Repository navigation
Conversation
StreamReadHandle::stopSending unregistered the QUIC read callback but left a pending readStreamData() promise open. The peer's solicited RESET_STREAM then had no callback to arrive on, so the read did not complete until session teardown. Fail the pending read, as WtStreamManager's ReadHandle does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FakeStreamHandle::stopSending fired the write cancel token before it finished updating the handle. The fake wires both sessions together on one thread, so the peer reacts inside that call. A peer that closes the session then destroys the handle, and the rest of stopSending touched freed memory. Fire the token last, as resetStream already does. Add deliverPeerReset, which fails a read half without resetting the writer. resetStream also fires the writer's cancel token, so a session on the writing side reacts to its own reset and the reader cannot see a RESET_STREAM alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
|
@afrind has imported this pull request. If you are a Meta employee, you can view this in D129465241. |
afrind
added a commit
to afrind/moxygen
that referenced
this pull request
Nov 5, 2026
Cancelling a draft-18 request stream reset our write half, but STOP_SENDING waited until the read loop exited, which could take until session close. A peer that treats STOP_SENDING as the cancel kept the subscription and refused the next SUBSCRIBE as a duplicate. The read loop now sends STOP_SENDING as soon as the request is cancelled, whether by us or by a peer STOP_SENDING. The loop exits promptly once WebTransportImpl's stopSending fails the pending read (meta/proxygen#661). On session close, FETCH now cancels its data stream before its request stream, so the peer sees the data stream's own error code. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
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
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.
StreamReadHandle::stopSending unregistered the QUIC read callback but left a
pending readStreamData() open. The peer's RESET_STREAM needed that callback to
fail the read, so the read stayed pending until the session closed.
stopSending now fails the pending read with its error code, as
WtStreamManager's ReadHandle does.
FakeSharedWebTransport's stopSending fired the write cancel token before it
finished updating the handle. A peer that closes the session in response
destroyed the handle mid-call, so the token now fires last. The fake also adds
deliverPeerReset, which lets a test deliver a RESET_STREAM to the read half
without resetting the writer.
🤖 Generated with Claude Code