A major feature of MLIR is modeling higher-level control flow operation. By treating 'if's and loops as first-class citizens the implementation of transformation passes is simplified and further analysis of loops made easier and incremental. However, depending on the input language, lowering to structured control flow is non-trivial compared to CFGs. This talk therefore presents the recent upstream implementation of lifting control flow graphs to structured control flow. After elaborating what structured control flow is and showing use-cases for lifting from a CFG, the talk further goes into how the upstream implementation works, how it can be used with custom dialects, and any input constraints and guarantees given by the algorithm.