Merge pull request #3068 from tree-sitter/load-old-language-via-wasm-…

…crash Fix crash when attempting to load ancient languages via wasm
tree-sitter · Feb 23, 2024 · 9301c1e · 9301c1e
2 parents 1f751bb + 8dded3a
commit 9301c1e
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 3 deletions.
diff --git a/lib/binding_rust/wasm_language.rs b/lib/binding_rust/wasm_language.rs
@@ -134,7 +134,7 @@ impl fmt::Display for WasmError {
             WasmErrorKind::Instantiate => "Failed to instantiate wasm module",
             WasmErrorKind::Other => "Unknown error",
         };
-        write!(f, "{kind} {}", self.message)
+        write!(f, "{kind}: {}", self.message)
     }
 }
 

diff --git a/lib/src/language.h b/lib/src/language.h
@@ -10,6 +10,9 @@ extern "C" {
 
 #define ts_builtin_sym_error_repeat (ts_builtin_sym_error - 1)
 
+#define LANGUAGE_VERSION_WITH_PRIMARY_STATES 14
+#define LANGUAGE_VERSION_USABLE_VIA_WASM 13
+
 typedef struct {
   const TSParseAction *actions;
   uint32_t action_count;
@@ -186,7 +189,7 @@ static inline bool ts_language_state_is_primary(
   const TSLanguage *self,
   TSStateId state
 ) {
-  if (self->version >= 14) {
+  if (self->version >= LANGUAGE_VERSION_WITH_PRIMARY_STATES) {
     return state == self->primary_state_ids[state];
   } else {
     return true;

diff --git a/lib/src/wasm.c b/lib/src/wasm.c
@@ -10,6 +10,7 @@
 #include "./alloc.h"
 #include "./array.h"
 #include "./atomic.h"
+#include "./language.h"
 #include "./lexer.h"
 #include "./wasm.h"
 #include "./wasm/wasm-stdlib.h"
@@ -1057,6 +1058,12 @@ const TSLanguage *ts_wasm_store_load_language(
   const uint8_t *memory = wasmtime_memory_data(context, &self->memory);
   memcpy(&wasm_language, &memory[language_address], sizeof(LanguageInWasmMemory));
 
+  if (wasm_language.version < LANGUAGE_VERSION_USABLE_VIA_WASM) {
+      wasm_error->kind = TSWasmErrorKindInstantiate;
+      format(&wasm_error->message, "language version %u is too old for wasm", wasm_language.version);
+      goto error;
+  }
+
   int32_t addresses[] = {
     wasm_language.alias_map,
     wasm_language.alias_sequences,
@@ -1188,7 +1195,7 @@ const TSLanguage *ts_wasm_store_load_language(
     );
   }
 
-  if (language->version >= 14) {
+  if (language->version >= LANGUAGE_VERSION_WITH_PRIMARY_STATES) {
     language->primary_state_ids = copy(
       &memory[wasm_language.primary_state_ids],
       wasm_language.state_count * sizeof(TSStateId)