Memory issue when sharing CancelToken between posts for IO implementation

[kunstmusik]

Package

dio

Version

5.4.0

Operating-System

Android

Output of flutter doctor -v

[✓] Flutter (Channel stable, 3.13.6, on macOS 14.2.1 23C71 darwin-arm64, locale en-US)
    • Flutter version 3.13.6 on channel stable at /Users/stevenyi/fvm/versions/3.13.6
    • Upstream repository https://github.com/flutter/flutter.git
    • Framework revision ead455963c (5 months ago), 2023-09-26 18:28:17 -0700
    • Engine revision a794cf2681
    • Dart version 3.1.3
    • DevTools version 2.25.0

[✓] Android toolchain - develop for Android devices (Android SDK version 33.0.0)
    • Android SDK at /Users/stevenyi/Library/Android/sdk
    • Platform android-33, build-tools 33.0.0
    • Java binary at: /Applications/Android Studio.app/Contents/jbr/Contents/Home/bin/java
    • Java version OpenJDK Runtime Environment (build 17.0.7+0-17.0.7b1000.6-10550314)
    • All Android licenses accepted.

[!] Xcode - develop for iOS and macOS (Xcode 15.2)
    • Xcode at /Applications/Xcode.app/Contents/Developer
    • Build 15C500b
    ✗ Unable to get list of installed Simulator runtimes.
    • CocoaPods version 1.15.2

[✓] Chrome - develop for the web
    • Chrome at /Applications/Google Chrome.app/Contents/MacOS/Google Chrome

[✓] Android Studio (version 2023.1)
    • Android Studio at /Applications/Android Studio.app/Contents
    • Flutter plugin can be installed from:
      🔨 https://plugins.jetbrains.com/plugin/9212-flutter
    • Dart plugin can be installed from:
      🔨 https://plugins.jetbrains.com/plugin/6351-dart
    • Java version OpenJDK Runtime Environment (build 17.0.7+0-17.0.7b1000.6-10550314)

[✓] IntelliJ IDEA Ultimate Edition (version 2023.2)
    • IntelliJ at /Applications/IntelliJ IDEA.app
    • Flutter plugin can be installed from:
      🔨 https://plugins.jetbrains.com/plugin/9212-flutter
    • Dart plugin can be installed from:
      🔨 https://plugins.jetbrains.com/plugin/6351-dart

[✓] VS Code (version 1.86.0)
    • VS Code at /Applications/Visual Studio Code.app/Contents
    • Flutter extension version 3.82.0

[✓] Connected device (2 available)
    • macOS (desktop) • macos  • darwin-arm64   • macOS 14.2.1 23C71 darwin-arm64
    • Chrome (web)    • chrome • web-javascript • Google Chrome 121.0.6167.184

[✓] Network resources
    • All expected network resources are available.

! Doctor found issues in 1 category.

Dart Version

3.1.3

Steps to Reproduce

1. Create a CancelToken
2. Make multiiple requests passing in the same CancelToken

Expected Result

No out of memory issues on Android.

Actual Result

Out of memory issues occur.

Dio's use of cancelFuture in dio/lib/src/adapters/io_adapter.dart differs from how it is used in browser_adapter.dart and exhibits different behavior. In IO, it captures on request in the callback to whenComplete in _fetch, which does not happen in the browser implementation.

This took us some time to discover as a memory leak issue. It would be great if this could be fixed, otherwise would love to see this documented so that future users will be aware of this issue.

[AlexV525]

Could you attach the minimal reproducible example and some analysis graphs or details?

[AlexV525]

@kuhnroyal I have been thinking about the cause of this issue for the last few days, it could be caused by referencing the resources of the request in callers like .whenComplete so they will only get released once the CancelToken canceled, but it doesn't, so memory leaks. Any ideas?

[AlexV525]

If yes, we might add a global map to cache the necessary instances for the corresponding cancel token.

[kuhnroyal]

I tried to enable the new leak tracking but didn't get very far yet...

[kunstmusik]

@AlexV525 I haven't had time to followup, but yes, as mentioned in my first message, we found that the .whenComplete callback captures over the Request:

      if (cancelFuture != null) {
        cancelFuture.whenComplete(() => request.abort());
      }

and whenComplete() also adds additional handlers each time it is called. So if a CancelToken is passed in to multiple fetches, it will keep accruing whenComplete handlers that all capture over individual requests.