[flang-commits] [flang] [flang][cuda] Allow to set the stack limit size (PR #124859)

[via flang-commits]

================
@@ -16,13 +16,17 @@
 #include "flang/Optimizer/Dialect/FIRType.h"
 #include "flang/Runtime/main.h"
 #include "flang/Runtime/stop.h"
+#ifdef FLANG_CUDA_SUPPORT
+#include "flang/Runtime/CUDA/init.h"
+#endif
----------------
jeanPerier wrote:

What is the reason for protecting this include with an ifdef, if this header contains only a runtime signature, it does not seem like it require some CUDA support when building flang.

The `initCuda` dynamic flag seems enough to me to control the feature.


https://github.com/llvm/llvm-project/pull/124859