[Mlir-commits] [mlir] [Flang][OpenMP][MLIR] Initial array section mapping MLIR -> LLVM-IR lowering utilising omp.bounds (PR #68689)
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Tue Oct 10 04:30:03 PDT 2023
https://github.com/agozillon created https://github.com/llvm/llvm-project/pull/68689
This patch seeks to add initial lowering of OpenMP array sections within target region map clauses from MLIR to LLVM IR.
This patch seeks to support fixed sized arrays initially, before looking toward assumed size and shaped arrays.
Although, assumed size works in some fashion (dummy arguments) with some minor alterations to the OMPEarlyOutliner, so it is possible changes made in the IsolatedFromAbove series may allow this to work with no further required patches.
It utilises the generated omp.bounds to calculate the size of the mapped OpenMP array (both for sectioned and un-sectioned arrays) as well as the offset to be passed to the kernel argument structure.
Alongside these changes some refactoring of how map data is handled is attempted, using a new MapData structure to keep track of information utilised in the lowering of mapped values.
The initial addition of a more complex createDeviceArgumentAccessor that utilises capture kinds similarly to (and loosely based on) Clang to generate different kernel argument accesses is also added.
A similar function for altering how the kernel argument is passed to the kernel argument structure on the host is also utilised (createAlteredByCaptureMap), which allows modification of the pointer/basePointer based on their capture (and bounds information). It's of note ByRef, is the default for explicit mappings and ByCopy will be the default for implicit captures, so the former is currently tested in this patch and the latter is not for the moment.
>From 1e02ffe581b3cf49d609c0f36ef58ddc29c8b95d Mon Sep 17 00:00:00 2001
From: Andrew Gozillon <Andrew.Gozillon at amd.com>
Date: Tue, 10 Oct 2023 06:09:54 -0500
Subject: [PATCH] [Flang][OpenMP][MLIR] Initial array section mapping MLIR ->
LLVM-IR lowering utilising omp.bounds
This patch seeks to add initial lowering of OpenMP array
sections within target region map clauses from MLIR to
LLVM IR.
This patch seeks to support fixed sized arrays initially,
before looking toward assumed size and shaped arrays.
Although, assumed size works in some fashion (dummy
arguments) with some minor alterations to the OMPEarlyOutliner,
so it is possible changes made in the IsolatedFromAbove series
may allow this to work with no further required patches.
It utilises the generated omp.bounds to calculate the size
of the mapped OpenMP array (both for sectioned and
un-sectioned arrays) as well as the offset to be passed to
the kernel argument structure.
Alongside these changes some refactoring of how map data is
handled is attempted, using a new MapData structure to keep
track of information utilised in the lowering of mapped values.
The initial addition of a more complex
createDeviceArgumentAccessor that utilises capture
kinds similarly to (and loosely based on) Clang to
generate different kernel argument accesses is also
added.
A similar function for altering how the kernel argument
is passed to the kernel argument structure on the host
is also utilised (createAlteredByCaptureMap), which
allows modification of the pointer/basePointer based
on their capture (and bounds information). It's of note
ByRef, is the default for explicit mappings and ByCopy
will be the default for implicit captures, so the former
is currently tested in this patch and the latter is not
for the moment.
---
.../OpenMP/OpenMPToLLVMIRTranslation.cpp | 512 +++++++++++++-----
.../omptarget-array-sectioning-device.mlir | 78 +++
.../omptarget-array-sectioning-host.mlir | 80 +++
mlir/test/Target/LLVMIR/omptarget-llvm.mlir | 61 ++-
.../basic-target-region-1D-array-section.f90 | 27 +
.../basic-target-region-3D-array-section.f90 | 39 ++
.../fortran/basic-target-region-3D-array.f90 | 45 ++
.../fortran/basic-target-region-array.f90 | 29 +
8 files changed, 713 insertions(+), 158 deletions(-)
create mode 100644 mlir/test/Target/LLVMIR/omptarget-array-sectioning-device.mlir
create mode 100644 mlir/test/Target/LLVMIR/omptarget-array-sectioning-host.mlir
create mode 100644 openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
create mode 100644 openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index 1ec3bb8e7562a9e..e2a408ffd810e2e 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -1528,13 +1528,6 @@ convertOmpThreadprivate(Operation &opInst, llvm::IRBuilderBase &builder,
return success();
}
-int64_t getSizeInBytes(DataLayout &DL, const Type &type) {
- if (isa<LLVM::LLVMPointerType>(type))
- return DL.getTypeSize(cast<LLVM::LLVMPointerType>(type).getElementType());
-
- return 0;
-}
-
static llvm::OffloadEntriesInfoManager::OMPTargetDeviceClauseKind
convertToDeviceClauseKind(mlir::omp::DeclareTargetDeviceType deviceClause) {
switch (deviceClause) {
@@ -1629,13 +1622,153 @@ getRefPtrIfDeclareTarget(mlir::Value value,
return nullptr;
}
+// A small helper structure to contain data gathered
+// for map lowering and coalese it into one area and
+// avoiding extra computations such as searches in the
+// llvm module for lowered mapped varibles or checking
+// if something is declare target (and retrieving the
+// value).
+struct MapData {
+ bool isDeclareTarget = false;
+ mlir::Operation *mapClause;
+ llvm::Value *basePointer;
+ llvm::Value *pointer;
+ llvm::Value *kernelValue;
+ llvm::Type *underlyingType;
+ llvm::Value *sizeInBytes;
+};
+
+uint64_t getArrayElementSizeInBits(LLVM::LLVMArrayType arrTy, DataLayout &dl) {
+ if (auto nestedArrTy = llvm::dyn_cast_if_present<LLVM::LLVMArrayType>(
+ arrTy.getElementType()))
+ return getArrayElementSizeInBits(nestedArrTy, dl);
+ return dl.getTypeSizeInBits(arrTy.getElementType());
+}
+
+// This function calculates the size to be offloaded for a specified type, given
+// its associated map clause (which can contain bounds information which affects
+// the total size), this size is calculated based on the underlying element type
+// e.g. given a 1-D array of ints, we will calculate the size from the integer
+// type * number of elements in the array. This size can be used in other
+// calculations but is ultimately used as an argument to the OpenMP runtimes
+// kernel argument structure which is generated through the combinedInfo data
+// structures.
+// This function is somewhat equivalent to Clang's getExprTypeSize inside of
+// CGOpenMPRuntime.cpp.
+llvm::Value *getSizeInBytes(DataLayout &dl, const mlir::Type &type,
+ Operation *clauseOp, llvm::IRBuilderBase &builder,
+ LLVM::ModuleTranslation &moduleTranslation) {
+ // utilising getTypeSizeInBits instead of getTypeSize as getTypeSize gives
+ // the size in inconsistent byte or bit format.
+ uint64_t underlyingTypeSzInBits = dl.getTypeSizeInBits(type);
+ if (auto ptrTy = llvm::dyn_cast_if_present<LLVM::LLVMPointerType>(type)) {
+ underlyingTypeSzInBits = dl.getTypeSizeInBits(ptrTy.getElementType());
+ if (auto arrTy = llvm::dyn_cast_if_present<LLVM::LLVMArrayType>(
+ ptrTy.getElementType())) {
+ underlyingTypeSzInBits = getArrayElementSizeInBits(arrTy, dl);
+ }
+ }
+
+ if (auto memberClause =
+ mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(clauseOp)) {
+ // This calculates the size to transfer based on bounds and the underlying
+ // element type, provided bounds have been specified (Fortran
+ // pointers/allocatables/target and arrays that have sections specified fall
+ // into this as well).
+ if (!memberClause.getBounds().empty()) {
+ llvm::Value *elementCount = nullptr;
+ for (auto bounds : memberClause.getBounds()) {
+ if (auto boundOp = mlir::dyn_cast_if_present<mlir::omp::DataBoundsOp>(
+ bounds.getDefiningOp())) {
+
+ if (!elementCount) {
+ elementCount = builder.CreateAdd(
+ builder.CreateSub(
+ moduleTranslation.lookupValue(boundOp.getUpperBound()),
+ moduleTranslation.lookupValue(boundOp.getLowerBound())),
+ builder.getInt64(1));
+ } else {
+ elementCount = builder.CreateMul(
+ elementCount,
+ builder.CreateAdd(
+ builder.CreateSub(
+ moduleTranslation.lookupValue(boundOp.getUpperBound()),
+ moduleTranslation.lookupValue(boundOp.getLowerBound())),
+ builder.getInt64(1)));
+ }
+ }
+ }
+
+ // The size in bytes x number of elements, the sizeInBytes stored is
+ // the underyling types size, e.g. if ptr<i32>, it'll be the i32's
+ // size, so we do some on the fly runtime math to get the size in
+ // bytes from the extent (ub - lb) * sizeInBytes. NOTE: This may need
+ // some adjustment for members with more complex types.
+ return builder.CreateMul(elementCount,
+ builder.getInt64(underlyingTypeSzInBits / 8));
+ }
+ }
+
+ return builder.getInt64(underlyingTypeSzInBits / 8);
+}
+
+// Get the underlying LLVM type, this will bypass the pointer and
+// access the underlying type. Which bypasses llvm's opaque pointers
+// to get the underlying type via MLIR.
+llvm::Type *getLLVMIRType(mlir::Type inputType,
+ LLVM::ModuleTranslation &moduleTranslation) {
+ llvm::Type *type = moduleTranslation.convertType(inputType);
+ if (auto pointerType =
+ llvm::dyn_cast<mlir::omp::PointerLikeType>(inputType)) {
+ if (auto eleType = pointerType.getElementType()) {
+ type = moduleTranslation.convertType(eleType);
+ }
+ }
+ return type;
+}
+
+void collectMapDataFromMapOperands(llvm::SmallVectorImpl<MapData> &mapData,
+ llvm::SmallVectorImpl<Value> &mapOperands,
+ LLVM::ModuleTranslation &moduleTranslation,
+ DataLayout &dl,
+ llvm::IRBuilderBase &builder) {
+ auto getMapData = [&](Value val, Operation *clauseOp) -> MapData {
+ MapData map;
+ if (llvm::Value *refPtr = getRefPtrIfDeclareTarget(
+ val, moduleTranslation)) { // declare target
+ map.kernelValue = moduleTranslation.lookupValue(val);
+ map.isDeclareTarget = true;
+ map.basePointer = refPtr;
+ map.pointer = map.kernelValue;
+ } else { // regular mapped variable
+ map.kernelValue = moduleTranslation.lookupValue(val);
+ map.isDeclareTarget = false;
+ map.basePointer = map.kernelValue;
+ map.pointer = map.kernelValue;
+ }
+
+ map.sizeInBytes =
+ getSizeInBytes(dl, val.getType(), clauseOp, builder, moduleTranslation);
+ map.underlyingType = getLLVMIRType(val.getType(), moduleTranslation);
+ map.mapClause = clauseOp;
+
+ return map;
+ };
+
+ for (mlir::Value mapValue : mapOperands) {
+ if (auto mapOp = mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(
+ mapValue.getDefiningOp())) {
+ mapData.push_back(getMapData(mapOp.getVarPtr(), mapOp));
+ }
+ }
+}
+
// Generate all map related information and fill the combinedInfo.
static void genMapInfos(llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation,
- DataLayout &DL,
+ DataLayout &dl,
llvm::OpenMPIRBuilder::MapInfosTy &combinedInfo,
- const SmallVector<Value> &mapOperands,
- const ArrayAttr &mapTypes,
+ llvm::SmallVectorImpl<MapData> &mapData,
const SmallVector<Value> &devPtrOperands = {},
const SmallVector<Value> &devAddrOperands = {},
bool isTargetParams = false) {
@@ -1650,58 +1783,39 @@ static void genMapInfos(llvm::IRBuilderBase &builder,
combinedInfo.Names.clear();
};
- auto findMapInfo = [&combinedInfo](llvm::Value *val, unsigned &index) {
- index = 0;
- for (auto basePtr : combinedInfo.BasePointers) {
- if (basePtr == val)
- return true;
- index++;
- }
- return false;
- };
-
- unsigned index = 0;
- for (const auto &mapOp : mapOperands) {
- // Unlike dev_ptr and dev_addr operands these map operands point
- // to a map entry operation which contains further information
- // on the variable being mapped and how it should be mapped.
- auto mapInfoOp =
- mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp());
-
- // TODO: Only LLVMPointerTypes are handled.
- if (!mapInfoOp.getType().isa<LLVM::LLVMPointerType>())
- return fail();
-
- llvm::Value *mapOpValue =
- moduleTranslation.lookupValue(mapInfoOp.getVarPtr());
-
- llvm::Value *refPtr =
- getRefPtrIfDeclareTarget(mapInfoOp.getVarPtr(), moduleTranslation);
-
- combinedInfo.BasePointers.emplace_back(refPtr ? refPtr : mapOpValue);
- combinedInfo.Pointers.emplace_back(mapOpValue);
- combinedInfo.DevicePointers.emplace_back(
- llvm::OpenMPIRBuilder::DeviceInfoTy::None);
- combinedInfo.Names.emplace_back(LLVM::createMappingInformation(
- mapInfoOp.getVarPtr().getLoc(), *ompBuilder));
-
- auto mapFlag = llvm::omp::OpenMPOffloadMappingFlags(
- mapTypes[index].cast<IntegerAttr>().getUInt());
+ for (MapData &mapValue : mapData) {
+ auto mapOp = mlir::dyn_cast<mlir::omp::MapInfoOp>(mapValue.mapClause);
+ auto mapFlag =
+ llvm::omp::OpenMPOffloadMappingFlags(mapOp.getMapType().value());
// 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.
- if (refPtr)
+ if (mapValue.isDeclareTarget)
mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_PTR_AND_OBJ;
else if (isTargetParams)
mapFlag |= llvm::omp::OpenMPOffloadMappingFlags::OMP_MAP_TARGET_PARAM;
+ combinedInfo.BasePointers.emplace_back(mapValue.basePointer);
+ combinedInfo.Pointers.emplace_back(mapValue.pointer);
+ combinedInfo.DevicePointers.emplace_back(
+ llvm::OpenMPIRBuilder::DeviceInfoTy::None);
+ combinedInfo.Names.emplace_back(
+ LLVM::createMappingInformation(mapOp.getLoc(), *ompBuilder));
combinedInfo.Types.emplace_back(mapFlag);
- combinedInfo.Sizes.emplace_back(
- builder.getInt64(getSizeInBytes(DL, mapInfoOp.getVarPtr().getType())));
- index++;
+ combinedInfo.Sizes.emplace_back(mapValue.sizeInBytes);
}
+ auto findMapInfo = [&combinedInfo](llvm::Value *val, unsigned &index) {
+ index = 0;
+ for (llvm::Value *basePtr : combinedInfo.BasePointers) {
+ if (basePtr == val)
+ return true;
+ index++;
+ }
+ return false;
+ };
+
auto addDevInfos = [&, fail](auto devOperands, auto devOpType) -> void {
for (const auto &devOp : devOperands) {
// TODO: Only LLVMPointerTypes are handled.
@@ -1747,19 +1861,6 @@ convertOmpTargetData(Operation *op, llvm::IRBuilderBase &builder,
llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
- auto getMapTypes = [](mlir::OperandRange mapOperands,
- mlir::MLIRContext *ctx) {
- SmallVector<mlir::Attribute> mapTypes;
- for (auto mapValue : mapOperands) {
- if (mapValue.getDefiningOp()) {
- auto mapOp =
- mlir::dyn_cast<mlir::omp::MapInfoOp>(mapValue.getDefiningOp());
- mapTypes.push_back(mapOp.getMapTypeAttr());
- }
- }
- return mlir::ArrayAttr::get(ctx, mapTypes);
- };
-
LogicalResult result =
llvm::TypeSwitch<Operation *, LogicalResult>(op)
.Case([&](omp::DataOp dataOp) {
@@ -1773,8 +1874,6 @@ convertOmpTargetData(Operation *op, llvm::IRBuilderBase &builder,
deviceID = intAttr.getInt();
mapOperands = dataOp.getMapOperands();
- mapTypes =
- getMapTypes(dataOp.getMapOperands(), dataOp->getContext());
useDevPtrOperands = dataOp.getUseDevicePtr();
useDevAddrOperands = dataOp.getUseDeviceAddr();
return success();
@@ -1793,8 +1892,6 @@ convertOmpTargetData(Operation *op, llvm::IRBuilderBase &builder,
deviceID = intAttr.getInt();
RTLFn = llvm::omp::OMPRTL___tgt_target_data_begin_mapper;
mapOperands = enterDataOp.getMapOperands();
- mapTypes = getMapTypes(enterDataOp.getMapOperands(),
- enterDataOp->getContext());
return success();
})
.Case([&](omp::ExitDataOp exitDataOp) {
@@ -1812,8 +1909,6 @@ convertOmpTargetData(Operation *op, llvm::IRBuilderBase &builder,
RTLFn = llvm::omp::OMPRTL___tgt_target_data_end_mapper;
mapOperands = exitDataOp.getMapOperands();
- mapTypes = getMapTypes(exitDataOp.getMapOperands(),
- exitDataOp->getContext());
return success();
})
.Default([&](Operation *op) {
@@ -1826,17 +1921,20 @@ convertOmpTargetData(Operation *op, llvm::IRBuilderBase &builder,
using InsertPointTy = llvm::OpenMPIRBuilder::InsertPointTy;
+ llvm::SmallVector<MapData> mapData;
+ collectMapDataFromMapOperands(mapData, mapOperands, moduleTranslation, DL,
+ builder);
+
// Fill up the arrays with all the mapped variables.
llvm::OpenMPIRBuilder::MapInfosTy combinedInfo;
auto genMapInfoCB =
[&](InsertPointTy codeGenIP) -> llvm::OpenMPIRBuilder::MapInfosTy & {
builder.restoreIP(codeGenIP);
- if (auto DataOp = dyn_cast<omp::DataOp>(op)) {
- genMapInfos(builder, moduleTranslation, DL, combinedInfo, mapOperands,
- mapTypes, useDevPtrOperands, useDevAddrOperands);
+ if (auto dataOp = dyn_cast<omp::DataOp>(op)) {
+ genMapInfos(builder, moduleTranslation, DL, combinedInfo, mapData,
+ useDevPtrOperands, useDevAddrOperands);
} else {
- genMapInfos(builder, moduleTranslation, DL, combinedInfo, mapOperands,
- mapTypes);
+ genMapInfos(builder, moduleTranslation, DL, combinedInfo, mapData);
}
return combinedInfo;
};
@@ -1988,38 +2086,53 @@ static bool targetOpSupported(Operation &opInst) {
}
static void
-handleDeclareTargetMapVar(llvm::ArrayRef<Value> mapOperands,
+handleDeclareTargetMapVar(llvm::ArrayRef<MapData> mapData,
LLVM::ModuleTranslation &moduleTranslation,
llvm::IRBuilderBase &builder) {
- for (const mlir::Value &mapOp : mapOperands) {
- auto mapInfoOp =
- mlir::dyn_cast<mlir::omp::MapInfoOp>(mapOp.getDefiningOp());
- llvm::Value *mapOpValue =
- moduleTranslation.lookupValue(mapInfoOp.getVarPtr());
- if (auto *declareTarget = getRefPtrIfDeclareTarget(mapInfoOp.getVarPtr(),
- moduleTranslation)) {
- // The user's iterator will get invalidated if we modify an element,
+ for (const auto &mapValue : mapData) {
+ // In the case of declare target mapped variables, the basePointer is
+ // the reference pointer generated by the convertDeclareTargetAttr
+ // method. Whereas the kernelValue is the original variable, so for
+ // the device we must replace all uses of this original global variable
+ // (stored in kernelValue) with the reference pointer (stored in
+ // basePointer for declare target mapped variables), as for device the
+ // data is mapped into this reference pointer and should be loaded
+ // from it, the original variable is discarded. On host both exist and
+ // metadata is generated (elsewhere in the convertDeclareTargetAttr)
+ // function to link the two variables in the runtime and then both the
+ // reference pointer and the pointer are assigned in the kernel argument
+ // structure for the host.
+ if (mapValue.isDeclareTarget) {
+ // The users iterator will get invalidated if we modify an element,
// so we populate this vector of uses to alter each user on an individual
// basis to emit its own load (rather than one load for all).
llvm::SmallVector<llvm::User *> userVec;
- for (llvm::User *user : mapOpValue->users())
+ for (llvm::User *user : mapValue.kernelValue->users())
userVec.push_back(user);
- for (llvm::User *user : userVec) {
+ for (auto *user : userVec) {
if (auto *insn = dyn_cast<llvm::Instruction>(user)) {
- auto *load = builder.CreateLoad(
- moduleTranslation.convertType(mapInfoOp.getVarPtr().getType()),
- declareTarget);
+ auto *load = builder.CreateLoad(mapValue.basePointer->getType(),
+ mapValue.basePointer);
load->moveBefore(insn);
- user->replaceUsesOfWith(mapOpValue, load);
+ user->replaceUsesOfWith(mapValue.kernelValue, load);
}
}
}
}
}
+// This currently implements a very light version of Clang's
+// EmitParmDecl's handling of direct argument handling as well
+// as a portion of the argument access generation based on
+// capture types found at the end of emitOutlinedFunctionPrologue
+// in Clang. The indirect path handling of EmitParmDecl's may be
+// required for future work, but a direct 1-to-1 copy doesn't seem
+// possible as the logic is rather scattered throughout Clang's
+// lowering and perhaps we wish to deviate slightly.
static llvm::IRBuilderBase::InsertPoint
-createDeviceArgumentAccessor(llvm::Argument &arg, llvm::Value *input,
+createDeviceArgumentAccessor(llvm::SmallVectorImpl<MapData> &mapData,
+ llvm::Argument &arg, llvm::Value *input,
llvm::Value *&retVal, llvm::IRBuilderBase &builder,
llvm::OpenMPIRBuilder &ompBuilder,
LLVM::ModuleTranslation &moduleTranslation,
@@ -2027,22 +2140,152 @@ createDeviceArgumentAccessor(llvm::Argument &arg, llvm::Value *input,
llvm::IRBuilderBase::InsertPoint codeGenIP) {
builder.restoreIP(allocaIP);
- llvm::Value *addr =
+ mlir::omp::VariableCaptureKind capture =
+ mlir::omp::VariableCaptureKind::ByRef;
+ llvm::Type *inputType = input->getType();
+
+ // Find the associated MapData entry for the current input
+ for (const MapData &map : mapData) {
+ if (map.kernelValue == input) {
+ if (auto mapOp =
+ mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(map.mapClause)) {
+ capture = mapOp.getMapCaptureType().value_or(
+ mlir::omp::VariableCaptureKind::ByRef);
+ }
+
+ inputType = map.underlyingType;
+ break;
+ }
+ }
+
+ unsigned int allocaAS = ompBuilder.M.getDataLayout().getAllocaAddrSpace();
+ unsigned int defaultAS =
+ ompBuilder.M.getDataLayout().getProgramAddressSpace();
+
+ // Create the alloca for the argument the current point.
+ llvm::Value *v =
builder.CreateAlloca(arg.getType()->isPointerTy()
? arg.getType()
: llvm::Type::getInt64Ty(builder.getContext()),
- ompBuilder.M.getDataLayout().getAllocaAddrSpace());
- llvm::Value *addrAscast =
- builder.CreatePointerBitCastOrAddrSpaceCast(addr, input->getType());
- builder.CreateStore(&arg, addrAscast);
+ allocaAS);
+
+ if (allocaAS != defaultAS) {
+ v = builder.CreatePointerBitCastOrAddrSpaceCast(
+ v, arg.getType()->getPointerTo(defaultAS));
+ }
+
+ builder.CreateStore(&arg, v);
builder.restoreIP(codeGenIP);
- retVal = builder.CreateLoad(arg.getType(), addrAscast);
+ switch (capture) {
+ case mlir::omp::VariableCaptureKind::ByCopy: {
+ if (inputType->isPointerTy()) {
+ retVal = v;
+ return builder.saveIP();
+ }
+
+ // Ignore conversions like int -> uint.
+ if (v->getType() == inputType->getPointerTo()) {
+ retVal = v;
+ return builder.saveIP();
+ }
+
+ assert(false && "Currently unsupported OMPTargetVarCaptureByCopy Type");
+ break;
+ }
+ case mlir::omp::VariableCaptureKind::ByRef: {
+ retVal = builder.CreateAlignedLoad(
+ v->getType(), v,
+ ompBuilder.M.getDataLayout().getPrefTypeAlign(v->getType()));
+ break;
+ }
+ case mlir::omp::VariableCaptureKind::This:
+ case mlir::omp::VariableCaptureKind::VLAType:
+ assert(false && "Currently unsupported capture kind");
+ break;
+ }
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(llvm::SmallVectorImpl<MapData> &mapData,
+ LLVM::ModuleTranslation &moduleTranslation,
+ llvm::IRBuilderBase &builder) {
+ for (MapData &mapValue : mapData) {
+ // if it's declare target, skip it, it's handled seperately.
+ if (!mapValue.isDeclareTarget) {
+ mlir::omp::VariableCaptureKind captureKind =
+ mlir::omp::VariableCaptureKind::ByRef;
+
+ if (auto mapOp = mlir::dyn_cast_if_present<mlir::omp::MapInfoOp>(
+ mapValue.mapClause)) {
+ 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>(
+ mapValue.mapClause)) {
+ if (!mapOp.getBounds().empty() &&
+ mapValue.underlyingType->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()));
+ }
+ }
+
+ mapValue.pointer = builder.CreateInBoundsGEP(
+ mapValue.underlyingType, mapValue.pointer, idx);
+ }
+ }
+ } break;
+ case mlir::omp::VariableCaptureKind::ByCopy: {
+ llvm::Type *type = mapValue.underlyingType;
+ llvm::Value *newV = builder.CreateLoad(type, mapValue.pointer);
+
+ if (!type->isPointerTy()) {
+ auto curInsert = builder.saveIP();
+ builder.restoreIP(findAllocaInsertPoint(builder, moduleTranslation));
+ auto *memTempAlloc =
+ builder.CreateAlloca(builder.getInt8PtrTy(), nullptr, ".casted");
+ builder.restoreIP(curInsert);
+
+ builder.CreateStore(newV, memTempAlloc);
+ newV = builder.CreateLoad(builder.getInt8PtrTy(), memTempAlloc);
+ }
+
+ mapValue.pointer = newV;
+ mapValue.basePointer = newV;
+ } break;
+ case mlir::omp::VariableCaptureKind::This:
+ case mlir::omp::VariableCaptureKind::VLAType:
+ mapValue.mapClause->emitOpError("Unhandled capture kind");
+ break;
+ }
+ }
+ }
+}
+
static LogicalResult
convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation) {
@@ -2053,26 +2296,6 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
auto targetOp = cast<omp::TargetOp>(opInst);
auto &targetRegion = targetOp.getRegion();
- // This function filters out kernel data that will not show up as kernel
- // input arguments to the generated kernel function but will still need
- // explicitly mapped through supplying information to the OpenMP runtime
- // (declare target). It also prepares some data used for generating the
- // kernel and populating the associated OpenMP runtime data structures.
- auto getKernelArguments =
- [&](const llvm::SetVector<Value> &operandSet,
- llvm::SmallVectorImpl<llvm::Value *> &llvmInputs) {
- for (Value operand : operandSet) {
- if (!getRefPtrIfDeclareTarget(operand, moduleTranslation))
- llvmInputs.push_back(moduleTranslation.lookupValue(operand));
- }
- };
-
- llvm::SetVector<Value> operandSet;
- getUsedValuesDefinedAbove(targetRegion, operandSet);
-
- llvm::SmallVector<llvm::Value *> inputs;
- getKernelArguments(operandSet, inputs);
-
LogicalResult bodyGenStatus = success();
using InsertPointTy = llvm::OpenMPIRBuilder::InsertPointTy;
@@ -2106,31 +2329,32 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
llvm::OpenMPIRBuilder::InsertPointTy allocaIP =
findAllocaInsertPoint(builder, moduleTranslation);
- DataLayout DL = DataLayout(opInst.getParentOfType<ModuleOp>());
- SmallVector<Value> mapOperands = targetOp.getMapOperands();
-
- auto getMapTypes = [](mlir::OperandRange mapOperands,
- mlir::MLIRContext *ctx) {
- SmallVector<mlir::Attribute> mapTypes;
- for (auto mapValue : mapOperands) {
- if (mapValue.getDefiningOp()) {
- auto mapOp =
- mlir::dyn_cast<mlir::omp::MapInfoOp>(mapValue.getDefiningOp());
- mapTypes.push_back(mapOp.getMapTypeAttr());
- }
- }
- return mlir::ArrayAttr::get(ctx, mapTypes);
- };
-
- ArrayAttr mapTypes =
- getMapTypes(targetOp.getMapOperands(), targetOp->getContext());
+ DataLayout dl = DataLayout(opInst.getParentOfType<ModuleOp>());
+ llvm::SmallVector<Value> mapOperands = targetOp.getMapOperands();
+ llvm::SmallVector<MapData> mapData;
+ 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 MapData, 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 & {
builder.restoreIP(codeGenIP);
- genMapInfos(builder, moduleTranslation, DL, combinedInfos, mapOperands,
- mapTypes, {}, {}, true);
+ genMapInfos(builder, moduleTranslation, dl, combinedInfos, mapData, {}, {},
+ true);
return combinedInfos;
};
@@ -2149,20 +2373,28 @@ convertOmpTarget(Operation &opInst, llvm::IRBuilderBase &builder,
return codeGenIP;
}
- return createDeviceArgumentAccessor(arg, input, retVal, builder,
+ return createDeviceArgumentAccessor(mapData, arg, input, retVal, builder,
*ompBuilder, moduleTranslation,
allocaIP, codeGenIP);
};
+ llvm::SmallVector<llvm::Value *, 4> kernelInput;
+ for (auto &data : mapData) {
+ // declare target arguments are not passed to kernels as arguments
+ if (!data.isDeclareTarget) {
+ kernelInput.push_back(data.kernelValue);
+ }
+ }
+
builder.restoreIP(moduleTranslation.getOpenMPBuilder()->createTarget(
ompLoc, allocaIP, builder.saveIP(), entryInfo, defaultValTeams,
- defaultValThreads, inputs, genMapInfoCB, bodyCB, argAccessorCB));
+ defaultValThreads, kernelInput, genMapInfoCB, bodyCB, argAccessorCB));
// Remap access operations to declare target reference pointers for the
// device, essentially generating extra loadop's as necessary
if (moduleTranslation.getOpenMPBuilder()->Config.isTargetDevice()) {
SmallVector<Value> mapOperands = targetOp.getMapOperands();
- handleDeclareTargetMapVar(llvm::ArrayRef(mapOperands), moduleTranslation,
+ handleDeclareTargetMapVar(llvm::ArrayRef(mapData), moduleTranslation,
builder);
}
return bodyGenStatus;
diff --git a/mlir/test/Target/LLVMIR/omptarget-array-sectioning-device.mlir b/mlir/test/Target/LLVMIR/omptarget-array-sectioning-device.mlir
new file mode 100644
index 000000000000000..eee1374eea57078
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/omptarget-array-sectioning-device.mlir
@@ -0,0 +1,78 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+module attributes {omp.is_target_device = true} {
+ llvm.func @_3d_target_array_section() attributes {} {
+ %0 = llvm.mlir.addressof @_QFEinarray : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>
+ %1 = llvm.mlir.addressof @_QFEoutarray : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>
+ %2 = llvm.mlir.constant(1 : index) : i64
+ %3 = llvm.mlir.constant(0 : index) : i64
+ %4 = llvm.mlir.constant(2 : index) : i64
+ %5 = omp.bounds lower_bound(%3 : i64) upper_bound(%4 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %6 = llvm.mlir.constant(0 : index) : i64
+ %7 = llvm.mlir.constant(2 : index) : i64
+ %8 = omp.bounds lower_bound(%6 : i64) upper_bound(%7 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %9 = llvm.mlir.constant(1 : index) : i64
+ %10 = llvm.mlir.constant(1 : index) : i64
+ %11 = omp.bounds lower_bound(%9 : i64) upper_bound(%10 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %12 = omp.map_info var_ptr(%0 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) map_clauses(tofrom) capture(ByRef) bounds(%5, %8, %11) -> !llvm.ptr<array<3 x array<3 x array<3 x i32>>>> {name = "inarray(1:3,1:3,2:2)"}
+ %13 = llvm.mlir.constant(1 : index) : i64
+ %14 = llvm.mlir.constant(0 : index) : i64
+ %15 = llvm.mlir.constant(2 : index) : i64
+ %16 = omp.bounds lower_bound(%14 : i64) upper_bound(%15 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %17 = llvm.mlir.constant(0 : index) : i64
+ %18 = llvm.mlir.constant(2 : index) : i64
+ %19 = omp.bounds lower_bound(%17 : i64) upper_bound(%18 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %20 = llvm.mlir.constant(0 : index) : i64
+ %21 = llvm.mlir.constant(2 : index) : i64
+ %22 = omp.bounds lower_bound(%20 : i64) upper_bound(%21 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %23 = omp.map_info var_ptr(%1 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) map_clauses(tofrom) capture(ByRef) bounds(%16, %19, %22) -> !llvm.ptr<array<3 x array<3 x array<3 x i32>>>> {name = "outarray(1:3,1:3,1:3)"}
+ omp.target map_entries(%12, %23 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) {
+ %24 = llvm.mlir.constant(1 : i64) : i64
+ %25 = llvm.mlir.constant(1 : i64) : i64
+ %26 = llvm.mlir.constant(0 : i64) : i64
+ %27 = llvm.mlir.constant(1 : i64) : i64
+ %28 = llvm.mlir.constant(1 : i64) : i64
+ %29 = llvm.mlir.constant(0 : i64) : i64
+ %30 = llvm.mlir.constant(2 : i64) : i64
+ %31 = llvm.mlir.constant(1 : i64) : i64
+ %32 = llvm.mlir.constant(1 : i64) : i64
+ %33 = llvm.getelementptr %0[0, %32, %29, %26] : (!llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, i64, i64, i64) -> !llvm.ptr<i32>
+ %34 = llvm.load %33 : !llvm.ptr<i32>
+ %35 = llvm.mlir.constant(1 : i64) : i64
+ %36 = llvm.mlir.constant(1 : i64) : i64
+ %37 = llvm.mlir.constant(0 : i64) : i64
+ %38 = llvm.mlir.constant(1 : i64) : i64
+ %39 = llvm.mlir.constant(1 : i64) : i64
+ %40 = llvm.mlir.constant(0 : i64) : i64
+ %41 = llvm.mlir.constant(2 : i64) : i64
+ %42 = llvm.mlir.constant(1 : i64) : i64
+ %43 = llvm.mlir.constant(1 : i64) : i64
+ %44 = llvm.getelementptr %1[0, %43, %40, %37] : (!llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, i64, i64, i64) -> !llvm.ptr<i32>
+ llvm.store %34, %44 : !llvm.ptr<i32>
+ omp.terminator
+ }
+ llvm.return
+ }
+ llvm.mlir.global internal @_QFEinarray() {addr_space = 0 : i32} : !llvm.array<3 x array<3 x array<3 x i32>>> {
+ %0 = llvm.mlir.zero : !llvm.array<3 x array<3 x array<3 x i32>>>
+ llvm.return %0 : !llvm.array<3 x array<3 x array<3 x i32>>>
+ }
+ llvm.mlir.global internal @_QFEoutarray() {addr_space = 0 : i32} : !llvm.array<3 x array<3 x array<3 x i32>>> {
+ %0 = llvm.mlir.zero : !llvm.array<3 x array<3 x array<3 x i32>>>
+ llvm.return %0 : !llvm.array<3 x array<3 x array<3 x i32>>>
+ }
+}
+
+
+// CHECK: define weak_odr protected void @__omp_offloading_fd00_4944f87__3d_target_array_section_l29(ptr %0, ptr %1)
+
+// CHECK: %2 = alloca ptr, align 8
+// CHECK: store ptr %0, ptr %2, align 8
+// CHECK: %3 = alloca ptr, align 8
+// CHECK: store ptr %1, ptr %3, align 8
+
+// CHECK: %5 = load ptr, ptr %2, align 8
+// CHECK: %6 = load ptr, ptr %3, align 8
+
+// CHECK: %7 = load i32, ptr getelementptr inbounds ([3 x [3 x [3 x i32]]], ptr @_QFEinarray, i32 0, i64 1, i64 0, i64 0), align 4
+// CHECK: store i32 %7, ptr getelementptr inbounds ([3 x [3 x [3 x i32]]], ptr @_QFEoutarray, i32 0, i64 1, i64 0, i64 0), align 4
diff --git a/mlir/test/Target/LLVMIR/omptarget-array-sectioning-host.mlir b/mlir/test/Target/LLVMIR/omptarget-array-sectioning-host.mlir
new file mode 100644
index 000000000000000..177167587da152b
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/omptarget-array-sectioning-host.mlir
@@ -0,0 +1,80 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+module attributes {omp.is_target_device = false} {
+ llvm.func @_3d_target_array_section() attributes {} {
+ %0 = llvm.mlir.addressof @_QFEinarray : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>
+ %1 = llvm.mlir.addressof @_QFEoutarray : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>
+ %2 = llvm.mlir.constant(1 : index) : i64
+ %3 = llvm.mlir.constant(0 : index) : i64
+ %4 = llvm.mlir.constant(2 : index) : i64
+ %5 = omp.bounds lower_bound(%3 : i64) upper_bound(%4 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %6 = llvm.mlir.constant(0 : index) : i64
+ %7 = llvm.mlir.constant(2 : index) : i64
+ %8 = omp.bounds lower_bound(%6 : i64) upper_bound(%7 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %9 = llvm.mlir.constant(1 : index) : i64
+ %10 = llvm.mlir.constant(1 : index) : i64
+ %11 = omp.bounds lower_bound(%9 : i64) upper_bound(%10 : i64) stride(%2 : i64) start_idx(%2 : i64)
+ %12 = omp.map_info var_ptr(%0 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) map_clauses(tofrom) capture(ByRef) bounds(%5, %8, %11) -> !llvm.ptr<array<3 x array<3 x array<3 x i32>>>> {name = "inarray(1:3,1:3,2:2)"}
+ %13 = llvm.mlir.constant(1 : index) : i64
+ %14 = llvm.mlir.constant(0 : index) : i64
+ %15 = llvm.mlir.constant(2 : index) : i64
+ %16 = omp.bounds lower_bound(%14 : i64) upper_bound(%15 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %17 = llvm.mlir.constant(0 : index) : i64
+ %18 = llvm.mlir.constant(2 : index) : i64
+ %19 = omp.bounds lower_bound(%17 : i64) upper_bound(%18 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %20 = llvm.mlir.constant(0 : index) : i64
+ %21 = llvm.mlir.constant(2 : index) : i64
+ %22 = omp.bounds lower_bound(%20 : i64) upper_bound(%21 : i64) stride(%13 : i64) start_idx(%13 : i64)
+ %23 = omp.map_info var_ptr(%1 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) map_clauses(tofrom) capture(ByRef) bounds(%16, %19, %22) -> !llvm.ptr<array<3 x array<3 x array<3 x i32>>>> {name = "outarray(1:3,1:3,1:3)"}
+ omp.target map_entries(%12, %23 : !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, !llvm.ptr<array<3 x array<3 x array<3 x i32>>>>) {
+ %24 = llvm.mlir.constant(1 : i64) : i64
+ %25 = llvm.mlir.constant(1 : i64) : i64
+ %26 = llvm.mlir.constant(0 : i64) : i64
+ %27 = llvm.mlir.constant(1 : i64) : i64
+ %28 = llvm.mlir.constant(1 : i64) : i64
+ %29 = llvm.mlir.constant(0 : i64) : i64
+ %30 = llvm.mlir.constant(2 : i64) : i64
+ %31 = llvm.mlir.constant(1 : i64) : i64
+ %32 = llvm.mlir.constant(1 : i64) : i64
+ %33 = llvm.getelementptr %0[0, %32, %29, %26] : (!llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, i64, i64, i64) -> !llvm.ptr<i32>
+ %34 = llvm.load %33 : !llvm.ptr<i32>
+ %35 = llvm.mlir.constant(1 : i64) : i64
+ %36 = llvm.mlir.constant(1 : i64) : i64
+ %37 = llvm.mlir.constant(0 : i64) : i64
+ %38 = llvm.mlir.constant(1 : i64) : i64
+ %39 = llvm.mlir.constant(1 : i64) : i64
+ %40 = llvm.mlir.constant(0 : i64) : i64
+ %41 = llvm.mlir.constant(2 : i64) : i64
+ %42 = llvm.mlir.constant(1 : i64) : i64
+ %43 = llvm.mlir.constant(1 : i64) : i64
+ %44 = llvm.getelementptr %1[0, %43, %40, %37] : (!llvm.ptr<array<3 x array<3 x array<3 x i32>>>>, i64, i64, i64) -> !llvm.ptr<i32>
+ llvm.store %34, %44 : !llvm.ptr<i32>
+ omp.terminator
+ }
+ llvm.return
+ }
+ llvm.mlir.global internal @_QFEinarray() {addr_space = 0 : i32} : !llvm.array<3 x array<3 x array<3 x i32>>> {
+ %0 = llvm.mlir.zero : !llvm.array<3 x array<3 x array<3 x i32>>>
+ llvm.return %0 : !llvm.array<3 x array<3 x array<3 x i32>>>
+ }
+ llvm.mlir.global internal @_QFEoutarray() {addr_space = 0 : i32} : !llvm.array<3 x array<3 x array<3 x i32>>> {
+ %0 = llvm.mlir.zero : !llvm.array<3 x array<3 x array<3 x i32>>>
+ llvm.return %0 : !llvm.array<3 x array<3 x array<3 x i32>>>
+ }
+}
+
+// CHECK: @.offload_sizes = private unnamed_addr constant [2 x i64] [i64 36, i64 108]
+// CHECK: @.offload_maptypes = private unnamed_addr constant [2 x i64] [i64 35, i64 35]
+// CHECKL: @.offload_mapnames = private constant [2 x ptr] [ptr @0, ptr @1]
+
+// CHECK: define void @_3d_target_array_section()
+
+// CHECK: %1 = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 0
+// CHECK: store ptr @_QFEinarray, ptr %1, align 8
+// CHECK: %2 = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 0
+// CHECK: store ptr getelementptr inbounds ([3 x [3 x [3 x i32]]], ptr @_QFEinarray, i64 0, i64 1, i64 0, i64 0), ptr %2, align 8
+
+// CHECK: %4 = getelementptr inbounds [2 x ptr], ptr %.offload_baseptrs, i32 0, i32 1
+// CHECK: store ptr @_QFEoutarray, ptr %4, align 8
+// CHECK: %5 = getelementptr inbounds [2 x ptr], ptr %.offload_ptrs, i32 0, i32 1
+// CHECK: store ptr @_QFEoutarray, ptr %5, align 8
diff --git a/mlir/test/Target/LLVMIR/omptarget-llvm.mlir b/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
index 5ed76fd5a1edaff..7e151902dad9d00 100644
--- a/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/omptarget-llvm.mlir
@@ -38,15 +38,20 @@ llvm.func @_QPopenmp_target_data() {
// -----
-llvm.func @_QPopenmp_target_data_region(%1 : !llvm.ptr<array<1024 x i32>>) {
- %2 = omp.map_info var_ptr(%1 : !llvm.ptr<array<1024 x i32>>) map_clauses(from) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
- omp.target_data map_entries(%2 : !llvm.ptr<array<1024 x i32>>) {
- %3 = llvm.mlir.constant(99 : i32) : i32
- %4 = llvm.mlir.constant(1 : i64) : i64
- %5 = llvm.mlir.constant(1 : i64) : i64
- %6 = llvm.mlir.constant(0 : i64) : i64
- %7 = llvm.getelementptr %1[0, %6] : (!llvm.ptr<array<1024 x i32>>, i64) -> !llvm.ptr<i32>
- llvm.store %3, %7 : !llvm.ptr<i32>
+llvm.func @_QPopenmp_target_data_region(%0 : !llvm.ptr<array<1024 x i32>>) {
+ %1 = llvm.mlir.constant(1023 : index) : i64
+ %2 = llvm.mlir.constant(0 : index) : i64
+ %3 = llvm.mlir.constant(1024 : index) : i64
+ %4 = llvm.mlir.constant(1 : index) : i64
+ %5 = omp.bounds lower_bound(%2 : i64) upper_bound(%1 : i64) extent(%3 : i64) stride(%4 : i64) start_idx(%4 : i64)
+ %6 = omp.map_info var_ptr(%0 : !llvm.ptr<array<1024 x i32>>) map_clauses(from) capture(ByRef) bounds(%5) -> !llvm.ptr<array<1024 x i32>> {name = ""}
+ omp.target_data map_entries(%6 : !llvm.ptr<array<1024 x i32>>) {
+ %7 = llvm.mlir.constant(99 : i32) : i32
+ %8 = llvm.mlir.constant(1 : i64) : i64
+ %9 = llvm.mlir.constant(1 : i64) : i64
+ %10 = llvm.mlir.constant(0 : i64) : i64
+ %11 = llvm.getelementptr %0[0, %10] : (!llvm.ptr<array<1024 x i32>>, i64) -> !llvm.ptr<i32>
+ llvm.store %7, %11 : !llvm.ptr<i32>
omp.terminator
}
llvm.return
@@ -92,16 +97,36 @@ llvm.func @_QPomp_target_enter_exit(%1 : !llvm.ptr<array<1024 x i32>>, %3 : !llv
%11 = llvm.mlir.constant(10 : i32) : i32
%12 = llvm.icmp "slt" %10, %11 : i32
%13 = llvm.load %5 : !llvm.ptr<i32>
- %map1 = omp.map_info var_ptr(%1 : !llvm.ptr<array<1024 x i32>>) map_clauses(to) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
- %map2 = omp.map_info var_ptr(%3 : !llvm.ptr<array<512 x i32>>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) -> !llvm.ptr<array<512 x i32>> {name = ""}
+ %14 = llvm.mlir.constant(1023 : index) : i64
+ %15 = llvm.mlir.constant(0 : index) : i64
+ %16 = llvm.mlir.constant(1024 : index) : i64
+ %17 = llvm.mlir.constant(1 : index) : i64
+ %18 = omp.bounds lower_bound(%15 : i64) upper_bound(%14 : i64) extent(%16 : i64) stride(%17 : i64) start_idx(%17 : i64)
+ %map1 = omp.map_info var_ptr(%1 : !llvm.ptr<array<1024 x i32>>) map_clauses(to) capture(ByRef) bounds(%18) -> !llvm.ptr<array<1024 x i32>> {name = ""}
+ %19 = llvm.mlir.constant(511 : index) : i64
+ %20 = llvm.mlir.constant(0 : index) : i64
+ %21 = llvm.mlir.constant(512 : index) : i64
+ %22 = llvm.mlir.constant(1 : index) : i64
+ %23 = omp.bounds lower_bound(%20 : i64) upper_bound(%19 : i64) extent(%21 : i64) stride(%22 : i64) start_idx(%22 : i64)
+ %map2 = omp.map_info var_ptr(%3 : !llvm.ptr<array<512 x i32>>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) bounds(%23) -> !llvm.ptr<array<512 x i32>> {name = ""}
omp.target_enter_data if(%12 : i1) device(%13 : i32) map_entries(%map1, %map2 : !llvm.ptr<array<1024 x i32>>, !llvm.ptr<array<512 x i32>>)
- %14 = llvm.load %7 : !llvm.ptr<i32>
- %15 = llvm.mlir.constant(10 : i32) : i32
- %16 = llvm.icmp "sgt" %14, %15 : i32
- %17 = llvm.load %5 : !llvm.ptr<i32>
- %map3 = omp.map_info var_ptr(%1 : !llvm.ptr<array<1024 x i32>>) map_clauses(from) capture(ByRef) -> !llvm.ptr<array<1024 x i32>> {name = ""}
- %map4 = omp.map_info var_ptr(%3 : !llvm.ptr<array<512 x i32>>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) -> !llvm.ptr<array<512 x i32>> {name = ""}
- omp.target_exit_data if(%16 : i1) device(%17 : i32) map_entries(%map3, %map4 : !llvm.ptr<array<1024 x i32>>, !llvm.ptr<array<512 x i32>>)
+ %24 = llvm.load %7 : !llvm.ptr<i32>
+ %25 = llvm.mlir.constant(10 : i32) : i32
+ %26 = llvm.icmp "sgt" %24, %25 : i32
+ %27 = llvm.load %5 : !llvm.ptr<i32>
+ %28 = llvm.mlir.constant(1023 : index) : i64
+ %29 = llvm.mlir.constant(0 : index) : i64
+ %30 = llvm.mlir.constant(1024 : index) : i64
+ %31 = llvm.mlir.constant(1 : index) : i64
+ %32 = omp.bounds lower_bound(%29 : i64) upper_bound(%28 : i64) extent(%30 : i64) stride(%31 : i64) start_idx(%31 : i64)
+ %map3 = omp.map_info var_ptr(%1 : !llvm.ptr<array<1024 x i32>>) map_clauses(from) capture(ByRef) bounds(%32) -> !llvm.ptr<array<1024 x i32>> {name = ""}
+ %33 = llvm.mlir.constant(511 : index) : i64
+ %34 = llvm.mlir.constant(0 : index) : i64
+ %35 = llvm.mlir.constant(512 : index) : i64
+ %36 = llvm.mlir.constant(1 : index) : i64
+ %37 = omp.bounds lower_bound(%34 : i64) upper_bound(%33 : i64) extent(%35 : i64) stride(%36 : i64) start_idx(%36 : i64)
+ %map4 = omp.map_info var_ptr(%3 : !llvm.ptr<array<512 x i32>>) map_clauses(exit_release_or_enter_alloc) capture(ByRef) bounds(%37) -> !llvm.ptr<array<512 x i32>> {name = ""}
+ omp.target_exit_data if(%26 : i1) device(%27 : i32) map_entries(%map3, %map4 : !llvm.ptr<array<1024 x i32>>, !llvm.ptr<array<512 x i32>>)
llvm.return
}
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
new file mode 100644
index 000000000000000..4362587c139795f
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
@@ -0,0 +1,27 @@
+! Basic offloading test of arrays with provided lower
+! and upper bounds as specified by OpenMP's sectioning
+! 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
+ implicit none
+ integer :: write_arr(10) = (/0,0,0,0,0,0,0,0,0,0/)
+ integer :: read_arr(10) = (/1,2,3,4,5,6,7,8,9,10/)
+ integer :: i = 2
+
+ !$omp target map(to:read_arr(2:5)) map(from:write_arr(2:5)) map(tofrom:i)
+ do i = 2, 5
+ write_arr(i) = read_arr(i)
+ end do
+ !$omp end target
+
+ print *, write_arr(:)
+end program
+
+! CHECK: 0 2 3 4 5 0 0 0 0 0
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
new file mode 100644
index 000000000000000..83775958e5d4454
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
@@ -0,0 +1,39 @@
+! Basic offloading test of a regular array explicitly
+! passed within a target region
+! 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
+ implicit none
+ integer :: inArray(3,3,3)
+ integer :: outArray(3,3,3)
+ integer :: i, j, k
+
+ do i = 1, 3
+ do j = 1, 3
+ do k = 1, 3
+ inArray(i, j, k) = 42
+ outArray(i, j, k) = 0
+ end do
+ end do
+ end do
+
+!$omp target map(tofrom:inArray(1:3, 1:3, 2:2), outArray(1:3, 1:3, 1:3), j, k)
+ do j = 1, 3
+ do k = 1, 3
+ outArray(k, j, 2) = inArray(k, j, 2)
+ end do
+ end do
+!$omp end target
+
+ print *, outArray
+
+end program
+
+! 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 0
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
new file mode 100644
index 000000000000000..261603a8d648f5e
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
@@ -0,0 +1,45 @@
+! Basic offloading test of a regular array explicitly
+! passed within a target region
+! 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
+ implicit none
+ integer :: x(2,2,2)
+ integer :: i = 1, j = 1, k = 1
+ integer :: counter = 1
+ do i = 1, 2
+ do j = 1, 2
+ do k = 1, 2
+ x(i, j, k) = 0
+ end do
+ end do
+ end do
+
+!$omp target map(tofrom:x, i, j, k, counter)
+ do i = 1, 2
+ do j = 1, 2
+ do k = 1, 2
+ x(i, j, k) = counter
+ counter = counter + 1
+ end do
+ end do
+ end do
+!$omp end target
+
+ do i = 1, 2
+ do j = 1, 2
+ do k = 1, 2
+ print *, x(i, j, k)
+ end do
+ end do
+ end do
+end program main
+
+! CHECK: 1 2 3 4 5 6 7 8
\ No newline at end of file
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
new file mode 100644
index 000000000000000..44663c0b30a5fb9
--- /dev/null
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
@@ -0,0 +1,29 @@
+! Basic offloading test of a regular array explicitly
+! passed within a target region
+! 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 :: x(10) = (/0,0,0,0,0,0,0,0,0,0/)
+ integer :: i = 1
+ integer :: j = 11
+
+ !$omp target map(tofrom:x, i, j)
+ do while (i <= j)
+ x(i) = i;
+ i = i + 1
+ end do
+ !$omp end target
+
+ PRINT *, x(:)
+end program main
+
+! CHECK: 1 2 3 4 5 6 7 8 9 10
+
+
More information about the Mlir-commits
mailing list