[wasm] Deduplicate wasm scripts

R=clemensb@chromium.org

Bug: v8:6847
Change-Id: I4509a7011a0d32f5bbd2eabf3d2ee6ef5304263d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2124320
Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#66981}

src/wasm/module-compiler.cc

@@ -1604,10 +1604,9 @@ void AsyncCompileJob::PrepareRuntimeObjects() {
   DCHECK(module_object_.is_null());
   const WasmModule* module = native_module_->module();
   auto source_url = stream_ ? stream_->url() : Vector<const char>();
-  Handle<Script> script = CreateWasmScript(
-      isolate_, native_module_->wire_bytes(), VectorOf(module->source_map_url),
+  auto script = isolate_->wasm_engine()->GetOrCreateScript(
+      isolate_, native_module_.get(), VectorOf(module->source_map_url),
       module->name, source_url);
-
   Handle<WasmModuleObject> module_object =
       WasmModuleObject::New(isolate_, native_module_, script);
 

src/wasm/wasm-engine.cc

@@ -131,6 +131,31 @@ class WasmGCForegroundTask : public CancelableTask {
   Isolate* isolate_;
 };
 
+class WeakScriptHandle {
+ public:
+  explicit WeakScriptHandle(Handle<Script> handle) {
+    auto global_handle =
+        handle->GetIsolate()->global_handles()->Create(*handle);
+    location_ = std::make_unique<Address*>(global_handle.location());
+    GlobalHandles::MakeWeak(location_.get());
+  }
+
+  WeakScriptHandle(WeakScriptHandle&&) V8_NOEXCEPT = default;
+
+  ~WeakScriptHandle() {
+    if (location_) {
+      GlobalHandles::Destroy(*location_);
+    }
+  }
+
+  Handle<Script> handle() { return Handle<Script>(*location_); }
+
+ private:
+  // Store the location in a unique_ptr so that its address stays the same even
+  // when this object is moved/copied.
+  std::unique_ptr<Address*> location_;
+};
+
 }  // namespace
 
 std::shared_ptr<NativeModule> NativeModuleCache::MaybeGetNativeModule(
@@ -320,6 +345,9 @@ struct WasmEngine::IsolateInfo {
   // grows, never shrinks).
   std::set<NativeModule*> native_modules;
 
+  // Scripts created for each native module in this isolate.
+  std::unordered_map<NativeModule*, WeakScriptHandle> scripts;
+
   // Caches whether code needs to be logged on this isolate.
   bool log_codes;
 
@@ -462,9 +490,9 @@ MaybeHandle<WasmModuleObject> WasmEngine::SyncCompile(
 #endif
 
   Handle<Script> script =
-      CreateWasmScript(isolate, bytes.module_bytes(),
-                       VectorOf(native_module->module()->source_map_url),
-                       native_module->module()->name);
+      GetOrCreateScript(isolate, native_module.get(),
+                        VectorOf(native_module->module()->source_map_url),
+                        native_module->module()->name);
 
   // Create the compiled module object and populate with compiled functions
   // and information needed at instantiation time. This object needs to be
@@ -928,6 +956,7 @@ void WasmEngine::FreeNativeModule(NativeModule* native_module) {
     IsolateInfo* info = isolates_[isolate].get();
     DCHECK_EQ(1, info->native_modules.count(native_module));
     info->native_modules.erase(native_module);
+    info->scripts.erase(native_module);
     // If there are {WasmCode} objects of the deleted {NativeModule}
     // outstanding to be logged in this isolate, remove them. Decrementing the
     // ref count is not needed, since the {NativeModule} dies anyway.
@@ -1089,6 +1118,29 @@ void WasmEngine::FreeDeadCodeLocked(const DeadCodeMap& dead_code) {
   }
 }
 
+Handle<Script> WasmEngine::GetOrCreateScript(Isolate* isolate,
+                                             NativeModule* native_module,
+                                             Vector<const char> source_map_url,
+                                             WireBytesRef name,
+                                             Vector<const char> source_url) {
+  base::MutexGuard guard(&mutex_);
+  DCHECK_EQ(1, isolates_.count(isolate));
+  auto& scripts = isolates_[isolate]->scripts;
+  auto it = scripts.find(native_module);
+  if (it != scripts.end()) {
+    Handle<Script> weak_global_handle = it->second.handle();
+    if (weak_global_handle.is_null()) {
+      scripts.erase(it);
+    } else {
+      return Handle<Script>::New(*weak_global_handle, isolate);
+    }
+  }
+  auto script = CreateWasmScript(isolate, native_module->wire_bytes(),
+                                 source_map_url, name, source_url);
+  scripts.emplace(native_module, WeakScriptHandle(script));
+  return script;
+}
+
 void WasmEngine::TriggerGC(int8_t gc_sequence_index) {
   DCHECK_NULL(current_gc_info_);
   DCHECK(FLAG_wasm_code_gc);

src/wasm/wasm-engine.h

@@ -335,6 +335,11 @@ class V8_EXPORT_PRIVATE WasmEngine {
   void FreeDeadCode(const DeadCodeMap&);
   void FreeDeadCodeLocked(const DeadCodeMap&);
 
+  Handle<Script> GetOrCreateScript(Isolate*, NativeModule*,
+                                   Vector<const char> source_map_url,
+                                   WireBytesRef name,
+                                   Vector<const char> source_url = {});
+
   // Call on process start and exit.
   static void InitializeOncePerProcess();
   static void GlobalTearDown();

test/inspector/debugger/wasm-scripts-expected.txt

 Tests how wasm scripts are reported
 Check that each inspector gets a wasm script at module creation time.
-Session #1: Script #0 parsed. URL: wasm://wasm/7b04570e. Source map URL: , module begin: 0, module end: 0, code offset: 34
-Session #2: Script #0 parsed. URL: wasm://wasm/7b04570e. Source map URL: , module begin: 0, module end: 0, code offset: 34
-Session #1: Script #1 parsed. URL: wasm://wasm/ba7c35be. Source map URL: wasm://dwarf, module begin: 0, module end: 0, code offset: 34
-Session #2: Script #1 parsed. URL: wasm://wasm/ba7c35be. Source map URL: wasm://dwarf, module begin: 0, module end: 0, code offset: 34
-Session #1: Script #2 parsed. URL: wasm://wasm/1baa71fe. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
-Session #2: Script #2 parsed. URL: wasm://wasm/1baa71fe. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
-Session #1: Script #3 parsed. URL: wasm://wasm/95e97206. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
-Session #2: Script #3 parsed. URL: wasm://wasm/95e97206. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
-Session #1: Script #4 parsed. URL: wasm://wasm/7ab47392. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
-Session #2: Script #4 parsed. URL: wasm://wasm/7ab47392. Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #1: Script #0 parsed. URL: wasm://wasm/7b04570e. Script ID: 0, Source map URL: , module begin: 0, module end: 0, code offset: 34
+Session #2: Script #0 parsed. URL: wasm://wasm/7b04570e. Script ID: 0, Source map URL: , module begin: 0, module end: 0, code offset: 34
+Session #1: Script #1 parsed. URL: wasm://wasm/7b04570e. Script ID: 0, Source map URL: , module begin: 0, module end: 0, code offset: 34
+Session #2: Script #1 parsed. URL: wasm://wasm/7b04570e. Script ID: 0, Source map URL: , module begin: 0, module end: 0, code offset: 34
+Session #1: Script #2 parsed. URL: wasm://wasm/ba7c35be. Script ID: 1, Source map URL: wasm://dwarf, module begin: 0, module end: 0, code offset: 34
+Session #2: Script #2 parsed. URL: wasm://wasm/ba7c35be. Script ID: 1, Source map URL: wasm://dwarf, module begin: 0, module end: 0, code offset: 34
+Session #1: Script #3 parsed. URL: wasm://wasm/1baa71fe. Script ID: 2, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #2: Script #3 parsed. URL: wasm://wasm/1baa71fe. Script ID: 2, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #1: Script #4 parsed. URL: wasm://wasm/95e97206. Script ID: 3, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #2: Script #4 parsed. URL: wasm://wasm/95e97206. Script ID: 3, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #1: Script #5 parsed. URL: wasm://wasm/7ab47392. Script ID: 4, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #2: Script #5 parsed. URL: wasm://wasm/7ab47392. Script ID: 4, Source map URL: abc, module begin: 0, module end: 0, code offset: 34
+Session #1: Source for wasm://wasm/7b04570e:
+Raw: 00 61 73 6d 01 00 00 00 01 07 02 60 00 00 60 00 00 03 03 02 00 01 07 08 01 04 6d 61 69 6e 00 01 0a 0e 02 03 00 01 0b 08 00 02 40 41 02 1a 0b 0b 00 1b 04 6e 61 6d 65 01 14 02 00 0b 6e 6f 70 46 75 6e 63 74 69 6f 6e 01 04 6d 61 69 6e
+Imports: []
+Exports: [main: function]
+Session #2: Source for wasm://wasm/7b04570e:
+Raw: 00 61 73 6d 01 00 00 00 01 07 02 60 00 00 60 00 00 03 03 02 00 01 07 08 01 04 6d 61 69 6e 00 01 0a 0e 02 03 00 01 0b 08 00 02 40 41 02 1a 0b 0b 00 1b 04 6e 61 6d 65 01 14 02 00 0b 6e 6f 70 46 75 6e 63 74 69 6f 6e 01 04 6d 61 69 6e
+Imports: []
+Exports: [main: function]
 Session #1: Source for wasm://wasm/7b04570e:
 Raw: 00 61 73 6d 01 00 00 00 01 07 02 60 00 00 60 00 00 03 03 02 00 01 07 08 01 04 6d 61 69 6e 00 01 0a 0e 02 03 00 01 0b 08 00 02 40 41 02 1a 0b 0b 00 1b 04 6e 61 6d 65 01 14 02 00 0b 6e 6f 70 46 75 6e 63 74 69 6f 6e 01 04 6d 61 69 6e
 Imports: []

test/inspector/debugger/wasm-scripts.js

@@ -35,6 +35,15 @@ function testFunction(bytes) {
   new WebAssembly.Module(new Uint8Array(bytes));
 }
 
+// Generate stable IDs.
+let scriptIds = {};
+function nextStableId(id) {
+  if (!(id in scriptIds)) {
+    scriptIds[id] = Object.keys(scriptIds).length;
+  }
+  return scriptIds[id];
+}
+
 contextGroup.addScript(testFunction.toString(), 0, 0, 'v8://test/testFunction');
 
 InspectorTest.log(
@@ -55,6 +64,9 @@ sessions[0].Protocol.Runtime
       // no debug info
       testFunction([${createModule()}]);
 
+      // shared script for identical modules
+      testFunction([${createModule()}]);
+
       // DWARF
       testFunction([${createModule(dwarfSection)}]);
 
@@ -87,8 +99,9 @@ function trackScripts(debuggerParams) {
 
   async function loadScript(
       {url, scriptId, sourceMapURL, startColumn, endColumn, codeOffset}) {
+    let stableId = nextStableId(scriptId);
     InspectorTest.log(`Session #${sessionId}: Script #${
-        scripts.length} parsed. URL: ${url}. Source map URL: ${
+        scripts.length} parsed. URL: ${url}. Script ID: ${stableId}, Source map URL: ${
         sourceMapURL}, module begin: ${startColumn}, module end: ${endColumn}, code offset: ${codeOffset}`);
     let {result: {scriptSource, bytecode}} =
         await Protocol.Debugger.getScriptSource({scriptId});