[clang] [compiler-rt] [libc] [libcxx] [llvm] [AMDGPU] Fix code object version not being set to 'none' (PR #135036)

[Joseph Huber via cfe-commits]

================
@@ -62,7 +62,7 @@ Value *EmitAMDGPUWorkGroupSize(CodeGenFunction &CGF, unsigned Index) {
 
   auto Cov = CGF.getTarget().getTargetOpts().CodeObjectVersion;
 
-  if (Cov == CodeObjectVersionKind::COV_None) {
+  if (Cov == CodeObjectVersionKind::COV_None && !CGF.getLangOpts().OpenMP) {
----------------
jhuber6 wrote:

Yeah, I'm realizing that now. This just needs to turn off the module flag but I don't want to have this lowering at all. I think it's still used in the ROCm Device Libs? So probably can't just delete it like I did before. Can I just add a flag to suppress the module metadata instead?

https://github.com/llvm/llvm-project/pull/135036