Merged
Conversation
…rate() and unload() (#1056) ## Description Add thread-safety to `VoiceActivityDetection` using the same mutex pattern already established by `VisionModel`. Without this, `BaseModel::unload()` can destroy `module_` on the JS thread while `generate()` is still calling `forward()` on a worker thread, causing `SIGILL` / `SIGSEGV` crashes. ### Introduces a breaking change? - [x] No ### Type of change - [x] Bug fix (change which fixes an issue) ### Tested on - [x] Android ### Testing instructions Use `useVAD({ model: FSMN_VAD })` with continuous audio input and trigger rapid mount/unmount cycles. Confirm no crashes or deadlocks. ### Related issues Fixes #1055 --------- Co-authored-by: Bartosz Hanc <bartosz.hanc02@gmail.com>
## Description Fix crashes when loading LLMs by using mmap and avoiding reporting model file size as external memory pressure. This PR changes how LLM models are loaded to prevent crashes with large models. Previously, reporting the full model file size via setExternalMemoryPressure() would cause Hermes to crash because it breaks the GC's heap accounting when external memory exceeds or approaches the 3GB max heap size. We also set the LoadMode to Mmap instead of File, causing the ET runtime to lazy-load weights to RAM on-demand instead of storing the entire file content in memory, preventing the OS from killing the app. ### Introduces a breaking change? - [ ] Yes - [x] No ### Type of change - [x] Bug fix (change which fixes an issue) - [ ] New feature (change which adds functionality) - [ ] Documentation update (improves or adds clarity to existing documentation) - [ ] Other (chores, tests, code style improvements etc.) ### Tested on - [x] iOS - [x] Android ### Testing instructions - [ ] Take a large model, verify it crashes the app on main and note the memory consumption - [ ] Try running the same model on this branch, make sure it doesn't crash and note the memory consumption - [ ] Verify models that would usually fit in your RAM are not slowed down significantly. ### Screenshots <!-- Add screenshots here, if applicable --> ### Related issues <!-- Link related issues here using #issue-number --> ### Checklist - [x] I have performed a self-review of my code - [x] I have commented my code, particularly in hard-to-understand areas - [ ] I have updated the documentation accordingly - [ ] My changes generate no new warnings ### Additional notes <!-- Include any additional information, assumptions, or context that reviewers might need to understand this PR. --> --------- Co-authored-by: Mateusz Sluszniak <56299341+msluszniak@users.noreply.github.com>
## Description
Adds thread-safety to Text Embeddings and Text-to-Image models mirroring
what was already done for models inheriting from VisionModel and VAD.
### Introduces a breaking change?
- [ ] Yes
- [x] No
### Type of change
- [x] Bug fix (change which fixes an issue)
- [ ] New feature (change which adds functionality)
- [ ] Documentation update (improves or adds clarity to existing
documentation)
- [ ] Other (chores, tests, code style improvements etc.)
### Tested on
- [x] iOS
- [x] Android
### Testing instructions
Use the following app screen and try to trigger the race condition
before fix and verify that it doesn't occur after applying the fix. You
can use `adb logcat | grep -E "FATAL|SIGSEGV|backtrace"` to observe the
error on Android.
```ts
import React, { useState } from 'react';
import { Button, ScrollView, Text, View } from 'react-native';
import {
BK_SDM_TINY_VPRED_512,
TextToImageModule,
} from 'react-native-executorch';
import {
CLIP_VIT_BASE_PATCH32_TEXT,
TextEmbeddingsModule,
} from 'react-native-executorch';
import { FSMN_VAD, VADModule } from 'react-native-executorch';
const DELAY_MS = 50; // tune this so that forward() is running when delete() is called
const MODEL_VAD = {
name: 'VAD',
load: (onProgress: (p: number) => void) =>
VADModule.fromModelName(FSMN_VAD, onProgress),
input: () => new Float32Array(16000 * 300),
};
const MODEL_TEXT_EMBEDDINGS = {
name: 'TextEmbeddings',
load: (onProgress: (p: number) => void) =>
TextEmbeddingsModule.fromModelName(CLIP_VIT_BASE_PATCH32_TEXT, onProgress),
input: () => 'hello world',
};
const MODEL_TEXT_TO_IMAGE = {
name: 'TextToImage',
load: (onProgress: (p: number) => void) =>
TextToImageModule.fromModelName(BK_SDM_TINY_VPRED_512, onProgress),
input: () => 'a red apple',
};
const MODEL = MODEL_TEXT_EMBEDDINGS;
export default function RaceTest() {
const [lines, setLines] = useState<string[]>([]);
const [downloadProgress, setDownloadProgress] = useState<number | null>(null);
const log = (line: string) => setLines((prev) => [line, ...prev]);
const run = async () => {
setLines([]);
setDownloadProgress(null);
log(`model: ${MODEL.name}`);
log('loading');
const model = await MODEL.load((p) => setDownloadProgress(p));
setDownloadProgress(null);
log('running forward()');
const result = model.forward(MODEL.input());
log(`waiting ${DELAY_MS} ms`);
await new Promise((r) => setTimeout(r, DELAY_MS));
log('calling delete()');
model.delete();
try {
await result;
log('forward() completed successfully');
} catch (e: any) {
log('error: ' + (e?.message ?? String(e)));
}
};
return (
<View>
<Button title="Run Race Test" onPress={run} />
{downloadProgress !== null && (
<Text>downloading: {Math.round(downloadProgress * 100)}%</Text>
)}
<ScrollView>
{lines.map((l, i) => (
<Text key={i}>{l}</Text>
))}
</ScrollView>
</View>
);
}
```
### Screenshots
<!-- Add screenshots here, if applicable -->
### Related issues
#1055
### Checklist
- [x] I have performed a self-review of my code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have updated the documentation accordingly
- [x] My changes generate no new warnings
### Additional notes
<!-- Include any additional information, assumptions, or context that
reviewers might need to understand this PR. -->
mkopcins
approved these changes
Apr 10, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Patch release v0.8.3 — cherry-picks the following bug fixes from
mainintorelease/0.8:generate()andunload()(fix: add mutex to VoiceActivityDetection to prevent race between generate() and unload() #1056)Checklist
mainin chronological order0.8.3inpackages/react-native-executorch/package.json🤖 Generated with Claude Code