Repository navigation
Fix false QUIC flow-control errors during bulk downloads - #153
Merged
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Browser QUIC fingerprints advertise receive credit that differs from quic-go defaults. Chrome permits 6 MiB per stream while the runtime initially allowed 512 KiB, so a compliant server could trigger a local FLOW_CONTROL_ERROR and close every stream in the session, including iperf3's control connection.
Initialize flow control from the actual uTLS transport parameters before connection setup, preserving distinct local/remote bidirectional and unidirectional limits and connection credit. Ordinary connections retain configured windows. Browser wire parameters are unchanged and excess data is still rejected.
Validation: deterministic boundary regression failed for all four advertised-credit cases before the fix and passes afterward; a 32 MiB download verifies payload integrity and a surviving control stream for Chrome, Firefox and randomized profiles.
bundle exec fastlane android native_testspasses all 370 tests with the race detector. On the isolated Linux VM, the original HTTP/3 reverse failure reproduces with explicit FLOW_CONTROL_ERROR; the instrumented fixed helper passes three reverse samples using the same iperf3 3.16, with three sing-box controls also passing. The same fixed helper passes all six MASQUE reverse cases (three samples, one/eight streams), and all six corresponding sing-box controls pass. These VM controls use the original iperf3 3.16 to isolate the QUIC fix. Diagnostic figures are not new performance claims. Android/device execution remains CI coverage, not a completed local check.Discovered while investigating comparison PR #152. The separate sing-box nested-QUIC DATAGRAM-size failure and old iperf3 segfault are not fixed by this patch.
APK artifacts
Built from commit
f7e6523cby CI run #492.Artifacts expire after 14 days. Neither APK uses the MegaProxy release key.