1
# Copyright (C) 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""On-disk mutex protecting a resource
19
bzr on-disk objects are locked by the existence of a directory with a
20
particular name within the control directory. We use this rather than OS
21
internal locks (such as flock etc) because they can be seen across all
22
transports, including http.
24
Objects can be read if there is only physical read access; therefore
25
readers can never be required to create a lock, though they will
26
check whether a writer is using the lock. Writers can't detect
27
whether anyone else is reading from the resource as they write.
28
This works because of ordering constraints that make sure readers
29
see a consistent view of existing data.
31
Waiting for a lock must be done by polling; this can be aborted after
34
Locks must always be explicitly released, typically from a try/finally
35
block -- they are not released from a finalizer or when Python
38
Locks may fail to be released if the process is abruptly terminated
39
(machine stop, SIGKILL) or if a remote transport becomes permanently
40
disconnected. There is therefore a method to break an existing lock.
41
This should rarely be used, and generally only with user approval.
42
Locks contain some information on when the lock was taken and by who
43
which may guide in deciding whether it can safely be broken. (This is
44
similar to the messages displayed by emacs and vim.) Note that if the
45
lock holder is still alive they will get no notification that the lock
46
has been broken and will continue their work -- so it is important to be
47
sure they are actually dead.
49
A lock is represented on disk by a directory of a particular name,
50
containing an information file. Taking a lock is done by renaming a
51
temporary directory into place. We use temporary directories because
52
for all known transports and filesystems we believe that exactly one
53
attempt to claim the lock will succeed and the others will fail. (Files
54
won't do because some filesystems or transports only have
55
rename-and-overwrite, making it hard to tell who won.)
57
The desired characteristics are:
59
* Locks are not reentrant. (That is, a client that tries to take a
60
lock it already holds may deadlock or fail.)
61
* Stale locks can be guessed at by a heuristic
62
* Lost locks can be broken by any client
63
* Failed lock operations leave little or no mess
64
* Deadlocks are avoided by having a timeout always in use, clients
65
desiring indefinite waits can retry or set a silly big timeout.
67
Storage formats use the locks, and also need to consider concurrency
68
issues underneath the lock. A format may choose not to use a lock
69
at all for some operations.
71
LockDirs always operate over a Transport. The transport may be readonly, in
72
which case the lock can be queried but not acquired.
74
Locks are identified by a path name, relative to a base transport.
76
Calling code will typically want to make sure there is exactly one LockDir
77
object per actual lock on disk. This module does nothing to prevent aliasing
78
and deadlocks will likely occur if the locks are aliased.
80
In the future we may add a "freshen" method which can be called
81
by a lock holder to check that their lock has not been broken, and to
82
update the timestamp within it.
86
>>> from bzrlib.transport.memory import MemoryTransport
87
>>> # typically will be obtained from a BzrDir, Branch, etc
88
>>> t = MemoryTransport()
89
>>> l = LockDir(t, 'sample-lock')
92
>>> # do something here
99
from cStringIO import StringIO
105
from bzrlib.errors import (
117
from bzrlib.trace import mutter, note
118
from bzrlib.transport import Transport
119
from bzrlib.osutils import rand_chars, format_delta
120
from bzrlib.rio import read_stanza, Stanza
123
# XXX: At the moment there is no consideration of thread safety on LockDir
124
# objects. This should perhaps be updated - e.g. if two threads try to take a
125
# lock at the same time they should *both* get it. But then that's unlikely
128
# TODO: Perhaps store some kind of note like the bzr command line in the lock
131
# TODO: Some kind of callback run while polling a lock to show progress
134
# TODO: Make sure to pass the right file and directory mode bits to all
135
# files/dirs created.
138
_DEFAULT_TIMEOUT_SECONDS = 300
139
_DEFAULT_POLL_SECONDS = 1.0
142
class LockDir(object):
143
"""Write-lock guarding access to data."""
145
__INFO_NAME = '/info'
147
def __init__(self, transport, path, file_modebits=0644, dir_modebits=0755):
148
"""Create a new LockDir object.
150
The LockDir is initially unlocked - this just creates the object.
152
:param transport: Transport which will contain the lock
154
:param path: Path to the lock within the base directory of the
157
assert isinstance(transport, Transport), \
158
("not a transport: %r" % transport)
159
self.transport = transport
161
self._lock_held = False
162
self._fake_read_lock = False
163
self._held_dir = path + '/held'
164
self._held_info_path = self._held_dir + self.__INFO_NAME
165
self._file_modebits = file_modebits
166
self._dir_modebits = dir_modebits
167
self.nonce = rand_chars(20)
169
self._report_function = note
172
return '%s(%s%s)' % (self.__class__.__name__,
176
is_held = property(lambda self: self._lock_held)
178
def create(self, mode=None):
179
"""Create the on-disk lock.
181
This is typically only called when the object/directory containing the
182
directory is first created. The lock is not held when it's created.
184
if self.transport.is_readonly():
185
raise UnlockableTransport(self.transport)
186
self.transport.mkdir(self.path, mode=mode)
188
def attempt_lock(self):
189
"""Take the lock; fail if it's already held.
191
If you wish to block until the lock can be obtained, call wait_lock()
194
if self._fake_read_lock:
195
raise LockContention(self)
196
if self.transport.is_readonly():
197
raise UnlockableTransport(self.transport)
199
tmpname = '%s/pending.%s.tmp' % (self.path, rand_chars(20))
201
self.transport.mkdir(tmpname)
203
# This may raise a FileExists exception
204
# which is okay, it will be caught later and determined
205
# to be a LockContention.
206
self.create(mode=self._dir_modebits)
208
# After creating the lock directory, try again
209
self.transport.mkdir(tmpname)
211
info_bytes = self._prepare_info()
212
# We use put_file_non_atomic because we just created a new unique
213
# directory so we don't have to worry about files existing there.
214
# We'll rename the whole directory into place to get atomic
216
self.transport.put_bytes_non_atomic(tmpname + self.__INFO_NAME,
219
self.transport.rename(tmpname, self._held_dir)
220
self._lock_held = True
222
except (PathError, DirectoryNotEmpty, FileExists, ResourceBusy), e:
223
mutter("contention on %r: %s", self, e)
224
raise LockContention(self)
227
"""Release a held lock
229
if self._fake_read_lock:
230
self._fake_read_lock = False
232
if not self._lock_held:
233
raise LockNotHeld(self)
234
# rename before deleting, because we can't atomically remove the whole
236
tmpname = '%s/releasing.%s.tmp' % (self.path, rand_chars(20))
237
# gotta own it to unlock
239
self.transport.rename(self._held_dir, tmpname)
240
self._lock_held = False
241
self.transport.delete(tmpname + self.__INFO_NAME)
242
self.transport.rmdir(tmpname)
244
def break_lock(self):
245
"""Break a lock not held by this instance of LockDir.
247
This is a UI centric function: it uses the bzrlib.ui.ui_factory to
248
prompt for input if a lock is detected and there is any doubt about
249
it possibly being still active.
251
self._check_not_locked()
252
holder_info = self.peek()
253
if holder_info is not None:
254
lock_info = '\n'.join(self._format_lock_info(holder_info))
255
if bzrlib.ui.ui_factory.get_boolean("Break %s" % lock_info):
256
self.force_break(holder_info)
258
def force_break(self, dead_holder_info):
259
"""Release a lock held by another process.
261
WARNING: This should only be used when the other process is dead; if
262
it still thinks it has the lock there will be two concurrent writers.
263
In general the user's approval should be sought for lock breaks.
265
dead_holder_info must be the result of a previous LockDir.peek() call;
266
this is used to check that it's still held by the same process that
267
the user decided was dead. If this is not the current holder,
268
LockBreakMismatch is raised.
270
After the lock is broken it will not be held by any process.
271
It is possible that another process may sneak in and take the
272
lock before the breaking process acquires it.
274
if not isinstance(dead_holder_info, dict):
275
raise ValueError("dead_holder_info: %r" % dead_holder_info)
276
self._check_not_locked()
277
current_info = self.peek()
278
if current_info is None:
279
# must have been recently released
281
if current_info != dead_holder_info:
282
raise LockBreakMismatch(self, current_info, dead_holder_info)
283
tmpname = '%s/broken.%s.tmp' % (self.path, rand_chars(20))
284
self.transport.rename(self._held_dir, tmpname)
285
# check that we actually broke the right lock, not someone else;
286
# there's a small race window between checking it and doing the
288
broken_info_path = tmpname + self.__INFO_NAME
289
broken_info = self._read_info_file(broken_info_path)
290
if broken_info != dead_holder_info:
291
raise LockBreakMismatch(self, broken_info, dead_holder_info)
292
self.transport.delete(broken_info_path)
293
self.transport.rmdir(tmpname)
295
def _check_not_locked(self):
296
"""If the lock is held by this instance, raise an error."""
298
raise AssertionError("can't break own lock: %r" % self)
301
"""Make sure that the lock is still held by this locker.
303
This should only fail if the lock was broken by user intervention,
304
or if the lock has been affected by a bug.
306
If the lock is not thought to be held, raises LockNotHeld. If
307
the lock is thought to be held but has been broken, raises
310
if not self._lock_held:
311
raise LockNotHeld(self)
314
# no lock there anymore!
315
raise LockBroken(self)
316
if info.get('nonce') != self.nonce:
317
# there is a lock, but not ours
318
raise LockBroken(self)
320
def _read_info_file(self, path):
321
"""Read one given info file.
323
peek() reads the info file of the lock holder, if any.
325
return self._parse_info(self.transport.get(path))
328
"""Check if the lock is held by anyone.
330
If it is held, this returns the lock info structure as a rio Stanza,
331
which contains some information about the current lock holder.
332
Otherwise returns None.
335
info = self._read_info_file(self._held_info_path)
336
assert isinstance(info, dict), \
337
"bad parse result %r" % info
339
except NoSuchFile, e:
342
def _prepare_info(self):
343
"""Write information about a pending lock to a temporary file.
346
# XXX: is creating this here inefficient?
347
config = bzrlib.config.GlobalConfig()
349
user = config.user_email()
350
except errors.NoEmailInUsername:
351
user = config.username()
352
s = Stanza(hostname=socket.gethostname(),
353
pid=str(os.getpid()),
354
start_time=str(int(time.time())),
360
def _parse_info(self, info_file):
361
return read_stanza(info_file.readlines()).as_dict()
363
def wait_lock(self, timeout=None, poll=None):
364
"""Wait a certain period for a lock.
366
If the lock can be acquired within the bounded time, it
367
is taken and this returns. Otherwise, LockContention
368
is raised. Either way, this function should return within
369
approximately `timeout` seconds. (It may be a bit more if
370
a transport operation takes a long time to complete.)
373
timeout = _DEFAULT_TIMEOUT_SECONDS
375
poll = _DEFAULT_POLL_SECONDS
377
# XXX: the transport interface doesn't let us guard
378
# against operations there taking a long time.
379
deadline = time.time() + timeout
386
except LockContention:
388
new_info = self.peek()
389
mutter('last_info: %s, new info: %s', last_info, new_info)
390
if new_info is not None and new_info != last_info:
391
if last_info is None:
392
start = 'Unable to obtain'
394
start = 'Lock owner changed for'
396
formatted_info = self._format_lock_info(new_info)
397
if deadline_str is None:
398
deadline_str = time.strftime('%H:%M:%S',
399
time.localtime(deadline))
400
self._report_function('%s %s\n'
402
'%s\n' # locked ... ago
403
'Will continue to try until %s\n',
410
if time.time() + poll < deadline:
413
raise LockContention(self)
415
def lock_write(self):
416
"""Wait for and acquire the lock."""
420
"""Compatibility-mode shared lock.
422
LockDir doesn't support shared read-only locks, so this
423
just pretends that the lock is taken but really does nothing.
425
# At the moment Branches are commonly locked for read, but
426
# we can't rely on that remotely. Once this is cleaned up,
427
# reenable this warning to prevent it coming back in
429
## warn("LockDir.lock_read falls back to write lock")
430
if self._lock_held or self._fake_read_lock:
431
raise LockContention(self)
432
self._fake_read_lock = True
434
def wait(self, timeout=20, poll=0.5):
435
"""Wait a certain period for a lock to be released."""
436
# XXX: the transport interface doesn't let us guard
437
# against operations there taking a long time.
438
deadline = time.time() + timeout
442
if time.time() + poll < deadline:
445
raise LockContention(self)
447
def _format_lock_info(self, info):
448
"""Turn the contents of peek() into something for the user"""
449
lock_url = self.transport.abspath(self.path)
450
delta = time.time() - int(info['start_time'])
452
'lock %s' % (lock_url,),
453
'held by %(user)s on host %(hostname)s [process #%(pid)s]' % info,
454
'locked %s' % (format_delta(delta),),