vm: avoid starting the same VM multiple times simultaneously

While libvirt handle locking itself, there is also Qubes-specific
startup part. Especially starting qrexec-daemon and waiting until
qrexec-agent connect to it. When someone will attempt to start VM the
second time (or simply assume it's already running) - qrexec will not be
connected yet and the operation will fail. Solve the problem by wrapping
the whole vm.start() function with a lock, including a check if VM is
running and waiting for qrexec.

Also, do not throw exception if VM is already running.

This way, after a call to vm.start(), VM will be started with qrexec
connected - regardless of who really started it.
Note that, it will not solve the situation when someone check if VM is
running manually, like:

    if not vm.is_running():
        yield from vm.start()

Such code should be changed to simply:

    yield from vm.start()

Fixes QubesOS/qubes-issues#2001
Fixes QubesOS/qubes-issues#2666
This commit is contained in:
Marek Marczykowski-Górecki 2017-06-01 03:31:59 +02:00
parent 1f86c9253c
commit 018877a19c
No known key found for this signature in database
GPG Key ID: 063938BA42CFA724

View File

@ -704,6 +704,10 @@ class QubesVM(qubes.vm.mix.net.NetVMMixin, qubes.vm.BaseVM):
# will be initialized after loading all the properties
#: operations which shouldn't happen simultaneously with qube startup
# (including another startup of the same qube)
self.startup_lock = asyncio.Lock()
# fire hooks
if xml is None:
self.events_enabled = True
@ -841,88 +845,99 @@ class QubesVM(qubes.vm.mix.net.NetVMMixin, qubes.vm.BaseVM):
:param int mem_required: FIXME
'''
# Intentionally not used is_running(): eliminate also "Paused",
# "Crashed", "Halting"
if self.get_power_state() != 'Halted':
raise qubes.exc.QubesVMNotHaltedError(self)
with (yield from self.startup_lock):
# Intentionally not used is_running(): eliminate also "Paused",
# "Crashed", "Halting"
if self.get_power_state() != 'Halted':
return
self.log.info('Starting {}'.format(self.name))
self.log.info('Starting {}'.format(self.name))
self.fire_event_pre('domain-pre-start', preparing_dvm=preparing_dvm,
start_guid=start_guid, mem_required=mem_required)
self.fire_event_pre('domain-pre-start', preparing_dvm=preparing_dvm,
start_guid=start_guid, mem_required=mem_required)
yield from self.storage.verify()
yield from self.storage.verify()
if self.netvm is not None:
# pylint: disable = no-member
if self.netvm.qid != 0:
if not self.netvm.is_running():
yield from self.netvm.start(start_guid=start_guid,
notify_function=notify_function)
if self.netvm is not None:
# pylint: disable = no-member
if self.netvm.qid != 0:
if not self.netvm.is_running():
yield from self.netvm.start(start_guid=start_guid,
notify_function=notify_function)
# TODO: lock
qmemman_client = yield from asyncio.get_event_loop().\
run_in_executor(None, self.request_memory, mem_required)
qmemman_client = yield from asyncio.get_event_loop().run_in_executor(
None, self.request_memory, mem_required)
try:
yield from self.storage.start()
self._update_libvirt_domain()
try:
yield from self.storage.start()
self._update_libvirt_domain()
self.libvirt_domain.createWithFlags(
libvirt.VIR_DOMAIN_START_PAUSED)
finally:
if qmemman_client:
qmemman_client.close()
self.libvirt_domain.createWithFlags(libvirt.VIR_DOMAIN_START_PAUSED)
finally:
if qmemman_client:
qmemman_client.close()
try:
self.fire_event('domain-spawn',
preparing_dvm=preparing_dvm, start_guid=start_guid)
try:
self.fire_event('domain-spawn',
preparing_dvm=preparing_dvm, start_guid=start_guid)
self.log.info('Setting Qubes DB info for the VM')
yield from self.start_qubesdb()
self.create_qdb_entries()
self.log.info('Setting Qubes DB info for the VM')
yield from self.start_qubesdb()
self.create_qdb_entries()
if preparing_dvm:
self.qdb.write('/dvm', '1')
if preparing_dvm:
self.qdb.write('/dvm', '1')
self.log.warning('Activating the {} VM'.format(self.name))
self.libvirt_domain.resume()
self.log.warning('Activating the {} VM'.format(self.name))
self.libvirt_domain.resume()
# close() is not really needed, because the descriptor is
# close-on-exec anyway, the reason to postpone close() is that
# possibly xl is not done constructing the domain after its main
# process exits so we close() when we know the domain is up the
# successful unpause is some indicator of it
if qmemman_client:
qmemman_client.close()
qmemman_client = None
# close() is not really needed, because the descriptor is
# close-on-exec anyway, the reason to postpone close() is that
# possibly xl is not done constructing the domain after its main
# process exits so we close() when we know the domain is up the
# successful unpause is some indicator of it
if qmemman_client:
qmemman_client.close()
qmemman_client = None
#if self._start_guid_first and start_guid and not preparing_dvm \
# and os.path.exists('/var/run/shm.id'):
# self.start_guid()
# if self._start_guid_first and start_guid and not preparing_dvm \
# and os.path.exists('/var/run/shm.id'):
# self.start_guid()
if not preparing_dvm:
yield from self.start_qrexec_daemon()
if not preparing_dvm:
yield from self.start_qrexec_daemon()
self.fire_event('domain-start',
preparing_dvm=preparing_dvm, start_guid=start_guid)
self.fire_event('domain-start',
preparing_dvm=preparing_dvm, start_guid=start_guid)
except: # pylint: disable=bare-except
if self.is_running() or self.is_paused():
# This avoids losing the exception if an exception is raised in
# self.force_shutdown(), because the vm is not running or paused
yield from self.kill() # pylint: disable=not-an-iterable
raise
finally:
if qmemman_client:
qmemman_client.close()
except: # pylint: disable=bare-except
if self.is_running() or self.is_paused():
# This avoids losing the exception if an exception is
# raised in self.force_shutdown(), because the vm is not
# running or paused
yield from self.kill() # pylint: disable=not-an-iterable
raise
finally:
if qmemman_client:
qmemman_client.close()
return self
@asyncio.coroutine
def on_domain_shutdown_coro(self):
'''Coroutine for executing cleanup after domain shutdown.
Do not allow domain to be started again until this finishes.
'''
with (yield from self.startup_lock):
yield from self.storage.stop()
@qubes.events.handler('domain-shutdown')
def on_domain_shutdown(self, _event, **_kwargs):
'''Cleanup after domain shutdown'''
asyncio.ensure_future(self.storage.stop())
# TODO: ensure that domain haven't been started _before_ this
# coroutine got a chance to acquire a lock
asyncio.ensure_future(self.on_domain_shutdown_coro())
@asyncio.coroutine
def shutdown(self, force=False, wait=False):