Adding Changes for invoking Masked Operation #98423
Merged
Conversation
This file contains 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
|
@llvm/pr-subscribers-flang-openmp @llvm/pr-subscribers-flang-fir-hlfir Author: Anchu Rajendran S (anchuraj) ChangesChange adds the lowering support for Omp masked directive. Omp masked is introduced in 5.2 standard and allows a parallel region to be executed by threads specified by a programmer. This is achieved with the help of filter clause which helps to specify thread id expected to execute the region. Other related PRs:
Full diff: https://github.com/llvm/llvm-project/pull/98423.diff 5 Files Affected:
diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
index d507e58b164dd..f1e049c15c5c3 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.cpp
@@ -332,6 +332,16 @@ bool ClauseProcessor::processDistSchedule(
return false;
}
+bool ClauseProcessor::processFilter(lower::StatementContext &stmtCtx,
+ mlir::omp::FilterClauseOps &result) const {
+ if (auto *clause = findUniqueClause<omp::clause::Filter>()) {
+ result.filteredThreadIdVar =
+ fir::getBase(converter.genExprValue(clause->v, stmtCtx));
+ return true;
+ }
+ return false;
+}
+
bool ClauseProcessor::processFinal(lower::StatementContext &stmtCtx,
mlir::omp::FinalClauseOps &result) const {
const parser::CharBlock *source = nullptr;
diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.h b/flang/lib/Lower/OpenMP/ClauseProcessor.h
index 43795d5c25399..a8c0f26759565 100644
--- a/flang/lib/Lower/OpenMP/ClauseProcessor.h
+++ b/flang/lib/Lower/OpenMP/ClauseProcessor.h
@@ -63,6 +63,8 @@ class ClauseProcessor {
bool processDeviceType(mlir::omp::DeviceTypeClauseOps &result) const;
bool processDistSchedule(lower::StatementContext &stmtCtx,
mlir::omp::DistScheduleClauseOps &result) const;
+ bool processFilter(lower::StatementContext &stmtCtx,
+ mlir::omp::FilterClauseOps &result) const;
bool processFinal(lower::StatementContext &stmtCtx,
mlir::omp::FinalClauseOps &result) const;
bool processHasDeviceAddr(
diff --git a/flang/lib/Lower/OpenMP/OpenMP.cpp b/flang/lib/Lower/OpenMP/OpenMP.cpp
index dffdb834d4e66..63740835a131a 100644
--- a/flang/lib/Lower/OpenMP/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP/OpenMP.cpp
@@ -1067,6 +1067,15 @@ genLoopNestClauses(lower::AbstractConverter &converter,
clauseOps.loopInclusiveAttr = converter.getFirOpBuilder().getUnitAttr();
}
+static void genMaskedClauses(lower::AbstractConverter &converter,
+ semantics::SemanticsContext &semaCtx,
+ lower::StatementContext &stmtCtx,
+ const List<Clause> &clauses, mlir::Location loc,
+ mlir::omp::MaskedClauseOps &clauseOps) {
+ ClauseProcessor cp(converter, semaCtx, clauses);
+ cp.processFilter(stmtCtx, clauseOps);
+}
+
static void
genOrderedRegionClauses(lower::AbstractConverter &converter,
semantics::SemanticsContext &semaCtx,
@@ -1375,6 +1384,21 @@ genLoopNestOp(lower::AbstractConverter &converter, lower::SymMap &symTable,
queue, item, clauseOps);
}
+static mlir::omp::MaskedOp
+genMaskedOp(lower::AbstractConverter &converter, lower::SymMap &symTable,
+ semantics::SemanticsContext &semaCtx, lower::pft::Evaluation &eval,
+ mlir::Location loc, const ConstructQueue &queue,
+ ConstructQueue::iterator item) {
+ lower::StatementContext stmtCtx;
+ mlir::omp::MaskedClauseOps clauseOps;
+ genMaskedClauses(converter, semaCtx, stmtCtx, item->clauses, loc, clauseOps);
+
+ return genOpWithBody<mlir::omp::MaskedOp>(
+ OpWithBodyGenInfo(converter, symTable, semaCtx, loc, eval,
+ llvm::omp::Directive::OMPD_masked),
+ queue, item, clauseOps);
+}
+
static mlir::omp::MasterOp
genMasterOp(lower::AbstractConverter &converter, lower::SymMap &symTable,
semantics::SemanticsContext &semaCtx, lower::pft::Evaluation &eval,
@@ -2106,9 +2130,11 @@ static void genOMPDispatch(lower::AbstractConverter &converter,
*loopDsp);
break;
case llvm::omp::Directive::OMPD_loop:
- case llvm::omp::Directive::OMPD_masked:
TODO(loc, "Unhandled directive " + llvm::omp::getOpenMPDirectiveName(dir));
break;
+ case llvm::omp::Directive::OMPD_masked:
+ genMaskedOp(converter, symTable, semaCtx, eval, loc, queue, item);
+ break;
case llvm::omp::Directive::OMPD_master:
genMasterOp(converter, symTable, semaCtx, eval, loc, queue, item);
break;
@@ -2444,6 +2470,7 @@ static void genOMP(lower::AbstractConverter &converter, lower::SymMap &symTable,
!std::holds_alternative<clause::Copyprivate>(clause.u) &&
!std::holds_alternative<clause::Default>(clause.u) &&
!std::holds_alternative<clause::Depend>(clause.u) &&
+ !std::holds_alternative<clause::Filter>(clause.u) &&
!std::holds_alternative<clause::Final>(clause.u) &&
!std::holds_alternative<clause::Firstprivate>(clause.u) &&
!std::holds_alternative<clause::HasDeviceAddr>(clause.u) &&
diff --git a/flang/test/Lower/OpenMP/Todo/masked-directive.f90 b/flang/test/Lower/OpenMP/Todo/masked-directive.f90
deleted file mode 100644
index 77767715af522..0000000000000
--- a/flang/test/Lower/OpenMP/Todo/masked-directive.f90
+++ /dev/null
@@ -1,13 +0,0 @@
-! This test checks lowering of OpenMP masked Directive.
-
-! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-
-! CHECK: not yet implemented: Unhandled directive masked
-subroutine test_masked()
- integer :: c = 1
- !$omp masked
- c = c + 1
- !$omp end masked
-end subroutine
-
diff --git a/flang/test/Lower/OpenMP/masked.f90 b/flang/test/Lower/OpenMP/masked.f90
new file mode 100644
index 0000000000000..0d67c08d2d9f4
--- /dev/null
+++ b/flang/test/Lower/OpenMP/masked.f90
@@ -0,0 +1,25 @@
+!RUN: %flang_fc1 -emit-hlfir -fopenmp %s -o - | FileCheck %s
+
+!CHECK-LABEL: func @_QPomp_masked
+subroutine omp_masked(threadId)
+integer :: threadId
+
+!CHECK: omp.masked {
+!$omp masked
+
+ !CHECK: fir.call @_QPmasked() {{.*}}: () -> ()
+ call masked()
+
+!CHECK: omp.terminator
+!$omp end masked
+
+!CHECK: omp.masked filter({{.*}}) {
+!$omp masked filter(threadId)
+
+ !CHECK: fir.call @_QPmaskedwithfilter() {{.*}}: () -> ()
+ call maskedWithFilter()
+
+!CHECK: omp.terminator
+!$omp end masked
+end subroutine omp_masked
+
|
anchuraj
requested review from
kiranchandramohan,
mjklemm,
raghavendhra and
agozillon
|
Thank you for the reviews. Will merge this change after, lowering support is merged. |
aaryanshukla
pushed a commit
to aaryanshukla/llvm-project
that referenced
this pull request
Jul 14, 2024
PR adds changes to the flang frontend to create the `MaskedOp` when `masked` directive is used in the input program. Omp masked is introduced in 5.2 standard and allows a parallel region to be executed by threads specified by a programmer. This is achieved with the help of filter clause which helps to specify thread id expected to execute the region. Other related PRs: - [Fortran Parsing and Semantic Support](llvm#91432) - Merged - [MLIR Support](https://github.com/llvm/llvm-project/pull/96022/files) - Merged - [Lowering Support](llvm#98401) - Under Review
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.
PR adds changes to the flang frontend to create the
MaskedOpwhenmaskeddirective is used in the input program. Omp masked is introduced in 5.2 standard and allows a parallel region to be executed by threads specified by a programmer. This is achieved with the help of filter clause which helps to specify thread id expected to execute the region.Other related PRs: