Module rustc_mir_dataflow::framework

Expand description

A framework that can express both gen-kill and generic dataflow problems.

To use this framework, implement either the Analysis or the GenKillAnalysis trait. If your transfer function can be expressed with only gen/kill operations, prefer GenKillAnalysis since it will run faster while iterating to fixpoint. The impls module contains several examples of gen/kill dataflow analyses.

Create an Engine for your analysis using the into_engine method on the Analysis trait, then call iterate_to_fixpoint. From there, you can use a ResultsCursor to inspect the fixpoint solution to your dataflow problem, or implement the ResultsVisitor interface and use visit_results. The following example uses the ResultsCursor approach.

use rustc_const_eval::dataflow::Analysis; // Makes `into_engine` available.

fn do_my_analysis(tcx: TyCtxt<'tcx>, body: &mir::Body<'tcx>) {
    let analysis = MyAnalysis::new()
        .into_engine(tcx, body)
        .iterate_to_fixpoint()
        .into_results_cursor(body);

    // Print the dataflow state *after* each statement in the start block.
    for (_, statement_index) in body.block_data[START_BLOCK].statements.iter_enumerated() {
        cursor.seek_after(Location { block: START_BLOCK, statement_index });
        let state = cursor.get();
        println!("{:?}", state);
    }
}