transmutability: remove NFA intermediate representation

Prior to this commit, the transmutability analysis used an intermediate
NFA representation of type layout. We then determinized this
representation into a DFA, upon which we ran the core transmutability
analysis. Unfortunately, determinizing NFAs is expensive. In this
commit, we avoid NFAs entirely by observing that Rust `union`s are the
only source of nondeterminism and that it is comparatively cheap to
compute the DFA union of DFAs.

We also implement Graphviz DOT debug formatting of DFAs.

Fixes rust-lang/project-safe-transmute#23
Fixes rust-lang/project-safe-transmute#24

Author: jswrenn

compiler/rustc_transmute/src/layout/dfa.rs

@@ -1,12 +1,10 @@
 use std::fmt;
 use std::sync::atomic::{AtomicU32, Ordering};
 
-use tracing::instrument;
-
-use super::{Byte, Nfa, Ref, nfa};
+use super::{Byte, Ref, Tree, Uninhabited};
 use crate::Map;
 
-#[derive(PartialEq, Clone, Debug)]
+#[derive(PartialEq, Clone)]
 pub(crate) struct Dfa<R>
 where
     R: Ref,
@@ -34,35 +32,15 @@ where
     }
 }
 
-impl<R> Transitions<R>
-where
-    R: Ref,
-{
-    #[cfg(test)]
-    fn insert(&mut self, transition: Transition<R>, state: State) {
-        match transition {
-            Transition::Byte(b) => {
-                self.byte_transitions.insert(b, state);
-            }
-            Transition::Ref(r) => {
-                self.ref_transitions.insert(r, state);
-            }
-        }
-    }
-}
-
 /// The states in a `Nfa` represent byte offsets.
 #[derive(Hash, Eq, PartialEq, PartialOrd, Ord, Copy, Clone)]
-pub(crate) struct State(u32);
+pub(crate) struct State(pub(crate) u32);
 
-#[cfg(test)]
-#[derive(Hash, Eq, PartialEq, Clone, Copy)]
-pub(crate) enum Transition<R>
-where
-    R: Ref,
-{
-    Byte(Byte),
-    Ref(R),
+impl State {
+    pub(crate) fn new() -> Self {
+        static COUNTER: AtomicU32 = AtomicU32::new(0);
+        Self(COUNTER.fetch_add(1, Ordering::SeqCst))
+    }
 }
 
 impl fmt::Debug for State {
@@ -71,19 +49,6 @@ impl fmt::Debug for State {
     }
 }
 
-#[cfg(test)]
-impl<R> fmt::Debug for Transition<R>
-where
-    R: Ref,
-{
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match &self {
-            Self::Byte(b) => b.fmt(f),
-            Self::Ref(r) => r.fmt(f),
-        }
-    }
-}
-
 impl<R> Dfa<R>
 where
     R: Ref,
@@ -94,58 +59,153 @@ where
         let start = State::new();
         let accepting = State::new();
 
-        transitions.entry(start).or_default().insert(Transition::Byte(Byte::Init(0x00)), accepting);
+        transitions.entry(start).or_default().byte_transitions.insert(Byte::Init(0x00), accepting);
 
-        transitions.entry(start).or_default().insert(Transition::Byte(Byte::Init(0x01)), accepting);
+        transitions.entry(start).or_default().byte_transitions.insert(Byte::Init(0x01), accepting);
 
         Self { transitions, start, accepting }
     }
 
-    #[instrument(level = "debug")]
-    pub(crate) fn from_nfa(nfa: Nfa<R>) -> Self {
-        let Nfa { transitions: nfa_transitions, start: nfa_start, accepting: nfa_accepting } = nfa;
+    pub(crate) fn unit() -> Self {
+        let transitions: Map<State, Transitions<R>> = Map::default();
+        let start = State::new();
+        let accepting = start;
+
+        Self { transitions, start, accepting }
+    }
 
-        let mut dfa_transitions: Map<State, Transitions<R>> = Map::default();
-        let mut nfa_to_dfa: Map<nfa::State, State> = Map::default();
-        let dfa_start = State::new();
-        nfa_to_dfa.insert(nfa_start, dfa_start);
+    pub(crate) fn from_byte(byte: Byte) -> Self {
+        let mut transitions: Map<State, Transitions<R>> = Map::default();
+        let start = State::new();
+        let accepting = State::new();
 
-        let mut queue = vec![(nfa_start, dfa_start)];
+        transitions.entry(start).or_default().byte_transitions.insert(byte, accepting);
 
-        while let Some((nfa_state, dfa_state)) = queue.pop() {
-            if nfa_state == nfa_accepting {
-                continue;
-            }
+        Self { transitions, start, accepting }
+    }
+
+    pub(crate) fn from_ref(r: R) -> Self {
+        let mut transitions: Map<State, Transitions<R>> = Map::default();
+        let start = State::new();
+        let accepting = State::new();
+
+        transitions.entry(start).or_default().ref_transitions.insert(r, accepting);
 
-            for (nfa_transition, next_nfa_states) in nfa_transitions[&nfa_state].iter() {
-                let dfa_transitions =
-                    dfa_transitions.entry(dfa_state).or_insert_with(Default::default);
-
-                let mapped_state = next_nfa_states.iter().find_map(|x| nfa_to_dfa.get(x).copied());
-
-                let next_dfa_state = match nfa_transition {
-                    &nfa::Transition::Byte(b) => *dfa_transitions
-                        .byte_transitions
-                        .entry(b)
-                        .or_insert_with(|| mapped_state.unwrap_or_else(State::new)),
-                    &nfa::Transition::Ref(r) => *dfa_transitions
-                        .ref_transitions
-                        .entry(r)
-                        .or_insert_with(|| mapped_state.unwrap_or_else(State::new)),
-                };
-
-                for &next_nfa_state in next_nfa_states {
-                    nfa_to_dfa.entry(next_nfa_state).or_insert_with(|| {
-                        queue.push((next_nfa_state, next_dfa_state));
-                        next_dfa_state
-                    });
+        Self { transitions, start, accepting }
+    }
+
+    pub(crate) fn from_tree(tree: Tree<!, R>) -> Result<Self, Uninhabited> {
+        Ok(match tree {
+            Tree::Byte(b) => Self::from_byte(b),
+            Tree::Ref(r) => Self::from_ref(r),
+            Tree::Alt(alts) => {
+                let mut alts = alts.into_iter().map(Self::from_tree);
+                let mut dfa = alts.next().ok_or(Uninhabited)??;
+                for alt in alts {
+                    dfa = dfa.union(alt?, State::new);
                 }
+                dfa
+            }
+            Tree::Seq(elts) => {
+                let mut dfa = Self::unit();
+                for elt in elts.into_iter().map(Self::from_tree) {
+                    dfa = dfa.concat(elt?);
+                }
+                dfa
+            }
+        })
+    }
+
+    /// Concatenate two `Dfa`s.
+    pub(crate) fn concat(self, other: Self) -> Self {
+        if self.start == self.accepting {
+            return other;
+        } else if other.start == other.accepting {
+            return self;
+        }
+
+        let start = self.start;
+        let accepting = other.accepting;
+
+        let mut transitions: Map<State, Transitions<R>> = self.transitions;
+
+        for (source, transition) in other.transitions {
+            let fix_state = |state| if state == other.start { self.accepting } else { state };
+            let entry = transitions.entry(fix_state(source)).or_default();
+            for (edge, destination) in transition.byte_transitions {
+                entry.byte_transitions.insert(edge, fix_state(destination));
+            }
+            for (edge, destination) in transition.ref_transitions {
+                entry.ref_transitions.insert(edge, fix_state(destination));
             }
         }
 
-        let dfa_accepting = nfa_to_dfa[&nfa_accepting];
+        Self { transitions, start, accepting }
+    }
+
+    /// Compute the union of two `Nfa`s.
+    pub(crate) fn union(self, other: Self, mut new_state: impl FnMut() -> State) -> Self {
+        let a = self;
+        let b = other;
+
+        let accepting = new_state();
+
+        let mut mapping: Map<(Option<State>, Option<State>), State> = Map::default();
+
+        let mut mapped = |(a_state, b_state)| {
+            if Some(a.accepting) == a_state || Some(b.accepting) == b_state {
+                accepting
+            } else {
+                *mapping.entry((a_state, b_state)).or_insert_with(&mut new_state)
+            }
+        };
+
+        let start = mapped((Some(a.start), Some(b.start)));
+        let mut transitions: Map<State, Transitions<R>> = Map::default();
+        let mut queue = vec![(Some(a.start), Some(b.start))];
+        let empty_transitions = Transitions::default();
+
+        while let Some((a_src, b_src)) = queue.pop() {
+            let a_transitions =
+                a_src.and_then(|a_src| a.transitions.get(&a_src)).unwrap_or(&empty_transitions);
+            let b_transitions =
+                b_src.and_then(|b_src| b.transitions.get(&b_src)).unwrap_or(&empty_transitions);
+
+            let byte_transitions =
+                a_transitions.byte_transitions.keys().chain(b_transitions.byte_transitions.keys());
+
+            for byte_transition in byte_transitions {
+                let a_dst = a_transitions.byte_transitions.get(byte_transition).copied();
+
+                let b_dst = b_transitions.byte_transitions.get(byte_transition).copied();
+
+                assert!(a_dst.is_some() || b_dst.is_some());
+                let src = mapped((a_src, b_src));
+                let dst = mapped((a_dst, b_dst));
+
+                transitions.entry(src).or_default().byte_transitions.insert(*byte_transition, dst);
+
+                queue.push((a_dst, b_dst))
+            }
+
+            let ref_transitions =
+                a_transitions.ref_transitions.keys().chain(b_transitions.ref_transitions.keys());
 
-        Self { transitions: dfa_transitions, start: dfa_start, accepting: dfa_accepting }
+            for ref_transition in ref_transitions {
+                let a_dst = a_transitions.ref_transitions.get(ref_transition).copied();
+
+                let b_dst = b_transitions.ref_transitions.get(ref_transition).copied();
+
+                let src = mapped((a_src, b_src));
+                let dst = mapped((a_dst, b_dst));
+
+                transitions.entry(src).or_default().ref_transitions.insert(*ref_transition, dst);
+
+                queue.push((a_dst, b_dst))
+            }
+        }
+
+        Self { transitions, start, accepting }
     }
 
     pub(crate) fn bytes_from(&self, start: State) -> Option<&Map<Byte, State>> {
@@ -159,24 +219,47 @@ where
     pub(crate) fn refs_from(&self, start: State) -> Option<&Map<R, State>> {
         Some(&self.transitions.get(&start)?.ref_transitions)
     }
-}
 
-impl State {
-    pub(crate) fn new() -> Self {
-        static COUNTER: AtomicU32 = AtomicU32::new(0);
-        Self(COUNTER.fetch_add(1, Ordering::SeqCst))
+    #[cfg(test)]
+    pub(crate) fn from_edges<B: Copy + Into<Byte>>(
+        start: u32,
+        accept: u32,
+        edges: &[(u32, B, u32)],
+    ) -> Self {
+        let start = State(start);
+        let accepting = State(accept);
+        let mut transitions: Map<State, Transitions<R>> = Map::default();
+
+        for &(src, edge, dst) in edges {
+            let src = State(src);
+            let dst = State(dst);
+            let old = transitions.entry(src).or_default().byte_transitions.insert(edge.into(), dst);
+            assert!(old.is_none());
+        }
+
+        Self { start, accepting, transitions }
     }
 }
 
-#[cfg(test)]
-impl<R> From<nfa::Transition<R>> for Transition<R>
+impl<R> fmt::Debug for Dfa<R>
 where
     R: Ref,
 {
-    fn from(nfa_transition: nfa::Transition<R>) -> Self {
-        match nfa_transition {
-            nfa::Transition::Byte(byte) => Transition::Byte(byte),
-            nfa::Transition::Ref(r) => Transition::Ref(r),
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        writeln!(f, "digraph {{")?;
+        writeln!(f, "    {:?} [shape = doublecircle]", self.start)?;
+        writeln!(f, "    {:?} [shape = doublecircle]", self.accepting)?;
+
+        for (src, transitions) in self.transitions.iter() {
+            for (t, dst) in transitions.byte_transitions.iter() {
+                writeln!(f, "    {src:?} -> {dst:?} [label=\"{t:?}\"]")?;
+            }
+
+            for (t, dst) in transitions.ref_transitions.iter() {
+                writeln!(f, "    {src:?} -> {dst:?} [label=\"{t:?}\"]")?;
+            }
         }
+
+        writeln!(f, "}}")
     }
 }

compiler/rustc_transmute/src/layout/mod.rs

@@ -4,8 +4,8 @@ use std::hash::Hash;
 pub(crate) mod tree;
 pub(crate) use tree::Tree;
 
-pub(crate) mod nfa;
-pub(crate) use nfa::Nfa;
+//pub(crate) mod nfa;
+//pub(crate) use nfa::Nfa;
 
 pub(crate) mod dfa;
 pub(crate) use dfa::Dfa;
@@ -29,6 +29,13 @@ impl fmt::Debug for Byte {
     }
 }
 
+#[cfg(test)]
+impl From<u8> for Byte {
+    fn from(src: u8) -> Self {
+        Self::Init(src)
+    }
+}
+
 pub(crate) trait Def: Debug + Hash + Eq + PartialEq + Copy + Clone {
     fn has_safety_invariants(&self) -> bool;
 }