[flang][OpenACC] lower acc loops with early exits #164992
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
llvmbot
added
flang
Member
|
@llvm/pr-subscribers-flang-fir-hlfir Author: None (jeanPerier) ChangesLower acc loop with early exit using the newly added "unstructured" attribute. The core change of this patch is to refactor the loop control variable so that for loop with early exits, the induction variables are privatized, but no bounds operands are added to the acc.loop. The logic of the loop is implemented by the FIR loop lowering logic by generating explicit control flow. Full diff: https://github.com/llvm/llvm-project/pull/164992.diff 2 Files Affected:
diff --git a/flang/lib/Lower/OpenACC.cpp b/flang/lib/Lower/OpenACC.cpp
index d7861ac6463c8..7901971088f70 100644
--- a/flang/lib/Lower/OpenACC.cpp
+++ b/flang/lib/Lower/OpenACC.cpp
@@ -2251,6 +2251,49 @@ static void determineDefaultLoopParMode(
}
}
+// Helper to visit Bounds of DO LOOP nest.
+static void visitLoopControl(
+ Fortran::lower::AbstractConverter &converter,
+ const Fortran::parser::DoConstruct &outerDoConstruct,
+ uint64_t loopsToProcess, Fortran::lower::pft::Evaluation &eval,
+ std::function<void(const Fortran::parser::LoopControl::Bounds &,
+ mlir::Location)>
+ callback) {
+ Fortran::lower::pft::Evaluation *crtEval = &eval.getFirstNestedEvaluation();
+ for (uint64_t i = 0; i < loopsToProcess; ++i) {
+ const Fortran::parser::LoopControl *loopControl;
+ if (i == 0) {
+ loopControl = &*outerDoConstruct.GetLoopControl();
+ mlir::Location loc = converter.genLocation(
+ Fortran::parser::FindSourceLocation(outerDoConstruct));
+ callback(std::get<Fortran::parser::LoopControl::Bounds>(loopControl->u),
+ loc);
+ } else {
+ // Safely locate the next inner DoConstruct within this eval.
+ const Fortran::parser::DoConstruct *innerDo = nullptr;
+ if (crtEval && crtEval->hasNestedEvaluations()) {
+ for (Fortran::lower::pft::Evaluation &child :
+ crtEval->getNestedEvaluations()) {
+ if (auto *stmt = child.getIf<Fortran::parser::DoConstruct>()) {
+ innerDo = stmt;
+ // Prepare to descend for the next iteration
+ crtEval = &child;
+ break;
+ }
+ }
+ }
+ if (!innerDo)
+ break; // No deeper loop; stop collecting collapsed bounds.
+
+ loopControl = &*innerDo->GetLoopControl();
+ mlir::Location loc =
+ converter.genLocation(Fortran::parser::FindSourceLocation(*innerDo));
+ callback(std::get<Fortran::parser::LoopControl::Bounds>(loopControl->u),
+ loc);
+ }
+ }
+}
+
// Extract loop bounds, steps, induction variables, and privatization info
// for both DO CONCURRENT and regular do loops
static void processDoLoopBounds(
@@ -2272,7 +2315,6 @@ static void processDoLoopBounds(
llvm::SmallVector<mlir::Location> &locs, uint64_t loopsToProcess) {
assert(loopsToProcess > 0 && "expect at least one loop");
locs.push_back(currentLocation); // Location of the directive
- Fortran::lower::pft::Evaluation *crtEval = &eval.getFirstNestedEvaluation();
bool isDoConcurrent = outerDoConstruct.IsDoConcurrent();
if (isDoConcurrent) {
@@ -2313,57 +2355,29 @@ static void processDoLoopBounds(
inclusiveBounds.push_back(true);
}
} else {
- for (uint64_t i = 0; i < loopsToProcess; ++i) {
- const Fortran::parser::LoopControl *loopControl;
- if (i == 0) {
- loopControl = &*outerDoConstruct.GetLoopControl();
- locs.push_back(converter.genLocation(
- Fortran::parser::FindSourceLocation(outerDoConstruct)));
- } else {
- // Safely locate the next inner DoConstruct within this eval.
- const Fortran::parser::DoConstruct *innerDo = nullptr;
- if (crtEval && crtEval->hasNestedEvaluations()) {
- for (Fortran::lower::pft::Evaluation &child :
- crtEval->getNestedEvaluations()) {
- if (auto *stmt = child.getIf<Fortran::parser::DoConstruct>()) {
- innerDo = stmt;
- // Prepare to descend for the next iteration
- crtEval = &child;
- break;
- }
- }
- }
- if (!innerDo)
- break; // No deeper loop; stop collecting collapsed bounds.
-
- loopControl = &*innerDo->GetLoopControl();
- locs.push_back(converter.genLocation(
- Fortran::parser::FindSourceLocation(*innerDo)));
- }
-
- const Fortran::parser::LoopControl::Bounds *bounds =
- std::get_if<Fortran::parser::LoopControl::Bounds>(&loopControl->u);
- assert(bounds && "Expected bounds on the loop construct");
- lowerbounds.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->lower), stmtCtx)));
- upperbounds.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->upper), stmtCtx)));
- if (bounds->step)
- steps.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->step), stmtCtx)));
- else // If `step` is not present, assume it is `1`.
- steps.push_back(builder.createIntegerConstant(
- currentLocation, upperbounds[upperbounds.size() - 1].getType(), 1));
-
- Fortran::semantics::Symbol &ivSym =
- bounds->name.thing.symbol->GetUltimate();
- privatizeIv(converter, ivSym, currentLocation, ivTypes, ivLocs,
- privateOperands, ivPrivate, privatizationRecipes);
-
- inclusiveBounds.push_back(true);
-
- // crtEval already updated when descending; no blind increment here.
- }
+ visitLoopControl(
+ converter, outerDoConstruct, loopsToProcess, eval,
+ [&](const Fortran::parser::LoopControl::Bounds &bounds,
+ mlir::Location loc) {
+ locs.push_back(loc);
+ lowerbounds.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.lower), stmtCtx)));
+ upperbounds.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.upper), stmtCtx)));
+ if (bounds.step)
+ steps.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.step), stmtCtx)));
+ else // If `step` is not present, assume it is `1`.
+ steps.push_back(builder.createIntegerConstant(
+ currentLocation, upperbounds[upperbounds.size() - 1].getType(),
+ 1));
+ Fortran::semantics::Symbol &ivSym =
+ bounds.name.thing.symbol->GetUltimate();
+ privatizeIv(converter, ivSym, currentLocation, ivTypes, ivLocs,
+ privateOperands, ivPrivate, privatizationRecipes);
+
+ inclusiveBounds.push_back(true);
+ });
}
}
@@ -2499,6 +2513,34 @@ static void remapDataOperandSymbols(
}
}
+static void privatizeInductionVariables(
+ Fortran::lower::AbstractConverter &converter,
+ mlir::Location currentLocation,
+ const Fortran::parser::DoConstruct &outerDoConstruct,
+ Fortran::lower::pft::Evaluation &eval,
+ llvm::SmallVector<mlir::Value> &privateOperands,
+ llvm::SmallVector<std::pair<mlir::Value, Fortran::semantics::SymbolRef>>
+ &ivPrivate,
+ llvm::SmallVector<mlir::Attribute> &privatizationRecipes,
+ llvm::SmallVector<mlir::Location> &locs, uint64_t loopsToProcess) {
+ // ivTypes and locs will be ignored since no acc.loop control arguments will
+ // be created.
+ llvm::SmallVector<mlir::Type> ivTypes;
+ llvm::SmallVector<mlir::Location> ivLocs;
+ assert(!outerDoConstruct.IsDoConcurrent() &&
+ "do concurrent loops are not expected to contained earlty exits");
+ visitLoopControl(converter, outerDoConstruct, loopsToProcess, eval,
+ [&](const Fortran::parser::LoopControl::Bounds &bounds,
+ mlir::Location loc) {
+ locs.push_back(loc);
+ Fortran::semantics::Symbol &ivSym =
+ bounds.name.thing.symbol->GetUltimate();
+ privatizeIv(converter, ivSym, currentLocation, ivTypes,
+ ivLocs, privateOperands, ivPrivate,
+ privatizationRecipes);
+ });
+}
+
static mlir::acc::LoopOp buildACCLoopOp(
Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation,
@@ -2528,13 +2570,22 @@ static mlir::acc::LoopOp buildACCLoopOp(
llvm::SmallVector<mlir::Location> locs;
llvm::SmallVector<mlir::Value> lowerbounds, upperbounds, steps;
- // Look at the do/do concurrent loops to extract bounds information.
- processDoLoopBounds(converter, currentLocation, stmtCtx, builder,
- outerDoConstruct, eval, lowerbounds, upperbounds, steps,
- privateOperands, ivPrivate, privatizationRecipes, ivTypes,
- ivLocs, inclusiveBounds, locs, loopsToProcess);
-
- // Prepare the operand segment size attribute and the operands value range.
+ // Look at the do/do concurrent loops to extract bounds information unless
+ // this loop is lowered in an unstructured fashion, in which case bounds are
+ // not represented on acc.loop and explicit control flow is used inside body.
+ if (!eval.lowerAsUnstructured()) {
+ processDoLoopBounds(converter, currentLocation, stmtCtx, builder,
+ outerDoConstruct, eval, lowerbounds, upperbounds, steps,
+ privateOperands, ivPrivate, privatizationRecipes,
+ ivTypes, ivLocs, inclusiveBounds, locs, loopsToProcess);
+ } else {
+ // When the loop contains early exits, privatize induction variables, but do
+ // not create acc.loop bounds. The control flow of the loop will be
+ // generated explicitly in the acc.loop body that is just a container.
+ privatizeInductionVariables(converter, currentLocation, outerDoConstruct,
+ eval, privateOperands, ivPrivate,
+ privatizationRecipes, locs, loopsToProcess);
+ }
llvm::SmallVector<mlir::Value> operands;
llvm::SmallVector<int32_t> operandSegments;
addOperands(operands, operandSegments, lowerbounds);
@@ -2563,20 +2614,36 @@ static mlir::acc::LoopOp buildACCLoopOp(
// Remap symbols from data clauses to use data operation results
remapDataOperandSymbols(converter, builder, loopOp, dataOperandSymbolPairs);
- for (auto [arg, iv] :
- llvm::zip(loopOp.getLoopRegions().front()->front().getArguments(),
- ivPrivate)) {
- // Store block argument to the related iv private variable.
- mlir::Value privateValue =
- converter.getSymbolAddress(std::get<Fortran::semantics::SymbolRef>(iv));
- fir::StoreOp::create(builder, currentLocation, arg, privateValue);
+ if (!eval.lowerAsUnstructured()) {
+ for (auto [arg, iv] :
+ llvm::zip(loopOp.getLoopRegions().front()->front().getArguments(),
+ ivPrivate)) {
+ // Store block argument to the related iv private variable.
+ mlir::Value privateValue = converter.getSymbolAddress(
+ std::get<Fortran::semantics::SymbolRef>(iv));
+ fir::StoreOp::create(builder, currentLocation, arg, privateValue);
+ }
+ loopOp.setInclusiveUpperbound(inclusiveBounds);
+ } else {
+ loopOp.setUnstructuredAttr(builder.getUnitAttr());
}
- loopOp.setInclusiveUpperbound(inclusiveBounds);
-
return loopOp;
}
+static bool hasEarlyReturn(Fortran::lower::pft::Evaluation &eval) {
+ bool hasReturnStmt = false;
+ for (auto &e : eval.getNestedEvaluations()) {
+ e.visit(Fortran::common::visitors{
+ [&](const Fortran::parser::ReturnStmt &) { hasReturnStmt = true; },
+ [&](const auto &s) {},
+ });
+ if (e.hasNestedEvaluations())
+ hasReturnStmt = hasEarlyReturn(e);
+ }
+ return hasReturnStmt;
+}
+
static mlir::acc::LoopOp createLoopOp(
Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation,
@@ -2586,8 +2653,7 @@ static mlir::acc::LoopOp createLoopOp(
Fortran::lower::pft::Evaluation &eval,
const Fortran::parser::AccClauseList &accClauseList,
std::optional<mlir::acc::CombinedConstructsType> combinedConstructs =
- std::nullopt,
- bool needEarlyReturnHandling = false) {
+ std::nullopt) {
fir::FirOpBuilder &builder = converter.getFirOpBuilder();
llvm::SmallVector<mlir::Value> tileOperands, privateOperands,
reductionOperands, cacheOperands, vectorOperands, workerNumOperands,
@@ -2763,7 +2829,10 @@ static mlir::acc::LoopOp createLoopOp(
llvm::SmallVector<mlir::Type> retTy;
mlir::Value yieldValue;
- if (needEarlyReturnHandling) {
+ if (eval.lowerAsUnstructured() && hasEarlyReturn(eval)) {
+ // When there is a return statement inside the loop, add a result to the
+ // acc.loop that will be used in a conditional branch after the loop to
+ // return.
mlir::Type i1Ty = builder.getI1Type();
yieldValue = builder.createIntegerConstant(currentLocation, i1Ty, 0);
retTy.push_back(i1Ty);
@@ -2844,19 +2913,6 @@ static mlir::acc::LoopOp createLoopOp(
return loopOp;
}
-static bool hasEarlyReturn(Fortran::lower::pft::Evaluation &eval) {
- bool hasReturnStmt = false;
- for (auto &e : eval.getNestedEvaluations()) {
- e.visit(Fortran::common::visitors{
- [&](const Fortran::parser::ReturnStmt &) { hasReturnStmt = true; },
- [&](const auto &s) {},
- });
- if (e.hasNestedEvaluations())
- hasReturnStmt = hasEarlyReturn(e);
- }
- return hasReturnStmt;
-}
-
static mlir::Value
genACC(Fortran::lower::AbstractConverter &converter,
Fortran::semantics::SemanticsContext &semanticsContext,
@@ -2870,17 +2926,6 @@ genACC(Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation =
converter.genLocation(beginLoopDirective.source);
- bool needEarlyExitHandling = false;
- if (eval.lowerAsUnstructured()) {
- needEarlyExitHandling = hasEarlyReturn(eval);
- // If the loop is lowered in an unstructured fashion, lowering generates
- // explicit control flow that duplicates the looping semantics of the
- // loops.
- if (!needEarlyExitHandling)
- TODO(currentLocation,
- "loop with early exit inside OpenACC loop construct");
- }
-
Fortran::lower::StatementContext stmtCtx;
assert(loopDirective.v == llvm::acc::ACCD_loop &&
@@ -2893,8 +2938,8 @@ genACC(Fortran::lower::AbstractConverter &converter,
std::get<std::optional<Fortran::parser::DoConstruct>>(loopConstruct.t);
auto loopOp = createLoopOp(converter, currentLocation, semanticsContext,
stmtCtx, *outerDoConstruct, eval, accClauseList,
- /*combinedConstructs=*/{}, needEarlyExitHandling);
- if (needEarlyExitHandling)
+ /*combinedConstructs=*/{});
+ if (loopOp.getNumResults() == 1)
return loopOp.getResult(0);
return mlir::Value{};
@@ -3679,10 +3724,6 @@ genACC(Fortran::lower::AbstractConverter &converter,
converter.genLocation(beginCombinedDirective.source);
Fortran::lower::StatementContext stmtCtx;
- if (eval.lowerAsUnstructured())
- TODO(currentLocation,
- "loop with early exit inside OpenACC combined construct");
-
if (combinedDirective.v == llvm::acc::ACCD_kernels_loop) {
createComputeOp<mlir::acc::KernelsOp>(
converter, currentLocation, eval, semanticsContext, stmtCtx,
diff --git a/flang/test/Lower/OpenACC/acc-unstructured.f90 b/flang/test/Lower/OpenACC/acc-unstructured.f90
index c42c7dddc5ca1..829ed5486c196 100644
--- a/flang/test/Lower/OpenACC/acc-unstructured.f90
+++ b/flang/test/Lower/OpenACC/acc-unstructured.f90
@@ -1,5 +1,4 @@
! RUN: bbc -fopenacc -emit-hlfir %s -o - | FileCheck %s
-! XFAIL: *
subroutine test_unstructured1(a, b, c)
integer :: i, j, k
@@ -55,10 +54,11 @@ subroutine test_unstructured2(a, b, c)
! CHECK-LABEL: func.func @_QPtest_unstructured2
! CHECK: acc.parallel
-! CHECK: acc.loop
+! CHECK: acc.loop combined(parallel) private(@privatization_ref_i32 -> %{{.*}} : !fir.ref<i32>) {
! CHECK: fir.call @_FortranAStopStatementText
! CHECK: acc.yield
! CHECK: acc.yield
+! CHECK: } attributes {independent = [#acc.device_type<none>], unstructured}
! CHECK: acc.yield
end subroutine
|
Member
|
@llvm/pr-subscribers-openacc Author: None (jeanPerier) ChangesLower acc loop with early exit using the newly added "unstructured" attribute. The core change of this patch is to refactor the loop control variable so that for loop with early exits, the induction variables are privatized, but no bounds operands are added to the acc.loop. The logic of the loop is implemented by the FIR loop lowering logic by generating explicit control flow. Full diff: https://github.com/llvm/llvm-project/pull/164992.diff 2 Files Affected:
diff --git a/flang/lib/Lower/OpenACC.cpp b/flang/lib/Lower/OpenACC.cpp
index d7861ac6463c8..7901971088f70 100644
--- a/flang/lib/Lower/OpenACC.cpp
+++ b/flang/lib/Lower/OpenACC.cpp
@@ -2251,6 +2251,49 @@ static void determineDefaultLoopParMode(
}
}
+// Helper to visit Bounds of DO LOOP nest.
+static void visitLoopControl(
+ Fortran::lower::AbstractConverter &converter,
+ const Fortran::parser::DoConstruct &outerDoConstruct,
+ uint64_t loopsToProcess, Fortran::lower::pft::Evaluation &eval,
+ std::function<void(const Fortran::parser::LoopControl::Bounds &,
+ mlir::Location)>
+ callback) {
+ Fortran::lower::pft::Evaluation *crtEval = &eval.getFirstNestedEvaluation();
+ for (uint64_t i = 0; i < loopsToProcess; ++i) {
+ const Fortran::parser::LoopControl *loopControl;
+ if (i == 0) {
+ loopControl = &*outerDoConstruct.GetLoopControl();
+ mlir::Location loc = converter.genLocation(
+ Fortran::parser::FindSourceLocation(outerDoConstruct));
+ callback(std::get<Fortran::parser::LoopControl::Bounds>(loopControl->u),
+ loc);
+ } else {
+ // Safely locate the next inner DoConstruct within this eval.
+ const Fortran::parser::DoConstruct *innerDo = nullptr;
+ if (crtEval && crtEval->hasNestedEvaluations()) {
+ for (Fortran::lower::pft::Evaluation &child :
+ crtEval->getNestedEvaluations()) {
+ if (auto *stmt = child.getIf<Fortran::parser::DoConstruct>()) {
+ innerDo = stmt;
+ // Prepare to descend for the next iteration
+ crtEval = &child;
+ break;
+ }
+ }
+ }
+ if (!innerDo)
+ break; // No deeper loop; stop collecting collapsed bounds.
+
+ loopControl = &*innerDo->GetLoopControl();
+ mlir::Location loc =
+ converter.genLocation(Fortran::parser::FindSourceLocation(*innerDo));
+ callback(std::get<Fortran::parser::LoopControl::Bounds>(loopControl->u),
+ loc);
+ }
+ }
+}
+
// Extract loop bounds, steps, induction variables, and privatization info
// for both DO CONCURRENT and regular do loops
static void processDoLoopBounds(
@@ -2272,7 +2315,6 @@ static void processDoLoopBounds(
llvm::SmallVector<mlir::Location> &locs, uint64_t loopsToProcess) {
assert(loopsToProcess > 0 && "expect at least one loop");
locs.push_back(currentLocation); // Location of the directive
- Fortran::lower::pft::Evaluation *crtEval = &eval.getFirstNestedEvaluation();
bool isDoConcurrent = outerDoConstruct.IsDoConcurrent();
if (isDoConcurrent) {
@@ -2313,57 +2355,29 @@ static void processDoLoopBounds(
inclusiveBounds.push_back(true);
}
} else {
- for (uint64_t i = 0; i < loopsToProcess; ++i) {
- const Fortran::parser::LoopControl *loopControl;
- if (i == 0) {
- loopControl = &*outerDoConstruct.GetLoopControl();
- locs.push_back(converter.genLocation(
- Fortran::parser::FindSourceLocation(outerDoConstruct)));
- } else {
- // Safely locate the next inner DoConstruct within this eval.
- const Fortran::parser::DoConstruct *innerDo = nullptr;
- if (crtEval && crtEval->hasNestedEvaluations()) {
- for (Fortran::lower::pft::Evaluation &child :
- crtEval->getNestedEvaluations()) {
- if (auto *stmt = child.getIf<Fortran::parser::DoConstruct>()) {
- innerDo = stmt;
- // Prepare to descend for the next iteration
- crtEval = &child;
- break;
- }
- }
- }
- if (!innerDo)
- break; // No deeper loop; stop collecting collapsed bounds.
-
- loopControl = &*innerDo->GetLoopControl();
- locs.push_back(converter.genLocation(
- Fortran::parser::FindSourceLocation(*innerDo)));
- }
-
- const Fortran::parser::LoopControl::Bounds *bounds =
- std::get_if<Fortran::parser::LoopControl::Bounds>(&loopControl->u);
- assert(bounds && "Expected bounds on the loop construct");
- lowerbounds.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->lower), stmtCtx)));
- upperbounds.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->upper), stmtCtx)));
- if (bounds->step)
- steps.push_back(fir::getBase(converter.genExprValue(
- *Fortran::semantics::GetExpr(bounds->step), stmtCtx)));
- else // If `step` is not present, assume it is `1`.
- steps.push_back(builder.createIntegerConstant(
- currentLocation, upperbounds[upperbounds.size() - 1].getType(), 1));
-
- Fortran::semantics::Symbol &ivSym =
- bounds->name.thing.symbol->GetUltimate();
- privatizeIv(converter, ivSym, currentLocation, ivTypes, ivLocs,
- privateOperands, ivPrivate, privatizationRecipes);
-
- inclusiveBounds.push_back(true);
-
- // crtEval already updated when descending; no blind increment here.
- }
+ visitLoopControl(
+ converter, outerDoConstruct, loopsToProcess, eval,
+ [&](const Fortran::parser::LoopControl::Bounds &bounds,
+ mlir::Location loc) {
+ locs.push_back(loc);
+ lowerbounds.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.lower), stmtCtx)));
+ upperbounds.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.upper), stmtCtx)));
+ if (bounds.step)
+ steps.push_back(fir::getBase(converter.genExprValue(
+ *Fortran::semantics::GetExpr(bounds.step), stmtCtx)));
+ else // If `step` is not present, assume it is `1`.
+ steps.push_back(builder.createIntegerConstant(
+ currentLocation, upperbounds[upperbounds.size() - 1].getType(),
+ 1));
+ Fortran::semantics::Symbol &ivSym =
+ bounds.name.thing.symbol->GetUltimate();
+ privatizeIv(converter, ivSym, currentLocation, ivTypes, ivLocs,
+ privateOperands, ivPrivate, privatizationRecipes);
+
+ inclusiveBounds.push_back(true);
+ });
}
}
@@ -2499,6 +2513,34 @@ static void remapDataOperandSymbols(
}
}
+static void privatizeInductionVariables(
+ Fortran::lower::AbstractConverter &converter,
+ mlir::Location currentLocation,
+ const Fortran::parser::DoConstruct &outerDoConstruct,
+ Fortran::lower::pft::Evaluation &eval,
+ llvm::SmallVector<mlir::Value> &privateOperands,
+ llvm::SmallVector<std::pair<mlir::Value, Fortran::semantics::SymbolRef>>
+ &ivPrivate,
+ llvm::SmallVector<mlir::Attribute> &privatizationRecipes,
+ llvm::SmallVector<mlir::Location> &locs, uint64_t loopsToProcess) {
+ // ivTypes and locs will be ignored since no acc.loop control arguments will
+ // be created.
+ llvm::SmallVector<mlir::Type> ivTypes;
+ llvm::SmallVector<mlir::Location> ivLocs;
+ assert(!outerDoConstruct.IsDoConcurrent() &&
+ "do concurrent loops are not expected to contained earlty exits");
+ visitLoopControl(converter, outerDoConstruct, loopsToProcess, eval,
+ [&](const Fortran::parser::LoopControl::Bounds &bounds,
+ mlir::Location loc) {
+ locs.push_back(loc);
+ Fortran::semantics::Symbol &ivSym =
+ bounds.name.thing.symbol->GetUltimate();
+ privatizeIv(converter, ivSym, currentLocation, ivTypes,
+ ivLocs, privateOperands, ivPrivate,
+ privatizationRecipes);
+ });
+}
+
static mlir::acc::LoopOp buildACCLoopOp(
Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation,
@@ -2528,13 +2570,22 @@ static mlir::acc::LoopOp buildACCLoopOp(
llvm::SmallVector<mlir::Location> locs;
llvm::SmallVector<mlir::Value> lowerbounds, upperbounds, steps;
- // Look at the do/do concurrent loops to extract bounds information.
- processDoLoopBounds(converter, currentLocation, stmtCtx, builder,
- outerDoConstruct, eval, lowerbounds, upperbounds, steps,
- privateOperands, ivPrivate, privatizationRecipes, ivTypes,
- ivLocs, inclusiveBounds, locs, loopsToProcess);
-
- // Prepare the operand segment size attribute and the operands value range.
+ // Look at the do/do concurrent loops to extract bounds information unless
+ // this loop is lowered in an unstructured fashion, in which case bounds are
+ // not represented on acc.loop and explicit control flow is used inside body.
+ if (!eval.lowerAsUnstructured()) {
+ processDoLoopBounds(converter, currentLocation, stmtCtx, builder,
+ outerDoConstruct, eval, lowerbounds, upperbounds, steps,
+ privateOperands, ivPrivate, privatizationRecipes,
+ ivTypes, ivLocs, inclusiveBounds, locs, loopsToProcess);
+ } else {
+ // When the loop contains early exits, privatize induction variables, but do
+ // not create acc.loop bounds. The control flow of the loop will be
+ // generated explicitly in the acc.loop body that is just a container.
+ privatizeInductionVariables(converter, currentLocation, outerDoConstruct,
+ eval, privateOperands, ivPrivate,
+ privatizationRecipes, locs, loopsToProcess);
+ }
llvm::SmallVector<mlir::Value> operands;
llvm::SmallVector<int32_t> operandSegments;
addOperands(operands, operandSegments, lowerbounds);
@@ -2563,20 +2614,36 @@ static mlir::acc::LoopOp buildACCLoopOp(
// Remap symbols from data clauses to use data operation results
remapDataOperandSymbols(converter, builder, loopOp, dataOperandSymbolPairs);
- for (auto [arg, iv] :
- llvm::zip(loopOp.getLoopRegions().front()->front().getArguments(),
- ivPrivate)) {
- // Store block argument to the related iv private variable.
- mlir::Value privateValue =
- converter.getSymbolAddress(std::get<Fortran::semantics::SymbolRef>(iv));
- fir::StoreOp::create(builder, currentLocation, arg, privateValue);
+ if (!eval.lowerAsUnstructured()) {
+ for (auto [arg, iv] :
+ llvm::zip(loopOp.getLoopRegions().front()->front().getArguments(),
+ ivPrivate)) {
+ // Store block argument to the related iv private variable.
+ mlir::Value privateValue = converter.getSymbolAddress(
+ std::get<Fortran::semantics::SymbolRef>(iv));
+ fir::StoreOp::create(builder, currentLocation, arg, privateValue);
+ }
+ loopOp.setInclusiveUpperbound(inclusiveBounds);
+ } else {
+ loopOp.setUnstructuredAttr(builder.getUnitAttr());
}
- loopOp.setInclusiveUpperbound(inclusiveBounds);
-
return loopOp;
}
+static bool hasEarlyReturn(Fortran::lower::pft::Evaluation &eval) {
+ bool hasReturnStmt = false;
+ for (auto &e : eval.getNestedEvaluations()) {
+ e.visit(Fortran::common::visitors{
+ [&](const Fortran::parser::ReturnStmt &) { hasReturnStmt = true; },
+ [&](const auto &s) {},
+ });
+ if (e.hasNestedEvaluations())
+ hasReturnStmt = hasEarlyReturn(e);
+ }
+ return hasReturnStmt;
+}
+
static mlir::acc::LoopOp createLoopOp(
Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation,
@@ -2586,8 +2653,7 @@ static mlir::acc::LoopOp createLoopOp(
Fortran::lower::pft::Evaluation &eval,
const Fortran::parser::AccClauseList &accClauseList,
std::optional<mlir::acc::CombinedConstructsType> combinedConstructs =
- std::nullopt,
- bool needEarlyReturnHandling = false) {
+ std::nullopt) {
fir::FirOpBuilder &builder = converter.getFirOpBuilder();
llvm::SmallVector<mlir::Value> tileOperands, privateOperands,
reductionOperands, cacheOperands, vectorOperands, workerNumOperands,
@@ -2763,7 +2829,10 @@ static mlir::acc::LoopOp createLoopOp(
llvm::SmallVector<mlir::Type> retTy;
mlir::Value yieldValue;
- if (needEarlyReturnHandling) {
+ if (eval.lowerAsUnstructured() && hasEarlyReturn(eval)) {
+ // When there is a return statement inside the loop, add a result to the
+ // acc.loop that will be used in a conditional branch after the loop to
+ // return.
mlir::Type i1Ty = builder.getI1Type();
yieldValue = builder.createIntegerConstant(currentLocation, i1Ty, 0);
retTy.push_back(i1Ty);
@@ -2844,19 +2913,6 @@ static mlir::acc::LoopOp createLoopOp(
return loopOp;
}
-static bool hasEarlyReturn(Fortran::lower::pft::Evaluation &eval) {
- bool hasReturnStmt = false;
- for (auto &e : eval.getNestedEvaluations()) {
- e.visit(Fortran::common::visitors{
- [&](const Fortran::parser::ReturnStmt &) { hasReturnStmt = true; },
- [&](const auto &s) {},
- });
- if (e.hasNestedEvaluations())
- hasReturnStmt = hasEarlyReturn(e);
- }
- return hasReturnStmt;
-}
-
static mlir::Value
genACC(Fortran::lower::AbstractConverter &converter,
Fortran::semantics::SemanticsContext &semanticsContext,
@@ -2870,17 +2926,6 @@ genACC(Fortran::lower::AbstractConverter &converter,
mlir::Location currentLocation =
converter.genLocation(beginLoopDirective.source);
- bool needEarlyExitHandling = false;
- if (eval.lowerAsUnstructured()) {
- needEarlyExitHandling = hasEarlyReturn(eval);
- // If the loop is lowered in an unstructured fashion, lowering generates
- // explicit control flow that duplicates the looping semantics of the
- // loops.
- if (!needEarlyExitHandling)
- TODO(currentLocation,
- "loop with early exit inside OpenACC loop construct");
- }
-
Fortran::lower::StatementContext stmtCtx;
assert(loopDirective.v == llvm::acc::ACCD_loop &&
@@ -2893,8 +2938,8 @@ genACC(Fortran::lower::AbstractConverter &converter,
std::get<std::optional<Fortran::parser::DoConstruct>>(loopConstruct.t);
auto loopOp = createLoopOp(converter, currentLocation, semanticsContext,
stmtCtx, *outerDoConstruct, eval, accClauseList,
- /*combinedConstructs=*/{}, needEarlyExitHandling);
- if (needEarlyExitHandling)
+ /*combinedConstructs=*/{});
+ if (loopOp.getNumResults() == 1)
return loopOp.getResult(0);
return mlir::Value{};
@@ -3679,10 +3724,6 @@ genACC(Fortran::lower::AbstractConverter &converter,
converter.genLocation(beginCombinedDirective.source);
Fortran::lower::StatementContext stmtCtx;
- if (eval.lowerAsUnstructured())
- TODO(currentLocation,
- "loop with early exit inside OpenACC combined construct");
-
if (combinedDirective.v == llvm::acc::ACCD_kernels_loop) {
createComputeOp<mlir::acc::KernelsOp>(
converter, currentLocation, eval, semanticsContext, stmtCtx,
diff --git a/flang/test/Lower/OpenACC/acc-unstructured.f90 b/flang/test/Lower/OpenACC/acc-unstructured.f90
index c42c7dddc5ca1..829ed5486c196 100644
--- a/flang/test/Lower/OpenACC/acc-unstructured.f90
+++ b/flang/test/Lower/OpenACC/acc-unstructured.f90
@@ -1,5 +1,4 @@
! RUN: bbc -fopenacc -emit-hlfir %s -o - | FileCheck %s
-! XFAIL: *
subroutine test_unstructured1(a, b, c)
integer :: i, j, k
@@ -55,10 +54,11 @@ subroutine test_unstructured2(a, b, c)
! CHECK-LABEL: func.func @_QPtest_unstructured2
! CHECK: acc.parallel
-! CHECK: acc.loop
+! CHECK: acc.loop combined(parallel) private(@privatization_ref_i32 -> %{{.*}} : !fir.ref<i32>) {
! CHECK: fir.call @_FortranAStopStatementText
! CHECK: acc.yield
! CHECK: acc.yield
+! CHECK: } attributes {independent = [#acc.device_type<none>], unstructured}
! CHECK: acc.yield
end subroutine
|
Base automatically changed from
users/jeanPerier/mlir_acc_unstructured_loop
to
main
November 4, 2025 00:02
vinay-deshmukh
pushed a commit
to vinay-deshmukh/llvm-project
that referenced
this pull request
Nov 8, 2025
Lower acc loop with early exit using the newly added "unstructured" attribute. The core change of this patch is to refactor the loop control variable so that for loop with early exits, the induction variables are privatized, but no bounds operands are added to the acc.loop. The logic of the loop is implemented by the FIR loop lowering logic by generating explicit control flow.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Lower acc loop with early exit using the newly added "unstructured" attribute.
The core change of this patch is to refactor the loop control variable so that for loop with early exits, the induction variables are privatized, but no bounds operands are added to the acc.loop.
The logic of the loop is implemented by the FIR loop lowering logic by generating explicit control flow.