CUDA: Now pass correct FLAGS when device link cuda executables.

Previously we had a two issues when building cuda executables
that required separable compilation. The first was that we didn't
propagate FLAGS causing any -arch / -gencode flags to be dropped, and
secondly generators such as ninja would use the CXX language flags
instead of CUDA when the executable was mixed language.

1 files changed, 6 insertions, 0 deletions

diff --git a/Source/cmLinkLineComputer.cxx b/Source/cmLinkLineComputer.cxx
index cf0cf88a4e..e728632bd4 100644
--- a/Source/cmLinkLineComputer.cxx
+++ b/Source/cmLinkLineComputer.cxx
@@ -184,3 +184,9 @@ std::string cmLinkLineComputer::ComputeLinkLibraries(
 
   return fout.str();
 }
+
+std::string cmLinkLineComputer::GetLinkerLanguage(cmGeneratorTarget* target,
+                                                  std::string const& config)
+{
+  return target->GetLinkerLanguage(config);
+}