Drop the query result memmap before serializing it back.

Author: cjgillot

compiler/rustc_incremental/src/persist/file_format.rs

@@ -52,7 +52,10 @@ where
     // Delete the old file, if any.
     // Note: It's important that we actually delete the old file and not just
     // truncate and overwrite it, since it might be a shared hard-link, the
-    // underlying data of which we don't want to modify
+    // underlying data of which we don't want to modify.
+    //
+    // We have to ensure we have dropped the memory maps to this file
+    // before performing this removal.
     match fs::remove_file(&path_buf) {
         Ok(()) => {
             debug!("save: remove old file");
@@ -114,6 +117,12 @@ pub fn read_file(
         Err(err) if err.kind() == io::ErrorKind::NotFound => return Ok(None),
         Err(err) => return Err(err),
     };
+    // SAFETY: This process must not modify nor remove the backing file while the memory map lives.
+    // For the dep-graph and the work product index, it is as soon as the decoding is done.
+    // For the query result cache, the memory map is dropped in save_dep_graph before calling
+    // save_in and trying to remove the backing file.
+    //
+    // There is no way to prevent another process from modifying this file.
     let mmap = unsafe { Mmap::map(file) }?;
 
     let mut file = io::Cursor::new(&*mmap);

compiler/rustc_incremental/src/persist/save.rs

@@ -42,6 +42,11 @@ pub fn save_dep_graph(tcx: TyCtxt<'_>) {
         join(
             move || {
                 sess.time("incr_comp_persist_result_cache", || {
+                    // Drop the memory map so that we can remove the file and write to it.
+                    if let Some(odc) = &tcx.on_disk_cache {
+                        odc.drop_serialized_data(tcx);
+                    }
+
                     file_format::save_in(sess, query_cache_path, "query cache", |e| {
                         encode_query_cache(tcx, e)
                     });

compiler/rustc_middle/src/ty/context.rs

@@ -101,6 +101,8 @@ pub trait OnDiskCache<'tcx>: rustc_data_structures::sync::Sync {
     fn register_reused_dep_node(&self, tcx: TyCtxt<'tcx>, dep_node: &DepNode);
     fn store_foreign_def_id_hash(&self, def_id: DefId, hash: DefPathHash);
 
+    fn drop_serialized_data(&self, tcx: TyCtxt<'tcx>);
+
     fn serialize(&self, tcx: TyCtxt<'tcx>, encoder: &mut FileEncoder) -> FileEncodeResult;
 }
 

compiler/rustc_query_impl/src/on_disk_cache.rs

@@ -1,7 +1,7 @@
 use crate::QueryCtxt;
 use rustc_data_structures::fx::{FxHashMap, FxHashSet, FxIndexSet};
 use rustc_data_structures::memmap::Mmap;
-use rustc_data_structures::sync::{HashMapExt, Lock, Lrc, OnceCell};
+use rustc_data_structures::sync::{HashMapExt, Lock, Lrc, OnceCell, RwLock};
 use rustc_data_structures::unhash::UnhashMap;
 use rustc_hir::def_id::{CrateNum, DefId, DefIndex, LocalDefId, StableCrateId, LOCAL_CRATE};
 use rustc_hir::definitions::DefPathHash;
@@ -43,7 +43,7 @@ const TAG_EXPN_DATA: u8 = 1;
 /// any side effects that have been emitted during a query.
 pub struct OnDiskCache<'sess> {
     // The complete cache data in serialized form.
-    serialized_data: Option<Mmap>,
+    serialized_data: RwLock<Option<Mmap>>,
 
     // Collects all `QuerySideEffects` created during the current compilation
     // session.
@@ -206,7 +206,7 @@ impl<'sess> rustc_middle::ty::OnDiskCache<'sess> for OnDiskCache<'sess> {
         };
 
         Self {
-            serialized_data: Some(data),
+            serialized_data: RwLock::new(Some(data)),
             file_index_to_stable_id: footer.file_index_to_stable_id,
             file_index_to_file: Default::default(),
             cnum_map: OnceCell::new(),
@@ -227,7 +227,7 @@ impl<'sess> rustc_middle::ty::OnDiskCache<'sess> for OnDiskCache<'sess> {
 
     fn new_empty(source_map: &'sess SourceMap) -> Self {
         Self {
-            serialized_data: None,
+            serialized_data: RwLock::new(None),
             file_index_to_stable_id: Default::default(),
             file_index_to_file: Default::default(),
             cnum_map: OnceCell::new(),
@@ -246,7 +246,26 @@ impl<'sess> rustc_middle::ty::OnDiskCache<'sess> for OnDiskCache<'sess> {
         }
     }
 
-    fn serialize(&self, tcx: TyCtxt<'sess>, encoder: &mut FileEncoder) -> FileEncodeResult {
+    fn drop_serialized_data(&self, tcx: TyCtxt<'tcx>) {
+        // Register any dep nodes that we reused from the previous session,
+        // but didn't `DepNode::construct` in this session. This ensures
+        // that their `DefPathHash` to `RawDefId` mappings are registered
+        // in 'latest_foreign_def_path_hashes' if necessary, since that
+        // normally happens in `DepNode::construct`.
+        tcx.dep_graph.register_reused_dep_nodes(tcx);
+
+        // Load everything into memory so we can write it out to the on-disk
+        // cache. The vast majority of cacheable query results should already
+        // be in memory, so this should be a cheap operation.
+        // Do this *before* we clone 'latest_foreign_def_path_hashes', since
+        // loading existing queries may cause us to create new DepNodes, which
+        // may in turn end up invoking `store_foreign_def_id_hash`
+        tcx.dep_graph.exec_cache_promotions(QueryCtxt::from_tcx(tcx));
+
+        *self.serialized_data.write() = None;
+    }
+
+    fn serialize<'tcx>(&self, tcx: TyCtxt<'tcx>, encoder: &mut FileEncoder) -> FileEncodeResult {
         // Serializing the `DepGraph` should not modify it.
         tcx.dep_graph.with_ignore(|| {
             // Allocate `SourceFileIndex`es.
@@ -268,21 +287,6 @@ impl<'sess> rustc_middle::ty::OnDiskCache<'sess> for OnDiskCache<'sess> {
                 (file_to_file_index, file_index_to_stable_id)
             };
 
-            // Register any dep nodes that we reused from the previous session,
-            // but didn't `DepNode::construct` in this session. This ensures
-            // that their `DefPathHash` to `RawDefId` mappings are registered
-            // in 'latest_foreign_def_path_hashes' if necessary, since that
-            // normally happens in `DepNode::construct`.
-            tcx.dep_graph.register_reused_dep_nodes(tcx);
-
-            // Load everything into memory so we can write it out to the on-disk
-            // cache. The vast majority of cacheable query results should already
-            // be in memory, so this should be a cheap operation.
-            // Do this *before* we clone 'latest_foreign_def_path_hashes', since
-            // loading existing queries may cause us to create new DepNodes, which
-            // may in turn end up invoking `store_foreign_def_id_hash`
-            tcx.dep_graph.exec_cache_promotions(QueryCtxt::from_tcx(tcx));
-
             let latest_foreign_def_path_hashes = self.latest_foreign_def_path_hashes.lock().clone();
             let hygiene_encode_context = HygieneEncodeContext::default();
 
@@ -566,7 +570,7 @@ impl<'sess> OnDiskCache<'sess> {
         })
     }
 
-    fn with_decoder<'a, 'tcx, T, F: FnOnce(&mut CacheDecoder<'sess, 'tcx>) -> T>(
+    fn with_decoder<'a, 'tcx, T, F: for<'s> FnOnce(&mut CacheDecoder<'s, 'tcx>) -> T>(
         &'sess self,
         tcx: TyCtxt<'tcx>,
         pos: AbsoluteBytePos,
@@ -577,12 +581,10 @@ impl<'sess> OnDiskCache<'sess> {
     {
         let cnum_map = self.cnum_map.get_or_init(|| Self::compute_cnum_map(tcx));
 
+        let serialized_data = self.serialized_data.read();
         let mut decoder = CacheDecoder {
             tcx,
-            opaque: opaque::Decoder::new(
-                self.serialized_data.as_deref().unwrap_or(&[]),
-                pos.to_usize(),
-            ),
+            opaque: opaque::Decoder::new(serialized_data.as_deref().unwrap_or(&[]), pos.to_usize()),
             source_map: self.source_map,
             cnum_map,
             file_index_to_file: &self.file_index_to_file,