+ "OMPMapInfoFinalization currently only supports single users "
+ "of a MapInfoOp");
+
+ if (!op.getMembers().empty()) {
+ addImplicitMembersToTarget(op, builder, *op->getUsers().begin());
+ } else if (fir::isTypeWithDescriptor(op.getVarType()) ||
+ mlir::isa_and_present<fir::BoxAddrOp>(
+ op.getVarPtr().getDefiningOp())) {
+ builder.setInsertionPoint(op);
+ genDescriptorMemberMaps(op, builder, *op->getUsers().begin());
+ }
+ });
+ }
+};
+
+} // namespace
+
+namespace fir {
+std::unique_ptr<mlir::Pass> createOMPMapInfoFinalizationPass() {
+ return std::make_unique<OMPMapInfoFinalizationPass>();
+}
+} // namespace fir
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 beb399ec3ac05e..9648eda2255464 100644
--- a/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
+++ b/flang/test/Fir/convert-to-llvm-openmp-and-fir.fir
@@ -902,13 +902,40 @@ func.func @omp_critical_() {
func.func @omp_map_info_descriptor_type_conversion(%arg0 : !fir.ref<!fir.box<!fir.heap<i32>>>) {
// CHECK: %[[GEP:.*]] = llvm.getelementptr %[[ARG_0]][0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>
%0 = fir.box_offset %arg0 base_addr : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> !fir.llvm_ptr<!fir.ref<i32>>
- // CHECK: %[[MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[GEP]] : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
- %1 = omp.map_info var_ptr(%0 : !fir.llvm_ptr<!fir.ref<i32>>, i32) map_clauses(tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
- // CHECK: %[[DESC_MAP:.*]] = omp.map_info var_ptr(%[[ARG_0]] : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(always, delete) capture(ByRef) members(%[[MEMBER_MAP]] : !llvm.ptr) -> !llvm.ptr {name = ""}
- %2 = omp.map_info var_ptr(%arg0 : !fir.ref<!fir.box<!fir.heap<i32>>>, !fir.box<!fir.heap<i32>>) map_clauses(always, delete) capture(ByRef) members(%1 : !fir.llvm_ptr<!fir.ref<i32>>) -> !fir.ref<!fir.box<!fir.heap<i32>>> {name = ""}
+ // CHECK: %[[MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[GEP]] : !llvm.ptr, i32) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+ %1 = omp.map_info var_ptr(%0 : !fir.llvm_ptr<!fir.ref<i32>>, i32) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
+ // CHECK: %[[DESC_MAP:.*]] = omp.map_info var_ptr(%[[ARG_0]] : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(always, delete) capture(ByRef) members(%[[MEMBER_MAP]] : !llvm.ptr : [0]) -> !llvm.ptr {name = ""}
+ %2 = omp.map_info var_ptr(%arg0 : !fir.ref<!fir.box<!fir.heap<i32>>>, !fir.box<!fir.heap<i32>>) map_clauses(always, delete) capture(ByRef) members(%1 : !fir.llvm_ptr<!fir.ref<i32>> : [0]) -> !fir.ref<!fir.box<!fir.heap<i32>>> {name = ""}
// CHECK: omp.target_exit_data map_entries(%[[DESC_MAP]] : !llvm.ptr)
omp.target_exit_data map_entries(%2 : !fir.ref<!fir.box<!fir.heap<i32>>>)
return
}
// -----
+
+// CHECK-LABEL: llvm.func @omp_map_info_derived_type_explicit_member_conversion
+// CHECK-SAME: %[[ARG_0:.*]]: !llvm.ptr)
+
+func.func @omp_map_info_derived_type_explicit_member_conversion(%arg0 : !fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+ // CHECK: %[[GEP:.*]] = llvm.getelementptr %[[ARG_0]][0, 2] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<"_QFderived_type", (f32, array<10 x i32>, i32)>
+ %0 = fir.field_index int, !fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>
+ %1 = fir.coordinate_of %arg0, %0 : (!fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.field) -> !fir.ref<i32>
+ // CHECK: %[[MAP_MEMBER_1:.*]] = omp.map_info var_ptr(%[[GEP]] : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = "dtype%int"}
+ %2 = omp.map_info var_ptr(%1 : !fir.ref<i32>, i32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "dtype%int"}
+ // CHECK: %[[GEP_2:.*]] = llvm.getelementptr %[[ARG_0]][0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<"_QFderived_type", (f32, array<10 x i32>, i32)>
+ %3 = fir.field_index real, !fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>
+ %4 = fir.coordinate_of %arg0, %3 : (!fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.field) -> !fir.ref<f32>
+ // CHECK: %[[MAP_MEMBER_2:.*]] = omp.map_info var_ptr(%[[GEP_2]] : !llvm.ptr, f32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = "dtype%real"}
+ %5 = omp.map_info var_ptr(%4 : !fir.ref<f32>, f32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<f32> {name = "dtype%real"}
+ // CHECK: %[[MAP_PARENT:.*]] = omp.map_info var_ptr(%[[ARG_0]] : !llvm.ptr, !llvm.struct<"_QFderived_type", (f32, array<10 x i32>, i32)>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBER_1]], %[[MAP_MEMBER_2]] : !llvm.ptr, !llvm.ptr : [2, 0]) -> !llvm.ptr {name = "dtype", partial_map = true}
+ %6 = omp.map_info var_ptr(%arg0 : !fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(tofrom) capture(ByRef) members(%2, %5 : !fir.ref<i32>, !fir.ref<f32> : [2, 0]) -> !fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "dtype", partial_map = true}
+ // CHECK: omp.target map_entries(%[[MAP_MEMBER_1]] -> %arg1, %[[MAP_MEMBER_2]] -> %arg2, %[[MAP_PARENT]] -> %arg3 : !llvm.ptr, !llvm.ptr, !llvm.ptr) {
+ // CHECK: ^bb0(%arg1: !llvm.ptr, %arg2: !llvm.ptr, %arg3: !llvm.ptr):
+ omp.target map_entries(%2 -> %arg1, %5 -> %arg2, %6 -> %arg3 : !fir.ref<i32>, !fir.ref<f32>, !fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+ ^bb0(%arg1: !fir.ref<f32>, %arg2: !fir.ref<i32>, %arg3: !fir.ref<!fir.type<_QFderived_type{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+ omp.terminator
+ }
+ return
+}
+
+// -----
diff --git a/flang/test/Integration/OpenMP/map-types-and-sizes.f90 b/flang/test/Integration/OpenMP/map-types-and-sizes.f90
index 283ac227f34364..d950d724056f0c 100644
--- a/flang/test/Integration/OpenMP/map-types-and-sizes.f90
+++ b/flang/test/Integration/OpenMP/map-types-and-sizes.f90
@@ -69,6 +69,81 @@ subroutine mapType_allocatable_explicit
!$omp end target
deallocate(a)
end subroutine mapType_allocatable_explicit
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 48]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 547]
+subroutine mapType_derived_implicit
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_implicit
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 48]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 35]
+subroutine mapType_derived_explicit
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr)
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_explicit
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [2 x i64] [i64 0, i64 40]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [2 x i64] [i64 32, i64 281474976710659]
+subroutine mapType_derived_explicit_single_member
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%array)
+ scalar_arr%array(1) = 1
+ !$omp end target
+end subroutine mapType_derived_explicit_single_member
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [3 x i64] [i64 0, i64 4, i64 4]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710659]
+subroutine mapType_derived_explicit_multiple_members
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%int, scalar_arr%real)
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_explicit_multiple_members
+
+!CHECK: @.offload_sizes{{.*}} = private unnamed_addr constant [2 x i64] [i64 0, i64 16]
+!CHECK: @.offload_maptypes{{.*}} = private unnamed_addr constant [2 x i64] [i64 32, i64 281474976710659]
+subroutine mapType_derived_explicit_member_with_bounds
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%array(2:5))
+ scalar_arr%array(3) = 3
+ !$omp end target
+end subroutine mapType_derived_explicit_member_with_bounds
!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]
@@ -100,7 +175,6 @@ end subroutine mapType_char
!CHECK: %[[OFFLOAD_SIZE_ARR:.*]] = getelementptr inbounds [3 x i64], ptr %.offload_sizes, i32 0, i32 0
!CHECK: store i64 %[[DIV]], ptr %[[OFFLOAD_SIZE_ARR]], align 8
-
!CHECK-LABEL: define {{.*}} @{{.*}}maptype_allocatable_explicit_{{.*}}
!CHECK: %[[ALLOCA:.*]] = alloca { ptr, i64, i32, i8, i8, i8, i8 }, i64 1, align 8
!CHECK: %[[ALLOCA_GEP:.*]] = getelementptr { ptr, i64, i32, i8, i8, i8, i8 }, ptr %[[ALLOCA]], i32 1
@@ -110,3 +184,81 @@ end subroutine mapType_char
!CHECK: %[[DIV:.*]] = sdiv exact i64 %[[SIZE_DIFF]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
!CHECK: %[[OFFLOAD_SIZE_ARR:.*]] = getelementptr inbounds [3 x i64], ptr %.offload_sizes, i32 0, i32 0
!CHECK: store i64 %[[DIV]], ptr %[[OFFLOAD_SIZE_ARR]], align 8
+
+!CHECK-LABEL: define {{.*}} @{{.*}}maptype_derived_implicit_{{.*}}
+!CHECK: %[[ALLOCA:.*]] = alloca %_QFmaptype_derived_implicitTscalar_and_array, i64 1, align 8
+!CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [1 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR:.*]] = getelementptr inbounds [1 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[OFFLOAD_PTR_ARR]], align 8
+
+!CHECK-LABEL: define {{.*}} @{{.*}}maptype_derived_explicit_{{.*}}
+!CHECK: %[[ALLOCA:.*]] = alloca %_QFmaptype_derived_explicitTscalar_and_array, i64 1, align 8
+!CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [1 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR:.*]] = getelementptr inbounds [1 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[OFFLOAD_PTR_ARR]], align 8
+
+!CHECK-LABEL: define {{.*}} @{{.*}}maptype_derived_explicit_single_member_{{.*}}
+!CHECK: %[[ALLOCA:.*]] = alloca %_QFmaptype_derived_explicit_single_memberTscalar_and_array, i64 1, align 8
+!CHECK: %[[MEMBER_ACCESS:.*]] = getelementptr %_QFmaptype_derived_explicit_single_memberTscalar_and_array, ptr %[[ALLOCA]], i32 0, i32 1
+!CHECK: %[[ARR_OFF:.*]] = getelementptr inbounds [10 x i32], ptr %[[MEMBER_ACCESS]], i64 0, i64 0
+!CHECK: %[[ARR_END_OFF:.*]] = getelementptr [10 x i32], ptr %[[ARR_OFF]], i64 1
+!CHECK: %[[ARR_END:.*]] = ptrtoint ptr %[[ARR_END_OFF]] to i64
+!CHECK: %[[ARR_BEGIN:.*]] = ptrtoint ptr %[[ARR_OFF]] to i64
+!CHECK: %[[SIZE_DIFF:.*]] = sub i64 %[[ARR_END]], %[[ARR_BEGIN]]
+!CHECK: %[[SIZE:.*]] = sdiv exact i64 %[[SIZE_DIFF]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
+!CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+!CHECK: store ptr %[[ARR_OFF]], ptr %[[OFFLOAD_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_SIZE_ARR:.*]] = getelementptr inbounds [2 x i64], ptr %.offload_sizes, i32 0, i32 0
+!CHECK: store i64 %[[SIZE]], ptr %[[OFFLOAD_SIZE_ARR]], align 8
+!CHECK: %[[BASE_PTR_ARR_2:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 1
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_2]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR_2:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 1
+!CHECK: store ptr %[[ARR_OFF]], ptr %[[OFFLOAD_PTR_ARR_2]], align 8
+
+!CHECK-LABEL: define {{.*}} @{{.*}}maptype_derived_explicit_multiple_members_{{.*}}
+!CHECK: %[[ALLOCA:.*]] = alloca %_QFmaptype_derived_explicit_multiple_membersTscalar_and_array, i64 1, align 8
+!CHECK: %[[MEMBER_ACCESS_1:.*]] = getelementptr %_QFmaptype_derived_explicit_multiple_membersTscalar_and_array, ptr %[[ALLOCA]], i32 0, i32 2
+!CHECK: %[[MEMBER_ACCESS_2:.*]] = getelementptr %_QFmaptype_derived_explicit_multiple_membersTscalar_and_array, ptr %[[ALLOCA]], i32 0, i32 0
+!CHECK: %[[ARR_END_OFF:.*]] = getelementptr i32, ptr %[[MEMBER_ACCESS_1]], i64 1
+!CHECK: %[[ARR_END:.*]] = ptrtoint ptr %[[ARR_END_OFF]] to i64
+!CHECK: %[[FIRST_MEMBER:.*]] = ptrtoint ptr %[[MEMBER_ACCESS_2]] to i64
+!CHECK: %[[SIZE_DIFF:.*]] = sub i64 %[[ARR_END]], %[[FIRST_MEMBER]]
+!CHECK: %[[SIZE:.*]] = sdiv exact i64 %[[SIZE_DIFF]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
+!CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+!CHECK: store ptr %[[MEMBER_ACCESS_2]], ptr %[[OFFLOAD_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_SIZE_ARR:.*]] = getelementptr inbounds [3 x i64], ptr %.offload_sizes, i32 0, i32 0
+!CHECK: store i64 %[[SIZE]], ptr %[[OFFLOAD_SIZE_ARR]], align 8
+!CHECK: %[[BASE_PTR_ARR_2:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 1
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_2]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR_2:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 1
+!CHECK: store ptr %[[MEMBER_ACCESS_1]], ptr %[[OFFLOAD_PTR_ARR_2]], align 8
+!CHECK: %[[BASE_PTR_ARR_3:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 2
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_3]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR_3:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 2
+!CHECK: store ptr %[[MEMBER_ACCESS_2]], ptr %[[OFFLOAD_PTR_ARR_3]], align 8
+
+!CHECK-LABEL: define {{.*}} @{{.*}}maptype_derived_explicit_member_with_bounds_{{.*}}
+!CHECK: %[[ALLOCA:.*]] = alloca %_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array, i64 1, align 8
+!CHECK: %[[MEMBER_ACCESS:.*]] = getelementptr %_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array, ptr %[[ALLOCA]], i32 0, i32 1
+!CHECK: %[[ARR_OFF:.*]] = getelementptr inbounds [10 x i32], ptr %[[MEMBER_ACCESS]], i64 0, i64 1
+!CHECK: %[[ARR_END_OFF:.*]] = getelementptr [10 x i32], ptr %[[ARR_OFF]], i64 1
+!CHECK: %[[ARR_END:.*]] = ptrtoint ptr %[[ARR_END_OFF]] to i64
+!CHECK: %[[ARR_BEGIN:.*]] = ptrtoint ptr %[[ARR_OFF]] to i64
+!CHECK: %[[SIZE_DIFF:.*]] = sub i64 %[[ARR_END]], %[[ARR_BEGIN]]
+!CHECK: %[[SIZE:.*]] = sdiv exact i64 %[[SIZE_DIFF]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
+!CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+!CHECK: store ptr %[[ARR_OFF]], ptr %[[OFFLOAD_PTR_ARR]], align 8
+!CHECK: %[[OFFLOAD_SIZE_ARR:.*]] = getelementptr inbounds [2 x i64], ptr %.offload_sizes, i32 0, i32 0
+!CHECK: store i64 %[[SIZE]], ptr %[[OFFLOAD_SIZE_ARR]], align 8
+!CHECK: %[[BASE_PTR_ARR_2:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 1
+!CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_2]], align 8
+!CHECK: %[[OFFLOAD_PTR_ARR_2:.*]] = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 1
+!CHECK: store ptr %[[ARR_OFF]], ptr %[[OFFLOAD_PTR_ARR_2]], align 8
diff --git a/flang/test/Lower/OpenMP/FIR/array-bounds.f90 b/flang/test/Lower/OpenMP/FIR/array-bounds.f90
index 3cd284c46e727d..bd61f1eac5e02a 100644
--- a/flang/test/Lower/OpenMP/FIR/array-bounds.f90
+++ b/flang/test/Lower/OpenMP/FIR/array-bounds.f90
@@ -46,8 +46,8 @@ module assumed_array_routines
!ALL: %[[DIMS1:.*]]:3 = fir.box_dims %arg0, %[[C0_1]] : (!fir.box<!fir.array<?xi32>>, index) -> (index, index, index)
!ALL: %[[BOUNDS:.*]] = omp.bounds lower_bound(%[[C3]] : index) upper_bound(%[[C4]] : index) extent(%[[DIMS1]]#1 : index) stride(%[[DIMS0]]#2 : index) start_idx(%[[C0]] : index) {stride_in_bytes = true}
!ALL: %[[BOXADDRADDR:.*]] = fir.box_offset %0 base_addr : (!fir.ref<!fir.box<!fir.array<?xi32>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!ALL: %[[MAP_MEMBER:.*]] = omp.map_info var_ptr(%0 : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[BOXADDRADDR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!ALL: %[[MAP:.*]] = omp.map_info var_ptr(%0 : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
+!ALL: %[[MAP_MEMBER:.*]] = omp.map_info var_ptr(%0 : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[BOXADDRADDR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!ALL: %[[MAP:.*]] = omp.map_info var_ptr(%0 : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !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_MEMBER]] -> %{{.*}}, %[[MAP]] -> %{{.*}}, %[[MAP2]] -> %{{.*}} : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>, !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>) {
subroutine assumed_shape_array(arr_read_write)
diff --git a/flang/test/Lower/OpenMP/FIR/map-component-ref.f90 b/flang/test/Lower/OpenMP/FIR/map-component-ref.f90
index af8c6fa3c5e877..27d95fbe8d6d16 100644
--- a/flang/test/Lower/OpenMP/FIR/map-component-ref.f90
+++ b/flang/test/Lower/OpenMP/FIR/map-component-ref.f90
@@ -6,7 +6,7 @@
! CHECK: %[[V2:[0-9]+]] = fir.field_index a1, !fir.type<_QFfooTt0{a0:i32,a1:i32}>
! CHECK: %[[V3:[0-9]+]] = fir.coordinate_of %[[V1]], %[[V2]] : (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.field) -> !fir.ref<i32>
! CHECK: %[[V4:[0-9]+]] = omp.map_info var_ptr(%[[V3]] : !fir.ref<i32>, i32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "a%a1"}
-! CHECK: %[[V5:[0-9]+]] = omp.map_info var_ptr(%[[V1]] : !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.type<_QFfooTt0{a0:i32,a1:i32}>) map_clauses(implicit, tofrom) capture(ByRef) -> !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>> {name = "a"}
+! CHECK: %[[V5:[0-9]+]] = omp.map_info var_ptr(%[[V1]] : !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.type<_QFfooTt0{a0:i32,a1:i32}>) map_clauses(tofrom) capture(ByRef) members(%[[V4]] : !fir.ref<i32> : [1]) -> !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>> {name = "a", partial_map = true}
! CHECK: omp.target map_entries(%[[V4]] -> %arg0, %[[V5]] -> %arg1 : !fir.ref<i32>, !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>) {
! CHECK: ^bb0(%arg0: !fir.ref<i32>, %arg1: !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>):
! CHECK: %c0_i32 = arith.constant 0 : i32
@@ -19,7 +19,7 @@
subroutine foo()
implicit none
-
+
type t0
integer :: a0, a1
end type
diff --git a/flang/test/Lower/OpenMP/FIR/target.f90 b/flang/test/Lower/OpenMP/FIR/target.f90
index 06772771647de8..5f2d71fcbf3881 100644
--- a/flang/test/Lower/OpenMP/FIR/target.f90
+++ b/flang/test/Lower/OpenMP/FIR/target.f90
@@ -450,8 +450,8 @@ end subroutine omp_target_device_ptr
subroutine omp_target_device_addr
integer, pointer :: a
!CHECK: %[[VAL_0:.*]] = fir.alloca !fir.box<!fir.ptr<i32>> {bindc_name = "a", uniq_name = "_QFomp_target_device_addrEa"}
- !CHECK: %[[MAP_MEMBERS:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr({{.*}} : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
- !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBERS]] : !fir.llvm_ptr<!fir.ref<i32>>) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "a"}
+ !CHECK: %[[MAP_MEMBERS:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr({{.*}} : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
+ !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBERS]] : !fir.llvm_ptr<!fir.ref<i32>> : [0]) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "a"}
!CHECK: omp.target_data map_entries(%[[MAP_MEMBERS]], %[[MAP]] : {{.*}}) use_device_addr(%[[VAL_0]] : !fir.ref<!fir.box<!fir.ptr<i32>>>) {
!$omp target data map(tofrom: a) use_device_addr(a)
!CHECK: ^bb0(%[[VAL_1:.*]]: !fir.ref<!fir.box<!fir.ptr<i32>>>):
diff --git a/flang/test/Lower/OpenMP/allocatable-array-bounds.f90 b/flang/test/Lower/OpenMP/allocatable-array-bounds.f90
index adf74efa9b596e..f79185e7bf0c1e 100644
--- a/flang/test/Lower/OpenMP/allocatable-array-bounds.f90
+++ b/flang/test/Lower/OpenMP/allocatable-array-bounds.f90
@@ -23,8 +23,8 @@
!HOST: %[[BOX_3:.*]]:3 = fir.box_dims %[[LOAD_3]], %[[CONSTANT_3]] : (!fir.box<!fir.heap<!fir.array<?xi32>>>, index) -> (index, index, index)
!HOST: %[[BOUNDS_1:.*]] = omp.bounds lower_bound(%[[LB_1]] : index) upper_bound(%[[UB_1]] : index) extent(%[[BOX_3]]#1 : index) stride(%[[BOX_2]]#2 : index) start_idx(%[[BOX_1]]#0 : index) {stride_in_bytes = true}
!HOST: %[[VAR_PTR_PTR:.*]] = fir.box_offset %[[DECLARE_1]]#1 base_addr : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS_1]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!HOST: %[[MAP_INFO_1:.*]] = omp.map_info var_ptr(%[[DECLARE_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "sp_read(2:5)"}
+!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS_1]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!HOST: %[[MAP_INFO_1:.*]] = omp.map_info var_ptr(%[[DECLARE_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "sp_read(2:5)"}
!HOST: %[[LOAD_3:.*]] = fir.load %[[DECLARE_2]]#0 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
!HOST: %[[LOAD_4:.*]] = fir.load %[[DECLARE_2]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
@@ -41,8 +41,8 @@
!HOST: %[[BOX_5:.*]]:3 = fir.box_dims %[[LOAD_5]], %[[CONSTANT_5]] : (!fir.box<!fir.heap<!fir.array<?xi32>>>, index) -> (index, index, index)
!HOST: %[[BOUNDS_2:.*]] = omp.bounds lower_bound(%[[LB_2]] : index) upper_bound(%[[UB_2]] : index) extent(%[[BOX_5]]#1 : index) stride(%[[BOX_4]]#2 : index) start_idx(%[[BOX_3]]#0 : index) {stride_in_bytes = true}
!HOST: %[[VAR_PTR_PTR:.*]] = fir.box_offset %[[DECLARE_2]]#1 base_addr : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE_2]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS_2]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!HOST: %[[MAP_INFO_2:.*]] = omp.map_info var_ptr(%[[DECLARE_2]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "sp_write(2:5)"}
+!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE_2]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS_2]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!HOST: %[[MAP_INFO_2:.*]] = omp.map_info var_ptr(%[[DECLARE_2]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "sp_write(2:5)"}
subroutine read_write_section()
integer, allocatable :: sp_read(:)
@@ -80,8 +80,8 @@ module assumed_allocatable_array_routines
!HOST: %[[BOX_3:.*]]:3 = fir.box_dims %[[LOAD_3]], %[[CONSTANT_3]] : (!fir.box<!fir.heap<!fir.array<?xi32>>>, index) -> (index, index, index)
!HOST: %[[BOUNDS:.*]] = omp.bounds lower_bound(%[[LB]] : index) upper_bound(%[[UB]] : index) extent(%[[BOX_3]]#1 : index) stride(%[[BOX_2]]#2 : index) start_idx(%[[BOX_1]]#0 : index) {stride_in_bytes = true}
!HOST: %[[VAR_PTR_PTR:.*]] = fir.box_offset %[[DECLARE]]#1 base_addr : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!HOST: %[[MAP_INFO:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "arr_read_write(2:5)"}
+!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!HOST: %[[MAP_INFO:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.box<!fir.heap<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> {name = "arr_read_write(2:5)"}
subroutine assumed_shape_array(arr_read_write)
integer, allocatable, intent(inout) :: arr_read_write(:)
diff --git a/flang/test/Lower/OpenMP/allocatable-map.f90 b/flang/test/Lower/OpenMP/allocatable-map.f90
index ddc20b582b26e6..31fc86d9fb631c 100644
--- a/flang/test/Lower/OpenMP/allocatable-map.f90
+++ b/flang/test/Lower/OpenMP/allocatable-map.f90
@@ -2,12 +2,12 @@
!HLFIRDIALECT: %[[POINTER:.*]]:2 = hlfir.declare %0 {fortran_attrs = #fir.var_attrs<pointer>, uniq_name = "_QFpointer_routineEpoint"} : (!fir.ref<!fir.box<!fir.ptr<i32>>>) -> (!fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.ref<!fir.box<!fir.ptr<i32>>>)
!HLFIRDIALECT: %[[BOX_OFF:.*]] = fir.box_offset %[[POINTER]]#1 base_addr : (!fir.ref<!fir.box<!fir.ptr<i32>>>) -> !fir.llvm_ptr<!fir.ref<i32>>
-!HLFIRDIALECT: %[[POINTER_MAP_MEMBER:.*]] = omp.map_info var_ptr(%[[POINTER]]#1 : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr(%[[BOX_OFF]] : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(implicit, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
-!HLFIRDIALECT: %[[POINTER_MAP:.*]] = omp.map_info var_ptr(%[[POINTER]]#1 : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(implicit, tofrom) capture(ByRef) members(%[[POINTER_MAP_MEMBER]] : !fir.llvm_ptr<!fir.ref<i32>>) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "point"}
-!HLFIRDIALECT: omp.target map_entries({{.*}}, %[[POINTER_MAP_MEMBER]] -> {{.*}}, %[[POINTER_MAP]] -> {{.*}} : {{.*}}, !fir.llvm_ptr<!fir.ref<i32>>, !fir.ref<!fir.box<!fir.ptr<i32>>>) {
+!HLFIRDIALECT: %[[POINTER_MAP_MEMBER:.*]] = omp.map_info var_ptr(%[[POINTER]]#1 : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr(%[[BOX_OFF]] : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
+!HLFIRDIALECT: %[[POINTER_MAP:.*]] = omp.map_info var_ptr(%[[POINTER]]#1 : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[POINTER_MAP_MEMBER]] : !fir.llvm_ptr<!fir.ref<i32>> : [0]) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "point"}
+!HLFIRDIALECT: omp.target map_entries(%[[POINTER_MAP_MEMBER]] -> {{.*}}, %[[POINTER_MAP]] -> {{.*}} : !fir.llvm_ptr<!fir.ref<i32>>, !fir.ref<!fir.box<!fir.ptr<i32>>>) {
subroutine pointer_routine()
integer, pointer :: point
-!$omp target map(tofrom:pointer)
+!$omp target map(tofrom:point)
point = 1
!$omp end target
end subroutine pointer_routine
diff --git a/flang/test/Lower/OpenMP/array-bounds.f90 b/flang/test/Lower/OpenMP/array-bounds.f90
index 7d76ff4b106a0c..20b3b0754a990b 100644
--- a/flang/test/Lower/OpenMP/array-bounds.f90
+++ b/flang/test/Lower/OpenMP/array-bounds.f90
@@ -51,8 +51,8 @@ module assumed_array_routines
!HOST: %[[DIMS1:.*]]:3 = fir.box_dims %[[ARG0_DECL]]#1, %[[C0_1]] : (!fir.box<!fir.array<?xi32>>, index) -> (index, index, index)
!HOST: %[[BOUNDS:.*]] = omp.bounds lower_bound(%[[C3]] : index) upper_bound(%[[C4]] : index) extent(%[[DIMS1]]#1 : index) stride(%[[DIMS0]]#2 : index) start_idx(%[[C0]] : index) {stride_in_bytes = true}
!HOST: %[[VAR_PTR_PTR:.*]] = fir.box_offset %0 base_addr : (!fir.ref<!fir.box<!fir.array<?xi32>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
+!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
!HOST: omp.target map_entries(%[[MAP_INFO_MEMBER]] -> %{{.*}}, %[[MAP]] -> %{{.*}}, {{.*}} -> {{.*}} : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>, !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>) {
subroutine assumed_shape_array(arr_read_write)
integer, intent(inout) :: arr_read_write(:)
@@ -76,8 +76,8 @@ end subroutine assumed_shape_array
!HOST: %[[EXT:.*]] = arith.addi %[[C4_1]], %c1{{.*}} : index
!HOST: %[[BOUNDS:.*]] = omp.bounds lower_bound(%c1{{.*}} : index) upper_bound(%c4{{.*}} : index) extent(%[[EXT]] : index) stride(%[[DIMS0]]#2 : index) start_idx(%c1{{.*}} : index) {stride_in_bytes = true}
!HOST: %[[VAR_PTR_PTR:.*]] = fir.box_offset %[[INTERMEDIATE_ALLOCA]] base_addr : (!fir.ref<!fir.box<!fir.array<?xi32>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
+!HOST: %[[MAP_INFO_MEMBER:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[VAR_PTR_PTR]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+!HOST: %[[MAP:.*]] = omp.map_info var_ptr(%[[INTERMEDIATE_ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_INFO_MEMBER]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.array<?xi32>> {name = "arr_read_write(2:5)"}
!HOST: omp.target map_entries(%[[MAP_INFO_MEMBER]] -> %{{.*}}, %[[MAP]] -> %{{.*}}, {{.*}} -> {{.*}} : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>, !fir.ref<!fir.array<?xi32>>, !fir.ref<i32>) {
subroutine assumed_size_array(arr_read_write)
integer, intent(inout) :: arr_read_write(*)
diff --git a/flang/test/Lower/OpenMP/derived-type-map.f90 b/flang/test/Lower/OpenMP/derived-type-map.f90
new file mode 100644
index 00000000000000..b124c196dd215c
--- /dev/null
+++ b/flang/test/Lower/OpenMP/derived-type-map.f90
@@ -0,0 +1,105 @@
+!RUN: %flang_fc1 -emit-hlfir -fopenmp %s -o - | FileCheck %s
+
+
+!CHECK: %[[ALLOCA:.*]] = fir.alloca !fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}> {bindc_name = "scalar_arr", uniq_name = "_QFmaptype_derived_implicitEscalar_arr"}
+!CHECK: %[[DECLARE:.*]]:2 = hlfir.declare %[[ALLOCA]] {uniq_name = "_QFmaptype_derived_implicitEscalar_arr"} : (!fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> (!fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>)
+!CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(implicit, tofrom) capture(ByRef) -> !fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "scalar_arr"}
+!CHECK: omp.target map_entries(%[[MAP]] -> %[[ARG0:.*]] : !fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+!CHECK: ^bb0(%[[ARG0]]: !fir.ref<!fir.type<_QFmaptype_derived_implicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+subroutine mapType_derived_implicit
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_implicit
+
+!CHECK: %[[ALLOCA:.*]] = fir.alloca !fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}> {bindc_name = "scalar_arr", uniq_name = "_QFmaptype_derived_explicitEscalar_arr"}
+!CHECK: %[[DECLARE:.*]]:2 = hlfir.declare %[[ALLOCA]] {uniq_name = "_QFmaptype_derived_explicitEscalar_arr"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> (!fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>)
+!CHECK: %[[MAP:.*]] = omp.map_info var_ptr(%[[DECLARE]]#0 : !fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(tofrom) capture(ByRef) -> !fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "scalar_arr"}
+!CHECK: omp.target map_entries(%[[MAP]] -> %[[ARG0:.*]] : !fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+!CHECK: ^bb0(%[[ARG0]]: !fir.ref<!fir.type<_QFmaptype_derived_explicitTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+subroutine mapType_derived_explicit
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr)
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_explicit
+
+!CHECK: %[[ALLOCA:.*]] = fir.alloca !fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}> {bindc_name = "scalar_arr", uniq_name = "_QFmaptype_derived_explicit_single_memberEscalar_arr"}
+!CHECK: %[[DECLARE:.*]]:2 = hlfir.declare %[[ALLOCA]] {uniq_name = "_QFmaptype_derived_explicit_single_memberEscalar_arr"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> (!fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>)
+!CHECK: %[[MEMBER:.*]] = hlfir.designate %[[DECLARE]]#0{"array"} shape %{{.*}} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.shape<1>) -> !fir.ref<!fir.array<10xi32>>
+!CHECK: %[[BOUNDS:.*]] = omp.bounds lower_bound(%{{.*}} : index) upper_bound(%{{.*}} : index) extent(%{{.*}} : index) stride(%{{.*}} : index) start_idx(%{{.*}} : index)
+!CHECK: %[[MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[MEMBER]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>) map_clauses(tofrom) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.ref<!fir.array<10xi32>> {name = "scalar_arr%array"}
+!CHECK: %[[PARENT_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(tofrom) capture(ByRef) members(%[[MEMBER_MAP]] : !fir.ref<!fir.array<10xi32>> : [1]) -> !fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "scalar_arr", partial_map = true}
+!CHECK: omp.target map_entries(%[[MEMBER_MAP]] -> %[[ARG0:.*]], %[[PARENT_MAP]] -> %[[ARG1:.*]] : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+!CHECK: ^bb0(%[[ARG0]]: !fir.ref<!fir.array<10xi32>>, %[[ARG1]]: !fir.ref<!fir.type<_QFmaptype_derived_explicit_single_memberTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+subroutine mapType_derived_explicit_single_member
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%array)
+ scalar_arr%array(1) = 1
+ !$omp end target
+end subroutine mapType_derived_explicit_single_member
+
+!CHECK: %[[ALLOCA:.*]] = fir.alloca !fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}> {bindc_name = "scalar_arr", uniq_name = "_QFmaptype_derived_explicit_multiple_membersEscalar_arr"}
+!CHECK: %[[DECLARE:.*]]:2 = hlfir.declare %[[ALLOCA]] {uniq_name = "_QFmaptype_derived_explicit_multiple_membersEscalar_arr"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> (!fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>)
+!CHECK: %[[MEMBER1:.*]] = hlfir.designate %[[DECLARE]]#0{"int"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> !fir.ref<i32>
+!CHECK: %[[MEMBER_MAP_1:.*]] = omp.map_info var_ptr(%[[MEMBER1]] : !fir.ref<i32>, i32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "scalar_arr%int"}
+!CHECK: %[[MEMBER2:.*]] = hlfir.designate %[[DECLARE]]#0{"real"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> !fir.ref<f32>
+!CHECK: %[[MEMBER_MAP_2:.*]] = omp.map_info var_ptr(%[[MEMBER2]] : !fir.ref<f32>, f32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<f32> {name = "scalar_arr%real"}
+!CHECK: %[[PARENT_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(tofrom) capture(ByRef) members(%[[MEMBER_MAP_1]], %[[MEMBER_MAP_2]] : !fir.ref<i32>, !fir.ref<f32> : [2, 0]) -> !fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "scalar_arr", partial_map = true}
+!CHECK: omp.target map_entries(%[[MEMBER_MAP_1]] -> %[[ARG0:.*]], %[[MEMBER_MAP_2]] -> %[[ARG1:.*]], %[[PARENT_MAP]] -> %[[ARG2:.*]] : !fir.ref<i32>, !fir.ref<f32>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+!CHECK: ^bb0(%[[ARG0]]: !fir.ref<f32>, %[[ARG1]]: !fir.ref<i32>, %[[ARG2]]: !fir.ref<!fir.type<_QFmaptype_derived_explicit_multiple_membersTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+subroutine mapType_derived_explicit_multiple_members
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%int, scalar_arr%real)
+ scalar_arr%int = 1
+ !$omp end target
+end subroutine mapType_derived_explicit_multiple_members
+
+!CHECK: %[[ALLOCA:.*]] = fir.alloca !fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}> {bindc_name = "scalar_arr", uniq_name = "_QFmaptype_derived_explicit_member_with_boundsEscalar_arr"}
+!CHECK: %[[DECLARE:.*]]:2 = hlfir.declare %[[ALLOCA]] {uniq_name = "_QFmaptype_derived_explicit_member_with_boundsEscalar_arr"} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) -> (!fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>)
+!CHECK: %[[MEMBER:.*]] = hlfir.designate %[[DECLARE]]#0{"array"} shape %{{.*}} : (!fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.shape<1>) -> !fir.ref<!fir.array<10xi32>>
+!CHECK: %{{.*}} = arith.constant 1 : index
+!CHECK: %[[LB:.*]] = arith.constant 1 : index
+!CHECK: %[[UB:.*]] = arith.constant 4 : index
+!CHECK: %[[BOUNDS:.*]] = omp.bounds lower_bound(%[[LB]] : index) upper_bound(%[[UB]] : index) extent(%{{.*}} : index) stride(%{{.*}} : index) start_idx(%{{.*}} : index)
+!CHECK: %[[MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[MEMBER]] : !fir.ref<!fir.array<10xi32>>, !fir.array<10xi32>) map_clauses(tofrom) capture(ByRef) bounds(%20) -> !fir.ref<!fir.array<10xi32>> {name = "scalar_arr%array(2:5)"}
+!CHECK: %[[PARENT_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE]]#1 : !fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>, !fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>) map_clauses(tofrom) capture(ByRef) members(%[[MEMBER_MAP]] : !fir.ref<!fir.array<10xi32>> : [1]) -> !fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>> {name = "scalar_arr", partial_map = true}
+!CHECK: omp.target map_entries(%[[MEMBER_MAP]] -> %[[ARG0:.*]], %[[PARENT_MAP]] -> %[[ARG1:.*]] : !fir.ref<!fir.array<10xi32>>, !fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>) {
+!CHECK: ^bb0(%[[ARG0]]: !fir.ref<!fir.array<10xi32>>, %[[ARG1]]: !fir.ref<!fir.type<_QFmaptype_derived_explicit_member_with_boundsTscalar_and_array{real:f32,array:!fir.array<10xi32>,int:i32}>>):
+subroutine mapType_derived_explicit_member_with_bounds
+ type :: scalar_and_array
+ real(4) :: real
+ integer(4) :: array(10)
+ integer(4) :: int
+ end type scalar_and_array
+ type(scalar_and_array) :: scalar_arr
+
+ !$omp target map(tofrom: scalar_arr%array(2:5))
+ scalar_arr%array(3) = 3
+ !$omp end target
+end subroutine mapType_derived_explicit_member_with_bounds
diff --git a/flang/test/Lower/OpenMP/map-component-ref.f90 b/flang/test/Lower/OpenMP/map-component-ref.f90
index 1ed37e73158025..027be7c6955724 100644
--- a/flang/test/Lower/OpenMP/map-component-ref.f90
+++ b/flang/test/Lower/OpenMP/map-component-ref.f90
@@ -5,7 +5,7 @@
! CHECK: %[[V1:[0-9]+]]:2 = hlfir.declare %[[V0]] {uniq_name = "_QFfooEa"} : (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>) -> (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>)
! CHECK: %[[V2:[0-9]+]] = hlfir.designate %[[V1]]#0{"a1"} : (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>) -> !fir.ref<i32>
! CHECK: %[[V3:[0-9]+]] = omp.map_info var_ptr(%[[V2]] : !fir.ref<i32>, i32) map_clauses(tofrom) capture(ByRef) -> !fir.ref<i32> {name = "a%a1"}
-! CHECK: %[[V4:[0-9]+]] = omp.map_info var_ptr(%[[V1]]#1 : !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.type<_QFfooTt0{a0:i32,a1:i32}>) map_clauses(implicit, tofrom) capture(ByRef) -> !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>> {name = "a"}
+! CHECK: %[[V4:[0-9]+]] = omp.map_info var_ptr(%[[V1]]#1 : !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.type<_QFfooTt0{a0:i32,a1:i32}>) map_clauses(tofrom) capture(ByRef) members(%[[V3]] : !fir.ref<i32> : [1]) -> !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>> {name = "a", partial_map = true}
! CHECK: omp.target map_entries(%[[V3]] -> %arg0, %[[V4]] -> %arg1 : !fir.ref<i32>, !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>) {
! CHECK: ^bb0(%arg0: !fir.ref<i32>, %arg1: !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>):
! CHECK: %[[V5:[0-9]+]]:2 = hlfir.declare %arg1 {uniq_name = "_QFfooEa"} : (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>) -> (!fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>, !fir.ref<!fir.type<_QFfooTt0{a0:i32,a1:i32}>>)
diff --git a/flang/test/Lower/OpenMP/target.f90 b/flang/test/Lower/OpenMP/target.f90
index fa07b7f71d514e..a148b7275804f5 100644
--- a/flang/test/Lower/OpenMP/target.f90
+++ b/flang/test/Lower/OpenMP/target.f90
@@ -445,8 +445,8 @@ subroutine omp_target_device_addr
integer, pointer :: a
!CHECK: %[[VAL_0:.*]] = fir.alloca !fir.box<!fir.ptr<i32>> {bindc_name = "a", uniq_name = "_QFomp_target_device_addrEa"}
!CHECK: %[[VAL_0_DECL:.*]]:2 = hlfir.declare %0 {fortran_attrs = #fir.var_attrs<pointer>, uniq_name = "_QFomp_target_device_addrEa"} : (!fir.ref<!fir.box<!fir.ptr<i32>>>) -> (!fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.ref<!fir.box<!fir.ptr<i32>>>)
- !CHECK: %[[MAP_MEMBERS:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr({{.*}} : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
- !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBERS]] : !fir.llvm_ptr<!fir.ref<i32>>) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "a"}
+ !CHECK: %[[MAP_MEMBERS:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, i32) var_ptr_ptr({{.*}} : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
+ !CHECK: %[[MAP:.*]] = omp.map_info var_ptr({{.*}} : !fir.ref<!fir.box<!fir.ptr<i32>>>, !fir.box<!fir.ptr<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[MAP_MEMBERS]] : !fir.llvm_ptr<!fir.ref<i32>> : [0]) -> !fir.ref<!fir.box<!fir.ptr<i32>>> {name = "a"}
!CHECK: omp.target_data map_entries(%[[MAP_MEMBERS]], %[[MAP]] : {{.*}}) use_device_addr(%[[VAL_0_DECL]]#1 : !fir.ref<!fir.box<!fir.ptr<i32>>>) {
!$omp target data map(tofrom: a) use_device_addr(a)
!CHECK: ^bb0(%[[VAL_1:.*]]: !fir.ref<!fir.box<!fir.ptr<i32>>>):
diff --git a/flang/test/Transforms/omp-descriptor-map-info-gen.fir b/flang/test/Transforms/omp-map-info-finalization.fir
similarity index 53%
rename from flang/test/Transforms/omp-descriptor-map-info-gen.fir
rename to flang/test/Transforms/omp-map-info-finalization.fir
index 22594ec88c9cb1..dcc3b80cf6e404 100644
--- a/flang/test/Transforms/omp-descriptor-map-info-gen.fir
+++ b/flang/test/Transforms/omp-map-info-finalization.fir
@@ -1,4 +1,4 @@
-// RUN: fir-opt --omp-descriptor-map-info-gen %s | FileCheck %s
+// RUN: fir-opt --split-input-file --omp-map-info-finalization %s | FileCheck %s
module attributes {omp.is_target_device = false} {
func.func @test_descriptor_expansion_pass(%arg0: !fir.box<!fir.array<?xi32>>) {
@@ -34,11 +34,35 @@ module attributes {omp.is_target_device = false} {
// CHECK: %[[DECLARE2:.*]]:2 = hlfir.declare %[[ALLOCA2]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "test2"} : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> (!fir.ref<!fir.box<!fir.heap<i32>>>, !fir.ref<!fir.box<!fir.heap<i32>>>)
// CHECK: %[[BOUNDS:.*]] = omp.bounds lower_bound(%{{.*}} : index) upper_bound(%{{.*}} : index) extent(%{{.*}} : index) stride(%{{.*}} : index) start_idx(%{{.*}} : index) {stride_in_bytes = true}
// CHECK: %[[BASE_ADDR_OFF:.*]] = fir.box_offset %[[DECLARE2]]#1 base_addr : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> !fir.llvm_ptr<!fir.ref<i32>>
-// CHECK: %[[DESC_MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE2]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>>, i32) var_ptr_ptr(%[[BASE_ADDR_OFF]] : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
-// CHECK: %[[DESC_PARENT_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE2]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>>, !fir.box<!fir.heap<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[DESC_MEMBER_MAP]] : !fir.llvm_ptr<!fir.ref<i32>>) -> !fir.ref<!fir.box<!fir.heap<i32>>>
+// CHECK: %[[DESC_MEMBER_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE2]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>>, i32) var_ptr_ptr(%[[BASE_ADDR_OFF]] : !fir.llvm_ptr<!fir.ref<i32>>) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !fir.llvm_ptr<!fir.ref<i32>> {name = ""}
+// CHECK: %[[DESC_PARENT_MAP:.*]] = omp.map_info var_ptr(%[[DECLARE2]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>>, !fir.box<!fir.heap<i32>>) map_clauses(tofrom) capture(ByRef) members(%[[DESC_MEMBER_MAP]] : !fir.llvm_ptr<!fir.ref<i32>> : [0]) -> !fir.ref<!fir.box<!fir.heap<i32>>>
// CHECK: fir.store %[[DECLARE1]]#1 to %[[ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>
// CHECK: %[[BASE_ADDR_OFF_2:.*]] = fir.box_offset %[[ALLOCA]] base_addr : (!fir.ref<!fir.box<!fir.array<?xi32>>>) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>
-// CHECK: %[[DESC_MEMBER_MAP_2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[BASE_ADDR_OFF_2]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(from) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
-// CHECK: %[[DESC_PARENT_MAP_2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(from) capture(ByRef) members(%15 : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) -> !fir.ref<!fir.array<?xi32>>
+// CHECK: %[[DESC_MEMBER_MAP_2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.array<?xi32>) var_ptr_ptr(%[[BASE_ADDR_OFF_2]] : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>) map_clauses(ptr_and_obj, from) capture(ByRef) bounds(%[[BOUNDS]]) -> !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> {name = ""}
+// CHECK: %[[DESC_PARENT_MAP_2:.*]] = omp.map_info var_ptr(%[[ALLOCA]] : !fir.ref<!fir.box<!fir.array<?xi32>>>, !fir.box<!fir.array<?xi32>>) map_clauses(from) capture(ByRef) members(%15 : !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>> : [0]) -> !fir.ref<!fir.array<?xi32>>
// CHECK: omp.target map_entries(%[[DESC_MEMBER_MAP]] -> %[[ARG1:.*]], %[[DESC_PARENT_MAP]] -> %[[ARG2:.*]], %[[DESC_MEMBER_MAP_2]] -> %[[ARG3:.*]], %[[DESC_PARENT_MAP_2]] -> %[[ARG4:.*]] : {{.*}}) {
// CHECK: ^bb0(%[[ARG1]]: !fir.llvm_ptr<!fir.ref<i32>>, %[[ARG2]]: !fir.ref<!fir.box<!fir.heap<i32>>>, %[[ARG3]]: !fir.llvm_ptr<!fir.ref<!fir.array<?xi32>>>, %[[ARG4]]: !fir.ref<!fir.array<?xi32>>):
+
+// -----
+
+module attributes {omp.is_target_device = false} {
+ func.func @test_derived_type_map_operand_and_block_addition(%arg0: !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) {
+ %0 = hlfir.designate %arg0{"rx"} : (!fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) -> !fir.ref<f32>
+ %1 = omp.map_info var_ptr(%0 : !fir.ref<f32>, f32) map_clauses(from) capture(ByRef) -> !fir.ref<f32> {name = "scalar_struct%rx"}
+ %2 = hlfir.designate %arg0{"ry"} : (!fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) -> !fir.ref<f32>
+ %3 = omp.map_info var_ptr(%2 : !fir.ref<f32>, f32) map_clauses(from) capture(ByRef) -> !fir.ref<f32> {name = "scalar_struct%ry"}
+ %4 = omp.map_info var_ptr(%arg0 : !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>, !fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>) map_clauses(from) capture(ByRef) members(%1, %3 : !fir.ref<f32>, !fir.ref<f32> : [1, 4]) -> !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>> {name = "scalar_struct", partial_map = true}
+ omp.target map_entries(%4 -> %arg1 : !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) {
+ ^bb0(%arg1: !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>):
+ omp.terminator
+ }
+ return
+ }
+}
+
+// CHECK: func.func @test_derived_type_map_operand_and_block_addition(%{{.*}}: !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) {
+// CHECK: %[[MAP_MEMBER_1:.*]] = omp.map_info var_ptr(%{{.*}} : !fir.ref<f32>, f32) map_clauses(from) capture(ByRef) -> !fir.ref<f32> {name = "scalar_struct%rx"}
+// CHECK: %[[MAP_MEMBER_2:.*]] = omp.map_info var_ptr(%{{.*}} : !fir.ref<f32>, f32) map_clauses(from) capture(ByRef) -> !fir.ref<f32> {name = "scalar_struct%ry"}
+// CHECK: %[[MAP_PARENT:.*]] = omp.map_info var_ptr(%{{.*}} : !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>, !fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>) map_clauses(from) capture(ByRef) members(%1, %3 : !fir.ref<f32>, !fir.ref<f32> : [1, 4]) -> !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>> {name = "scalar_struct", partial_map = true}
+// CHECK: omp.target map_entries(%[[MAP_MEMBER_1]] -> %arg1, %[[MAP_MEMBER_2]] -> %arg2, %[[MAP_PARENT]] -> %arg3 : !fir.ref<f32>, !fir.ref<f32>, !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>) {
+// CHECK: ^bb0(%arg1: !fir.ref<f32>, %arg2: !fir.ref<f32>, %arg3: !fir.ref<!fir.type<_QFTdtype{ix:i32,rx:f32,zx:!fir.complex<4>,nested:!fir.box<!fir.heap<!fir.type<_QFTdtype>>>,ry:f32}>>):
diff --git a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
index ca363505485773..cdfe6a8f8edf48 100644
--- a/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
+++ b/mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td
@@ -1195,10 +1195,12 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
TypeAttr:$var_type,
Optional<OpenMP_PointerLikeType>:$var_ptr_ptr,
Variadic<OpenMP_PointerLikeType>:$members,
+ OptionalAttr<I64ArrayAttr>:$members_index,
Variadic<DataBoundsType>:$bounds, /* rank-0 to rank-{n-1} */
OptionalAttr<UI64Attr>:$map_type,
OptionalAttr<VariableCaptureKindAttr>:$map_capture_type,
- OptionalAttr<StrAttr>:$name);
+ OptionalAttr<StrAttr>:$name,
+ DefaultValuedAttr<BoolAttr, "false">:$partial_map);
let results = (outs OpenMP_PointerLikeType:$omp_ptr);
let description = [{
@@ -1234,10 +1236,14 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
- `var_type`: The type of the 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.
- - `members`: Used to indicate mapped child members for the current MapInfoOp,
+ - `members`: Used to indicate mapped child members for the current MapInfoOp,
represented as other MapInfoOp's, utilised in cases where a parent structure
type and members of the structure type are being mapped at the same time.
- For example: map(to: parent, parent->member, parent->member2[:10])
+ For example: map(to: parent, parent->member, parent->member2[:10])
+ - `members_index`: Used to indicate the ordering of members within the containing
+ parent (generally a record type such as a structure, class or derived type),
+ e.g. struct {int x, float y, double z}, x would be 0, y would be 1, and z
+ would be 2. This aids the mapping
- `bounds`: Used when copying slices of array's, pointers or pointer members of
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
@@ -1248,6 +1254,8 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
- 'map_capture_type': Capture type for the variable e.g. this, byref, byvalue, byvla
this can affect how the variable is lowered.
- `name`: Holds the name of variable as specified in user clause (including bounds).
+ - `partial_map`: The record type being mapped will not be mapped in its entirety,
+ it may be used however, in a mapping to bind it's mapped components together.
}];
let assemblyFormat = [{
@@ -1256,7 +1264,7 @@ def MapInfoOp : OpenMP_Op<"map_info", [AttrSizedOperandSegments]> {
`var_ptr_ptr` `(` $var_ptr_ptr `:` type($var_ptr_ptr) `)`
| `map_clauses` `(` custom<MapClause>($map_type) `)`
| `capture` `(` custom<CaptureType>($map_capture_type) `)`
- | `members` `(` $members `:` type($members) `)`
+ | `members` `(` $members `:` type($members) `:` $members_index `)`
| `bounds` `(` $bounds `)`
) `->` type($omp_ptr) attr-dict
}];
diff --git a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
index 381f17d0804191..7d166b1f78bff3 100644
--- a/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
+++ b/mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp
@@ -709,6 +709,9 @@ static ParseResult parseMapClause(OpAsmParser &parser, IntegerAttr &mapType) {
if (mapTypeMod == "delete")
mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_DELETE;
+ if (mapTypeMod == "ptr_and_obj")
+ mapTypeBits |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
+
return success();
};
@@ -745,6 +748,10 @@ static void printMapClause(OpAsmPrinter &p, Operation *op,
if (mapTypeToBitFlag(mapTypeBits,
llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PRESENT))
mapTypeStrs.push_back("present");
+ if (mapTypeToBitFlag(
+ mapTypeBits,
+ llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ))
+ mapTypeStrs.push_back("ptr_and_obj");
// special handling of to/from/tofrom/delete and release/alloc, release +
// alloc are the abscense of one of the other flags, whereas tofrom requires
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index 79956f82ed141a..1ba8099135c35f 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -1750,9 +1750,9 @@ void collectMapDataFromMapOperands(MapInfoData &mapData,
mapData.BaseType.push_back(
moduleTranslation.convertType(mapOp.getVarType()));
- mapData.Sizes.push_back(getSizeInBytes(
- dl, mapOp.getVarType(), mapOp, mapData.BasePointers.back(),
- mapData.BaseType.back(), builder, moduleTranslation));
+ mapData.Sizes.push_back(
+ getSizeInBytes(dl, mapOp.getVarType(), mapOp, mapData.Pointers.back(),
+ mapData.BaseType.back(), builder, moduleTranslation));
mapData.MapClause.push_back(mapOp.getOperation());
mapData.Types.push_back(
llvm::omp::OpenMPOffloadMappingFlags(mapOp.getMapType().value()));
@@ -1783,6 +1783,98 @@ void collectMapDataFromMapOperands(MapInfoData &mapData,
}
}
+static int getMapDataMemberIdx(MapInfoData &mapData,
+ mlir::omp::MapInfoOp memberOp) {
+ int memberDataIdx = -1;
+ for (size_t i = 0; i < mapData.MapClause.size(); ++i) {
+ if (mapData.MapClause[i] == memberOp)
+ memberDataIdx = i;
+ }
+ return memberDataIdx;
+}
+
+static mlir::omp::MapInfoOp
+getFirstOrLastMappedMemberPtr(mlir::omp::MapInfoOp mapInfo, bool first) {
+ // Only 1 member has been mapped, we can return it.
+ if (mapInfo.getMembersIndex()->size() == 1)
+ if (auto mapOp = mlir::dyn_cast<mlir::omp::MapInfoOp>(
+ mapInfo.getMembers()[0].getDefiningOp()))
+ return mapOp;
+
+ int64_t curPos =
+ mapInfo.getMembersIndex()->begin()->cast<mlir::IntegerAttr>().getInt();
+
+ int64_t idx = 1, curIdx = 0, memberPlacement = 0;
+ for (const auto *iter = std::next(mapInfo.getMembersIndex()->begin());
+ iter != mapInfo.getMembersIndex()->end(); iter++) {
+ memberPlacement = iter->cast<mlir::IntegerAttr>().getInt();
+ if (first) {
+ if (memberPlacement < curPos) {
+ curIdx = idx;
+ curPos = memberPlacement;
+ }
+ } else {
+ if (memberPlacement > curPos) {
+ curIdx = idx;
+ curPos = memberPlacement;
+ }
+ }
+ idx++;
+ }
+
+ if (auto mapOp = mlir::dyn_cast<mlir::omp::MapInfoOp>(
+ mapInfo.getMembers()[curIdx].getDefiningOp()))
+ return mapOp;
+
+ return {};
+}
+
+std::vector<llvm::Value *>
+calculateBoundsOffset(LLVM::ModuleTranslation &moduleTranslation,
+ llvm::IRBuilderBase &builder, bool isArrayTy,
+ mlir::OperandRange bounds) {
+ std::vector<llvm::Value *> idx;
+ llvm::Value *offsetAddress = nullptr;
+ if (!bounds.empty()) {
+ idx.push_back(builder.getInt64(0));
+ if (isArrayTy) {
+ for (int i = bounds.size() - 1; i >= 0; --i) {
+ if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
+ bounds[i].getDefiningOp())) {
+ idx.push_back(moduleTranslation.lookupValue(boundOp.getLowerBound()));
+ }
+ }
+ } else {
+ std::vector<llvm::Value *> dimensionIndexSizeOffset{builder.getInt64(1)};
+ for (size_t i = 1; i < bounds.size(); ++i) {
+ if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
+ bounds[i].getDefiningOp())) {
+ dimensionIndexSizeOffset.push_back(builder.CreateMul(
+ moduleTranslation.lookupValue(boundOp.getExtent()),
+ dimensionIndexSizeOffset[i - 1]));
+ }
+ }
+
+ for (int i = bounds.size() - 1; i >= 0; --i) {
+ if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
+ bounds[i].getDefiningOp())) {
+ if (!offsetAddress)
+ offsetAddress = builder.CreateMul(
+ moduleTranslation.lookupValue(boundOp.getLowerBound()),
+ dimensionIndexSizeOffset[i]);
+ else
+ offsetAddress = builder.CreateAdd(
+ offsetAddress, builder.CreateMul(moduleTranslation.lookupValue(
+ boundOp.getLowerBound()),
+ dimensionIndexSizeOffset[i]));
+ }
+ }
+ }
+ }
+
+ return offsetAddress ? std::vector<llvm::Value *>{offsetAddress} : idx;
+}
+
// This creates two insertions into the MapInfosTy data structure for the
// "parent" of a set of members, (usually a container e.g.
// class/structure/derived type) when subsequent members have also been
@@ -1810,7 +1902,6 @@ static llvm::omp::OpenMPOffloadMappingFlags mapParentWithMembers(
combinedInfo.Names.emplace_back(LLVM::createMappingInformation(
mapData.MapClause[mapDataIndex]->getLoc(), ompBuilder));
combinedInfo.BasePointers.emplace_back(mapData.BasePointers[mapDataIndex]);
- combinedInfo.Pointers.emplace_back(mapData.Pointers[mapDataIndex]);
// Calculate size of the parent object being mapped based on the
// addresses at runtime, highAddr - lowAddr = size. This of course
@@ -1819,24 +1910,40 @@ static llvm::omp::OpenMPOffloadMappingFlags mapParentWithMembers(
// Fortran pointers and allocatables, the mapping of the pointed to
// data by the descriptor (which itself, is a structure containing
// runtime information on the dynamically allocated data).
- llvm::Value *lowAddr = builder.CreatePointerCast(
- mapData.Pointers[mapDataIndex], builder.getPtrTy());
- llvm::Value *highAddr = builder.CreatePointerCast(
- builder.CreateConstGEP1_32(mapData.BaseType[mapDataIndex],
- mapData.Pointers[mapDataIndex], 1),
- builder.getPtrTy());
+ auto parentClause =
+ mlir::dyn_cast<mlir::omp::MapInfoOp>(mapData.MapClause[mapDataIndex]);
+
+ llvm::Value *lowAddr, *highAddr;
+ if (!parentClause.getPartialMap()) {
+ lowAddr = builder.CreatePointerCast(mapData.Pointers[mapDataIndex],
+ builder.getPtrTy());
+ highAddr = builder.CreatePointerCast(
+ builder.CreateConstGEP1_32(mapData.BaseType[mapDataIndex],
+ mapData.Pointers[mapDataIndex], 1),
+ builder.getPtrTy());
+ combinedInfo.Pointers.emplace_back(mapData.Pointers[mapDataIndex]);
+ } else {
+ auto mapOp =
+ mlir::dyn_cast<mlir::omp::MapInfoOp>(mapData.MapClause[mapDataIndex]);
+ int firstMemberIdx = getMapDataMemberIdx(
+ mapData, getFirstOrLastMappedMemberPtr(mapOp, true));
+ lowAddr = builder.CreatePointerCast(mapData.Pointers[firstMemberIdx],
+ builder.getPtrTy());
+ int lastMemberIdx = getMapDataMemberIdx(
+ mapData, getFirstOrLastMappedMemberPtr(mapOp, false));
+ highAddr = builder.CreatePointerCast(
+ builder.CreateGEP(mapData.BaseType[lastMemberIdx],
+ mapData.Pointers[lastMemberIdx], builder.getInt64(1)),
+ builder.getPtrTy());
+ combinedInfo.Pointers.emplace_back(mapData.Pointers[firstMemberIdx]);
+ }
+
llvm::Value *size = builder.CreateIntCast(
builder.CreatePtrDiff(builder.getInt8Ty(), highAddr, lowAddr),
builder.getInt64Ty(),
/*isSigned=*/false);
combinedInfo.Sizes.push_back(size);
- // This creates the initial MEMBER_OF mapping that consists of
- // the parent/top level container (same as above effectively, except
- // with a fixed initial compile time size and seperate maptype which
- // indicates the true mape type (tofrom etc.) and that it is a part
- // of a larger mapping and indicating the link between it and it's
- // members that are also explicitly mapped).
llvm::omp::OpenMPOffloadMappingFlags mapFlag =
llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TO;
if (isTargetParams)
@@ -1846,15 +1953,22 @@ static llvm::omp::OpenMPOffloadMappingFlags mapParentWithMembers(
ompBuilder.getMemberOfFlag(combinedInfo.BasePointers.size() - 1);
ompBuilder.setCorrectMemberOfFlag(mapFlag, memberOfFlag);
- combinedInfo.Types.emplace_back(mapFlag);
- combinedInfo.DevicePointers.emplace_back(
- llvm::OpenMPIRBuilder::DeviceInfoTy::None);
- combinedInfo.Names.emplace_back(LLVM::createMappingInformation(
- mapData.MapClause[mapDataIndex]->getLoc(), ompBuilder));
- combinedInfo.BasePointers.emplace_back(mapData.BasePointers[mapDataIndex]);
- combinedInfo.Pointers.emplace_back(mapData.Pointers[mapDataIndex]);
- combinedInfo.Sizes.emplace_back(mapData.Sizes[mapDataIndex]);
-
+ // This creates the initial MEMBER_OF mapping that consists of
+ // the parent/top level container (same as above effectively, except
+ // with a fixed initial compile time size and seperate maptype which
+ // indicates the true mape type (tofrom etc.). This parent mapping is
+ // only relevant if the structure in it's totality is being mapped,
+ // otherwise the above suffices.
+ if (!parentClause.getPartialMap()) {
+ combinedInfo.Types.emplace_back(mapFlag);
+ combinedInfo.DevicePointers.emplace_back(
+ llvm::OpenMPIRBuilder::DeviceInfoTy::None);
+ combinedInfo.Names.emplace_back(LLVM::createMappingInformation(
+ mapData.MapClause[mapDataIndex]->getLoc(), ompBuilder));
+ combinedInfo.BasePointers.emplace_back(mapData.BasePointers[mapDataIndex]);
+ combinedInfo.Pointers.emplace_back(mapData.Pointers[mapDataIndex]);
+ combinedInfo.Sizes.emplace_back(mapData.Sizes[mapDataIndex]);
+ }
return memberOfFlag;
}
@@ -1871,86 +1985,99 @@ static void processMapMembersWithParent(
for (auto mappedMembers : parentClause.getMembers()) {
auto memberClause =
mlir::dyn_cast<mlir::omp::MapInfoOp>(mappedMembers.getDefiningOp());
- int memberDataIdx = -1;
- for (size_t i = 0; i < mapData.MapClause.size(); ++i) {
- if (mapData.MapClause[i] == memberClause)
- memberDataIdx = i;
- }
+ int memberDataIdx = getMapDataMemberIdx(mapData, memberClause);
assert(memberDataIdx >= 0 && "could not find mapped member of structure");
// Same MemberOfFlag to indicate its link with parent and other members
- // of, and we flag that it's part of a pointer and object coupling.
+ // of
auto mapFlag =
llvm::omp::OpenMPOffloadMappingFlags(memberClause.getMapType().value());
mapFlag &= ~llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TARGET_PARAM;
+ mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_MEMBER_OF;
ompBuilder.setCorrectMemberOfFlag(mapFlag, memberOfFlag);
- mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
+
combinedInfo.Types.emplace_back(mapFlag);
combinedInfo.DevicePointers.emplace_back(
llvm::OpenMPIRBuilder::DeviceInfoTy::None);
combinedInfo.Names.emplace_back(
LLVM::createMappingInformation(memberClause.getLoc(), ompBuilder));
-
- combinedInfo.BasePointers.emplace_back(mapData.BasePointers[memberDataIdx]);
-
- std::vector<llvm::Value *> idx{builder.getInt64(0)};
- llvm::Value *offsetAddress = nullptr;
- if (!memberClause.getBounds().empty()) {
- if (mapData.BaseType[memberDataIdx]->isArrayTy()) {
- for (int i = memberClause.getBounds().size() - 1; i >= 0; --i) {
- if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
- memberClause.getBounds()[i].getDefiningOp())) {
- idx.push_back(
- moduleTranslation.lookupValue(boundOp.getLowerBound()));
- }
- }
- } else {
- std::vector<llvm::Value *> dimensionIndexSizeOffset{
- builder.getInt64(1)};
- for (size_t i = 1; i < memberClause.getBounds().size(); ++i) {
- if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
- memberClause.getBounds()[i].getDefiningOp())) {
- dimensionIndexSizeOffset.push_back(builder.CreateMul(
- moduleTranslation.lookupValue(boundOp.getExtent()),
- dimensionIndexSizeOffset[i - 1]));
- }
- }
-
- for (int i = memberClause.getBounds().size() - 1; i >= 0; --i) {
- if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
- memberClause.getBounds()[i].getDefiningOp())) {
- if (!offsetAddress)
- offsetAddress = builder.CreateMul(
- moduleTranslation.lookupValue(boundOp.getLowerBound()),
- dimensionIndexSizeOffset[i]);
- else
- offsetAddress = builder.CreateAdd(
- offsetAddress,
- builder.CreateMul(
- moduleTranslation.lookupValue(boundOp.getLowerBound()),
- dimensionIndexSizeOffset[i]));
- }
- }
- }
- }
-
- llvm::Value *memberIdx =
- builder.CreateLoad(builder.getPtrTy(), mapData.Pointers[memberDataIdx]);
- memberIdx = builder.CreateInBoundsGEP(
- mapData.BaseType[memberDataIdx], memberIdx,
- offsetAddress ? std::vector<llvm::Value *>{offsetAddress} : idx,
- "member_idx");
- combinedInfo.Pointers.emplace_back(memberIdx);
+ combinedInfo.BasePointers.emplace_back(mapData.BasePointers[mapDataIndex]);
+ combinedInfo.Pointers.emplace_back(mapData.Pointers[memberDataIdx]);
combinedInfo.Sizes.emplace_back(mapData.Sizes[memberDataIdx]);
}
}
+// This may be a bit of a naive check, the intent is to verify if the
+// mapped data being passed is a pointer -> pointee that requires special
+// handling in certain cases. There may be a better way to verify this, but
+// unfortunately with opaque pointers we lose the ability to easily check if
+// something is a pointer whilst maintaining access to the underlying type.
+static bool checkIfPointerMap(llvm::omp::OpenMPOffloadMappingFlags mapFlag) {
+ return static_cast<
+ std::underlying_type_t<llvm::omp::OpenMPOffloadMappingFlags>>(
+ mapFlag &
+ llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ) != 0;
+}
+
+static void
+processIndividualMap(MapInfoData &mapData, size_t mapDataIdx,
+ llvm::OpenMPIRBuilder::MapInfosTy &combinedInfo,
+ bool isTargetParams, int mapDataParentIdx = -1) {
+ // Declare Target Mappings are excluded from being marked as
+ // OMP_MAP_TARGET_PARAM as they are not passed as parameters, they're
+ // marked with OMP_MAP_PTR_AND_OBJ instead.
+ auto mapFlag = mapData.Types[mapDataIdx];
+ if (isTargetParams && !mapData.IsDeclareTarget[mapDataIdx])
+ mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TARGET_PARAM;
+
+ if (auto mapInfoOp =
+ dyn_cast<mlir::omp::MapInfoOp>(mapData.MapClause[mapDataIdx]))
+ if (mapInfoOp.getMapCaptureType().value() ==
+ mlir::omp::VariableCaptureKind::ByCopy &&
+ !checkIfPointerMap(mapFlag))
+ mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_LITERAL;
+
+ // if we're provided a mapDataParentIdx, then the data being mapped is
+ // part of a larger object (in a parent <-> member mapping) and in this
+ // case our BasePointer should be the parent.
+ if (mapDataParentIdx >= 0)
+ combinedInfo.BasePointers.emplace_back(
+ mapData.BasePointers[mapDataParentIdx]);
+ else
+ combinedInfo.BasePointers.emplace_back(mapData.BasePointers[mapDataIdx]);
+
+ combinedInfo.Pointers.emplace_back(mapData.Pointers[mapDataIdx]);
+ combinedInfo.DevicePointers.emplace_back(mapData.DevicePointers[mapDataIdx]);
+ combinedInfo.Names.emplace_back(mapData.Names[mapDataIdx]);
+ combinedInfo.Types.emplace_back(mapFlag);
+ combinedInfo.Sizes.emplace_back(mapData.Sizes[mapDataIdx]);
+}
+
static void processMapWithMembersOf(
LLVM::ModuleTranslation &moduleTranslation, llvm::IRBuilderBase &builder,
llvm::OpenMPIRBuilder &ompBuilder, DataLayout &dl,
llvm::OpenMPIRBuilder::MapInfosTy &combinedInfo, MapInfoData &mapData,
uint64_t mapDataIndex, bool isTargetParams) {
+ auto parentClause =
+ mlir::dyn_cast<mlir::omp::MapInfoOp>(mapData.MapClause[mapDataIndex]);
+ // If we have a partial map (no parent referneced in the map clauses of the
+ // directive, only members) and only a single member, we do not need to bind
+ // the map of the member to the parent, we can pass the member seperately.
+ if (parentClause.getMembers().size() == 1 && parentClause.getPartialMap()) {
+ auto memberClause = mlir::dyn_cast<mlir::omp::MapInfoOp>(
+ parentClause.getMembers()[0].getDefiningOp());
+ int memberDataIdx = getMapDataMemberIdx(mapData, memberClause);
+ // Primarily only scalars can be optimised this way it seems, array's
+ // need to be mapped as a regular record <-> member map even if partially
+ // mapping.
+ if (!mapData.BaseType[memberDataIdx]->isArrayTy()) {
+ processIndividualMap(mapData, memberDataIdx, combinedInfo, isTargetParams,
+ mapDataIndex);
+ return;
+ }
+ }
+
llvm::omp::OpenMPOffloadMappingFlags memberOfParentFlag =
mapParentWithMembers(moduleTranslation, builder, ompBuilder, dl,
combinedInfo, mapData, mapDataIndex, isTargetParams);
@@ -1959,6 +2086,79 @@ static void processMapWithMembersOf(
memberOfParentFlag);
}
+// This is a variation on Clang's GenerateOpenMPCapturedVars, which
+// generates different operation (e.g. load/store) combinations for
+// arguments to the kernel, based on map capture kinds which are then
+// utilised in the combinedInfo in place of the original Map value.
+static void
+createAlteredByCaptureMap(MapInfoData &mapData,
+ LLVM::ModuleTranslation &moduleTranslation,
+ llvm::IRBuilderBase &builder) {
+ for (size_t i = 0; i < mapData.MapClause.size(); ++i) {
+ // if it's declare target, skip it, it's handled seperately.
+ if (!mapData.IsDeclareTarget[i]) {
+ mlir::omp::VariableCaptureKind captureKind =
+ mlir::omp::VariableCaptureKind::ByRef;
+
+ auto mapOp =
+ mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(mapData.MapClause[i]);
+ captureKind = mapOp.getMapCaptureType().value_or(
+ mlir::omp::VariableCaptureKind::ByRef);
+
+ bool isPtrTy = checkIfPointerMap(
+ llvm::omp::OpenMPOffloadMappingFlags(mapOp.getMapType().value()));
+
+ // Currently handles array sectioning lowerbound case, but more
+ // logic may be required in the future. Clang invokes EmitLValue,
+ // which has specialised logic for special Clang types such as user
+ // defines, so it is possible we will have to extend this for
+ // structures or other complex types. As the general idea is that this
+ // function mimics some of the logic from Clang that we require for
+ // kernel argument passing from host -> device.
+ switch (captureKind) {
+ case mlir::omp::VariableCaptureKind::ByRef: {
+ llvm::Value *newV = mapData.Pointers[i];
+ std::vector<llvm::Value *> offsetIdx = calculateBoundsOffset(
+ moduleTranslation, builder, mapData.BaseType[i]->isArrayTy(),
+ mapOp.getBounds());
+ if (isPtrTy)
+ newV = builder.CreateLoad(builder.getPtrTy(), newV);
+
+ if (!offsetIdx.empty())
+ newV = builder.CreateInBoundsGEP(mapData.BaseType[i], newV, offsetIdx,
+ "array_offset");
+ mapData.Pointers[i] = newV;
+ } break;
+ case mlir::omp::VariableCaptureKind::ByCopy: {
+ llvm::Value *newV;
+ if (mapData.Pointers[i]->getType()->isPointerTy())
+ newV = builder.CreateLoad(mapData.BaseType[i], mapData.Pointers[i]);
+ else
+ newV = mapData.Pointers[i];
+
+ if (!isPtrTy) {
+ auto curInsert = builder.saveIP();
+ builder.restoreIP(findAllocaInsertPoint(builder, moduleTranslation));
+ auto *memTempAlloc =
+ builder.CreateAlloca(builder.getPtrTy(), nullptr, ".casted");
+ builder.restoreIP(curInsert);
+
+ builder.CreateStore(newV, memTempAlloc);
+ newV = builder.CreateLoad(builder.getPtrTy(), memTempAlloc);
+ }
+
+ mapData.Pointers[i] = newV;
+ mapData.BasePointers[i] = newV;
+ } break;
+ case mlir::omp::VariableCaptureKind::This:
+ case mlir::omp::VariableCaptureKind::VLAType:
+ mapData.MapClause[i]->emitOpError("Unhandled capture kind");
+ break;
+ }
+ }
+ }
+}
+
// Generate all map related information and fill the combinedInfo.
static void genMapInfos(llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation,
@@ -1968,6 +2168,20 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
const SmallVector<Value> &devPtrOperands = {},
const SmallVector<Value> &devAddrOperands = {},
bool isTargetParams = false) {
+ // We wish to modify some of the methods in which kernel arguments are
+ // passed based on their capture type by the target region, this can
+ // involve generating new loads and stores, which changes the
+ // MLIR value to LLVM value mapping, however, we only wish to do this
+ // locally for the current function/target and also avoid altering
+ // ModuleTranslation, so we remap the base pointer or pointer stored
+ // in the map infos corresponding MapInfoData, which is later accessed
+ // by genMapInfos and createTarget to help generate the kernel and
+ // kernel arg structure. It primarily becomes relevant in cases like
+ // bycopy, or byref range'd arrays. In the default case, we simply
+ // pass thee pointer byref as both basePointer and pointer.
+ if (!moduleTranslation.getOpenMPBuilder()->Config.isTargetDevice())
+ createAlteredByCaptureMap(mapData, moduleTranslation, builder);
+
llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
auto fail = [&combinedInfo]() -> void {
@@ -1985,12 +2199,8 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
// utilise the size from any component of MapInfoData, if we can't
// something is missing from the initial MapInfoData construction.
for (size_t i = 0; i < mapData.MapClause.size(); ++i) {
- // NOTE/TODO: We currently do not handle member mapping seperately from it's
- // parent or explicit mapping of a parent and member in the same operation,
- // this will need to change in the near future, for now we primarily handle
- // descriptor mapping from fortran, generalised as mapping record types
- // with implicit member maps. This lowering needs further generalisation to
- // fully support fortran derived types, and C/C++ structures and classes.
+ // NOTE/TODO: We currently do not support arbitrary depth record
+ // type mapping.
if (mapData.IsAMember[i])
continue;
@@ -2001,27 +2211,7 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
continue;
}
- // Declare Target Mappings are excluded from being marked as
- // OMP_MAP_TARGET_PARAM as they are not passed as parameters, they're
- // marked with OMP_MAP_PTR_AND_OBJ instead.
- auto mapFlag = mapData.Types[i];
- if (mapData.IsDeclareTarget[i])
- mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
- else if (isTargetParams)
- mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TARGET_PARAM;
-
- if (auto mapInfoOp = dyn_cast<mlir::omp::MapInfoOp>(mapData.MapClause[i]))
- if (mapInfoOp.getMapCaptureType().value() ==
- mlir::omp::VariableCaptureKind::ByCopy &&
- !mapInfoOp.getVarType().isa<LLVM::LLVMPointerType>())
- mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_LITERAL;
-
- combinedInfo.BasePointers.emplace_back(mapData.BasePointers[i]);
- combinedInfo.Pointers.emplace_back(mapData.Pointers[i]);
- combinedInfo.DevicePointers.emplace_back(mapData.DevicePointers[i]);
- combinedInfo.Names.emplace_back(mapData.Names[i]);
- combinedInfo.Types.emplace_back(mapFlag);
- combinedInfo.Sizes.emplace_back(mapData.Sizes[i]);
+ processIndividualMap(mapData, i, combinedInfo, isTargetParams);
}
auto findMapInfo = [&combinedInfo](llvm::Value *val, unsigned &index) {
@@ -2462,86 +2652,6 @@ createDeviceArgumentAccessor(MapInfoData &mapData, llvm::Argument &arg,
return builder.saveIP();
}
-// This is a variation on Clang's GenerateOpenMPCapturedVars, which
-// generates different operation (e.g. load/store) combinations for
-// arguments to the kernel, based on map capture kinds which are then
-// utilised in the combinedInfo in place of the original Map value.
-static void
-createAlteredByCaptureMap(MapInfoData &mapData,
- LLVM::ModuleTranslation &moduleTranslation,
- llvm::IRBuilderBase &builder) {
- for (size_t i = 0; i < mapData.MapClause.size(); ++i) {
- // if it's declare target, skip it, it's handled seperately.
- if (!mapData.IsDeclareTarget[i]) {
- mlir::omp::VariableCaptureKind captureKind =
- mlir::omp::VariableCaptureKind::ByRef;
-
- if (auto mapOp = mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(
- mapData.MapClause[i])) {
- captureKind = mapOp.getMapCaptureType().value_or(
- mlir::omp::VariableCaptureKind::ByRef);
- }
-
- switch (captureKind) {
- case mlir::omp::VariableCaptureKind::ByRef: {
- // Currently handles array sectioning lowerbound case, but more
- // logic may be required in the future. Clang invokes EmitLValue,
- // which has specialised logic for special Clang types such as user
- // defines, so it is possible we will have to extend this for
- // structures or other complex types. As the general idea is that this
- // function mimics some of the logic from Clang that we require for
- // kernel argument passing from host -> device.
- if (auto mapOp = mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(
- mapData.MapClause[i])) {
- if (!mapOp.getBounds().empty() && mapData.BaseType[i]->isArrayTy()) {
-
- std::vector<llvm::Value *> idx =
- std::vector<llvm::Value *>{builder.getInt64(0)};
- for (int i = mapOp.getBounds().size() - 1; i >= 0; --i) {
- if (auto boundOp =
- mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
- mapOp.getBounds()[i].getDefiningOp())) {
- idx.push_back(
- moduleTranslation.lookupValue(boundOp.getLowerBound()));
- }
- }
-
- mapData.Pointers[i] = builder.CreateInBoundsGEP(
- mapData.BaseType[i], mapData.Pointers[i], idx);
- }
- }
- } break;
- case mlir::omp::VariableCaptureKind::ByCopy: {
- llvm::Type *type = mapData.BaseType[i];
- llvm::Value *newV;
- if (mapData.Pointers[i]->getType()->isPointerTy())
- newV = builder.CreateLoad(type, mapData.Pointers[i]);
- else
- newV = mapData.Pointers[i];
-
- if (!type->isPointerTy()) {
- auto curInsert = builder.saveIP();
- builder.restoreIP(findAllocaInsertPoint(builder, moduleTranslation));
- auto *memTempAlloc =
- builder.CreateAlloca(builder.getPtrTy(), nullptr, ".casted");
- builder.restoreIP(curInsert);
-
- builder.CreateStore(newV, memTempAlloc);
- newV = builder.CreateLoad(builder.getPtrTy(), memTempAlloc);
- }
-
- mapData.Pointers[i] = newV;
- mapData.BasePointers[i] = newV;
- } break;
- case mlir::omp::VariableCaptureKind::This:
- case mlir::omp::VariableCaptureKind::VLAType:
- mapData.MapClause[i]->emitOpError("Unhandled capture kind");
- break;
- }
- }
- }
-}
-
static LogicalResult
convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation) {
@@ -2610,20 +2720,6 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
collectMapDataFromMapOperands(mapData, mapOperands, moduleTranslation, dl,
builder);
- // We wish to modify some of the methods in which kernel arguments are
- // passed based on their capture type by the target region, this can
- // involve generating new loads and stores, which changes the
- // MLIR value to LLVM value mapping, however, we only wish to do this
- // locally for the current function/target and also avoid altering
- // ModuleTranslation, so we remap the base pointer or pointer stored
- // in the map infos corresponding MapInfoData, which is later accessed
- // by genMapInfos and createTarget to help generate the kernel and
- // kernel arg structure. It primarily becomes relevant in cases like
- // bycopy, or byref range'd arrays. In the default case, we simply
- // pass thee pointer byref as both basePointer and pointer.
- if (!moduleTranslation.getOpenMPBuilder()->Config.isTargetDevice())
- createAlteredByCaptureMap(mapData, moduleTranslation, builder);
-
llvm::OpenMPIRBuilder::MapInfosTy combinedInfos;
auto genMapInfoCB = [&](llvm::OpenMPIRBuilder::InsertPointTy codeGenIP)
-> llvm::OpenMPIRBuilder::MapInfosTy & {
diff --git a/mlir/test/Dialect/OpenMP/ops.mlir b/mlir/test/Dialect/OpenMP/ops.mlir
index 65a704d18107b5..fe47190c4b9c3d 100644
--- a/mlir/test/Dialect/OpenMP/ops.mlir
+++ b/mlir/test/Dialect/OpenMP/ops.mlir
@@ -2019,8 +2019,6 @@ func.func @omp_requires_multiple() -> ()
return
}
-// -----
-
// CHECK-LABEL: @opaque_pointers_atomic_rwu
// CHECK-SAME: (%[[v:.*]]: !llvm.ptr, %[[x:.*]]: !llvm.ptr)
func.func @opaque_pointers_atomic_rwu(%v: !llvm.ptr, %x: !llvm.ptr) {
@@ -2127,10 +2125,10 @@ func.func @omp_target_update_data (%if_cond : i1, %device : si32, %map1: memref<
// CHECK-LABEL: omp_targets_is_allocatable
// CHECK-SAME: (%[[ARG0:.*]]: !llvm.ptr, %[[ARG1:.*]]: !llvm.ptr)
func.func @omp_targets_is_allocatable(%arg0: !llvm.ptr, %arg1: !llvm.ptr) -> () {
- // CHECK: %[[MAP0:.*]] = omp.map_info var_ptr(%[[ARG0]] : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
- %mapv1 = omp.map_info var_ptr(%arg0 : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
- // CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG1]] : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%[[MAP0]] : !llvm.ptr) -> !llvm.ptr {name = ""}
- %mapv2 = omp.map_info var_ptr(%arg1 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%mapv1 : !llvm.ptr) -> !llvm.ptr {name = ""}
+ // CHECK: %[[MAP0:.*]] = omp.map_info var_ptr(%[[ARG0]] : !llvm.ptr, i32) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+ %mapv1 = omp.map_info var_ptr(%arg0 : !llvm.ptr, i32) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !llvm.ptr {name = ""}
+ // CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG1]] : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%[[MAP0]] : !llvm.ptr : [0]) -> !llvm.ptr {name = ""}
+ %mapv2 = omp.map_info var_ptr(%arg1 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%mapv1 : !llvm.ptr : [0]) -> !llvm.ptr {name = ""}
// CHECK: omp.target map_entries(%[[MAP0]] -> {{.*}}, %[[MAP1]] -> {{.*}} : !llvm.ptr, !llvm.ptr)
omp.target map_entries(%mapv1 -> %arg2, %mapv2 -> %arg3 : !llvm.ptr, !llvm.ptr) {
^bb0(%arg2: !llvm.ptr, %arg3 : !llvm.ptr):
@@ -2138,3 +2136,21 @@ func.func @omp_targets_is_allocatable(%arg0: !llvm.ptr, %arg1: !llvm.ptr) -> ()
}
return
}
+
+// CHECK-LABEL: omp_map_with_members
+// CHECK-SAME: (%[[ARG0:.*]]: !llvm.ptr, %[[ARG1:.*]]: !llvm.ptr, %[[ARG2:.*]]: !llvm.ptr)
+func.func @omp_map_with_members(%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) -> () {
+ // CHECK: %[[MAP0:.*]] = omp.map_info var_ptr(%[[ARG0]] : !llvm.ptr, i32) map_clauses(to) capture(ByRef) -> !llvm.ptr {name = ""}
+ %mapv1 = omp.map_info var_ptr(%arg0 : !llvm.ptr, i32) map_clauses(to) capture(ByRef) -> !llvm.ptr {name = ""}
+
+ // CHECK: %[[MAP1:.*]] = omp.map_info var_ptr(%[[ARG1]] : !llvm.ptr, f32) map_clauses(to) capture(ByRef) -> !llvm.ptr {name = ""}
+ %mapv2 = omp.map_info var_ptr(%arg1 : !llvm.ptr, f32) map_clauses(to) capture(ByRef) -> !llvm.ptr {name = ""}
+
+ // CHECK: %[[MAP2:.*]] = omp.map_info var_ptr(%[[ARG2]] : !llvm.ptr, !llvm.struct<(i32, f32)>) map_clauses(to) capture(ByRef) members(%[[MAP0]], %[[MAP1]] : !llvm.ptr, !llvm.ptr : [0, 1]) -> !llvm.ptr {name = "", partial_map = true}
+ %mapv3 = omp.map_info var_ptr(%arg2 : !llvm.ptr, !llvm.struct<(i32, f32)>) map_clauses(to) capture(ByRef) members(%mapv1, %mapv2 : !llvm.ptr, !llvm.ptr : [0, 1]) -> !llvm.ptr {name = "", partial_map = true}
+
+ // CHECK: omp.target_enter_data map_entries(%[[MAP0]], %[[MAP1]], %[[MAP2]] : !llvm.ptr, !llvm.ptr, !llvm.ptr)
+ omp.target_enter_data map_entries(%mapv1, %mapv2, %mapv3 : !llvm.ptr, !llvm.ptr, !llvm.ptr){}
+
+ return
+}
diff --git a/mlir/test/Target/LLVMIR/omptarget-fortran-allocatable-types-host.mlir b/mlir/test/Target/LLVMIR/omptarget-fortran-allocatable-types-host.mlir
index 831cd05871c4e4..30a76795952f7d 100644
--- a/mlir/test/Target/LLVMIR/omptarget-fortran-allocatable-types-host.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-fortran-allocatable-types-host.mlir
@@ -2,10 +2,9 @@
// This test checks the offload sizes, map types and base pointers and pointers
// provided to the OpenMP kernel argument structure are correct when lowering
-// to LLVM-IR from MLIR when the fortran allocatables flag is switched on and
-// a fortran allocatable descriptor type is provided alongside the omp.map_info,
-// the test utilises mapping of array sections, full arrays and individual
-// allocated scalars.
+// to LLVM-IR from MLIR when a fortran allocatable descriptor type is provided
+// alongside the omp.map_info, the test utilises mapping of array sections,
+// full arrays and individual allocated scalars.
module attributes {omp.is_target_device = false} {
llvm.func @_QQmain() {
@@ -26,8 +25,8 @@ module attributes {omp.is_target_device = false} {
%14 = llvm.sub %11, %2 : i64
%15 = omp.bounds lower_bound(%7 : i64) upper_bound(%14 : i64) extent(%11 : i64) stride(%13 : i64) start_idx(%9 : i64) {stride_in_bytes = true}
%16 = llvm.getelementptr %3[0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
- %17 = omp.map_info var_ptr(%16 : !llvm.ptr, f32) map_clauses(tofrom) capture(ByRef) bounds(%15) -> !llvm.ptr {name = "full_arr"}
- %18 = omp.map_info var_ptr(%3 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>) map_clauses(tofrom) capture(ByRef) members(%17 : !llvm.ptr) -> !llvm.ptr {name = "full_arr"}
+ %17 = omp.map_info var_ptr(%16 : !llvm.ptr, f32) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%15) -> !llvm.ptr {name = "full_arr"}
+ %18 = omp.map_info var_ptr(%3 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>) map_clauses(tofrom) capture(ByRef) members(%17 : !llvm.ptr : [0]) -> !llvm.ptr {name = "full_arr"}
%19 = llvm.getelementptr %6[0, 7, %7, 0] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
%20 = llvm.load %19 : !llvm.ptr -> i64
%21 = llvm.getelementptr %6[0, 7, %7, 1] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
@@ -38,11 +37,11 @@ module attributes {omp.is_target_device = false} {
%26 = llvm.sub %0, %20 : i64
%27 = omp.bounds lower_bound(%25 : i64) upper_bound(%26 : i64) extent(%22 : i64) stride(%24 : i64) start_idx(%20 : i64) {stride_in_bytes = true}
%28 = llvm.getelementptr %6[0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
- %29 = omp.map_info var_ptr(%6 : !llvm.ptr, i32) var_ptr_ptr(%28 : !llvm.ptr) map_clauses(tofrom) capture(ByRef) bounds(%27) -> !llvm.ptr {name = "sect_arr(2:5)"}
- %30 = omp.map_info var_ptr(%6 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>) map_clauses(tofrom) capture(ByRef) members(%29 : !llvm.ptr) -> !llvm.ptr {name = "sect_arr(2:5)"}
+ %29 = omp.map_info var_ptr(%6 : !llvm.ptr, i32) var_ptr_ptr(%28 : !llvm.ptr) map_clauses(ptr_and_obj, tofrom) capture(ByRef) bounds(%27) -> !llvm.ptr {name = "sect_arr(2:5)"}
+ %30 = omp.map_info var_ptr(%6 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>) map_clauses(tofrom) capture(ByRef) members(%29 : !llvm.ptr : [0]) -> !llvm.ptr {name = "sect_arr(2:5)"}
%31 = llvm.getelementptr %5[0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>
- %32 = omp.map_info var_ptr(%5 : !llvm.ptr, f32) var_ptr_ptr(%31 : !llvm.ptr) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr {name = "scalar"}
- %33 = omp.map_info var_ptr(%5 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%32 : !llvm.ptr) -> !llvm.ptr {name = "scalar"}
+ %32 = omp.map_info var_ptr(%5 : !llvm.ptr, f32) var_ptr_ptr(%31 : !llvm.ptr) map_clauses(ptr_and_obj, tofrom) capture(ByRef) -> !llvm.ptr {name = "scalar"}
+ %33 = omp.map_info var_ptr(%5 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8)>) map_clauses(tofrom) capture(ByRef) members(%32 : !llvm.ptr : [0]) -> !llvm.ptr {name = "scalar"}
omp.target map_entries(%17 -> %arg0, %18 -> %arg1, %29 -> %arg2, %30 -> %arg3, %32 -> %arg4, %33 -> %arg5 : !llvm.ptr, !llvm.ptr, !llvm.ptr, !llvm.ptr, !llvm.ptr, !llvm.ptr) {
^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr, %arg3: !llvm.ptr, %arg4: !llvm.ptr, %arg5: !llvm.ptr):
omp.terminator
@@ -81,20 +80,19 @@ module attributes {omp.is_target_device = false} {
// CHECK: %[[ARR_SECT_SIZE2:.*]] = add i64 %[[ARR_SECT_SIZE3]], 1
// CHECK: %[[ARR_SECT_SIZE1:.*]] = mul i64 1, %[[ARR_SECT_SIZE2]]
// CHECK: %[[ARR_SECT_SIZE:.*]] = mul i64 %[[ARR_SECT_SIZE1]], 4
-// CHECK: %[[FULL_ARR_DESC_SIZE:.*]] = sdiv exact i64 sub (i64 ptrtoint (ptr getelementptr inbounds ({ ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]] }, ptr @_QFEfull_arr, i32 1) to i64), i64 ptrtoint (ptr @_QFEfull_arr to i64)), ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK: %[[LFULL_ARR:.*]] = load ptr, ptr @_QFEfull_arr, align 8
// CHECK: %[[FULL_ARR_PTR:.*]] = getelementptr inbounds float, ptr %[[LFULL_ARR]], i64 0
-// CHECK: %[[ARR_SECT_DESC_SIZE:.*]] = sdiv exact i64 sub (i64 ptrtoint (ptr getelementptr inbounds ({ ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]] }, ptr @_QFEsect_arr, i32 1) to i64), i64 ptrtoint (ptr @_QFEsect_arr to i64)), ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK: %[[ARR_SECT_OFFSET1:.*]] = mul i64 %[[ARR_SECT_OFFSET2]], 1
// CHECK: %[[LARR_SECT:.*]] = load ptr, ptr @_QFEsect_arr, align 8
// CHECK: %[[ARR_SECT_PTR:.*]] = getelementptr inbounds i32, ptr %[[LARR_SECT]], i64 %[[ARR_SECT_OFFSET1]]
+// CHECK: %[[SCALAR_PTR_LOAD:.*]] = load ptr, ptr %[[SCALAR_BASE]], align 8
+// CHECK: %[[FULL_ARR_DESC_SIZE:.*]] = sdiv exact i64 sub (i64 ptrtoint (ptr getelementptr inbounds ({ ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]] }, ptr @_QFEfull_arr, i32 1) to i64), i64 ptrtoint (ptr @_QFEfull_arr to i64)), ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
+// CHECK: %[[ARR_SECT_DESC_SIZE:.*]] = sdiv exact i64 sub (i64 ptrtoint (ptr getelementptr inbounds ({ ptr, i64, i32, i8, i8, i8, i8, [1 x [3 x i64]] }, ptr @_QFEsect_arr, i32 1) to i64), i64 ptrtoint (ptr @_QFEsect_arr to i64)), ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK: %[[SCALAR_DESC_SZ4:.*]] = getelementptr { ptr, i64, i32, i8, i8, i8, i8 }, ptr %[[SCALAR_ALLOCA]], i32 1
// CHECK: %[[SCALAR_DESC_SZ3:.*]] = ptrtoint ptr %[[SCALAR_DESC_SZ4]] to i64
// CHECK: %[[SCALAR_DESC_SZ2:.*]] = ptrtoint ptr %[[SCALAR_ALLOCA]] to i64
// CHECK: %[[SCALAR_DESC_SZ1:.*]] = sub i64 %[[SCALAR_DESC_SZ3]], %[[SCALAR_DESC_SZ2]]
// CHECK: %[[SCALAR_DESC_SZ:.*]] = sdiv exact i64 %[[SCALAR_DESC_SZ1]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
-// CHECK: %[[SCALAR_PTR_LOAD:.*]] = load ptr, ptr %[[SCALAR_BASE]], align 8
-// CHECK: %[[SCALAR_PTR:.*]] = getelementptr inbounds float, ptr %[[SCALAR_PTR_LOAD]], i64 0
// CHECK: %[[OFFLOADBASEPTRS:.*]] = getelementptr inbounds [9 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
// CHECK: store ptr @_QFEfull_arr, ptr %[[OFFLOADBASEPTRS]], align 8
@@ -143,6 +141,6 @@ module attributes {omp.is_target_device = false} {
// CHECK: %[[OFFLOADPTRS:.*]] = getelementptr inbounds [9 x ptr], ptr %.offload_ptrs, i32 0, i32 7
// CHECK: store ptr %[[SCALAR_ALLOCA]], ptr %[[OFFLOADPTRS]], align 8
// CHECK: %[[OFFLOADBASEPTRS:.*]] = getelementptr inbounds [9 x ptr], ptr %.offload_baseptrs, i32 0, i32 8
-// CHECK: store ptr %[[SCALAR_BASE]], ptr %[[OFFLOADBASEPTRS]], align 8
+// CHECK: store ptr %[[SCALAR_ALLOCA]], ptr %[[OFFLOADBASEPTRS]], align 8
// CHECK: %[[OFFLOADPTRS:.*]] = getelementptr inbounds [9 x ptr], ptr %.offload_ptrs, i32 0, i32 8
-// CHECK: store ptr %[[SCALAR_PTR]], ptr %[[OFFLOADPTRS]], align 8
+// CHECK: store ptr %[[SCALAR_PTR_LOAD]], ptr %[[OFFLOADPTRS]], align 8
diff --git a/mlir/test/Target/LLVMIR/omptarget-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
index b089d47f795df3..7f575e9b9be2b5 100644
--- a/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
@@ -66,20 +66,21 @@ llvm.func @_QPopenmp_target_data_region(%0 : !llvm.ptr) {
// CHECK: %[[VAL_2:.*]] = alloca [1 x ptr], align 8
// CHECK: br label %[[VAL_3:.*]]
// CHECK: entry: ; preds = %[[VAL_4:.*]]
+// CHECK: %[[ARR_OFFSET:.*]] = getelementptr inbounds [1024 x i32], ptr %[[ARR_DATA:.*]], i64 0, i64 0
// CHECK: %[[VAL_5:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_0]], i32 0, i32 0
-// CHECK: store ptr %[[VAL_6:.*]], ptr %[[VAL_5]], align 8
-// CHECK: %[[VAL_7:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
-// CHECK: store ptr %[[VAL_6]], ptr %[[VAL_7]], align 8
-// CHECK: %[[VAL_8:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_2]], i64 0, i64 0
-// CHECK: store ptr null, ptr %[[VAL_8]], align 8
-// CHECK: %[[VAL_9:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_0]], i32 0, i32 0
-// CHECK: %[[VAL_10:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
-// CHECK: call void @__tgt_target_data_begin_mapper(ptr @2, i64 -1, i32 1, ptr %[[VAL_9]], ptr %[[VAL_10]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr @.offload_mapnames, ptr null)
-// CHECK: %[[VAL_11:.*]] = getelementptr [1024 x i32], ptr %[[VAL_6]], i32 0, i64 0
-// CHECK: store i32 99, ptr %[[VAL_11]], align 4
-// CHECK: %[[VAL_12:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_0]], i32 0, i32 0
-// CHECK: %[[VAL_13:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
-// CHECK: call void @__tgt_target_data_end_mapper(ptr @2, i64 -1, i32 1, ptr %[[VAL_12]], ptr %[[VAL_13]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr @.offload_mapnames, ptr null)
+// CHECK: store ptr %[[ARR_DATA]], ptr %[[VAL_5]], align 8
+// CHECK: %[[VAL_6:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
+// CHECK: store ptr %[[ARR_OFFSET]], ptr %[[VAL_6]], align 8
+// CHECK: %[[VAL_7:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_2]], i64 0, i64 0
+// CHECK: store ptr null, ptr %[[VAL_7]], align 8
+// CHECK: %[[VAL_8:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_0]], i32 0, i32 0
+// CHECK: %[[VAL_9:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
+// CHECK: call void @__tgt_target_data_begin_mapper(ptr @2, i64 -1, i32 1, ptr %[[VAL_8]], ptr %[[VAL_9]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr @.offload_mapnames, ptr null)
+// CHECK: %[[VAL_10:.*]] = getelementptr [1024 x i32], ptr %[[ARR_DATA]], i32 0, i64 0
+// CHECK: store i32 99, ptr %[[VAL_10]], align 4
+// CHECK: %[[VAL_11:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_0]], i32 0, i32 0
+// CHECK: %[[VAL_12:.*]] = getelementptr inbounds [1 x ptr], ptr %[[VAL_1]], i32 0, i32 0
+// CHECK: call void @__tgt_target_data_end_mapper(ptr @2, i64 -1, i32 1, ptr %[[VAL_11]], ptr %[[VAL_12]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr @.offload_mapnames, ptr null)
// CHECK: ret void
// -----
@@ -153,16 +154,18 @@ llvm.func @_QPomp_target_enter_exit(%1 : !llvm.ptr, %3 : !llvm.ptr) {
// CHECK: entry: ; preds = %[[VAL_12:.*]]
// CHECK: br i1 %[[VAL_9]], label %[[VAL_13:.*]], label %[[VAL_14:.*]]
// CHECK: omp_if.then: ; preds = %[[VAL_11]]
+// CHECK: %[[ARR_OFFSET1:.*]] = getelementptr inbounds [1024 x i32], ptr %[[VAL_16:.*]], i64 0, i64 0
+// CHECK: %[[ARR_OFFSET2:.*]] = getelementptr inbounds [512 x i32], ptr %[[VAL_20:.*]], i64 0, i64 0
// CHECK: %[[VAL_15:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_3]], i32 0, i32 0
-// CHECK: store ptr %[[VAL_16:.*]], ptr %[[VAL_15]], align 8
+// CHECK: store ptr %[[VAL_16]], ptr %[[VAL_15]], align 8
// CHECK: %[[VAL_17:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_4]], i32 0, i32 0
-// CHECK: store ptr %[[VAL_16]], ptr %[[VAL_17]], align 8
+// CHECK: store ptr %[[ARR_OFFSET1]], ptr %[[VAL_17]], align 8
// CHECK: %[[VAL_18:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_5]], i64 0, i64 0
// CHECK: store ptr null, ptr %[[VAL_18]], align 8
// CHECK: %[[VAL_19:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_3]], i32 0, i32 1
-// CHECK: store ptr %[[VAL_20:.*]], ptr %[[VAL_19]], align 8
+// CHECK: store ptr %[[VAL_20]], ptr %[[VAL_19]], align 8
// CHECK: %[[VAL_21:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_4]], i32 0, i32 1
-// CHECK: store ptr %[[VAL_20]], ptr %[[VAL_21]], align 8
+// CHECK: store ptr %[[ARR_OFFSET2]], ptr %[[VAL_21]], align 8
// CHECK: %[[VAL_22:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_5]], i64 0, i64 1
// CHECK: store ptr null, ptr %[[VAL_22]], align 8
// CHECK: %[[VAL_23:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_3]], i32 0, i32 0
@@ -176,26 +179,28 @@ llvm.func @_QPomp_target_enter_exit(%1 : !llvm.ptr, %3 : !llvm.ptr) {
// CHECK: %[[VAL_27:.*]] = icmp sgt i32 %[[VAL_26]], 10
// CHECK: %[[VAL_28:.*]] = load i32, ptr %[[VAL_6]], align 4
// CHECK: br i1 %[[VAL_27]], label %[[VAL_29:.*]], label %[[VAL_30:.*]]
-// CHECK: omp_if.then1: ; preds = %[[VAL_25]]
+// CHECK: omp_if.then2: ; preds = %[[VAL_25]]
+// CHECK: %[[ARR_OFFSET3:.*]] = getelementptr inbounds [1024 x i32], ptr %[[VAL_16]], i64 0, i64 0
+// CHECK: %[[ARR_OFFSET4:.*]] = getelementptr inbounds [512 x i32], ptr %[[VAL_20]], i64 0, i64 0
// CHECK: %[[VAL_31:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_0]], i32 0, i32 0
// CHECK: store ptr %[[VAL_16]], ptr %[[VAL_31]], align 8
// CHECK: %[[VAL_32:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_1]], i32 0, i32 0
-// CHECK: store ptr %[[VAL_16]], ptr %[[VAL_32]], align 8
+// CHECK: store ptr %[[ARR_OFFSET3]], ptr %[[VAL_32]], align 8
// CHECK: %[[VAL_33:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_2]], i64 0, i64 0
// CHECK: store ptr null, ptr %[[VAL_33]], align 8
// CHECK: %[[VAL_34:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_0]], i32 0, i32 1
// CHECK: store ptr %[[VAL_20]], ptr %[[VAL_34]], align 8
// CHECK: %[[VAL_35:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_1]], i32 0, i32 1
-// CHECK: store ptr %[[VAL_20]], ptr %[[VAL_35]], align 8
+// CHECK: store ptr %[[ARR_OFFSET4]], ptr %[[VAL_35]], align 8
// CHECK: %[[VAL_36:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_2]], i64 0, i64 1
// CHECK: store ptr null, ptr %[[VAL_36]], align 8
// CHECK: %[[VAL_37:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_0]], i32 0, i32 0
// CHECK: %[[VAL_38:.*]] = getelementptr inbounds [2 x ptr], ptr %[[VAL_1]], i32 0, i32 0
// CHECK: call void @__tgt_target_data_end_mapper(ptr @3, i64 -1, i32 2, ptr %[[VAL_37]], ptr %[[VAL_38]], ptr @.offload_sizes.1, ptr @.offload_maptypes.2, ptr @.offload_mapnames.3, ptr null)
// CHECK: br label %[[VAL_39:.*]]
-// CHECK: omp_if.else5: ; preds = %[[VAL_25]]
+// CHECK: omp_if.else8: ; preds = %[[VAL_25]]
// CHECK: br label %[[VAL_39]]
-// CHECK: omp_if.end6: ; preds = %[[VAL_30]], %[[VAL_29]]
+// CHECK: omp_if.end9: ; preds = %[[VAL_30]], %[[VAL_29]]
// CHECK: ret void
// -----
diff --git a/mlir/test/Target/LLVMIR/omptarget-record-type-mapping-host.mlir b/mlir/test/Target/LLVMIR/omptarget-record-type-mapping-host.mlir
new file mode 100644
index 00000000000000..12e98acf6d82e5
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/omptarget-record-type-mapping-host.mlir
@@ -0,0 +1,63 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// This test checks the offload sizes, map types and base pointers and pointers
+// provided to the OpenMP kernel argument structure are correct when lowering
+// to LLVM-IR from MLIR when performing explicit member mapping of a recrod type
+// (C++/C class/structure, Fortran derived type) where only members of the record
+// type are mapped.
+
+module attributes {omp.is_target_device = false} {
+llvm.func @_QQmain() {
+ %0 = llvm.mlir.constant(10 : index) : i64
+ %1 = llvm.mlir.constant(4 : index) : i64
+ %2 = llvm.mlir.constant(1 : index) : i64
+ %3 = llvm.mlir.constant(1 : i64) : i64
+ %4 = llvm.alloca %3 x !llvm.struct<(f32, array<10 x i32>, i32)> : (i64) -> !llvm.ptr
+ %5 = llvm.mlir.constant(2 : i32) : i32
+ %6 = llvm.getelementptr %4[0, 2] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(f32, array<10 x i32>, i32)>
+ %7 = omp.map_info var_ptr(%6 : !llvm.ptr, i32) map_clauses(tofrom) capture(ByRef) -> !llvm.ptr
+ %8 = llvm.mlir.constant(1 : i32) : i32
+ %9 = llvm.getelementptr %4[0, 1] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(f32, array<10 x i32>, i32)>
+ %10 = omp.bounds lower_bound(%2 : i64) upper_bound(%1 : i64) extent(%0 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %11 = omp.map_info var_ptr(%9 : !llvm.ptr, !llvm.array<10 x i32>) map_clauses(tofrom) capture(ByRef) bounds(%10) -> !llvm.ptr
+ %12 = omp.map_info var_ptr(%4 : !llvm.ptr, !llvm.struct<(f32, array<10 x i32>, i32)>) map_clauses(tofrom) capture(ByRef) members(%7, %11 : !llvm.ptr, !llvm.ptr : [2, 1]) -> !llvm.ptr {partial_map = true}
+ omp.target map_entries(%7 -> %arg0, %11 -> %arg1, %12 -> %arg2 : !llvm.ptr, !llvm.ptr, !llvm.ptr) {
+ ^bb0(%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr):
+ omp.terminator
+ }
+ llvm.return
+ }
+}
+
+// CHECK: @.offload_sizes = private unnamed_addr constant [3 x i64] [i64 0, i64 4, i64 16]
+// CHECK: @.offload_maptypes = private unnamed_addr constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710659]
+
+// CHECK: define void @_QQmain()
+// CHECK: %[[ALLOCA:.*]] = alloca { float, [10 x i32], i32 }, i64 1, align 8
+// CHECK: %[[MEMBER_ACCESS_1:.*]] = getelementptr { float, [10 x i32], i32 }, ptr %[[ALLOCA]], i32 0, i32 2
+// CHECK: %[[MEMBER_ACCESS_2:.*]] = getelementptr { float, [10 x i32], i32 }, ptr %[[ALLOCA]], i32 0, i32 1
+
+// CHECK: %[[LAST_MEMBER:.*]] = getelementptr inbounds [10 x i32], ptr %[[MEMBER_ACCESS_2]], i64 0, i64 1
+// CHECK: %[[FIRST_MEMBER:.*]] = getelementptr i32, ptr %[[MEMBER_ACCESS_1]], i64 1
+// CHECK: %[[FIRST_MEMBER_OFF:.*]] = ptrtoint ptr %[[FIRST_MEMBER]] to i64
+// CHECK: %[[SECOND_MEMBER_OFF:.*]] = ptrtoint ptr %[[LAST_MEMBER]] to i64
+// CHECK: %[[MEMBER_DIFF:.*]] = sub i64 %[[FIRST_MEMBER_OFF]], %[[SECOND_MEMBER_OFF]]
+// CHECK: %[[OFFLOAD_SIZE:.*]] = sdiv exact i64 %[[MEMBER_DIFF]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
+
+// CHECK: %[[BASE_PTR_ARR:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+// CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR]], align 8
+// CHECK: %[[PTR_ARR:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+// CHECK: store ptr %[[LAST_MEMBER]], ptr %[[PTR_ARR]], align 8
+// CHECK: %[[SIZE_ARR:.*]] = getelementptr inbounds [3 x i64], ptr %.offload_sizes, i32 0, i32 0
+// CHECK: store i64 %[[OFFLOAD_SIZE]], ptr %[[SIZE_ARR]], align 8
+
+// CHECK: %[[BASE_PTR_ARR_2:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 1
+// CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_2]], align 8
+// CHECK: %[[PTR_ARR_2:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 1
+// CHECK: store ptr %[[MEMBER_ACCESS_1]], ptr %[[PTR_ARR_2]], align 8
+
+// CHECK: %[[BASE_PTR_ARR_3:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_baseptrs, i32 0, i32 2
+// CHECK: store ptr %[[ALLOCA]], ptr %[[BASE_PTR_ARR_3]], align 8
+// CHECK: %[[PTR_ARR_3:.*]] = getelementptr inbounds [3 x ptr], ptr %.offload_ptrs, i32 0, i32 2
+// CHECK: store ptr %[[LAST_MEMBER]], ptr %[[PTR_ARR_3]], align 8
+
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-1.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-1.f90
new file mode 100644
index 00000000000000..cb03708554fed0
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-1.f90
@@ -0,0 +1,45 @@
+! Offloading test checking interaction of an
+! explicit derived type mapping when mapped
+! to target and assinging one derived type
+! to another
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ end type scalar
+
+ type(scalar) :: in
+ type(scalar) :: out
+ in%ix = 10
+ in%rx = 2.0
+ in%zx = (2, 10)
+
+ !$omp target map(from:out) map(to:in)
+ out = in
+ !$omp end target
+
+ print*, in%ix
+ print*, in%rx
+ write (*,*) in%zx
+
+ print*, out%ix
+ print*, out%rx
+ write (*,*) out%zx
+end program main
+
+!CHECK: 10
+!CHECK: 2.
+!CHECK: (2.,10.)
+!CHECK: 10
+!CHECK: 2.
+!CHECK: (2.,10.)
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-2.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-2.f90
new file mode 100644
index 00000000000000..0095b0fdf86a66
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-2.f90
@@ -0,0 +1,60 @@
+! Offloading test checking interaction of an
+! explicit derived type mapping when mapped to
+! target and assigning to individual members
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ integer(4) :: array(5)
+ end type scalar
+
+ type(scalar) :: out
+ type(scalar) :: in
+
+ in%ix = 10
+ in%rx = 2.0
+ in%zx = (2, 10)
+
+ do i = 1, 5
+ in%array(i) = i
+ end do
+
+ !$omp target map(from:out) map(to:in)
+ out%ix = in%ix
+ out%rx = in%rx
+ out%zx = in%zx
+
+ do i = 1, 5
+ out%array(i) = in%array(i)
+ end do
+ !$omp end target
+
+ print*, in%ix
+ print*, in%rx
+ print*, in%array
+ write (*,*) in%zx
+
+ print*, out%ix
+ print*, out%rx
+ print*, out%array
+ write (*,*) out%zx
+end program main
+
+!CHECK: 10
+!CHECK: 2.
+!CHECK: 1 2 3 4 5
+!CHECK: (2.,10.)
+!CHECK: 10
+!CHECK: 2.
+!CHECK: 1 2 3 4 5
+!CHECK: (2.,10.)
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-1.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-1.f90
new file mode 100644
index 00000000000000..f57e2c70d155e9
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-1.f90
@@ -0,0 +1,46 @@
+! Offloading test checking interaction of an
+! implicit derived type mapping when mapped
+! to target and assinging one derived type
+! to another
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ end type scalar
+
+ type(scalar) :: in
+ type(scalar) :: out
+ in%ix = 10
+ in%rx = 2.0
+ in%zx = (2, 10)
+
+ !$omp target map(from:out)
+ out = in
+ !$omp end target
+
+ print*, in%ix
+ print*, in%rx
+ write (*,*) in%zx
+
+ print*, out%ix
+ print*, out%rx
+ write (*,*) out%zx
+ end program main
+
+!CHECK: 10
+!CHECK: 2.
+!CHECK: (2.,10.)
+!CHECK: 10
+!CHECK: 2.
+!CHECK: (2.,10.)
+
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-2.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-2.f90
new file mode 100644
index 00000000000000..92d3454d462a7d
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-derived-type-full-implicit-2.f90
@@ -0,0 +1,61 @@
+! Offloading test checking interaction of an
+! explicit derived type mapping when mapped
+! to target and assinging one derived type
+! to another
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ integer(4) :: array(5)
+ end type scalar
+
+ type(scalar) :: out
+ type(scalar) :: in
+
+ in%ix = 10
+ in%rx = 2.0
+ in%zx = (2, 10)
+
+ do i = 1, 5
+ in%array(i) = i
+ end do
+
+ !$omp target
+ out%ix = in%ix
+ out%rx = in%rx
+ out%zx = in%zx
+
+ do i = 1, 5
+ out%array(i) = in%array(i)
+ end do
+ !$omp end target
+
+ print*, in%ix
+ print*, in%rx
+ print*, in%array
+ write (*,*) in%zx
+
+ print*, out%ix
+ print*, out%rx
+ print*, out%array
+ write (*,*) out%zx
+end program main
+
+!CHECK: 10
+!CHECK: 2.
+!CHECK: 1 2 3 4 5
+!CHECK: (2.,10.)
+!CHECK: 10
+!CHECK: 2.
+!CHECK: 1 2 3 4 5
+!CHECK: (2.,10.)
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit-update.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit-update.f90
new file mode 100644
index 00000000000000..3b3ec96b9babfd
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit-update.f90
@@ -0,0 +1,49 @@
+! Offloading test checking interaction of an
+! explicit derived type member mapping of
+! an array with bounds when mapped to
+! target using a combination of update,
+! enter and exit directives.
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ integer(4) :: array(10)
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ do I = 1, 10
+ scalar_arr%array(I) = I + I
+ end do
+
+ !$omp target enter data map(to: scalar_arr%array(3:6))
+
+ ! overwrite our target data with an update.
+ do I = 1, 10
+ scalar_arr%array(I) = 10
+ end do
+
+ !$omp target update to(scalar_arr%array(3:6))
+
+ ! The compiler/runtime is less friendly about read/write out of
+ ! bounds when using enter and exit, we have to specifically loop
+ ! over the correct range
+ !$omp target
+ do i=3,6
+ scalar_arr%array(i) = scalar_arr%array(i) + i
+ end do
+ !$omp end target
+
+ !$omp target exit data map(from: scalar_arr%array(3:6))
+
+ print*, scalar_arr%array
+end program
+
+!CHECK: 10 10 13 14 15 16 10 10 10 10
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit.f90
new file mode 100644
index 00000000000000..5f7e9f94682659
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-arr-bounds-member-enter-exit.f90
@@ -0,0 +1,49 @@
+! Offloading test checking interaction of an
+! explicit derived type member mapping of
+! an array with bounds when mapped to
+! target using a combination of enter and
+! exit directives.
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ integer(4) :: array(10)
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ do I = 1, 10
+ scalar_arr%array(I) = I + I
+ end do
+
+ !$omp target enter data map(to: scalar_arr%array(3:6))
+
+ ! Shouldn't overwrite data already locked in
+ ! on target via enter, which will then be
+ ! overwritten by our exit
+ do I = 1, 10
+ scalar_arr%array(I) = 10
+ end do
+
+ ! The compiler/runtime is less friendly about read/write out of
+ ! bounds when using enter and exit, we have to specifically loop
+ ! over the correct range
+ !$omp target
+ do i=3,6
+ scalar_arr%array(i) = scalar_arr%array(i) + i
+ end do
+ !$omp end target
+
+ !$omp target exit data map(from: scalar_arr%array(3:6))
+
+ print*, scalar_arr%array
+end program
+
+!CHECK: 10 10 9 12 15 18 10 10 10 10
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-explicit-individual-array-member.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-explicit-individual-array-member.f90
new file mode 100644
index 00000000000000..907b16ffedf524
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-explicit-individual-array-member.f90
@@ -0,0 +1,33 @@
+! Offloading test checking interaction of an
+! explicit derived type member mapping of
+! an array when mapped to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ !$omp target map(tofrom:scalar_arr%array_y)
+ do i = 1, 10
+ scalar_arr%array_y(i) = i
+ end do
+ !$omp end target
+
+ print *, scalar_arr%array_y
+end program main
+
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-3D-member-bounds.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-3D-member-bounds.f90
new file mode 100644
index 00000000000000..110fb648980cdc
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-3D-member-bounds.f90
@@ -0,0 +1,45 @@
+! Offloading test checking interaction of an
+! explicit derived type member mapping of
+! two arrays with explicit bounds when
+! mapped to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ integer(4) :: array_x(3,3,3)
+ real(4) :: break_1
+ integer(4) :: array_y(3,3,3)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ do i = 1, 3
+ do j = 1, 3
+ do k = 1, 3
+ scalar_arr%array_x(i, j, k) = 42
+ scalar_arr%array_y(i, j, k) = 0 ! Will get overwritten by garbage values in target
+ end do
+ end do
+ end do
+
+ !$omp target map(tofrom:scalar_arr%array_x(1:3, 1:3, 2:2), scalar_arr%array_y(1:3, 1:3, 1:3))
+ do j = 1, 3
+ do k = 1, 3
+ scalar_arr%array_y(k, j, 2) = scalar_arr%array_x(k, j, 2)
+ end do
+ end do
+ !$omp end target
+
+ print *, scalar_arr%array_y
+end program main
+
+!CHECK: 0 0 0 0 0 0 0 0 0 42 42 42 42 42 42 42 42 42 0 0 0 0 0 0 0 0
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member-bounds.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member-bounds.f90
new file mode 100644
index 00000000000000..b7f6e2ddfb3bcb
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member-bounds.f90
@@ -0,0 +1,38 @@
+! Offloading test checking interaction of an
+! explicit derived type member mapping of
+! two arrays with explicit bounds when
+! mapped to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ do i = 1, 10
+ scalar_arr%array_x(i) = i
+ end do
+
+ !$omp target map(tofrom:scalar_arr%array_x(3:6), scalar_arr%array_y(3:6))
+ do i = 1, 10
+ scalar_arr%array_y(i) = scalar_arr%array_x(i)
+ end do
+ !$omp end target
+
+ print*, scalar_arr%array_y
+end program main
+
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member.f90
new file mode 100644
index 00000000000000..c44a58dbebc8e4
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-array-member.f90
@@ -0,0 +1,39 @@
+! Offloading test checking interaction of an
+! derived type mapping of two explicit array
+! members to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr
+
+ do i = 1, 10
+ scalar_arr%array_x(i) = i
+ end do
+
+ !$omp target map(tofrom:scalar_arr%array_x, scalar_arr%array_y)
+ do i = 1, 10
+ scalar_arr%array_y(i) = scalar_arr%array_x(i)
+ end do
+ !$omp end target
+
+ print*, scalar_arr%array_x
+ print*, scalar_arr%array_y
+end program main
+
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-member.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-member.f90
new file mode 100644
index 00000000000000..a4205bc0179e20
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-dtype-multi-explicit-member.f90
@@ -0,0 +1,33 @@
+! Offloading test checking interaction of an
+! derived type mapping of two explicit
+! members to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ real(4) :: ry = 1.0
+ end type scalar
+
+ type(scalar) :: scalar_struct
+
+ !$omp target map(from:scalar_struct%rx, scalar_struct%ry)
+ scalar_struct%rx = 21.0
+ scalar_struct%ry = 27.0
+ !$omp end target
+
+ print*, scalar_struct%rx
+ print*, scalar_struct%ry
+end program main
+
+!CHECK: 21.
+!CHECK: 27.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-2.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-2.f90
new file mode 100644
index 00000000000000..d4ee3665a1e6a5
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-2.f90
@@ -0,0 +1,39 @@
+! Offloading test checking interaction of an
+! enter and exit map of an array of scalars
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ integer :: array(10)
+
+ do I = 1, 10
+ array(I) = I + I
+ end do
+
+ !$omp target enter data map(to: array)
+
+ ! Shouldn't overwrite data already locked in
+ ! on target via enter, this will then be
+ ! overwritten by our exit
+ do I = 1, 10
+ array(I) = 10
+ end do
+
+ !$omp target
+ do i=1,10
+ array(i) = array(i) + i
+ end do
+ !$omp end target
+
+ !$omp target exit data map(from: array)
+
+ print*, array
+end program
+
+!CHECK: 3 6 9 12 15 18 21 24 27 30
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-bounds.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-bounds.f90
new file mode 100644
index 00000000000000..93c5189a47afcd
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-array-bounds.f90
@@ -0,0 +1,44 @@
+! Offloading test checking interaction of an
+! enter and exit map of an array of scalars
+! with specified bounds
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+
+program main
+ integer :: array(10)
+
+ do I = 1, 10
+ array(I) = I + I
+ end do
+
+ !$omp target enter data map(to: array(3:6))
+
+ ! Shouldn't overwrite data already locked in
+ ! on target via enter, which will then be
+ ! overwritten by our exit
+ do I = 1, 10
+ array(I) = 10
+ end do
+
+ ! The compiler/runtime is less lenient about read/write out of
+ ! bounds when using enter and exit, we have to specifically loop
+ ! over the correctly mapped range
+ !$omp target
+ do i=3,6
+ array(i) = array(i) + i
+ end do
+ !$omp end target
+
+ !$omp target exit data map(from: array(3:6))
+
+ print *, array
+end program
+
+!CHECK: 10 10 9 12 15 18 10 10 10 10
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-scalar.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-scalar.f90
new file mode 100644
index 00000000000000..f3a18614f37122
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-enter-exit-scalar.f90
@@ -0,0 +1,33 @@
+! Offloading test checking interaction of an
+! enter and exit map of an scalar
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ integer :: scalar
+ scalar = 10
+
+ !$omp target enter data map(to: scalar)
+
+ !ignored, as we've already attached
+ scalar = 20
+
+ !$omp target
+ scalar = scalar + 50
+ !$omp end target
+
+ !$omp target exit data map(from: scalar)
+
+ ! not the answer one may expect, but it is the same
+ ! answer Clang gives so we are correctly on par with
+ ! Clang for the moment.
+ print *, scalar
+end program
+
+!CHECK: 10
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-individual-dtype-member-map.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-individual-dtype-member-map.f90
new file mode 100644
index 00000000000000..4cdf41db70d43b
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-individual-dtype-member-map.f90
@@ -0,0 +1,33 @@
+! Offloading test checking interaction of an
+! single explicit member map from a single
+! derived type.
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ real :: test
+ type :: scalar
+ integer(4) :: ix = 0
+ real(4) :: rx = 0.0
+ complex(4) :: zx = (0,0)
+ real(4) :: ry = 1.0
+ end type scalar
+
+ type(scalar) :: scalar_struct
+ scalar_struct%rx = 2.0
+ test = 21.0
+
+ !$omp target map(from:scalar_struct%rx)
+ scalar_struct%rx = test
+ !$omp end target
+
+ print *, scalar_struct%rx
+end program main
+
+!CHECK: 21.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-explicit-member.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-explicit-member.f90
new file mode 100644
index 00000000000000..b080b437e381b5
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-explicit-member.f90
@@ -0,0 +1,35 @@
+! Offloading test checking interaction of two
+! derived type's with one explicit member
+! each being mapped with bounds to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr1
+ type(scalar_array) :: scalar_arr2
+
+ !$omp target map(tofrom:scalar_arr1%break_1, scalar_arr2%break_3)
+ scalar_arr2%break_3 = 10
+ scalar_arr1%break_1 = 15
+ !$omp end target
+
+ print*, scalar_arr1%break_1
+ print*, scalar_arr2%break_3
+end program main
+
+!CHECK: 15.
+!CHECK: 10.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-individual-member-array-1D-bounds.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-individual-member-array-1D-bounds.f90
new file mode 100644
index 00000000000000..1e96207c2baaff
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-individual-member-array-1D-bounds.f90
@@ -0,0 +1,39 @@
+! Offloading test checking interaction of two
+! derived type's with a single explicit array
+! member each being mapped with bounds to
+! target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr1
+ type(scalar_array) :: scalar_arr2
+
+
+ !$omp target map(tofrom:scalar_arr1%array_x(3:6), scalar_arr2%array_x(3:6))
+ do i = 1, 10
+ scalar_arr2%array_x(i) = i
+ scalar_arr1%array_x(i) = i
+ end do
+ !$omp end target
+
+ print*, scalar_arr1%array_x
+ print*, scalar_arr2%array_x
+end program main
+
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-1.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-1.f90
new file mode 100644
index 00000000000000..fbe6b305464a64
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-1.f90
@@ -0,0 +1,35 @@
+! Offloading test checking interaction of two
+! derived type's with a mix of explicit and
+! implicit member mapping to target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr1
+ type(scalar_array) :: scalar_arr2
+
+ !$omp target map(tofrom:scalar_arr1%break_1)
+ scalar_arr2%break_3 = 10
+ scalar_arr1%break_1 = 15
+ !$omp end target
+
+ print*, scalar_arr1%break_1
+ print*, scalar_arr2%break_3
+end program main
+
+!CHECK: 15.
+!CHECK: 10.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-2.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-2.f90
new file mode 100644
index 00000000000000..d01fd1e595118a
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-mixed-implicit-explicit-capture-2.f90
@@ -0,0 +1,41 @@
+! Offloading test checking interaction of two
+! derived type's with a mix of explicit and
+! implicit member mapping of arrays to target
+! one with bounds.
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr1
+ type(scalar_array) :: scalar_arr2
+
+ do i = 1, 10
+ scalar_arr1%array_x(i) = i
+ end do
+
+ !$omp target map(tofrom:scalar_arr2%array_x(3:6))
+ do i = 1, 10
+ scalar_arr2%array_x(i) = scalar_arr1%array_x(i)
+ end do
+ !$omp end target
+
+ print*, scalar_arr1%array_x
+ print*, scalar_arr2%array_x
+end program main
+
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
diff --git a/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-multi-member-array-1D-bounds.f90 b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-multi-member-array-1D-bounds.f90
new file mode 100644
index 00000000000000..ed350c54dcb259
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/target-map-two-dtype-multi-member-array-1D-bounds.f90
@@ -0,0 +1,51 @@
+! Offloading test checking interaction of two
+! derived type's with two explicit array
+! members each being mapped with bounds to
+! target
+! REQUIRES: flang, amdgcn-amd-amdhsa
+! UNSUPPORTED: nvptx64-nvidia-cuda
+! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
+! UNSUPPORTED: aarch64-unknown-linux-gnu
+! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
+! UNSUPPORTED: x86_64-pc-linux-gnu
+! UNSUPPORTED: x86_64-pc-linux-gnu-LTO
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ type :: scalar_array
+ real(4) :: break_0
+ real(4) :: array_x(10)
+ real(4) :: break_1
+ real(4) :: array_y(10)
+ real(4) :: break_3
+ end type scalar_array
+
+ type(scalar_array) :: scalar_arr1
+ type(scalar_array) :: scalar_arr2
+
+ do i = 1, 10
+ scalar_arr1%array_x(i) = i
+ scalar_arr2%array_x(i) = i
+ end do
+
+ !$omp target map(tofrom:scalar_arr1%array_x(3:6), scalar_arr1%array_y(3:6), scalar_arr2%array_x(3:6), scalar_arr2%array_y(3:6))
+ do i = 1, 10
+ scalar_arr2%array_y(i) = scalar_arr1%array_x(i)
+ end do
+
+ do i = 1, 10
+ scalar_arr1%array_y(i) = scalar_arr2%array_x(i)
+ end do
+ !$omp end target
+
+ print*, scalar_arr1%array_x
+ print*, scalar_arr2%array_y
+
+ print*, scalar_arr2%array_x
+ print*, scalar_arr1%array_y
+end program main
+
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
+!CHECK: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
+!CHECK: 0. 0. 3. 4. 5. 6. 0. 0. 0. 0.
More information about the flang-commits
mailing list