[llvm] [mlir] [OpenMP][MLIR] Fix threadprivate lowering when compiling for target when target operations are in use (PR #119310)
via llvm-commits
llvm-commits at lists.llvm.org
Fri Jan 3 08:59:17 PST 2025
https://github.com/agozillon updated https://github.com/llvm/llvm-project/pull/119310
>From cc6a0f3cf1d23537c0b41bc462905af6914baa3d Mon Sep 17 00:00:00 2001
From: agozillon <Andrew.Gozillon at amd.com>
Date: Fri, 6 Dec 2024 10:35:28 -0600
Subject: [PATCH] [OpenMP][MLIR] Fix threadprivate lowering when compiling for
target when target operations are in use
Currently the compiler will ICE in programs like the following on the device lowering pass:
program main
implicit none
type i1_t
integer :: val(1000)
end type i1_t
integer :: i
type(i1_t), pointer :: newi1
type(i1_t), pointer :: tab=>null()
integer, dimension(:), pointer :: tabval
!$omp THREADPRIVATE(tab)
allocate(newi1)
tab=>newi1
tab%val(:)=1
tabval=>tab%val
!$omp target teams distribute parallel do
do i = 1, 1000
tabval(i) = i
end do
!$omp end target teams distribute parallel do
end program main
This is due to the fact that THREADPRIVATE returns a result operation, and this operation can actually be used by other LLVM dialect (or other dialect) operations. However, we currently skip the lowering of threadprivate, so we effectively never generate and bind an LLVM-IR result to the threadprivate operation result. So when we later go on to lower dependent LLVM dialect operations, we are missing the required LLVM-IR result, try to access and use it and then ICE. The fix in this particular PR is to allow compilation of threadprivate for device as well as host, and simply treat the device compilation as
a no-op, binding the LLVM-IR result of threadprivate with no alterations and binding it, which will allow the rest of the compilation to proceed, where we'll eventually discard the host segment in any case.
The other possible solution to this I can think of, is doing something similar to Flang's passes that occur prior to CodeGen to the LLVM dialect, where they erase/no-op certain unrequired operations or transform them to lower level series of operations. And we would erase/no-op threadprivate on device as we'd never have these in target regions.
The main issues I can see with this are that we currently do not specialise this stage based on wether we're compiling for device or host, so it's setting a precedent and adding another point of having to understand the separation between target and host compilation. I am also not sure we'd necessarily want to enforce this at a dialect level incase someone else wishes to add a different lowering flow or translation flow. Another possible issue is that a target operation we have/utilise would depend on the result of threadprivate, meaning we'd not be allowed to entirely erase/no-op it, I am not sure of any situations where this may be an issue currently though.
---
.../OpenMP/OpenMPToLLVMIRTranslation.cpp | 38 +++++++++++++------
...ptarget-threadprivate-device-lowering.mlir | 30 +++++++++++++++
.../fortran/target-with-threadprivate.f90 | 37 ++++++++++++++++++
3 files changed, 94 insertions(+), 11 deletions(-)
create mode 100644 mlir/test/Target/LLVMIR/omptarget-threadprivate-device-lowering.mlir
create mode 100644 offload/test/offloading/fortran/target-with-threadprivate.f90
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index ce129417fc5b25..87cb7f03fec6aa 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -2588,6 +2588,7 @@ static LogicalResult
convertOmpThreadprivate(Operation &opInst, llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation) {
llvm::OpenMPIRBuilder::LocationDescription ompLoc(builder);
+ llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
auto threadprivateOp = cast<omp::ThreadprivateOp>(opInst);
if (failed(checkImplementationStatus(opInst)))
@@ -2595,6 +2596,10 @@ convertOmpThreadprivate(Operation &opInst, llvm::IRBuilderBase &builder,
Value symAddr = threadprivateOp.getSymAddr();
auto *symOp = symAddr.getDefiningOp();
+
+ if (auto asCast = dyn_cast<LLVM::AddrSpaceCastOp>(symOp))
+ symOp = asCast.getOperand().getDefiningOp();
+
if (!isa<LLVM::AddressOfOp>(symOp))
return opInst.emitError("Addressing symbol not found");
LLVM::AddressOfOp addressOfOp = dyn_cast<LLVM::AddressOfOp>(symOp);
@@ -2602,17 +2607,20 @@ convertOmpThreadprivate(Operation &opInst, llvm::IRBuilderBase &builder,
LLVM::GlobalOp global =
addressOfOp.getGlobal(moduleTranslation.symbolTable());
llvm::GlobalValue *globalValue = moduleTranslation.lookupGlobal(global);
- llvm::Type *type = globalValue->getValueType();
- llvm::TypeSize typeSize =
- builder.GetInsertBlock()->getModule()->getDataLayout().getTypeStoreSize(
- type);
- llvm::ConstantInt *size = builder.getInt64(typeSize.getFixedValue());
- llvm::StringRef suffix = llvm::StringRef(".cache", 6);
- std::string cacheName = (Twine(global.getSymName()).concat(suffix)).str();
- llvm::Value *callInst =
- moduleTranslation.getOpenMPBuilder()->createCachedThreadPrivate(
- ompLoc, globalValue, size, cacheName);
- moduleTranslation.mapValue(opInst.getResult(0), callInst);
+
+ if (!ompBuilder->Config.isTargetDevice()) {
+ llvm::Type *type = globalValue->getValueType();
+ llvm::TypeSize typeSize =
+ builder.GetInsertBlock()->getModule()->getDataLayout().getTypeStoreSize(
+ type);
+ llvm::ConstantInt *size = builder.getInt64(typeSize.getFixedValue());
+ llvm::Value *callInst = ompBuilder->createCachedThreadPrivate(
+ ompLoc, globalValue, size, global.getSymName() + ".cache");
+ moduleTranslation.mapValue(opInst.getResult(0), callInst);
+ } else {
+ moduleTranslation.mapValue(opInst.getResult(0), globalValue);
+ }
+
return success();
}
@@ -4212,6 +4220,14 @@ static bool isTargetDeviceOp(Operation *op) {
if (op->getParentOfType<omp::TargetOp>())
return true;
+ // Certain operations return results, and whether utilised in host or
+ // target there is a chance an LLVM Dialect operation depends on it
+ // by taking it in as an operand, so we must always lower these in
+ // some manner or result in an ICE (whether they end up in a no-op
+ // or otherwise).
+ if (mlir::isa<omp::ThreadprivateOp>(op))
+ return true;
+
if (auto parentFn = op->getParentOfType<LLVM::LLVMFuncOp>())
if (auto declareTargetIface =
llvm::dyn_cast<mlir::omp::DeclareTargetInterface>(
diff --git a/mlir/test/Target/LLVMIR/omptarget-threadprivate-device-lowering.mlir b/mlir/test/Target/LLVMIR/omptarget-threadprivate-device-lowering.mlir
new file mode 100644
index 00000000000000..279ecb3f8e998f
--- /dev/null
+++ b/mlir/test/Target/LLVMIR/omptarget-threadprivate-device-lowering.mlir
@@ -0,0 +1,30 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// Not intended to be a functional example, the aim of this test is to verify
+// omp.threadprivate does not crash on lowering during the OpenMP target device
+// pass when used in conjunction with target code in the same module.
+
+module attributes {omp.is_target_device = true } {
+ llvm.func @func() attributes {omp.declare_target = #omp.declaretarget<device_type = (host), capture_clause = (to)>} {
+ %0 = llvm.mlir.addressof @_QFEpointer2 : !llvm.ptr
+ %1 = omp.threadprivate %0 : !llvm.ptr -> !llvm.ptr
+ %2 = omp.map.info var_ptr(%1 : !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>) map_clauses(implicit, to) capture(ByRef) -> !llvm.ptr
+ omp.target map_entries(%2 -> %arg0 : !llvm.ptr) {
+ %3 = llvm.mlir.constant(1 : i32) : i32
+ %4 = llvm.getelementptr %arg0[0, 0] : (!llvm.ptr) -> !llvm.ptr, !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
+ llvm.store %3, %4 : i32, !llvm.ptr
+ omp.terminator
+ }
+ llvm.return
+ }
+ llvm.mlir.global internal @_QFEpointer2() {addr_space = 0 : i32} : !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)> {
+ %0 = llvm.mlir.undef : !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
+ llvm.return %0 : !llvm.struct<(ptr, i64, i32, i8, i8, i8, i8, array<1 x array<3 x i64>>)>
+ }
+}
+
+// CHECK: define weak_odr protected void @{{.*}}(ptr %{{.*}}, ptr %[[ARG1:.*]]) {
+// CHECK: %[[ALLOCA:.*]] = alloca ptr, align 8
+// CHECK: store ptr %[[ARG1]], ptr %[[ALLOCA]], align 8
+// CHECK: %[[LOAD_ALLOCA:.*]] = load ptr, ptr %[[ALLOCA]], align 8
+// CHECK: store i32 1, ptr %[[LOAD_ALLOCA]], align 4
diff --git a/offload/test/offloading/fortran/target-with-threadprivate.f90 b/offload/test/offloading/fortran/target-with-threadprivate.f90
new file mode 100644
index 00000000000000..10c7cecf084123
--- /dev/null
+++ b/offload/test/offloading/fortran/target-with-threadprivate.f90
@@ -0,0 +1,37 @@
+! Basic offloading test that makes sure we can use the predominantly host
+! pragma threadprivate in the same program as target code
+! REQUIRES: flang, amdgpu
+
+! RUN: %libomptarget-compile-fortran-run-and-check-generic
+program main
+ implicit none
+
+ type dtype
+ integer :: val(10)
+ end type dtype
+
+ integer :: i
+ type(dtype), pointer :: pointer1
+ type(dtype), pointer :: pointer2=>null()
+ integer, dimension(:), pointer :: data_pointer
+
+!$omp threadprivate(pointer2)
+
+nullify(pointer1)
+allocate(pointer1)
+
+pointer2=>pointer1
+pointer2%val(:)=1
+data_pointer=>pointer2%val
+
+!$omp target
+ do i = 1, 10
+ data_pointer(i) = i
+ end do
+!$omp end target
+
+print *, data_pointer
+
+end program main
+
+! CHECK: 1 2 3 4 5 6 7 8 9 10
More information about the llvm-commits
mailing list