-
Notifications
You must be signed in to change notification settings - Fork 1.9k
[android] fix memory leak in TabbedPage
#21218
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Fixes: dotnet#17959 Context: https://github.com/VoznichkaSviatoslav1/TabbedPageMemoryLeak In the sample app, I added some `GC.Collect()` calls to force the GC to run, but the following log message never prints: public partial class MyTabbedPage : TabbedPage { public MyTabbedPage() { InitializeComponent(); Console.WriteLine("Constructor was called"); } // Doesn't print!!! ~MyTabbedPage() => Console.WriteLine("Destructor was called"); I took a memory snapshot with `dotnet-gcdump` and found a tree like: SampleApp.MyTabbedPage -> Microsoft.Maui.Controls.Platform.MultiPageFragmentStateAdapter<Microsoft.Maui.Controls.Page> -> Action<Microsoft.Maui.Controls.Platform.AdapterItemKey> I didn't find an explanation of what was holding `MultiPageFragmentStateAdapter`. I tried eliminating the `Action<T>`, but that didn't solve the underlying issue; it just made the object tree shorter: SampleApp.MyTabbedPage -> Microsoft.Maui.Controls.Platform.MultiPageFragmentStateAdapter<Microsoft.Maui.Controls.Page> I could also reproduce the problem in a new device test. Reviewing the code, I found we were doing: _viewPager.Adapter = new MultiPageFragmentStateAdapter<Page>(tabbedPage, FragmentManager, _context) { CountOverride = tabbedPage.Children.Count }; But then in the cleanup code, we never set `Adapter` to `null`: _viewPager.Adapter = null; After this change, the memory is released as expected. I am not exactly sure *why* this leaked, but the change does seem appropriate and fixes the problem. Now the new device test passes & the sample app prints: 03-14 15:06:32.394 6055 6055 I DOTNET : Constructor was called 03-14 15:06:37.008 6055 6089 I DOTNET : Destructor was called
I spent 2.5 hours on this, and did not figure it out! Comment for now as original issue is Android.
|
I disabled the test on Windows for now, since the original issue is Android. I spent 2 hours commenting out code, and can't seem to figure it out. I don't understand what the |
|
WindowsAppSDK looks like it makes extensive use of Probably worth me learning about it, might be useful for other things. |



Fixes: #17959
Context: https://github.com/VoznichkaSviatoslav1/TabbedPageMemoryLeak
In the sample app, I added some
GC.Collect()calls to force the GC to run, but the following log message never prints:I took a memory snapshot with
dotnet-gcdumpand found a tree like:I didn't find an explanation of what was holding
MultiPageFragmentStateAdapter. I tried eliminating theAction<T>, but that didn't solve the underlying issue; it just made the object tree shorter:I could also reproduce the problem in a new device test.
Reviewing the code, I found we were doing:
But then in the cleanup code, we never set
Adaptertonull:After this change, the memory is released as expected. I am not exactly sure why this leaked, but the change does seem appropriate and fixes the problem.
Now the new device test passes & the sample app prints: