Use a unique id instead of by-address indexing

Author: Nadrieril

compiler/rustc_pattern_analysis/src/pat.rs

@@ -9,6 +9,17 @@ use crate::TypeCx;
 
 use self::Constructor::*;
 
+/// A globally unique id to distinguish patterns.
+#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash)]
+pub(crate) struct PatId(u32);
+impl PatId {
+    fn new() -> Self {
+        use std::sync::atomic::{AtomicU32, Ordering};
+        static PAT_ID: AtomicU32 = AtomicU32::new(0);
+        PatId(PAT_ID.fetch_add(1, Ordering::SeqCst))
+    }
+}
+
 /// Values and patterns can be represented as a constructor applied to some fields. This represents
 /// a pattern in this form. A `DeconstructedPat` will almost always come from user input; the only
 /// exception are some `Wildcard`s introduced during pattern lowering.
@@ -24,11 +35,13 @@ pub struct DeconstructedPat<Cx: TypeCx> {
     /// Extra data to store in a pattern. `None` if the pattern is a wildcard that does not
     /// correspond to a user-supplied pattern.
     data: Option<Cx::PatData>,
+    /// Globally-unique id used to track usefulness at the level of subpatterns.
+    pub(crate) uid: PatId,
 }
 
 impl<Cx: TypeCx> DeconstructedPat<Cx> {
     pub fn wildcard(ty: Cx::Ty) -> Self {
-        DeconstructedPat { ctor: Wildcard, fields: Vec::new(), ty, data: None }
+        DeconstructedPat { ctor: Wildcard, fields: Vec::new(), ty, data: None, uid: PatId::new() }
     }
 
     pub fn new(
@@ -37,7 +50,7 @@ impl<Cx: TypeCx> DeconstructedPat<Cx> {
         ty: Cx::Ty,
         data: Cx::PatData,
     ) -> Self {
-        DeconstructedPat { ctor, fields, ty, data: Some(data) }
+        DeconstructedPat { ctor, fields, ty, data: Some(data), uid: PatId::new() }
     }
 
     pub(crate) fn is_or_pat(&self) -> bool {

compiler/rustc_pattern_analysis/src/usefulness.rs

@@ -713,10 +713,9 @@ use rustc_hash::FxHashSet;
 use rustc_index::bit_set::BitSet;
 use smallvec::{smallvec, SmallVec};
 use std::fmt;
-use std::ops::Deref;
 
 use crate::constructor::{Constructor, ConstructorSet, IntRange};
-use crate::pat::{DeconstructedPat, PatOrWild, WitnessPat};
+use crate::pat::{DeconstructedPat, PatId, PatOrWild, WitnessPat};
 use crate::{Captures, MatchArm, TypeCx};
 
 use self::ValidityConstraint::*;
@@ -728,36 +727,13 @@ pub fn ensure_sufficient_stack<R>(f: impl FnOnce() -> R) -> R {
     f()
 }
 
-/// Wrapper type for by-address hashing. Comparison and hashing of the wrapped pointer type will be
-/// based on the address of its contents, rather than their value.
-struct ByAddress<T>(T);
-
-impl<T: Deref> ByAddress<T> {
-    fn addr(&self) -> *const T::Target {
-        (&*self.0) as *const _
-    }
-}
-/// Raw pointer hashing and comparison.
-impl<T: Deref> std::hash::Hash for ByAddress<T> {
-    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
-        self.addr().hash(state)
-    }
-}
-impl<T: Deref> PartialEq for ByAddress<T> {
-    fn eq(&self, other: &Self) -> bool {
-        std::ptr::eq(self.addr(), other.addr())
-    }
-}
-impl<T: Deref> Eq for ByAddress<T> {}
-
 /// Context that provides information for usefulness checking.
-struct UsefulnessCtxt<'a, 'p, Cx: TypeCx> {
+struct UsefulnessCtxt<'a, Cx: TypeCx> {
     /// The context for type information.
     tycx: &'a Cx,
     /// Collect the patterns found useful during usefulness checking. This is used to lint
-    /// unreachable (sub)patterns. We distinguish patterns by their address to avoid needing to
-    /// inspect the contents. They'll all be distinct anyway since they carry a `Span`.
-    useful_subpatterns: FxHashSet<ByAddress<&'p DeconstructedPat<Cx>>>,
+    /// unreachable (sub)patterns.
+    useful_subpatterns: FxHashSet<PatId>,
 }
 
 /// Context that provides information local to a place under investigation.
@@ -1378,7 +1354,7 @@ impl<Cx: TypeCx> WitnessMatrix<Cx> {
 /// We can however get false negatives because exhaustiveness does not explore all cases. See the
 /// section on relevancy at the top of the file.
 fn collect_overlapping_range_endpoints<'p, Cx: TypeCx>(
-    mcx: &mut UsefulnessCtxt<'_, 'p, Cx>,
+    mcx: &mut UsefulnessCtxt<'_, Cx>,
     overlap_range: IntRange,
     matrix: &Matrix<'p, Cx>,
     specialized_matrix: &Matrix<'p, Cx>,
@@ -1451,7 +1427,7 @@ fn collect_overlapping_range_endpoints<'p, Cx: TypeCx>(
 /// This is all explained at the top of the file.
 #[instrument(level = "debug", skip(mcx, is_top_level), ret)]
 fn compute_exhaustiveness_and_usefulness<'a, 'p, Cx: TypeCx>(
-    mcx: &mut UsefulnessCtxt<'a, 'p, Cx>,
+    mcx: &mut UsefulnessCtxt<'a, Cx>,
     matrix: &mut Matrix<'p, Cx>,
     is_top_level: bool,
 ) -> Result<WitnessMatrix<Cx>, Cx::Error> {
@@ -1580,7 +1556,7 @@ fn compute_exhaustiveness_and_usefulness<'a, 'p, Cx: TypeCx>(
     for row in matrix.rows() {
         if row.useful {
             if let PatOrWild::Pat(pat) = row.head() {
-                mcx.useful_subpatterns.insert(ByAddress(pat));
+                mcx.useful_subpatterns.insert(pat.uid);
             }
         }
     }
@@ -1602,14 +1578,14 @@ pub enum Usefulness<'p, Cx: TypeCx> {
 
 /// Report whether this pattern was found useful, and its subpatterns that were not useful if any.
 fn collect_pattern_usefulness<'p, Cx: TypeCx>(
-    useful_subpatterns: &FxHashSet<ByAddress<&'p DeconstructedPat<Cx>>>,
+    useful_subpatterns: &FxHashSet<PatId>,
     pat: &'p DeconstructedPat<Cx>,
 ) -> Usefulness<'p, Cx> {
     fn pat_is_useful<'p, Cx: TypeCx>(
-        useful_subpatterns: &FxHashSet<ByAddress<&'p DeconstructedPat<Cx>>>,
+        useful_subpatterns: &FxHashSet<PatId>,
         pat: &'p DeconstructedPat<Cx>,
     ) -> bool {
-        if useful_subpatterns.contains(&ByAddress(pat)) {
+        if useful_subpatterns.contains(&pat.uid) {
             true
         } else if pat.is_or_pat() && pat.iter_fields().any(|f| pat_is_useful(useful_subpatterns, f))
         {