[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop #139385

NimishMishra · 2025-05-10T15:24:55Z

This patch adds support for MLIR lowering of linear clause on omp.wsloop (except for linear modifiers).

llvmbot · 2025-05-10T15:25:29Z

@llvm/pr-subscribers-mlir-openmp
@llvm/pr-subscribers-mlir-llvm

@llvm/pr-subscribers-flang-fir-hlfir

Author: None (NimishMishra)

Changes

This patch adds support for MLIR lowering of linear clause on omp.wsloop (except for linear modifiers).

Full diff: https://github.com/llvm/llvm-project/pull/139385.diff

5 Files Affected:

(modified) flang/lib/Lower/OpenMP/ClauseProcessor.cpp (+34)
(modified) flang/lib/Lower/OpenMP/ClauseProcessor.h (+1)
(modified) flang/lib/Lower/OpenMP/DataSharingProcessor.cpp (+3-2)
(modified) flang/lib/Lower/OpenMP/OpenMP.cpp (+2-2)
(added) flang/test/Lower/OpenMP/wsloop-linear.f90 (+57)

diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
index 79b5087e4da68..8ba2f604df80a 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
@@ -1060,6 +1060,40 @@ bool ClauseProcessor::processIsDevicePtr(
       });
 }
 
+bool ClauseProcessor::processLinear(mlir::omp::LinearClauseOps &result) const {
+  lower::StatementContext stmtCtx;
+  return findRepeatableClause<
+      omp::clause::Linear>([&](const omp::clause::Linear &clause,
+                               const parser::CharBlock &) {
+    auto &objects = std::get<omp::ObjectList>(clause.t);
+    for (const omp::Object &object : objects) {
+      semantics::Symbol *sym = object.sym();
+      const mlir::Value variable = converter.getSymbolAddress(*sym);
+      result.linearVars.push_back(variable);
+    }
+    if (objects.size()) {
+      if (auto &mod =
+              std::get<std::optional<omp::clause::Linear::StepComplexModifier>>(
+                  clause.t)) {
+        mlir::Value operand =
+            fir::getBase(converter.genExprValue(toEvExpr(*mod), stmtCtx));
+        result.linearStepVars.append(objects.size(), operand);
+      } else if (std::get<std::optional<omp::clause::Linear::LinearModifier>>(
+                     clause.t)) {
+        mlir::Location currentLocation = converter.getCurrentLocation();
+        TODO(currentLocation, "Linear modifiers not yet implemented");
+      } else {
+        // If nothing is present, add the default step of 1.
+        fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
+        mlir::Location currentLocation = converter.getCurrentLocation();
+        mlir::Value operand = firOpBuilder.createIntegerConstant(
+            currentLocation, firOpBuilder.getI32Type(), 1);
+        result.linearStepVars.append(objects.size(), operand);
+      }
+    }
+  });
+}
+
 bool ClauseProcessor::processLink(
     llvm::SmallVectorImpl<DeclareTargetCapturePair> &result) const {
   return findRepeatableClause<omp::clause::Link>(
diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.h b/flang/lib/Lower/OpenMP/ClauseProcessor.h
index 7857ba3fd0845..0ec41bdd33256 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.h
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.h
@@ -122,6 +122,7 @@ class ClauseProcessor {
   bool processIsDevicePtr(
       mlir::omp::IsDevicePtrClauseOps &result,
       llvm::SmallVectorImpl<const semantics::Symbol *> &isDeviceSyms) const;
+  bool processLinear(mlir::omp::LinearClauseOps &result) const;
   bool
   processLink(llvm::SmallVectorImpl<DeclareTargetCapturePair> &result) const;
 
diff --git a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
index 7eec598645eac..2a1c94407e1c8 100644
--- a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
@@ -213,14 +213,15 @@ void DataSharingProcessor::collectSymbolsForPrivatization() {
   // so, we won't need to explicitely handle block objects (or forget to do
   // so).
   for (auto *sym : explicitlyPrivatizedSymbols)
-    allPrivatizedSymbols.insert(sym);
+    if (!sym->test(Fortran::semantics::Symbol::Flag::OmpLinear))
+      allPrivatizedSymbols.insert(sym);
 }
 
 bool DataSharingProcessor::needBarrier() {
   // Emit implicit barrier to synchronize threads and avoid data races on
   // initialization of firstprivate variables and post-update of lastprivate
   // variables.
-  // Emit implicit barrier for linear clause. Maybe on somewhere else.
+  // Emit implicit barrier for linear clause in the OpenMPIRBuilder.
   for (const semantics::Symbol *sym : allPrivatizedSymbols) {
     if (sym->test(semantics::Symbol::Flag::OmpLastPrivate) &&
         (sym->test(semantics::Symbol::Flag::OmpFirstPrivate) ||
diff --git a/flang/lib/Lower/OpenMP/OpenMP.cpp b/flang/lib/Lower/OpenMP/OpenMP.cpp
index 54560729eb4af..6fa915b4364f9 100644
--- a/flang/lib/Lower/OpenMP/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP/OpenMP.cpp
@@ -1841,13 +1841,13 @@ static void genWsloopClauses(
     llvm::SmallVectorImpl<const semantics::Symbol *> &reductionSyms) {
   ClauseProcessor cp(converter, semaCtx, clauses);
   cp.processNowait(clauseOps);
+  cp.processLinear(clauseOps);
   cp.processOrder(clauseOps);
   cp.processOrdered(clauseOps);
   cp.processReduction(loc, clauseOps, reductionSyms);
   cp.processSchedule(stmtCtx, clauseOps);
 
-  cp.processTODO<clause::Allocate, clause::Linear>(
-      loc, llvm::omp::Directive::OMPD_do);
+  cp.processTODO<clause::Allocate>(loc, llvm::omp::Directive::OMPD_do);
 }
 
 //===----------------------------------------------------------------------===//
diff --git a/flang/test/Lower/OpenMP/wsloop-linear.f90 b/flang/test/Lower/OpenMP/wsloop-linear.f90
new file mode 100644
index 0000000000000..b99677108be2f
--- /dev/null
+++ b/flang/test/Lower/OpenMP/wsloop-linear.f90
@@ -0,0 +1,57 @@
+! This test checks lowering of OpenMP DO Directive (Worksharing)
+! with linear clause
+
+! RUN: %flang_fc1 -fopenmp -emit-hlfir %s -o - 2>&1 | FileCheck %s
+
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFsimple_linearEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFsimple_linearEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+!CHECK: %[[const:.*]] = arith.constant 1 : i32
+subroutine simple_linear
+    implicit none
+    integer :: x, y, i
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[const]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x)
+    !CHECK: %[[LOAD:.*]] = fir.load %[[X]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 2 : i32
+    !CHECK: %[[RESULT:.*]] = arith.addi %[[LOAD]], %[[const]] : i32
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine
+
+
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFlinear_stepEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFlinear_stepEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+subroutine linear_step
+    implicit none
+    integer :: x, y, i
+    !CHECK: %[[const:.*]] = arith.constant 4 : i32
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[const]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x:4)
+    !CHECK: %[[LOAD:.*]] = fir.load %[[X]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 2 : i32
+    !CHECK: %[[RESULT:.*]] = arith.addi %[[LOAD]], %[[const]] : i32   
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine
+
+!CHECK: %[[A_alloca:.*]] = fir.alloca i32 {bindc_name = "a", uniq_name = "_QFlinear_exprEa"}
+!CHECK: %[[A:.*]]:2 = hlfir.declare %[[A_alloca]] {uniq_name = "_QFlinear_exprEa"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFlinear_exprEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFlinear_exprEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+subroutine linear_expr
+    implicit none
+    integer :: x, y, i, a
+    !CHECK: %[[LOAD_A:.*]] = fir.load %[[A]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 4 : i32
+    !CHECK: %[[LINEAR_EXPR:.*]] = arith.addi %[[LOAD_A]], %[[const]] : i32
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[LINEAR_EXPR]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x:a+4)
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine

llvmbot · 2025-05-10T15:25:29Z

@llvm/pr-subscribers-flang-openmp

Author: None (NimishMishra)

Changes

This patch adds support for MLIR lowering of linear clause on omp.wsloop (except for linear modifiers).

Full diff: https://github.com/llvm/llvm-project/pull/139385.diff

5 Files Affected:

(modified) flang/lib/Lower/OpenMP/ClauseProcessor.cpp (+34)
(modified) flang/lib/Lower/OpenMP/ClauseProcessor.h (+1)
(modified) flang/lib/Lower/OpenMP/DataSharingProcessor.cpp (+3-2)
(modified) flang/lib/Lower/OpenMP/OpenMP.cpp (+2-2)
(added) flang/test/Lower/OpenMP/wsloop-linear.f90 (+57)

diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
index 79b5087e4da68..8ba2f604df80a 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
@@ -1060,6 +1060,40 @@ bool ClauseProcessor::processIsDevicePtr(
       });
 }
 
+bool ClauseProcessor::processLinear(mlir::omp::LinearClauseOps &result) const {
+  lower::StatementContext stmtCtx;
+  return findRepeatableClause<
+      omp::clause::Linear>([&](const omp::clause::Linear &clause,
+                               const parser::CharBlock &) {
+    auto &objects = std::get<omp::ObjectList>(clause.t);
+    for (const omp::Object &object : objects) {
+      semantics::Symbol *sym = object.sym();
+      const mlir::Value variable = converter.getSymbolAddress(*sym);
+      result.linearVars.push_back(variable);
+    }
+    if (objects.size()) {
+      if (auto &mod =
+              std::get<std::optional<omp::clause::Linear::StepComplexModifier>>(
+                  clause.t)) {
+        mlir::Value operand =
+            fir::getBase(converter.genExprValue(toEvExpr(*mod), stmtCtx));
+        result.linearStepVars.append(objects.size(), operand);
+      } else if (std::get<std::optional<omp::clause::Linear::LinearModifier>>(
+                     clause.t)) {
+        mlir::Location currentLocation = converter.getCurrentLocation();
+        TODO(currentLocation, "Linear modifiers not yet implemented");
+      } else {
+        // If nothing is present, add the default step of 1.
+        fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
+        mlir::Location currentLocation = converter.getCurrentLocation();
+        mlir::Value operand = firOpBuilder.createIntegerConstant(
+            currentLocation, firOpBuilder.getI32Type(), 1);
+        result.linearStepVars.append(objects.size(), operand);
+      }
+    }
+  });
+}
+
 bool ClauseProcessor::processLink(
     llvm::SmallVectorImpl<DeclareTargetCapturePair> &result) const {
   return findRepeatableClause<omp::clause::Link>(
diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.h b/flang/lib/Lower/OpenMP/ClauseProcessor.h
index 7857ba3fd0845..0ec41bdd33256 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.h
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.h
@@ -122,6 +122,7 @@ class ClauseProcessor {
   bool processIsDevicePtr(
       mlir::omp::IsDevicePtrClauseOps &result,
       llvm::SmallVectorImpl<const semantics::Symbol *> &isDeviceSyms) const;
+  bool processLinear(mlir::omp::LinearClauseOps &result) const;
   bool
   processLink(llvm::SmallVectorImpl<DeclareTargetCapturePair> &result) const;
 
diff --git a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
index 7eec598645eac..2a1c94407e1c8 100644
--- a/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/DataSharingProcessor.cpp
@@ -213,14 +213,15 @@ void DataSharingProcessor::collectSymbolsForPrivatization() {
   // so, we won't need to explicitely handle block objects (or forget to do
   // so).
   for (auto *sym : explicitlyPrivatizedSymbols)
-    allPrivatizedSymbols.insert(sym);
+    if (!sym->test(Fortran::semantics::Symbol::Flag::OmpLinear))
+      allPrivatizedSymbols.insert(sym);
 }
 
 bool DataSharingProcessor::needBarrier() {
   // Emit implicit barrier to synchronize threads and avoid data races on
   // initialization of firstprivate variables and post-update of lastprivate
   // variables.
-  // Emit implicit barrier for linear clause. Maybe on somewhere else.
+  // Emit implicit barrier for linear clause in the OpenMPIRBuilder.
   for (const semantics::Symbol *sym : allPrivatizedSymbols) {
     if (sym->test(semantics::Symbol::Flag::OmpLastPrivate) &&
         (sym->test(semantics::Symbol::Flag::OmpFirstPrivate) ||
diff --git a/flang/lib/Lower/OpenMP/OpenMP.cpp b/flang/lib/Lower/OpenMP/OpenMP.cpp
index 54560729eb4af..6fa915b4364f9 100644
--- a/flang/lib/Lower/OpenMP/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP/OpenMP.cpp
@@ -1841,13 +1841,13 @@ static void genWsloopClauses(
     llvm::SmallVectorImpl<const semantics::Symbol *> &reductionSyms) {
   ClauseProcessor cp(converter, semaCtx, clauses);
   cp.processNowait(clauseOps);
+  cp.processLinear(clauseOps);
   cp.processOrder(clauseOps);
   cp.processOrdered(clauseOps);
   cp.processReduction(loc, clauseOps, reductionSyms);
   cp.processSchedule(stmtCtx, clauseOps);
 
-  cp.processTODO<clause::Allocate, clause::Linear>(
-      loc, llvm::omp::Directive::OMPD_do);
+  cp.processTODO<clause::Allocate>(loc, llvm::omp::Directive::OMPD_do);
 }
 
 //===----------------------------------------------------------------------===//
diff --git a/flang/test/Lower/OpenMP/wsloop-linear.f90 b/flang/test/Lower/OpenMP/wsloop-linear.f90
new file mode 100644
index 0000000000000..b99677108be2f
--- /dev/null
+++ b/flang/test/Lower/OpenMP/wsloop-linear.f90
@@ -0,0 +1,57 @@
+! This test checks lowering of OpenMP DO Directive (Worksharing)
+! with linear clause
+
+! RUN: %flang_fc1 -fopenmp -emit-hlfir %s -o - 2>&1 | FileCheck %s
+
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFsimple_linearEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFsimple_linearEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+!CHECK: %[[const:.*]] = arith.constant 1 : i32
+subroutine simple_linear
+    implicit none
+    integer :: x, y, i
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[const]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x)
+    !CHECK: %[[LOAD:.*]] = fir.load %[[X]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 2 : i32
+    !CHECK: %[[RESULT:.*]] = arith.addi %[[LOAD]], %[[const]] : i32
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine
+
+
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFlinear_stepEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFlinear_stepEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+subroutine linear_step
+    implicit none
+    integer :: x, y, i
+    !CHECK: %[[const:.*]] = arith.constant 4 : i32
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[const]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x:4)
+    !CHECK: %[[LOAD:.*]] = fir.load %[[X]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 2 : i32
+    !CHECK: %[[RESULT:.*]] = arith.addi %[[LOAD]], %[[const]] : i32   
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine
+
+!CHECK: %[[A_alloca:.*]] = fir.alloca i32 {bindc_name = "a", uniq_name = "_QFlinear_exprEa"}
+!CHECK: %[[A:.*]]:2 = hlfir.declare %[[A_alloca]] {uniq_name = "_QFlinear_exprEa"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+!CHECK: %[[X_alloca:.*]] = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFlinear_exprEx"}
+!CHECK: %[[X:.*]]:2 = hlfir.declare %[[X_alloca]] {uniq_name = "_QFlinear_exprEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+subroutine linear_expr
+    implicit none
+    integer :: x, y, i, a
+    !CHECK: %[[LOAD_A:.*]] = fir.load %[[A]]#0 : !fir.ref<i32>
+    !CHECK: %[[const:.*]] = arith.constant 4 : i32
+    !CHECK: %[[LINEAR_EXPR:.*]] = arith.addi %[[LOAD_A]], %[[const]] : i32
+    !CHECK: omp.wsloop linear(%[[X]]#0 = %[[LINEAR_EXPR]] : !fir.ref<i32>) {{.*}}
+    !$omp do linear(x:a+4)
+    do i = 1, 10
+        y = x + 2
+    end do
+    !$omp end do
+end subroutine

NimishMishra · 2025-05-10T15:28:17Z

Current TODOs in linear clause following this patch:

Add support for translation to LLVM IR (support added in PR [llvm][mlir][OpenMP] Support translation for linear clause in omp.wsloop #139386)
Extend support to omp.simd (WIP)
Extend support for linear modifiers (WIP)

tblah

The flang lowering changes look good to me. If you like, these can be merged without the mlir->llvm translation because that should issue its own TODO message anyway.

NimishMishra added 2 commits May 10, 2025 19:34

[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop

e4f3cb2

[mlir][llvm][OpenMP] Support translation for linear clause in omp.wsloop

616d637

NimishMishra requested review from ergawy, tblah, kiranchandramohan, mjklemm, kiranktp, kaviya2510 and Thirumalai-Shaktivel May 10, 2025 15:24

llvmbot added flang Flang issues not falling into any other category flang:fir-hlfir flang:openmp labels May 10, 2025

NimishMishra mentioned this pull request May 10, 2025

[llvm][mlir][OpenMP] Support translation for linear clause in omp.wsloop #139386

Open

Drop unused InsertionPt restoration

5b5bba6

llvmbot added mlir:llvm mlir mlir:openmp clang:openmp OpenMP related changes to Clang labels May 10, 2025

tblah approved these changes May 12, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop #139385

[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop #139385

NimishMishra commented May 10, 2025

llvmbot commented May 10, 2025 •

edited

Loading

llvmbot commented May 10, 2025

NimishMishra commented May 10, 2025

tblah left a comment

[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop #139385

Are you sure you want to change the base?

[flang][OpenMP] Support MLIR lowering of linear clause for omp.wsloop #139385

Conversation

NimishMishra commented May 10, 2025

llvmbot commented May 10, 2025 • edited Loading

llvmbot commented May 10, 2025

NimishMishra commented May 10, 2025

tblah left a comment

Choose a reason for hiding this comment

llvmbot commented May 10, 2025 •

edited

Loading