Source code for gevent._semaphore

# cython: auto_pickle=False,embedsignature=True,always_allow_keywords=False
###
# This file is ``gevent._semaphore`` so that it can be compiled by Cython
# individually. However, this is not the place to import from. Everyone,
# gevent internal code included, must import from ``gevent.lock``.
# The only exception are .pxd files which need access to the
# C code; the PURE_PYTHON things that have to happen and which are
# handled in ``gevent.lock``, do not apply to them.
###
from __future__ import print_function, absolute_import, division

__all__ = [
    'Semaphore',
    'BoundedSemaphore',
]

from time import sleep as _native_sleep

from gevent._compat import monotonic
from gevent.exceptions import InvalidThreadUseError
from gevent.exceptions import LoopExit
from gevent.timeout import Timeout

def _get_linkable():
    x = __import__('gevent._abstract_linkable')
    return x._abstract_linkable.AbstractLinkable
locals()['AbstractLinkable'] = _get_linkable()
del _get_linkable

from gevent._hub_local import get_hub_if_exists
from gevent._hub_local import get_hub
from gevent.hub import spawn_raw

class _LockReleaseLink(object):
    __slots__ = (
        'lock',
    )

    def __init__(self, lock):
        self.lock = lock

    def __call__(self, _):
        self.lock.release()

_UNSET = object()
_MULTI = object()

class Semaphore(AbstractLinkable): # pylint:disable=undefined-variable
    """
    Semaphore(value=1) -> Semaphore

    .. seealso:: :class:`BoundedSemaphore` for a safer version that prevents
       some classes of bugs. If unsure, most users should opt for `BoundedSemaphore`.

    A semaphore manages a counter representing the number of `release`
    calls minus the number of `acquire` calls, plus an initial value.
    The `acquire` method blocks if necessary until it can return
    without making the counter negative. A semaphore does not track ownership
    by greenlets; any greenlet can call `release`, whether or not it has previously
    called `acquire`.

    If not given, ``value`` defaults to 1.

    The semaphore is a context manager and can be used in ``with`` statements.

    This Semaphore's ``__exit__`` method does not call the trace function
    on CPython, but does under PyPy.

    .. versionchanged:: 1.4.0
        Document that the order in which waiters are awakened is not specified. It was not
        specified previously, but due to CPython implementation quirks usually went in FIFO order.
    .. versionchanged:: 1.5a3
       Waiting greenlets are now awakened in the order in which they waited.
    .. versionchanged:: 1.5a3
       The low-level ``rawlink`` method (most users won't use this) now automatically
       unlinks waiters before calling them.
    .. versionchanged:: 20.12.0
       Improved support for multi-threaded usage. When multi-threaded usage is detected,
       instances will no longer create the thread's hub if it's not present.
    """

    __slots__ = (
        'counter',
        # long integer, signed (Py2) or unsigned (Py3); see comments
        # in the .pxd file for why we store as Python object. Set to ``_UNSET``
        # initially. Set to the ident of the first thread that
        # acquires us. If we later see a different thread ident, set
        # to ``_MULTI``.
        '_multithreaded',
    )

    def __init__(self, value=1, hub=None):
        self.counter = value
        if self.counter < 0: # Do the check after Cython native int conversion
            raise ValueError("semaphore initial value must be >= 0")
        super(Semaphore, self).__init__(hub)
        self._notify_all = False
        self._multithreaded = _UNSET

    def __str__(self):
        return '<%s at 0x%x counter=%s _links[%s]>' % (
            self.__class__.__name__,
            id(self),
            self.counter,
            self.linkcount()
        )

[docs] def locked(self): """ Return a boolean indicating whether the semaphore can be acquired (`False` if the semaphore *can* be acquired). Most useful with binary semaphores (those with an initial value of 1). :rtype: bool """ return self.counter <= 0
def release(self): """ Release the semaphore, notifying any waiters if needed. There is no return value. .. note:: This can be used to over-release the semaphore. (Release more times than it has been acquired or was initially created with.) This is usually a sign of a bug, but under some circumstances it can be used deliberately, for example, to model the arrival of additional resources. :rtype: None """ self.counter += 1 self._check_and_notify() return self.counter
[docs] def ready(self): """ Return a boolean indicating whether the semaphore can be acquired (`True` if the semaphore can be acquired). :rtype: bool """ return self.counter > 0
def _start_notify(self): self._check_and_notify() def _wait_return_value(self, waited, wait_success): if waited: return wait_success # We didn't even wait, we must be good to go. # XXX: This is probably dead code, we're careful not to go into the wait # state if we don't expect to need to return True def wait(self, timeout=None): """ Wait until it is possible to acquire this semaphore, or until the optional *timeout* elapses. .. note:: If this semaphore was initialized with a *value* of 0, this method will block forever if no timeout is given. :keyword float timeout: If given, specifies the maximum amount of seconds this method will block. :return: A number indicating how many times the semaphore can be acquired before blocking. *This could be 0,* if other waiters acquired the semaphore. :rtype: int """ if self.counter > 0: return self.counter self._wait(timeout) # return value irrelevant, whether we got it or got a timeout return self.counter def acquire(self, blocking=True, timeout=None): """ acquire(blocking=True, timeout=None) -> bool Acquire the semaphore. .. note:: If this semaphore was initialized with a *value* of 0, this method will block forever (unless a timeout is given or blocking is set to false). :keyword bool blocking: If True (the default), this function will block until the semaphore is acquired. :keyword float timeout: If given, and *blocking* is true, specifies the maximum amount of seconds this method will block. :return: A `bool` indicating whether the semaphore was acquired. If ``blocking`` is True and ``timeout`` is None (the default), then (so long as this semaphore was initialized with a size greater than 0) this will always return True. If a timeout was given, and it expired before the semaphore was acquired, False will be returned. (Note that this can still raise a ``Timeout`` exception, if some other caller had already started a timer.) """ # pylint:disable=too-many-return-statements,too-many-branches # Sadly, the body of this method is rather complicated. if self._multithreaded is _UNSET: self._multithreaded = self._get_thread_ident() elif self._multithreaded != self._get_thread_ident(): self._multithreaded = _MULTI # We conceptually now belong to the hub of the thread that # called this, whether or not we have to block. Note that we # cannot force it to be created yet, because Semaphore is used # by importlib.ModuleLock which is used when importing the hub # itself! This also checks for cross-thread issues. invalid_thread_use = None try: self._capture_hub(False) except InvalidThreadUseError as e: # My hub belongs to some other thread. We didn't release the GIL/object lock # by raising the exception, so we know this is still true. invalid_thread_use = e.args e = None if not self.counter and blocking: # We would need to block. So coordinate with the main hub. return self.__acquire_from_other_thread(invalid_thread_use, blocking, timeout) if self.counter > 0: self.counter -= 1 return True if not blocking: return False if self._multithreaded is not _MULTI and self.hub is None: # pylint:disable=access-member-before-definition self.hub = get_hub() # pylint:disable=attribute-defined-outside-init if self.hub is None and not invalid_thread_use: # Someone else is holding us. There's not a hub here, # nor is there a hub in that thread. We'll need to use regular locks. # This will be unfair to yet a third thread that tries to use us with greenlets. return self.__acquire_from_other_thread( (None, None, self._getcurrent(), "NoHubs"), blocking, timeout ) # self._wait may drop both the GIL and the _lock_lock. # By the time we regain control, both have been reacquired. try: success = self._wait(timeout) except LoopExit as ex: args = ex.args ex = None if self.counter: success = True else: # Avoid using ex.hub property to keep holding the GIL if len(args) == 3 and args[1].main_hub: # The main hub, meaning the main thread. We probably can do nothing with this. raise return self.__acquire_from_other_thread( (self.hub, get_hub_if_exists(), self._getcurrent(), "LoopExit"), blocking, timeout) if not success: assert timeout is not None # Our timer expired. return False # Neither our timer or another one expired, so we blocked until # awoke. Therefore, the counter is ours assert self.counter > 0, (self.counter, blocking, timeout, success,) self.counter -= 1 return True _py3k_acquire = acquire # PyPy needs this; it must be static for Cython def __enter__(self): self.acquire() def __exit__(self, t, v, tb): self.release() def _handle_unswitched_notifications(self, unswitched): # If we fail to switch to a greenlet in another thread to send # a notification, just re-queue it, in the hopes that the # other thread will eventually run notifications itself. # # We CANNOT do what the ``super()`` does and actually allow # this notification to get run sometime in the future by # scheduling a callback in the other thread. The algorithm # that we use to handle cross-thread locking/unlocking was # designed before the schedule-a-callback mechanism was # implemented. If we allow this to be run as a callback, we # can find ourself the victim of ``InvalidSwitchError`` (or # worse, silent corruption) because the switch can come at an # unexpected time: *after* the destination thread has already # acquired the lock. # # This manifests in a fairly reliable test failure, # ``gevent.tests.test__semaphore`` # ``TestSemaphoreMultiThread.test_dueling_threads_with_hub``, # but ONLY when running in PURE_PYTHON mode. # # TODO: Maybe we can rewrite that part of the algorithm to be friendly to # running the callbacks? self._links.extend(unswitched) def __add_link(self, link): if not self._notifier: self.rawlink(link) else: self._notifier.args[0].append(link) def __acquire_from_other_thread(self, ex_args, blocking, timeout): assert blocking # Some other hub owns this object. We must ask it to wake us # up. In general, we can't use a Python-level ``Lock`` because # # (1) it doesn't support a timeout on all platforms; and # (2) we don't want to block this hub from running. # # So we need to do so in a way that cooperates with *two* # hubs. That's what an async watcher is built for. # # Of course, if we don't actually have two hubs, then we must find some other # solution. That involves using a lock. # We have to take an action that drops the GIL and drops the object lock # to allow the main thread (the thread for our hub) to advance. owning_hub = ex_args[0] hub_for_this_thread = ex_args[1] current_greenlet = ex_args[2] if owning_hub is None and hub_for_this_thread is None: return self.__acquire_without_hubs(timeout) if hub_for_this_thread is None: # Probably a background worker thread. We don't want to create # the hub if not needed, and since it didn't exist there are no # other greenlets that we could yield to anyway, so there's nothing # to block and no reason to try to avoid blocking, so using a native # lock is the simplest way to go. return self.__acquire_using_other_hub(owning_hub, timeout) # We have a hub we don't want to block. Use an async watcher # and ask the next releaser of this object to wake us up. return self.__acquire_using_two_hubs(hub_for_this_thread, current_greenlet, timeout) def __acquire_using_two_hubs(self, hub_for_this_thread, current_greenlet, timeout): # Allocating and starting the watcher *could* release the GIL. # with the libev corcext, allocating won't, but starting briefly will. # With other backends, allocating might, and starting might also. # So... watcher = hub_for_this_thread.loop.async_() send = watcher.send_ignoring_arg watcher.start(current_greenlet.switch, self) try: with Timeout._start_new_or_dummy(timeout) as timer: # ... now that we're back holding the GIL, we need to verify our # state. try: while 1: if self.counter > 0: self.counter -= 1 assert self.counter >= 0, (self,) return True self.__add_link(send) # Releases the object lock self._switch_to_hub(hub_for_this_thread) # We waited and got notified. We should be ready now, so a non-blocking # acquire() should succeed. But sometimes we get spurious notifications? # It's not entirely clear how. So we need to loop until we get it, or until # the timer expires result = self.acquire(0) if result: return result except Timeout as tex: if tex is not timer: raise return False finally: self._quiet_unlink_all(send) watcher.stop() watcher.close() def __acquire_from_other_thread_cb(self, results, blocking, timeout, thread_lock): try: result = self.acquire(blocking, timeout) results.append(result) finally: thread_lock.release() return result def __acquire_using_other_hub(self, owning_hub, timeout): assert owning_hub is not get_hub_if_exists() thread_lock = self._allocate_lock() thread_lock.acquire() results = [] owning_hub.loop.run_callback_threadsafe( spawn_raw, self.__acquire_from_other_thread_cb, results, 1, # blocking, timeout, # timeout, thread_lock) # We MUST use a blocking acquire here, or at least be sure we keep going # until we acquire it. If we timed out waiting here, # just before the callback runs, then we would be out of sync. self.__spin_on_native_lock(thread_lock, None) return results[0] def __acquire_without_hubs(self, timeout): thread_lock = self._allocate_lock() thread_lock.acquire() absolute_expiration = 0 begin = 0 if timeout: absolute_expiration = monotonic() + timeout # Cython won't compile a lambda here link = _LockReleaseLink(thread_lock) while 1: self.__add_link(link) if absolute_expiration: begin = monotonic() got_native = self.__spin_on_native_lock(thread_lock, timeout) self._quiet_unlink_all(link) if got_native: if self.acquire(0): return True if absolute_expiration: now = monotonic() if now >= absolute_expiration: return False duration = now - begin timeout -= duration if timeout <= 0: return False def __spin_on_native_lock(self, thread_lock, timeout): expiration = 0 if timeout: expiration = monotonic() + timeout self._drop_lock_for_switch_out() try: # TODO: When timeout is given and the lock supports that # (Python 3), pass that. # Python 2 has terrible behaviour where lock acquires can't # be interrupted, so we use a spin loop while not thread_lock.acquire(0): if expiration and monotonic() >= expiration: return False _native_sleep(0.001) return True finally: self._acquire_lock_for_switch_in() class BoundedSemaphore(Semaphore): """ BoundedSemaphore(value=1) -> BoundedSemaphore A bounded semaphore checks to make sure its current value doesn't exceed its initial value. If it does, :class:`ValueError` is raised. In most situations semaphores are used to guard resources with limited capacity. If the semaphore is released too many times it's a sign of a bug. If not given, *value* defaults to 1. """ __slots__ = ( '_initial_value', ) #: For monkey-patching, allow changing the class of error we raise _OVER_RELEASE_ERROR = ValueError def __init__(self, *args, **kwargs): Semaphore.__init__(self, *args, **kwargs) self._initial_value = self.counter def release(self): """ Like :meth:`Semaphore.release`, but raises :class:`ValueError` if the semaphore is being over-released. """ if self.counter >= self._initial_value: raise self._OVER_RELEASE_ERROR("Semaphore released too many times") counter = Semaphore.release(self) # When we are absolutely certain that no one holds this semaphore, # release our hub and go back to floating. This assists in cross-thread # uses. if counter == self._initial_value: self.hub = None # pylint:disable=attribute-defined-outside-init return counter def _at_fork_reinit(self): super(BoundedSemaphore, self)._at_fork_reinit() self.counter = self._initial_value # By building the semaphore with Cython under PyPy, we get # atomic operations (specifically, exiting/releasing), at the # cost of some speed (one trivial semaphore micro-benchmark put the pure-python version # at around 1s and the compiled version at around 4s). Some clever subclassing # and having only the bare minimum be in cython might help reduce that penalty. # NOTE: You must use version 0.23.4 or later to avoid a memory leak. # https://mail.python.org/pipermail/cython-devel/2015-October/004571.html # However, that's all for naught on up to and including PyPy 4.0.1 which # have some serious crashing bugs with GC interacting with cython. # It hasn't been tested since then, and PURE_PYTHON is assumed to be true # for PyPy in all cases anyway, so this does nothing. from gevent._util import import_c_accel import_c_accel(globals(), 'gevent.__semaphore')