[flang][cuda] Move interface to __cuda_device

flang/lib/Semantics/resolve-names.cpp

@@ -4015,7 +4015,9 @@ bool SubprogramVisitor::Pre(const parser::PrefixSpec::Attributes &attrs) {
           *attrs == common::CUDASubprogramAttrs::Device) {
         const Scope &scope{currScope()};
         const Scope *mod{FindModuleContaining(scope)};
-        if (mod && mod->GetName().value() == "cudadevice") {
+        if (mod &&
+            (mod->GetName().value() == "cudadevice" ||
+                mod->GetName().value() == "__cuda_device")) {
           return false;
         }
         // Implicitly USE the cudadevice module by copying its symbols in the

flang/module/__cuda_device.f90

@@ -0,0 +1,32 @@
+!===-- module/__cuda_device.f90 --------------------------------------------===!
+!
+! 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
+!
+!===------------------------------------------------------------------------===!
+
+! This module contains CUDA Fortran interfaces used in cudadevice.f90.
+
+module __cuda_device
+implicit none
+
+  ! Set PRIVATE by default to explicitly only export what is meant
+  ! to be exported by this MODULE.
+
+  interface
+    attributes(device) function __fadd_rd(x, y) bind(c, name='__nv_fadd_rd')
+      real, intent(in), value :: x, y
+      real :: __fadd_rd
+    end function
+  end interface
+  public :: __fadd_rd
+
+  interface
+    attributes(device) function __fadd_ru(x, y) bind(c, name='__nv_fadd_ru')
+      real, intent(in), value :: x, y
+      real :: __fadd_ru
+    end function
+  end interface
+  public :: __fadd_ru
+end module

flang/module/cudadevice.f90

@@ -9,6 +9,7 @@
 ! CUDA Fortran procedures available in device subprogram
 
 module cudadevice
+  use __cuda_device, only: __fadd_rd, __fadd_ru
 implicit none
 
   ! Set PRIVATE by default to explicitly only export what is meant
@@ -71,20 +72,4 @@ attributes(device) subroutine threadfence_system()
   end interface
   public :: threadfence_system
 
-  interface
-    attributes(device) function __fadd_rd(x, y) bind(c, name='__nv_fadd_rd')
-      real, intent(in) :: x, y
-      real :: __fadd_rd
-    end function
-  end interface
-  public :: __fadd_rd
-
-  interface
-    attributes(device) function __fadd_ru(x, y) bind(c, name='__nv_fadd_ru')
-      real, intent(in) :: x, y
-      real :: __fadd_ru
-    end function
-  end interface
-  public :: __fadd_ru
-
 end module

flang/tools/f18/CMakeLists.txt

@@ -21,6 +21,7 @@ set(MODULES_WITHOUT_IMPLEMENTATION
   "__ppc_intrinsics"
   "mma"
   "__cuda_builtins"
+  "__cuda_device"
   "cudadevice"
   "ieee_arithmetic"
   "ieee_exceptions"
@@ -67,9 +68,12 @@ if (NOT CMAKE_CROSSCOMPILING)
     elseif(${filename} STREQUAL "__ppc_intrinsics" OR
            ${filename} STREQUAL "mma")
       set(depends ${FLANG_INTRINSIC_MODULES_DIR}/__ppc_types.mod)
-    elseif(${filename} STREQUAL "cudadevice")
+    elseif(${filename} STREQUAL "__cuda_device")
       set(opts -fc1 -xcuda)
       set(depends ${FLANG_INTRINSIC_MODULES_DIR}/__cuda_builtins.mod)
+    elseif(${filename} STREQUAL "cudadevice")
+      set(opts -fc1 -xcuda)
+      set(depends ${FLANG_INTRINSIC_MODULES_DIR}/__cuda_device.mod)
     else()
       set(depends ${FLANG_INTRINSIC_MODULES_DIR}/__fortran_builtins.mod)
       if(${filename} STREQUAL "iso_fortran_env")