Rollup merge of #118029 - saethlin:allocid-gc, r=RalfJung

Expand Miri's BorTag GC to a Provenance GC

As suggested in https://github.com/rust-lang/miri/issues/3080#issuecomment-1732505573

We previously solved memory growth issues associated with the Stacked Borrows and Tree Borrows runtimes with a GC. But of course we also have state accumulation associated with whole allocations elsewhere in the interpreter, and this PR starts tackling those.

To do this, we expand the visitor for the GC so that it can visit a BorTag or an AllocId. Instead of collecting all live AllocIds into a single HashSet, we just collect from the Machine itself then go through an accessor `InterpCx::is_alloc_live` which checks a number of allocation data structures in the core interpreter. This avoids the overhead of all the inserts that collecting their keys would require.

r? ``@RalfJung``

3 files changed, 25 insertions, 0 deletions

diff --git a/compiler/rustc_const_eval/src/const_eval/machine.rs b/compiler/rustc_const_eval/src/const_eval/machine.rs
index 4b447229c5f..739dbcc5aba 100644
--- a/compiler/rustc_const_eval/src/const_eval/machine.rs
+++ b/compiler/rustc_const_eval/src/const_eval/machine.rs
@@ -108,6 +108,14 @@ impl<K: Hash + Eq, V> interpret::AllocMap<K, V> for FxIndexMap<K, V> {
     }
 
     #[inline(always)]
+    fn contains_key_ref<Q: ?Sized + Hash + Eq>(&self, k: &Q) -> bool
+    where
+        K: Borrow<Q>,
+    {
+        FxIndexMap::contains_key(self, k)
+    }
+
+    #[inline(always)]
     fn insert(&mut self, k: K, v: V) -> Option<V> {
         FxIndexMap::insert(self, k, v)
     }
diff --git a/compiler/rustc_const_eval/src/interpret/machine.rs b/compiler/rustc_const_eval/src/interpret/machine.rs
index 61fe9151d8b..6617f6f2ffb 100644
--- a/compiler/rustc_const_eval/src/interpret/machine.rs
+++ b/compiler/rustc_const_eval/src/interpret/machine.rs
@@ -49,6 +49,14 @@ pub trait AllocMap<K: Hash + Eq, V> {
     where
         K: Borrow<Q>;
 
+    /// Callers should prefer [`AllocMap::contains_key`] when it is possible to call because it may
+    /// be more efficient. This function exists for callers that only have a shared reference
+    /// (which might make it slightly less efficient than `contains_key`, e.g. if
+    /// the data is stored inside a `RefCell`).
+    fn contains_key_ref<Q: ?Sized + Hash + Eq>(&self, k: &Q) -> bool
+    where
+        K: Borrow<Q>;
+
     /// Inserts a new entry into the map.
     fn insert(&mut self, k: K, v: V) -> Option<V>;
 
diff --git a/compiler/rustc_const_eval/src/interpret/memory.rs b/compiler/rustc_const_eval/src/interpret/memory.rs
index 16905e93bf1..d5b165a7415 100644
--- a/compiler/rustc_const_eval/src/interpret/memory.rs
+++ b/compiler/rustc_const_eval/src/interpret/memory.rs
@@ -692,6 +692,15 @@ impl<'mir, 'tcx: 'mir, M: Machine<'mir, 'tcx>> InterpCx<'mir, 'tcx, M> {
         Ok((&mut alloc.extra, machine))
     }
 
+    /// Check whether an allocation is live. This is faster than calling
+    /// [`InterpCx::get_alloc_info`] if all you need to check is whether the kind is
+    /// [`AllocKind::Dead`] because it doesn't have to look up the type and layout of statics.
+    pub fn is_alloc_live(&self, id: AllocId) -> bool {
+        self.tcx.try_get_global_alloc(id).is_some()
+            || self.memory.alloc_map.contains_key_ref(&id)
+            || self.memory.extra_fn_ptr_map.contains_key(&id)
+    }
+
     /// Obtain the size and alignment of an allocation, even if that allocation has
     /// been deallocated.
     pub fn get_alloc_info(&self, id: AllocId) -> (Size, Align, AllocKind) {