[flang-commits] [flang] [flang][cuda] Pass assumed-shape arrays of bind(c) attributes(global) kernels by base address (PR #199313)

[Zhen Wang via flang-commits]

================
@@ -0,0 +1,51 @@
+<!--===- docs/CUDA.md
+
+   Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+   See https://llvm.org/LICENSE.txt for license information.
+   SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+-->
+
+# CUDA Fortran lowering notes
+
+```{contents}
+---
+local:
+---
+```
+
+Catalog of CUDA Fortran lowering decisions in Flang that diverge from the
+Fortran 2018 standard, for cases the [CUDA Fortran Programming
+Guide](https://docs.nvidia.com/hpc-sdk/compilers/cuda-fortran-prog-guide/index.html)
+does not specify.
+
+## `BIND(C) ATTRIBUTES(GLOBAL)` shape-only dummy arrays
+
+For a `BIND(C)` procedure with `ATTRIBUTES(GLOBAL)` or
+`ATTRIBUTES(GRID_GLOBAL)`, dummies whose only descriptor requirement is shape
+(`AssumedShape`, `DeferredShape`, or `AssumedRank`) are passed by base address
+(`!fir.ref<T>`) instead of by `CFI_cdesc_t *` (`!fir.box<T>`). `ALLOCATABLE`,
+`POINTER`, polymorphic, coarray, and `BIND(C)` assumed-length character dummies
----------------
wangzpgi wrote:

dropped polymorphic / coarray / assumed-length character. Only ALLOCATABLE / POINTER left.

https://github.com/llvm/llvm-project/pull/199313