Struct move_stackless_bytecode::function_target_pipeline::FunctionTargetPipeline

source · [−]

pub struct FunctionTargetPipeline { /* private fields */ }

Expand description

A processing pipeline for function targets.

Implementations

source

impl FunctionTargetPipeline

source

pub fn add_processor(&mut self, processor: Box<dyn FunctionTargetProcessor>)

Adds a processor to this pipeline. Processor will be called in the order they have been added.

source

pub fn last_processor(&self) -> &dyn FunctionTargetProcessor

Gets the last processor in the pipeline, for testing.

source

pub fn sort_targets_in_topological_order<'env>(
 env: &'env GlobalEnv,
 targets: &FunctionTargetsHolder
) -> Vec<FunctionEnv<'env>>ⓘNotable traits for Vec<u8, A>`impl<A> Write for Vec<u8, A> where A: Allocator,`

Sort functions in topological order. This is important for the function target processors. In programs without recursion or mutual recursion, processing functions in topological order means that when a processor sees a caller function, it is guaranteed that all the callees have already been analyzed.

source

pub fn run_with_hook<H1, H2>(
 &self,
 env: &GlobalEnv,
 targets: &mut FunctionTargetsHolder,
 hook_before_pipeline: H1,
 hook_after_each_processor: H2
) where
 H1: Fn(&FunctionTargetsHolder),
 H2: Fn(usize, &dyn FunctionTargetProcessor, &FunctionTargetsHolder),

Runs the pipeline on all functions in the targets holder. Processors are run on each individual function in breadth-first fashion; i.e. a processor can expect that processors preceding it in the pipeline have been executed for all functions before it is called.

source

pub fn run(&self, env: &GlobalEnv, targets: &mut FunctionTargetsHolder)

Run the pipeline on all functions in the targets holder, with no hooks in effect

source

pub fn run_with_dump(
    &self,
    env: &GlobalEnv,
    targets: &mut FunctionTargetsHolder,
    dump_base_name: &str,
    dump_cfg: bool
)

Runs the pipeline on all functions in the targets holder, dump the bytecode before the pipeline as well as after each processor pass. If dump_cfg is set, dump the per-function control-flow graph (in dot format) too.