https://github.com/tblah updated
https://github.com/llvm/llvm-project/pull/125901
>From 8a7e449cfa357e18ba094cc61d14bf481668ddcd Mon Sep 17 00:00:00 2001
From: Tom Eccles <tom.ecc...@arm.com>
Date: Wed, 5 Feb 2025 17:29:42 +0000
Subject: [PATCH] [flang][Lower][OpenMP] Don't read moldarg for static sized
array
This should further reduce the number of spurious barriers
---
.../lib/Lower/OpenMP/DataSharingProcessor.cpp | 5 +-
.../Lower/OpenMP/PrivateReductionUtils.cpp | 61 +++++++++++++------
.../lib/Lower/OpenMP/PrivateReductionUtils.h | 6 +-
.../OpenMP/delayed-privatization-array.f90 | 7 +--
4 files changed, 51 insertions(+), 28 deletions(-)
diff --git a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
index 55f543ca38178d1..800d332b74e3182 100644
--- a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
@@ -508,6 +508,8 @@ void DataSharingProcessor::doPrivatize(const
semantics::Symbol *sym,
lower::SymbolBox hsb = converter.lookupOneLevelUpSymbol(*sym);
assert(hsb && "Host symbol box not found");
+ hlfir::Entity entity{hsb.getAddr()};
+ bool cannotHaveNonDefaultLowerBounds =
!entity.mayHaveNonDefaultLowerBounds();
mlir::Location symLoc = hsb.getAddr().getLoc();
std::string privatizerName = sym->name().ToString() + ".privatizer";
@@ -528,7 +530,6 @@ void DataSharingProcessor::doPrivatize(const
semantics::Symbol *sym,
// an alloca for a fir.array type there. Get around this by boxing all
// arrays.
if (mlir::isa<fir::SequenceType>(allocType)) {
- hlfir::Entity entity{hsb.getAddr()};
entity = genVariableBox(symLoc, firOpBuilder, entity);
privVal = entity.getBase();
allocType = privVal.getType();
@@ -590,7 +591,7 @@ void DataSharingProcessor::doPrivatize(const
semantics::Symbol *sym,
result.getDeallocRegion(),
isFirstPrivate ? DeclOperationKind::FirstPrivate
: DeclOperationKind::Private,
- sym);
+ sym, cannotHaveNonDefaultLowerBounds);
// TODO: currently there are false positives from dead uses of the mold
// arg
if (!result.getInitMoldArg().getUses().empty())
diff --git a/flang/lib/Lower/OpenMP/PrivateReductionUtils.cpp
b/flang/lib/Lower/OpenMP/PrivateReductionUtils.cpp
index 951293b133677d3..e970968eb72887b 100644
--- a/flang/lib/Lower/OpenMP/PrivateReductionUtils.cpp
+++ b/flang/lib/Lower/OpenMP/PrivateReductionUtils.cpp
@@ -122,25 +122,40 @@ static void
createCleanupRegion(Fortran::lower::AbstractConverter &converter,
typeError();
}
-fir::ShapeShiftOp Fortran::lower::omp::getShapeShift(fir::FirOpBuilder
&builder,
- mlir::Location loc,
- mlir::Value box) {
+fir::ShapeShiftOp
+Fortran::lower::omp::getShapeShift(fir::FirOpBuilder &builder,
+ mlir::Location loc, mlir::Value box,
+ bool cannotHaveNonDefaultLowerBounds) {
fir::SequenceType sequenceType = mlir::cast<fir::SequenceType>(
hlfir::getFortranElementOrSequenceType(box.getType()));
const unsigned rank = sequenceType.getDimension();
+
llvm::SmallVector<mlir::Value> lbAndExtents;
lbAndExtents.reserve(rank * 2);
-
mlir::Type idxTy = builder.getIndexType();
- for (unsigned i = 0; i < rank; ++i) {
- // TODO: ideally we want to hoist box reads out of the critical section.
- // We could do this by having box dimensions in block arguments like
- // OpenACC does
- mlir::Value dim = builder.createIntegerConstant(loc, idxTy, i);
- auto dimInfo =
- builder.create<fir::BoxDimsOp>(loc, idxTy, idxTy, idxTy, box, dim);
- lbAndExtents.push_back(dimInfo.getLowerBound());
- lbAndExtents.push_back(dimInfo.getExtent());
+
+ if (cannotHaveNonDefaultLowerBounds && !sequenceType.hasDynamicExtents()) {
+ // We don't need fir::BoxDimsOp if all of the extents are statically known
+ // and we can assume default lower bounds. This helps avoids reads from the
+ // mold arg.
+ mlir::Value one = builder.createIntegerConstant(loc, idxTy, 1);
+ for (int64_t extent : sequenceType.getShape()) {
+ assert(extent != sequenceType.getUnknownExtent());
+ mlir::Value extentVal = builder.createIntegerConstant(loc, idxTy,
extent);
+ lbAndExtents.push_back(one);
+ lbAndExtents.push_back(extentVal);
+ }
+ } else {
+ for (unsigned i = 0; i < rank; ++i) {
+ // TODO: ideally we want to hoist box reads out of the critical section.
+ // We could do this by having box dimensions in block arguments like
+ // OpenACC does
+ mlir::Value dim = builder.createIntegerConstant(loc, idxTy, i);
+ auto dimInfo =
+ builder.create<fir::BoxDimsOp>(loc, idxTy, idxTy, idxTy, box, dim);
+ lbAndExtents.push_back(dimInfo.getLowerBound());
+ lbAndExtents.push_back(dimInfo.getExtent());
+ }
}
auto shapeShiftTy = fir::ShapeShiftType::get(builder.getContext(), rank);
@@ -249,12 +264,13 @@ class PopulateInitAndCleanupRegionsHelper {
mlir::Type argType, mlir::Value scalarInitValue,
mlir::Value allocatedPrivVarArg, mlir::Value moldArg,
mlir::Block *initBlock, mlir::Region &cleanupRegion,
- DeclOperationKind kind, const Fortran::semantics::Symbol *sym)
+ DeclOperationKind kind, const Fortran::semantics::Symbol *sym,
+ bool cannotHaveLowerBounds)
: converter{converter}, builder{converter.getFirOpBuilder()}, loc{loc},
argType{argType}, scalarInitValue{scalarInitValue},
allocatedPrivVarArg{allocatedPrivVarArg}, moldArg{moldArg},
initBlock{initBlock}, cleanupRegion{cleanupRegion}, kind{kind},
- sym{sym} {
+ sym{sym}, cannotHaveNonDefaultLowerBounds{cannotHaveLowerBounds} {
valType = fir::unwrapRefType(argType);
}
@@ -296,6 +312,10 @@ class PopulateInitAndCleanupRegionsHelper {
/// Any length parameters which have been fetched for the type
mlir::SmallVector<mlir::Value> lenParams;
+ /// If the source variable being privatized definately can't have non-default
+ /// lower bounds then we don't need to generate code to read them.
+ bool cannotHaveNonDefaultLowerBounds;
+
void createYield(mlir::Value ret) {
builder.create<mlir::omp::YieldOp>(loc, ret);
}
@@ -433,7 +453,8 @@ void
PopulateInitAndCleanupRegionsHelper::initAndCleanupBoxedArray(
// Special case for (possibly allocatable) arrays of polymorphic types
// e.g. !fir.class<!fir.heap<!fir.array<?x!fir.type<>>>>
if (source.isPolymorphic()) {
- fir::ShapeShiftOp shape = getShapeShift(builder, loc, source);
+ fir::ShapeShiftOp shape =
+ getShapeShift(builder, loc, source, cannotHaveNonDefaultLowerBounds);
mlir::Type arrayType = source.getElementOrSequenceType();
mlir::Value allocatedArray = builder.create<fir::AllocMemOp>(
loc, arrayType, /*typeparams=*/mlir::ValueRange{}, shape.getExtents());
@@ -472,8 +493,8 @@ void
PopulateInitAndCleanupRegionsHelper::initAndCleanupBoxedArray(
// Put the temporary inside of a box:
// hlfir::genVariableBox doesn't handle non-default lower bounds
mlir::Value box;
- fir::ShapeShiftOp shapeShift =
- getShapeShift(builder, loc, getLoadedMoldArg());
+ fir::ShapeShiftOp shapeShift = getShapeShift(builder, loc,
getLoadedMoldArg(),
+
cannotHaveNonDefaultLowerBounds);
mlir::Type boxType = getLoadedMoldArg().getType();
if (mlir::isa<fir::BaseBoxType>(temp.getType()))
// the box created by the declare form createTempFromMold is missing
@@ -608,10 +629,10 @@ void
Fortran::lower::omp::populateByRefInitAndCleanupRegions(
mlir::Type argType, mlir::Value scalarInitValue, mlir::Block *initBlock,
mlir::Value allocatedPrivVarArg, mlir::Value moldArg,
mlir::Region &cleanupRegion, DeclOperationKind kind,
- const Fortran::semantics::Symbol *sym) {
+ const Fortran::semantics::Symbol *sym, bool cannotHaveLowerBounds) {
PopulateInitAndCleanupRegionsHelper helper(
converter, loc, argType, scalarInitValue, allocatedPrivVarArg, moldArg,
- initBlock, cleanupRegion, kind, sym);
+ initBlock, cleanupRegion, kind, sym, cannotHaveLowerBounds);
helper.populateByRefInitAndCleanupRegions();
// Often we load moldArg to check something (e.g. length parameters, shape)
diff --git a/flang/lib/Lower/OpenMP/PrivateReductionUtils.h
b/flang/lib/Lower/OpenMP/PrivateReductionUtils.h
index fcd36392a29e0a5..0a3513bff19b0cb 100644
--- a/flang/lib/Lower/OpenMP/PrivateReductionUtils.h
+++ b/flang/lib/Lower/OpenMP/PrivateReductionUtils.h
@@ -55,11 +55,13 @@ void populateByRefInitAndCleanupRegions(
mlir::Value scalarInitValue, mlir::Block *initBlock,
mlir::Value allocatedPrivVarArg, mlir::Value moldArg,
mlir::Region &cleanupRegion, DeclOperationKind kind,
- const Fortran::semantics::Symbol *sym = nullptr);
+ const Fortran::semantics::Symbol *sym = nullptr,
+ bool cannotHaveNonDefaultLowerBounds = false);
/// Generate a fir::ShapeShift op describing the provided boxed array.
fir::ShapeShiftOp getShapeShift(fir::FirOpBuilder &builder, mlir::Location loc,
- mlir::Value box);
+ mlir::Value box,
+ bool cannotHaveNonDefaultLowerBounds = false);
} // namespace omp
} // namespace lower
diff --git a/flang/test/Lower/OpenMP/delayed-privatization-array.f90
b/flang/test/Lower/OpenMP/delayed-privatization-array.f90
index 95fa3f9e0305274..c447fa6f27a7592 100644
--- a/flang/test/Lower/OpenMP/delayed-privatization-array.f90
+++ b/flang/test/Lower/OpenMP/delayed-privatization-array.f90
@@ -108,15 +108,14 @@ program main
! ONE_DIM_DEFAULT_LB-SAME: @[[PRIVATIZER_SYM:.*]] :
[[BOX_TYPE:!fir.box<!fir.array<10xi32>>]] init {
! ONE_DIM_DEFAULT_LB-NEXT: ^bb0(%[[PRIV_ARG:.*]]:
[[TYPE:!fir.ref<!fir.box<!fir.array<10xi32>>>]], %[[PRIV_BOX_ALLOC:.*]]:
[[TYPE]]):
-! ONE_DIM_DEFAULT_LB-NEXT: %[[PRIV_ARG_VAL:.*]] = fir.load %[[PRIV_ARG]]
! ONE_DIM_DEFAULT_LB-NEXT: %[[C10:.*]] = arith.constant 10 : index
! ONE_DIM_DEFAULT_LB-NEXT: %[[SHAPE:.*]] = fir.shape %[[C10]]
! ONE_DIM_DEFAULT_LB-NEXT: %[[ARRAY_ALLOC:.*]] = fir.allocmem
!fir.array<10xi32>
! ONE_DIM_DEFAULT_LB-NEXT: %[[TRUE:.*]] = arith.constant true
! ONE_DIM_DEFAULT_LB-NEXT: %[[DECL:.*]]:2 = hlfir.declare
%[[ARRAY_ALLOC]](%[[SHAPE]])
-! ONE_DIM_DEFAULT_LB-NEXT: %[[C0_0:.*]] = arith.constant 0
-! ONE_DIM_DEFAULT_LB-NEXT: %[[DIMS2:.*]]:3 = fir.box_dims %[[PRIV_ARG_VAL]],
%[[C0_0]]
-! ONE_DIM_DEFAULT_LB-NEXT: %[[SHAPE_SHIFT:.*]] = fir.shape_shift
%[[DIMS2]]#0, %[[DIMS2]]#1
+! ONE_DIM_DEFAULT_LB-NEXT: %[[ONE:.*]] = arith.constant 1 : index
+! ONE_DIM_DEFAULT_LB-NEXT: %[[TEN:.*]] = arith.constant 10 : index
+! ONE_DIM_DEFAULT_LB-NEXT: %[[SHAPE_SHIFT:.*]] = fir.shape_shift %[[ONE]],
%[[TEN]]
! ONE_DIM_DEFAULT_LB-NEXT: %[[EMBOX:.*]] = fir.embox
%[[DECL]]#0(%[[SHAPE_SHIFT]])
! ONE_DIM_DEFAULT_LB-NEXT: fir.store %[[EMBOX]] to %[[PRIV_BOX_ALLOC]]
! ONE_DIM_DEFAULT_LB-NEXT: omp.yield(%[[PRIV_BOX_ALLOC]] : [[TYPE]])
_______________________________________________
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
