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]>
This commit is contained in:
silverwind authored and bircni committed 2026-11-04 16:32:22 +00:00
1 parent 142d9f8c88
commit dcba546cb3
32 files changed
+374 -593

No files matched your search

-263
View File
@@ -1,263 +0,0 @@
// Copyright 2026 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package process
import (
"context"
"errors"
"fmt"
"os"
"strings"
"unsafe"
"golang.org/x/sys/windows"
)
func KillProcessesWithPaths(ctx context.Context, dirs []string, trackingID string) (int, error) {
if len(dirs) == 0 || trackingID == "" {
return 0, nil
}
pids, err := listProcessIDs()
if err != nil {
return 0, err
}
self := uint32(os.Getpid())
var ownSession uint32
if err := windows.ProcessIdToSessionId(self, &ownSession); err != nil {
return 0, fmt.Errorf("look up own session: %w", err)
}
var killed int
var errs []error
for _, pid := range pids {
if ctx.Err() != nil {
break
}
if pid == self || pid == 0 || pid == 4 {
continue
}
var session uint32
if err := windows.ProcessIdToSessionId(pid, &session); err != nil || session != ownSession {
continue
}
ok, err := terminateIfPathsMatch(pid, dirs, trackingID)
if err != nil {
errs = append(errs, err)
continue
}
if ok {
killed++
}
}
return killed, errors.Join(errs...)
}
func terminateIfPathsMatch(pid uint32, dirs []string, trackingID string) (bool, error) {
handle, err := windows.OpenProcess(windows.PROCESS_QUERY_INFORMATION|windows.PROCESS_VM_READ|windows.PROCESS_TERMINATE, false, pid)
if err != nil {
return false, nil
}
defer func() { _ = windows.CloseHandle(handle) }()
params, err := processParametersHandle(handle)
if err != nil {
return false, nil
}
matches := false
fields:
for _, field := range []struct {
offset uintptr
wow64 uintptr
command bool
}{
{unsafe.Offsetof(windows.RTL_USER_PROCESS_PARAMETERS{}.CurrentDirectory), 0x24, false},
{unsafe.Offsetof(windows.RTL_USER_PROCESS_PARAMETERS{}.ImagePathName), 0x38, false},
{unsafe.Offsetof(windows.RTL_USER_PROCESS_PARAMETERS{}.CommandLine), 0x40, true},
} {
value, err := params.unicodeString(handle, field.offset, field.wow64)
if err != nil {
continue
}
for _, dir := range dirs {
if !field.command && dirContainsPath(dir, value) || field.command && strings.Contains(strings.ToLower(value), strings.ToLower(strings.TrimRight(dir, `\/`)+`\`)) {
matches = true
break fields
}
}
}
if !matches {
return false, nil
}
value, err := params.environmentVariable(handle, "RUNNER_TRACKING_ID")
if err != nil || value != trackingID {
return false, nil
}
if err := windows.TerminateProcess(handle, 1); err != nil {
return false, fmt.Errorf("terminate pid %d: %w", pid, err)
}
return true, nil
}
// listProcessIDs returns the PIDs of every process in a Toolhelp snapshot.
func listProcessIDs() ([]uint32, error) {
snap, err := windows.CreateToolhelp32Snapshot(windows.TH32CS_SNAPPROCESS, 0)
if err != nil {
return nil, err
}
defer func() { _ = windows.CloseHandle(snap) }()
var entry windows.ProcessEntry32
entry.Size = uint32(unsafe.Sizeof(entry))
pids := make([]uint32, 0, 256)
err = windows.Process32First(snap, &entry)
for err == nil {
pids = append(pids, entry.ProcessID)
err = windows.Process32Next(snap, &entry)
}
if !errors.Is(err, windows.ERROR_NO_MORE_FILES) {
return pids, err
}
return pids, nil
}
type remoteParameters struct {
address uintptr
wow64 bool
}
func processParametersHandle(handle windows.Handle) (remoteParameters, error) {
var pbi windows.PROCESS_BASIC_INFORMATION
var retLen uint32
if err := windows.NtQueryInformationProcess(handle, windows.ProcessBasicInformation, unsafe.Pointer(&pbi), uint32(unsafe.Sizeof(pbi)), &retLen); err != nil {
return remoteParameters{}, err
}
pebAddr := uintptr(unsafe.Pointer(pbi.PebBaseAddress))
wow64 := false
if unsafe.Sizeof(uintptr(0)) == 8 {
var peb32 uintptr
if err := windows.NtQueryInformationProcess(handle, windows.ProcessWow64Information, unsafe.Pointer(&peb32), uint32(unsafe.Sizeof(peb32)), &retLen); err != nil {
return remoteParameters{}, err
}
if peb32 != 0 {
pebAddr = peb32
wow64 = true
}
}
if pebAddr == 0 {
return remoteParameters{}, errors.New("nil PEB base address")
}
params := remoteParameters{wow64: wow64}
address, err := params.pointer(handle, pebAddr+params.offset(unsafe.Offsetof(windows.PEB{}.ProcessParameters), 0x10))
if err != nil {
return remoteParameters{}, err
}
if address == 0 {
return remoteParameters{}, errors.New("nil process parameters")
}
params.address = address
return params, nil
}
func (p remoteParameters) offset(native, wow64 uintptr) uintptr {
if p.wow64 {
return wow64
}
return native
}
func (p remoteParameters) pointer(handle windows.Handle, address uintptr) (uintptr, error) {
var value uintptr
size := unsafe.Sizeof(value)
if p.wow64 {
size = 4
}
var read uintptr
if err := windows.ReadProcessMemory(handle, address, (*byte)(unsafe.Pointer(&value)), size, &read); err != nil {
return 0, err
}
if read != size {
return 0, errors.New("short pointer read")
}
return value, nil
}
func (p remoteParameters) unicodeString(handle windows.Handle, native, wow64 uintptr) (string, error) {
address := p.address + p.offset(native, wow64)
length, err := readRemoteU16(handle, address)
if err != nil || length == 0 {
return "", err
}
buffer, err := p.pointer(handle, address+p.offset(unsafe.Offsetof(windows.NTUnicodeString{}.Buffer), 4))
if err != nil {
return "", err
}
value := make([]uint16, (int(length)+1)/2)
var read uintptr
if err := windows.ReadProcessMemory(handle, buffer, (*byte)(unsafe.Pointer(&value[0])), uintptr(length), &read); err != nil {
return "", err
}
if read != uintptr(length) {
return "", errors.New("short string read")
}
return windows.UTF16ToString(value), nil
}
func (p remoteParameters) environmentVariable(handle windows.Handle, name string) (string, error) {
address, err := p.pointer(handle, p.address+p.offset(unsafe.Offsetof(windows.RTL_USER_PROCESS_PARAMETERS{}.Environment), 0x48))
if err != nil || address == 0 {
return "", err
}
var entry []uint16
for consumed := uintptr(0); consumed < 32*1024*1024; {
var region windows.MemoryBasicInformation
if err := windows.VirtualQueryEx(handle, address+consumed, &region, unsafe.Sizeof(region)); err != nil {
return "", err
}
size := min(uintptr(4096), region.BaseAddress+region.RegionSize-address-consumed)
if size < 2 {
return "", errors.New("invalid environment region")
}
buffer := make([]uint16, size/2)
var read uintptr
if err := windows.ReadProcessMemory(handle, address+consumed, (*byte)(unsafe.Pointer(&buffer[0])), uintptr(len(buffer)*2), &read); err != nil {
return "", err
}
if read != uintptr(len(buffer)*2) {
return "", errors.New("short environment read")
}
for _, value := range buffer {
if value != 0 {
entry = append(entry, value)
continue
}
if len(entry) == 0 {
return "", nil
}
key, value, ok := strings.Cut(windows.UTF16ToString(entry), "=")
if ok && strings.EqualFold(key, name) {
return value, nil
}
entry = entry[:0]
}
consumed += read
}
return "", errors.New("environment exceeds read limit")
}
// readRemoteU16 reads a single uint16 from another process.
func readRemoteU16(handle windows.Handle, addr uintptr) (uint16, error) {
var v uint16
var n uintptr
if err := windows.ReadProcessMemory(handle, addr, (*byte)(unsafe.Pointer(&v)), unsafe.Sizeof(v), &n); err != nil {
return 0, err
}
if n != unsafe.Sizeof(v) {
return 0, errors.New("short uint16 read")
}
return v, nil
}