Add a footer in FileEncoder and check for it in MemDecoder

3 files changed, 25 insertions, 4 deletions

diff --git a/compiler/rustc_incremental/messages.ftl b/compiler/rustc_incremental/messages.ftl
index e74173b24a9..de2177ebb6e 100644
--- a/compiler/rustc_incremental/messages.ftl
+++ b/compiler/rustc_incremental/messages.ftl
@@ -21,6 +21,8 @@ incremental_cargo_help_2 =
 incremental_copy_workproduct_to_cache =
     error copying object file `{$from}` to incremental directory as `{$to}`: {$err}
 
+incremental_corrupt_file = corrupt incremental compilation artifact found at `{$path}`. This file will automatically be ignored and deleted. If you see this message repeatedly or can provoke it without manually manipulating the compiler's artifacts, please file an issue. The incremental compilation system relies on hardlinks and filesystem locks behaving correctly, and may not deal well with OS crashes, so whatever information you can provide about your filesystem or other state may be very relevant.
+
 incremental_create_dep_graph = failed to create dependency graph at `{$path}`: {$err}
 
 incremental_create_incr_comp_dir =
diff --git a/compiler/rustc_incremental/src/errors.rs b/compiler/rustc_incremental/src/errors.rs
index 61bb0353a9f..e94a7fb876b 100644
--- a/compiler/rustc_incremental/src/errors.rs
+++ b/compiler/rustc_incremental/src/errors.rs
@@ -306,3 +306,9 @@ pub struct DeleteWorkProduct<'a> {
     pub path: &'a Path,
     pub err: std::io::Error,
 }
+
+#[derive(Diagnostic)]
+#[diag(incremental_corrupt_file)]
+pub struct CorruptFile<'a> {
+    pub path: &'a Path,
+}
diff --git a/compiler/rustc_incremental/src/persist/load.rs b/compiler/rustc_incremental/src/persist/load.rs
index 26aaa24771f..c96d22ce45d 100644
--- a/compiler/rustc_incremental/src/persist/load.rs
+++ b/compiler/rustc_incremental/src/persist/load.rs
@@ -115,7 +115,12 @@ fn load_dep_graph(sess: &Session) -> LoadResult<(Arc<SerializedDepGraph>, WorkPr
 
         if let LoadResult::Ok { data: (work_products_data, start_pos) } = load_result {
             // Decode the list of work_products
-            let mut work_product_decoder = MemDecoder::new(&work_products_data[..], start_pos);
+            let Some(mut work_product_decoder) =
+                MemDecoder::new(&work_products_data[..], start_pos)
+            else {
+                sess.dcx().emit_warn(errors::CorruptFile { path: &work_products_path });
+                return LoadResult::DataOutOfDate;
+            };
             let work_products: Vec<SerializedWorkProduct> =
                 Decodable::decode(&mut work_product_decoder);
 
@@ -145,7 +150,10 @@ fn load_dep_graph(sess: &Session) -> LoadResult<(Arc<SerializedDepGraph>, WorkPr
         LoadResult::DataOutOfDate => LoadResult::DataOutOfDate,
         LoadResult::LoadDepGraph(path, err) => LoadResult::LoadDepGraph(path, err),
         LoadResult::Ok { data: (bytes, start_pos) } => {
-            let mut decoder = MemDecoder::new(&bytes, start_pos);
+            let Some(mut decoder) = MemDecoder::new(&bytes, start_pos) else {
+                sess.dcx().emit_warn(errors::CorruptFile { path: &path });
+                return LoadResult::DataOutOfDate;
+            };
             let prev_commandline_args_hash = u64::decode(&mut decoder);
 
             if prev_commandline_args_hash != expected_hash {
@@ -181,9 +189,14 @@ pub fn load_query_result_cache(sess: &Session) -> Option<OnDiskCache<'_>> {
 
     let _prof_timer = sess.prof.generic_activity("incr_comp_load_query_result_cache");
 
-    match load_data(&query_cache_path(sess), sess) {
+    let path = query_cache_path(sess);
+    match load_data(&path, sess) {
         LoadResult::Ok { data: (bytes, start_pos) } => {
-            Some(OnDiskCache::new(sess, bytes, start_pos))
+            let cache = OnDiskCache::new(sess, bytes, start_pos).unwrap_or_else(|| {
+                sess.dcx().emit_warn(errors::CorruptFile { path: &path });
+                OnDiskCache::new_empty(sess.source_map())
+            });
+            Some(cache)
         }
         _ => Some(OnDiskCache::new_empty(sess.source_map())),
     }