src/allmydata/mutable/layout.py

   1
   2 import struct
   3 from allmydata.mutable.common import NeedMoreDataError, UnknownVersionError
   4 from allmydata.interfaces import HASH_SIZE, SALT_SIZE, SDMF_VERSION, \
   5                                  MDMF_VERSION, IMutableSlotWriter
   6 from allmydata.util import mathutil
   7 from twisted.python import failure
   8 from twisted.internet import defer
   9 from zope.interface import implements
  10
  11
  12 # These strings describe the format of the packed structs they help process
  13 # Here's what they mean:
  14 #
  15 #  PREFIX:
  16 #    >: Big-endian byte order; the most significant byte is first (leftmost).
  17 #    B: The version information; an 8 bit version identifier. Stored as
  18 #       an unsigned char. This is currently 00 00 00 00; our modifications
  19 #       will turn it into 00 00 00 01.
  20 #    Q: The sequence number; this is sort of like a revision history for
  21 #       mutable files; they start at 1 and increase as they are changed after
  22 #       being uploaded. Stored as an unsigned long long, which is 8 bytes in
  23 #       length.
  24 #  32s: The root hash of the share hash tree. We use sha-256d, so we use 32
  25 #       characters = 32 bytes to store the value.
  26 #  16s: The salt for the readkey. This is a 16-byte random value, stored as
  27 #       16 characters.
  28 #
  29 #  SIGNED_PREFIX additions, things that are covered by the signature:
  30 #    B: The "k" encoding parameter. We store this as an 8-bit character,
  31 #       which is convenient because our erasure coding scheme cannot
  32 #       encode if you ask for more than 255 pieces.
  33 #    B: The "N" encoding parameter. Stored as an 8-bit character for the
  34 #       same reasons as above.
  35 #    Q: The segment size of the uploaded file. This will essentially be the
  36 #       length of the file in SDMF. An unsigned long long, so we can store
  37 #       files of quite large size.
  38 #    Q: The data length of the uploaded file. Modulo padding, this will be
  39 #       the same of the data length field. Like the data length field, it is
  40 #       an unsigned long long and can be quite large.
  41 #
  42 #   HEADER additions:
  43 #     L: The offset of the signature of this. An unsigned long.
  44 #     L: The offset of the share hash chain. An unsigned long.
  45 #     L: The offset of the block hash tree. An unsigned long.
  46 #     L: The offset of the share data. An unsigned long.
  47 #     Q: The offset of the encrypted private key. An unsigned long long, to
  48 #        account for the possibility of a lot of share data.
  49 #     Q: The offset of the EOF. An unsigned long long, to account for the
  50 #        possibility of a lot of share data.
  51 #
  52 #  After all of these, we have the following:
  53 #    - The verification key: Occupies the space between the end of the header
  54 #      and the start of the signature (i.e.: data[HEADER_LENGTH:o['signature']].
  55 #    - The signature, which goes from the signature offset to the share hash
  56 #      chain offset.
  57 #    - The share hash chain, which goes from the share hash chain offset to
  58 #      the block hash tree offset.
  59 #    - The share data, which goes from the share data offset to the encrypted
  60 #      private key offset.
  61 #    - The encrypted private key offset, which goes until the end of the file.
  62 #
  63 #  The block hash tree in this encoding has only one share, so the offset of
  64 #  the share data will be 32 bits more than the offset of the block hash tree.
  65 #  Given this, we may need to check to see how many bytes a reasonably sized
  66 #  block hash tree will take up.
  67
  68 PREFIX = ">BQ32s16s" # each version has a different prefix
  69 SIGNED_PREFIX = ">BQ32s16s BBQQ" # this is covered by the signature
  70 SIGNED_PREFIX_LENGTH = struct.calcsize(SIGNED_PREFIX)
  71 HEADER = ">BQ32s16s BBQQ LLLLQQ" # includes offsets
  72 HEADER_LENGTH = struct.calcsize(HEADER)
  73 OFFSETS = ">LLLLQQ"
  74 OFFSETS_LENGTH = struct.calcsize(OFFSETS)
  75
  76 # These are still used for some tests.
  77 def unpack_header(data):
  78     o = {}
  79     (version,
  80      seqnum,
  81      root_hash,
  82      IV,
  83      k, N, segsize, datalen,
  84      o['signature'],
  85      o['share_hash_chain'],
  86      o['block_hash_tree'],
  87      o['share_data'],
  88      o['enc_privkey'],
  89      o['EOF']) = struct.unpack(HEADER, data[:HEADER_LENGTH])
  90     return (version, seqnum, root_hash, IV, k, N, segsize, datalen, o)
  91
  92 def unpack_share(data):
  93     assert len(data) >= HEADER_LENGTH
  94     o = {}
  95     (version,
  96      seqnum,
  97      root_hash,
  98      IV,
  99      k, N, segsize, datalen,
 100      o['signature'],
 101      o['share_hash_chain'],
 102      o['block_hash_tree'],
 103      o['share_data'],
 104      o['enc_privkey'],
 105      o['EOF']) = struct.unpack(HEADER, data[:HEADER_LENGTH])
 106
 107     if version != 0:
 108         raise UnknownVersionError("got mutable share version %d, but I only understand version 0" % version)
 109
 110     if len(data) < o['EOF']:
 111         raise NeedMoreDataError(o['EOF'],
 112                                 o['enc_privkey'], o['EOF']-o['enc_privkey'])
 113
 114     pubkey = data[HEADER_LENGTH:o['signature']]
 115     signature = data[o['signature']:o['share_hash_chain']]
 116     share_hash_chain_s = data[o['share_hash_chain']:o['block_hash_tree']]
 117     share_hash_format = ">H32s"
 118     hsize = struct.calcsize(share_hash_format)
 119     assert len(share_hash_chain_s) % hsize == 0, len(share_hash_chain_s)
 120     share_hash_chain = []
 121     for i in range(0, len(share_hash_chain_s), hsize):
 122         chunk = share_hash_chain_s[i:i+hsize]
 123         (hid, h) = struct.unpack(share_hash_format, chunk)
 124         share_hash_chain.append( (hid, h) )
 125     share_hash_chain = dict(share_hash_chain)
 126     block_hash_tree_s = data[o['block_hash_tree']:o['share_data']]
 127     assert len(block_hash_tree_s) % 32 == 0, len(block_hash_tree_s)
 128     block_hash_tree = []
 129     for i in range(0, len(block_hash_tree_s), 32):
 130         block_hash_tree.append(block_hash_tree_s[i:i+32])
 131
 132     share_data = data[o['share_data']:o['enc_privkey']]
 133     enc_privkey = data[o['enc_privkey']:o['EOF']]
 134
 135     return (seqnum, root_hash, IV, k, N, segsize, datalen,
 136             pubkey, signature, share_hash_chain, block_hash_tree,
 137             share_data, enc_privkey)
 138
 139 def get_version_from_checkstring(checkstring):
 140     (t, ) = struct.unpack(">B", checkstring[:1])
 141     return t
 142
 143 def unpack_sdmf_checkstring(checkstring):
 144     cs_len = struct.calcsize(PREFIX)
 145     version, seqnum, root_hash, IV = struct.unpack(PREFIX, checkstring[:cs_len])
 146     if version != 0: # TODO: just ignore the share
 147         raise UnknownVersionError("got mutable share version %d, but I only understand version 0" % version)
 148     return (seqnum, root_hash, IV)
 149
 150 def unpack_mdmf_checkstring(checkstring):
 151     cs_len = struct.calcsize(MDMFCHECKSTRING)
 152     checkstring = checkstring[:cs_len]
 153     version, seqnum, root_hash = struct.unpack(MDMFCHECKSTRING, checkstring)
 154
 155     assert version == 1
 156     return (seqnum, root_hash)
 157
 158 def pack_offsets(verification_key_length, signature_length,
 159                  share_hash_chain_length, block_hash_tree_length,
 160                  share_data_length, encprivkey_length):
 161     post_offset = HEADER_LENGTH
 162     offsets = {}
 163     o1 = offsets['signature'] = post_offset + verification_key_length
 164     o2 = offsets['share_hash_chain'] = o1 + signature_length
 165     o3 = offsets['block_hash_tree'] = o2 + share_hash_chain_length
 166     o4 = offsets['share_data'] = o3 + block_hash_tree_length
 167     o5 = offsets['enc_privkey'] = o4 + share_data_length
 168     offsets['EOF'] = o5 + encprivkey_length
 169
 170     return struct.pack(">LLLLQQ",
 171                        offsets['signature'],
 172                        offsets['share_hash_chain'],
 173                        offsets['block_hash_tree'],
 174                        offsets['share_data'],
 175                        offsets['enc_privkey'],
 176                        offsets['EOF'])
 177
 178 def pack_share(prefix, verification_key, signature,
 179                share_hash_chain, block_hash_tree,
 180                share_data, encprivkey):
 181     share_hash_chain_s = "".join([struct.pack(">H32s", i, share_hash_chain[i])
 182                                   for i in sorted(share_hash_chain.keys())])
 183     for h in block_hash_tree:
 184         assert len(h) == 32
 185     block_hash_tree_s = "".join(block_hash_tree)
 186
 187     offsets = pack_offsets(len(verification_key),
 188                            len(signature),
 189                            len(share_hash_chain_s),
 190                            len(block_hash_tree_s),
 191                            len(share_data),
 192                            len(encprivkey))
 193     final_share = "".join([prefix,
 194                            offsets,
 195                            verification_key,
 196                            signature,
 197                            share_hash_chain_s,
 198                            block_hash_tree_s,
 199                            share_data,
 200                            encprivkey])
 201     return final_share
 202
 203 def pack_prefix(seqnum, root_hash, IV,
 204                 required_shares, total_shares,
 205                 segment_size, data_length):
 206     prefix = struct.pack(SIGNED_PREFIX,
 207                          0, # version,
 208                          seqnum,
 209                          root_hash,
 210                          IV,
 211                          required_shares,
 212                          total_shares,
 213                          segment_size,
 214                          data_length,
 215                          )
 216     return prefix
 217
 218
 219 class SDMFSlotWriteProxy:
 220     implements(IMutableSlotWriter)
 221     """
 222     I represent a remote write slot for an SDMF mutable file. I build a
 223     share in memory, and then write it in one piece to the remote
 224     server. This mimics how SDMF shares were built before MDMF (and the
 225     new MDMF uploader), but provides that functionality in a way that
 226     allows the MDMF uploader to be built without much special-casing for
 227     file format, which makes the uploader code more readable.
 228     """
 229     def __init__(self,
 230                  shnum,
 231                  rref, # a remote reference to a storage server
 232                  storage_index,
 233                  secrets, # (write_enabler, renew_secret, cancel_secret)
 234                  seqnum, # the sequence number of the mutable file
 235                  required_shares,
 236                  total_shares,
 237                  segment_size,
 238                  data_length): # the length of the original file
 239         self.shnum = shnum
 240         self._rref = rref
 241         self._storage_index = storage_index
 242         self._secrets = secrets
 243         self._seqnum = seqnum
 244         self._required_shares = required_shares
 245         self._total_shares = total_shares
 246         self._segment_size = segment_size
 247         self._data_length = data_length
 248
 249         # This is an SDMF file, so it should have only one segment, so,
 250         # modulo padding of the data length, the segment size and the
 251         # data length should be the same.
 252         expected_segment_size = mathutil.next_multiple(data_length,
 253                                                        self._required_shares)
 254         assert expected_segment_size == segment_size
 255
 256         self._block_size = self._segment_size / self._required_shares
 257
 258         # This is meant to mimic how SDMF files were built before MDMF
 259         # entered the picture: we generate each share in its entirety,
 260         # then push it off to the storage server in one write. When
 261         # callers call set_*, they are just populating this dict.
 262         # finish_publishing will stitch these pieces together into a
 263         # coherent share, and then write the coherent share to the
 264         # storage server.
 265         self._share_pieces = {}
 266
 267         # This tells the write logic what checkstring to use when
 268         # writing remote shares.
 269         self._testvs = []
 270
 271         self._readvs = [(0, struct.calcsize(PREFIX))]
 272
 273
 274     def set_checkstring(self, checkstring_or_seqnum,
 275                               root_hash=None,
 276                               salt=None):
 277         """
 278         Set the checkstring that I will pass to the remote server when
 279         writing.
 280
 281             @param checkstring_or_seqnum: A packed checkstring to use,
 282                    or a sequence number. I will treat this as a checkstr
 283
 284         Note that implementations can differ in which semantics they
 285         wish to support for set_checkstring -- they can, for example,
 286         build the checkstring themselves from its constituents, or
 287         some other thing.
 288         """
 289         if root_hash and salt:
 290             checkstring = struct.pack(PREFIX,
 291                                       0,
 292                                       checkstring_or_seqnum,
 293                                       root_hash,
 294                                       salt)
 295         else:
 296             checkstring = checkstring_or_seqnum
 297         self._testvs = [(0, len(checkstring), "eq", checkstring)]
 298
 299
 300     def get_checkstring(self):
 301         """
 302         Get the checkstring that I think currently exists on the remote
 303         server.
 304         """
 305         if self._testvs:
 306             return self._testvs[0][3]
 307         return ""
 308
 309
 310     def put_block(self, data, segnum, salt):
 311         """
 312         Add a block and salt to the share.
 313         """
 314         # SDMF files have only one segment
 315         assert segnum == 0
 316         assert len(data) == self._block_size
 317         assert len(salt) == SALT_SIZE
 318
 319         self._share_pieces['sharedata'] = data
 320         self._share_pieces['salt'] = salt
 321
 322         # TODO: Figure out something intelligent to return.
 323         return defer.succeed(None)
 324
 325
 326     def put_encprivkey(self, encprivkey):
 327         """
 328         Add the encrypted private key to the share.
 329         """
 330         self._share_pieces['encprivkey'] = encprivkey
 331
 332         return defer.succeed(None)
 333
 334
 335     def put_blockhashes(self, blockhashes):
 336         """
 337         Add the block hash tree to the share.
 338         """
 339         assert isinstance(blockhashes, list)
 340         for h in blockhashes:
 341             assert len(h) == HASH_SIZE
 342
 343         # serialize the blockhashes, then set them.
 344         blockhashes_s = "".join(blockhashes)
 345         self._share_pieces['block_hash_tree'] = blockhashes_s
 346
 347         return defer.succeed(None)
 348
 349
 350     def put_sharehashes(self, sharehashes):
 351         """
 352         Add the share hash chain to the share.
 353         """
 354         assert isinstance(sharehashes, dict)
 355         for h in sharehashes.itervalues():
 356             assert len(h) == HASH_SIZE
 357
 358         # serialize the sharehashes, then set them.
 359         sharehashes_s = "".join([struct.pack(">H32s", i, sharehashes[i])
 360                                  for i in sorted(sharehashes.keys())])
 361         self._share_pieces['share_hash_chain'] = sharehashes_s
 362
 363         return defer.succeed(None)
 364
 365
 366     def put_root_hash(self, root_hash):
 367         """
 368         Add the root hash to the share.
 369         """
 370         assert len(root_hash) == HASH_SIZE
 371
 372         self._share_pieces['root_hash'] = root_hash
 373
 374         return defer.succeed(None)
 375
 376
 377     def put_salt(self, salt):
 378         """
 379         Add a salt to an empty SDMF file.
 380         """
 381         assert len(salt) == SALT_SIZE
 382
 383         self._share_pieces['salt'] = salt
 384         self._share_pieces['sharedata'] = ""
 385
 386
 387     def get_signable(self):
 388         """
 389         Return the part of the share that needs to be signed.
 390
 391         SDMF writers need to sign the packed representation of the
 392         first eight fields of the remote share, that is:
 393             - version number (0)
 394             - sequence number
 395             - root of the share hash tree
 396             - salt
 397             - k
 398             - n
 399             - segsize
 400             - datalen
 401
 402         This method is responsible for returning that to callers.
 403         """
 404         return struct.pack(SIGNED_PREFIX,
 405                            0,
 406                            self._seqnum,
 407                            self._share_pieces['root_hash'],
 408                            self._share_pieces['salt'],
 409                            self._required_shares,
 410                            self._total_shares,
 411                            self._segment_size,
 412                            self._data_length)
 413
 414
 415     def put_signature(self, signature):
 416         """
 417         Add the signature to the share.
 418         """
 419         self._share_pieces['signature'] = signature
 420
 421         return defer.succeed(None)
 422
 423
 424     def put_verification_key(self, verification_key):
 425         """
 426         Add the verification key to the share.
 427         """
 428         self._share_pieces['verification_key'] = verification_key
 429
 430         return defer.succeed(None)
 431
 432
 433     def get_verinfo(self):
 434         """
 435         I return my verinfo tuple. This is used by the ServermapUpdater
 436         to keep track of versions of mutable files.
 437
 438         The verinfo tuple for MDMF files contains:
 439             - seqnum
 440             - root hash
 441             - a blank (nothing)
 442             - segsize
 443             - datalen
 444             - k
 445             - n
 446             - prefix (the thing that you sign)
 447             - a tuple of offsets
 448
 449         We include the nonce in MDMF to simplify processing of version
 450         information tuples.
 451
 452         The verinfo tuple for SDMF files is the same, but contains a
 453         16-byte IV instead of a hash of salts.
 454         """
 455         return (self._seqnum,
 456                 self._share_pieces['root_hash'],
 457                 self._share_pieces['salt'],
 458                 self._segment_size,
 459                 self._data_length,
 460                 self._required_shares,
 461                 self._total_shares,
 462                 self.get_signable(),
 463                 self._get_offsets_tuple())
 464
 465     def _get_offsets_dict(self):
 466         post_offset = HEADER_LENGTH
 467         offsets = {}
 468
 469         verification_key_length = len(self._share_pieces['verification_key'])
 470         o1 = offsets['signature'] = post_offset + verification_key_length
 471
 472         signature_length = len(self._share_pieces['signature'])
 473         o2 = offsets['share_hash_chain'] = o1 + signature_length
 474
 475         share_hash_chain_length = len(self._share_pieces['share_hash_chain'])
 476         o3 = offsets['block_hash_tree'] = o2 + share_hash_chain_length
 477
 478         block_hash_tree_length = len(self._share_pieces['block_hash_tree'])
 479         o4 = offsets['share_data'] = o3 + block_hash_tree_length
 480
 481         share_data_length = len(self._share_pieces['sharedata'])
 482         o5 = offsets['enc_privkey'] = o4 + share_data_length
 483
 484         encprivkey_length = len(self._share_pieces['encprivkey'])
 485         offsets['EOF'] = o5 + encprivkey_length
 486         return offsets
 487
 488
 489     def _get_offsets_tuple(self):
 490         offsets = self._get_offsets_dict()
 491         return tuple([(key, value) for key, value in offsets.items()])
 492
 493
 494     def _pack_offsets(self):
 495         offsets = self._get_offsets_dict()
 496         return struct.pack(">LLLLQQ",
 497                            offsets['signature'],
 498                            offsets['share_hash_chain'],
 499                            offsets['block_hash_tree'],
 500                            offsets['share_data'],
 501                            offsets['enc_privkey'],
 502                            offsets['EOF'])
 503
 504
 505     def finish_publishing(self):
 506         """
 507         Do anything necessary to finish writing the share to a remote
 508         server. I require that no further publishing needs to take place
 509         after this method has been called.
 510         """
 511         for k in ["sharedata", "encprivkey", "signature", "verification_key",
 512                   "share_hash_chain", "block_hash_tree"]:
 513             assert k in self._share_pieces, (self.shnum, k, self._share_pieces.keys())
 514         # This is the only method that actually writes something to the
 515         # remote server.
 516         # First, we need to pack the share into data that we can write
 517         # to the remote server in one write.
 518         offsets = self._pack_offsets()
 519         prefix = self.get_signable()
 520         final_share = "".join([prefix,
 521                                offsets,
 522                                self._share_pieces['verification_key'],
 523                                self._share_pieces['signature'],
 524                                self._share_pieces['share_hash_chain'],
 525                                self._share_pieces['block_hash_tree'],
 526                                self._share_pieces['sharedata'],
 527                                self._share_pieces['encprivkey']])
 528
 529         # Our only data vector is going to be writing the final share,
 530         # in its entirely.
 531         datavs = [(0, final_share)]
 532
 533         if not self._testvs:
 534             # Our caller has not provided us with another checkstring
 535             # yet, so we assume that we are writing a new share, and set
 536             # a test vector that will allow a new share to be written.
 537             self._testvs = []
 538             self._testvs.append(tuple([0, 1, "eq", ""]))
 539
 540         tw_vectors = {}
 541         tw_vectors[self.shnum] = (self._testvs, datavs, None)
 542         return self._rref.callRemote("slot_testv_and_readv_and_writev",
 543                                      self._storage_index,
 544                                      self._secrets,
 545                                      tw_vectors,
 546                                      # TODO is it useful to read something?
 547                                      self._readvs)
 548
 549
 550 MDMFHEADER = ">BQ32sBBQQ QQQQQQQQ"
 551 MDMFHEADERWITHOUTOFFSETS = ">BQ32sBBQQ"
 552 MDMFHEADERSIZE = struct.calcsize(MDMFHEADER)
 553 MDMFHEADERWITHOUTOFFSETSSIZE = struct.calcsize(MDMFHEADERWITHOUTOFFSETS)
 554 MDMFCHECKSTRING = ">BQ32s"
 555 MDMFSIGNABLEHEADER = ">BQ32sBBQQ"
 556 MDMFOFFSETS = ">QQQQQQQQ"
 557 MDMFOFFSETS_LENGTH = struct.calcsize(MDMFOFFSETS)
 558
 559 PRIVATE_KEY_SIZE = 1220
 560 SIGNATURE_SIZE = 260
 561 VERIFICATION_KEY_SIZE = 292
 562 # We know we won't have more than 256 shares, and we know that we won't need
 563 # to store more than ln2(256) hash-chain nodes to validate, so that's our
 564 # bound. Each node requires 2 bytes of node-number plus 32 bytes of hash.
 565 SHARE_HASH_CHAIN_SIZE = (2+HASH_SIZE)*mathutil.log_ceil(256, 2)
 566
 567 class MDMFSlotWriteProxy:
 568     implements(IMutableSlotWriter)
 569
 570     """
 571     I represent a remote write slot for an MDMF mutable file.
 572
 573     I abstract away from my caller the details of block and salt
 574     management, and the implementation of the on-disk format for MDMF
 575     shares.
 576     """
 577     # Expected layout, MDMF:
 578     # offset:     size:       name:
 579     #-- signed part --
 580     # 0           1           version number (01)
 581     # 1           8           sequence number
 582     # 9           32          share tree root hash
 583     # 41          1           The "k" encoding parameter
 584     # 42          1           The "N" encoding parameter
 585     # 43          8           The segment size of the uploaded file
 586     # 51          8           The data length of the original plaintext
 587     #-- end signed part --
 588     # 59          8           The offset of the encrypted private key
 589     # 67          8           The offset of the share hash chain
 590     # 75          8           The offset of the signature
 591     # 83          8           The offset of the verification key
 592     # 91          8           The offset of the end of the v. key.
 593     # 99          8           The offset of the share data
 594     # 107         8           The offset of the block hash tree
 595     # 115         8           The offset of EOF
 596     # 123         var         encrypted private key
 597     # var         var         share hash chain
 598     # var         var         signature
 599     # var         var         verification key
 600     # var         large       share data
 601     # var         var         block hash tree
 602     #
 603     # We order the fields that way to make smart downloaders -- downloaders
 604     # which prempetively read a big part of the share -- possible.
 605     #
 606     # The checkstring is the first three fields -- the version number,
 607     # sequence number, root hash and root salt hash. This is consistent
 608     # in meaning to what we have with SDMF files, except now instead of
 609     # using the literal salt, we use a value derived from all of the
 610     # salts -- the share hash root.
 611     #
 612     # The salt is stored before the block for each segment. The block
 613     # hash tree is computed over the combination of block and salt for
 614     # each segment. In this way, we get integrity checking for both
 615     # block and salt with the current block hash tree arrangement.
 616     #
 617     # The ordering of the offsets is different to reflect the dependencies
 618     # that we'll run into with an MDMF file. The expected write flow is
 619     # something like this:
 620     #
 621     #   0: Initialize with the sequence number, encoding parameters and
 622     #      data length. From this, we can deduce the number of segments,
 623     #      and where they should go.. We can also figure out where the
 624     #      encrypted private key should go, because we can figure out how
 625     #      big the share data will be.
 626     #
 627     #   1: Encrypt, encode, and upload the file in chunks. Do something
 628     #      like
 629     #
 630     #       put_block(data, segnum, salt)
 631     #
 632     #      to write a block and a salt to the disk. We can do both of
 633     #      these operations now because we have enough of the offsets to
 634     #      know where to put them.
 635     #
 636     #   2: Put the encrypted private key. Use:
 637     #
 638     #        put_encprivkey(encprivkey)
 639     #
 640     #      Now that we know the length of the private key, we can fill
 641     #      in the offset for the block hash tree.
 642     #
 643     #   3: We're now in a position to upload the block hash tree for
 644     #      a share. Put that using something like:
 645     #
 646     #        put_blockhashes(block_hash_tree)
 647     #
 648     #      Note that block_hash_tree is a list of hashes -- we'll take
 649     #      care of the details of serializing that appropriately. When
 650     #      we get the block hash tree, we are also in a position to
 651     #      calculate the offset for the share hash chain, and fill that
 652     #      into the offsets table.
 653     #
 654     #   4: We're now in a position to upload the share hash chain for
 655     #      a share. Do that with something like:
 656     #
 657     #        put_sharehashes(share_hash_chain)
 658     #
 659     #      share_hash_chain should be a dictionary mapping shnums to
 660     #      32-byte hashes -- the wrapper handles serialization.
 661     #      We'll know where to put the signature at this point, also.
 662     #      The root of this tree will be put explicitly in the next
 663     #      step.
 664     #
 665     #   5: Before putting the signature, we must first put the
 666     #      root_hash. Do this with:
 667     #
 668     #        put_root_hash(root_hash).
 669     #
 670     #      In terms of knowing where to put this value, it was always
 671     #      possible to place it, but it makes sense semantically to
 672     #      place it after the share hash tree, so that's why you do it
 673     #      in this order.
 674     #
 675     #   6: With the root hash put, we can now sign the header. Use:
 676     #
 677     #        get_signable()
 678     #
 679     #      to get the part of the header that you want to sign, and use:
 680     #
 681     #        put_signature(signature)
 682     #
 683     #      to write your signature to the remote server.
 684     #
 685     #   6: Add the verification key, and finish. Do:
 686     #
 687     #        put_verification_key(key)
 688     #
 689     #      and
 690     #
 691     #        finish_publish()
 692     #
 693     # Checkstring management:
 694     #
 695     # To write to a mutable slot, we have to provide test vectors to ensure
 696     # that we are writing to the same data that we think we are. These
 697     # vectors allow us to detect uncoordinated writes; that is, writes
 698     # where both we and some other shareholder are writing to the
 699     # mutable slot, and to report those back to the parts of the program
 700     # doing the writing.
 701     #
 702     # With SDMF, this was easy -- all of the share data was written in
 703     # one go, so it was easy to detect uncoordinated writes, and we only
 704     # had to do it once. With MDMF, not all of the file is written at
 705     # once.
 706     #
 707     # If a share is new, we write out as much of the header as we can
 708     # before writing out anything else. This gives other writers a
 709     # canary that they can use to detect uncoordinated writes, and, if
 710     # they do the same thing, gives us the same canary. We them update
 711     # the share. We won't be able to write out two fields of the header
 712     # -- the share tree hash and the salt hash -- until we finish
 713     # writing out the share. We only require the writer to provide the
 714     # initial checkstring, and keep track of what it should be after
 715     # updates ourselves.
 716     #
 717     # If we haven't written anything yet, then on the first write (which
 718     # will probably be a block + salt of a share), we'll also write out
 719     # the header. On subsequent passes, we'll expect to see the header.
 720     # This changes in two places:
 721     #
 722     #   - When we write out the salt hash
 723     #   - When we write out the root of the share hash tree
 724     #
 725     # since these values will change the header. It is possible that we
 726     # can just make those be written in one operation to minimize
 727     # disruption.
 728     def __init__(self,
 729                  shnum,
 730                  rref, # a remote reference to a storage server
 731                  storage_index,
 732                  secrets, # (write_enabler, renew_secret, cancel_secret)
 733                  seqnum, # the sequence number of the mutable file
 734                  required_shares,
 735                  total_shares,
 736                  segment_size,
 737                  data_length): # the length of the original file
 738         self.shnum = shnum
 739         self._rref = rref
 740         self._storage_index = storage_index
 741         self._seqnum = seqnum
 742         self._required_shares = required_shares
 743         assert self.shnum >= 0 and self.shnum < total_shares
 744         self._total_shares = total_shares
 745         # We build up the offset table as we write things. It is the
 746         # last thing we write to the remote server.
 747         self._offsets = {}
 748         self._testvs = []
 749         # This is a list of write vectors that will be sent to our
 750         # remote server once we are directed to write things there.
 751         self._writevs = []
 752         self._secrets = secrets
 753         # The segment size needs to be a multiple of the k parameter --
 754         # any padding should have been carried out by the publisher
 755         # already.
 756         assert segment_size % required_shares == 0
 757         self._segment_size = segment_size
 758         self._data_length = data_length
 759
 760         # These are set later -- we define them here so that we can
 761         # check for their existence easily
 762
 763         # This is the root of the share hash tree -- the Merkle tree
 764         # over the roots of the block hash trees computed for shares in
 765         # this upload.
 766         self._root_hash = None
 767
 768         # We haven't yet written anything to the remote bucket. By
 769         # setting this, we tell the _write method as much. The write
 770         # method will then know that it also needs to add a write vector
 771         # for the checkstring (or what we have of it) to the first write
 772         # request. We'll then record that value for future use.  If
 773         # we're expecting something to be there already, we need to call
 774         # set_checkstring before we write anything to tell the first
 775         # write about that.
 776         self._written = False
 777
 778         # When writing data to the storage servers, we get a read vector
 779         # for free. We'll read the checkstring, which will help us
 780         # figure out what's gone wrong if a write fails.
 781         self._readv = [(0, struct.calcsize(MDMFCHECKSTRING))]
 782
 783         # We calculate the number of segments because it tells us
 784         # where the salt part of the file ends/share segment begins,
 785         # and also because it provides a useful amount of bounds checking.
 786         self._num_segments = mathutil.div_ceil(self._data_length,
 787                                                self._segment_size)
 788         self._block_size = self._segment_size / self._required_shares
 789         # We also calculate the share size, to help us with block
 790         # constraints later.
 791         tail_size = self._data_length % self._segment_size
 792         if not tail_size:
 793             self._tail_block_size = self._block_size
 794         else:
 795             self._tail_block_size = mathutil.next_multiple(tail_size,
 796                                                            self._required_shares)
 797             self._tail_block_size /= self._required_shares
 798
 799         # We already know where the sharedata starts; right after the end
 800         # of the header (which is defined as the signable part + the offsets)
 801         # We can also calculate where the encrypted private key begins
 802         # from what we know know.
 803         self._actual_block_size = self._block_size + SALT_SIZE
 804         data_size = self._actual_block_size * (self._num_segments - 1)
 805         data_size += self._tail_block_size
 806         data_size += SALT_SIZE
 807         self._offsets['enc_privkey'] = MDMFHEADERSIZE
 808
 809         # We don't define offsets for these because we want them to be
 810         # tightly packed -- this allows us to ignore the responsibility
 811         # of padding individual values, and of removing that padding
 812         # later. So nonconstant_start is where we start writing
 813         # nonconstant data.
 814         nonconstant_start = self._offsets['enc_privkey']
 815         nonconstant_start += PRIVATE_KEY_SIZE
 816         nonconstant_start += SIGNATURE_SIZE
 817         nonconstant_start += VERIFICATION_KEY_SIZE
 818         nonconstant_start += SHARE_HASH_CHAIN_SIZE
 819
 820         self._offsets['share_data'] = nonconstant_start
 821
 822         # Finally, we know how big the share data will be, so we can
 823         # figure out where the block hash tree needs to go.
 824         # XXX: But this will go away if Zooko wants to make it so that
 825         # you don't need to know the size of the file before you start
 826         # uploading it.
 827         self._offsets['block_hash_tree'] = self._offsets['share_data'] + \
 828                     data_size
 829
 830         # Done. We can snow start writing.
 831
 832
 833     def set_checkstring(self,
 834                         seqnum_or_checkstring,
 835                         root_hash=None,
 836                         salt=None):
 837         """
 838         Set checkstring checkstring for the given shnum.
 839
 840         This can be invoked in one of two ways.
 841
 842         With one argument, I assume that you are giving me a literal
 843         checkstring -- e.g., the output of get_checkstring. I will then
 844         set that checkstring as it is. This form is used by unit tests.
 845
 846         With two arguments, I assume that you are giving me a sequence
 847         number and root hash to make a checkstring from. In that case, I
 848         will build a checkstring and set it for you. This form is used
 849         by the publisher.
 850
 851         By default, I assume that I am writing new shares to the grid.
 852         If you don't explcitly set your own checkstring, I will use
 853         one that requires that the remote share not exist. You will want
 854         to use this method if you are updating a share in-place;
 855         otherwise, writes will fail.
 856         """
 857         # You're allowed to overwrite checkstrings with this method;
 858         # I assume that users know what they are doing when they call
 859         # it.
 860         if root_hash:
 861             checkstring = struct.pack(MDMFCHECKSTRING,
 862                                       1,
 863                                       seqnum_or_checkstring,
 864                                       root_hash)
 865         else:
 866             checkstring = seqnum_or_checkstring
 867
 868         if checkstring == "":
 869             # We special-case this, since len("") = 0, but we need
 870             # length of 1 for the case of an empty share to work on the
 871             # storage server, which is what a checkstring that is the
 872             # empty string means.
 873             self._testvs = []
 874         else:
 875             self._testvs = []
 876             self._testvs.append((0, len(checkstring), "eq", checkstring))
 877
 878
 879     def __repr__(self):
 880         return "MDMFSlotWriteProxy for share %d" % self.shnum
 881
 882
 883     def get_checkstring(self):
 884         """
 885         Given a share number, I return a representation of what the
 886         checkstring for that share on the server will look like.
 887
 888         I am mostly used for tests.
 889         """
 890         if self._root_hash:
 891             roothash = self._root_hash
 892         else:
 893             roothash = "\x00" * 32
 894         return struct.pack(MDMFCHECKSTRING,
 895                            1,
 896                            self._seqnum,
 897                            roothash)
 898
 899
 900     def put_block(self, data, segnum, salt):
 901         """
 902         I queue a write vector for the data, salt, and segment number
 903         provided to me. I return None, as I do not actually cause
 904         anything to be written yet.
 905         """
 906         if segnum >= self._num_segments:
 907             raise LayoutInvalid("I won't overwrite the block hash tree")
 908         if len(salt) != SALT_SIZE:
 909             raise LayoutInvalid("I was given a salt of size %d, but "
 910                                 "I wanted a salt of size %d")
 911         if segnum + 1 == self._num_segments:
 912             if len(data) != self._tail_block_size:
 913                 raise LayoutInvalid("I was given the wrong size block to write")
 914         elif len(data) != self._block_size:
 915             raise LayoutInvalid("I was given the wrong size block to write")
 916
 917         # We want to write at len(MDMFHEADER) + segnum * block_size.
 918         offset = self._offsets['share_data'] + \
 919             (self._actual_block_size * segnum)
 920         data = salt + data
 921
 922         self._writevs.append(tuple([offset, data]))
 923
 924
 925     def put_encprivkey(self, encprivkey):
 926         """
 927         I queue a write vector for the encrypted private key provided to
 928         me.
 929         """
 930         assert self._offsets
 931         assert self._offsets['enc_privkey']
 932         # You shouldn't re-write the encprivkey after the block hash
 933         # tree is written, since that could cause the private key to run
 934         # into the block hash tree. Before it writes the block hash
 935         # tree, the block hash tree writing method writes the offset of
 936         # the share hash chain. So that's a good indicator of whether or
 937         # not the block hash tree has been written.
 938         if "signature" in self._offsets:
 939             raise LayoutInvalid("You can't put the encrypted private key "
 940                                 "after putting the share hash chain")
 941
 942         self._offsets['share_hash_chain'] = self._offsets['enc_privkey'] + \
 943                 len(encprivkey)
 944
 945         self._writevs.append(tuple([self._offsets['enc_privkey'], encprivkey]))
 946
 947
 948     def put_blockhashes(self, blockhashes):
 949         """
 950         I queue a write vector to put the block hash tree in blockhashes
 951         onto the remote server.
 952
 953         The encrypted private key must be queued before the block hash
 954         tree, since we need to know how large it is to know where the
 955         block hash tree should go. The block hash tree must be put
 956         before the share hash chain, since its size determines the
 957         offset of the share hash chain.
 958         """
 959         assert self._offsets
 960         assert "block_hash_tree" in self._offsets
 961
 962         assert isinstance(blockhashes, list)
 963
 964         blockhashes_s = "".join(blockhashes)
 965         self._offsets['EOF'] = self._offsets['block_hash_tree'] + len(blockhashes_s)
 966
 967         self._writevs.append(tuple([self._offsets['block_hash_tree'],
 968                                   blockhashes_s]))
 969
 970
 971     def put_sharehashes(self, sharehashes):
 972         """
 973         I queue a write vector to put the share hash chain in my
 974         argument onto the remote server.
 975
 976         The block hash tree must be queued before the share hash chain,
 977         since we need to know where the block hash tree ends before we
 978         can know where the share hash chain starts. The share hash chain
 979         must be put before the signature, since the length of the packed
 980         share hash chain determines the offset of the signature. Also,
 981         semantically, you must know what the root of the block hash tree
 982         is before you can generate a valid signature.
 983         """
 984         assert isinstance(sharehashes, dict)
 985         assert self._offsets
 986         if "share_hash_chain" not in self._offsets:
 987             raise LayoutInvalid("You must put the block hash tree before "
 988                                 "putting the share hash chain")
 989
 990         # The signature comes after the share hash chain. If the
 991         # signature has already been written, we must not write another
 992         # share hash chain. The signature writes the verification key
 993         # offset when it gets sent to the remote server, so we look for
 994         # that.
 995         if "verification_key" in self._offsets:
 996             raise LayoutInvalid("You must write the share hash chain "
 997                                 "before you write the signature")
 998         sharehashes_s = "".join([struct.pack(">H32s", i, sharehashes[i])
 999                                   for i in sorted(sharehashes.keys())])
1000         self._offsets['signature'] = self._offsets['share_hash_chain'] + \
1001             len(sharehashes_s)
1002         self._writevs.append(tuple([self._offsets['share_hash_chain'],
1003                             sharehashes_s]))
1004
1005
1006     def put_root_hash(self, roothash):
1007         """
1008         Put the root hash (the root of the share hash tree) in the
1009         remote slot.
1010         """
1011         # It does not make sense to be able to put the root
1012         # hash without first putting the share hashes, since you need
1013         # the share hashes to generate the root hash.
1014         #
1015         # Signature is defined by the routine that places the share hash
1016         # chain, so it's a good thing to look for in finding out whether
1017         # or not the share hash chain exists on the remote server.
1018         if len(roothash) != HASH_SIZE:
1019             raise LayoutInvalid("hashes and salts must be exactly %d bytes"
1020                                  % HASH_SIZE)
1021         self._root_hash = roothash
1022         # To write both of these values, we update the checkstring on
1023         # the remote server, which includes them
1024         checkstring = self.get_checkstring()
1025         self._writevs.append(tuple([0, checkstring]))
1026         # This write, if successful, changes the checkstring, so we need
1027         # to update our internal checkstring to be consistent with the
1028         # one on the server.
1029
1030
1031     def get_signable(self):
1032         """
1033         Get the first seven fields of the mutable file; the parts that
1034         are signed.
1035         """
1036         if not self._root_hash:
1037             raise LayoutInvalid("You need to set the root hash "
1038                                 "before getting something to "
1039                                 "sign")
1040         return struct.pack(MDMFSIGNABLEHEADER,
1041                            1,
1042                            self._seqnum,
1043                            self._root_hash,
1044                            self._required_shares,
1045                            self._total_shares,
1046                            self._segment_size,
1047                            self._data_length)
1048
1049
1050     def put_signature(self, signature):
1051         """
1052         I queue a write vector for the signature of the MDMF share.
1053
1054         I require that the root hash and share hash chain have been put
1055         to the grid before I will write the signature to the grid.
1056         """
1057         if "signature" not in self._offsets:
1058             raise LayoutInvalid("You must put the share hash chain "
1059         # It does not make sense to put a signature without first
1060         # putting the root hash and the salt hash (since otherwise
1061         # the signature would be incomplete), so we don't allow that.
1062                        "before putting the signature")
1063         if not self._root_hash:
1064             raise LayoutInvalid("You must complete the signed prefix "
1065                                 "before computing a signature")
1066         # If we put the signature after we put the verification key, we
1067         # could end up running into the verification key, and will
1068         # probably screw up the offsets as well. So we don't allow that.
1069         if "verification_key_end" in self._offsets:
1070             raise LayoutInvalid("You can't put the signature after the "
1071                                 "verification key")
1072         # The method that writes the verification key defines the EOF
1073         # offset before writing the verification key, so look for that.
1074         self._offsets['verification_key'] = self._offsets['signature'] +\
1075             len(signature)
1076         self._writevs.append(tuple([self._offsets['signature'], signature]))
1077
1078
1079     def put_verification_key(self, verification_key):
1080         """
1081         I queue a write vector for the verification key.
1082
1083         I require that the signature have been written to the storage
1084         server before I allow the verification key to be written to the
1085         remote server.
1086         """
1087         if "verification_key" not in self._offsets:
1088             raise LayoutInvalid("You must put the signature before you "
1089                                 "can put the verification key")
1090
1091         self._offsets['verification_key_end'] = \
1092             self._offsets['verification_key'] + len(verification_key)
1093         assert self._offsets['verification_key_end'] <= self._offsets['share_data']
1094         self._writevs.append(tuple([self._offsets['verification_key'],
1095                             verification_key]))
1096
1097
1098     def _get_offsets_tuple(self):
1099         return tuple([(key, value) for key, value in self._offsets.items()])
1100
1101
1102     def get_verinfo(self):
1103         return (self._seqnum,
1104                 self._root_hash,
1105                 self._required_shares,
1106                 self._total_shares,
1107                 self._segment_size,
1108                 self._data_length,
1109                 self.get_signable(),
1110                 self._get_offsets_tuple())
1111
1112
1113     def finish_publishing(self):
1114         """
1115         I add a write vector for the offsets table, and then cause all
1116         of the write vectors that I've dealt with so far to be published
1117         to the remote server, ending the write process.
1118         """
1119         if "verification_key_end" not in self._offsets:
1120             raise LayoutInvalid("You must put the verification key before "
1121                                 "you can publish the offsets")
1122         offsets_offset = struct.calcsize(MDMFHEADERWITHOUTOFFSETS)
1123         offsets = struct.pack(MDMFOFFSETS,
1124                               self._offsets['enc_privkey'],
1125                               self._offsets['share_hash_chain'],
1126                               self._offsets['signature'],
1127                               self._offsets['verification_key'],
1128                               self._offsets['verification_key_end'],
1129                               self._offsets['share_data'],
1130                               self._offsets['block_hash_tree'],
1131                               self._offsets['EOF'])
1132         self._writevs.append(tuple([offsets_offset, offsets]))
1133         encoding_parameters_offset = struct.calcsize(MDMFCHECKSTRING)
1134         params = struct.pack(">BBQQ",
1135                              self._required_shares,
1136                              self._total_shares,
1137                              self._segment_size,
1138                              self._data_length)
1139         self._writevs.append(tuple([encoding_parameters_offset, params]))
1140         return self._write(self._writevs)
1141
1142
1143     def _write(self, datavs, on_failure=None, on_success=None):
1144         """I write the data vectors in datavs to the remote slot."""
1145         tw_vectors = {}
1146         if not self._testvs:
1147             self._testvs = []
1148             self._testvs.append(tuple([0, 1, "eq", ""]))
1149         if not self._written:
1150             # Write a new checkstring to the share when we write it, so
1151             # that we have something to check later.
1152             new_checkstring = self.get_checkstring()
1153             datavs.append((0, new_checkstring))
1154             def _first_write():
1155                 self._written = True
1156                 self._testvs = [(0, len(new_checkstring), "eq", new_checkstring)]
1157             on_success = _first_write
1158         tw_vectors[self.shnum] = (self._testvs, datavs, None)
1159         d = self._rref.callRemote("slot_testv_and_readv_and_writev",
1160                                   self._storage_index,
1161                                   self._secrets,
1162                                   tw_vectors,
1163                                   self._readv)
1164         def _result(results):
1165             if isinstance(results, failure.Failure) or not results[0]:
1166                 # Do nothing; the write was unsuccessful.
1167                 if on_failure: on_failure()
1168             else:
1169                 if on_success: on_success()
1170             return results
1171         d.addCallback(_result)
1172         return d
1173
1174
1175 class MDMFSlotReadProxy:
1176     """
1177     I read from a mutable slot filled with data written in the MDMF data
1178     format (which is described above).
1179
1180     I can be initialized with some amount of data, which I will use (if
1181     it is valid) to eliminate some of the need to fetch it from servers.
1182     """
1183     def __init__(self,
1184                  rref,
1185                  storage_index,
1186                  shnum,
1187                  data=""):
1188         # Start the initialization process.
1189         self._rref = rref
1190         self._storage_index = storage_index
1191         self.shnum = shnum
1192
1193         # Before doing anything, the reader is probably going to want to
1194         # verify that the signature is correct. To do that, they'll need
1195         # the verification key, and the signature. To get those, we'll
1196         # need the offset table. So fetch the offset table on the
1197         # assumption that that will be the first thing that a reader is
1198         # going to do.
1199
1200         # The fact that these encoding parameters are None tells us
1201         # that we haven't yet fetched them from the remote share, so we
1202         # should. We could just not set them, but the checks will be
1203         # easier to read if we don't have to use hasattr.
1204         self._version_number = None
1205         self._sequence_number = None
1206         self._root_hash = None
1207         # Filled in if we're dealing with an SDMF file. Unused
1208         # otherwise.
1209         self._salt = None
1210         self._required_shares = None
1211         self._total_shares = None
1212         self._segment_size = None
1213         self._data_length = None
1214         self._offsets = None
1215
1216         # If the user has chosen to initialize us with some data, we'll
1217         # try to satisfy subsequent data requests with that data before
1218         # asking the storage server for it. If
1219         self._data = data
1220         # The way callers interact with cache in the filenode returns
1221         # None if there isn't any cached data, but the way we index the
1222         # cached data requires a string, so convert None to "".
1223         if self._data == None:
1224             self._data = ""
1225
1226
1227     def _maybe_fetch_offsets_and_header(self, force_remote=False):
1228         """
1229         I fetch the offset table and the header from the remote slot if
1230         I don't already have them. If I do have them, I do nothing and
1231         return an empty Deferred.
1232         """
1233         if self._offsets:
1234             return defer.succeed(None)
1235         # At this point, we may be either SDMF or MDMF. Fetching 107
1236         # bytes will be enough to get header and offsets for both SDMF and
1237         # MDMF, though we'll be left with 4 more bytes than we
1238         # need if this ends up being MDMF. This is probably less
1239         # expensive than the cost of a second roundtrip.
1240         readvs = [(0, 123)]
1241         d = self._read(readvs, force_remote)
1242         d.addCallback(self._process_encoding_parameters)
1243         d.addCallback(self._process_offsets)
1244         return d
1245
1246
1247     def _process_encoding_parameters(self, encoding_parameters):
1248         assert self.shnum in encoding_parameters
1249         encoding_parameters = encoding_parameters[self.shnum][0]
1250         # The first byte is the version number. It will tell us what
1251         # to do next.
1252         (verno,) = struct.unpack(">B", encoding_parameters[:1])
1253         if verno == MDMF_VERSION:
1254             read_size = MDMFHEADERWITHOUTOFFSETSSIZE
1255             (verno,
1256              seqnum,
1257              root_hash,
1258              k,
1259              n,
1260              segsize,
1261              datalen) = struct.unpack(MDMFHEADERWITHOUTOFFSETS,
1262                                       encoding_parameters[:read_size])
1263             if segsize == 0 and datalen == 0:
1264                 # Empty file, no segments.
1265                 self._num_segments = 0
1266             else:
1267                 self._num_segments = mathutil.div_ceil(datalen, segsize)
1268
1269         elif verno == SDMF_VERSION:
1270             read_size = SIGNED_PREFIX_LENGTH
1271             (verno,
1272              seqnum,
1273              root_hash,
1274              salt,
1275              k,
1276              n,
1277              segsize,
1278              datalen) = struct.unpack(">BQ32s16s BBQQ",
1279                                 encoding_parameters[:SIGNED_PREFIX_LENGTH])
1280             self._salt = salt
1281             if segsize == 0 and datalen == 0:
1282                 # empty file
1283                 self._num_segments = 0
1284             else:
1285                 # non-empty SDMF files have one segment.
1286                 self._num_segments = 1
1287         else:
1288             raise UnknownVersionError("You asked me to read mutable file "
1289                                       "version %d, but I only understand "
1290                                       "%d and %d" % (verno, SDMF_VERSION,
1291                                                      MDMF_VERSION))
1292
1293         self._version_number = verno
1294         self._sequence_number = seqnum
1295         self._root_hash = root_hash
1296         self._required_shares = k
1297         self._total_shares = n
1298         self._segment_size = segsize
1299         self._data_length = datalen
1300
1301         self._block_size = self._segment_size / self._required_shares
1302         # We can upload empty files, and need to account for this fact
1303         # so as to avoid zero-division and zero-modulo errors.
1304         if datalen > 0:
1305             tail_size = self._data_length % self._segment_size
1306         else:
1307             tail_size = 0
1308         if not tail_size:
1309             self._tail_block_size = self._block_size
1310         else:
1311             self._tail_block_size = mathutil.next_multiple(tail_size,
1312                                                     self._required_shares)
1313             self._tail_block_size /= self._required_shares
1314
1315         return encoding_parameters
1316
1317
1318     def _process_offsets(self, offsets):
1319         if self._version_number == 0:
1320             read_size = OFFSETS_LENGTH
1321             read_offset = SIGNED_PREFIX_LENGTH
1322             end = read_size + read_offset
1323             (signature,
1324              share_hash_chain,
1325              block_hash_tree,
1326              share_data,
1327              enc_privkey,
1328              EOF) = struct.unpack(">LLLLQQ",
1329                                   offsets[read_offset:end])
1330             self._offsets = {}
1331             self._offsets['signature'] = signature
1332             self._offsets['share_data'] = share_data
1333             self._offsets['block_hash_tree'] = block_hash_tree
1334             self._offsets['share_hash_chain'] = share_hash_chain
1335             self._offsets['enc_privkey'] = enc_privkey
1336             self._offsets['EOF'] = EOF
1337
1338         elif self._version_number == 1:
1339             read_offset = MDMFHEADERWITHOUTOFFSETSSIZE
1340             read_length = MDMFOFFSETS_LENGTH
1341             end = read_offset + read_length
1342             (encprivkey,
1343              sharehashes,
1344              signature,
1345              verification_key,
1346              verification_key_end,
1347              sharedata,
1348              blockhashes,
1349              eof) = struct.unpack(MDMFOFFSETS,
1350                                   offsets[read_offset:end])
1351             self._offsets = {}
1352             self._offsets['enc_privkey'] = encprivkey
1353             self._offsets['block_hash_tree'] = blockhashes
1354             self._offsets['share_hash_chain'] = sharehashes
1355             self._offsets['signature'] = signature
1356             self._offsets['verification_key'] = verification_key
1357             self._offsets['verification_key_end']= \
1358                 verification_key_end
1359             self._offsets['EOF'] = eof
1360             self._offsets['share_data'] = sharedata
1361
1362
1363     def get_block_and_salt(self, segnum):
1364         """
1365         I return (block, salt), where block is the block data and
1366         salt is the salt used to encrypt that segment.
1367         """
1368         d = self._maybe_fetch_offsets_and_header()
1369         def _then(ignored):
1370             base_share_offset = self._offsets['share_data']
1371
1372             if segnum + 1 > self._num_segments:
1373                 raise LayoutInvalid("Not a valid segment number")
1374
1375             if self._version_number == 0:
1376                 share_offset = base_share_offset + self._block_size * segnum
1377             else:
1378                 share_offset = base_share_offset + (self._block_size + \
1379                                                     SALT_SIZE) * segnum
1380             if segnum + 1 == self._num_segments:
1381                 data = self._tail_block_size
1382             else:
1383                 data = self._block_size
1384
1385             if self._version_number == 1:
1386                 data += SALT_SIZE
1387
1388             readvs = [(share_offset, data)]
1389             return readvs
1390         d.addCallback(_then)
1391         d.addCallback(lambda readvs: self._read(readvs))
1392         def _process_results(results):
1393             assert self.shnum in results
1394             if self._version_number == 0:
1395                 # We only read the share data, but we know the salt from
1396                 # when we fetched the header
1397                 data = results[self.shnum]
1398                 if not data:
1399                     data = ""
1400                 else:
1401                     assert len(data) == 1
1402                     data = data[0]
1403                 salt = self._salt
1404             else:
1405                 data = results[self.shnum]
1406                 if not data:
1407                     salt = data = ""
1408                 else:
1409                     salt_and_data = results[self.shnum][0]
1410                     salt = salt_and_data[:SALT_SIZE]
1411                     data = salt_and_data[SALT_SIZE:]
1412             return data, salt
1413         d.addCallback(_process_results)
1414         return d
1415
1416
1417     def get_blockhashes(self, needed=None, force_remote=False):
1418         """
1419         I return the block hash tree
1420
1421         I take an optional argument, needed, which is a set of indices
1422         correspond to hashes that I should fetch. If this argument is
1423         missing, I will fetch the entire block hash tree; otherwise, I
1424         may attempt to fetch fewer hashes, based on what needed says
1425         that I should do. Note that I may fetch as many hashes as I
1426         want, so long as the set of hashes that I do fetch is a superset
1427         of the ones that I am asked for, so callers should be prepared
1428         to tolerate additional hashes.
1429         """
1430         # TODO: Return only the parts of the block hash tree necessary
1431         # to validate the blocknum provided?
1432         # This is a good idea, but it is hard to implement correctly. It
1433         # is bad to fetch any one block hash more than once, so we
1434         # probably just want to fetch the whole thing at once and then
1435         # serve it.
1436         if needed == set([]):
1437             return defer.succeed([])
1438         d = self._maybe_fetch_offsets_and_header()
1439         def _then(ignored):
1440             blockhashes_offset = self._offsets['block_hash_tree']
1441             if self._version_number == 1:
1442                 blockhashes_length = self._offsets['EOF'] - blockhashes_offset
1443             else:
1444                 blockhashes_length = self._offsets['share_data'] - blockhashes_offset
1445             readvs = [(blockhashes_offset, blockhashes_length)]
1446             return readvs
1447         d.addCallback(_then)
1448         d.addCallback(lambda readvs:
1449             self._read(readvs, force_remote=force_remote))
1450         def _build_block_hash_tree(results):
1451             assert self.shnum in results
1452
1453             rawhashes = results[self.shnum][0]
1454             results = [rawhashes[i:i+HASH_SIZE]
1455                        for i in range(0, len(rawhashes), HASH_SIZE)]
1456             return results
1457         d.addCallback(_build_block_hash_tree)
1458         return d
1459
1460
1461     def get_sharehashes(self, needed=None, force_remote=False):
1462         """
1463         I return the part of the share hash chain placed to validate
1464         this share.
1465
1466         I take an optional argument, needed. Needed is a set of indices
1467         that correspond to the hashes that I should fetch. If needed is
1468         not present, I will fetch and return the entire share hash
1469         chain. Otherwise, I may fetch and return any part of the share
1470         hash chain that is a superset of the part that I am asked to
1471         fetch. Callers should be prepared to deal with more hashes than
1472         they've asked for.
1473         """
1474         if needed == set([]):
1475             return defer.succeed([])
1476         d = self._maybe_fetch_offsets_and_header()
1477
1478         def _make_readvs(ignored):
1479             sharehashes_offset = self._offsets['share_hash_chain']
1480             if self._version_number == 0:
1481                 sharehashes_length = self._offsets['block_hash_tree'] - sharehashes_offset
1482             else:
1483                 sharehashes_length = self._offsets['signature'] - sharehashes_offset
1484             readvs = [(sharehashes_offset, sharehashes_length)]
1485             return readvs
1486         d.addCallback(_make_readvs)
1487         d.addCallback(lambda readvs:
1488             self._read(readvs, force_remote=force_remote))
1489         def _build_share_hash_chain(results):
1490             assert self.shnum in results
1491
1492             sharehashes = results[self.shnum][0]
1493             results = [sharehashes[i:i+(HASH_SIZE + 2)]
1494                        for i in range(0, len(sharehashes), HASH_SIZE + 2)]
1495             results = dict([struct.unpack(">H32s", data)
1496                             for data in results])
1497             return results
1498         d.addCallback(_build_share_hash_chain)
1499         return d
1500
1501
1502     def get_encprivkey(self):
1503         """
1504         I return the encrypted private key.
1505         """
1506         d = self._maybe_fetch_offsets_and_header()
1507
1508         def _make_readvs(ignored):
1509             privkey_offset = self._offsets['enc_privkey']
1510             if self._version_number == 0:
1511                 privkey_length = self._offsets['EOF'] - privkey_offset
1512             else:
1513                 privkey_length = self._offsets['share_hash_chain'] - privkey_offset
1514             readvs = [(privkey_offset, privkey_length)]
1515             return readvs
1516         d.addCallback(_make_readvs)
1517         d.addCallback(lambda readvs: self._read(readvs))
1518         def _process_results(results):
1519             assert self.shnum in results
1520             privkey = results[self.shnum][0]
1521             return privkey
1522         d.addCallback(_process_results)
1523         return d
1524
1525
1526     def get_signature(self):
1527         """
1528         I return the signature of my share.
1529         """
1530         d = self._maybe_fetch_offsets_and_header()
1531
1532         def _make_readvs(ignored):
1533             signature_offset = self._offsets['signature']
1534             if self._version_number == 1:
1535                 signature_length = self._offsets['verification_key'] - signature_offset
1536             else:
1537                 signature_length = self._offsets['share_hash_chain'] - signature_offset
1538             readvs = [(signature_offset, signature_length)]
1539             return readvs
1540         d.addCallback(_make_readvs)
1541         d.addCallback(lambda readvs: self._read(readvs))
1542         def _process_results(results):
1543             assert self.shnum in results
1544             signature = results[self.shnum][0]
1545             return signature
1546         d.addCallback(_process_results)
1547         return d
1548
1549
1550     def get_verification_key(self):
1551         """
1552         I return the verification key.
1553         """
1554         d = self._maybe_fetch_offsets_and_header()
1555
1556         def _make_readvs(ignored):
1557             if self._version_number == 1:
1558                 vk_offset = self._offsets['verification_key']
1559                 vk_length = self._offsets['verification_key_end'] - vk_offset
1560             else:
1561                 vk_offset = struct.calcsize(">BQ32s16sBBQQLLLLQQ")
1562                 vk_length = self._offsets['signature'] - vk_offset
1563             readvs = [(vk_offset, vk_length)]
1564             return readvs
1565         d.addCallback(_make_readvs)
1566         d.addCallback(lambda readvs: self._read(readvs))
1567         def _process_results(results):
1568             assert self.shnum in results
1569             verification_key = results[self.shnum][0]
1570             return verification_key
1571         d.addCallback(_process_results)
1572         return d
1573
1574
1575     def get_encoding_parameters(self):
1576         """
1577         I return (k, n, segsize, datalen)
1578         """
1579         d = self._maybe_fetch_offsets_and_header()
1580         d.addCallback(lambda ignored:
1581             (self._required_shares,
1582              self._total_shares,
1583              self._segment_size,
1584              self._data_length))
1585         return d
1586
1587
1588     def get_seqnum(self):
1589         """
1590         I return the sequence number for this share.
1591         """
1592         d = self._maybe_fetch_offsets_and_header()
1593         d.addCallback(lambda ignored:
1594             self._sequence_number)
1595         return d
1596
1597
1598     def get_root_hash(self):
1599         """
1600         I return the root of the block hash tree
1601         """
1602         d = self._maybe_fetch_offsets_and_header()
1603         d.addCallback(lambda ignored: self._root_hash)
1604         return d
1605
1606
1607     def get_checkstring(self):
1608         """
1609         I return the packed representation of the following:
1610
1611             - version number
1612             - sequence number
1613             - root hash
1614             - salt hash
1615
1616         which my users use as a checkstring to detect other writers.
1617         """
1618         d = self._maybe_fetch_offsets_and_header()
1619         def _build_checkstring(ignored):
1620             if self._salt:
1621                 checkstring = struct.pack(PREFIX,
1622                                           self._version_number,
1623                                           self._sequence_number,
1624                                           self._root_hash,
1625                                           self._salt)
1626             else:
1627                 checkstring = struct.pack(MDMFCHECKSTRING,
1628                                           self._version_number,
1629                                           self._sequence_number,
1630                                           self._root_hash)
1631
1632             return checkstring
1633         d.addCallback(_build_checkstring)
1634         return d
1635
1636
1637     def get_prefix(self, force_remote):
1638         d = self._maybe_fetch_offsets_and_header(force_remote)
1639         d.addCallback(lambda ignored:
1640             self._build_prefix())
1641         return d
1642
1643
1644     def _build_prefix(self):
1645         # The prefix is another name for the part of the remote share
1646         # that gets signed. It consists of everything up to and
1647         # including the datalength, packed by struct.
1648         if self._version_number == SDMF_VERSION:
1649             return struct.pack(SIGNED_PREFIX,
1650                            self._version_number,
1651                            self._sequence_number,
1652                            self._root_hash,
1653                            self._salt,
1654                            self._required_shares,
1655                            self._total_shares,
1656                            self._segment_size,
1657                            self._data_length)
1658
1659         else:
1660             return struct.pack(MDMFSIGNABLEHEADER,
1661                            self._version_number,
1662                            self._sequence_number,
1663                            self._root_hash,
1664                            self._required_shares,
1665                            self._total_shares,
1666                            self._segment_size,
1667                            self._data_length)
1668
1669
1670     def _get_offsets_tuple(self):
1671         # The offsets tuple is another component of the version
1672         # information tuple. It is basically our offsets dictionary,
1673         # itemized and in a tuple.
1674         return self._offsets.copy()
1675
1676
1677     def get_verinfo(self):
1678         """
1679         I return my verinfo tuple. This is used by the ServermapUpdater
1680         to keep track of versions of mutable files.
1681
1682         The verinfo tuple for MDMF files contains:
1683             - seqnum
1684             - root hash
1685             - a blank (nothing)
1686             - segsize
1687             - datalen
1688             - k
1689             - n
1690             - prefix (the thing that you sign)
1691             - a tuple of offsets
1692
1693         We include the nonce in MDMF to simplify processing of version
1694         information tuples.
1695
1696         The verinfo tuple for SDMF files is the same, but contains a
1697         16-byte IV instead of a hash of salts.
1698         """
1699         d = self._maybe_fetch_offsets_and_header()
1700         def _build_verinfo(ignored):
1701             if self._version_number == SDMF_VERSION:
1702                 salt_to_use = self._salt
1703             else:
1704                 salt_to_use = None
1705             return (self._sequence_number,
1706                     self._root_hash,
1707                     salt_to_use,
1708                     self._segment_size,
1709                     self._data_length,
1710                     self._required_shares,
1711                     self._total_shares,
1712                     self._build_prefix(),
1713                     self._get_offsets_tuple())
1714         d.addCallback(_build_verinfo)
1715         return d
1716
1717
1718     def _read(self, readvs, force_remote=False):
1719         unsatisfiable = filter(lambda x: x[0] + x[1] > len(self._data), readvs)
1720         # TODO: It's entirely possible to tweak this so that it just
1721         # fulfills the requests that it can, and not demand that all
1722         # requests are satisfiable before running it.
1723         if not unsatisfiable and not force_remote:
1724             results = [self._data[offset:offset+length]
1725                        for (offset, length) in readvs]
1726             results = {self.shnum: results}
1727             return defer.succeed(results)
1728         else:
1729             return self._rref.callRemote("slot_readv",
1730                                          self._storage_index,
1731                                          [self.shnum],
1732                                          readvs)
1733
1734
1735     def is_sdmf(self):
1736         """I tell my caller whether or not my remote file is SDMF or MDMF
1737         """
1738         d = self._maybe_fetch_offsets_and_header()
1739         d.addCallback(lambda ignored:
1740             self._version_number == 0)
1741         return d
1742
1743
1744 class LayoutInvalid(Exception):
1745     """
1746     This isn't a valid MDMF mutable file
1747     """