[llvm] [AMDGPU] Add dynamic LDS size implicit kernel argument to CO-v5 (PR #65273)

[via llvm-commits]

================
@@ -663,7 +663,10 @@ void MetadataStreamerMsgPackV5::emitHiddenKernelArgs(
     Offset += 8; // Skipped.
   }
 
-  Offset += 72; // Reserved.
+  // emit argument for hidden dynamic lds size
----------------
skc7 wrote:

Runtime will always have the dynamic lds size passed in via hipLaunchKernelGGL(zero by default). So, it can fill this argument always. Also, there is no attribute present which I can find, which can be used to check if dynamic LDS is used in the function or not. Please let me know if we really need to skip adding this argument.

https://github.com/llvm/llvm-project/pull/65273