[llvm] [NVPTX] Add support for maxclusterrank in launch_bounds (PR #66496)

[Artem Belevich via llvm-commits]

================
@@ -5650,34 +5665,51 @@ static Expr *makeLaunchBoundsArgExpr(Sema &S, Expr *E,
 
 CUDALaunchBoundsAttr *
 Sema::CreateLaunchBoundsAttr(const AttributeCommonInfo &CI, Expr *MaxThreads,
-                             Expr *MinBlocks) {
-  CUDALaunchBoundsAttr TmpAttr(Context, CI, MaxThreads, MinBlocks);
+                             Expr *MinBlocks, Expr *MaxBlocks) {
+  CUDALaunchBoundsAttr TmpAttr(Context, CI, MaxThreads, MinBlocks, MaxBlocks);
   MaxThreads = makeLaunchBoundsArgExpr(*this, MaxThreads, TmpAttr, 0);
-  if (MaxThreads == nullptr)
+  if (!MaxThreads)
     return nullptr;
 
   if (MinBlocks) {
     MinBlocks = makeLaunchBoundsArgExpr(*this, MinBlocks, TmpAttr, 1);
-    if (MinBlocks == nullptr)
+    if (!MinBlocks)
       return nullptr;
   }
 
+  if (MaxBlocks) {
+    // Feature '.maxclusterrank' requires .target sm_90 or higher.
----------------
Artem-B wrote:

Nit: `.maxclusterrank` is a PTX directive. 

https://github.com/llvm/llvm-project/pull/66496