[Soham Das (SohamDas2021)]

📖 Description

Fixes the two correctness items left open when #1564 was approved.

CLI: stdin is closed only after the command is known to be stopped. If reading wxc-exec's stdin fails while a WSLc command runs, the executor stops the container and now waits for the command to exit before closing the command's stdin. Previously the result of the stop was ignored, so a command that was still running could read the cut-short input as complete, for example cat > staged && mv staged live publishing a truncated file.

• If the stop can't be confirmed, the command's stdin stays open until the container is torn down. The run fails with a message saying the command could not be confirmed stopped.
• A run with no timeout no longer hangs in that case.
• When a run times out, stdin is no longer closed before the command is known to be gone, so a command that ignores SIGTERM can't read early end-of-file either.

.NET: waits stay responsive while a finished process is cleaned up. On WSLc, a command that exits leaving a lot of stdin unread is reported about a minute late, and MxcProcess spends that time in cleanup. During it:

• cancelling WaitAsync now returns at once, and a later wait returns the result;
• Kill returns at once without doing anything;
• Dispose returns at once, and the process is released when cleanup finishes, before the wait returns;
• Warnings returns the warnings from before cleanup started, and OutputMetadata is null until cleanup finishes.

Before, all of these waited for the cleanup; #1564 fixed only Kill. The native API is unchanged.

🔗 References

Follow-up to #1564 (review findings 2 and 4). Finding 6 only concerned the #1564 description, which is corrected there.

🔍 Validation

Windows with WSL 3.0.4 and WSLc SDK 3.0.3 (vendored; the MxcDependencies feed returned 401).

Command	Result
cargo fmt --all -- --check	Passed
cargo clippy --workspace --all-targets -- -D warnings	Passed
cargo clippy -p mxc-sdk -p mxc_ffi --features wslc --all-targets -- -D warnings	Passed
cargo test -p mxc-sdk --features wslc --lib	Passed: 2835 passed, 0 failed
.NET suite (Microsoft.Mxc.Sdk.Tests, built with -p:MxcWithWslc=true)	Passed: 373 passed, 0 failed, 49 skipped (live-host tests)
Same, with MXC_ENABLE_WSLC_TESTS=1, class MxcContainerWslcE2ETests	Passed: 11/11
tests\scripts\run_wslc_all_tests.ps1 -Debug -SkipSetup	Passed: create-and-run 49/49 (9 stdin cases), container lifecycle 139/139
Node wslc-e2e.test.js with MXC_ENABLE_WSLC_TESTS=1	Passed: 8/8

New tests:

• Live .NET: cancelling WaitAsync, and calling Warnings and Dispose, while a wait is in the late-exit cleanup. Both tests fail without the MxcProcess change and pass with it. They are in the class the scheduled WSLc validation already runs.
• Rust unit: a forwarding failure whose stop can't be confirmed keeps the command's stdin open until teardown, and the run's wait ends on either the process exiting or the forwarding failure.

Not covered live: a failed or unconfirmed container stop, which can't be triggered on a real host; the unit tests cover it. The Node SDK unit tests were not run, since nothing in the Node SDK changed.

✅ Checklist

• Signed the Contributor License Agreement
• Linked to an issue
• Updated documentation (if applicable)
• Updated Copilot instructions (if build, architecture, or conventions changed)
• If this PR changes Cargo.lock, the dependency-feed-check check passes (see pull request builds)

📋 Issue Type

• Bug fix
• Feature
• Task

Microsoft Reviewers: Open in CodeFlow

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

🟡 Changes recommended

Unresolved stdin error paths, a V1 disposal compatibility break, timeout misreporting, and regression-test gaps prevent approval.

6 open issues

• Preserve synchronous disposal behavior for V1 · New
• Confirm session termination before releasing stdin · New
• Retain stdin until shutdown resources are prepared · New
• Add an independent deadline to exit polling · New
• Await discarded tasks to observe cleanup failures · New
• Recheck terminal status before reporting timeout · New

_{Review effort: Balanced}

[Copilot]

🟡 Changes recommended

Forwarding-setup failures can still deliver premature EOF, and timeout ordering lacks a functional regression that detects it.

3 open issues

• Quiesce closes stdin before confirming command termination · New
• Confirm session termination before releasing stdin
• Preserve synchronous disposal behavior for V1

1 previously missed issue

In code that hasn't changed since last review

Add regression for premature stdin EOF after timeout

src/​mxc-sdk/​src/​backends/​wslc/​common/​wsl_container_runner.rs:714

[P2] The timeout stdin-close ordering fix lacks a regression through the CLI/backend flow; the existing cat timeout case in tests/scripts/run_wslc_all_tests.ps1:628-637 checks only the timeout result and would still pass with premature EOF. Extend that scheduled suite with a command that ignores SIGTERM and writes a persistent marker only after EOF, keeping host stdin open through the timeout. Assert that the run times out without writing the marker, so restoring the unconditional close_stdin() fails the test.

4 resolved since last review

• Retain stdin until shutdown resources are prepared
• Recheck terminal status before reporting timeout
• Await discarded tasks to observe cleanup failures
• Add an independent deadline to exit polling

_{Review effort: Balanced}