Windows CI hang: in-box Windows PowerShell attach E2E test wedges after the 20260614 runner image

[andyleejordan]

Summary

The Windows leg of CI (CI Tests workflow) hangs on the E2E debugger-attach
test CanAttachScriptWithPathMappings and rides GitHub Actions' default
6-hour job timeout. The root cause is not a PSES code change — it is a
windows-latest runner-image refresh that broke in-box Windows PowerShell
5.1's cross-process attach. This issue tracks the real fix; PR #2318 is a
stopgap (skip the test on in-box WinPS + a 30-minute timeout-minutes
backstop).

Root cause: runner-image refresh 20260608 → 20260614

Comparing the last-green and first-red main runs:

• Last green (Correct example PowerShell -Uri argument name #2304) ran on image win25-vs2026/20260608.135.
• First red (Match strong-name identity when resolving PSES dependencies #2303) and every red after it ran on image 20260614.141.

Same image family, same runner agent (2.335.1), same -Preview PowerShell.
The only thing that moved at the boundary is the weekly OS-servicing patch in
the image. That refresh broke in-box Windows PowerShell 5.1's cross-process
Debug-Runspace / Enter-PSHostProcess attach — exactly the path this test
exercises. The precise servicing delta (likely a .NET Framework or Windows
named-pipe/IPC update) is still unidentified and is the main open question
here.

Symptoms

• The hang is specifically the in-box Windows PowerShell E2E suite
  (TestE2EPowerShell). PowerShell Core (TestE2EPwsh) and the preview pass
  the same attach test; macOS and Linux pass in ~5 minutes.
• The test wedges on the attach handshake and the run sits on that single
  test until the job is killed.
• Its xUnit [SkippableFact(Timeout = 15000)] never fires: the wedge blocks
  in a way xUnit's cooperative timeout cannot abort, so it is reported as a
  "Long Running Test" climbing past 20+ minutes.

What we ruled out

• Not Match strong-name identity when resolving PSES dependencies #2303 (strong-name identity change). The per-commit bisection
  fingered Match strong-name identity when resolving PSES dependencies #2303, but that was confounded: Match strong-name identity when resolving PSES dependencies #2303 was the first merge onto the
  new image, so the image bump and the code change moved together. Two
  independent proofs exonerate it:
  • PsesLoadContext.cs (the file Match strong-name identity when resolving PSES dependencies #2303 touched) is <Compile Remove>'d for
    net462 and uses the Core-only AssemblyLoadContext. In-box WinPS 5.1
    loads bin/Desktop/ (net462), so Match strong-name identity when resolving PSES dependencies #2303's code is not even compiled into
    the hanging configuration.
  • With Match strong-name identity when resolving PSES dependencies #2303 reverted on the Skip attach E2E test on in-box Windows PowerShell (20260614 image regression); cap CI job #2318 branch, the in-box WinPS attach suite
    still hung on the new image (killed by the 30-minute cap).
• Not the -Preview PowerShell. The preview pwsh suite passes; it is the
  in-box Windows PowerShell suite that hangs. (An earlier draft of this issue
  hypothesized the preview — that is disproven.)
• Not ReadScriptLogLineAsync. The test wedges on the attach handshake
  before it reaches that helper. (PR Skip attach E2E test on in-box Windows PowerShell (20260614 image regression); cap CI job #2318 still hardens that EOF/poll spin as
  general hygiene, but it is not the cause.)

Mitigation (PR #2318)

• Skip CanAttachScriptWithPathMappings on in-box Windows PowerShell
  (IsWindowsPowerShell, covering the WinPS and WinPS-CLM suites) so the
  windows-latest leg completes; Core / preview / macOS / Linux keep full
  coverage of the attach path.
• timeout-minutes: 30 on the matrix test job as a permanent backstop.

Next steps toward a real fix

• Identify the precise 20260614 servicing delta (.NET Framework / Windows
  IPC) that changed in-box WinPS 5.1 attach behavior.
• Capture a hung Windows run's process/runspace state or DAP transcript to see
  where the attach handshake stalls under the new image.
• Investigate the harness's readiness logic (RunWithAttachableProcess keys
  off RunspaceBase.AvailabilityChanged via reflection) and the
  Debug-Runspace handshake for an OS-version-sensitive race.
• Once fixed, remove the Skip.If from Skip attach E2E test on in-box Windows PowerShell (20260614 image regression); cap CI job #2318; the timeout-minutes backstop
  can stay.

References

• Stopgap PR: Skip attach E2E test on in-box Windows PowerShell (20260614 image regression); cap CI job #2318
• Original reported run: https://github.com/PowerShell/PowerShellEditorServices/actions/runs/27734424409