React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard

--- description: React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard - Enforces safe C++ HostObject reference pinning and deterministic lifecycle management across concurrent native-to-Hermes execution boundaries in React Native 0.77 bridgeless mode. Prevents Hermes heap corruption, race conditions, and use-after-free crashes caused by capturing raw JSI values across asynchronous native threads without CallInvoker synchronization. globs: src/**/*.{ts,tsx}, tm/**/*.{ts,tsx}, cpp/**/*.{h,cpp}, ios/**/*.{h,mm}, android/src/main/jni/**/*.{h,cpp} alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/rn-077-bridgeless-jsi-memory-pinning-standard" --- # React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard > **Severity**: CRITICAL | **Category**: mobile > **Anti-Pattern Warning**: Capturing Unpinned JSI Value Pointers Across Asynchronous Native Threads and Retaining Leaked HostObjects ## System Constraints You are a Principal Mobile Systems Architect enforcing the React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard. In React Native 0.77 Bridgeless mode, the legacy bridge is eliminated, and all communication relies on direct C++ JSI bindings. You must mandate strict memory invariants: never capture raw facebook::jsi::Value, jsi::Object, or jsi::Pointer across concurrent thread boundaries; always pin host state via std::shared_ptr<jsi::HostObject>; always marshal native-to-Hermes calls through react::CallInvoker; and enforce explicit disposal semantics on the TypeScript consumer layer to prevent native resource leaks. ## Strict Guidelines - Engineers MUST ALWAYS schedule Hermes VM access through facebook::react::CallInvoker::invokeAsync when returning callbacks from background threads. - Engineers MUST NOT store raw jsi::Value, jsi::Object, or jsi::String instances in native member variables or capture them by reference across asynchronous dispatch queues. - Code MUST pin native state using std::shared_ptr<facebook::jsi::HostObject> and expose pinned resources to JavaScript via TurboModule HostObject references. - TypeScript interfaces MUST expose an explicit dispose() or unpin() method for native-backed HostObjects and register them with FinalizationRegistry as a defensive safeguard. - Native C++ TurboModules MUST verify runtime validity and guard against torn reads by executing inside jsi::Runtime locking scopes where multi-threading is enabled. ## Bad vs Good Implementations ### TypeScript Native HostObject Lifecycle and Pinning Management #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: The implementation relies on primitive identifiers without explicit lifecycle controls or native finalization tracking. If the JS instance is garbage-collected, the native thread retains asynchronous callback listeners and orphaned native resources, causing leaks in Hermes Bridgeless runtimes. ```typescript import { TurboModuleRegistry } from 'react-native'; import type { TurboModule } from 'react-native'; interface Spec extends TurboModule { createSession(id: string): { sessionId: string }; subscribeSession(sessionId: string, callback: (event: string) => void): void; } const NativeSessionModule = TurboModuleRegistry.getEnforcing<Spec>('SessionModule'); export class SessionHandle { private sessionId: string; constructor(id: string) { // ANTI-PATTERN: Native HostObject is referenced only by raw primitive string without // lifecycle cleanup or deterministic native pinning contract, risking heap desync. const res = NativeSessionModule.createSession(id); this.sessionId = res.sessionId; NativeSessionModule.subscribeSession(this.sessionId, (evt) => { console.log('Event received:', evt); }); } } ``` #### Production Standard (Correct) > **Standard**: Uses a pinned JSI HostObject reference, captures unsubscription tokens directly, implements deterministic dispose() calls, and provides a FinalizationRegistry fallback to guarantee native-side C++ object unpinning. ```typescript import { TurboModuleRegistry } from 'react-native'; import type { TurboModule } from 'react-native'; interface SessionNativeHostObject { readonly sessionId: string; dispose(): void; } interface Spec extends TurboModule { createSessionHostObject(id: string): SessionNativeHostObject; subscribeSession( hostObject: SessionNativeHostObject, callback: (event: string) => void ): () => void; } const NativeSessionModule = TurboModuleRegistry.getEnforcing<Spec>('SessionModule'); export class SessionHandle { private hostObject: SessionNativeHostObject | null; private unsubscribe: (() => void) | null = null; private static registry = new FinalizationRegistry<(obj: SessionNativeHostObject) => void>( (cleanup) => cleanup(this as unknown as SessionNativeHostObject) ); constructor(id: string) { this.hostObject = NativeSessionModule.createSessionHostObject(id); this.unsubscribe = NativeSessionModule.subscribeSession(this.hostObject, (evt) => { this.handleEvent(evt); }); SessionHandle.registry.register(this, () => this.dispose()); } private handleEvent(event: string): void { if (!this.hostObject) return; // Safe processing while pinned HostObject is alive } public dispose(): void { if (this.unsubscribe) { this.unsubscribe(); this.unsubscribe = null; } if (this.hostObject) { this.hostObject.dispose(); this.hostObject = null; } SessionHandle.registry.unregister(this); } } ``` ### C++ TurboModule CallInvoker Synchronization vs Raw Pointer Capture #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Raw jsi::Value and jsi::Function instances are thread-confined to the Hermes runtime thread. Calling them from a std::thread or native background queue without CallInvoker invokes undefined behavior and immediate memory corruption. ```cpp // ANTI-PATTERN: Capturing jsi::Function or jsi::Value across threads without CallInvoker void SessionModule::subscribeSession( facebook::jsi::Runtime& rt, const facebook::jsi::Object& hostObj, const facebook::jsi::Function& callback) { // DANGEROUS: Capturing raw reference to JSI function directly into native worker queue std::thread([&rt, &callback]() { std::string payload = fetchBackgroundPayload(); // CRASH: Accessing jsi::Runtime directly from non-Hermes thread causes use-after-free and SIGSEGV callback.call(rt, facebook::jsi::String::createFromUtf8(rt, payload)); }).detach(); } ``` #### Production Standard (Correct) > **Standard**: Pins C++ state via std::shared_ptr<SessionHostObject> and safely transfers execution back to the Hermes runtime using CallInvoker::invokeAsync, ensuring thread-safe access to JSI values. ```cpp // PRODUCTION STANDARD: Pinning Native State via HostObject and CallInvoker Marshaling void SessionModule::subscribeSession( facebook::jsi::Runtime& rt, std::shared_ptr<SessionHostObject> sessionHostObject, facebook::jsi::Function callback) { // Convert JSI function to Value and pin into a shared CallInvoker callback auto pinnedCallback = std::make_shared<facebook::jsi::Value>(rt, callback); auto jsInvoker = this->jsInvoker_; nativeQueue_->async([sessionHostObject, pinnedCallback, jsInvoker]() { if (!sessionHostObject->isValid()) { return; } std::string payload = sessionHostObject->fetchBackgroundPayload(); // Safe asynchronous boundary crossing via CallInvoker jsInvoker->invokeAsync([pinnedCallback, payload](facebook::jsi::Runtime& runtime) { if (pinnedCallback->isObject() && pinnedCallback->getObject(runtime).isFunction(runtime)) { auto fn = pinnedCallback->getObject(runtime).getFunction(runtime); fn.call(runtime, facebook::jsi::String::createFromUtf8(runtime, payload)); } }); }); } ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard

React Native 0.77 Bridgeless TurboModule JSI Memory Pinning Standard defines an industrial-grade behavioral standard for mobile engineering. Enforces safe C++ HostObject reference pinning and deterministic lifecycle management across concurrent native-to-Hermes execution boundaries in React Native 0.77 bridgeless mode. Prevents Hermes heap corruption, race conditions, and use-after-free crashes caused by capturing raw JSI values across asynchronous native threads without CallInvoker synchronization. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: Capturing Unpinned JSI Value Pointers Across Asynchronous Native Threads and Retaining Leaked HostObjects Without an explicit system instruction file, developers spend hours debugging silent failures.

Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	Capturing Unpinned JSI Value Pointers Across Asynchronous Native Threads and Retaining Leaked HostObjects	Critical	Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production. Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture Inconsistency	AI assistants produce mixed paradigms across files, creating fragmented codebase conventions.	Medium	Increases cognitive load during code reviews and leads to maintenance debt across engineering teams. Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.

Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

Engineers MUST ALWAYS schedule Hermes VM access through facebook

Engineers MUST ALWAYS schedule Hermes VM access through facebook::react::CallInvoker::invokeAsync when returning callbacks from background threads.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.