[llvm] [LLVM][NVPTX] Add NVPTX codegen support for clusterlaunchcontrol instruction (PR #134568)

[Pradeep Kumar via llvm-commits]

================
@@ -1478,6 +1478,104 @@ similar but the latter uses generic addressing (see `Generic Addressing <https:/
 
 For more information, refer `PTX ISA <https://docs.nvidia.com/cuda/parallel-thread-execution/#data-movement-and-conversion-instructions-st-bulk>`__.
 
+
+clusterlaunchcontrol Intrinsics
+-------------------------------
+
+'``llvm.nvvm.clusterlaunchcontrol.try_cancel*``' Intrinsics
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Syntax:
+"""""""
+
+.. code-block:: llvm
+
+  declare void @llvm.nvvm.clusterlaunchcontrol.try_cancel.async(ptr %addr, ptr %mbar)
+  declare void @llvm.nvvm.clusterlaunchcontrol.try_cancel.async.shared(ptr addrspace(3) %addr, ptr addrspace(3) %mbar)
+  declare void @llvm.nvvm.clusterlaunchcontrol.try_cancel.async.multicast(ptr %addr, ptr %mbar)
+  declare void @llvm.nvvm.clusterlaunchcontrol.try_cancel.async.multicast.shared(ptr addrspace(3) %addr, ptr addrspace(3) %mbar)
----------------
schwarzschild-radius wrote:

That's right. We have used `shared` in the past to refer to `shared::cta` this keeps the name concise. If the instruction extends support for shared::cluster, we can add shared_cluster variants which would disambiguate. Also, NVPTX now support addrspace for shared::cluster, so the operand would disambiguate as well

https://github.com/llvm/llvm-project/pull/134568