fix(iocp)!: map remote RST to connection_reset, not cond::canceled (#…#319
Merged
mvandeberg merged 1 commit intoJul 17, 2026
Merged
Conversation
|
An automated preview of the documentation is available at https://319.corosio.prtest3.cppalliance.org/index.html If more commits are pushed to the pull request, the docs will rebuild at the same URL. 2026-07-17 17:12:05 UTC |
mvandeberg
force-pushed
the
pr/304-normalize-errors
branch
from
July 16, 2026 22:55
bfa1e65 to
fee9573
Compare
|
GCOVR code coverage report https://319.corosio.prtest3.cppalliance.org/gcovr/index.html Build time: 2026-07-17 17:21:45 UTC |
…ppalliance#304) Make cond::canceled mean only local, deliberate cancellation on every backend, and surface remote/connection failures as the portable std::errc condition consistently across epoll/io_uring/kqueue/select/IOCP. Previously the IOCP and kqueue backends diverged from POSIX: - IOCP mapped ERROR_NETNAME_DELETED unconditionally to canceled, so a remote RST looked identical to a local cancel. It is now disambiguated per operation kind in the new iocp_make_err (connection_reset for stream read/write, connection_aborted for accept); make_err no longer folds it into canceled. The cancelled flag, set before the raw error is consulted, keeps genuine local closes (close()/cancel()/stop-token) as canceled. - IOCP surfaced NTSTATUS-derived Win32 socket codes (e.g. ERROR_CONNECTION_REFUSED 1225) that MSVC's system_category does not map to std::errc. iocp_make_err now normalizes the observed ones to their Winsock equivalents (connection_refused, network_unreachable, timed_out) so they compare equal to the portable conditions like POSIX does. - IOCP close_socket() flagged only the aux-reactor wait op cancelled, relying on make_err's old NETNAME_DELETED->canceled fold for the overlapped read/write/connect/accept ops. With that fold gone, close_socket() now request_cancel()s every op before tearing down the handle (as cancel() does), so a local close stays canceled regardless of the CancelIoEx-vs- closesocket race. Applied across tcp, udp, and local-stream services. - kqueue stripped a user-set SO_LINGER before close (a macOS workaround), turning an abortive close into a graceful FIN so the peer saw eof instead of connection_reset. RST is in fact delivered reliably (EV_EOF carries ECONNRESET in fflags), so the strip is removed and SO_LINGER is passed through, matching the other backends. Add test/unit/error_conditions.cpp: a cross-backend audit (run under every backend) asserting the portable contract for the remote/connection error family -- FIN->eof, remote reset is never canceled, write-to-dead-peer is connection_reset|broken_pipe, refused connect is never canceled.
mvandeberg
force-pushed
the
pr/304-normalize-errors
branch
from
July 17, 2026 17:05
c54648d to
320dee4
Compare
mvandeberg
marked this pull request as ready for review
July 17, 2026 17:35
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #319 +/- ##
===========================================
+ Coverage 77.79% 77.80% +0.01%
===========================================
Files 96 96
Lines 6985 6984 -1
Branches 1682 1681 -1
===========================================
Hits 5434 5434
Misses 1055 1055
+ Partials 496 495 -1
Continue to review full report in Codecov by Harness.
🚀 New features to boost your workflow:
|
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
2 participants
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.
…304)
IOCP delivers ERROR_NETNAME_DELETED both when a local closesocket() cancels a pending overlapped op and when the peer hard-closes with a RST. detail::make_err mapped it unconditionally to capy::error::canceled, so a remote reset on Windows surfaced as cond::canceled -- conflating remote termination with deliberate local cancellation. POSIX already mapped RST to errc::connection_reset.
The cancelled atomic (set by the stop-token, cancel(), and close() paths) is consulted before the raw error at every decode site, so a surviving ERROR_NETNAME_DELETED is always a genuine remote reset. Remove it from make_err's canceled branch and remap it per operation kind at the IOCP decode sites via a new iocp_make_err helper: connection_reset for stream read/write, connection_aborted for accept (mirroring Asio). make_err stays generic; the winsock-dependent remap lives in the IOCP layer.
Add test/unit/error_conditions.cpp: a backend-parameterized audit of the remote/connection error family (FIN->eof, RST->connection_reset, write-to-dead-peer->connection_reset|broken_pipe, connect-refused-> connection_refused), asserting portable conditions on every backend so the cross-platform contract is enforced. The cancellation family is already covered across all backends elsewhere and is not duplicated.
normalize_openssl_shutdown_read_error is left unchanged: it already excludes canceled (#301), and reset-during-shutdown genuinely means no close_notify was received -> stream_truncated.