[clang] [CIR][AMDGPU] Honor __launch_bounds__ on HIP kernels (PR #233929)

[Steffen Larsen via cfe-commits]

================
@@ -29,7 +29,8 @@ namespace {
 static void
 handleAMDGPUFlatWorkGroupSizeAttr(const FunctionDecl *fd, cir::FuncOp func,
                                   CIRGenModule &cgm, CIRGenBuilderTy &builder,
-                                  bool isOpenCLKernel, bool isHIPKernel) {
+                                  bool isOpenCLKernel, bool isHIPKernel,
+                                  unsigned lbMaxThreads) {
   const auto *flatWGS = fd->getAttr<AMDGPUFlatWorkGroupSizeAttr>();
   const auto *reqdWGS =
       cgm.getLangOpts().OpenCL ? fd->getAttr<ReqdWorkGroupSizeAttr>() : nullptr;
----------------
steffenlarsen wrote:

Do we need an OpenCL test as well to ensure reqd_work_group precedence?

https://github.com/llvm/llvm-project/pull/233929