Merge topic 'cuda_separable_compilation_determined_by_compiler_id'

a233e4e7d9 CUDA: Compilers can now state they don't require a device linking step

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !3659

1 files changed, 10 insertions, 6 deletions

diff --git a/Help/prop_tgt/CUDA_RESOLVE_DEVICE_SYMBOLS.rst b/Help/prop_tgt/CUDA_RESOLVE_DEVICE_SYMBOLS.rst
index ef74ae2..dae960f 100644
--- a/Help/prop_tgt/CUDA_RESOLVE_DEVICE_SYMBOLS.rst
+++ b/Help/prop_tgt/CUDA_RESOLVE_DEVICE_SYMBOLS.rst
@@ -1,12 +1,16 @@
 CUDA_RESOLVE_DEVICE_SYMBOLS
 ---------------------------
 
-CUDA only: Enables device linking for the specific library target
-
-If set this will enable device linking on the library target. Normally
-device linking is deferred until a shared library or executable is generated,
-allowing for multiple static libraries to resolve device symbols at the same
-time when they are used by a shared library or executable.
+CUDA only: Enables device linking for the specific library target where
+required.
+
+If set, this will tell the required compilers to enable device linking
+on the library target. Device linking is an additional link step
+required by some CUDA compilers when :prop_tgt:`CUDA_SEPARABLE_COMPILATION` is
+enabled. Normally device linking is deferred until a shared library or
+executable is generated, allowing for multiple static libraries to resolve
+device symbols at the same time when they are used by a shared library or
+executable.
 
 By default static library targets have this property is disabled,
 while shared, module, and executable targets have this property enabled.