tests/killdaemons.py

worker: rewrite error handling so os._exit covers all cases

Previously the worker error handling is like:

    pid = os.fork()   --+
    if pid == 0:        |
        ....            | problematic
        ....          --+
        try:          --+
            ....        | worker error handling
                      --+

If a signal arrives when Python is executing the "problematic" lines, an
external error handling (dispatch.py) will take over the control flow and
it's no longer guaranteed "os._exit" is called (see 86cd09bc13ba for why it
is necessary).

This patch rewrites the error handling so it covers all possible code paths
for a worker even during fork.

Note: "os.getpid() == parentpid" is used to test if the process is parent or
not intentionally, instead of checking "pid", because "pid = os.fork()" may
be not atomic - it's possible that that a signal hits the worker before the
assignment completes [1].  The newly added test replaces "os.fork" to
exercise that extreme case.

[1]: CPython compiles "pid = os.fork()" to 2 byte codes: "CALL_FUNCTION" and
"STORE_FAST", so it's probably not atomic:

    def f():
        pid = os.fork()

    dis.dis(f)
      2           0 LOAD_GLOBAL              0 (os)
                  3 LOAD_ATTR                1 (fork)
                  6 CALL_FUNCTION            0
                  9 STORE_FAST               0 (pid)
                 12 LOAD_CONST               0 (None)
                 15 RETURN_VALUE

#!/usr/bin/env python

from __future__ import absolute_import
import errno
import os
import signal
import sys
import time

if os.name =='nt':
    import ctypes

    def _check(ret, expectederr=None):
        if ret == 0:
            winerrno = ctypes.GetLastError()
            if winerrno == expectederr:
                return True
            raise ctypes.WinError(winerrno)

    def kill(pid, logfn, tryhard=True):
        logfn('# Killing daemon process %d' % pid)
        PROCESS_TERMINATE = 1
        PROCESS_QUERY_INFORMATION = 0x400
        SYNCHRONIZE = 0x00100000
        WAIT_OBJECT_0 = 0
        WAIT_TIMEOUT = 258
        handle = ctypes.windll.kernel32.OpenProcess(
                PROCESS_TERMINATE|SYNCHRONIZE|PROCESS_QUERY_INFORMATION,
                False, pid)
        if handle == 0:
            _check(0, 87) # err 87 when process not found
            return # process not found, already finished
        try:
            r = ctypes.windll.kernel32.WaitForSingleObject(handle, 100)
            if r == WAIT_OBJECT_0:
                pass # terminated, but process handle still available
            elif r == WAIT_TIMEOUT:
                _check(ctypes.windll.kernel32.TerminateProcess(handle, -1))
            else:
                _check(r)

            # TODO?: forcefully kill when timeout
            #        and ?shorter waiting time? when tryhard==True
            r = ctypes.windll.kernel32.WaitForSingleObject(handle, 100)
                                                       # timeout = 100 ms
            if r == WAIT_OBJECT_0:
                pass # process is terminated
            elif r == WAIT_TIMEOUT:
                logfn('# Daemon process %d is stuck')
            else:
                _check(r) # any error
        except: #re-raises
            ctypes.windll.kernel32.CloseHandle(handle) # no _check, keep error
            raise
        _check(ctypes.windll.kernel32.CloseHandle(handle))

else:
    def kill(pid, logfn, tryhard=True):
        try:
            os.kill(pid, 0)
            logfn('# Killing daemon process %d' % pid)
            os.kill(pid, signal.SIGTERM)
            if tryhard:
                for i in range(10):
                    time.sleep(0.05)
                    os.kill(pid, 0)
            else:
                time.sleep(0.1)
                os.kill(pid, 0)
            logfn('# Daemon process %d is stuck - really killing it' % pid)
            os.kill(pid, signal.SIGKILL)
        except OSError as err:
            if err.errno != errno.ESRCH:
                raise

def killdaemons(pidfile, tryhard=True, remove=False, logfn=None):
    if not logfn:
        logfn = lambda s: s
    # Kill off any leftover daemon processes
    try:
        fp = open(pidfile)
        for line in fp:
            try:
                pid = int(line)
                if pid <= 0:
                    raise ValueError
            except ValueError:
                logfn('# Not killing daemon process %s - invalid pid'
                      % line.rstrip())
                continue
            kill(pid, logfn, tryhard)
        fp.close()
        if remove:
            os.unlink(pidfile)
    except IOError:
        pass

if __name__ == '__main__':
    if len(sys.argv) > 1:
        path, = sys.argv[1:]
    else:
        path = os.environ["DAEMON_PIDS"]

    killdaemons(path)