[Mlir-commits] [mlir] [mlir][amdgpu] Add `rocdl.s.waitcnt` wrapper (PR #149670)

[Jakub Kuderski]

================
@@ -419,6 +419,105 @@ struct RawBufferOpLowering : public ConvertOpToLLVMPattern<GpuOp> {
   }
 };
 
+// TODO: AMDGPU backend already have all this bitpacking logic, we should move
+// it to some common place.
+/// \details \p Vmcnt, \p Expcnt and \p Lgkmcnt are decoded as follows:
----------------
kuhar wrote:

nit: mlir uses markdown for comment formatting instead of the older doxygen style

https://github.com/llvm/llvm-project/pull/149670