From 70102e5ebd9ace8b006389b533ca91efd7382233 Mon Sep 17 00:00:00 2001
From: Ryan Ziegler <ryanjz2@illinois.edu>
Date: Tue, 26 Nov 2024 12:29:08 -0600
Subject: [PATCH] Dce dead calls
---
hercules_ir/src/build.rs | 2 +
hercules_ir/src/ir.rs | 1 +
hercules_ir/src/parse.rs | 4 +-
hercules_opt/src/delete_uncalled.rs | 81 ++++++++++++++++++++++++++++
hercules_opt/src/lib.rs | 2 +
hercules_opt/src/pass.rs | 56 +++++++++++++++++++
juno_frontend/src/codegen.rs | 1 +
juno_frontend/src/labeled_builder.rs | 11 ++--
juno_frontend/src/lib.rs | 9 ++++
9 files changed, 163 insertions(+), 4 deletions(-)
create mode 100644 hercules_opt/src/delete_uncalled.rs
diff --git a/hercules_ir/src/build.rs b/hercules_ir/src/build.rs
index ad130c1b..6dd5a3e6 100644
--- a/hercules_ir/src/build.rs
+++ b/hercules_ir/src/build.rs
@@ -457,6 +457,7 @@ impl<'a> Builder<'a> {
param_types: Vec<TypeID>,
return_type: TypeID,
num_dynamic_constants: u32,
+ entry: bool,
) -> BuilderResult<(FunctionID, NodeID)> {
if let Some(_) = self.function_ids.get(name) {
Err(format!("Can't create a function with name \"{}\", because a function with the same name has already been created.", name))?
@@ -469,6 +470,7 @@ impl<'a> Builder<'a> {
return_type,
nodes: vec![Node::Start],
num_dynamic_constants,
+ entry,
});
Ok((id, NodeID::new(0)))
}
diff --git a/hercules_ir/src/ir.rs b/hercules_ir/src/ir.rs
index c4ab9b3e..23d4b067 100644
--- a/hercules_ir/src/ir.rs
+++ b/hercules_ir/src/ir.rs
@@ -42,6 +42,7 @@ pub struct Function {
pub return_type: TypeID,
pub nodes: Vec<Node>,
pub num_dynamic_constants: u32,
+ pub entry: bool,
}
/*
diff --git a/hercules_ir/src/parse.rs b/hercules_ir/src/parse.rs
index ccd4b13f..41d59441 100644
--- a/hercules_ir/src/parse.rs
+++ b/hercules_ir/src/parse.rs
@@ -123,7 +123,8 @@ fn parse_module<'a>(ir_text: &'a str, context: Context<'a>) -> nom::IResult<&'a
param_types: vec![],
return_type: TypeID::new(0),
nodes: vec![],
- num_dynamic_constants: 0
+ num_dynamic_constants: 0,
+ entry: true
};
context.function_ids.len()
];
@@ -258,6 +259,7 @@ fn parse_function<'a>(
return_type,
nodes: fixed_nodes,
num_dynamic_constants,
+ entry: true,
},
))
}
diff --git a/hercules_opt/src/delete_uncalled.rs b/hercules_opt/src/delete_uncalled.rs
new file mode 100644
index 00000000..fe99b5fb
--- /dev/null
+++ b/hercules_opt/src/delete_uncalled.rs
@@ -0,0 +1,81 @@
+extern crate bitvec;
+extern crate hercules_ir;
+use self::bitvec::prelude::*;
+
+use self::hercules_ir::callgraph::*;
+use self::hercules_ir::ir::*;
+
+use crate::*;
+
+/**
+ * First renumber functions by deleting all functions unreachable from any entry function.
+ * Then, updates all functions to use the new function numbering.
+ * Returns a vec where the element at index i is the new idx of the ith function (if it was not
+ * deleted).
+ */
+pub fn delete_uncalled(
+ editors: &mut Vec<FunctionEditor>,
+ callgraph: &CallGraph,
+) -> Vec<Option<usize>> {
+ // Step 1. Identify which functions are not reachable from entry nodes using BFS.
+ let mut reachable = bitvec![u8, Lsb0; 0; editors.len()];
+ let mut worklist = vec![];
+ for (idx, editor) in editors.iter().enumerate() {
+ if editor.func().entry {
+ worklist.push(idx);
+ reachable.set(idx, true);
+ }
+ }
+
+ while let Some(func_idx) = worklist.pop() {
+ for callee in callgraph.get_callees(FunctionID::new(func_idx)) {
+ if !reachable[callee.idx()] {
+ reachable.set(callee.idx(), true);
+ worklist.push(callee.idx());
+ }
+ }
+ }
+
+ // Step 2. Compute the new index of each function, which is obtained by
+ // deleteting all unreachable non-entry functions before it
+ let mut new_idx = vec![None; editors.len()];
+ let mut deleted_count = 0;
+ for idx in 0..editors.len() {
+ if !reachable[idx] {
+ deleted_count += 1;
+ } else {
+ new_idx[idx] = Some(idx - deleted_count);
+ }
+ }
+
+ // Step 3. Update all function callsites to use new indices. We assume
+ // that all nodes in all functions will be mutable, so panic if any
+ // edit fails.
+ for editor in editors.iter_mut() {
+ let callsites: Vec<_> = editor
+ .node_ids()
+ .filter(|id| editor.func().nodes[id.idx()].is_call())
+ .collect();
+
+ for callsite in callsites {
+ let success = editor.edit(|mut edit| {
+ let (control, function, dynamic_constants, args) =
+ edit.get_node(callsite).try_call().unwrap();
+ let new_node = edit.add_node(Node::Call {
+ control,
+ function: FunctionID::new(new_idx[function.idx()].unwrap()),
+ dynamic_constants: dynamic_constants.clone(),
+ args: args.clone(),
+ });
+ let edit = edit.delete_node(callsite)?;
+ edit.replace_all_uses(callsite, new_node)
+ });
+ assert!(
+ success,
+ "Expected all delete_uncalled callsite edits to succeed!"
+ );
+ }
+ }
+
+ new_idx
+}
diff --git a/hercules_opt/src/lib.rs b/hercules_opt/src/lib.rs
index ab30050d..032a2619 100644
--- a/hercules_opt/src/lib.rs
+++ b/hercules_opt/src/lib.rs
@@ -2,6 +2,7 @@
pub mod ccp;
pub mod dce;
+pub mod delete_uncalled;
pub mod editor;
pub mod fork_guard_elim;
pub mod forkify;
@@ -16,6 +17,7 @@ pub mod utils;
pub use crate::ccp::*;
pub use crate::dce::*;
+pub use crate::delete_uncalled::*;
pub use crate::editor::*;
pub use crate::fork_guard_elim::*;
pub use crate::forkify::*;
diff --git a/hercules_opt/src/pass.rs b/hercules_opt/src/pass.rs
index 2798ebc6..82adfeb5 100644
--- a/hercules_opt/src/pass.rs
+++ b/hercules_opt/src/pass.rs
@@ -43,6 +43,7 @@ pub enum Pass {
// Parameterized over where to serialize module to.
Serialize(String),
InterproceduralSROA,
+ DeleteUncalled,
}
/*
@@ -656,6 +657,50 @@ impl PassManager {
}
self.clear_analyses();
}
+ Pass::DeleteUncalled => {
+ self.make_def_uses();
+ self.make_callgraph();
+ let def_uses = self.def_uses.as_ref().unwrap();
+ let callgraph = self.callgraph.as_ref().unwrap();
+ let constants_ref = RefCell::new(std::mem::take(&mut self.module.constants));
+ let dynamic_constants_ref =
+ RefCell::new(std::mem::take(&mut self.module.dynamic_constants));
+ let types_ref = RefCell::new(std::mem::take(&mut self.module.types));
+
+ // By default in an editor all nodes are mutable, which is desired in this case
+ // since we are only modifying the IDs of functions that we call.
+ let mut editors: Vec<_> =
+ zip(self.module.functions.iter_mut(), def_uses.iter())
+ .map(|(func, def_use)| {
+ FunctionEditor::new(
+ func,
+ &constants_ref,
+ &dynamic_constants_ref,
+ &types_ref,
+ def_use,
+ )
+ })
+ .collect();
+
+ let new_idx = delete_uncalled(&mut editors, callgraph);
+ self.module.constants = constants_ref.take();
+ self.module.dynamic_constants = dynamic_constants_ref.take();
+ self.module.types = types_ref.take();
+
+ let edits: Vec<_> = editors.into_iter().map(|editor| editor.edits()).collect();
+ for idx in 0..edits.len() {
+ if let Some(plans) = self.plans.as_mut() {
+ repair_plan(&mut plans[idx], &self.module.functions[idx], &edits[idx]);
+ }
+ let grave_mapping = self.module.functions[idx].delete_gravestones();
+ if let Some(plans) = self.plans.as_mut() {
+ plans[idx].fix_gravestones(&grave_mapping);
+ }
+ }
+
+ self.fix_deleted_functions(&new_idx);
+ self.clear_analyses();
+ }
Pass::Verify => {
let (
def_uses,
@@ -850,4 +895,15 @@ impl PassManager {
pub fn get_manifests(self) -> HashMap<String, Manifest> {
self.manifests.unwrap()
}
+
+ fn fix_deleted_functions(&mut self, id_mapping: &[Option<usize>]) {
+ let mut idx = 0;
+
+ // Rust does not like enumerate here, so use
+ // idx outside as a hack to make it happy.
+ self.module.functions.retain(|_| {
+ idx += 1;
+ id_mapping[idx - 1].is_some()
+ });
+ }
}
diff --git a/juno_frontend/src/codegen.rs b/juno_frontend/src/codegen.rs
index c99978c8..31c85ae9 100644
--- a/juno_frontend/src/codegen.rs
+++ b/juno_frontend/src/codegen.rs
@@ -139,6 +139,7 @@ impl CodeGenerator<'_> {
return_type,
func.num_dyn_consts as u32,
func.num_labels,
+ func.entry,
)
.unwrap();
diff --git a/juno_frontend/src/labeled_builder.rs b/juno_frontend/src/labeled_builder.rs
index 2386d81c..3cbf36da 100644
--- a/juno_frontend/src/labeled_builder.rs
+++ b/juno_frontend/src/labeled_builder.rs
@@ -54,10 +54,15 @@ impl<'a> LabeledBuilder<'a> {
return_type: TypeID,
num_dynamic_constants: u32,
num_labels: usize,
+ entry: bool,
) -> Result<(FunctionID, NodeID), String> {
- let (func, entry) =
- self.builder
- .create_function(name, param_types, return_type, num_dynamic_constants)?;
+ let (func, entry) = self.builder.create_function(
+ name,
+ param_types,
+ return_type,
+ num_dynamic_constants,
+ entry,
+ )?;
self.label_tree.insert(
func,
diff --git a/juno_frontend/src/lib.rs b/juno_frontend/src/lib.rs
index b2af3381..4713cfeb 100644
--- a/juno_frontend/src/lib.rs
+++ b/juno_frontend/src/lib.rs
@@ -154,6 +154,15 @@ pub fn compile_ir(
pm.add_pass(hercules_opt::pass::Pass::Xdot(true));
}
add_pass!(pm, verify, Inline);
+ if x_dot {
+ pm.add_pass(hercules_opt::pass::Pass::Xdot(true));
+ }
+ // Inlining may make some functions uncalled, so run this pass.
+ // In general, this should always be run after inlining.
+ add_pass!(pm, verify, DeleteUncalled);
+ if x_dot {
+ pm.add_pass(hercules_opt::pass::Pass::Xdot(true));
+ }
// Run SROA pretty early (though after inlining which can make SROA more effective) so that
// CCP, GVN, etc. can work on the result of SROA
add_pass!(pm, verify, InterproceduralSROA);
--
GitLab