3 Commits
Author SHA1 Message Date
silverwind dcba546cb3 fix: align process handling and fix Windows crashes (#1376)
The intermittent Windows test crashes came from step errors formatting a nil `rc.Run`. `fmt` recovers the nil dereference, but on CPUs with AMX the recovered fault overflows into the heap (https://github.com/golang/go/issues/81238). The errors now name the job, and the Windows test VM hides AMX and runs on UTC, so cached boots no longer jump back 8 hours mid-test.

Windows commands now start suspended and resume once they are in their job objects, so no child escapes them. That drops the process scan reading other processes' memory and the last `unsafe` code, which `depguard` now denies. Step processes outlive a crashed runner, and commands are no longer looked up in the runner's working directory, both as with GitHub's runner.

Like GitHub's runner, host jobs on Linux and macOS now end the processes carrying the job's `RUNNER_TRACKING_ID`, and image pulls and Dockerfile action builds are retried.

Reviewed-on: https://gitea.com/gitea/runner/pulls/1376
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
2026-11-04 16:32:22 +00:00
silverwind d6f5b68a0d enhance: test on Windows and fix what it found (#1348)
Adds a weekly and on-demand workflow that runs the cross-compiled tests on Windows Server 2025 Core in a [dockur/windows](https://github.com/dockur/windows) QEMU VM, and fixes the Windows bugs it found in host and UNC paths, env handling, shell quoting, post-task scripts, health checks, checkout, action symlinks, step summaries, Git client certificates and docker mode from a Windows host.

Host-mode steps now emit UTF-8, `bash` falls back to Git's bash instead of the WSL launcher, the runner can run as a native Windows service, steps can keep background helpers with `RUNNER_TRACKING_ID`, and Windows arm64 binaries are released.

Fixes: https://gitea.com/gitea/runner/issues/1339
Fixes: https://gitea.com/gitea/runner/issues/637
Fixes: https://gitea.com/gitea/runner/issues/1172
Fixes: https://gitea.com/gitea/runner/issues/347
Reviewed-on: https://gitea.com/gitea/runner/pulls/1348
Reviewed-by: bircni <[email protected]>
2026-11-03 01:26:10 +00:00
bircniandsilverwind 7bec310002 fix: stop host-mode jobs from leaking processes on Windows (#1080)
## Problem

On host-mode Windows runners, a job can leave processes running after it finishes,
and those leftovers hold file handles that block deletion of the workspace.

Today a step's tree is only torn down when the step is *cancelled*
(`process.Killer`). A step that completes leaves whatever it spawned alive, and
the existing workspace scan in `terminateRunningProcesses` misses two shapes of
leftover: orphans whose parent already exited (no tree to walk, and their
executable often lives outside the workspace), and processes that merely *run in*
the workspace but reference no path from it — `Win32_Process` exposes no working
directory, so the scan cannot match them.

## Solution

Two additions in `internal/pkg/process`, both no-ops outside Windows:

- **`process.Group`** — a job-scoped Windows Job Object created with
  `JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE`. Step processes are assigned to it before
  they get their own `Killer`, so the step's job nests inside the job's:
  cancellation still kills exactly the step's tree, while `Remove` closing the
  group makes the kernel terminate everything still assigned, whatever its
  parentage. The kernel also drops the handle when the runner exits, so a crashed
  runner cannot strand processes.

- **`process.KillProcessesWithCWDUnder`** — a best-effort net for processes that
  never joined the job (started via a service or scheduled task). It reads each
  process's working directory from its PEB and terminates those under a workspace
  dir. Processes it cannot open are skipped.

---------

Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1080
Reviewed-by: silverwind <[email protected]>
2026-07-22 15:04:07 +00:00