onEnded listeners are never removed, so every AudioBufferSourceNode stays alive (and its buffer counts as Hermes external memory) until the app runs out of memory
Los mantenedores suelen responder en 1 día
Nadie ha tomado este issue todavía.
Evaluación
- Dificultad
- 4/5
- Tiempo estimado
- 3-5 días
- Aptitud para principiantes
- 57/100
- Tipo de issue
- Error
- Claridad
- Bien especificado
- Estado de actividad
- Activo
- Stack tecnológico
- android, cpp, ios, javascript, react-native
Línea de trabajo
Start by tracing AudioScheduledSourceNode's onEnded setter and the native AudioEventEmitter subscription it creates; the issue identifies both as the leak path. Check how subscriptions are stored and removed when the handler changes, is set to null, or ended fires. Done means replaced and cleared handlers release their subscriptions so finished nodes can be collected; the reproduction's Hermes external memory should stop growing.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
Description
Setting node.onEnded = callback on an AudioBufferSourceNode (any AudioScheduledSourceNode) registers a listener with the native AudioEventEmitter (addAudioEventListener('ended', …)), but the returned subscription is never removed:
- not when the
endedevent fires, - not when
onEndedis set tonull(the setter only sets the node's native listener id to'0'; the subscription stays in the emitter).
The emitter keeps the jsi::Function, the closure keeps the JS node, and the JS node keeps the native node. Each AudioBufferSourceNodeHostObject also calls setExternalMemoryPressure with its buffer's full size, so Hermes' external memory grows with every node that ever had onEnded set, and it never comes back down.
Impact: an app that schedules overlapping buffer segments (a seamless ambient loop: a new AudioBufferSourceNode every 16 s, each with onEnded to disconnect it) climbs steadily: Hermes js_externalBytes reached 2.9 GB after 24 minutes (Pixel 10 Pro XL) and 4.0 GB within minutes on a Pixel 8a, after which the OS killed the app with no JS error. iOS crashed the same way (Hermes out of memory at 2.2 GB).
Expected: the subscription is removed when onEnded is replaced or set to null (and ideally after ended fires for a node that can't start again), so finished nodes can be garbage-collected.
Suggested fix: keep the AudioEventSubscription returned by addAudioEventListener on the node and call remove() on it in the onEnded setter before registering a new one or when the value is null; consider removing it automatically after the ended event of a buffer source.
Workaround we use: no onEnded on segment nodes; a 1-second timer disconnects each one a second after its scheduled end.
Steps to reproduce
- In a development build (Hermes), run:
const ctx = new AudioContext();
const buf = ctx.createBuffer(2, ctx.sampleRate * 16, ctx.sampleRate);
setInterval(() => {
const src = ctx.createBufferSource();
src.buffer = buf;
src.connect(ctx.destination);
src.onEnded = () => { src.disconnect(); src.onEnded = null; };
src.start(); src.stop(ctx.currentTime + 1);
}, 1000);
- Log
HermesInternal.getInstrumentedStats().js_externalBytesevery few seconds: it grows ~12 MB per second and is never collected. - Remove the
src.onEnded = …line and run again:js_externalBytesstays flat (the nodes are collected).
Snack or a link to a repository
https://github.com/jessenicholsfromsanfrancisco/rnaa-onended-leak-repro
React Native Audio API version
0.13.6
React Native version
0.88.1
Platforms
Android, iOS
JavaScript runtime
Hermes
Workflow
Expo Dev Client
Architecture
Fabric (New Architecture)
Build type
Release app & dev bundle
Device
Real device
Device model
iPhone 15
Acknowledgements
Yes
- Lenguaje dominante
- C++
- Estrellas
- 856
- Forks
- 100
- Merge medio
- 2 d 1 h
- PR fusionados (30 d)
- 34
Preparar el entorno
- Sin Dockerfile ni archivo de Docker Compose
- Tiene una plantilla de pull request
- Leer la guía de contribución
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de software-mansion/react-native-audio-api
-
Android: audio focus listener posts to a dead looper after JS reload, interruptions (calls) are missedPosiblemente ocupada @JaydeepAgravat la tomó hace 1 día. Abiertoandroid bug
Dificultad 2/5 1-3 horas Aptitud para principiantes 76/100
software-mansion/react-native-audio-api#1358 · 2 comentarios ·
Los mantenedores suelen responder en 1 día
-
android bug
Dificultad 2/5 1-3 horas Aptitud para principiantes 82/100
software-mansion/react-native-audio-api#1273 · 5 comentarios ·
Los mantenedores suelen responder en 1 día
-
bug ios
Dificultad 2/5 1-3 horas Aptitud para principiantes 78/100
software-mansion/react-native-audio-api#1238 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
Dificultad 4/5 3-5 días Aptitud para principiantes 48/100
software-mansion/react-native-audio-api#1416 ·
Los mantenedores suelen responder en 1 día
-
feature-request ios
Dificultad 5/5 Más de una semana Aptitud para principiantes 45/100
software-mansion/react-native-audio-api#1398 ·
Los mantenedores suelen responder en 1 día
Todos los issues de software-mansion/react-native-audio-api
Issues similares
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 74/100
arvidn/libtorrent#9126 ·
Los mantenedores suelen responder en 1 día
-
bug
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
Sunrisepeak/mcpp-language-server#80 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 75/100
-
#2754 make deepseek api not workAbierto
Dificultad 2/5 1-3 horas Aptitud para principiantes 66/100
xiaoyifang/goldendict-ng#3148 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 1/5 Menos de una hora Aptitud para principiantes 91/100
triton-inference-server/server#9069 ·
Los mantenedores suelen responder en 1 día