backup.py 31 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862
  1. #
  2. # The Qubes OS Project, http://www.qubes-os.org
  3. #
  4. # Copyright (C) 2013-2017 Marek Marczykowski-Górecki
  5. # <marmarek@invisiblethingslab.com>
  6. # Copyright (C) 2013 Olivier Médoc <o_medoc@yahoo.fr>
  7. #
  8. # This library is free software; you can redistribute it and/or
  9. # modify it under the terms of the GNU Lesser General Public
  10. # License as published by the Free Software Foundation; either
  11. # version 2.1 of the License, or (at your option) any later version.
  12. #
  13. # This library is distributed in the hope that it will be useful,
  14. # but WITHOUT ANY WARRANTY; without even the implied warranty of
  15. # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  16. # Lesser General Public License for more details.
  17. #
  18. # You should have received a copy of the GNU Lesser General Public
  19. # License along with this library; if not, see <https://www.gnu.org/licenses/>.
  20. #
  21. #
  22. from __future__ import unicode_literals
  23. import itertools
  24. import logging
  25. import functools
  26. import string
  27. import termios
  28. import asyncio
  29. from qubes.utils import size_to_human
  30. import stat
  31. import os
  32. import fcntl
  33. import subprocess
  34. import re
  35. import shutil
  36. import tempfile
  37. import time
  38. import grp
  39. import pwd
  40. import datetime
  41. import qubes
  42. import qubes.core2migration
  43. import qubes.storage
  44. import qubes.storage.file
  45. import qubes.vm.templatevm
  46. QUEUE_ERROR = "ERROR"
  47. QUEUE_FINISHED = "FINISHED"
  48. HEADER_FILENAME = 'backup-header'
  49. DEFAULT_CRYPTO_ALGORITHM = 'aes-256-cbc'
  50. # 'scrypt' is not exactly HMAC algorithm, but a tool we use to
  51. # integrity-protect the data
  52. DEFAULT_HMAC_ALGORITHM = 'scrypt'
  53. DEFAULT_COMPRESSION_FILTER = 'gzip'
  54. CURRENT_BACKUP_FORMAT_VERSION = '4'
  55. # Maximum size of error message get from process stderr (including VM process)
  56. MAX_STDERR_BYTES = 1024
  57. # header + qubes.xml max size
  58. HEADER_QUBES_XML_MAX_SIZE = 1024 * 1024
  59. # hmac file max size - regardless of backup format version!
  60. HMAC_MAX_SIZE = 4096
  61. BLKSIZE = 512
  62. _re_alphanum = re.compile(r'^[A-Za-z0-9-]*$')
  63. class BackupCanceledError(qubes.exc.QubesException):
  64. def __init__(self, msg, tmpdir=None):
  65. super(BackupCanceledError, self).__init__(msg)
  66. self.tmpdir = tmpdir
  67. class BackupHeader(object):
  68. '''Structure describing backup-header file included as the first file in
  69. backup archive
  70. '''
  71. # pylint: disable=too-few-public-methods
  72. header_keys = {
  73. 'version': 'version',
  74. 'encrypted': 'encrypted',
  75. 'compressed': 'compressed',
  76. 'compression-filter': 'compression_filter',
  77. 'crypto-algorithm': 'crypto_algorithm',
  78. 'hmac-algorithm': 'hmac_algorithm',
  79. 'backup-id': 'backup_id'
  80. }
  81. bool_options = ['encrypted', 'compressed']
  82. int_options = ['version']
  83. def __init__(self,
  84. version=None,
  85. encrypted=None,
  86. compressed=None,
  87. compression_filter=None,
  88. hmac_algorithm=None,
  89. crypto_algorithm=None,
  90. backup_id=None):
  91. # repeat the list to help code completion...
  92. self.version = version
  93. self.encrypted = encrypted
  94. self.compressed = compressed
  95. # Options introduced in backup format 3+, which always have a header,
  96. # so no need for fallback in function parameter
  97. self.compression_filter = compression_filter
  98. self.hmac_algorithm = hmac_algorithm
  99. self.crypto_algorithm = crypto_algorithm
  100. self.backup_id = backup_id
  101. def save(self, filename):
  102. with open(filename, "w") as f_header:
  103. # make sure 'version' is the first key
  104. f_header.write('version={}\n'.format(self.version))
  105. for key, attr in self.header_keys.items():
  106. if key == 'version':
  107. continue
  108. if getattr(self, attr) is None:
  109. continue
  110. f_header.write("{!s}={!s}\n".format(key, getattr(self, attr)))
  111. class SendWorker(object):
  112. # pylint: disable=too-few-public-methods
  113. def __init__(self, queue, base_dir, backup_stdout):
  114. super(SendWorker, self).__init__()
  115. self.queue = queue
  116. self.base_dir = base_dir
  117. self.backup_stdout = backup_stdout
  118. self.log = logging.getLogger('qubes.backup')
  119. @asyncio.coroutine
  120. def run(self):
  121. self.log.debug("Started sending thread")
  122. while True:
  123. filename = yield from self.queue.get()
  124. if filename in (QUEUE_FINISHED, QUEUE_ERROR):
  125. break
  126. self.log.debug("Sending file {}".format(filename))
  127. # This tar used for sending data out need to be as simple, as
  128. # simple, as featureless as possible. It will not be
  129. # verified before untaring.
  130. tar_final_cmd = ["tar", "-cO", "--posix",
  131. "-C", self.base_dir, filename]
  132. final_proc = yield from asyncio.create_subprocess_exec(
  133. *tar_final_cmd,
  134. stdout=self.backup_stdout)
  135. retcode = yield from final_proc.wait()
  136. if retcode >= 2:
  137. # handle only exit code 2 (tar fatal error) or
  138. # greater (call failed?)
  139. raise qubes.exc.QubesException(
  140. "ERROR: Failed to write the backup, out of disk space? "
  141. "Check console output or ~/.xsession-errors for details.")
  142. # Delete the file as we don't need it anymore
  143. self.log.debug("Removing file {}".format(filename))
  144. os.remove(os.path.join(self.base_dir, filename))
  145. self.log.debug("Finished sending thread")
  146. @asyncio.coroutine
  147. def launch_proc_with_pty(args, stdin=None, stdout=None, stderr=None, echo=True):
  148. """Similar to pty.fork, but handle stdin/stdout according to parameters
  149. instead of connecting to the pty
  150. :return tuple (subprocess.Popen, pty_master)
  151. """
  152. def set_ctty(ctty_fd, master_fd):
  153. os.setsid()
  154. os.close(master_fd)
  155. fcntl.ioctl(ctty_fd, termios.TIOCSCTTY, 0)
  156. if not echo:
  157. termios_p = termios.tcgetattr(ctty_fd)
  158. # termios_p.c_lflags
  159. termios_p[3] &= ~termios.ECHO
  160. termios.tcsetattr(ctty_fd, termios.TCSANOW, termios_p)
  161. (pty_master, pty_slave) = os.openpty()
  162. p = yield from asyncio.create_subprocess_exec(*args,
  163. stdin=stdin,
  164. stdout=stdout,
  165. stderr=stderr,
  166. preexec_fn=lambda: set_ctty(pty_slave, pty_master))
  167. os.close(pty_slave)
  168. return p, open(pty_master, 'wb+', buffering=0)
  169. @asyncio.coroutine
  170. def launch_scrypt(action, input_name, output_name, passphrase):
  171. '''
  172. Launch 'scrypt' process, pass passphrase to it and return
  173. subprocess.Popen object.
  174. :param action: 'enc' or 'dec'
  175. :param input_name: input path or '-' for stdin
  176. :param output_name: output path or '-' for stdout
  177. :param passphrase: passphrase
  178. :type passphrase: bytes
  179. :return: subprocess.Popen object
  180. '''
  181. command_line = ['scrypt', action, input_name, output_name]
  182. (p, pty) = yield from launch_proc_with_pty(command_line,
  183. stdin=subprocess.PIPE if input_name == '-' else None,
  184. stdout=subprocess.PIPE if output_name == '-' else None,
  185. stderr=subprocess.PIPE,
  186. echo=False)
  187. if action == 'enc':
  188. prompts = (b'Please enter passphrase: ', b'Please confirm passphrase: ')
  189. else:
  190. prompts = (b'Please enter passphrase: ',)
  191. for prompt in prompts:
  192. actual_prompt = yield from p.stderr.read(len(prompt))
  193. if actual_prompt != prompt:
  194. raise qubes.exc.QubesException(
  195. 'Unexpected prompt from scrypt: {}'.format(actual_prompt))
  196. pty.write(passphrase + b'\n')
  197. pty.flush()
  198. # save it here, so garbage collector would not close it (which would kill
  199. # the child)
  200. p.pty = pty
  201. return p
  202. class Backup(object):
  203. '''Backup operation manager. Usage:
  204. >>> app = qubes.Qubes()
  205. >>> # optional - you can use 'None' to use default list (based on
  206. >>> # vm.include_in_backups property)
  207. >>> vms = [app.domains[name] for name in ['my-vm1', 'my-vm2', 'my-vm3']]
  208. >>> exclude_vms = []
  209. >>> options = {
  210. >>> 'encrypted': True,
  211. >>> 'compressed': True,
  212. >>> 'passphrase': 'This is very weak backup passphrase',
  213. >>> 'target_vm': app.domains['sys-usb'],
  214. >>> 'target_dir': '/media/disk',
  215. >>> }
  216. >>> backup_op = Backup(app, vms, exclude_vms, **options)
  217. >>> print(backup_op.get_backup_summary())
  218. >>> asyncio.get_event_loop().run_until_complete(backup_op.backup_do())
  219. See attributes of this object for all available options.
  220. '''
  221. # pylint: disable=too-many-instance-attributes
  222. class FileToBackup(object):
  223. # pylint: disable=too-few-public-methods
  224. def __init__(self, file_path, subdir=None, name=None, size=None):
  225. if size is None:
  226. size = qubes.storage.file.get_disk_usage(file_path)
  227. if subdir is None:
  228. abs_file_path = os.path.abspath(file_path)
  229. abs_base_dir = os.path.abspath(
  230. qubes.config.system_path["qubes_base_dir"]) + '/'
  231. abs_file_dir = os.path.dirname(abs_file_path) + '/'
  232. (nothing, directory, subdir) = \
  233. abs_file_dir.partition(abs_base_dir)
  234. assert nothing == ""
  235. assert directory == abs_base_dir
  236. else:
  237. if subdir and not subdir.endswith('/'):
  238. subdir += '/'
  239. #: real path to the file
  240. self.path = file_path
  241. #: size of the file
  242. self.size = size
  243. #: directory in backup archive where file should be placed
  244. self.subdir = subdir
  245. #: use this name in the archive (aka rename)
  246. self.name = os.path.basename(file_path)
  247. if name is not None:
  248. self.name = name
  249. class VMToBackup(object):
  250. # pylint: disable=too-few-public-methods
  251. def __init__(self, vm, files, subdir):
  252. self.vm = vm
  253. self.files = files
  254. self.subdir = subdir
  255. @property
  256. def size(self):
  257. return functools.reduce(lambda x, y: x + y.size, self.files, 0)
  258. def __init__(self, app, vms_list=None, exclude_list=None, **kwargs):
  259. """
  260. If vms = None, include all (sensible) VMs;
  261. exclude_list is always applied
  262. """
  263. super(Backup, self).__init__()
  264. #: progress of the backup - bytes handled of the current VM
  265. self.chunk_size = 100 * 1024 * 1024
  266. self._current_vm_bytes = 0
  267. #: progress of the backup - bytes handled of finished VMs
  268. self._done_vms_bytes = 0
  269. #: total backup size (set by :py:meth:`get_files_to_backup`)
  270. self.total_backup_bytes = 0
  271. #: application object
  272. self.app = app
  273. #: directory for temporary files - set after creating the directory
  274. self.tmpdir = None
  275. # Backup settings - defaults
  276. #: should the backup be compressed?
  277. self.compressed = True
  278. #: what passphrase should be used to intergrity protect (and encrypt)
  279. #: the backup; required
  280. self.passphrase = None
  281. #: custom compression filter; a program which process stdin to stdout
  282. self.compression_filter = DEFAULT_COMPRESSION_FILTER
  283. #: VM to which backup should be sent (if any)
  284. self.target_vm = None
  285. #: directory to save backup in (either in dom0 or target VM,
  286. #: depending on :py:attr:`target_vm`
  287. self.target_dir = None
  288. #: callback for progress reporting. Will be called with one argument
  289. #: - progress in percents
  290. self.progress_callback = None
  291. #: backup ID, needs to be unique (for a given user),
  292. #: not necessary unpredictable; automatically generated
  293. self.backup_id = datetime.datetime.now().strftime(
  294. '%Y%m%dT%H%M%S-' + str(os.getpid()))
  295. for key, value in kwargs.items():
  296. if hasattr(self, key):
  297. setattr(self, key, value)
  298. else:
  299. raise AttributeError(key)
  300. self.log = logging.getLogger('qubes.backup')
  301. if exclude_list is None:
  302. exclude_list = []
  303. if vms_list is None:
  304. vms_list = [vm for vm in app.domains if vm.include_in_backups]
  305. # Apply exclude list
  306. self.vms_for_backup = [vm for vm in vms_list
  307. if vm.name not in exclude_list]
  308. self._files_to_backup = self.get_files_to_backup()
  309. def __del__(self):
  310. if self.tmpdir and os.path.exists(self.tmpdir):
  311. shutil.rmtree(self.tmpdir)
  312. def get_files_to_backup(self):
  313. files_to_backup = {}
  314. for vm in self.vms_for_backup:
  315. if vm.qid == 0:
  316. # handle dom0 later
  317. continue
  318. subdir = 'vm%d/' % vm.qid
  319. vm_files = []
  320. for name, volume in vm.volumes.items():
  321. if not volume.save_on_stop:
  322. continue
  323. vm_files.append(self.FileToBackup(
  324. volume.export(),
  325. subdir,
  326. name + '.img',
  327. volume.usage))
  328. vm_files.extend(self.FileToBackup(i, subdir)
  329. for i in vm.fire_event('backup-get-files'))
  330. firewall_conf = os.path.join(vm.dir_path, vm.firewall_conf)
  331. if os.path.exists(firewall_conf):
  332. vm_files.append(self.FileToBackup(firewall_conf, subdir))
  333. files_to_backup[vm.qid] = self.VMToBackup(vm, vm_files, subdir)
  334. # Dom0 user home
  335. if 0 in [vm.qid for vm in self.vms_for_backup]:
  336. local_user = grp.getgrnam('qubes').gr_mem[0]
  337. home_dir = pwd.getpwnam(local_user).pw_dir
  338. # Home dir should have only user-owned files, so fix it now
  339. # to prevent permissions problems - some root-owned files can
  340. # left after 'sudo bash' and similar commands
  341. subprocess.check_call(['sudo', 'chown', '-R', local_user, home_dir])
  342. home_to_backup = [
  343. self.FileToBackup(home_dir, 'dom0-home/')]
  344. vm_files = home_to_backup
  345. files_to_backup[0] = self.VMToBackup(self.app.domains[0],
  346. vm_files,
  347. os.path.join('dom0-home', os.path.basename(home_dir)))
  348. self.total_backup_bytes = functools.reduce(
  349. lambda x, y: x + y.size, files_to_backup.values(), 0)
  350. return files_to_backup
  351. def get_backup_summary(self):
  352. summary = ""
  353. fields_to_display = [
  354. {"name": "VM", "width": 16},
  355. {"name": "type", "width": 12},
  356. {"name": "size", "width": 12}
  357. ]
  358. # Display the header
  359. for field in fields_to_display:
  360. fmt = "{{0:-^{0}}}-+".format(field["width"] + 1)
  361. summary += fmt.format('-')
  362. summary += "\n"
  363. for field in fields_to_display:
  364. fmt = "{{0:>{0}}} |".format(field["width"] + 1)
  365. summary += fmt.format(field["name"])
  366. summary += "\n"
  367. for field in fields_to_display:
  368. fmt = "{{0:-^{0}}}-+".format(field["width"] + 1)
  369. summary += fmt.format('-')
  370. summary += "\n"
  371. files_to_backup = self._files_to_backup
  372. for qid, vm_info in files_to_backup.items():
  373. summary_line = ""
  374. fmt = "{{0:>{0}}} |".format(fields_to_display[0]["width"] + 1)
  375. summary_line += fmt.format(vm_info.vm.name)
  376. fmt = "{{0:>{0}}} |".format(fields_to_display[1]["width"] + 1)
  377. if qid == 0:
  378. summary_line += fmt.format("User home")
  379. elif isinstance(vm_info.vm, qubes.vm.templatevm.TemplateVM):
  380. summary_line += fmt.format("Template VM")
  381. else:
  382. summary_line += fmt.format("VM" + (" + Sys" if
  383. vm_info.vm.updateable else ""))
  384. vm_size = vm_info.size
  385. fmt = "{{0:>{0}}} |".format(fields_to_display[2]["width"] + 1)
  386. summary_line += fmt.format(size_to_human(vm_size))
  387. if qid != 0 and vm_info.vm.is_running():
  388. summary_line += " <-- The VM is running, backup will contain " \
  389. "its state from before its start!"
  390. summary += summary_line + "\n"
  391. for field in fields_to_display:
  392. fmt = "{{0:-^{0}}}-+".format(field["width"] + 1)
  393. summary += fmt.format('-')
  394. summary += "\n"
  395. fmt = "{{0:>{0}}} |".format(fields_to_display[0]["width"] + 1)
  396. summary += fmt.format("Total size:")
  397. fmt = "{{0:>{0}}} |".format(
  398. fields_to_display[1]["width"] + 1 + 2 + fields_to_display[2][
  399. "width"] + 1)
  400. summary += fmt.format(size_to_human(self.total_backup_bytes))
  401. summary += "\n"
  402. for field in fields_to_display:
  403. fmt = "{{0:-^{0}}}-+".format(field["width"] + 1)
  404. summary += fmt.format('-')
  405. summary += "\n"
  406. vms_not_for_backup = [vm.name for vm in self.app.domains
  407. if vm not in self.vms_for_backup]
  408. summary += "VMs not selected for backup:\n - " + "\n - ".join(
  409. sorted(vms_not_for_backup)) + "\n"
  410. return summary
  411. @asyncio.coroutine
  412. def _prepare_backup_header(self):
  413. header_file_path = os.path.join(self.tmpdir, HEADER_FILENAME)
  414. backup_header = BackupHeader(
  415. version=CURRENT_BACKUP_FORMAT_VERSION,
  416. hmac_algorithm=DEFAULT_HMAC_ALGORITHM,
  417. encrypted=True,
  418. compressed=self.compressed,
  419. compression_filter=self.compression_filter,
  420. backup_id=self.backup_id,
  421. )
  422. backup_header.save(header_file_path)
  423. # Start encrypt, scrypt will also handle integrity
  424. # protection
  425. scrypt_passphrase = '{filename}!'.format(
  426. filename=HEADER_FILENAME).encode() + self.passphrase
  427. scrypt = yield from launch_scrypt(
  428. 'enc', header_file_path, header_file_path + '.hmac',
  429. scrypt_passphrase)
  430. retcode = yield from scrypt.wait()
  431. if retcode:
  432. raise qubes.exc.QubesException(
  433. "Failed to compute hmac of header file: "
  434. + scrypt.stderr.read())
  435. return HEADER_FILENAME, HEADER_FILENAME + ".hmac"
  436. def _send_progress_update(self):
  437. if not self.total_backup_bytes:
  438. return
  439. if callable(self.progress_callback):
  440. progress = (
  441. 100 * (self._done_vms_bytes + self._current_vm_bytes) /
  442. self.total_backup_bytes)
  443. # pylint: disable=not-callable
  444. self.progress_callback(progress)
  445. def _add_vm_progress(self, bytes_done):
  446. self._current_vm_bytes += bytes_done
  447. self._send_progress_update()
  448. @asyncio.coroutine
  449. def _split_and_send(self, input_stream, file_basename,
  450. output_queue):
  451. '''Split *input_stream* into parts of max *chunk_size* bytes and send
  452. to *output_queue*.
  453. :param input_stream: stream (asyncio reader stream) of data to split
  454. :param file_basename: basename (i.e. without part number and '.enc')
  455. of output files
  456. :param output_queue: asyncio.Queue instance to put produced files to
  457. - queue will get only filenames of written chunks
  458. '''
  459. # Wait for compressor (tar) process to finish or for any
  460. # error of other subprocesses
  461. i = 0
  462. run_error = "size_limit"
  463. scrypt = None
  464. while run_error == "size_limit":
  465. # Prepare a first chunk
  466. chunkfile = file_basename + ".%03d.enc" % i
  467. i += 1
  468. # Start encrypt, scrypt will also handle integrity
  469. # protection
  470. scrypt_passphrase = \
  471. '{backup_id}!{filename}!'.format(
  472. backup_id=self.backup_id,
  473. filename=os.path.relpath(chunkfile[:-4],
  474. self.tmpdir)).encode() + self.passphrase
  475. try:
  476. scrypt = yield from launch_scrypt(
  477. "enc", "-", chunkfile, scrypt_passphrase)
  478. run_error = yield from handle_streams(
  479. input_stream,
  480. scrypt.stdin,
  481. self.chunk_size,
  482. self._add_vm_progress
  483. )
  484. self.log.debug(
  485. "handle_streams returned: {}".format(run_error))
  486. except:
  487. scrypt.terminate()
  488. raise
  489. scrypt.stdin.close()
  490. yield from scrypt.wait()
  491. self.log.debug("scrypt return code: {}".format(
  492. scrypt.returncode))
  493. # Send the chunk to the backup target
  494. yield from output_queue.put(
  495. os.path.relpath(chunkfile, self.tmpdir))
  496. @asyncio.coroutine
  497. def _wrap_and_send_files(self, files_to_backup, output_queue):
  498. for vm_info in files_to_backup:
  499. for file_info in vm_info.files:
  500. self.log.debug("Backing up {}".format(file_info))
  501. backup_tempfile = os.path.join(
  502. self.tmpdir, file_info.subdir,
  503. file_info.name)
  504. self.log.debug("Using temporary location: {}".format(
  505. backup_tempfile))
  506. # Ensure the temporary directory exists
  507. if not os.path.isdir(os.path.dirname(backup_tempfile)):
  508. os.makedirs(os.path.dirname(backup_tempfile))
  509. # The first tar cmd can use any complex feature as we want.
  510. # Files will be verified before untaring this.
  511. # Prefix the path in archive with filename["subdir"] to have it
  512. # verified during untar
  513. tar_cmdline = (["tar", "-Pc", '--sparse',
  514. '-C', os.path.dirname(file_info.path)] +
  515. (['--dereference'] if
  516. file_info.subdir != "dom0-home/" else []) +
  517. ['--xform=s:^%s:%s\\0:' % (
  518. os.path.basename(file_info.path),
  519. file_info.subdir),
  520. os.path.basename(file_info.path)
  521. ])
  522. file_stat = os.stat(file_info.path)
  523. if stat.S_ISBLK(file_stat.st_mode) or \
  524. file_info.name != os.path.basename(file_info.path):
  525. # tar doesn't handle content of block device, use our
  526. # writer
  527. # also use our tar writer when renaming file
  528. assert not stat.S_ISDIR(file_stat.st_mode), \
  529. "Renaming directories not supported"
  530. tar_cmdline = ['python3', '-m', 'qubes.tarwriter',
  531. '--override-name=%s' % (
  532. os.path.join(file_info.subdir, os.path.basename(
  533. file_info.name))),
  534. file_info.path]
  535. if self.compressed:
  536. tar_cmdline.insert(-2,
  537. "--use-compress-program=%s" % self.compression_filter)
  538. self.log.debug(" ".join(tar_cmdline))
  539. # Pipe: tar-sparse | scrypt | tar | backup_target
  540. # TODO: log handle stderr
  541. tar_sparse = yield from asyncio.create_subprocess_exec(
  542. *tar_cmdline, stdout=subprocess.PIPE)
  543. try:
  544. yield from self._split_and_send(
  545. tar_sparse.stdout,
  546. backup_tempfile,
  547. output_queue)
  548. except:
  549. try:
  550. tar_sparse.terminate()
  551. except ProcessLookupError:
  552. pass
  553. raise
  554. yield from tar_sparse.wait()
  555. if tar_sparse.returncode:
  556. raise qubes.exc.QubesException(
  557. 'Failed to archive {} file'.format(file_info.path))
  558. # This VM done, update progress
  559. self._done_vms_bytes += vm_info.size
  560. self._current_vm_bytes = 0
  561. self._send_progress_update()
  562. yield from output_queue.put(QUEUE_FINISHED)
  563. @staticmethod
  564. @asyncio.coroutine
  565. def _monitor_process(proc, error_message):
  566. try:
  567. yield from proc.wait()
  568. except:
  569. proc.terminate()
  570. raise
  571. if proc.returncode:
  572. if proc.stderr is not None:
  573. proc_stderr = (yield from proc.stderr.read())
  574. proc_stderr = proc_stderr.decode('ascii', errors='ignore')
  575. proc_stderr = ''.join(
  576. c for c in proc_stderr if c in string.printable and
  577. c not in '\r\n%{}')
  578. error_message += ': ' + proc_stderr
  579. raise qubes.exc.QubesException(error_message)
  580. @staticmethod
  581. @asyncio.coroutine
  582. def _cancel_on_error(future, previous_task):
  583. '''If further element of chain fail, cancel previous one to
  584. avoid deadlock.
  585. When earlier element of chain fail, it will be handled by
  586. :py:meth:`backup_do`.
  587. The chain is:
  588. :py:meth:`_wrap_and_send_files` -> :py:class:`SendWorker` -> vmproc
  589. '''
  590. try:
  591. yield from future
  592. except: # pylint: disable=bare-except
  593. previous_task.cancel()
  594. @asyncio.coroutine
  595. def backup_do(self):
  596. # pylint: disable=too-many-statements
  597. if self.passphrase is None:
  598. raise qubes.exc.QubesException("No passphrase set")
  599. if not isinstance(self.passphrase, bytes):
  600. self.passphrase = self.passphrase.encode('utf-8')
  601. qubes_xml = self.app.store
  602. self.tmpdir = tempfile.mkdtemp()
  603. shutil.copy(qubes_xml, os.path.join(self.tmpdir, 'qubes.xml'))
  604. qubes_xml = os.path.join(self.tmpdir, 'qubes.xml')
  605. backup_app = qubes.Qubes(qubes_xml, offline_mode=True)
  606. backup_app.events_enabled = False
  607. files_to_backup = self._files_to_backup
  608. # make sure backup_content isn't set initially
  609. for vm in backup_app.domains:
  610. vm.events_enabled = False
  611. vm.features['backup-content'] = False
  612. for qid, vm_info in files_to_backup.items():
  613. # VM is included in the backup
  614. backup_app.domains[qid].features['backup-content'] = True
  615. backup_app.domains[qid].features['backup-path'] = vm_info.subdir
  616. backup_app.domains[qid].features['backup-size'] = vm_info.size
  617. backup_app.save()
  618. del backup_app
  619. vmproc = None
  620. if self.target_vm is not None:
  621. # Prepare the backup target (Qubes service call)
  622. # If APPVM, STDOUT is a PIPE
  623. read_fd, write_fd = os.pipe()
  624. vmproc = yield from self.target_vm.run_service('qubes.Backup',
  625. stdin=read_fd,
  626. stderr=subprocess.PIPE,
  627. stdout=subprocess.DEVNULL)
  628. os.close(read_fd)
  629. os.write(write_fd, (self.target_dir.
  630. replace("\r", "").replace("\n", "") + "\n").encode())
  631. backup_stdout = write_fd
  632. else:
  633. # Prepare the backup target (local file)
  634. if os.path.isdir(self.target_dir):
  635. backup_target = self.target_dir + "/qubes-{0}". \
  636. format(time.strftime("%Y-%m-%dT%H%M%S"))
  637. else:
  638. backup_target = self.target_dir
  639. # Create the target directory
  640. if not os.path.exists(os.path.dirname(self.target_dir)):
  641. raise qubes.exc.QubesException(
  642. "ERROR: the backup directory for {0} does not exists".
  643. format(self.target_dir))
  644. # If not APPVM, STDOUT is a local file
  645. backup_stdout = open(backup_target, 'wb')
  646. # Tar with tape length does not deals well with stdout
  647. # (close stdout between two tapes)
  648. # For this reason, we will use named pipes instead
  649. self.log.debug("Working in {}".format(self.tmpdir))
  650. self.log.debug("Will backup: {}".format(files_to_backup))
  651. header_files = yield from self._prepare_backup_header()
  652. # Setup worker to send encrypted data chunks to the backup_target
  653. to_send = asyncio.Queue(10)
  654. send_proc = SendWorker(to_send, self.tmpdir, backup_stdout)
  655. send_task = asyncio.ensure_future(send_proc.run())
  656. vmproc_task = None
  657. if vmproc is not None:
  658. vmproc_task = asyncio.ensure_future(
  659. self._monitor_process(vmproc,
  660. 'Writing backup to VM {} failed'.format(
  661. self.target_vm.name)))
  662. asyncio.ensure_future(self._cancel_on_error(
  663. vmproc_task, send_task))
  664. for file_name in header_files:
  665. yield from to_send.put(file_name)
  666. qubes_xml_info = self.VMToBackup(
  667. None,
  668. [self.FileToBackup(qubes_xml, '')],
  669. ''
  670. )
  671. inner_archive_task = asyncio.ensure_future(
  672. self._wrap_and_send_files(
  673. itertools.chain([qubes_xml_info], files_to_backup.values()),
  674. to_send
  675. ))
  676. asyncio.ensure_future(
  677. self._cancel_on_error(send_task, inner_archive_task))
  678. try:
  679. try:
  680. yield from inner_archive_task
  681. except:
  682. yield from to_send.put(QUEUE_ERROR)
  683. # in fact we may be handling CancelledError, induced by
  684. # exception in send_task or vmproc_task (and propagated by
  685. # self._cancel_on_error call above); in such a case this
  686. # yield from will raise exception, covering CancelledError -
  687. # this is intended behaviour
  688. if vmproc_task:
  689. yield from vmproc_task
  690. yield from send_task
  691. raise
  692. yield from send_task
  693. finally:
  694. if isinstance(backup_stdout, int):
  695. os.close(backup_stdout)
  696. else:
  697. backup_stdout.close()
  698. try:
  699. if vmproc_task:
  700. yield from vmproc_task
  701. finally:
  702. shutil.rmtree(self.tmpdir)
  703. # Save date of last backup, only when backup succeeded
  704. for qid, vm_info in files_to_backup.items():
  705. if vm_info.vm:
  706. vm_info.vm.backup_timestamp = datetime.datetime.now()
  707. self.app.save()
  708. @asyncio.coroutine
  709. def handle_streams(stream_in, stream_out, size_limit=None,
  710. progress_callback=None):
  711. '''
  712. Copy stream_in to all streams_out and monitor all mentioned processes.
  713. If any of them terminate with non-zero code, interrupt the process. Copy
  714. at most `size_limit` data (if given).
  715. :param stream_in: StreamReader object to read data from
  716. :param stream_out: StreamWriter object to write data to
  717. :param size_limit: int maximum data amount to process
  718. :param progress_callback: callable function to report progress, will be
  719. given copied data size (it should accumulate internally)
  720. :return: "size_limit" or None (no error)
  721. '''
  722. buffer_size = 409600
  723. bytes_copied = 0
  724. while True:
  725. if size_limit:
  726. to_copy = min(buffer_size, size_limit - bytes_copied)
  727. if to_copy <= 0:
  728. return "size_limit"
  729. else:
  730. to_copy = buffer_size
  731. buf = yield from stream_in.read(to_copy)
  732. if not buf:
  733. # done
  734. break
  735. if callable(progress_callback):
  736. progress_callback(len(buf))
  737. stream_out.write(buf)
  738. bytes_copied += len(buf)
  739. return None
  740. # vim:sw=4:et: