[flang-commits] [flang] [flang][cuda] Synchronize context before finalization (PR #223058)

[via flang-commits]

github-actions[bot] wrote:

<!--LLVM CODE FORMAT COMMENT: {clang-format}-->


:warning: C/C++ code formatter, clang-format found issues in your code. :warning:

<details>
<summary>
You can test this locally with the following command:
</summary>

``````````bash
git-clang-format --diff origin/main HEAD --extensions h,cpp -- flang/include/flang/Optimizer/Builder/Runtime/CUDA/Support.h flang/lib/Optimizer/Builder/Runtime/CUDA/Support.cpp flang/lib/Lower/Bridge.cpp --diff_from_common_commit
``````````

:warning:
The reproduction instructions above might return results for more than one PR
in a stack if you are using a stacked PR workflow. You can limit the results by
changing `origin/main` to the base branch/commit you want to compare against.
:warning:

</details>

<details>
<summary>
View the diff from clang-format here.
</summary>

``````````diff
diff --git a/flang/lib/Lower/Bridge.cpp b/flang/lib/Lower/Bridge.cpp
index b6b8b0307..30e790396 100644
--- a/flang/lib/Lower/Bridge.cpp
+++ b/flang/lib/Lower/Bridge.cpp
@@ -2006,9 +2006,9 @@ private:
         mlir::Value active =
             fir::runtime::cuda::genDeviceIsActive(*builder, loc);
         builder->genIfThen(loc, active)
-            .genThen([&]() { 
+            .genThen([&]() {
               fir::runtime::cuda::getCUDADeviceSynchronize(*builder, loc);
-              bridge.cudaCleanupCtx().finalizeAndKeep(); 
+              bridge.cudaCleanupCtx().finalizeAndKeep();
             })
             .end();
       }
diff --git a/flang/lib/Optimizer/Builder/Runtime/CUDA/Support.cpp b/flang/lib/Optimizer/Builder/Runtime/CUDA/Support.cpp
index 4fc5fa8e3..097317857 100644
--- a/flang/lib/Optimizer/Builder/Runtime/CUDA/Support.cpp
+++ b/flang/lib/Optimizer/Builder/Runtime/CUDA/Support.cpp
@@ -12,22 +12,30 @@
 
 using namespace fir::runtime::cuda;
 
-static constexpr llvm::StringRef kCudaDeviceSynchronizeName = "_QPcudadevicesynchronize";
+static constexpr llvm::StringRef kCudaDeviceSynchronizeName =
+    "_QPcudadevicesynchronize";
 
 void fir::runtime::cuda::getCUDADeviceSynchronize(fir::FirOpBuilder &builder,
                                                   mlir::Location loc) {
-  mlir::func::FuncOp func = builder.getNamedFunction(kCudaDeviceSynchronizeName);
+  mlir::func::FuncOp func =
+      builder.getNamedFunction(kCudaDeviceSynchronizeName);
   if (!func) {
-    mlir::FunctionType funcType = mlir::FunctionType::get(builder.getContext(), {}, {builder.getI32Type()});
+    mlir::FunctionType funcType = mlir::FunctionType::get(
+        builder.getContext(), {}, {builder.getI32Type()});
     func = builder.createFunction(loc, kCudaDeviceSynchronizeName, funcType);
-    func->setAttr(fir::getFortranProcedureFlagsAttrName(), fir::FortranProcedureFlagsEnumAttr::get(builder.getContext(), fir::FortranProcedureFlagsEnum::intrinsic));
+    func->setAttr(
+        fir::getFortranProcedureFlagsAttrName(),
+        fir::FortranProcedureFlagsEnumAttr::get(
+            builder.getContext(), fir::FortranProcedureFlagsEnum::intrinsic));
     func.setPrivate();
   }
   auto call = fir::CallOp::create(builder, loc, func, mlir::ValueRange{});
-  call.setProcedureAttrsAttr(fir::FortranProcedureFlagsEnumAttr::get(builder.getContext(), fir::FortranProcedureFlagsEnum::intrinsic));
+  call.setProcedureAttrsAttr(fir::FortranProcedureFlagsEnumAttr::get(
+      builder.getContext(), fir::FortranProcedureFlagsEnum::intrinsic));
 }
 
+//  %1315 = fir.call @_QPcudadevicesynchronize() proc_attrs<intrinsic>
+//  fastmath<contract> : () -> i32
 
-//  %1315 = fir.call @_QPcudadevicesynchronize() proc_attrs<intrinsic> fastmath<contract> : () -> i32
-
-//  func.func private @_QPcudadevicesynchronize() -> i32 attributes {fir.proc_attrs = #fir.proc_attrs<intrinsic>}
+//  func.func private @_QPcudadevicesynchronize() -> i32 attributes
+//  {fir.proc_attrs = #fir.proc_attrs<intrinsic>}

``````````

</details>


https://github.com/llvm/llvm-project/pull/223058