[flang-commits] [flang] [flang][cuda] Allow compiler directive in cuda code (PR #141991)

[via flang-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-flang-semantics

Author: Valentin Clement (バレンタイン クレメン) (clementval)

<details>
<summary>Changes</summary>



---
Full diff: https://github.com/llvm/llvm-project/pull/141991.diff


2 Files Affected:

- (modified) flang/lib/Semantics/check-cuda.cpp (+3) 
- (modified) flang/test/Semantics/cuf09.cuf (+11) 


``````````diff
diff --git a/flang/lib/Semantics/check-cuda.cpp b/flang/lib/Semantics/check-cuda.cpp
index fd1ec2b2c69f8..c024640af1220 100644
--- a/flang/lib/Semantics/check-cuda.cpp
+++ b/flang/lib/Semantics/check-cuda.cpp
@@ -321,6 +321,9 @@ template <bool IsCUFKernelDo> class DeviceContextChecker {
                 Check(std::get<parser::Block>(c.t));
               }
             },
+            [&](const common::Indirection<parser::CompilerDirective> &x) {
+              // TODO(CUDA): Check for unsupported compiler directive here.
+            },
             [&](const auto &x) {
               if (auto source{parser::GetSource(x)}) {
                 context_.Say(*source,
diff --git a/flang/test/Semantics/cuf09.cuf b/flang/test/Semantics/cuf09.cuf
index 4a6d9ab09387d..1e23819f9afe8 100644
--- a/flang/test/Semantics/cuf09.cuf
+++ b/flang/test/Semantics/cuf09.cuf
@@ -18,6 +18,17 @@ module m
     !WARNING: I/O statement might not be supported on device
     write(12,'(10F4.1)'), x
   end
+  attributes(global) subroutine devsub3(n)
+    implicit none
+    integer :: n
+    integer :: i, ig, iGrid
+    iGrid = gridDim%x*blockDim%x
+    ig = (blockIdx%x-1)*blockDim%x + threadIdx%x
+ 
+    !dir$ nounroll
+    do i = ig, n, iGrid
+    end do
+  end subroutine
   attributes(global) subroutine hostglobal(a)
     integer :: a(*)
     i = threadIdx%x

``````````

</details>


https://github.com/llvm/llvm-project/pull/141991