On Fri, 8 Nov 2024 17:42:22 GMT, Daniel Fuchs <dfu...@openjdk.org> wrote:
> [JDK-8342075](https://bugs.openjdk.org/browse/JDK-8342075) has introduced > more flow controls checks, but also introduced a race condition where > DataFrames for closed streams may fail to be discounted from the connection > window. > > The consequence is that WINDOW_UPDATE frames for the connection window may > not be sent when they should, preventing the server from making progress and > stalling the connection. > > This can be shown by modifying the StreamFlowControlTest to send less but > bigger frames (e.g. chunks of 1600 bytes instead of chunks of 12 bytes). With > such a modification the test can be seen failing intermittently, when > sameClient=true. > > The race happens when frames that have been added to Stream::inputQ fail to > be drained after the stream is closed (or continue to be added to the inputQ > after the stream is closed). > > The fix ensures that Stream::drainInputQueue() is called when the stream is > closed, and that no further data farme will be added to the inputQ after the > stream is marked closed. > > The modified StreamFlowControlTest could be observed failing relatively > frequently on linux-aarch64 without the fix. > With the fix the test no longer fails. This pull request has now been integrated. Changeset: bd6152f5 Author: Daniel Fuchs <dfu...@openjdk.org> URL: https://git.openjdk.org/jdk/commit/bd6152f5967107d7b32db9bcfa224fc07314f098 Stats: 186 lines in 5 files changed: 146 ins; 6 del; 34 mod 8343855: HTTP/2 ConnectionWindowUpdateSender may miss some unprocessed DataFrames from closed streams Reviewed-by: jpai ------------- PR: https://git.openjdk.org/jdk/pull/21991
