✅ regression test for SkiaAnimatedImageView close ordering#4444
Merged
guiyanakuang merged 4 commits intoMay 20, 2026
Merged
Conversation
Decoding frames on ioDispatcher (#4422) made codec.readPixels a synchronous JNI call that ignores coroutine cancellation, while DisposableEffect.onDispose closed bitmaps and produceState.awaitDispose closed the codec — both could free native pixel buffers under an active decode and crash the process. Fold codec + bitmap ownership into a single LaunchedEffect whose finally runs only after the last withContext returns, so close happens after readPixels regardless of when composition leaves.
- Nest try/finally so codec.close() still runs if Bitmap.allocPixels
throws (e.g. OOM) before the inner try is entered.
- Replace the 20ms !playing polling delay with snapshotFlow { playing }
.first { it } so a hidden preview suspends instead of waking ~50x/s.
Extract a testable seam (FrameDecoder / FrameBuffer + runAnimationLoop) so the bb9a381 UAF fix has explicit test coverage. A future refactor that re-splits codec/bitmap ownership across sibling effects would silently reintroduce the SIGSEGV; these tests catch that. The fake decoder simulates readPixels' uninterruptible JNI semantics via withContext(NonCancellable), then asserts buffer.close and decoder.close fire only after the in-flight decodeFrame returns. Closes #4443
…animated-image-close-ordering-regression # Conflicts: # app/src/desktopMain/kotlin/com/crosspaste/ui/paste/side/preview/SkiaAnimatedImageView.kt
guiyanakuang
deleted the
test/issue-4443-skia-animated-image-close-ordering-regression
branch
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
1 participant
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.
Closes #4443
Note
Stacked on #4438 (bb9a381 + 85a49bd). Diff is only the test seam + regression tests; the base will retarget to
mainonce #4438 lands.Summary
FrameDecoder/FrameBufferinterfaces +suspend fun runAnimationLoop(...)so the codec/bitmap lifecycle is testable without a real GIF on disk. Production wrapsCodec/BitmapasSkiaCodecDecoder/SkiaBitmapBuffer— semantics unchanged.buffer.closeanddecoder.closefire only after the in-flightdecodeFramereturns, anddecoder.closeis the final event (proving the nested-finally ordering).readPixels' uninterruptible JNI semantics viawithContext(NonCancellable)around aCompletableDeferred— so the test exercises the same cancellation race the bug came from.runAnimationLoopthat names commit bb9a381 and the patterns (siblingDisposableEffect.onDispose,produceState.awaitDispose) that would silently reintroduce the UAF.Test plan
./gradlew ktlintFormatclean./gradlew app:desktopTest --tests "com.crosspaste.ui.paste.side.preview.SkiaAnimatedImageViewTest"— 17/17 passing (14 existing + 3 new)decoder.close()out ofrunAnimationLoop's outerfinallyinto a sibling effect would causedecoder.closeto fire beforedecode.end(...)on cancellation, failing both close-ordering tests.