CLI: Fix hang with exec/run -i with self-exiting command and open stdin

[David Bennett (dkbennett)]

Summary of the Pull Request

wslc exec -i <container> echo hello  didn't return until the client's stdin was closed, while  docker exec -i  returns as soon as the process exits. The hang is entirely client‑side.

When wslc's stdin is a pipe (not a console), the stdin relay runs on a worker thread blocked in a synchronous  ReadFile . That read can't be interrupted by the relay's exit event. So when the remote process exits and teardown tries to  join()  the worker, the join blocks until stdin is closed externally — the hang.

The fix

A shared teardown helper,  InterruptAndJoinInputThread  (used by both the TTY and non‑TTY relay paths), now signals the exit event and then calls  CancelSynchronousIo  on the worker (retried until it exits) before joining — which aborts the otherwise‑uninterruptible read. A matching one‑line change in  relay.cpp  treats the resulting  ERROR_OPERATION_ABORTED  as EOF so the relay loop ends cleanly. Termination is guaranteed: every state the worker can block in is broken by either the cancel or the already‑set exit event.

Two regression tests are also added which exercise this hang and I confirmed these tests fail without the fix, and pass with the fix.

PR Checklist

• Closes: Link to issue #xxx
• Communication: I've discussed this with core contributors already. If work hasn't been agreed, this work might be rejected
• Tests: Added/updated if needed and all pass
• Localization: All end user facing strings can be localized
• Dev docs: Added/updated if needed
• Documentation updated: If checked, please file a pull request on our docs repo and link it here: #xxx

Detailed Description of the Pull Request / Additional comments

Validation Steps Performed

Tests

Two regression tests run a self‑exiting command ( echo hello ) and assert the process exits before stdin is closed — one for  exec -i , one for  exec -it , since the two paths have separate teardown logic that could regress independently. Both detect the symptom (process must exit within the timeout), so any teardown hang fails the test.

These tests fail without the fix and pass with the fix.

[Copilot]

Pull request overview

This PR fixes a client-side hang in wslc container exec -i/-it when stdin is redirected to a synchronous (non-overlapped) pipe and the remote process exits without consuming stdin. The fix ensures the stdin relay worker can always be interrupted during teardown, matching the expected docker exec -i behavior.

Changes:

• Add a shared teardown helper in ConsoleService that signals the exit event and repeatedly calls CancelSynchronousIo before joining the stdin relay thread.
• Treat ERROR_OPERATION_ABORTED from ReadFile as EOF in the relay read path so cancellation cleanly terminates the relay loop.
• Add two E2E regression tests covering exec -i and exec -it with a self-exiting command while keeping stdin open.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
test/windows/wslc/e2e/WSLCE2EContainerExecTests.cpp	Adds regression tests asserting exec -i/-it echo hello exits without requiring client stdin closure.
src/windows/wslc/services/ConsoleService.cpp	Introduces teardown logic to interrupt synchronous stdin reads (via CancelSynchronousIo) before joining relay worker threads.
src/windows/common/relay.cpp	Maps ERROR_OPERATION_ABORTED from ReadFile to EOF (0 bytes) to allow canceled reads to end relay loops cleanly.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.